diff --git a/history/disorders/Neuroblastoma/2026-10-21T101350Z-claude-code-1e2920.yaml b/history/disorders/Neuroblastoma/2026-10-21T101350Z-claude-code-1e2920.yaml new file mode 100644 index 00000000000..b2322987749 --- /dev/null +++ b/history/disorders/Neuroblastoma/2026-10-21T101350Z-claude-code-1e2920.yaml @@ -0,0 +1,43 @@ +history_version: 1 +target: + kind: disorder + slug: Neuroblastoma + path: kb/disorders/Neuroblastoma.yaml +session: + id: 2026-10-21T101350Z-claude-code-1e2920 + timestamp: '2026-10-21T10:13:50Z' + actors: + - type: ai_agent + name: claude-code + model: claude-opus-5-5 + agent_tool: claude-code + agent_version: 2.1.241 +links: + issues: + - https://github.com/monarch-initiative/dismech/issues/14976 + prs: [] + urls: [] +events: +- type: EDIT + outcome: changed + sections: + - pathophysiology + - genetic + - description + - references + - evidence + summary: Add telomere maintenance mechanism nodes and evidence + details: | + Added the telomere maintenance mechanism layer to the neuroblastoma entry, on top of the TERT Rearrangement and ATRX Mutation genetic records added in #7114. + + Pathophysiology. Four new nodes: Telomerase Activation (genes TERT; molecular function GO:0003720 telomerase activity, increased; GO:0007004; conforms to enabling_replicative_immortality#Telomere Maintenance Reactivation), Reduced ATRX/DAXX Complex Abundance (genes ATRX; GO:0140658 ATP-dependent chromatin remodeler activity, decreased), Alternative Lengthening of Telomeres (GO:0000722; no conformance, because the module's ALT anchor exists only in open PR #13924) and Unlimited Replicative Potential (GO:0008283, the binding of its anchor; conforms to enabling_replicative_immortality#Replicative Immortality). New edges: MYCN-Driven Proliferation -> Telomerase Activation (intermediate: increased TERT transcription driven by amplified MYCN), Telomerase Activation -> Unlimited Replicative Potential, Reduced ATRX/DAXX Complex Abundance -> Alternative Lengthening of Telomeres (intermediate: loss of ATRX/DAXX-dependent H3.3 deposition at telomeric repeats) and Alternative Lengthening of Telomeres -> Unlimited Replicative Potential. Counter-evidence is recorded as REFUTE items: Dagg 2017 (PMID:28636942; a high-risk subset without telomere lengthening activity) on Unlimited Replicative Potential, and Koneru 2020 (PMID:32291317; MYCN-amplified tumours with low TERT expression, and MYCN overexpression not overcoming TERT repression) on the MYCN edge. + + Genetic records. ATRX Mutation: notes rewritten; the unsourced 'treatment-refractory' and 'essentially absent in infants' were replaced by statements scoped to Sun 2026 (PMID:42558321), Kurihara 2014 (PMID:25487495) and Cheung 2012 (PMID:22416102), and six evidence items were added. The existing PMID:22416102 age item's explanation was corrected: whole-genome sequencing covered 40 of the 104 patients (the other 64 by PCR and Sanger), and its pointer to an adolescent/young-adult subtype framing was dropped along with the notes rewrite; directness and quote_role were added after assessment. TERT Rearrangement: the notes and the first item's explanation were reworded from 'three alternative routes to telomere maintenance' to the two-mechanism framing, with TERT rearrangement as a route into telomerase activation. + + Description and references. The telomere sentence of the top-level description now names the two mechanisms, defines the third telomerase route as high TERT expression without either alteration, and scopes the chemotherapy-response finding to large ATRX deletions and truncating mutations. In references, the bioRxiv preprint DOI:10.1101/2022.08.16.504100 was replaced by its published version, PMID:39635126; the preprint's cache file is left in place, uncited. + + Totals: 54 new evidence items; 15 PMIDs new to the entry (38 in all). + + Not done. No ALT conformance until #13924 merges (follow-up: add it and re-anchor Leiomyosarcoma's ALT node to the same anchor). No variant_origin on any new node, so the entry stays NO_ORIGIN for check-cancer-origin. No telomere-directed treatments. Phenotypes are not yet wired to these nodes. + + Validation: just validate, validate-terms, count-verified-snippets (106/106), validate-disorders, the offline gates, check_gene_activity_grounding against origin/main, node_classes --verify-kb, prose-figure-audit (no new flags). A four-lens pre-review with two independent verifiers per finding was run before the PR, and the findings both verifiers upheld were applied, including the REFUTE regrades. diff --git a/kb/disorders/Neuroblastoma.yaml b/kb/disorders/Neuroblastoma.yaml index d8927bf4c86..f30c0cfbb72 100644 --- a/kb/disorders/Neuroblastoma.yaml +++ b/kb/disorders/Neuroblastoma.yaml @@ -11,10 +11,14 @@ description: >- stage, age, histologic category and grade of differentiation, MYCN status, chromosome 11q status, and DNA ploidy to guide treatment intensity. A distinct axis of neuroblastoma biology is the mechanism of telomere - maintenance: MYCN amplification, TERT rearrangement, and ATRX - mutation/alternative lengthening of telomeres are largely mutually - exclusive routes to telomere maintenance, with ATRX-mutant disease defining - an indolent but treatment-refractory adolescent and young-adult subtype. + maintenance: telomerase is activated through TERT rearrangement, MYCN + amplification or high TERT expression without either alteration, while + alternative lengthening of telomeres (ALT), often with ATRX alteration, + maintains telomeres without telomerase. The two mechanisms are almost mutually + exclusive. ATRX-altered disease is enriched in older children, adolescents and + young adults and often runs a slow but progressive course, with poor response + to induction chemotherapy reported for large ATRX deletions and truncating + mutations. Low-risk neuroblastoma may require only observation, while high-risk disease requires intensive multimodal therapy with immunotherapy. categories: @@ -147,6 +151,101 @@ pathophysiology: term: id: GO:0008283 label: cell population proliferation + downstream: + - target: Telomerase Activation + description: >- + Amplified MYCN raises TERT transcription, which makes MYCN amplification a + route to telomerase activation rather than a separate telomere maintenance + mechanism. TERT expression was higher in MYCN-amplified tumours, and in a + MYCN-amplified cell line TERT was the gene most strongly downregulated + when MYCN was knocked down. Not every MYCN-amplified tumour takes this + route: some express TERT as low as ALT tumours do, and forced MYCN + overexpression did not induce TERT where the promoter was repressed. + causal_link_type: INDIRECT_KNOWN_INTERMEDIATES + intermediate_mechanisms: + - "Increased TERT transcription driven by amplified MYCN" + evidence: + - reference: PMID:26466568 + reference_title: "Telomerase activation by genomic rearrangements in high-risk neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In addition, TERT expression was higher in MYCN-amplified tumours than in the remaining samples (P = 0.021, Fig. 2d)." + explanation: >- + Result from this study's series of patient tumours: TERT expression was + higher in MYCN-amplified neuroblastomas than in tumours carrying neither + MYCN amplification nor a TERT rearrangement (P = 0.021). Support for + MYCN driving telomerase activation is indirect. The sentence reports + that MYCN amplification and higher TERT RNA levels occur together in + tumours, so it shows neither that MYCN causes the rise nor that + telomerase enzyme activity increases; the paper measured that activity + only in cell lines. + - reference: PMID:26466568 + reference_title: "Telomerase activation by genomic rearrangements in high-risk neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Thus, genomic rearrangements at 5p15.33 abrogate silencing of gene expression in this chromosomal region, whereas amplified MYCN selectively upregulates TERT by transcriptional activation." + explanation: >- + The authors' conclusion from their tumour expression data: + MYCN-amplified neuroblastomas upregulated TERT and one neighbouring + AceView-annotated gene (bleyplaby) but not SLC6A18 or SLC6A19, which + TERT-rearranged tumours also upregulated, and TERT was among the most + highly upregulated genes in 92 MYCN-amplified versus 238 low-risk + tumours. So amplified MYCN raises TERT selectively, whereas + rearrangements lift silencing more broadly across the region. + Limitation: the paper does not measure MYCN binding at TERT, so + 'transcriptional activation' is inferred from expression data and + earlier work it cites, and the sentence concerns TERT expression, not + telomerase activity. + - reference: PMID:26466568 + reference_title: "Telomerase activation by genomic rearrangements in high-risk neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: IN_VITRO + snippet: "TERT was the most strongly downregulated gene upon short-interfering-RNA-mediated MYCN knockdown in a MYCN-amplified neuroblastoma cell line" + explanation: >- + Transcriptome sequencing of a cultured MYCN-amplified neuroblastoma line + (IMR-5/75) before and after MYCN knockdown made TERT the most strongly + downregulated gene of all, so TERT expression in an amplified background + depends on MYCN rather than the reverse. That direction is what the + tumour expression comparisons reported elsewhere in the same study + cannot establish on their own. The limitation is scope: a single + cultured cell line, with TERT transcript level as the readout rather + than enzymatic telomerase activity. + - reference: PMID:32291317 + reference_title: "Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma." + supports: REFUTE + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Surprisingly, some MYCN-Amp tumors had TERT expression as low as seen in ALT tumors" + explanation: >- + Bears against this edge holding in every MYCN-amplified tumour. In this + high-risk cohort some MYCN-amplified tumours expressed TERT as low as + ALT tumours do, so the edge's intermediate, increased TERT transcription + in an amplified background, is not reached in every MYCN-amplified + tumour. It qualifies the edge's universality rather than the route's + existence: most MYCN-amplified tumours in the same comparison had high + TERT expression. + - reference: PMID:32291317 + reference_title: "Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma." + supports: REFUTE + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: IN_VITRO + snippet: "MYCN overexpression by itself is not sufficient to overcome the TERT repression in TERT non-expressing cells" + explanation: >- + Forced MYCN overexpression in neuroblastoma cell lines without MYCN + amplification raised TERT expression only where the TERT promoter was + intact and not repressed, so the authors conclude that MYCN cannot by + itself overcome TERT repression. This bears against the edge holding in + every tumour. Indirect for this edge: overexpression in non-amplified + cell lines is not the amplified state the edge describes, and the + readout is TERT expression rather than telomerase activity. evidence: - reference: PMID:41560679 reference_title: "Coexpression of MYCN and ALK Induces Neuroblastoma-Like Tumors From Human iPS Cell-Derived Cranial Neural Crest Cells." @@ -250,6 +349,876 @@ pathophysiology: the citing study's abstract, says that ALK signalling and MYCN amplification together enhance tumor malignancy; that ALK signalling feeds the MYCN-driven proliferative program is an inference from it. +- name: Telomerase Activation + description: >- + Telomerase-mediated telomere maintenance, one of the two telomere + maintenance mechanisms in neuroblastoma. TERT rearrangements at 5p15.33 + juxtapose the TERT coding sequence to strong enhancer elements, with + chromatin remodelling and DNA methylation of the affected region, and + despite their structural diversity they all induced massive transcriptional + upregulation of TERT. MYCN amplification raises TERT transcription, and some + tumours have high TERT expression with neither alteration. In 457 + pretreatment neuroblastomas, TERT rearrangement was found in 46 and MYCN + amplification in 93, and high TERT expression without either in 10 of 223 + tumours with expression data. TERT promoter mutations are rare: none was + found in 131 primary tumours representing the whole spectrum of the disease, + and a single-centre series found them in 17 of 603 tumours. Activated + telomerase, defined by these routes, was associated with the poorest overall + survival and was an independent prognostic marker in multivariable analyses. + Enzymatic telomerase activity has been shown in neuroblastoma cell lines + that carry a TERT rearrangement or MYCN amplification. + biological_scale: MOLECULAR + conforms_to: "enabling_replicative_immortality#Telomere Maintenance Reactivation" + genes: + - preferred_term: TERT + term: + id: hgnc:11730 + label: TERT + molecular_functions: + - preferred_term: telomerase activity + modifier: INCREASED + term: + id: GO:0003720 + label: telomerase activity + biological_processes: + - preferred_term: telomere maintenance via telomerase + modifier: INCREASED + term: + id: GO:0007004 + label: telomere maintenance via telomerase + evidence: + - reference: PMID:26466568 + reference_title: "Telomerase activation by genomic rearrangements in high-risk neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Despite a large structural diversity of these rearrangements, they all induced massive transcriptional upregulation of TERT." + explanation: >- + Abstract result in patient tumours (whole-genome sequencing of 56, an + extended series of 217, plus expression profiling): the TERT + rearrangements were structurally diverse, yet all came with massive + transcriptional upregulation of TERT. 'All' can only cover rearranged + tumours with expression data, such as the 10 in the transcriptome + comparison; 28 rearrangements were detected. 'Induced' is the authors' + causal reading, backed by enhancer juxtaposition and TERT mRNA not + tracking copy number, not by engineering a rearrangement. The quote + concerns TERT mRNA; telomerase enzyme activity was measured only in cell + lines. + - reference: PMID:26466568 + reference_title: "Telomerase activation by genomic rearrangements in high-risk neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "The 5p15.33 rearrangements juxtapose the TERT coding sequence to strong enhancer elements, resulting in massive chromatin remodelling and DNA methylation of the affected region." + explanation: >- + This abstract sentence summarises the paper's own results, mainly from + patient tumour tissue: 5p15.33 rearrangements place TERT next to strong + enhancer elements, with chromatin remodelling and DNA methylation of the + affected region. The enhancer mapping rested on ChIP-seq of only three + TERT-rearranged tumours, and part of the histone-mark survey used cell + lines. The rise in CpG methylation across the TERT locus, measured in 39 + primary tumours (six rearranged), was also seen in MYCN-amplified tumours, + so it is not specific to rearrangement. + - reference: PMID:35100718 + reference_title: "Telomerase Is a Prognostic Marker of Poor Outcome and a Therapeutic Target in Neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "We detected TERT rearrangements in 46 of 457 cases (10.1%), MYCN amplification in 93 of 457 cases (20.4%), and elevated TERT expression in tumors lacking TERT or MYCN alterations in 10 of 223 cases (4.5%)." + explanation: >- + In a series of 457 pretreatment neuroblastomas, the study found TERT + rearrangement in 46 and MYCN amplification in 93 by fluorescence in situ + hybridization, and elevated TERT expression without either alteration in + 10 of 223 tumours profiled by microarray, a subset of the 273 also + examined for ALT. The denominators differ, so the three percentages cannot + be added together. The sentence itself does not mention telomerase: the + authors class these alterations as activated telomerase in a later + sentence, and the abstract reports no telomerase-activity assay on the + tumours. + - reference: PMID:35100718 + reference_title: "Telomerase Is a Prognostic Marker of Poor Outcome and a Therapeutic Target in Neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "The presence of activated telomerase (ie, TERT rearrangements, MYCN amplification, or high TERT expression without these alterations) was associated with poorest overall survival and was an independent prognostic marker in multivariable analyses." + explanation: >- + In an outcome analysis of pretreatment neuroblastomas, activated + telomerase (TERT rearrangement, MYCN amplification, or high TERT + expression without either) was associated with the poorest overall + survival and was an independent prognostic marker in multivariable + analyses. Activation is defined by genomic and expression markers, not + enzyme activity, and the abstract names neither the covariates nor the + number of tumours analysed (FISH for TERT and MYCN covered 457 tumours; + microarray TERT expression covered 223). As MYCN amplification is a + defining route, this does not show prognostic value beyond MYCN + amplification. + - reference: PMID:40638875 + reference_title: "Clinical Significance of TERT Promoter Mutations in Neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "TERT-PM and TERT-RA were detected in 17 of 603 (2.8%) and 31of 168 (18.4%) tumors from individual patients, respectively." + explanation: >- + In a retrospective single-centre (MSK) review of sequenced neuroblastomas, + TERT promoter mutations were found in 17 of 603 tumours profiled on the + MSK-IMPACT targeted panel (2.8%) and TERT rearrangements in 31 of 168 + whole-genome-sequenced tumours (18.4%), so promoter mutations occur, but + less often than rearrangements. The two rates come from different tumour + sets and assays, because rearrangements were scored only by whole-genome + sequencing. The sentence shows frequency, not telomerase activation: + promoter-mutant tumours showed no statistically significant TERT mRNA + difference from TERT-wild-type tumours, only a trend towards higher + expression. + - reference: PMID:26171145 + reference_title: "Absence of telomerase reverse transcriptase promoter mutations in neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In 131 neuroblastoma primary tumors representing the whole spectrum of neuroblastoma, no TERT promoter mutations were detected." + explanation: >- + Sanger sequencing of a 474-bp TERT promoter region covering the C228T and + C250T hotspots found no mutation in any of 131 pretreatment primary + neuroblastomas from a German tumour bank, a cohort the authors describe as + representing the whole spectrum of the disease; this is the description's + 131-tumour observation. Main limitation: zero of 131 cannot rule out a low + frequency, and a later series of 603 tumours (PMID:40638875) found + promoter mutations in 2.8%, mostly in high-risk disease, so the mutation + is rare rather than absent. + - reference: PMID:26171145 + reference_title: "Absence of telomerase reverse transcriptase promoter mutations in neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In conclusion, the TERT promoter mutations are not a frequent mechanism of TERT overexpression in neuroblastoma." + explanation: >- + Drawing chiefly on the 131 primary tumours, none with a promoter mutation, + the authors conclude that promoter mutation is not a frequent mechanism of + TERT overexpression in neuroblastoma, which supports the description's + statement that TERT promoter mutations are rare. Their cell-line finding + (C228T in 3 of 19 lines, all SK-N-SH and two of its subclones, with no + correlation to TERT expression) is separate in-vitro evidence. Limitation: + only a 474-bp region around the TERT start codon was sequenced; the + authors name epigenetic changes and mutations in more distant, unanalysed + regulatory elements as possible alternative mechanisms. + - reference: PMID:26466568 + reference_title: "Telomerase activation by genomic rearrangements in high-risk neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: IN_VITRO + snippet: "Supporting a functional role of TERT, neuroblastoma cell lines bearing rearrangements or amplified MYCN exhibited both upregulated TERT expression and enzymatic telomerase activity." + explanation: >- + In this primary research paper, neuroblastoma cell lines with a TERT + rearrangement or MYCN amplification showed both raised TERT expression and + enzymatic telomerase activity; the Methods state that activity in cell + lines was measured with a PCR-based telomeric repeat amplification + protocol (TRAP) ELISA. The node's telomerase activity (GO:0003720) binding + therefore rests on a measured enzyme activity, not on TERT expression + alone. Limitation: activity was assayed only in cultured cell lines; the + paper's tumour data report TERT expression and telomere content, not + enzyme activity. + downstream: + - target: Unlimited Replicative Potential + description: >- + Telomerase keeps telomeres from eroding, so tumour cells can go on + dividing. High-risk neuroblastoma cell lines with neither telomerase + activity nor ALT showed continuous telomere shortening, and most lost + viability in culture. + causal_link_type: DIRECT + evidence: + - reference: PMID:32291317 + reference_title: "Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: IN_VITRO + snippet: "high-risk neuroblastoma cell lines that were non-ALT, low TERT expressing, and lacked telomerase activity showed continuous shortening of telomeres with growth in vitro and 2 of 3 TERT-L/non-ALT cell lines spontaneously lost viability after multiple passages in culture" + explanation: >- + Three patient-derived high-risk neuroblastoma cell lines that lacked ALT + and were negative for telomerase activity, not merely low in TERT mRNA, + shortened their telomeres continuously with passage in culture, and 2 of + 3 spontaneously lost viability. Losing both telomere maintenance + mechanisms therefore costs replicative capacity, which supports this + edge by its converse. Viability held up in the third line, which had + very high telomere content, so the finding establishes that a + maintenance mechanism is required, not that telomerase alone suffices. +- name: Reduced ATRX/DAXX Complex Abundance + description: >- + Reduced abundance of the ATRX/DAXX chromatin remodelling complex, a + recurrent molecular feature of alternative lengthening of telomeres (ALT) in + neuroblastoma. ATRX encodes an ATP-dependent helicase of the SWI/SNF family + of chromatin remodellers and, with DAXX, deposits the histone variant H3.3 + at telomeric regions. ALT-positive neuroblastomas show reduced ATRX/DAXX + complex abundance, due either to ATRX mutation (55% of them in one cohort) + or to low protein expression, while mutations in the complex members DAXX + and H3F3A were extremely rare. In a pooled analysis of published + ATRX-altered neuroblastomas, 68% of ATRX aberrations were multi-exon + deletions, and 75% of these were predicted to be in-frame. ATRX mutation is + strongly age-dependent: in 104 patients with metastatic disease, sampled to + represent each age group, it was found in tumours from 14 of 32 adolescents + and young adults, 9 of 54 children and none of 18 infants, although infant + cases are reported. In the pooled analysis, each of the three types of ATRX + aberration was more frequent in patients diagnosed at an older age, yet + about 40% of the ATRX-aberrant patients were five years or younger at + diagnosis. + biological_scale: MOLECULAR + genes: + - preferred_term: ATRX + term: + id: hgnc:886 + label: ATRX + molecular_functions: + - preferred_term: ATP-dependent chromatin remodeler activity + modifier: DECREASED + term: + id: GO:0140658 + label: ATP-dependent chromatin remodeler activity + evidence: + - reference: PMID:33627664 + reference_title: "Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "They display reduced ATRX/DAXX complex abundance, due to either ATRX mutations (55%) or low protein expression." + explanation: >- + In the abstract of a multi-omics study that screened 760 primary and + relapsed neuroblastoma tumours for ALT, 'They' refers to the ALT-positive + tumours named in the sentence before. The quote states that these tumours + have reduced ATRX/DAXX complex abundance, due to ATRX mutation (55%) or + low protein expression. Limitations: the 55% is 33 of the 60 ALT-positive + tumours in the sequenced discovery cohort, and abundance was read from + separate ATRX and DAXX protein levels in a 34-tumour mass-spectrometry + proteome, with protein data for only five ATRX wild-type ALT tumours. + - reference: PMID:22416102 + reference_title: "Association of age at diagnosis and genetic mutations in patients with neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: BACKGROUND + evidence_source: IN_VITRO + snippet: "ATRX is part of a multi-protein complex that includes DAXX and plays a role in regulating ATP-dependent chromatin remodeling, nucleosome assembly, and telomere maintenance." + explanation: >- + In the discussion of a tumour-sequencing study of 104 patients with + metastatic neuroblastoma, the authors state that ATRX is part of a + multi-protein complex that includes DAXX and plays a role in regulating + ATP-dependent chromatin remodelling, nucleosome assembly and telomere + maintenance. This directly supports the ATRX/DAXX chromatin remodelling + complex named in the description, though the sentence names neither the + SWI/SNF helicase family nor H3.3 deposition at telomeric regions. The + sentence restates established biochemistry and cell biology without a + citation and reports no patient data, and the same discussion calls the + telomere-maintenance role proposed. + - reference: PMID:35384159 + reference_title: "Mutational spectrum of ATRX aberrations in neuroblastoma and associated patient and tumor characteristics." + supports: SUPPORT + directness: DIRECT + quote_role: BACKGROUND + evidence_source: IN_VITRO + snippet: "ATRX, located on the X-chromosome, is a gene that encodes for an ATP-dependent helicase of the SWI/SNF family of chromatin remodelers. ... ATRX forms a chromatin remodeling complex with DAXX, which performs the replication-independent deposition of the histone variant H3.3 at pericentromeric and telomeric regions." + explanation: >- + The introduction of this pooled analysis of published ATRX-altered + neuroblastoma cases states that ATRX encodes an ATP-dependent helicase of + the SWI/SNF family of chromatin remodellers and, citing earlier work, that + it forms a chromatin-remodelling complex with DAXX which deposits histone + H3.3 at pericentromeric and telomeric regions independently of + replication. This states directly the description's account of what the + complex normally does. It restates earlier work rather than the paper's + own data, and says nothing about how abundant the complex is in + neuroblastoma. + - reference: PMID:35384159 + reference_title: "Mutational spectrum of ATRX aberrations in neuroblastoma and associated patient and tumor characteristics." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "We subdivide the ATRX aberrations into nonsense, missense, and multiexon deletions (MEDs) and show that 68% of them are MEDs. Of these MEDs, 75% are predicted to be in-frame." + explanation: >- + This meta-analysis pooled 127 neuroblastoma patients and three cell lines + with ATRX aberrations from 20 published papers; all but two tumours were + stage 4. Of the aberrations, 68% were multi-exon deletions, and 75% of the + deletions analysed in detail (59 of 79) were shown or predicted to be + in-frame, the figures the description gives. In-frame status is mostly + predicted, the paper measures neither ATRX protein nor ATRX/DAXX complex + levels, and a possible gain-of-function effect of the in-frame deletions, + raised by the authors, still needs experimental confirmation. + - reference: PMID:33627664 + reference_title: "Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Mutations in the ATRX complex members DAXX and H3F3A were extremely rare" + explanation: >- + In whole-genome sequencing of an ALT-enriched discovery cohort of 165 + patient tumours, 60 of them ALT-positive, mutations in the ATRX complex + members DAXX and H3F3A were extremely rare: one ATRX-wild-type tumour had + a large inversion affecting DAXX, and another had an A92S H3F3A mutation + alongside mutated ATRX. The sentence states this rarity directly for the + ALT-positive tumours of the cohort. It is a single ALT-enriched cohort, so + it gives no frequency for neuroblastoma as a whole and does not itself + show how the complex is reduced. + - reference: PMID:22416102 + reference_title: "Association of age at diagnosis and genetic mutations in patients with neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In both cohorts (N = 104), mutations in the ATRX gene were identified in 44% (95% CI, 28%-62%) of tumors from patients in the adolescent and young adult group (14 of 32), in 17% (95% CI, 9%-29%) of tumors from children (9 of 54), and in 0% (95% CI, 0%-17%) of tumors from infants (0 of 18)." + explanation: >- + Retrospective tumour sequencing in 104 patients with metastatic + neuroblastoma (whole-genome sequencing in 40, PCR/Sanger in 64) found ATRX + mutations in tumours from 14 of 32 adolescents and young adults (12 years + or older), 9 of 54 children (18 months to under 12 years) and 0 of 18 + infants (under 18 months), the age gradient the description gives. All but + one tumour came from one centre's registry, chosen at random to represent + each age group, so these are per-age-group rates, not an overall + frequency; the less sensitive PCR/Sanger assay may also undercount. + - reference: PMID:35384159 + reference_title: "Mutational spectrum of ATRX aberrations in neuroblastoma and associated patient and tumor characteristics." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "We also confirm that all three ATRX mutation types are more often identified in patients diagnosed at an older age, but still approximately 40% of the patients are aged 5 years or younger at diagnosis." + explanation: >- + This pooled analysis of ATRX-aberrant neuroblastomas from 20 published + papers (127 patients, all but two stage 4) compared age at diagnosis with + ATRX wild-type stage 4 tumours from one sequencing cohort. Nonsense, + missense and multi-exon-deletion cases were each diagnosed at an older + age, yet about 40% of ATRX-aberrant patients were five years or younger at + diagnosis. The cases are pooled from published reports rather than one + population-based cohort, and the authors note that their reference cohort + might introduce a systematic bias. + - reference: PMID:25487495 + reference_title: "Clinical features of ATRX or DAXX mutated neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In ALT cases, three infants showed ATRX deletions, and all seven cases detected after 18months of age showed poor prognosis." + explanation: >- + In this series of 121 neuroblastomas, next-generation sequencing found + ATRX deletions in three infants among the 11 ALT-positive cases, so the + lesion does occur in infancy. The abstract does not define 'infant', but + it contrasts these three with the seven ALT cases detected after 18 months + of age. It gives no count of infants in the series, so it shows that + infant cases exist, not how common they are, and the ALT subset is small. + - reference: PMID:27793328 + reference_title: "Telomere biology including TERT rearrangements in neuroblastoma: a useful indicator for surgical treatments." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In 81 infant cases, MYCN amplification, TERT rearrangements and ATRX mutations were detected in 3, 4, and 3 cases, respectively." + explanation: >- + In this cohort of 121 neuroblastomas, ATRX mutations were found in 3 of 81 + infant cases, so in infancy the lesion is present but rare. The cohort's + ATRX status was already known and its 11 ATRX/DAXX-mutated ALT cases match + those of PMID:25487495, so these are almost certainly the same three + infants, not an independent finding. The abstract does not define 'infant' + or say how many of its 67 mass-screening cases were infants, so the rate + may not hold for clinically detected infant tumours. + downstream: + - target: Alternative Lengthening of Telomeres + description: >- + Loss of ATRX/DAXX complex function permits alternative lengthening of + telomeres. ATRX mutation is associated with loss of nuclear ATRX protein + and with ALT, and in one series every ALT tumour carried an ATRX or DAXX + alteration. Without ATRX, histone H3.3 is not efficiently deposited at + telomeric repeats, replication forks stall there, and the telomeres + undergo the homologous recombination that leads to ALT. + causal_link_type: INDIRECT_KNOWN_INTERMEDIATES + intermediate_mechanisms: + - "Loss of ATRX/DAXX-dependent deposition of histone H3.3 at telomeric repeats" + evidence: + - reference: PMID:22416102 + reference_title: "Association of age at diagnosis and genetic mutations in patients with neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "ATRX mutations were associated with loss of nuclear ATRX protein, longer telomeres and ALT." + explanation: >- + In patients' stage 4 neuroblastomas, all 8 ATRX-mutant tumours with + tissue had complete or mosaic loss of nuclear ATRX protein and the + ultrabright telomeric FISH signal that is a hallmark of ALT, while only + 1 of 20 ATRX wild-type tumours had ALT. One wild-type tumour with mosaic + nuclear ATRX loss had no ALT. Longer telomeres by sequencing did not + separate the groups significantly (8 of 10 mutant, 12 of 30 wild-type). + These are co-occurrences, not a causal test, and the paper reports ATRX + immunostaining, not DAXX protein or complex levels. + - reference: PMID:25487495 + reference_title: "Clinical features of ATRX or DAXX mutated neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In NBLs, ALT was caused by ATRX or DAXX alterations." + explanation: >- + In a separate clinical series of 121 neuroblastomas, 10 of the 11 + tumours with ALT carried ATRX alterations and 1 a DAXX mutation, while + only 1 of the 110 tumours with normal or shortened telomeres carried + either. The causal wording is the authors' reading of this co-occurrence + in 11 ALT cases, and the DAXX evidence is a single tumour. The abstract + reports gene alterations found by sequencing, not ATRX or DAXX protein + levels, so reading them as reduced complex abundance is an inference. + - reference: PMID:32060267 + reference_title: "MYCN amplification and ATRX mutations are incompatible in neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: BACKGROUND + evidence_source: IN_VITRO + snippet: "In cells lacking ATRX, H3.3 is not efficiently deposited at the telomeric G-rich regions, G4 structures form, and replication forks stall ... Consequently, telomeres undergo homologous recombination leading to alternative lengthening of telomeres (ALT)" + explanation: >- + The introduction of this neuroblastoma study, citing earlier work, + states the step named as this edge's intermediate: in cells lacking + ATRX, histone H3.3 is not efficiently deposited at telomeric G-rich + regions, G4 structures form and replication forks stall, and the + telomeres then undergo homologous recombination, leading to ALT. Support + is indirect because this restates cell biology from earlier work rather + than a measurement in neuroblastoma tumours, and it concerns loss of + ATRX, not of DAXX. In the paper's own experiments, ATRX knockdown + changed neither telomere content nor ALT status in any cell line tested + after two weeks. +- name: Alternative Lengthening of Telomeres + description: >- + Alternative lengthening of telomeres (ALT), the recombination-based telomere + maintenance mechanism that works without telomerase. Its measured frequency + depends on cohort and assay: C-circle screening classified 9.2% of a + 720-tumour screening cohort as ALT-positive, ALT-associated PML nuclear + bodies marked 49 of 273 pretreatment neuroblastomas, and the C-circle assay + marked 25 of 107 high-risk tumours from patients older than 18 months. In a + registry cohort of 40 relapsed tumours, 47.5% were ALT-positive. ATRX + alterations are found in only about half to 60% of ALT tumours. In primary + tumours, ALT and the routes to telomerase activation occur almost mutually + exclusively, and ATRX mutation and MYCN amplification co-occurred in only 1 + of 819 tumours. ALT-positive neuroblastomas proliferate slowly, with a + protracted clinical course, yet children with these tumours have a dismal + outcome. About a third of ALT-positive tumours present with localized + disease classified as low or intermediate risk, and in one cohort survival + did not differ between ALT-positive tumours classified high risk and those + classified low or intermediate risk. + biological_scale: MOLECULAR + biological_processes: + - preferred_term: telomere maintenance via recombination + modifier: INCREASED + term: + id: GO:0000722 + label: telomere maintenance via recombination + evidence: + - reference: PMID:32291317 + reference_title: "Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "ALT was identified in 23.4% of high-risk neuroblastoma tumors and genomic alterations in ATRX were detected in 60% of ALT tumors; 40% of ALT tumors lacked genomic alterations in known ALT-associated genes." + explanation: >- + In a Children's Oncology Group cohort of patient primary tumours taken at + diagnosis, this abstract sentence states the paper's own result: + C-circle-defined ALT in 23.4% of high-risk tumours, with ATRX alterations + in 60% of ALT tumours. The other 40% were ATRX wild-type by Sanger + sequencing, and no DAXX, H3F3A or SMARCAL1 alteration was detected in + them. The 60% (15 of 25) includes two cases called only from RNA-seq + overexpression of in-frame-fusion ATRX transcripts; DNA sequencing alone + found ATRX alterations in 13 of 23 sequenced ALT tumours (57%). + - reference: PMID:32291317 + reference_title: "Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Of the 107 high-risk tumors (age > 18 months), 25 (23.4%; 95% CI: 15.3-31.4%) were positive for the ALT biomarker C-circles" + explanation: >- + This Results sentence gives the count behind the 23.4%: among 134 primary + tumours taken at diagnosis and assayed, 25 of the 107 high-risk tumours + from patients older than 18 months were positive for C-circles, a highly + specific DNA marker of ALT measured by qPCR. It supports the frequency, + assay and population, not the share of ALT tumours with ATRX alterations. + The figure is specific to this assay: three APB-positive tumours fell + below the C-circle cutoff, and counting either marker gives 28 of 107 ALT + tumours. + - reference: PMID:33627664 + reference_title: "Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "9.2% of the neuroblastomas in a screening cohort of 720 tumors were classified as ALT-positive" + explanation: >- + C-circle screening of a cohort of 720 neuroblastomas from German and + Austrian trials and registries classified 9.2% as ALT-positive. Because + the cohort was not restricted to high-risk disease, this is the + broadest-spectrum frequency among the cohorts cited here. It is a trial + and registry series rather than a population-based sample, and C-circle + frequencies are not directly comparable with those from ALT-associated PML + nuclear bodies. + - reference: PMID:33627664 + reference_title: "Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Importantly, 47.5% of the INFORM tumors were ALT-activated, indicating a strong enrichment of this molecular subgroup in relapsed cases" + explanation: >- + The same C-circle assay applied to 40 relapsed neuroblastomas from the + INFORM registry found 47.5% ALT-positive, which the authors read as strong + enrichment of ALT among relapsed cases. The cohort is small and not a + random sample of relapses: the authors note that registry entry requires + sufficient general condition and a life expectancy of at least three + months, which may underrepresent the most aggressive, mostly + MYCN-amplified or TERT-rearranged tumours. + - reference: PMID:39635126 + reference_title: "Copy-number dosage regulates telomere maintenance and disease-associated pathways in neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: BACKGROUND + evidence_source: HUMAN_CLINICAL + snippet: "Alterations at the ATRX locus are key drivers of ALT but only present in ~50% of ALT tumors." + explanation: >- + The abstract of this whole-genome and transcriptome study of 115 primary + neuroblastomas states that alterations at the ATRX locus are key drivers + of ALT yet are present in only about half of ALT tumours. The sentence + frames the study and restates earlier work rather than reporting the + paper's own measurement; the introduction credits the frequency (ATRX + mutations in 50-60% of ALT-positive neuroblastomas) to earlier studies. + The figure counts genomic alterations only; the discussion cites earlier + work finding reduced ATRX protein in ATRX-wild-type ALT-positive + neuroblastomas. + - reference: PMID:39635126 + reference_title: "Copy-number dosage regulates telomere maintenance and disease-associated pathways in neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Among ALT tumors 9/20 (45%) harbored ATRX alterations and a single tumor (NBL49) a H3F3A missense mutation (p.A48E)." + explanation: >- + In the paper's own cohort of 115 primary neuroblastomas, profiled by + tumour and matched-normal whole-genome sequencing plus tumour RNA-seq, 9 + of the 20 tumours classified as ALT (45%) carried ATRX alterations. This + supports only the frequency, not that ATRX alterations drive ALT. The ALT + group is small, and ALT status was inferred from telomere content in the + sequencing data, checked against ALT-associated PML nuclear bodies and + C-circles in subsets of 52 and 36 tumours. + - reference: PMID:35100718 + reference_title: "Telomerase Is a Prognostic Marker of Poor Outcome and a Therapeutic Target in Neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "ALT activation was found in 49 of 273 cases (17.9%). All these alterations occurred almost mutually exclusively and were associated with unfavorable prognostic variables and adverse outcome." + explanation: >- + This patient series covered 457 pretreatment neuroblastomas, 273 of them + assessed for ALT by detecting ALT-associated PML nuclear bodies. The + authors report that ALT and the three events they count as activated + telomerase (TERT rearrangement, MYCN amplification, or high TERT + expression without either) occurred almost mutually exclusively and were + associated with unfavourable prognostic variables and adverse outcome. + 'Almost' means some overlap the abstract does not count, comparisons + involving ALT cover only the 273 assessed tumours, and only activated + telomerase was reported as an independent prognostic marker in + multivariable analyses. + - reference: PMID:32060267 + reference_title: "MYCN amplification and ATRX mutations are incompatible in neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Here we show that ATRX mutations and MYCN amplification are mutually exclusive across all ages and stages in neuroblastoma." + explanation: >- + The authors' abstract summary of their own analysis of patient tumours, + made by merging the COG, TARGET and PCGP data (819 tumours), states the + exclusivity directly and extends it across all ages and stages. The + exclusivity is statistical rather than absolute: one of the 819 tumours + carried both lesions. For infants the result is uninformative, because + none of the 213 COG patients diagnosed before 18 months had an ATRX + mutation. + - reference: PMID:32060267 + reference_title: "MYCN amplification and ATRX mutations are incompatible in neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In the cohort of 819 tumors, ATRX mutations (n = 64) and MYCN amplification (n = 140) showed statistically significant mutual exclusivity (p = 0.0037, Cochrane-Mantel-Haenszel test) (Fig. 1e and Supplementary Note 1). Only 1 of 819 (0.1%) tumors had both lesions." + explanation: >- + The authors' pooled analysis of patient tumours (COG, TARGET and PCGP data + merged) found significant mutual exclusivity of ATRX mutation and MYCN + amplification among 819 tumours, with only one tumour carrying both, which + supports near-complete rather than absolute exclusivity. That tumour was + MYCN-amplified by FISH at the COG neuroblastoma reference lab but not by + sequencing read depth, and repeat FISH found regional heterogeneity of + MYCN amplification. In the COG cohort alone, ATRX mutation frequency did + not differ significantly by MYCN status, so the significance comes from + the merged data. + - reference: PMID:33627664 + reference_title: "Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "ALT-positive neuroblastomas proliferate slowly, which is reflected by a protracted clinical course of disease." + explanation: >- + This abstract sentence, from a study screening 760 primary and relapsed + neuroblastomas for ALT, states both slow proliferation and a protracted + clinical course. Slow proliferation rests on proteomics of 34 tumours, + where proliferation-associated protein pathways were lower than in + telomerase-activated tumours, and on fewer cycling cells than in + MYCN-amplified tumours. Survival curves (60 ALT-positive vs 34 + MYCN-amplified tumours) showed significantly longer overall survival time + but similarly poor event-free survival. Both findings are relative to + telomerase-activated tumours, and the link between them is the authors' + interpretation rather than a tested relationship. + - reference: PMID:33627664 + reference_title: "Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Nevertheless, children with an ALT-positive neuroblastoma have dismal outcome." + explanation: >- + Closing abstract sentence of a retrospective neuroblastoma study, resting + on the authors' own Kaplan-Meier analysis (60 ALT-positive vs 34 + MYCN-amplified tumours). It states that despite the slow proliferation and + protracted course reported before it, children with ALT-positive + neuroblastoma have a dismal outcome. In that analysis event-free survival + was similarly poor to that with MYCN amplification, but overall survival + was significantly longer, so 'dismal' means long-term outcome. Limitation: + the cohort was deliberately enriched for ALT tumours, with survival data + from German GPOH-trial patients (1991-2018), so it gives no + population-based estimate. + - reference: PMID:33627664 + reference_title: "Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "about one third of the ALT-positive neuroblastomas had localized disease and were stratified at diagnosis as low or intermediate risk using a standard risk classification system" + explanation: >- + Although ALT was associated with stage 4, high-risk disease and older age + at diagnosis, about a third of the ALT-positive tumours in this study had + localized disease and were classified as low or intermediate risk at + diagnosis, so ALT is not confined to high-risk disease at presentation. + None of the 83 stage 4S tumours screened was ALT-positive. The sentence + reports the authors' own cohort and says nothing about outcome. + - reference: PMID:33627664 + reference_title: "Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "No difference in event-free or overall survival was observed between patients with ALT-positive tumors stratified into high-risk and those predicted as low/intermediate risk" + explanation: >- + In the sequenced, ALT-enriched cohort, event-free and overall survival did + not differ between patients with ALT-positive tumours classified high risk + and those classified low or intermediate risk, which the authors read as + current risk stratification underestimating the poor prognosis of + ALT-positive tumours. The comparison is retrospective with small groups, + so the absence of a difference is weak evidence of equivalent outcome, and + the cohort was deliberately enriched for ALT. + downstream: + - target: Unlimited Replicative Potential + description: >- + ALT maintains telomere length without telomerase, giving the tumour cells + that use it an unlimited replicative capacity. High-risk neuroblastoma + cell lines with neither ALT nor telomerase activity showed continuous + telomere shortening, and most lost viability in culture. + causal_link_type: DIRECT + evidence: + - reference: PMID:40638875 + reference_title: "Clinical Significance of TERT Promoter Mutations in Neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: BACKGROUND + evidence_source: HUMAN_CLINICAL + snippet: "Advances in molecular techniques have highlighted the critical role of telomere maintenance mechanisms that stabilize telomeres for unlimited replicative capacity and immortality in the pathogenesis of high-risk neuroblastoma ... Telomere maintenance is mediated by telomerase-dependent (classical) and independent (alternate) pathways." + explanation: >- + The introduction of this neuroblastoma clinical cohort study states that + telomere maintenance mechanisms, which stabilise telomeres for unlimited + replicative capacity and immortality, have a critical role in high-risk + neuroblastoma, and that telomere maintenance is mediated by + telomerase-dependent and telomerase-independent ('alternate') pathways; + later in the paragraph the authors name this route the alternate + lengthening of telomeres (ALT) pathway. Support for ALT is indirect + because the first quoted sentence covers telomere maintenance mechanisms + as a class. It is cited background about high-risk disease, not this + study's finding. + - reference: PMID:24792489 + reference_title: "Alternative lengthening of telomeres in neuroblastoma cell lines is associated with a lack of MYCN genomic amplification and with p53 pathway aberrations." + supports: SUPPORT + directness: INDIRECT + quote_role: BACKGROUND + evidence_source: IN_VITRO + snippet: "Alternative lengthening of telomeres (ALT) is a telomerase-independent telomere length maintenance mechanism that enables the unlimited proliferation of a subset of cancer cells." + explanation: >- + The opening sentence of this study of 40 neuroblastoma cell lines + describes alternative lengthening of telomeres (ALT) as a + telomerase-independent telomere length maintenance mechanism that + enables the unlimited proliferation of a subset of cancer cells. It + names ALT and unlimited proliferation together, but as general + background about cancer cells rather than a neuroblastoma result, so + support for neuroblastoma is indirect. The abstract's next sentence says + only that some neuroblastoma tumours appear to maintain telomere length + by activating ALT. + - reference: PMID:32291317 + reference_title: "Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: IN_VITRO + snippet: "high-risk neuroblastoma cell lines that were non-ALT, low TERT expressing, and lacked telomerase activity showed continuous shortening of telomeres with growth in vitro and 2 of 3 TERT-L/non-ALT cell lines spontaneously lost viability after multiple passages in culture" + explanation: >- + Three patient-derived high-risk neuroblastoma cell lines that lacked ALT + as well as telomerase activity shortened their telomeres continuously in + culture, and 2 of 3 spontaneously lost viability, so losing both + telomere maintenance mechanisms costs replicative capacity, which + supports this edge by its converse. Viability held up in the third line, + which had very high telomere content, so the result shows that a + maintenance mechanism is required for sustained growth, not that ALT + alone suffices. +- name: Unlimited Replicative Potential + description: >- + The ability of tumour cells to keep dividing indefinitely, which a telomere + maintenance mechanism (telomerase activation or ALT) provides. Its binding, + increased cell population proliferation, follows the module node it conforms + to and means expansion continuing past the limit that telomere shortening + sets on normal cells, not a faster division rate: ALT-positive + neuroblastomas proliferate more slowly than telomerase-activated ones. + Aggressive neuroblastomas show a prolonged evolution during which they + acquire telomere maintenance mechanisms, whereas tumours with a favourable + prognosis expand clonally after a short evolution. In pretreatment + neuroblastomas, patients whose tumours lacked a telomere maintenance + mechanism had an excellent prognosis and those whose tumours had one did + substantially worse. Within high-risk disease the picture is less clear: in + one cohort the telomere maintenance groups differed in overall but not in + event-free survival, and a subset of high-risk tumours lacks significant + telomere lengthening activity, with cells derived from them proliferating + for more than 200 population doublings while their telomeres shortened. + biological_scale: CELLULAR + conforms_to: "enabling_replicative_immortality#Replicative Immortality" + biological_processes: + - preferred_term: cell population proliferation + modifier: INCREASED + term: + id: GO:0008283 + label: cell population proliferation + evidence: + - reference: PMID:30523111 + reference_title: "A mechanistic classification of clinical phenotypes in neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "We found that patients whose tumors lacked telomere maintenance mechanisms had an excellent prognosis, whereas the prognosis of patients whose tumors harbored telomere maintenance mechanisms was substantially worse." + explanation: >- + In a patient cohort, 208 of 416 sequenced pretreatment neuroblastomas were + assessed for both telomerase activation (MYCN amplification, TERT + rearrangement or high TERT expression) and ALT (ALT-associated PML nuclear + bodies). Patients whose tumours lacked a telomere maintenance mechanism + had an excellent prognosis, while those whose tumours had one did + substantially worse. Indirect for this node: an outcome association does + not measure replicative capacity, and the link rests on telomere + maintenance being essential for immortal proliferation, a premise the + paper takes from earlier work. + - reference: PMID:32291317 + reference_title: "Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Event-free survival was similar among TERT-high, ALT+, or TERT-low/non-ALT patients. However, overall survival (OS) for TERT-low/non-ALT patients was significantly higher relative to TERT-high or ALT patients (log-rank test; P < 0.01) independent of current clinical and molecular prognostic markers." + explanation: >- + In 107 stage 4 high-risk patients diagnosed after 18 months of age, + grouped by C-circle (ALT) status and TERT mRNA above or below the median, + event-free survival did not differ significantly across the three groups + (P = 0.137), while overall survival was significantly higher for + TERT-low/non-ALT patients. The authors attribute this to more successful + salvage therapy after an event, and hypothesise that tumours lacking a + telomere maintenance mechanism keep eroding their telomeres, which + eventually lets salvage therapy eradicate them. Indirect for this node: + survival does not measure replicative capacity, and the link rests on that + telomere-erosion hypothesis. Limitations are a single cohort not uniformly + treated on current protocols and a median TERT cutoff that may not be + optimal. + - reference: PMID:28636942 + reference_title: "Extensive Proliferation of Human Cancer Cells with Ever-Shorter Telomeres." + supports: REFUTE + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Here, we show that a subset of high-risk neuroblastomas (with ~50% 5-year mortality) lacked significant TLM activity." + explanation: >- + Bears against this node as a universal feature of high-risk neuroblastoma, + though not against it in tumours that maintain their telomeres. In patient + tumours the authors identify a subset of high-risk neuroblastomas that + lacked significant telomere lengthening mechanism (TLM) activity, TLM + being their term for telomerase or alternative lengthening of telomeres. + Indirect from this sentence alone, because the abstract also raises an + unrecognised lengthening mechanism or a false-negative assay; the + continued telomere shortening in cells derived from these tumours argues + against both. The ~50% 5-year mortality describes the high-risk category + the subset was drawn from, and the abstract gives neither the subset's + size nor the assay. + - reference: PMID:28636942 + reference_title: "Extensive Proliferation of Human Cancer Cells with Ever-Shorter Telomeres." + supports: REFUTE + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: IN_VITRO + snippet: "Cancer cells derived from these highly aggressive tumors initially had long telomeres and proliferated for >200 population doublings with ever-shorter telomeres." + explanation: >- + Cell lines derived from the high-risk neuroblastomas that lacked + significant TLM activity grew for more than 200 population doublings while + their telomeres kept shortening, so their extensive proliferation drew + down a finite telomere reserve rather than reflecting unlimited + replicative potential. Continued shortening over that many doublings + argues against a false-negative assay or an unrecognised maintenance + mechanism. Indirect because the inference to tumour behaviour is from cell + culture. + - reference: PMID:36973454 + reference_title: "Neuroblastoma arises in early fetal development and its evolutionary duration predicts outcome." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Neuroblastomas with favorable prognosis expand clonally after short evolution, whereas aggressive neuroblastomas show prolonged evolution during which they acquire telomere maintenance mechanisms." + explanation: >- + The quoted conclusion comes from deep whole-genome sequencing and + molecular-clock analysis of patient neuroblastomas covering the clinical + spectrum: aggressive tumours evolve over a prolonged period and acquire + telomere maintenance mechanisms during it, whereas favourable-prognosis + tumours expand clonally after a short evolution. An independent cohort (n + = 86) was assembled to test the outcome prediction from the duration of + evolution; the telomere maintenance enrichment behind this conclusion was + computed on both cohorts pooled. Indirect for this node: the sentence is + about when telomere maintenance is acquired, and the link to replicative + capacity rests on telomere maintenance being what confers it. The order of + events is reconstructed from single tumour samples, and the rearrangements + conferring telomere maintenance cannot themselves be timed from mutation + densities. + - reference: PMID:36973454 + reference_title: "Neuroblastoma arises in early fetal development and its evolutionary duration predicts outcome." + supports: SUPPORT + directness: INDIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "12% of early-MRCA tumors but 81% of late-MRCA tumors gained telomere maintenance - via MYCN amplification, ALT or TERT rearrangement - in the combined discovery and validation cohorts" + explanation: >- + Across the pooled discovery and validation cohorts, telomere maintenance + - by MYCN amplification, ALT or TERT rearrangement - had been gained in + 81% of patient tumours whose most recent common ancestor emerged late, + after a long evolution, against 12% of those with an early common + ancestor. This quantifies the enrichment behind the timing conclusion. It + is prevalence by evolutionary class rather than a timing measurement, and + the authors state that these small rearrangements cannot be timed from + mutation densities, so acquisition during the prolonged evolution follows + by inference. histopathology: - name: Neural Crest Tumor finding_term: @@ -695,11 +1664,12 @@ genetic: id: hgnc:11730 label: TERT notes: >- - Genomic rearrangements juxtaposing enhancers to the TERT locus drive - massive TERT upregulation and telomerase-mediated telomere maintenance. - TERT rearrangement occurs in high-risk disease and is largely mutually - exclusive with MYCN amplification and ATRX mutation, defining one of three - alternative routes to telomere maintenance in neuroblastoma. + Genomic rearrangements juxtaposing enhancers to the TERT locus drive massive + TERT upregulation and telomerase-mediated telomere maintenance. TERT + rearrangement occurs in high-risk disease and is largely mutually exclusive + with MYCN amplification and ATRX mutation. Like MYCN amplification, it is a + genomic route into telomerase activation, one of the two telomere + maintenance mechanisms in neuroblastoma. evidence: - reference: PMID:26466568 reference_title: "Telomerase activation by genomic rearrangements in high-risk neuroblastoma." @@ -707,9 +1677,9 @@ genetic: evidence_source: HUMAN_CLINICAL snippet: "These rearrangements occurred only in high-risk neuroblastomas (12/39, 31%) in a mutually exclusive fashion with MYCN amplifications and ATRX mutations, which are known genetic events in this tumour type." explanation: >- - Establishes TERT rearrangement as a high-risk-restricted event that is - mutually exclusive with MYCN amplification and ATRX mutation, defining - the three-way telomere-maintenance structure of neuroblastoma genomics. + Establishes TERT rearrangement as a high-risk-restricted event in the + whole-genome discovery series, where it was mutually exclusive with MYCN + amplification and ATRX mutation. - reference: PMID:26466568 reference_title: "Telomerase activation by genomic rearrangements in high-risk neuroblastoma." supports: SUPPORT @@ -724,12 +1694,16 @@ genetic: id: hgnc:886 label: ATRX notes: >- - Somatic ATRX loss-of-function drives telomere maintenance by the - alternative lengthening of telomeres (ALT) pathway. ATRX-mutant - neuroblastoma is strongly age-dependent, characterising an - adolescent/young-adult subtype with indolent but chronically progressive, - treatment-refractory disease. ATRX mutation is essentially absent in - infants and is mutually exclusive with MYCN amplification. + Somatic ATRX alterations, most often multi-exon deletions, are the commonest + known lesion behind alternative lengthening of telomeres (ALT) in + neuroblastoma, although many ALT tumours carry none. ATRX alteration is + strongly age-dependent, rare in infancy and most frequent in adolescents and + young adults, and it is almost mutually exclusive with MYCN amplification. + The clinical course is often slow but progressive. Poor response to standard + induction chemotherapy has been reported for large ATRX deletions and + truncating mutations in one two-centre retrospective cohort, and + chemo-resistance with late recurrence in a small series of ATRX- or + DAXX-altered tumours. evidence: - reference: PMID:22416102 reference_title: "Association of age at diagnosis and genetic mutations in patients with neuroblastoma." @@ -742,12 +1716,108 @@ genetic: - reference: PMID:22416102 reference_title: "Association of age at diagnosis and genetic mutations in patients with neuroblastoma." supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT evidence_source: HUMAN_CLINICAL snippet: "ATRX mutations were associated with age at diagnosis in children and young adults with stage 4 neuroblastoma." explanation: >- - Whole-genome sequencing across 104 patients establishes the age - dependence of ATRX mutation, supporting the adolescent/young-adult - subtype framing. + This is the conclusion of a retrospective study of 104 patients with stage + 4 neuroblastoma. ATRX mutations were found by whole-genome sequencing in a + 40-patient discovery cohort and by PCR and Sanger sequencing in a + 64-patient validation cohort, in tumours from 14 of 32 adolescents and + young adults, 9 of 54 children and none of 18 infants, which supports the + age dependence of ATRX alteration. Patients were sampled to represent each + age group, so these are per-age-group rates, not an overall frequency. + - reference: PMID:42558321 + reference_title: "ATRX large deletions and truncating mutations are associated with primary induction chemotherapy resistance in high-risk neuroblastoma: a multicenter propensity-weighted cohort study." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Compared with the WT group (71.0%), the LDT group exhibited a substantially lower ORR (36.0%; P < 0.001), with a primary induction-refractory rate of 64.0%." + explanation: >- + A retrospective cohort at two Chinese hospitals compared 33 high-risk + neuroblastomas carrying large ATRX deletions or truncating mutations with + 92 ATRX wild-type tumours, scoring response to multi-agent, + platinum-containing induction by blinded central review. The objective + response rate was 36% against 71%, and the quoted 64% induction-refractory + rate is the complement of that response rate rather than a separate + measurement. The figures are propensity-weighted estimates from one + unreplicated cohort of 33 such tumours, and missense ATRX variants did not + differ from wild-type. + - reference: PMID:25487495 + reference_title: "Clinical features of ATRX or DAXX mutated neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "ATRX altered cases without MYCN amplification detected at greater than 18months showed poor prognosis, suggesting that ATRX or DAXX alterations are a particular NBL subtype. Since these tumors showed chemo-resistance and late recurrence, complete resection in a surgical approach should be performed to improve patient prognosis." + explanation: >- + Among 121 neuroblastomas, all 11 ALT tumours carried ATRX (10) or DAXX (1) + alterations, and the authors conclude that this ATRX/DAXX-altered subtype + showed chemo-resistance and late recurrence. Only that observation is + cited, not the complete-resection recommendation that follows it. The + abstract presents no chemotherapy-response or relapse data and reports + poor prognosis only for the seven ALT cases detected after 18 months of + age. Its purpose statement says the group had already reported ALT as a + possible biomarker of chemotherapy response and late recurrence; the + ATRX/DAXX genotype is the new element. + - reference: PMID:22416102 + reference_title: "Association of age at diagnosis and genetic mutations in patients with neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "Specifically, patients with ATRX mutations, ultrabright telomere FISH signal characteristic of ALT, loss of nuclear ATRX protein and absence of MYCN gain may be more likely to have a chronic but progressive clinical course with standard therapeutic approaches and may require a different treatment strategy." + explanation: >- + This retrospective genomic study of 104 patients with advanced-stage + neuroblastoma hedges its own conclusion: patients whose tumors combine + ATRX mutation, the ultrabright telomere FISH signal of ALT, nuclear ATRX + loss and no MYCN gain 'may be more likely' to follow a chronic but + progressive course with standard therapy and 'may require' a different + treatment strategy. The study tests neither: its course data are + descriptive (4 of 5 ATRX-mutant children in the discovery cohort lived at + least twice as long as their time to first relapse), and it has no + treatment-response analysis. + - reference: PMID:35384159 + reference_title: "Mutational spectrum of ATRX aberrations in neuroblastoma and associated patient and tumor characteristics." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "We subdivide the ATRX aberrations into nonsense, missense, and multiexon deletions (MEDs) and show that 68% of them are MEDs. Of these MEDs, 75% are predicted to be in-frame." + explanation: >- + This meta-analysis pooled 127 neuroblastoma patients and three cell lines + with ATRX aberrations from 20 published papers and found that 68% of the + aberrations were multi-exon deletions, the commonest class, and that 75% + of these were predicted to be in-frame. All but two of the tumours were + stage 4, and the pooled series are published cases rather than an + unselected cohort. + - reference: PMID:32291317 + reference_title: "Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "ALT was identified in 23.4% of high-risk neuroblastoma tumors and genomic alterations in ATRX were detected in 60% of ALT tumors; 40% of ALT tumors lacked genomic alterations in known ALT-associated genes." + explanation: >- + In a Children's Oncology Group cohort of high-risk primary tumours, ATRX + alterations were found in 60% of the tumours with ALT, and the other 40% + had no alteration in known ALT-associated genes. ATRX is therefore the + commonest known altered gene behind ALT, while many ALT tumours carry no + known lesion. The cohort is high-risk only. + - reference: PMID:32060267 + reference_title: "MYCN amplification and ATRX mutations are incompatible in neuroblastoma." + supports: SUPPORT + directness: DIRECT + quote_role: PRIMARY_RESULT + evidence_source: HUMAN_CLINICAL + snippet: "In the cohort of 819 tumors, ATRX mutations (n = 64) and MYCN amplification (n = 140) showed statistically significant mutual exclusivity (p = 0.0037, Cochrane-Mantel-Haenszel test) (Fig. 1e and Supplementary Note 1). Only 1 of 819 (0.1%) tumors had both lesions." + explanation: >- + In patient tumours pooled from three sequencing datasets (819 tumours), + ATRX mutation and MYCN amplification were significantly mutually exclusive + and only one tumour carried both, so the exclusivity is near-complete + rather than absolute. - name: PHOX2B association: Germline Predisposition gene_term: @@ -1065,9 +2135,8 @@ references: - reference: DOI:10.1038/s44318-024-00206-1 title: Targeting SWI/SNF ATPases reduces neuroblastoma cell plasticity findings: [] -- reference: DOI:10.1101/2022.08.16.504100 - title: Copy-number dosage regulates telomere maintenance and - disease-associated pathways in neuroblastoma +- reference: PMID:39635126 + title: "Copy-number dosage regulates telomere maintenance and disease-associated pathways in neuroblastoma." findings: [] - reference: DOI:10.3389/fimmu.2023.1268645 title: 'Two bullets in the gun: combining immunotherapy with chemotherapy to defeat diff --git a/references_cache/PMID_24792489.md b/references_cache/PMID_24792489.md new file mode 100644 index 00000000000..b2b5adbd36f --- /dev/null +++ b/references_cache/PMID_24792489.md @@ -0,0 +1,41 @@ +--- +reference_id: PMID:24792489 +extractor_version: 1 +title: Alternative lengthening of telomeres in neuroblastoma cell lines is associated with a lack of MYCN genomic amplification and with p53 pathway aberrations. +authors: +- Farooqi AS +- Dagg RA +- Choi LM +- Shay JW +- Reynolds CP +- Lau LM +journal: J Neurooncol +year: '2014' +doi: 10.1007/s11060-014-1456-8 +keywords: +- "Cell Line, Tumor" +- Gene Amplification +- Humans +- N-Myc Proto-Oncogene Protein +- "Nuclear Proteins/genetics, metabolism" +- "Oncogene Proteins/genetics, metabolism" +- Signal Transduction/genetics +- "Telomere/genetics, metabolism" +- Telomere Homeostasis +- "Tumor Suppressor Protein p53/genetics, metabolism" +publication_types: +- Journal Article +- "Research Support, N.I.H., Extramural" +- "Research Support, Non-U.S. Gov't" +content_type: abstract_only +full_text_attempted: true +--- + +# Alternative lengthening of telomeres in neuroblastoma cell lines is associated with a lack of MYCN genomic amplification and with p53 pathway aberrations. +**Authors:** Farooqi AS, Dagg RA, Choi LM, Shay JW, Reynolds CP, Lau LM +**Journal:** J Neurooncol (2014) +**DOI:** [10.1007/s11060-014-1456-8](https://doi.org/10.1007/s11060-014-1456-8) + +## Content + +Alternative lengthening of telomeres (ALT) is a telomerase-independent telomere length maintenance mechanism that enables the unlimited proliferation of a subset of cancer cells. Some neuroblastoma (NB) tumors appear to maintain telomere length by activating ALT. Of 40 NB cell lines, we identified four potential ALT cell lines (CHLA-90, SK-N-FI, LA-N-6, and COG-N-291) that were telomerase-negative and had long telomeres (a feature of ALT cells). All four cell lines lacked MYCN amplification and were p53 non-functional upon irradiation. Two of these cell lines (CHLA-90 and SK-N-FI) were positive for C-circles (telomeric DNA circles) and ALT-associated promyelocytic leukemia nuclear bodies, both of which are phenotypic characteristics of ALT. Mutation of ATRX (associated with ALT in tumors) was only found in CHLA-90. Thus, the ALT phenotype in NB may not be limited to tumors with ATRX mutations but is associated with a lack of MYCN amplification and alterations in the p53 pathway. \ No newline at end of file diff --git a/references_cache/PMID_25487495.md b/references_cache/PMID_25487495.md new file mode 100644 index 00000000000..a3929f82b54 --- /dev/null +++ b/references_cache/PMID_25487495.md @@ -0,0 +1,55 @@ +--- +reference_id: PMID:25487495 +extractor_version: 1 +title: Clinical features of ATRX or DAXX mutated neuroblastoma. +authors: +- Kurihara S +- Hiyama E +- Onitake Y +- Yamaoka E +- Hiyama K +journal: J Pediatr Surg +year: '2014' +doi: 10.1016/j.jpedsurg.2014.09.029 +keywords: +- "Adaptor Proteins, Signal Transducing/genetics" +- Adolescent +- Child +- "Child, Preschool" +- Co-Repressor Proteins +- DNA Helicases/genetics +- DNA Mutational Analysis +- Female +- Humans +- Infant +- "Infant, Newborn" +- Male +- Molecular Chaperones +- Mutation +- "Neoplasm Recurrence, Local/genetics" +- Neuroblastoma/genetics +- Nuclear Proteins/genetics +- Prognosis +- Telomere Homeostasis +- X-linked Nuclear Protein +publication_types: +- Journal Article +- "Research Support, Non-U.S. Gov't" +content_type: abstract_only +full_text_attempted: true +--- + +# Clinical features of ATRX or DAXX mutated neuroblastoma. +**Authors:** Kurihara S, Hiyama E, Onitake Y, Yamaoka E, Hiyama K +**Journal:** J Pediatr Surg (2014) +**DOI:** [10.1016/j.jpedsurg.2014.09.029](https://doi.org/10.1016/j.jpedsurg.2014.09.029) + +## Content + +PURPOSE: Previously, we reported that alternative lengthening of telomere (ALT) may be a biomarker for chemo-sensitivity and late recurrence in neuroblastoma (NBL). In this study, alterations of ATRX or DAXX, which both encode chromatin remodeling proteins in telomeric region, and their relationship to ALT were examined in NBLs. + +METHODS: Our previous report on 121 NBLs revealed 11 NBLs with elongated telomeres by ALT. In these NBLs, ATRX or DAXX gene alterations were identified using next-generation sequencing and compared to clinical and other biological factors. + +RESULTS: In 11 ALT cases, DAXX mutations were detected in one case, and ATRX alterations were detected in 10 cases. Except for one case, no DAXX or ATRX alterations were detected in 110 tumors with normal or shortened telomeres. MYCN amplification was not detected in ATRX altered tumors. In ALT cases, three infants showed ATRX deletions, and all seven cases detected after 18months of age showed poor prognosis. + +CONCLUSIONS: In NBLs, ALT was caused by ATRX or DAXX alterations. ATRX altered cases without MYCN amplification detected at greater than 18months showed poor prognosis, suggesting that ATRX or DAXX alterations are a particular NBL subtype. Since these tumors showed chemo-resistance and late recurrence, complete resection in a surgical approach should be performed to improve patient prognosis. \ No newline at end of file diff --git a/references_cache/PMID_26171145.md b/references_cache/PMID_26171145.md new file mode 100644 index 00000000000..226b4e89d3e --- /dev/null +++ b/references_cache/PMID_26171145.md @@ -0,0 +1,85 @@ +--- +reference_id: PMID:26171145 +extractor_version: 1 +html_full_text_version: 1 +title: Absence of telomerase reverse transcriptase promoter mutations in neuroblastoma. +authors: +- Lindner S +- Bachmann HS +- Odersky A +- Schaefers S +- Klein-Hitpass L +- Hero B +- Fischer M +- Eggert A +- Schramm A +- Schulte JH +journal: Biomed Rep +year: '2015' +doi: 10.3892/br.2015.463 +publication_types: +- Journal Article +content_type: full_text_html +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# Absence of telomerase reverse transcriptase promoter mutations in neuroblastoma. +**Authors:** Lindner S, Bachmann HS, Odersky A, Schaefers S, Klein-Hitpass L, Hero B, Fischer M, Eggert A, Schramm A, Schulte JH +**Journal:** Biomed Rep (2015) +**DOI:** [10.3892/br.2015.463](https://doi.org/10.3892/br.2015.463) + +## Content + +Maintenance of telomere length is a critical hallmark of malignant transformation. While silenced in somatic cells, telomerase reverse transcriptase (TERT), the catalytic subunit of telomerase, is frequently overexpressed in malignant cells thereby maintaining their telomere length. Specific point mutations in the TERT promoter region have recently been identified in melanoma and other tumor entities resulting in high TERT expression. Neuroblastoma is the most common extracranial tumor of childhood, arising from neural-crest progenitor cells. TERT overexpression has been observed in the majority of neuroblastoma. Taking into consideration that TERT promoter mutations are frequently described in neural-crest-derived tumors such as melanoma, as well as a variety of other neuronal tumors, the present study analyzed the frequency of TERT promoter mutations in primary neuroblastoma and neuroblastoma cell lines. In 131 neuroblastoma primary tumors representing the whole spectrum of neuroblastoma, no TERT promoter mutations were detected. However, in 3 out of 19 neuroblastoma cell lines the previously described C228T TERT promoter mutation was present. In conclusion, the TERT promoter mutations are not a frequent mechanism of TERT overexpression in neuroblastoma. + +Maintenance of telomere length is a critical hallmark of malignant transformation. While silenced in somatic cells, telomerase reverse transcriptase (TERT), the catalytic subunit of telomerase, is frequently overexpressed in malignant cells thereby maintaining their telomere length. Specific point mutations in the TERT promoter region have recently been identified in melanoma and other tumor entities resulting in high TERT expression. Neuroblastoma is the most common extracranial tumor of childhood, arising from neural-crest progenitor cells. TERT overexpression has been observed in the majority of neuroblastoma. Taking into consideration that TERT promoter mutations are frequently described in neural-crest-derived tumors such as melanoma, as well as a variety of other neuronal tumors, the present study analyzed the frequency of TERT promoter mutations in primary neuroblastoma and neuroblastoma cell lines. In 131 neuroblastoma primary tumors representing the whole spectrum of neuroblastoma, no TERT promoter mutations were detected. However, in 3 out of 19 neuroblastoma cell lines the previously described C228T TERT promoter mutation was present. In conclusion, the TERT promoter mutations are not a frequent mechanism of TERT overexpression in neuroblastoma. + +Keywords: neuroblastoma, telomerase reverse transcriptase, alternative lengthening of telomeres, telomeres, sequencing + +Each mitotic cell division results in the shortening of telomeres (1). In somatic cells, subsequent shortening of telomeres eventually leads to senescence. The multiunit enzyme telomerase maintains telomere length by adding repetitive sequences to telomeres at each replicative cycle. However, the telomerase activity is absent in somatic cells. By contrast, malignant cells (re-)express telomerase or use an alternative mechanism to maintain telomere length, known as alternative lengthening of telomeres (ALT) (2). + +Telomerase reverse transcriptase (TERT) is an essential part of the telomerase enzyme complex (3). Most recently, specific point mutations were described in the promoter region of the TERT gene in familial and sporadic melanoma (4). These point mutations lead to a cytosine to thymine exchange at the positions 228 or 250 upstream of the start codon of TERT, further referred to as C228T and C250T, respectively (3–5). Through these mutations, the TERT promoter acquires novel binding sites for the E twenty-six transcription factor, resulting in increased TERT expression (6). Subsequently, TERT promoter mutations were described in a variety of tumors, including several neuronal tumors as medulloblastoma (7) and glioma, as well as other tumor entities (4,5). Of note, the presence of TERT promoter mutations often correlated with aggressive disease and an adverse outcome. + +Neuroblastoma is the most common extracranial tumor during childhood arising from the neural-crest (8,9). Only few recurrent mutations or genomic rearrangements have been described thus far (10). Among these are the genomic amplifications of MYCN and activating mutations of anaplastic lymphoma kinase. Most recently, two independent studies described ALT through mutations in the transcriptional regulator α-thalassemia/mental retardation X-linked (ATRX) in a minor fraction of neuroblastomas (11,12). Although the ALT phenotype may not be limited to neuroblastomas with ATRX mutations, the majority of neuroblastomas express telomerase and in particular TERT (13). Of note, high TERT expression was detected in MYCN-amplified tumors, but even in MYCN single copy neuroblastomas a high TERT expression correlated with adverse outcome (13). By contrast, low or absent TERT expression was detected in favorable neuroblastomas and absence of telomerase activity has been proposed as a potential mechanism to explain the high rate of spontaneous regression of favorable neuroblastomas (14). While MYCN is known to induce TERT transcription (15,16) and TERT amplification has been described in few neuroblastoma cases (17), the mechanism that drives TERT expression remains elusive in the majority of neuroblastomas. + +As TERT promoter mutations are frequent in neural-crest-derived melanoma, as well as in various neuronal tumors including medulloblastoma, the present study analyzed the frequency of TERT promoter mutations in neuroblastoma. + +Primary tumor samples and patient data were obtained from the German Society of Pediatric Oncology and Hematology Tumor Bank and Neuroblastoma Study Center (Cologne, Germany), respectively. All the patients were registered with the German neuroblastoma study and written informed consent was obtained. A tumor content of ≥60% was confirmed by a pathologist. DNA was isolated from ~20 mg of snap-frozen tissue obtained prior to cytotoxic treatment using the Puregene Blood Core kit B (Qiagen, Hilden, Germany), according to the manufacturer's instructions. + +Neuroblastoma cell lines BEC(2), CHP134, IMR-32, IMR-5-75, KCN, KCN(R), Kelly, LAN1, LAN5, LAN6, NGP, NLF, SH-EP, SH-SY5Y, SK-N-AS, SK-N-BE, SK-N-FI and SK-N-SH were cultured in RPMI-1640 supplemented with 10% fetal calf serum, 5% penicillin (100 U/ml) and streptomycin (100 µg/ml) until reaching a confluence of 70%. Cells were collected and DNA was isolated using the QIAamp DNA kit (Qiagen). All the cell lines were authenticated by short tandem repeat genotyping performed by DSMZ (Braunschweig, Germany) or IDEXX GmbH (Ludwigsburg, Germany). + +DNA (100 ng) was used for the PCR reaction. To each PCR reaction, TERT forward (ACGAACGTGGCCAGCGGCAG) and reverse primers (CTGGCGTCCCTGCACCCTGG) were added to amplify a 474-base pair (bp) long region of the TERT promoter (390-bp upstream and 80-bp downstream of the start codon). The PCR program used was as follows: 95°C for 180 sec, followed by 95°C for 30 sec, 62°C for 45 sec and 72°C for 60 sec repeated 35 times, and a final amplification step at 72°C for 600 sec. Sanger sequencing was performed by Sequence Laboratories Göttingen GmbH (Göttingen, Germany). + +Total RNA was isolated from cells using the RNeasyMini kit (Qiagen) and cDNA synthesis was performed using the SuperScript® RT kit (Invitrogen Life Technologies, Darmstadt, Germany). TERT expression was monitored using the QuantiTect Primer Assay™ (Qiagen). Expression values were normalized to the geometric mean of GAPDH (18). Data analysis and error propagation were performed using the qbasePLUS software version 1.5 (http://www.biogazelle.com). + +To determine the presence of the mutant C228T allele or the wild-type allele using pyrosequencing, the following primers were used to amplify a 151-bp amplicon: Biotinylated forward primer 5′-BIOTIN-CTTCACCTTCCAGCTCCGC-3′, and reverse primer 5′-CGCTGCCTGAAACTCGC-3′. The PCR products were analyzed by pyrosequencing using the primer 5′-GAGGGGCTGGGAGGGCCC-3′, on the PyroMark Q96 MD system according to the manufacturer's instructions (Qiagen) and as previously described (19). Results were analyzed using PyroMark MD 1.0 software (Biotage, Uppsala, Sweden). + +SNP array experiments were performed according to the standard protocol for Affymetrix CytoScanHD arrays (Affymetrix, Inc., Santa Clara, CA, USA). In brief, a 250-ng sample of genomic DNA was digested with NspI, ligated to adaptors, amplified by PCR, fragmented and biotin-labeled. The labeled samples were hybridized to Affymetrix CytoscanHD arrays, followed by washing, staining and scanning in the Affymetrix GeneChip Scanner 3000. Analysis was performed using the Affymetrix Chromosome Analysis Suite v2.1. + +To determine the frequency of TERT promoter mutations in neuroblastoma, including the previously described C228T and C250T mutations, the respective TERT promoter region of 131 primary neuroblastomas was sequenced. These tumors represented the whole spectrum of neuroblastoma regarding stage, MYCN and 1p36 status (Table I). Of note, none of the analyzed neuroblastomas harbored a C228T or a C250T mutation in the TERT promoter region. + +Patient data of the 131 sequenced primary neuroblastomas tumors. + +INSS, International Neuroblastoma Staging System. + +In addition to primary neuroblastomas, 19 neuroblastoma cell lines were analyzed. In 3 out of 19 analyzed cell lines (16%) a mutation of the TERT promoter was detected. The neuroblastoma cell lines SK-N-SH, SH-SY5Y and SH-EP harbor the C228T mutation (Fig. 1A–C). Of note, SH-SY5Y and SH-EP are subclones of SK-N-SH (20). SK-N-SH was established from a bone marrow biopsy of a 4-year-old girl (21,22). The presence of the C228T TERT promoter mutation in SK-N-SH, SH-SY5Y and SH-EP was independently validated using pyrosequencing (Fig. 1B). As the C228T mutation appeared to be homozygous in SK-N-SH, SH-SY5Y and SH-EP, Affymetrix Human SNP array 6.0 arrays were used to analyze zygosity. The 3 probes located within the TERT gene were homozygeous in all three cell lines, which could point to the presence of a small loss-of-heterozygosity region spanning the TERT gene (Table II). To analyze if the mutations in the TERT promoter correlate with increased TERT expression, the expression was analyzed in all the neuroblastoma cell lines using RT-qPCR. No correlation between mutations and TERT expression was detected (Fig. 1D). + +(A) Electropherogram examples of the cell lines SH-SY5Y, SH-EP and SK-N-SH harboring the C228T nucleotide exchange. By contrast, LAN5 shows the wild-type (WT) sequence at the mutation hotspots. (B) Representative pyrograms obtained for WT LAN5 cells, as well as SK-N-SH, SH-EP and SH-SY5Y cells, harboring the C228T telomerase reverse transcriptase (TERT) promoter mutation. Pyrograms were generated by sequential addition of the single-nucleotide polymorphism-specific nucleotides shown beneath each trace. Note that each peak corresponds to one or more identical bases added to the synthesized DNA-strand and that the height of each peak is correlated to the number of sequential nucleotides of the same type. (C) List of the neuroblastoma cell lines sequenced for TERT promoter mutations (C228T=guanine to adenine nucleotide exchange upstream of the TERT promoter at position-228). (D) TERT expression scaled to the lowest expression found in SK-N-SH (TERT expression=1). The cell lines that carry a TERT promoter mutation are marked with ‘mutated’. + +Location of the Affymetrix CytoScanHD array probes located within or near the TERT gene. The probes lying within the TERT gene showed a loss of heterozygousity for SH-EP, SH-SY5Y and SK-N-SH. Probes lying outside of the TERT gene showed a heterozygous status. + +Chromosomal position of TERT: Chr.5.1253167–1295047; Chromosomal position of the C228T mutation: Chr.5.1295228. TERT, telomerase reverse transcriptase. + +Analysis of the TERT gene promotor region in 20 neuroblastoma cell lines and 131 primary neuroblastoma tumors revealed mutations in 3 cell lines and no mutations in primary tumors. + +Of note, the only 3 mutated cell lines, which harbor the C228T mutation, are SK-N-SH and its 2 subclones SH-EP and SH-SY5Y, further limiting the general indication of this finding. It is uncommon that TERT mutations occur homozygously, as detected in SK-N-SH, SH-EP and SY5Y. However, further analysis confirmed a loss-of-heterozygosity region in the region of the TERT gene, pointing to a hemizygous rather than a homozygous status of the C228T mutation in these cases. Taking into consideration the variable levels of TERT expression in SK-N-SH, SH-EP and SH-SY5Y cells, the functional indication of the C228T mutations in these cell lines remain questionable. + +The absence of TERT mutations in primary neuroblastomas and the majority of analyzed cell lines is unexpected, as TERT expression is of pathogenic relevance in the majority of neuroblastomas (13) and TERT mutations are frequent in related neuroectodermal tumors, such as melanoma and medulloblastoma (4,7,23,24). However, the present results are in line with previous studies by Papathomas et al (25), who did not find TERT promoter mutations in 15 primary neuroblastomas and Killela et al (23), who identified only 2 TERT promoter mutations in 22 neuroblastomas. + +The mechanisms by which TERT expression is induced, particularly in MYCN non-amplified high-risk neuroblastomas, remains elusive and warrants further studies. Potential alternative mechanisms underlying TERT induction in neuroblastoma include epigenetic changes, as well as mutations in more distant regulatory elements that were not analyzed in the present study. + +Taken together, we conclude that TERT (core) promoter mutations are not relevant events in neuroblastoma pathogenesis. + +The authors acknowledge the funding from the German Ministry for Education and Research (e:MED grant nos. SMOOSE FKZ: 01ZX1303B and SYSMED-NB FKZ: 01ZX1307E to J.H.S.) and the German Cancer Aid (grant no. 111301 to J.H.S.). \ No newline at end of file diff --git a/references_cache/PMID_27793328.md b/references_cache/PMID_27793328.md new file mode 100644 index 00000000000..d244361e31d --- /dev/null +++ b/references_cache/PMID_27793328.md @@ -0,0 +1,52 @@ +--- +reference_id: PMID:27793328 +extractor_version: 1 +title: "Telomere biology including TERT rearrangements in neuroblastoma: a useful indicator for surgical treatments." +authors: +- Kawashima M +- Kojima M +- Ueda Y +- Kurihara S +- Hiyama E +journal: J Pediatr Surg +year: '2016' +doi: 10.1016/j.jpedsurg.2016.09.042 +keywords: +- Child +- "Child, Preschool" +- DNA Helicases/genetics +- DNA Mutational Analysis +- "DNA, Neoplasm/genetics" +- Disease Progression +- Female +- Humans +- Infant +- "Infant, Newborn" +- Male +- Mutation +- "Neuroblastoma/genetics, pathology, surgery" +- Prognosis +- "Telomerase/genetics, metabolism" +- Telomere/physiology +publication_types: +- Journal Article +content_type: abstract_only +full_text_attempted: true +--- + +# Telomere biology including TERT rearrangements in neuroblastoma: a useful indicator for surgical treatments. +**Authors:** Kawashima M, Kojima M, Ueda Y, Kurihara S, Hiyama E +**Journal:** J Pediatr Surg (2016) +**DOI:** [10.1016/j.jpedsurg.2016.09.042](https://doi.org/10.1016/j.jpedsurg.2016.09.042) + +## Content + +PURPOSE: Our telomere biology study of neuroblastomas (NBLs) has revealed that unfavorable NBLs acquired telomere stabilization by telomerase activation or ALT (alternative lengthening of telomeres). Recently, genomic rearrangements in a region proximal to the telomerase reverse transcriptase (TERT) gene have been discovered in NBLs. In this study, TERT rearrangements were examined in NBLs along with their relationship to other aspects of telomere biology. + +METHODS: In 121 NBLs, including 67 cases detected by mass-screening whose telomere length, telomerase activity, ALT with ATRX/DAXX alterations, and MYCN amplification were already known, TERT rearrangements were examined using GeneChip SNP arrays. + +RESULTS: The 11 ATRX/DAXX mutated ALT cases and 29 cases with high telomerase activity showed poor prognosis. MYCN amplification and TERT rearrangements were independently detected in 16 and 13 cases, respectively, and these alterations were significantly correlated with high telomerase activity. In 81 infant cases, MYCN amplification, TERT rearrangements and ATRX mutations were detected in 3, 4, and 3 cases, respectively. Among them, 6 cases showed progression or recurrences. + +CONCLUSIONS: Telomere stabilization in NBLs is acquired by telomerase activation through MYCN amplification, TERT rearrangements or by ALT. Since these tumors usually show progression and recurrence, complete resection should be considered, even in infant cases. + +LEVEL OF EVIDENCE: Prognosis study, level III. \ No newline at end of file diff --git a/references_cache/PMID_28636942.md b/references_cache/PMID_28636942.md new file mode 100644 index 00000000000..d1a9d674577 --- /dev/null +++ b/references_cache/PMID_28636942.md @@ -0,0 +1,46 @@ +--- +reference_id: PMID:28636942 +extractor_version: 1 +title: Extensive Proliferation of Human Cancer Cells with Ever-Shorter Telomeres. +authors: +- Dagg RA +- Pickett HA +- Neumann AA +- Napier CE +- Henson JD +- Teber ET +- Arthur JW +- Reynolds CP +- Murray J +- Haber M +- Sobinoff AP +- Lau LMS +- Reddel RR +journal: Cell Rep +year: '2017' +doi: 10.1016/j.celrep.2017.05.087 +keywords: +- "Cell Line, Tumor" +- Cell Proliferation +- Enzyme Activation +- Gene Amplification +- Humans +- N-Myc Proto-Oncogene Protein/genetics +- "Neuroblastoma/genetics, pathology" +- Telomerase/metabolism +- Telomere Shortening +publication_types: +- Journal Article +- "Research Support, Non-U.S. Gov't" +content_type: abstract_only +full_text_attempted: true +--- + +# Extensive Proliferation of Human Cancer Cells with Ever-Shorter Telomeres. +**Authors:** Dagg RA, Pickett HA, Neumann AA, Napier CE, Henson JD, Teber ET, Arthur JW, Reynolds CP, Murray J, Haber M, Sobinoff AP, Lau LMS, Reddel RR +**Journal:** Cell Rep (2017) +**DOI:** [10.1016/j.celrep.2017.05.087](https://doi.org/10.1016/j.celrep.2017.05.087) + +## Content + +Acquisition of replicative immortality is currently regarded as essential for malignant transformation. This is achieved by activating a telomere lengthening mechanism (TLM), either telomerase or alternative lengthening of telomeres, to counter normal telomere attrition. However, a substantial proportion of some cancer types, including glioblastomas, liposarcomas, retinoblastomas, and osteosarcomas, are reportedly TLM-negative. As serial samples of human tumors cannot usually be obtained to monitor telomere length changes, it has previously been impossible to determine whether tumors are truly TLM-deficient, there is a previously unrecognized TLM, or the assay results are false-negative. Here, we show that a subset of high-risk neuroblastomas (with ∼50% 5-year mortality) lacked significant TLM activity. Cancer cells derived from these highly aggressive tumors initially had long telomeres and proliferated for >200 population doublings with ever-shorter telomeres. This indicates that prevention of telomere shortening is not always required for oncogenesis, which has implications for inhibiting TLMs for cancer therapy. \ No newline at end of file diff --git a/references_cache/PMID_30523111.md b/references_cache/PMID_30523111.md new file mode 100644 index 00000000000..e1a979bbe92 --- /dev/null +++ b/references_cache/PMID_30523111.md @@ -0,0 +1,109 @@ +--- +reference_id: PMID:30523111 +extractor_version: 1 +xml_extraction_version: 1 +title: A mechanistic classification of clinical phenotypes in neuroblastoma. +authors: +- Ackermann S +- Cartolano M +- Hero B +- Welte A +- Kahlert Y +- Roderwieser A +- Bartenhagen C +- Walter E +- Gecht J +- Kerschke L +- Volland R +- Menon R +- Heuckmann JM +- Gartlgruber M +- Hartlieb S +- Henrich KO +- Okonechnikov K +- Altmüller J +- Nürnberg P +- Lefever S +- de Wilde B +- Sand F +- Ikram F +- Rosswog C +- Fischer J +- Theissen J +- Hertwig F +- Singhi AD +- Simon T +- Vogel W +- Perner S +- Krug B +- Schmidt M +- Rahmann S +- Achter V +- Lang U +- Vokuhl C +- Ortmann M +- Büttner R +- Eggert A +- Speleman F +- "O'Sullivan RJ" +- Thomas RK +- Berthold F +- Vandesompele J +- Schramm A +- Westermann F +- Schulte JH +- Peifer M +- Fischer M +journal: Science +year: '2018' +doi: 10.1126/science.aat6768 +keywords: +- Child +- "Child, Preschool" +- Disease-Free Survival +- Exome/genetics +- "Genome, Human" +- Humans +- Metabolic Networks and Pathways/genetics +- Mutation +- "Neuroblastoma/classification, drug therapy, genetics, mortality" +- Prognosis +- "Sequence Analysis, DNA" +- Telomere Homeostasis/genetics +- Tumor Suppressor Protein p53/genetics +- ras Proteins/genetics +publication_types: +- Journal Article +- "Research Support, Non-U.S. Gov't" +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# A mechanistic classification of clinical phenotypes in neuroblastoma. +**Authors:** Ackermann S, Cartolano M, Hero B, Welte A, Kahlert Y, Roderwieser A, Bartenhagen C, Walter E, Gecht J, Kerschke L, Volland R, Menon R, Heuckmann JM, Gartlgruber M, Hartlieb S, Henrich KO, Okonechnikov K, Altmüller J, Nürnberg P, Lefever S, de Wilde B, Sand F, Ikram F, Rosswog C, Fischer J, Theissen J, Hertwig F, Singhi AD, Simon T, Vogel W, Perner S, Krug B, Schmidt M, Rahmann S, Achter V, Lang U, Vokuhl C, Ortmann M, Büttner R, Eggert A, Speleman F, O'Sullivan RJ, Thomas RK, Berthold F, Vandesompele J, Schramm A, Westermann F, Schulte JH, Peifer M, Fischer M +**Journal:** Science (2018) +**DOI:** [10.1126/science.aat6768](https://doi.org/10.1126/science.aat6768) + +## Content + +Neuroblastoma is a pediatric tumor of the sympathetic nervous system. Its clinical course ranges from spontaneous tumor regression to fatal progression. To investigate the molecular features of the divergent tumor subtypes, we performed genome sequencing on 416 pretreatment neuroblastomas and assessed telomere maintenance mechanisms in 208 of these tumors. We found that patients whose tumors lacked telomere maintenance mechanisms had an excellent prognosis, whereas the prognosis of patients whose tumors harbored telomere maintenance mechanisms was substantially worse. Survival rates were lowest for neuroblastoma patients whose tumors harbored telomere maintenance mechanisms in combination with RAS and/or p53 pathway mutations. Spontaneous tumor regression occurred both in the presence and absence of these mutations in patients with telomere maintenance-negative tumors. On the basis of these data, we propose a mechanistic classification of neuroblastoma that may benefit the clinical management of patients. + +Neuroblastoma is a pediatric tumor of the sympathetic nervous system with substantially varying clinical courses (1). Roughly half of neuroblastoma patients have a dismal outcome despite intensive multimodal treatment, whereas other patients have an excellent outcome because their tumors either spontaneously regress or differentiate into benign ganglioneuromas. Patients are considered to be at high risk of death if they are diagnosed with metastatic disease when they are older than 18 months or when their tumor exhibits genomic amplification of the proto-oncogene MYCN (2). All other patients are classified as intermediate or low risk (referred to as non–high-risk patients in this study) and receive limited or no cytotoxic treatment. In addition to MYCN amplification, rearrangements of the TERT locus (encoding the catalytic subunit of telomerase) or inactivating mutations in ATRX (encoding a chromatin remodeling protein) have been found predominantly in high-risk tumors (3–7). Whereas both MYCN and TERT alterations lead to telomere maintenance by induction of telomerase, ATRX loss-of-function mutations have been associated with activation of the alternative lengthening of telomeres (ALT) pathway (5, 8). Neuroblastomas also harbor recurrent mutations in ALK (encoding a receptor tyrosine kinase) (9, 10). To date, these genomic data have not produced a coherent model of pathogenesis that can explain the extremely divergent clinical phenotypes of neuroblastoma. + +In the study, we aimed to evaluate whether the divergent clinical phenotypes in neuroblastoma are defined by specific genetic alterations. To this end, we examined 218 pretreatment tumors and matched normal control tissue (i.e., blood) by whole-exome or whole-genome sequencing (WES and WGS, respectively) from patients covering the entire spectrum of the disease (fig. S1 and tables S1 to S3). In line with previous studies, we found 14.9 somatic singlenucleotide variations (SNVs) per tumor exome on average (median, 12 SNVs per tumor exome; fig. S2) (4, 6). Because mutations in genes of the RAS and p53 pathways have been detected in relapsed neuroblastoma (11–13), we hypothesized that such alterations may not only be relevant at the time of relapse but may also determine the clinical course of neuroblastoma at diagnosis. We thus defined a panel of 17 genes related to the RAS pathway (11 genes including ALK) or the p53 pathway (6 genes) based on our own and published data (fig. S3 and tables S4 to S7) and examined their mutation frequency in pretreatment tumors (Fig. 1A). Focal amplifications, homozygous deletions, and variants of amino acids recorded in the Catalogue of Somatic Mutations in Cancer (14) were considered. We found alterations of these genes in 46 of 218 cases of the combined WES and WGS cohort. In an independent cohort of 198 pretreatment tumors examined by targeted sequencing (fig. S1 and tables S1 and S2), we detected alterations of these genes in 28 of 198 cases, resulting in an overall mutation frequency of 17.8% in the combined cohorts (74 of 416 cases; fig. S4 and tables S8 and S9). RAS and p53 pathway mutations were enriched in overall clonal cancer cell populations (95% versus 71% clonal events, P = 0.021; fig. S5), indicating their evolutionary selection during tumor development. + +Mutations in RAS and p53 pathway genes occurred in both high- and non–high-risk tumors, although at lower frequencies in the latter group (21.3% versus 13.3%, P = 0.048, fig. S6). Overall, the presence of such alterations was strongly associated with poor patient outcome (Fig. 1B and fig. S7). We did not observe significant differences between the prognostic effects of RAS and p53 pathway alterations; however, patients whose tumors had ALK mutations had better event-free survival than those whose tumors harbored other RAS pathway mutations (fig. S8). In high-risk patients, alterations of RAS or p53 pathway genes were also associated with poor outcome (Fig. 1C and fig. S9, A and B), both in MYCNamplified and non–MYCN-amplified cases (fig. S9C). Such alterations also identified patients with unfavorable clinical courses in the non–high-risk cohort (Fig. 1D and fig. S9D). The presence of these mutations predicted dismal outcome in multivariable analyses independently of prognostic markers currently used for neuroblastoma risk stratification (15) in the entire cohort and in both high-risk and non–high-risk patients (fig. S10). Together, our findings point to a crucial role of RAS and p53 pathway genes in the development of unfavorable neuroblastoma, which is in line with increased frequencies of such mutations at clinical relapse (11–13) and with data from genetically engineered mouse models showing that RAS pathway activation augments neuroblastoma aggressiveness (16–18). + +Despite the overall association of RAS and p53 pathway mutations with poor outcome, however, we noticed that the clinical courses of non–high-risk patients bearing such mutations varied greatly, ranging from spontaneous regression to fatal tumor progression (fig. S11). On the basis of previous work (5, 19–21), we hypothesized that these differences may be related to the presence or absence of telomere maintenance mechanisms. We therefore examined the genomic status of the MYCN and TERT loci, as well as ALT-associated promyelocytic leukemia nuclear bodies (APBs) and TERT expression in a cohort of 208 of 416 tumors (fig. S1). We observed MYCN amplification in 52 cases, TERT rearrangements in 21 cases, and APBs in 31 cases (Fig. 2A and table S10). In line with previous observations (5, 22), TERT expression was elevated in tumors bearing TERT rearrangements or MYCN amplification, indicating telomerase activation (fig. S12, A and B). In APB-positive tumors, TERT expression was low and telomere length ratios high, thus supporting an ALT phenotype (fig. S12, A, C, and D). We also assessed the genomic status of ATRX in 83 evaluable tumors and found mutations that were likely to be inactivating in eight of these, all of which were ALT positive (Fig. 2A and table S10); by contrast, mutations of DAXX, which encodes another protein participating in chromatin remodeling at telomeres, were not detected. We observed neither significant alterations in ATRX or DAXX gene methylation status or gene expression patterns in ALT-positive tumors (fig. S13) nor significant associations between ALT and p53 pathway mutations (3 of 31 ALT-positive cases mutated, 7 of 177 ALT-negative cases mutated; P = 0.173) (23). Immunohistochemical staining revealed loss of nuclear ATRX expression in one tumor bearing an ATRX nonsense mutation, whereas expression was retained in tumors with ATRX in-frame deletions (fig. S14) (23). Furthermore, we noticed that a small fraction of neuroblastomas lacking MYCN or TERT alterations had elevated TERT mRNA levels (fig. S12A). We therefore determined and validated a TERT expression threshold to identify wildtype MYCN and TERT (MYCNWT and TERTWT) tumors whose TERT mRNA levels are comparable to those of tumors bearing genomic MYCN or TERT alterations, pointing toward telomerase activation (fig. S15). In fact, high TERT mRNA levels corresponded to elevated enzymatic telomerase activity in these tumors, as well as in tumors harboring MYCN amplification or TERT rearrangements (Fig. 2B). On the basis of these and our previous observations (5), we considered tumors telomere maintenance positive if they harbored TERT rearrangements or MYCN amplification, elevated TERT expression in the absence of these alterations, or were positive for APBs as a marker of ALT (Fig. 2A). + +In the set of non–high-risk tumors bearing RAS or p53 pathway mutations (23 of 208 cases), we found evidence for telomerase or ALT activation in nine cases (fig. S16A). The outcome of these patients was poor, whereas all patients whose tumors lacked telomere maintenance mechanisms have survived to date, with no or limited cytotoxic therapy (fig. S16B). Importantly, this finding was validated in an additional series of 20 pretreatment non–high-risk neuroblastomas with RAS pathway gene mutations that had not been part of the initial WES and WGS or targeted sequencing cohorts (figs. S1 and S16, C and D, and table S11). Together, telomere maintenance mechanisms thus clearly discriminated the divergent clinical phenotypes occurring in non–high-risk tumors bearing RAS or p53 pathway mutations (Fig. 3, A and B). + +The prognostic dependence of RAS pathway mutations on telomere maintenance in non–high-risk disease was highlighted in a patient subgroup that was genetically defined by the presence of ALKR1275Q (R1275Q, Arg1275→Gln) mutations (n = 11 patients): Outcome was excellent only if telomere maintenance mechanisms were absent, and spontaneous regression had been documented in four of these children (Fig. 3C and fig. S16E). Similarly, complete regression of osteomedullary metastases without any chemotherapy had been noticed in a stage 4 patient whose tumor carried the particularly aggressive ALKF1174L (F1174L, Phe1174→Leu) mutation (Fig. 3D) (16). In two other patients with ALK-mutant tumors (NBL8 and NBL-V16), spontaneous differentiation into ganglioneuroblastoma had been found after partial regression in patient NBL8 (Fig. 3E and fig. S16F). Finally, long-term event-free survival without chemotherapy was also recorded in patient NBL59, whose tumor harbored both HRAS and TP53 mutations in the absence of telomere maintenance, whereas patients whose tumors harbored HRAS, NRAS, or TP53 mutations had fatal outcome when telomerase or ALT was activated (Fig. 3A). + +We hypothesized that a general pathogenetic hierarchy of telomere maintenance and RAS or p53 pathway mutations might mechanistically define the different clinical subgroups of neuroblastoma. Indeed, we observed that the outcome of patients whose tumors lacked telomere maintenance (n = 99) was excellent, irrespective of the presence of RAS or p53 pathway mutations (Fig. 4A). Fifty-seven of these patients had never received cytotoxic treatment, including 18 cases with documented spontaneous regression (Fig. 2A and table S10). Our data indicate that RAS or p53 pathway mutations are not sufficient for full malignant transformation and continuous growth of human neuroblastoma in the absence of telomere maintenance. Consistent with this observation, telomerase has been shown to be essential for full malignant transformation of human cells bearing oncogenic HRAS in experimental systems (24), whereas cellular senescence occurs in response to oncogenic HRAS in the absence of telomerase (25). Neuroblastomas lacking telomere maintenance were mainly derived from young patients (mean age at diagnosis, 378 days; fig. S17A) classified as clinical low or intermediate risk(96 of 99 cases; P < 0.001); the remaining three tumors had been obtained from young stage 4 patients (age at diagnosis, 732 to 1035 days) who all have survived event-free to date. By contrast, children whose tumors harbored telomerase or ALT activation were mainly clinical high-risk patients (92 of 109 cases; P < 0.001). Seventeen patients had been clinically classified as low or intermediate risk; however, their clinical course was as unfavorable as that of high-risk patients (fig. S17B), thus supporting the notion that telomere maintenance is a major determinant of neuroblastoma outcome. We also found no significant difference in the outcome of patients whose tumors displayed MYCN amplification compared with those whose tumors had other telomere maintenance mechanisms (fig. S18A). In addition, we observed that the outcome of patients whose tumors exhibited telomere maintenance was devastating when additional RAS or p53 pathway mutations were present, whereas survival was considerably better in their absence (Fig. 4A). Among the former patients, those whose tumors harbored ALK mutations tended to have a more favorable outcome than those whose tumors carried other RAS pathway mutations (fig. S18B). We also observed that the telomere maintenance status did not change over the disease course in 19 of 20 paired neuroblastoma samples biopsied at diagnosis and relapse or progression; in one case, de novo MYCN amplification accompanied by TERT up-regulation occurred at the time of relapse (table S12). This finding suggests that the telomere maintenance status is mostly fixed at diagnosis, which is in line with the notion that low-risk neuroblastoma rarely develops into high-risk disease (26, 27). The clinical relevance of telomere maintenance and RAS or p53 pathway alterations was substantiated by multivariable analysis, in which both alterations independently predicted unfavorable outcome (Fig. 4B). Additional backward selection of variables in this model identified only telomere maintenance and RAS or p53 pathway mutations as independent prognostic markers (telomere maintenance: hazard ratio, 5.184, confidence interval, 2.723 to 9.871, P < 0.001; RAS and/or p53 pathway mutation: hazard ratio, 2.056, confidence interval, 1.325 to 3.190, P = 0.001), whereas the established markers (stage, age, and chromosome 1p status) were not considered in the final model. + +Together, our findings demonstrate that the divergent clinical phenotypes of human neuroblastoma are driven by molecular alterations affecting telomere maintenance and RAS or p53 pathways, suggesting a mechanistic classification of this malignancy (Fig. 4C): High-risk neuroblastoma is defined by telomere maintenance caused by induction of telomerase or the ALT pathway. Additional mutations in genes of the RAS or p53 pathway increase tumor aggressiveness, resulting in a high likelihood of death from disease. By contrast, low-risk tumors invariably lack telomere maintenance mechanisms. Because telomere maintenance is essential for cancer cells to achieve immortal proliferation capacity (8, 28), its absence is likely a prerequisite for spontaneous regression and differentiation in neuroblastoma. Our data also indicate that mutations of RAS or p53 pathway genes in tumors without telomere maintenance do not affect patient outcome. + +Our findings may have important implications for the diagnosis and treatment of neuroblastoma patients, which should be validated in future prospective clinical trials. Assessment of telomere maintenance mechanisms and a limited set of RAS and p53 pathway genes may be sufficient to accurately estimate patient risk at diagnosis and to guide treatment stratification. In a clinical setting, telomerase activation may be readily determined by examining the genomic status of MYCN and TERT in the majority of cases and supplemented by analysis of TERT expression levels in MYCNWT and TERTWT tumors. It is important to note, though, that classification of patients based on a TERT expression threshold may bear a certain risk of misclassification because of potential confounding factors, such as tumor cell content or RNA integrity of the sample. In addition to analysis of telomerase activation, ALT can be assessed by detection of APBs or, potentially, by polymerase chain reaction (PCR) amplification of extrachromosomal circles of telomeric DNA (29). We propose that patients whose tumors lack telomere maintenance may require limited or no cytotoxic treatment, as suggested by the high prevalence of spontaneous regression in these cases, whereas patients whose tumors harbor such mechanisms need intensive therapy. Patients whose tumors carry both telomere maintenance and RAS or p53 pathway alterations, however, are at high risk of treatment failure and death (Fig. 4A). Nonetheless, the fact that these alterations can act in concert provides a rationale for developing novel combination therapies. Compounds interfering with aberrant RAS pathway signaling have shown promising antitumor effects in preclinical models of neuroblastoma (12, 16, 30, 31), and ALK inhibitors have entered clinical trials (32). In addition, therapeutic strategies directed against telomerase or the ALT pathway are the subject of current investigations (28, 33, 34). A combination of therapies targeting these two critical oncogenic pathways in neuroblastoma may thus merit investigation. \ No newline at end of file diff --git a/references_cache/PMID_32060267.md b/references_cache/PMID_32060267.md new file mode 100644 index 00000000000..47e16be6883 --- /dev/null +++ b/references_cache/PMID_32060267.md @@ -0,0 +1,480 @@ +--- +reference_id: PMID:32060267 +extractor_version: 1 +xml_extraction_version: 1 +title: MYCN amplification and ATRX mutations are incompatible in neuroblastoma. +authors: +- Zeineldin M +- Federico S +- Chen X +- Fan Y +- Xu B +- Stewart E +- Zhou X +- Jeon J +- Griffiths L +- Nguyen R +- Norrie J +- Easton J +- Mulder H +- Yergeau D +- Liu Y +- Wu J +- Van Ryn C +- Naranjo A +- Hogarty MD +- Kamiński MM +- Valentine M +- Pruett-Miller SM +- Pappo A +- Zhang J +- Clay MR +- Bahrami A +- Vogel P +- Lee S +- Shelat A +- Sarthy JF +- Meers MP +- George RE +- Mardis ER +- Wilson RK +- Henikoff S +- Downing JR +- Dyer MA +journal: Nat Commun +year: '2020' +doi: 10.1038/s41467-020-14682-6 +keywords: +- Animals +- "Child, Preschool" +- Cohort Studies +- Female +- Gene Amplification +- Humans +- Infant +- Male +- Mice +- "Mitochondria/genetics, metabolism" +- Mutation +- "N-Myc Proto-Oncogene Protein/genetics, metabolism" +- "Neuroblastoma/genetics, metabolism" +- Reactive Oxygen Species/metabolism +- "X-linked Nuclear Protein/genetics, metabolism" +publication_types: +- Journal Article +- "Research Support, N.I.H., Extramural" +- "Research Support, Non-U.S. Gov't" +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# MYCN amplification and ATRX mutations are incompatible in neuroblastoma. +**Authors:** Zeineldin M, Federico S, Chen X, Fan Y, Xu B, Stewart E, Zhou X, Jeon J, Griffiths L, Nguyen R, Norrie J, Easton J, Mulder H, Yergeau D, Liu Y, Wu J, Van Ryn C, Naranjo A, Hogarty MD, Kamiński MM, Valentine M, Pruett-Miller SM, Pappo A, Zhang J, Clay MR, Bahrami A, Vogel P, Lee S, Shelat A, Sarthy JF, Meers MP, George RE, Mardis ER, Wilson RK, Henikoff S, Downing JR, Dyer MA +**Journal:** Nat Commun (2020) +**DOI:** [10.1038/s41467-020-14682-6](https://doi.org/10.1038/s41467-020-14682-6) + +## Content + +Aggressive cancers often have activating mutations in growth-controlling oncogenes and inactivating mutations in tumor-suppressor genes. In neuroblastoma, amplification of the MYCN oncogene and inactivation of the ATRX tumor-suppressor gene correlate with high-risk disease and poor prognosis. Here we show that ATRX mutations and MYCN amplification are mutually exclusive across all ages and stages in neuroblastoma. Using human cell lines and mouse models, we found that elevated MYCN expression and ATRX mutations are incompatible. Elevated MYCN levels promote metabolic reprogramming, mitochondrial dysfunction, reactive-oxygen species generation, and DNA-replicative stress. The combination of replicative stress caused by defects in the ATRX-histone chaperone complex, and that induced by MYCN-mediated metabolic reprogramming, leads to synthetic lethality. Therefore, ATRX and MYCN represent an unusual example, where inactivation of a tumor-suppressor gene and activation of an oncogene are incompatible. This synthetic lethality may eventually be exploited to improve outcomes for patients with high-risk neuroblastoma. + +Neuroblastoma is the most common extracranial solid tumor of childhood1. MYCN amplification and age at diagnosis are the two most powerful predictors of outcome, with survival rates 5–10 times higher in infants than in adolescents or young adults1,2. Previous genomic analyses of stage 4 pediatric neuroblastoma samples identified the ATRX mutations in patients that were typically older than 5 y, had an indolent disease course, and poor overall survival (OS)1,3. + +One important function of ATRX is recognition of guanine (G)-rich stretches of DNA and deposition of the H3.3 histone variant to prevent the formation of G-quadruplex (G4) structures, which can block DNA replication or transcription4,5. G-rich repeats are also found at telomeres and centromeres; ATRX forms a complex with DAXX to deposit H3.3 in those regions to maintain their integrity4,5. In cells lacking ATRX, H3.3 is not efficiently deposited at the telomeric G-rich regions, G4 structures form, and replication forks stall4,5. Consequently, telomeres undergo homologous recombination leading to alternative lengthening of telomeres (ALT)6. + +The formation of G4 structures in other G-rich repetitive regions of the genome can cause replicative stress7,8 or block transcription9. Indeed, H3.3 is deposited at actively transcribed genes in addition to telomeres and pericentromeric DNA9. ATRX may also affect transcription by targeting the PRC2 complex to particular regions of the genome10. Consequently, in ATRX-deficient cells, PRC2-mediated modification of H3 to H3K27me3 lacks specificity, and genes that are normally repressed by polycomb are deregulated10. + +MYCN regulates diverse cellular processes during development and in cancer. For example, elevated MYCN leads to increased glycolytic flux and glutaminolysis to promote metabolic reprogramming associated with tumorigenesis in a variety of cancers including neuroblastomas11,12. MYCN-induced glutaminolysis in neuroblastoma elevates reactive-oxygen species (ROS) and DNA-replicative stress13,14. Indeed, one of the hallmarks of neuroblastoma is the DNA mutation signature associated with ROS induced DNA damage. Consequently, neuroblastoma cells exhibit increased sensitivity to pharmacological agents that induce oxidative stress13,14. Here we demonstrate that the DNA-replicative stress induced by ATRX mutations and MYCN amplification cause synthetic lethality in neuroblastoma. This is unusual because oncogene activation and tumor-suppressor inactivation often work in concert to promote tumorigenesis not cancer cell death. + +To complement previous neuroblastoma studies from the Therapeutically Applicable Research to Generate Effective Treatment (TARGET) initiative15 and the Pediatric Cancer Genome Project (PCGP)3,16, we obtained neuroblastoma samples from 473 patients (122 unpaired and 351 paired tumor/germline) from the Children’s Oncology Group (COG) (Table 1). We identified single-nucleotide variations, small indels, and other somatic mutations in the coding region of ATRX via custom capture and Illumina sequencing using probes spanning the entire ATRX locus and whole-exome sequencing of 828 germline and tumor samples. We also included MYCN in the capture probe set to determine its copy number. We identified 19 somatic ATRX mutations, 15 of those are internal deletions predicted to encode truncated proteins (Fig. 1a).Table 1Distribution of ATRX mutations in neuroblastoma.ClassificationaGroupingN(%)bATRX mutation (%)cp-valueINSS stageNon-stage 4312 (65.8)5 (1.6)0.0002Stage 4162 (34.2)14 (8.6)Risk GroupLow/intermediate375 (79.6)5 (1.3)4.4e-7High96 (20.4)14 (14.6)SexFemale253 (53.4)12 (4.7)0.3830Male221 (46.6)7 (3.2)MYCN (FISH)Not amplified431 (91.7)18 (4.2)1.0000Amplified39 (8.3)1 (2.6)MYCN (NGS)Not amplified447 (94.1)19 (4.3)0.6180Amplified28 (5.9)0 (0)ALKNo mutation454 (95.6)17 (3.7)0.2026Mutation21 (4.4)2 (9.5)PloidyHyperdiploid277 (60.7)10 (3.6)0.6454Diploid179 (39.3)8 (4.5)11q LOHNo301 (91.8)5 (1.7)4.9e-4Yes27 (8.2)5 (18.5)1p LOHNo290 (88.4)9 (3.1)1.0000Yes38 (11.6)1 (2.6)HistologyFavorable314 (69.8)1 (0.3)9.3e-9Unfavorable136 (30.2)17 (12.5)Age at diagnosis<18 months213 (44.9)0 (0)0.00001≥18 months261 (55.1)19 (7.3)Age at diagnosis<18 months213 (44.9)0 (0)9.0e-618 mo-5 yrs76 (16.0)1 (1.3)5–12 yrs152 (32.1)15 (9.9)>12 years33 (7.0)3 (9.1)GradeDifferentiating43 (13.2)3 (7.0)0.7168Poorly/undifferentiated283 (86.8)15 (5.3)MKILow/intermediate298 (90.3)18 (6.0)0.2356High32 (9.7)0 (0)RaceBlack55 (12.7)5 (9.1)0.0348other378 (87.3)12 (3.2)aINSS international neuroblastoma staging system, MYCN amplification was assessed by FISH at each COG participating center and by next-generation sequencing (NGS) in our sequence analysis, MKI mitosis–karyorrhexis index.bN number of patients with percentage in parentheses.cThere were 19 ATRX mutations identified but for some tumors, the classifier was not applied so the total number of samples and the total number of ATRX-mutant samples is lower.Fig. 1ATRX mutations are mutually exclusive of MYCN amplification in neuroblastoma.a Summary of ATRX mutations in the COG and PCGP cohorts. Black bars indicate the deleted amino acids; red indicates nonsense mutations; orange indicates frameshift mutations; and blue indicates missense mutations. The major protein domains are shaded in blue. b Bar plot of the percentage of the 819 patients analyzed in this study with ATRX mutations for each age and stage. c, d EFS for patients with neuroblastoma with or without ATRX mutations for stages 1–3 or stage 4 for all age groups. P values were calculated using Cox model. e Heatmap of the distribution of mutations in MYCN, ATRX, and ALK in the COG cohort. Only those tumors with a mutation in at least one of the three genes are indicated. f Micrograph of MYCN FISH (red) for the one neuroblastoma sample with an ATRX mutation and MYCN amplification showing regional amplification (dashed line). The PAX3 (2q35) probe (green) is the control. g Micrograph of a neuroblastoma cell with MYCN homogenously staining region within a region of the tumor that has MYCN amplification (box within the dashed line region in f. h Micrograph of a neuroblastoma cell with two copies of MYCN outside of the region with amplification (box outside of the dashed line region). i Representative photograph of neuroblastoma tumors from a LSL-MYCN;Dbh-iCre mouse. All the tumors developed in this mouse cohorts were harvested and examined. j Micrographs of the histology of the tumor shown in i. The boxed region in each micrograph is magnified in the lower left panel of each image. Three tumors from each group of mice were examined by both H&E and immune-staining. k, l Survival curves for two mouse models of neuroblastoma with or without conditional ATRX deletion. The numbers of mice in each group are indicated. P values were calculated using Log-rank test. Scale bar: f, 5 μm, g, h, 1 μm. COG Children’s Oncology Group, PCGP Pediatric Cancer Genome Project, INSS International Neuroblastoma Staging System, H&E hematoxylin and eosin, TH tyrosine hydroxylase. + +a Summary of ATRX mutations in the COG and PCGP cohorts. Black bars indicate the deleted amino acids; red indicates nonsense mutations; orange indicates frameshift mutations; and blue indicates missense mutations. The major protein domains are shaded in blue. b Bar plot of the percentage of the 819 patients analyzed in this study with ATRX mutations for each age and stage. c, d EFS for patients with neuroblastoma with or without ATRX mutations for stages 1–3 or stage 4 for all age groups. P values were calculated using Cox model. e Heatmap of the distribution of mutations in MYCN, ATRX, and ALK in the COG cohort. Only those tumors with a mutation in at least one of the three genes are indicated. f Micrograph of MYCN FISH (red) for the one neuroblastoma sample with an ATRX mutation and MYCN amplification showing regional amplification (dashed line). The PAX3 (2q35) probe (green) is the control. g Micrograph of a neuroblastoma cell with MYCN homogenously staining region within a region of the tumor that has MYCN amplification (box within the dashed line region in f. h Micrograph of a neuroblastoma cell with two copies of MYCN outside of the region with amplification (box outside of the dashed line region). i Representative photograph of neuroblastoma tumors from a LSL-MYCN;Dbh-iCre mouse. All the tumors developed in this mouse cohorts were harvested and examined. j Micrographs of the histology of the tumor shown in i. The boxed region in each micrograph is magnified in the lower left panel of each image. Three tumors from each group of mice were examined by both H&E and immune-staining. k, l Survival curves for two mouse models of neuroblastoma with or without conditional ATRX deletion. The numbers of mice in each group are indicated. P values were calculated using Log-rank test. Scale bar: f, 5 μm, g, h, 1 μm. COG Children’s Oncology Group, PCGP Pediatric Cancer Genome Project, INSS International Neuroblastoma Staging System, H&E hematoxylin and eosin, TH tyrosine hydroxylase. + +Significantly higher ATRX-mutation frequencies are detected in patients with International Neuroblastoma Staging System (INSS) stage 4 disease (8.6%), high-risk subgroup (14.6%), 11q loss of heterozygosity (LOH), or unfavorable histology and in those who are 18 mo or older at diagnosis (Table 1, Fig. 1b). ATRX-mutant neuroblastomas were significantly more likely to have ALT than ATRX wild-type tumors (89.5% (17/19) vs. 22.2% (4/18); p < 0.0001, two-tailed Fisher’s exact test). In addition, black patients had a significantly higher frequency of ATRX mutation than did those in other racial categories (9.1% vs. 3.3%; p = 0.0348, two-tailed Fisher’s exact test) (Table 1). ATRX mutation frequency did not differ based on sex, ALK status, tumor grade, mitosis–karyorrhexis index, or ploidy (Table 1). + +ATRX mutations were associated with significantly lower 4-year event-free survival (EFS) rates among the groups with non–INSS stage 4 disease (p = 0.0007, two-tailed Fisher’s exact test), INSS stage 4 disease (p = 0.0128, two-tailed Fisher’s exact test), ages 5–12 y (p = 0.0006, two-tailed Fisher’s exact test) and older than 12 y (p = 0.0038, two-tailed Fisher’s exact test) at diagnosis in the COG cohort (Supplementary Data 1 and Fig. 1c, d). Compared with those without a mutation, patients harboring an ATRX mutation were 5.0 times more likely to experience an adverse event (p < 0.0001) and 3.4 times more likely to die of their disease (p = 0.0046, two-tailed Fisher’s exact test) (Supplementary Data 1). + +In the cohort of 819 tumors, ATRX mutations (n = 64) and MYCN amplification (n = 140) showed statistically significant mutual exclusivity (p = 0.0037, Cochrane-Mantel-Haenszel test) (Fig. 1e and Supplementary Note 1). Only 1 of 819 (0.1%) tumors had both lesions. However, this was an unusual discordant sample because it was scored as MYCN-amplified by fluorescence in situ hybridization (FISH) in COG’s neuroblastoma reference lab but non-amplified based on normalized read depth from our custom capture Illumina next-generation sequencing data (Supplementary Fig. 1A). To resolve this discrepancy, we repeated the FISH analysis and found regional heterogeneity of MYCN amplification (Fig. 1f–h). Specifically, the region with MYCN amplification had a high mitosis–karyorrhexis index and was less likely to have ultrabright telomere foci, characteristic of ALT, in ATRX-mutant neuroblastoma cells (Fig. 1f–h and Supplementary Fig. 1B–E). + +To determine whether ATRX mutations and MYCN amplification are incompatible in vivo, we conditionally inactivated ATRX in two genetically engineered neuroblastoma mouse models (Fig. 1i–l)8,17,18. One model (LSL-MYCN;Dbh-iCre) is a transgenic line with conditional expression of MYCN from Rosa26a locus by Cre expressed in the dopaminergic cells of the sympathoadrenal lineage19. The other model includes the Th-ALKF1174L transgene that potentiates MYCN-mediated tumorigenesis20. For each of these models, tumor formation was reduced, and survival was significantly increased when Atrx was simultaneously inactivated with elevated MYCN expression (Fig. 1k–l). The few tumors that did form in the AtrxLox/Lox;LSL-MYCN;Dbh-iCre;ALKF1174L mice all had evidence of an intact Atrx allele, suggesting that rare cells that did not undergo Atrx inactivation contributed to tumor formation. It is also possible that ALK signaling partially rescued the synthetic lethality. Atrx inactivation has been shown to reduce cell survival in brain, muscles and testes8,21,22. Histopathological review of adrenal gland and paravertebral sympathetic ganglia from LSL-MYCN;Dbh-iCre and AtrxLox/Lox;LSL-MYCN;Dbh-iCre mice at 3 weeks and 1 year of age showed no differences (Supplementary Fig. 2, Supplementary Note 2), suggesting that the lack of tumor formation was not due to death of the NB cell of origin. + +To extend our data to humans, we characterized the expression of MYCN, ATRX, and DAXX proteins across 12 human cancer cell lines (Fig. 2a). We used eight neuroblastoma cell lines, including three MYCN-amplified lines (IMR32, SKNBE2, NB-5) or moderately elevated levels of MYCN (NBL-S) (Supplementary Data 2), two lines (SKNMM and CHLA90) produce ATRX with an in-frame deletion (ATRXIFD) and are diploid for MYCN (Fig. 2a), one osteosarcoma cell line with no detectable expression of ATRX (U2OS) and three lines as controls.Fig. 2ATRX mutations and MYCN amplification are incompatible.a Immunoblot of 12 cell lines. Boxes indicate ATRX-mutant lines. b Immunoblot for HeLa cells transfected with sc-shRNA or ATRX-specific shRNAs. The fraction of ATRX relative to sc-shRNA is indicated. c Bar plot of the number of colonies/well (mean of duplicates) after transfection of shRNAs, normalized to the sc-shRNA (dashed line). d Photograph of cresyl violet-stained colonies after transfection with shRNAs and bar plots of the average number of colonies/well. e Map of two gRNAs targeting ATRX. f Plot of the mutation frequency for gRNA-2, deletions (green) and insertions (red). g Bar plot of the proportion of ATRX-mutant alleles (100,000× coverage/sample). h Immunoblot for the cell lines with the doxycycline-inducible MYCN expression construct. Numbers indicate MYCN/GAPDH fluorescence intensity. The experiment was done three times with similar results. i Line graphs of the growth curves (mean ± SD) for each cell line ± doxycycline, n = 3. j Colony assay for SKNBE2MYCN and SKNMMMYCN cells ± doxycycline, the experiment was repeated twice with similar results. k Brightfield micrograph of SKNMMMYCN cells after 8 days in culture ± doxycycline, the experiment was repeated twice with similar results. l Brightfield micrographs of three individual SKNMMMYCN cells ± doxycycline at indicated timepoints, relative to the starting point of the movies (6 days + doxycycline), 44 cells were analyzed. m Immunoblots of SKNMMMYCN cells without doxycycline or after 2, 4, or 45 days in culture. The escapers are pools of cells that grew in the presence of doxycycline (n = 4). PCR for MYCN transgene and a control locus (RPL0) is indicated in the lower portion. n Xenogen image of a mouse with an orthotopic neuroblastoma tumor (photograph), that arose from a 1:1 mixture of SKNMMMYCN cells and SKNMMCONT–Luc:YFP cells (n = 5). o Line graph of xenogen image data described in n. p Xenogen image of a mouse with an orthotopic neuroblastoma tumor (photograph), that arose from a 1:1 mixture of SKNMMMYCN–Luc:YFP cells and SKNMMCONT cells (n = 5). q Line graph of xenogen image data described in p. Scale bars: k, 10-μm. DOX doxycycline, IFD in-frame deletion, sc scrambled. + +a Immunoblot of 12 cell lines. Boxes indicate ATRX-mutant lines. b Immunoblot for HeLa cells transfected with sc-shRNA or ATRX-specific shRNAs. The fraction of ATRX relative to sc-shRNA is indicated. c Bar plot of the number of colonies/well (mean of duplicates) after transfection of shRNAs, normalized to the sc-shRNA (dashed line). d Photograph of cresyl violet-stained colonies after transfection with shRNAs and bar plots of the average number of colonies/well. e Map of two gRNAs targeting ATRX. f Plot of the mutation frequency for gRNA-2, deletions (green) and insertions (red). g Bar plot of the proportion of ATRX-mutant alleles (100,000× coverage/sample). h Immunoblot for the cell lines with the doxycycline-inducible MYCN expression construct. Numbers indicate MYCN/GAPDH fluorescence intensity. The experiment was done three times with similar results. i Line graphs of the growth curves (mean ± SD) for each cell line ± doxycycline, n = 3. j Colony assay for SKNBE2MYCN and SKNMMMYCN cells ± doxycycline, the experiment was repeated twice with similar results. k Brightfield micrograph of SKNMMMYCN cells after 8 days in culture ± doxycycline, the experiment was repeated twice with similar results. l Brightfield micrographs of three individual SKNMMMYCN cells ± doxycycline at indicated timepoints, relative to the starting point of the movies (6 days + doxycycline), 44 cells were analyzed. m Immunoblots of SKNMMMYCN cells without doxycycline or after 2, 4, or 45 days in culture. The escapers are pools of cells that grew in the presence of doxycycline (n = 4). PCR for MYCN transgene and a control locus (RPL0) is indicated in the lower portion. n Xenogen image of a mouse with an orthotopic neuroblastoma tumor (photograph), that arose from a 1:1 mixture of SKNMMMYCN cells and SKNMMCONT–Luc:YFP cells (n = 5). o Line graph of xenogen image data described in n. p Xenogen image of a mouse with an orthotopic neuroblastoma tumor (photograph), that arose from a 1:1 mixture of SKNMMMYCN–Luc:YFP cells and SKNMMCONT cells (n = 5). q Line graph of xenogen image data described in p. Scale bars: k, 10-μm. DOX doxycycline, IFD in-frame deletion, sc scrambled. + +To knockdown ATRX, we used two different lentiviral vectors expressing shRNAs targeting ATRX (#9 and #20) (Fig. 2b). The optimal knockdown was ~ 86% in HeLa cells after 48 h with ATRX shRNA–20. Two weeks after introducing shRNAs, telomere content, ALT status and the proportions of cells in different phases of the cell cycle did not differ in any cell line tested (Supplementary Fig. 3A–I). Knocking down ATRX decreased colony formation in MYCN-amplified neuroblastoma cells (IMR32 and NB-5) but not cell lines expressing wild-type MYCN (Fig. 2c, d). + +To inactivate ATRX in the MYCN-amplified neuroblastoma cell lines, we designed and validated two guide RNAs (gRNA-2 and gRNA-5) to exon 6 of ATRX (Fig. 2e, f). We transfected 293T, SKNBE2, and IMR32 cells with each gRNA and a plasmid expressing Cas-9 and purified the transfected cells by flow cytometry. At 2, 10, and 14 days after plating the purified cells, we harvested DNA, performed PCR using primers spanning both of the two gRNA-target sequences, and performed Illumina MiSeq analysis to calculate the proportion of mutated alleles in each sample. Although the ATRX-mutant alleles were abundant in 293T and SKNBE2 cells 2 days after plating, those cells harboring ATRX-mutant alleles were lost by 10 days in culture (Fig. 2g and Supplementary Fig. 3J). In MYCN-amplified IMR32 cells, no mutant alleles were detected at any timepoint, though a control gRNA to the AAVS1 locus led to efficient mutagenesis. A similar study performed with a gRNA to DAXX had no effect on the viability of 293T, SKNBE2, or IMR32 cells (Supplementary Fig. 3K-M). Taken together, our data suggest that ATRX inactivation is incompatible with MYCN amplification in neuroblastoma, which is consistent with data from the Project Achilles (https://depmap.org/portal/achilles/). + +Next, we induced the expression of MYCN in ATRX-mutant neuroblastoma cells by using a doxycycline-inducible vector (Fig. 2h). Initially, we used two ATRX-mutant cell lines (SKNMM and U2OS), a MYCN-amplified cell line (SKNBE2), and a cell line that is wild-type for both genes (SKNFI) (Fig. 2a). The levels of MYCN achieved with doxycycline induction in both ATRX-mutant lines were similar to those in MYCN-amplified cell lines (Fig. 2h). + +The growth rates of SKNFIMYCN and SKNBE2MYCN cells, after 10 days in culture ± doxycycline, showed no difference with induction of MYCN expression, but by 6 days in culture, the two ATRX-mutant lines showed a marked loss of cells (Fig. 2i, j). A parallel experiment performed with stable cell lines that had only the doxycycline-inducible vector lacking MYCN showed no effect on cell growth. More than 95% of the cells in the SKNMMMYCN cultures were lost by 8–10 days in the presence of doxycycline (Fig. 2k). Live imaging of cells ± doxycycline on Days 6–8, the peak period of cell loss (Fig. 2l), enabled us to determine the timing of cell death in 44 cells from 13 different movies (Supplementary Data 3). In total, 54% of the cells died after initiating cytokinesis; the remaining cells died 20.2 ± 12.8 h (mean ± standard deviation) after cytokinesis. These data were extended to five other ATRX-mutant cell lines including two glioma cell lines that had isogenic ATRX-wild-type controls (Supplementary Fig. 4)23. A marker of DNA-replicative stress (pRPA32) was upregulated in the absence of ATRX and in the presence of doxycycline. There was no effect of ectopic MYCN expression on a DAXX-mutant cell line (Supplementary Fig. 4A, B). A small number of SKNMMMYCN cells escaped cell death and continued to grow for 45 days in the presence of doxycycline. The ATRXIFD was still expressed and the MYCN transgene was still present, but MYCN protein was lost (Fig. 2m). + +To determine whether an elevated level of MYCN is incompatible with ATRX mutation in human neuroblastoma cells in vivo, we performed an in vivo growth competition assay. In one cohort of mice, SKNMMMYCN cells labeled with luciferase and YFP were mixed in a 1:1 ratio with unlabeled control SKNMMCONT cells, which contained the doxycycline-inducible vector lacking MYCN. In a second cohort, the SKNMMCONT cells were labeled with luciferase and YFP and the SKNMMMYCN cells were unlabeled. Cells were injected into the para-adrenal space of 25 immunocompromised mice for each cohort. The luciferase was monitored every week, and ultrasound was performed every 2 weeks. Animals were randomized to +doxycycline and –doxycycline groups once the tumor was larger than 14 mm3 by three-dimensional ultrasound and/or 1.4 × 107 photons/s/cm2/sr by xenogen imaging (Supplementary Data 4). The SKNMMMYCN cells were outcompeted by the SKNMMCONT cells in vivo, as determined by xenogen and ultrasound (Fig. 2n–q). + +MYCN is a global transcriptional regulator that activates and represses genes controlling cell division, cell size, and cell differentiation during development19. MYCN can change gene expression by modifying the epigenetic status of target genes19. To analyze the epigenomic landscape in ATRX-mutant neuroblastoma cells, compared with that in wild-type ATRX neuroblastoma, we selected eight cell lines that encompass the major genetic groups, including those with MYCN amplification or ATRX mutations (Fig. 3a). We also included eight orthotopic patient-derived xenografts (O-PDXs) that were derived by para-adrenal injection of patient tumors into immunocompromised mice as described previously20,24 (Fig. 3a). One of those O-PDXs (SJNBL047443_X1) has an ATRX in-frame deletion (Fig. 3b, c and Supplementary Data 2). Autopsy tumor material from a patient with ATRX-mutant neuroblastoma (SJNBL030014_D) was also included (Fig. 3a and Supplementary Data 2). We performed whole-genome bisulfite sequencing (WGBS; Supplementary Fig. 5 and Data 5), RNA-sequencing, and ChIP-seq of eight histone marks (H3K4me1, H3k4me2, H3K4me3, H3K27me3, H3K27Ac, H3K36me3, H3K9/14Ac, and H3K9me3), CTCF, BRD4, and RNA polymerase II (PolII) across all eight cell lines, the eight O-PDX tumors, the autopsy sample, and normal fetal adrenal medulla. We also performed MYCN ChIP-seq on three O-PDXs with MYCN amplification (SJNBL046_X, SJNBL012407_X1, and SJNBL013762_X1) and the SKNFIMYCN, SKNBE2MYCN, and SKNMMMYCN cell lines with doxycycline on Day 4 in culture as well as matched controls. To relate gene expression changes to epigenetic changes induced by MYCN, we performed ChIP-seq for the eight histone marks, CTCF, Brd4, and RNA PolII in the SKNMMMYCN cells ± doxycycline on Day 4 in culture.Fig. 3Transcriptional and epigenetic changes induced by MYCN in ATRX-deficient neuroblastoma cells.a Heatmap of the mutations in the O-PDXs and cell lines used in this study. An autopsy sample was also used for molecular profiling. b Micrographs of histologic analysis of the ATRX-mutant neuroblastoma O-PDX and corresponding patient tumor. c Plot of sequence read depth from Illumina sequencing for the ATRX gene in the patient’s germline (blue) and tumor (red), showing somatic deletion of exons 2–11. d Heatmap of the chromHMM states used in this study with functional annotation. e ChromHMM plot of the DIRAS1 gene and a bar plot of the gene’s expression on the right. DNA methylation and H3K27me3 and H3K4me3 ChIP-seq for DIRAS1 for a MYCN-amplified tumor (SJNBL046_X1), an ATRX-mutant tumor (SJNBL047443_X1), and fetal adrenal medulla. f ChIP-seq tracks for a portion of the SLC3A2 gene spanning the promoter in SKNMMMYCN cells in the presence or absence of doxycycline after 4 days in culture. Dashed line indicates the start of transcription. The scales are indicated on the right. g ChromHMM and ChIP-seq for MYCN and RNA PolII for a portion of the SLC3A2 gene spanning the promoter in SKNMMMYCN cells in the presence or absence of doxycycline after 4 days. The dashed lines indicate the region with induction of ChromHMM state 13 that is expanded after MYCN binds the promoter. ChromHMM chromatin Hidden Markov Modeling, FPKM fragments per kilobase of transcript per million mapped reads, O-PDX orthotopic patient-derived xenograft. + +a Heatmap of the mutations in the O-PDXs and cell lines used in this study. An autopsy sample was also used for molecular profiling. b Micrographs of histologic analysis of the ATRX-mutant neuroblastoma O-PDX and corresponding patient tumor. c Plot of sequence read depth from Illumina sequencing for the ATRX gene in the patient’s germline (blue) and tumor (red), showing somatic deletion of exons 2–11. d Heatmap of the chromHMM states used in this study with functional annotation. e ChromHMM plot of the DIRAS1 gene and a bar plot of the gene’s expression on the right. DNA methylation and H3K27me3 and H3K4me3 ChIP-seq for DIRAS1 for a MYCN-amplified tumor (SJNBL046_X1), an ATRX-mutant tumor (SJNBL047443_X1), and fetal adrenal medulla. f ChIP-seq tracks for a portion of the SLC3A2 gene spanning the promoter in SKNMMMYCN cells in the presence or absence of doxycycline after 4 days in culture. Dashed line indicates the start of transcription. The scales are indicated on the right. g ChromHMM and ChIP-seq for MYCN and RNA PolII for a portion of the SLC3A2 gene spanning the promoter in SKNMMMYCN cells in the presence or absence of doxycycline after 4 days. The dashed lines indicate the region with induction of ChromHMM state 13 that is expanded after MYCN binds the promoter. ChromHMM chromatin Hidden Markov Modeling, FPKM fragments per kilobase of transcript per million mapped reads, O-PDX orthotopic patient-derived xenograft. + +To define the chromatin states and analyze the transitions thereof across the genome, we performed chromatin Hidden Markov Modeling (ChromHMM) using all 393 ChIP-seq data sets (Fig. 3d, e and Supplementary Data 6). To share these data, we have developed a cloud-based viewer (https://pecan.stjude.org/proteinpaint/study/mycn_nbl_2018). We identified 394 genes that were upregulated and 544 that were downregulated 4 days after adding doxycycline to the SKNMMMYCN or U2OSMYCN cell lines (Supplementary Data 7). Gene set-enrichment analysis of the upregulated genes identified MYC and MYCN targets as the most-enriched data sets (Supplementary Fig. 6A): 84% of the upregulated genes and 42% of the downregulated genes had a MYCN-binding site based on MYCN ChIP-seq for SKNMMMYCN cells in the presence of doxycycline (Supplementary Fig. 6B-E and Data 7). Analysis of the chromHMM for the upregulated MYCN targets showed that expansion of state 13 (strong transcribed gene body) downstream of the transcriptional start site was the most significant change (Fig. 3f, g and Supplementary Data 7). This is consistent with previous data showing that MYC regulates RNA PolII pause release not transcriptional initiation25. These upregulated target genes are enriched for the pathways involved in metabolism (e.g., SLC3A2, SLC7A5) and mitochondrial gene expression (e.g., PNPT1) (Fig. 3f, g, Supplementary Fig. 6B-E, and Data 7), consistent with previous studies, demonstrating that elevated MYC/MYCN induces metabolic reprogramming in cancer cells11. + +One form of metabolic reprogramming induced by MYC/MYCN is glutamine addiction26. The glutamine transporter (SLC1A5) involved in glutamine uptake and the bidirectional transporter (SLC7A5–SLC3A2) involved in glutamine efflux coupled with essential amino acid import are direct targets of MYCN and are upregulated in SKNMMMYCN cells in the presence of doxycycline (Fig. 3g, Supplementary Data 7 and 6D, E)27. Following MYCN induction, we quantified glutamine metabolism in SKNMMMYCN cells ± doxycycline on Day 4 in culture and measured eight tricarboxylic acid (TCA) cycle cellular metabolites28. The proportions of five carbon-labeled glutamate and α-ketoglutarate increased, consistent with glutamine utilization for the TCA cycle as well as reductive carboxylation to produce five carbon-labeled citrate (Fig. 4a, b).Fig. 4Expression of MYCN in ATRX-mutant cells leads to metabolic reprogramming and mitochondrial dysfunction.a Simplified drawing of the TCA cycle and reductive carboxylation, blue circles: 13C derived from 13C5-glutamine. b Bar plot (mean of two technical replicates) of 13C-labeled isotopomers in SKNMMMYCN cells on day 4 ± doxycycline after 5 h of labeling with 13C5-glutamine. The arrows indicate an increase in M + 5 glutamate and M + 5 α-ketoglutarate (the most abundant and direct derivatives of 13C5-glutamine) in the presence of doxycycline. c Photograph of culture media on day 3 ± doxycycline. d Line plots of media pH ± doxycycline for each day. The culture media were changed on Day 3. e Histogram (mean of two biological replicates) of the levels of lactate at Day 4 in SKNMMMYCN cells ± doxycycline. f Simplified drawing of the glycolysis highlighting the production of lactate from 13C6-glucose. g Bar plots of 13C-labeled isotopomers of lactate and α-ketoglutarate in the medium or cells after 18 h of labeling with 13C6-glucose or 13C5-glutamine, respectively, at Day 4 in SKNMMMYCN cells ± doxycycline, with concentrations indicated on the bars. h, i Box and Whiskers plots of MitoTracker Green (H) and TMRE (I) staining ± doxycycline on Day 4. Each Box shows the 10th to 90th percentiles range of data (at least 15,000 cells), the line represents the median and the Whiskers show the minimum and maximum data range. P values were calculated using two-tailed Mann–Whitney non-parametric test. j, k Representative results of oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) (mean and SEM of three technical replicates) for 2.5 × 104 SKNMMMYCN cells grown under the Mito-Stress assay conditions for the Seahorse. The experiment was done three times with similar results. l–o Representative electron micrograph of mitochondria from SKNMMMYCN cells after 4 days ± doxycycline l, n showing disruption in the mitochondrial cristae (arrows). Bar plot of scoring of mitochondrial morphology for the SKNMMMYCN cells ± doxycycline, numbers of scored mitochondria are indicated m, o. P value was calculated using Chi-square test. DOX doxycycline, FCCP carbonyl cyanide-p-trifluoromethoxy phenylhydrazone, Olig oligomycin, Rot/AA rotenone/antimycin A, TCA tricarboxylic acid, TMRE tetramethylrhodamine ethyl ester. + +a Simplified drawing of the TCA cycle and reductive carboxylation, blue circles: 13C derived from 13C5-glutamine. b Bar plot (mean of two technical replicates) of 13C-labeled isotopomers in SKNMMMYCN cells on day 4 ± doxycycline after 5 h of labeling with 13C5-glutamine. The arrows indicate an increase in M + 5 glutamate and M + 5 α-ketoglutarate (the most abundant and direct derivatives of 13C5-glutamine) in the presence of doxycycline. c Photograph of culture media on day 3 ± doxycycline. d Line plots of media pH ± doxycycline for each day. The culture media were changed on Day 3. e Histogram (mean of two biological replicates) of the levels of lactate at Day 4 in SKNMMMYCN cells ± doxycycline. f Simplified drawing of the glycolysis highlighting the production of lactate from 13C6-glucose. g Bar plots of 13C-labeled isotopomers of lactate and α-ketoglutarate in the medium or cells after 18 h of labeling with 13C6-glucose or 13C5-glutamine, respectively, at Day 4 in SKNMMMYCN cells ± doxycycline, with concentrations indicated on the bars. h, i Box and Whiskers plots of MitoTracker Green (H) and TMRE (I) staining ± doxycycline on Day 4. Each Box shows the 10th to 90th percentiles range of data (at least 15,000 cells), the line represents the median and the Whiskers show the minimum and maximum data range. P values were calculated using two-tailed Mann–Whitney non-parametric test. j, k Representative results of oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) (mean and SEM of three technical replicates) for 2.5 × 104 SKNMMMYCN cells grown under the Mito-Stress assay conditions for the Seahorse. The experiment was done three times with similar results. l–o Representative electron micrograph of mitochondria from SKNMMMYCN cells after 4 days ± doxycycline l, n showing disruption in the mitochondrial cristae (arrows). Bar plot of scoring of mitochondrial morphology for the SKNMMMYCN cells ± doxycycline, numbers of scored mitochondria are indicated m, o. P value was calculated using Chi-square test. DOX doxycycline, FCCP carbonyl cyanide-p-trifluoromethoxy phenylhydrazone, Olig oligomycin, Rot/AA rotenone/antimycin A, TCA tricarboxylic acid, TMRE tetramethylrhodamine ethyl ester. + +Inefficient use of glutamine and/or glucose by cancer cells can lead to the export of glycolytic intermediates such as lactate29. The pH of the medium was quickly acidified in subconfluent cultures of SKNMMMYCN or U2OSMYCN cells in the presence of doxycycline; however, SKNBE2MYCN and SKNFIMYCN cells showed no difference in the pH of their media (Fig. 4c, d). The reduced pH correlated with an increase in lactate in the culture medium and dependence on glutamine for survival (Fig. 4e and Supplementary Fig. 7A). To provide additional data on the metabolic reprogramming induced by MYCN in ATRX-deficient cells, we performed a more-comprehensive metabolomic profiling of 54 metabolites using 13C-labeled glucose and 13C-labeled glutamine (Fig. 4f, g and Supplementary Data 8). Glucose was the major source of lactate; relatively few of the carbons from glucose were used for the TCA cycle (Fig. 4f, g and Supplementary Data 8). In contrast, glutamine was converted to α-ketoglutarate for the TCA cycle and reductive carboxylation. + +Glutamine is also an important mitochondrial substrate; cells must precisely balance the expression and activity of proteins encoded by the nucleus with those encoded by the mitochondria to maintain homeostasis28. Cells also upregulate pathways required to mitigate the ROS that are a natural byproduct of mitochondrial metabolism to prevent excessive protein or DNA damage. For example, glutamine uptake in cancer cells can elevate glutathione levels because glutamine is converted to the glutathione precursor, glutamate30. Cell lines that were sensitive to glutamine depletion in the medium also had significantly elevated levels of glutathione (Supplementary Fig. 7B). ROS and DNA damage were increased in the absence of glutamine (Supplementary Fig. C, D). + +Pathways involved in mitochondrial homeostasis were upregulated in SKNMMMYCN or U2OSMYCN cells in the presence of doxycycline (Supplementary Data 7). Little change in mitochondrial mass was measured using MitoTracker Green in SKNMMMYCN or U2OSMYCN cells ± doxycycline, but the mitochondrial membrane potential (Δψm) measured with tetramethylrhodamine-ethyl ester (TMRE) was significantly lower in the presence of doxycycline (Fig. 4h, i). To assess mitochondrial function in MYCN-overexpressing SKNMM cells, we measured the oxygen consumption rate. Induction of MYCN in SKNMM cells decreased basal mitochondrial respiration and respiratory reserve capacity, relative to that seen in un-induced SKNMM cells (Fig. 4j, k and Supplementary Fig. 8). Transmission electron micrographs showed that SKNMMMYCN cells in the presence of doxycycline had more disrupted mitochondria on Day 4 and subsequent timepoints than did SKNMMMYCN cells maintained in culture without doxycycline or the other cell lines (Fig. 4l–o, Supplementary Fig. 9). Similar results were obtained for the ATRX-deficient U2OSMYCN cells but not ATRX-wild-type SKNBE2MYCN cells (Supplementary Fig. 9). + +The mitochondrial dysfunction described above may lead to the accumulation of ROS and in turn contribute to replicative stress through DNA damage. The ROS increase began at Day 4 and persisted until Day 6 in SKNMMMYCN cells in the presence of doxycycline (Fig. 5a). This was accompanied by increased expression of γH2AX protein, a biomarker for DNA double-strand breaks (Fig. 5b, c). Spectral karyotyping showed a significant increase in cells with DNA fragmentation and translocation in the SKNMMMYCN cells in the presence of doxycycline (Fig. 5d). We also detected increased DNA fragmentation in SKNMMMYCN and U2OSMYCN cells in the presence of doxycycline by using a single-cell gel electrophoresis assay (Fig. 5e–g).Fig. 5Expression of MYCN in ATRX-mutant cells leads to ROS production, DNA damage, and replicative stress.a Bar plot (mean±SD) of ROS levels on Day 6 ± doxycycline. P value was calculated using two-tailed Student t test. b Immunoblots on Day 4 ± doxycycline. The experiment was repeated with the same results. c Immunofluorescent detection of γH2AX (red) and nuclei (blue) in SKNMMMYCN cells on Day 4 ± doxycycline. The experiment was repeated with the same results. d Spectral karyotype analysis (SKY) of SKNMMMYCN cells ± DOX on Day 8. Chromosomes are shown adjacent to the pseudo-colored representation. Arrows indicate translocations. The pie charts show the proportion of cells with DNA fragmentation (n = 50). P value was calculated using Chi-square test. e, f Micrographs of single-cell electrophoresis of individual nuclei (dashed circles) and their COMET tail (arrows) ± doxycycline on Day 5. g Mean±SD of the COMET assay scoring. The number of analyzed cells are presented on the graph. h Photograph of cresyl violet-stained colonies from SKNMMMYCN cells ± doxycycline, with increasing concentrations of retinoic acid (RA). i Immunoblots of SKNMMMYCN cells ± doxycycline, with or without 5 μm RA. The level of MYCN protein was reduced by 30% (0.7×) in the presence of RA. j ChromHMM and ChIP-seq for MYCN, H3K4me3, and H3K27me3 for the CUX2 promoter for a MYCN-amplified neuroblastoma (SJNBL012407_X1) xenograft and SKNMMMYCN cells in the presence or absence of doxycycline after 4 days in culture. The gene expression (FPKM) is indicated, and the dashed line indicates the start of transcription. k Line plot of SKNMMMYCN cells in the presence or absence of doxycycline after 4, 6, 8, and 10 days in culture with or without ectopic expression of CUX2 or the GFP control from lentiviral infection. Each point is the mean and standard deviation of triplicate experiments, and the asterisks indicate statistical significance (P = 0.005, 0.001, and 0.008 at days 6, 8, and 10, respectively, two-tailed Student's t test). Scale bars e, f, 10 μm. DOX doxycycline, RA retinoic acid, ROS reactive-oxygen species, m marker chromosome fragments that could not be definitively identified. + +a Bar plot (mean±SD) of ROS levels on Day 6 ± doxycycline. P value was calculated using two-tailed Student t test. b Immunoblots on Day 4 ± doxycycline. The experiment was repeated with the same results. c Immunofluorescent detection of γH2AX (red) and nuclei (blue) in SKNMMMYCN cells on Day 4 ± doxycycline. The experiment was repeated with the same results. d Spectral karyotype analysis (SKY) of SKNMMMYCN cells ± DOX on Day 8. Chromosomes are shown adjacent to the pseudo-colored representation. Arrows indicate translocations. The pie charts show the proportion of cells with DNA fragmentation (n = 50). P value was calculated using Chi-square test. e, f Micrographs of single-cell electrophoresis of individual nuclei (dashed circles) and their COMET tail (arrows) ± doxycycline on Day 5. g Mean±SD of the COMET assay scoring. The number of analyzed cells are presented on the graph. h Photograph of cresyl violet-stained colonies from SKNMMMYCN cells ± doxycycline, with increasing concentrations of retinoic acid (RA). i Immunoblots of SKNMMMYCN cells ± doxycycline, with or without 5 μm RA. The level of MYCN protein was reduced by 30% (0.7×) in the presence of RA. j ChromHMM and ChIP-seq for MYCN, H3K4me3, and H3K27me3 for the CUX2 promoter for a MYCN-amplified neuroblastoma (SJNBL012407_X1) xenograft and SKNMMMYCN cells in the presence or absence of doxycycline after 4 days in culture. The gene expression (FPKM) is indicated, and the dashed line indicates the start of transcription. k Line plot of SKNMMMYCN cells in the presence or absence of doxycycline after 4, 6, 8, and 10 days in culture with or without ectopic expression of CUX2 or the GFP control from lentiviral infection. Each point is the mean and standard deviation of triplicate experiments, and the asterisks indicate statistical significance (P = 0.005, 0.001, and 0.008 at days 6, 8, and 10, respectively, two-tailed Student's t test). Scale bars e, f, 10 μm. DOX doxycycline, RA retinoic acid, ROS reactive-oxygen species, m marker chromosome fragments that could not be definitively identified. + +Cells with DNA-replicative stress are sensitive to hydroxyurea31. Thus, we maintained SKNMMMYCN cells cultured ± doxycycline with different concentrations of hydroxyurea (12.5, 25, 50, or 100 μM on Days 1–4) and measured cell viability using CellTiter-Glo on Day 4. The SKNMM cells expressing MYCN were significantly more sensitive to hydroxyurea at 50 or 100 μM concentrations (Supplementary Fig. 10A). The DNA content in each cell line, ±doxycycline at Day 4 in culture, showed an increase in S and G2/M phases in the SKNMMMYCN cells in the presence of doxycycline, which was consistent with cell cycle arrest as a result of DNA damage (Supplementary Fig. 10B). The ROS scavenger N-acetyl cysteine had a small but reproducible effect on reducing cell death induced by ectopic MYCN expression in SKNMM cells (Supplementary Fig. 10C, D). + +To identify other pathways that may modulate to MYCN-induced cell death in ATRX-deficient cells, we performed a dose–response screen using two drug libraries. The first was a collection of 177 oncology drugs and compounds in late clinical development (Phase II or later). The second was a customized library of 34 drugs and compounds that target proteins or pathways that have been reported to reduce MYC/MYCN oncogenic activity (Data S9). Within the oncology drug set, 37 compounds selectively potentiated killing of SKNMMMYCN cells in the presence of doxycycline (Supplementary Data 9); the majority (59.4%, 22/37) were drugs that induce DNA damage, inhibit DNA repair or replication, or inhibit cell cycle checkpoints (Supplementary Fig. 10E–G and Data 9). The only two drugs that partially rescued the cell death were a third-generation retinoid derivative (bexarotene) and an mTORC1/2 inhibitor (INK128) (Supplementary Fig. 10H and Data 9). The bexarotene result was particularly interesting because retinoids are an effective treatment for neuroblastoma (i.e., they reduce MYCN expression and induce differentiation). These results were independently validated using retinoic acid (Fig. 5h, i). + +Together, these data suggest that MYCN induction causes metabolic reprogramming and mitochondrial dysfunction that contribute to increased ROS and DNA-replicative stress. It may be synthetically lethal in ATRX-mutant cells because they already have DNA-replicative stress in part owing to reduced ability to resolve G4 DNA structures. A subset of genes that are normally induced in MYCN-amplified neuroblastomas also may have failed to be activated in the ATRX-mutant neuroblastomas owing to the DNA and/or histone modifications. We identified one gene that met those criteria. CUX2 encodes a homeodomain protein with three CUT repeats that is expressed in the developing nervous system and important for the repair of oxidative DNA damage32. CUX2 is a direct MYCN target and is upregulated in MYCN-amplified neuroblastomas but not ATRX-mutant neuroblastomas (Fig. 5j). The CUX2 promoter is epigenetically repressed by H3K27me3 in ATRX-mutant neuroblastomas and not induced in SKNMM cells in the presence of doxycycline (Fig. 5j). There was a partial rescue of cell death in SKNMMMYCN cells in the presence of doxycycline when CUX2 was ectopically expressed (Fig. 5k), suggesting that CUX2 is important for oxidative stress in MYCN-amplified neuroblastomas. CUX2 is not induced by retinoic acid in 11 different NB cell lines (Supplementary Fig. 11 and Data 10) so the mechanism of partial rescue by retinoic acid is distinct from that of CUX2. Combining NAC with ectopic expression of CUX2 did not significantly increase survival over NAC or CUX2 alone (Supplementary Fig. 11B). + +ATRX has an important role in H3.3 deposition in chromatin33. In addition to telomeres and pericentric heterochromatin, ATRX helps to resolve G-quadruplex (G4) structures in the DNA by deposition of H3.3 at those sites across the genome5,33. This is important because G4 structures can inhibit DNA replication and transcription33,34. G4 structures often overlap with noncanonical DNA:RNA hybrids called R-loops because both structures are favored in G-rich regions of the genome under negative torsional tension (e.g., active promoters)35,36. In cancer cells, R-loops can contribute to DNA breaks and genome instability37,38. To determine whether the pattern of H3.3 deposition is altered in ATRX-mutant NB, we performed CUT&RUN39–41 on six NB cell lines divided into three groups: MYCN-amplified cells (SKNBE2 and IMR32), ATRX-mutant cells (SKNMM and CHLA90) and MYCN non-amplified/ATRX wild-type cells (LAN6 and SKNFI). We identified the reproducible peaks in each group (ATRX, MYCN, and WT) and created a reference list of 51,919 H3.3 peaks across our data set. Among those peaks, 18,436 (35%) overlapped with DNA sequences that are predicted to form G4 structures and 70% (12,932) of those are predicted to form R-loops42. As expected, there were fewer H3.3 peaks in the ATRX group and nearly a twofold reduction in H3.3 peaks overlapping G4 sequences (Fig. 6a). For every H3.3 peak in the data set, we assigned it to a group defined by the pattern across the samples (ATRX, MYCN, and WT). The constitutive (C) group includes H3.3 peaks found at the same location across all samples (ATRX, MYCN, and WT) (Fig. 6b). The enriched (E) group includes H3.3 peaks that are higher in ATRX than MYCN and WT (Fig. 6b). The depleted (D) group includes H3.3 peaks that are lower in ATRX than in MYCN and WT and the overlapping (O) group has peaks in ATRX and either MYCN or WT (Fig. 6b). The D group was the largest (Fig. 6b) and 2428 of those had a correlation between ChromHMM state and the presence or absence of H3.3 peaks; 514 (21%) had G4 sequences and 69% of those (357) were predicted to form R-loops. The majority of H3.3 peaks that were depleted in ATRX-mutant NB relative to MYCN and WT that had a correlation with ChromHMM state were in enhancers, whereas those with G4 sequences (and R-loops) were in promoters (Fig. 6c). For both the non-genic and genic regions, there was a shift from active euchromatic states (states 1–4) to inactive heterochromatic states (states 15–17) in the ATRX group relative to MYCN and WT (Fig. 6d–h). There was overlap between H3.3 peaks and CTCF sites (state 18) consistent with the role of H3.3 in CTCF binding to establish insulated chromatin domains43–45. We performed pathway analysis on the genes that had promoters or enhancers that were depleted for H3.3 in ATRX-mutant NB and had correlation with ChromHMM states and reduced expression in ATRX-mutant relative to MYCN and WT. The most significant pathways were involved in neuronal differentiation and neural development (Supplementary Data 11). For example, the DUSP26 gene is a MYCN target that has G4 sequences immediately downstream of the promoter predicted to form R-loops and selective deposition of H3.3 in MYCN-amplified NB relative to ATRX-deficient NB cells (Fig. 6g). DUSP26 is a phosphatase that is expressed in neuroendocrine cells and can inhibit neuronal differentiation46,47. It has also been shown to dephosphorylate and inhibit p53 in neuroblastoma cells48. We performed the same H3.3 ChIP-seq on SKNMMMYCN with and without doxycycline and found similar chromatin changes at MYCN target genes (Fig. 6h). Taken together, our data show that there is marked reorganization of the chromatin landscape in ATRX-mutant NB beyond telomeres and centromeres. This is particularly notable at G4 sequences in promoters and enhancers and may impact the ability of ATRX-mutant NB cells to differentiate relative to those with wild-type ATRX.Fig. 6H3.3 deposition is altered at G4 sequences in ATRX-mutant neuroblastomas.a Stack bar plot of H3.3 peaks in the ATRX-deficient (CHLA90, SKNMM), MYCN amplified (SKNBE2, IMR32), and ATRX wild type, MYCN non-amplified (LAN6, SKNFI) cell lines. The H3.3 peaks that overlap with G4 sequences are shown in gray. b Stack bar plot of H3.3 peaks that are constitutive (C) across all three groups (ATRX, MYCN, WT), enriched (E) in ATRX mutant, depleted (D) in ATRX mutant, or overlap (O) between ATRX mutant and MYCN or WT. The H3.3 peaks that overlap with G4 sequences are shown in gray and the number of peaks for the D group are shown. c Piechart of the location of D group H3.3 peaks that have a correlation between the presence or absence of an H3.3 peak and ChromHMM state. Separate pie charts are shown for those that overlap with G4 sequences and those that lack G4 sequences. d Heatmap of the ChromHMM states used in this study with color coding. e Stack bar plot of the distribution of ChromHMM states for each cell line in the non-genic regions that have D group H3.3 peaks that correlate with ChromHMM state and overlap with G4 sequences or lack f G4 sequences. g ChromHMM, WGBS, and ChIP-seq tracks for the DUSP26 gene in SKNMM (ATRX mutant) and SKNBE2 (MYCN amplified) cells. Gene expression is indicated (FPKM) and G4 motif sequences are shown below the ChromHMM tracks. h ChromHMM, WGBS, and ChIP-seq tracks for the SLC3A2 gene in SKNMMMYCN cells with and without doxycycline. Gene expression is indicated (FPKM) and G4 motif sequences are shown below the ChromHMM tracks. DOX doxycycline, ChromHMM chromatin hidden markov modeling, G4 guanine quadruplex structure. + +a Stack bar plot of H3.3 peaks in the ATRX-deficient (CHLA90, SKNMM), MYCN amplified (SKNBE2, IMR32), and ATRX wild type, MYCN non-amplified (LAN6, SKNFI) cell lines. The H3.3 peaks that overlap with G4 sequences are shown in gray. b Stack bar plot of H3.3 peaks that are constitutive (C) across all three groups (ATRX, MYCN, WT), enriched (E) in ATRX mutant, depleted (D) in ATRX mutant, or overlap (O) between ATRX mutant and MYCN or WT. The H3.3 peaks that overlap with G4 sequences are shown in gray and the number of peaks for the D group are shown. c Piechart of the location of D group H3.3 peaks that have a correlation between the presence or absence of an H3.3 peak and ChromHMM state. Separate pie charts are shown for those that overlap with G4 sequences and those that lack G4 sequences. d Heatmap of the ChromHMM states used in this study with color coding. e Stack bar plot of the distribution of ChromHMM states for each cell line in the non-genic regions that have D group H3.3 peaks that correlate with ChromHMM state and overlap with G4 sequences or lack f G4 sequences. g ChromHMM, WGBS, and ChIP-seq tracks for the DUSP26 gene in SKNMM (ATRX mutant) and SKNBE2 (MYCN amplified) cells. Gene expression is indicated (FPKM) and G4 motif sequences are shown below the ChromHMM tracks. h ChromHMM, WGBS, and ChIP-seq tracks for the SLC3A2 gene in SKNMMMYCN cells with and without doxycycline. Gene expression is indicated (FPKM) and G4 motif sequences are shown below the ChromHMM tracks. DOX doxycycline, ChromHMM chromatin hidden markov modeling, G4 guanine quadruplex structure. + +Mutually exclusive mutation profiles are not uncommon in cancer cells; however, many are thought to result from targeting the same oncogenic pathway49. For example, inactivation of the RB1 tumor-suppressor gene is often exclusive of amplification of genes encoding cyclins (e.g., CCNE1) because they target the same pathway. Examples of synthetic lethality caused by oncogene activation and tumor-suppressor mutation are much less common, and few (if any) have been validated in vivo. + +Here, we show that amplification of the MYCN oncogene and inactivation of the ATRX tumor-suppressor gene are mutually exclusive in neuroblastomas from patients of all ages and stages of disease. One discrepant tumor sample may have contained two separate clones, but more-detailed analysis was not possible owing to limited tissue. In mouse models and cell lines, the combination of elevated MYCN expression and ATRX loss led to synthetic lethality. Ectopic MYCN caused metabolic reprogramming, mitochondrial dysfunction, ROS production, and DNA damage in ATRX-mutant neuroblastoma cells. We propose that MYCN amplification and ATRX mutations are incompatible in neuroblastoma, because both lead to DNA-replicative stress10,38,39. Consistent with this model, the synthetic lethality was partially rescued by genes that reduce oxidative stress (CUX2) and pharmacological agents that induce differentiation (retinoic acid) or reduce ROS levels (N-acetyl cysteine). Similarly, pharmacological agents that induced replicative stress through DNA damage exacerbated the synthetic lethality. + +ATRX-mutant neuroblastomas have several unique features, relative to other ATRX-mutant cancers. First, DAXX was not mutated in our cohort, whereas in pancreatic neuroendocrine tumors (PanNETs), ATRX, and DAXX mutations have approximately the same frequency and are mutually exclusive50. Second, ATRX-mutant neuroblastomas have worse outcome, but patients with ATRX/DAXX-mutant PanNETs tumors have prolonged survival50. Third, ATRX mutations in neuroblastomas are often in-frame deletions that remove approximately half of the amino terminus of the protein. In other cancers, the mutations are indels or nonsense mutations50. In the original study describing ATRX mutations in neuroblastoma3,16, there was no difference in outcome or clinical presentation for patients with in-frame deletions versus missense or nonsense mutations. However, that cohort was relatively small and a much larger study of ATRX-mutant neuroblastomas would be required to determine if there is any genotype–phenotype correlation for the type of ATRX mutation. + +Previous studies suggest that amino acids 1–841 of ATRX are sufficient for localization to heterochromatin33. This region is deleted in the ATRXIFDs in most ATRX-mutant neuroblastomas. The ADD domain (amino acids 168–293) binds to histone H3 tails, with preferential binding to H3K9me3 domains that lack H3K4 methylation. HP1 can bind H3K9me3 and recruit ATRX through the HP1-interacting domain (amino acids 586–590), even in the absence of the ADD domain in ATRX. Therefore, in neuroblastomas with in-frame deletions, we propose that they lack the heterochromatin-binding domain and have defects in H3.3-chaperone function that is required for genome maintenance. Our data also suggest that there are defects in H3.3 deposition at promoters and enhancers for genes involved in neuronal differentiation, particularly for those with G4 structures predicted to form R-loops. Indeed, one of the major pathways that was deregulated is involved in retinoic acid response. This is important because retinoic acid induces differentiation of NB cells and is part of the treatment regimen for NB patients. It is possible that there may be a block in retinoic acid response in ATRX-mutant NB cells and this may contribute to the poor outcome for those patients. + +Based on our data, we propose that ATRX mutation contributes to tumorigenesis in two ways. First, defects in H3.3 deposition at telomeres leads to telomere maintenance through ALT. Second, ATRX mutations lead to defects in H3.3 deposition at G4 structures in promoters and enhancers of genes involved in neuronal differentiation including retinoic acid-responsive genes. As a result, gene expression is attenuated and ATRX-mutant neuroblastoma cells continue to proliferate. Although DAXX mutations can also contribute to ALT, they do not alter H3.3 deposition at G4 structures. We propose that both mechanisms (ALT and G4 resolution) are essential for tumorigenesis and that is why we did not detect any DAXX mutations in our cohort. + +Although the inability of ATRX-mutant neuroblastoma cells to resolve G4 structures promotes tumorigenesis by preventing differentiation, it also causes DNA-replicative stress and slows tumor growth. Indeed, it was recently shown that ATRX can play a role in suppressing R-loop (RNA-DNA hybrid) formation51. R-loops can lead to collapse of the DNA replication fork and replicative stress52,53. This may be why ATRX-mutant neuroblastomas are slow growing and indolent. DNA-replicative stress leads to synthetic lethality with ectopic MYCN expression. This genetic synthetic lethality may eventually be exploited in the clinic by reducing the function of ATRX in patients with MYCN-amplified tumors or inducing metabolic (or other changes) caused by MYCN expression in ATRX-mutant tumors. To achieve this ambitious goal in high-risk neuroblastoma, much will need to be learned about the downstream targets of ATRX and MYCN that contribute to this phenotype. + +All uncropped gels and blots are shown in Supplementary Fig. 12. Flow cytometric analysis gating information is shown in Supplementary Fig. 13. + +Patients were eligible for inclusion in the analytic cohort if they enrolled in the COG Neuroblastoma Biology study ANBL00B1 before treatment; had a confirmed diagnosis of neuroblastoma; and had reported outcome data. Informed consent of the patients and/or parents/legal guardians was obtained at the time of enrollment to ANBL00B1. Detailed description of the statistical analysis is provided in the supplemental information. + +The cohort used in this analysis consists of 477 neuroblastoma patients (of which 476 had clinical data and 475 had outcome data available) representing a mixture of risk levels, disease stages, and ages at diagnosis. All are enrolled on COG protocol ANBL12B8 (“Analysis of ATRX Mutations in Neuroblastoma”) and were assayed for ATRX mutations. Roughly 80% of the patients were diagnosed between 2010 and 2012, with the remaining 20% diagnosed between 2001 and 2009. + +The frequency of ATRX mutations, along with Clopper–Pearson exact 95% confidence intervals, were calculated across the entire cohort and in subgroups formed by the combination of age categories (12 yrs at diagnosis) and INSS stage categories (1, 2 A/B or 3, 4 S, and 4). Chi-square tests or Fisher’s exact tests, depending on sample size, were used to test for associations between ATRX mutations and clinical factors. EFS and OS were compared between patients with vs. without ATRX mutations, across the entire cohort and in subgroups based on stage, age, and risk. EFS and OS Cox models were used to assess the prognostic strength of ATRX mutations in the presence of standard clinical risk factors. + +A forward-selection process was used to construct parsimonious Cox models. At the beginning of the forward-selection process, the following candidate variables were considered to be on equal footing: age at diagnosis (< 18 months vs. > 18 months) INSS stage (non-stage 4 vs. stage 4), sex, MYCN status (non-amplified vs. amplified), ALK mutation, ploidy (hyperdiploid vs. diploid), 1p (no LOH vs. LOH), 11q (no LOH vs. LOH), INPC histology (favorable vs. unfavorable), ATRX mutation, mitosis–karyorrhexis index (MKI, low/intermediate vs. high), and grade (totally undifferentiated/poorly differentiated vs. differentiating). Variables were entered into the model one at a time, with the variable chosen for entry being the one that is most significant at each step, per a Wald test. If at any point in the process histology was chosen to enter the model, then MKI, grade, and age at diagnosis were no longer considered for entry, as histology is confounded with these variables. Conversely, if at any point MKI, grade, or age at diagnosis entered the model, then histology was no longer considered for entry. The selection process ended when all remaining candidate variables failed to reach significance at the 0.05 level. + +EFS time was measured as the number of days from diagnosis to date of relapse, disease progression, secondary malignancy, death, or, if no event occurred, date of last contact. OS time was measured as the number of days from diagnosis to date of death, or, if the patient did not die, date of last contact. MYCN amplification status was assessed both by the neuroblastoma reference lab as part of ANBL00B1 enrollment and by St. Jude Children’s Research Hospital. There were some discrepancies between the two, particularly in one patient with an ATRX mutation (an aberration which previous research suggests is mutually exclusive from MYCN amplification), whom the six neuroblastoma reference lab determined to have MYCN amplification, whereas St. Jude detected no amplification. Owing to these discrepancies, both sets of MYCN amplification data were considered individually in this analysis. + +Outcome was not compared in the <18 months old at diagnosis subgroup, as there were no patients with ATRX mutations in that subgroup. Nor was outcome compared in the 18 mo–5yrs old at diagnosis subgroup, as only one patient in that subgroup had an ATRX mutation. + +The median follow-up time for patients who did not have an event was 3.9 years. The median follow-up time for patients who did not die was also 3.9 years. The forward-selection process substituting MYCN amplification data from Dr. Federico’s lab for those from the NBL reference lab yielded the same final EFS and OS Cox models. The effect of ATRX mutations on outcome was also tested among patients with INSS stage 4 disease who were at or above 5 years of age at the time of diagnosis. This subgroup was of interest because of its association with a chronic/indolent course of disease. A notable difference between this subgroup and the subgroups in which significant associations between outcome and ATRX mutation were observed, was the number of patients with high-risk disease; the subgroup of older INSS stage 4 patients was entirely high risk, whereas the other subgroups included a mixture of risk levels. This prompted us to perform a subgroup analysis based on risk group (high vs. low/intermediate). There were five low/intermediate risk patients with ATRX mutations, two of which experienced an event and one subsequently died. The differential effect of ATRX mutations between high- and non-high-risk disease groups was also detected in the Cox model of EFS that included an interaction term for mutation and risk level. + +Mutual exclusivity between ATRX mutations and MYCN amplification in neuroblastoma was done by merging the COG, TARGET, and PCGP data. + +Targeted enrichment was performed using the Seqcap EZchoice Kit (Roche) according to vendor instructions for the Kapa workflow with 500 ng of genomic DNA as the starting input for library construction. + +Whole-exome sequencing was conducted using the SeqCap EZ HGSC VCRome (Roche) according to manufacturers' instructions. + +Purified human NMYC BAC DNA (RP11-1183P10) was labeled with a red-dUTP (AF594, Molecular Probes) and human chromosome 2 control (2q11.2) BAC DNA (RP11-527J8) was labeled with a green-dUTP (AF488, Molecular Probes) both by nick translation. The paraffin slides were deparaffinized with xylene 2 × 10 min each at room temperature (RT), placed in ethyl alcohol 3 × 2 min each at RT, air-dried, placed in 10% buffered formalin for 1 h at RT. The slides were then placed in pepsin (8 mg/ml) in 0.1 n HCL for 3 min, rinsed in dH2O for 5 min at RT. One-hour incubation in Carnoy’s fixative at RT was performed. The labeled probes were combined with human sheared DNA and hybridized to the slides in a solution containing 50% formamide, 10% dextran sulfate, and 2× SSC. The probe and slide were co-denatured at 90 °C for 12 min and incubated overnight at 37 °C on a Thermobrite. In brief, washed in PN and then stained with 4,6-diamidino-2- phenylindole (DAPI) (1 µg/ml). Images were captured using a Plan-Apochromat × 63 objective on a Zeiss Axio Imager.Z2 microscope and GenASIs scanner (ScanView software version 7.2.7). + +ATRX immunohistochemistry was done on formalin-fixed 4-μm-thick paraffin sections using Leica Polymer Refine Detection kit (Leica Microsystems, Wetzlar, Germany) on a Leica Bond system. Epitope retrieval was done by heating and anti-ATRX antibodies polyclonal antibodies (1:600; Sigma-Aldrich) was used and incubated for 15 min3. + +Adrenals and paravertebral sympathetic ganglia from Atrxflox-LSL-MYCN:Dbh-iCre mice at the age of three weeks and 1 year by hematoxylin and eosin (HE) staining and immunohistochemical staining for Ki-67, cleaved caspase 3, tyrosine hydroxylase, and synaptophysin. Formalin-fixed adrenal glands and sympathetic ganglia were processed and embedded in paraffin by standard techniques, sectioned at 4 μm, mounted on positively charged glass slides (Superfrost Plus; Thermo Fisher Scientific, Waltham, MA), and dried at 60 °C for 20 min. Tissue sections were then stained with H&E or subjected to immunohistochemical (IHC) staining. For IHC, tissue sections underwent antigen retrieval in a prediluted Cell Conditioning Solution (CC1; Cat# 950-124; Ventana Medical Systems, Indianapolis, IN) for 32 min on a Discovery Ultra immunostainer (Ventana Medical Systems, Tucson, AZ) in order to detect tyrosine hydroxylase, synaptophysin, or Ki-67 antigen. The primary antibodies used in this study included the following: A rabbit polyclonal antibody raised against tyrosine hydroxylase diluted 1:500 (cat# AB152, EMD Millipore, Billerica, MA); a rabbit polyclonal antibody to synaptophysin diluted 1:400 (cat# ab7837; Abcam, Cambridge, UK); and a rabbit monoclonal antibody to Ki-67 antigen (Clone SP6) diluted 1:50 (cat# RM-9106-S; Thermo Scientific, Fremont, CA). Binding of these primary antibodies was detected using OmniMap anti-Rabbit (#760-4311; Ventana Medical Systems), with DISCOVERY ChromoMap DAB Kit (Ventana Medical Systems) as chromogenic substrate. Stained sections were examined by a pathologist blinded to the experimental group assignments. + +Mouse studies were performed in a strict accordance with the recommendations in the Guide to Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee (IACUC) at St. Jude Children’s Research Hospital. All efforts were made to minimize suffering. All mice were housed in accordance with approved IACUC protocols. Animals were housed on a 12–12 light cycle (light on 6:00, off 18:00) and provided food and water ad libitum. + +Patient-derived xenograft cells were generated and maintained orthotopically in athymic nude (Jackson Laboratories, strain code 007850), NSG (Jackson Laboratories, strain code 005557) or C57BL/6 scid mice (Jackson Laboratories, strain code 001913) as described previously54. Tumor cells were dissociated using Trypsin (10 mg/ml, Sigma Cat # T9935) filtered and then suspended in a Matrigel basement membrane matrix (BD Biosicences catalog number 354234) at a concentration of 2 × 105 cells per 10-μl and placed on ice for injection. The cells were injected in the para-adrenal region of the mice with ultrasound guidance under anesthesia. + +To test the incompatibility between ATRX mutations and MYCN overexpression in mice, we used LSL-MYCN:DbhiCre mice17. These mice express MYCN from the Rosa26 locus when a stop sequence is floxed out by Cre recombinase under the control of Dbh promoter in sympathetic ganglion cells. We breed LSL-MYCN:DbhiCre mice to AtrxFlox mice8 to inactivate Atrx in the same tissues. We also tested the consequences of Atrx inactivation when Th-ALKF1174L allele is added18 (LSL-MYCN: DbhiCre:Th-ALKF1174L). Th-ALKF1174L mice constitutively express ALKF1174L, the most common activating ALK mutation found in human neuroblastoma, under the control of Th promoter. + +Mice were genotyped using the published protocols8,17,18. Mice with the desired genotypes were enrolled in the study at the time of weaning and were followed up by abdominal palpation at least once every 2 weeks starting from age of 6 weeks (Table 2). Enrolled mice that died with no detectable tumors were censored out of the survival study. Mice were killed when the tumor is 20% of body weight or when the mouse has other morbidities, e.g., lethargic, hunched or dehydrated. After euthanasia, mice were opened longitudinally to expose abdomen, thorax, and neck to search for tumors. Tumors were dissected out and were cut into pieces for histological, immunohistochemical, DNA, RNA, and protein preparations.Table 2Numbers of enrolled mice for neuroblastoma genetic mouse model survival study.StrainEnrolledDid not reach the survival ageCensoredReported in the survival analysisMYCN4581#36MYCN-ATRX59132##44MYCN-ALK181017MYCN-ALK-ATRX47128*27#Found dead no tumors.##Hunched and lethargic no tumors, large spleen.*Malocclusion, four mice with fight wounds, paralysis hind limbs, found dead no tumors, large thymus (dyspnea). + +DNA from the tumors and matching germline samples was sequenced and analyzed as described previously54. In brief, 250–500 ng of genomic DNA was input for library construction using Illumina-compatible adapters, and four to six cycles of amplification were performed with KAPA HiFi Hotstart ReadyMix (KAPA Biosystems). The reference human genome assembly NCBI Build 37 was used for mapping samples. WGS mapping, coverage and quality assessment, single-nucleotide variation (SNV)/indel detection, tier annotation for sequence mutations, and identification of LOH have been described previously55. Structural variations were analyzed and annotated using CREST56. Copy number variations were identified by evaluating the difference of read depth for each tumor and its matching normal using CONSERTING57. WGS data were uploaded at EMBL with accession number EGAS00001002528. + +We did automated CUT&RUN (AutoCUT&RUN) for H3.3 as described in detail https://www.protocols.io/view/autocut-run-genome-wide-profiling-of-chromatin-pro-ufeetje. In brief, cells were bound to concanavalin A-coated magnetic beads (Bangs Laboratories), permeabilized with digitonin and then incubated with primary antibody against H3.3 (Abnova, clone 2D7-H1), followed by incubation with rabbit anti-mouse IgG (Abcam, ab46540). Cells were arrayed on a 96-well plate and the rest of the reaction was carried out on a Beckman Biomek FX, including digestion with proteinA-MNase, ligation of adapters and library preparation. Size distributions of prepared libraries were assessed using an Agilent 4200 TapeStation. Libraries were pooled and at equimolar concentrations and 25 × 25 bp paired-end sequencing on an Illumina HiSeq 2500 was performed. + +To find differential binding sites for H3.3, for each CUT&RUN sample, we called peaks twice by MACS2 (paired-end mode) that one using default FDR corrected p value cutoff 0.05 as high confidence peak set and the other using 0.5 as low confidence peak set. Then for each of the three groups, we compiled the reproducible peak set that required a high confidence peak also overlap a low confidence peak from the other samples within the sample group. Then we merged the reproducible peaks from three group to create a reference peak set and counted fragments (fragment size <2000 bp) overlap the reference peaks for each sample. At last, we performed statically tests using Voom58. To check their overlapping with G4 motifs and R-Loops, we downloaded G4 Motifs from supplementary data of ref. 59 and lifted over them from hg18 to hg19, we got 100,780 regions. We downloaded 169,222 “RLFS in genic and proximal regions” from http://rloop.bii.a-star.edu.sg/ for hg19 as predicted R-Loops regions60. + +Genomic DNA was extracted using the DNeasy Kit (QIAGEN #69504) according to the manufacturer’s protocol and quantified with a Nanodrop (Thermoscientific). Whole-genomic bisulfite conversion and sequencing were done by Hudson Alpha (Huntsville, AL). + +Sequencing data were aligned to the hg19 human genome using BSMAP2.7461. We first performed statistical test of differentially methylated locus (DML) between MYCN-amplified and non-MYCN-amplified samples using DMLtest function (smoothing = TRUE) in DSS (Wu et al. 2015), the results were then used to detect differentially methylated regions (DMRs) using CallDMR function in DSS, the p value threshold for calling DMR is 0.01. The minimum length for DMR is 50 bps and the minimum number of CpG sites for DMR is 3. The minimum methylation difference is 0.2. The DMRs that overlap with the promoters (1 kb flanking the TSS) of differentially expressed genes (at least twofold changes and adjust p value ≤ 0.1) were deemed to be associated with gene expressions. + +RNA was extracted from cultured cells and tissues using RNeasy Kit (QIAGEN #74104) or TRIzol (Life Technologies) preparations via a phenol–chloroform extraction or Direct-zol RNA MiniPrep (Zymo Research # R2052) following manufacturers’ protocols. RNA concentration was measured using a NanoDrop (Thermo Fisher scientific, Waltham, MA) and the quality of RNA was determined with a bioanalyzer (Agilent Technologies, Santa Clara, CA). Libraries were prepared using the TruSeq Stranded Total RNA Library Prep Kit (Illumina, San Diego, CA) from 500-ng total RNA. Paired-end deep sequencing was done using HiSeq 2500 sequencers (Illumina, San Diego, CA). + +FASTQ sequences were mapped to the human genome hg19 (GRCh37-lite) and Gene-based FPKM quantification was calculated similar to previously described57 using GENCODE gene annotation (v26lift37). We then used VOOM58 for differential analysis between +DOX vs −DOX. To confirm the results were meaningful, we ranked the genes by log2 fold change and use GSEA62 to check if any MYCN-related gene sets in MSigDB(v6.1) could be found enriched. Indeed, WEI_MYCN_TARGETS_WITH_E_BOX63 gene set were found top enriched for upregulated genes in +DOX (rank 46 out of 11371 gene sets, NES = 3.01, FDR corrected p value < 1e-4). + +Antibodies used in this study to generate ChromMM were previously validated and the validation data and protocols are available through St. Jude Childhood Solid Tumor Network website (https://www.stjude.org/CSTN/)20. + +For MYCN ChIP-Seq, we used rabbit polyclonal antibody (Active Motif Cat# 61185). This antibody was validated by small-scale and large-scale sequencing by active motif as well as by us in St. Jude Children’s Research Hospital. We tried different amounts of the antibodies and chromatin and determined that 20-µg of the antibody and 200-µg of the chromatin per ChIP-Seq reaction show the best results. + +Chromatin in cultured cells was fixed by changing the medium with fresh medium containing 1% formaldehyde (Thermo scientific # 28906). Snap-freezed xenografts were pulverized and hammered to powder in liquid Nitrogen, suspended in PBS and fixed by adding formaldehyde (Thermo scientific # 28906) to final concentration of 1%. Fixation was done for 10 min at room temperature. Cross-linking was stopped by adding glycine to a final concentration of 1.25 M. Cells and xenografts were washed in 1× PBS containing 1× protease inhibitor (Pierce # 78430). + +Chromatin preparation and shearing were done using TruChIP chromatin shearing kit (Covaris #520127) following the manufacturer’s protocol. Chromatin immune-precipitation (ChIP) reactions were performed using iDeal ChIP-seq kit (Diagenode # C01010051) with 10% of chromatin from each reaction was used as an input. Precipitated DNA was de-crosslinked, extracted using MinElute PCR-purification kit (QIAGEN #28006) and quantified using the Quant-iT PicoGreen ds DNA assay (Life Technologies #Q33120). Quantitative polymerase chain reaction (qPCR) was done to assess the efficiency of ChIP reactions as described in the validation protocols. Five to ten nanograms of the input and precipitated DNA were used to prepare the sequencing libraries using the NEBNext ChIP-Seq Library Prep Reagent Set for Illumina with NEBNext High-Fidelity 2× PCR Master Mix following the manufacturer’s instructions (New England Biolabs) with some modifications as described previously;64 after adaptor ligation, 1:1 Ampure cleanup was added twice, no Ampure size-selection step and the extension was done for 45 seconds. Inset-size distribution of the completed libraries was analyzed using a 2100 BioAnalyzer High Sensitivity kit (Agilent Technologies) or Caliper LabChip GX DNA High Sensitivity Reagent Kit (PerkinElmer). ChIP reactions and library preparation for RNA-Pol II, Brd4 H3K9me3 and MYCN were done by Active Motif. Quant-iT PicoGreen ds DNA assay (Life Technologies) and Kapa Library Quantificaiton kit (Kapa Biosystems) or low-pass sequencing on a MiSeq nano kit (Illumina) were used to quantify the libraries. Fifty-cycle single-end sequencing was performed on an Illumlina HiSeq 2500. + +ChIP-Seq analysis was done as described previously64. In brief, ChIP-Seq reads were aligned to human genome hg19 (GRCh37-lite) using BWA software (version 0.7.12-r1039, default parameter) and the duplicated reads were marked using Picard software. We kept only non-duplicated reads for the analysis using samtools (parameter “-q 1 -F 1024” version 1.2). The quality control of the data followed ENCODE criteria. We calculated relative strand correlation value (RSC) and estimated the fragment size under support of R (version 2.14.0) with packages caTools (version 1.17) and bitops (version 1.0-6). We required at least 10 m unique reads for point source factors H3K4me2/3, H3K9/14Ac, H3K27Ac, CTCF, RNA PolII, BRD4) with RSC > 1, 20 m unique reads for broad markers (H3K4me1, H3K9me3, H3K27me3, H3K36me3) and 10 m unique mapped reads for INPUTs with RSC < 1. All samples were manually inspected and the SPP (version 1.1) was used to generate the cross-correlation plot. Then we generated bigwig files from the best fragment size (the smallest fragment size estimated by SPP). Bigwig files were examined using IGV genome browser for clear peaks and low background noise. MACS2 (version 2.0.10.20131216)65 were used to call peaks for MYCN ChIP-Seq data with FDR corrected p value 0.05 (–nomodel –extsize FRAGMENTSIZE, where FRAGMENTSIZE was estimated by SPP as described above). Genes promoter (TSS ±2 kb) with MYCN peaks called were considering have MYCN-binding site. Enrichr66 were used for pathway analysis. + +Super-enhancers and core regulator circuit analysis have been performed as described previously64. We called auto-regulators by CRCMapper67 based on super-enhancers called by ROSE68. We provided two sets of results that one only used H3K27ac data, while the other one excluded H3K27ac peaks overlapping promoter (defined by H3K4me3 peaks) before calling super-enhancer by ROSE. + +ChromHMM models were generated for y as described before54. In brief, Non-duplicated aligned reads were extended by fragment size, and ChromHMM (version 1.10, with ‘-colfields 0,1,2,5 –center’ for BinarizeBed) was used for chromatin state modeling. To choose the state number, we first modeled all samples together from seven states to 33 states and selected the model with 18 states upon manual inspection. For better visualization of the dynamics of HMM state across stages, we normalized color intensity by the maximum total percentage of a state covering a gene and flanking region. We reduced the interval for an HMM state to half bar and the intensity to half the normalized amount if it did not rank in the top two HMM state for a gene. As HMM states could be assigned by multiple genes, the maximum total percentage across genes was used for normalization. + +We applied the Hidden Markov Model trained for 18 states from previously four types of solid tumors to the data from this manuscript. Expanded chromHMM state 13 were defined as proportion of promoter downstream (TSS ~ TSS + 2 kb) in SKNMMMYCN + Dox > twofold of SKNMMMYCN − DOX and ≥ 20% in SKNMMMYCN + DOX. + +The medium was removed and the cells were washed twice with PBS then the cells were scrapped in 1-ml PBS and centrifuged at 2000 × g for 5 min at 4 °C to pellet the cells. PBS was removed and the cells were flash-freezed in dry ice and stored at −80 °C until use. Cell pellets were thawed on ice and lysed for 15 min in radioimmunoprecipitation assay buffer (Sigma Cat. #R0278) supplemented with 1 × Halt protease inhibitor (Thermo Fischer, Cat. #7834) and 5 mM ethylenediaminetetraacetic acid (EDTA). Lysate were then centrifuged for 15 min at 15,000 rpm at 4 °C. Protein concentration in the cell lysates was measured using BCA protein assay kit (Thermo Fischer, Cat. #232225). Equal amounts of total protein were resolved on 4–15% sodium dodecyl sulfate polyacrylamide gel (Biorad Cat#4561086) and transferred to a nitrocellulose membrane by electrophoresis at 30 volts overnight at 4 °C. Non-specific binding was blocked by incubating the membrane with Odyssey Blocking buffer, PBS (LiCore Cat# 927-40000) for 1 h at room temperature. Then the membrane was incubated with the primary antibody overnight at 4 °C. The following antibodies were used: rabbit anti-ATRX (1:1000, Abcam, ab97508), mouse anti-MYCN (1:1000, Santa cruz Biotechnology, sc-53993), rabbit anti-GAPDH (1:2500, Abcam ab-9485), rabbit anti-DAXX (1:1000, Santa cruz Biotechnology, sc-7152) and mouse anti-β-actin (1:5000, Sigma, A1978). Membranes were washed three times in PBS with 0.1% Tween-20 at room temperature then were incubated in the secondary antibodies: IRDye 800CW goat anti-mouse (1:5000, LiCore, 925-68070) or IRDye 680CW goat anti-rabbit (1:5000, LiCore, 925-32211) for 30 min at room temperature. The membranes were then washed three times in phosphate-buffered saline with Tween; 5 min each at room temperatures and one time in PBS. The membranes were then scanned by Odyssey CLX infrared gel imaging system (Li-Cor Biosciences, Lincoln, NE). Band intensities were quantified using ImageStudio Lite software (Li-Cor Biosciences, Lincoln, NE). + +After a 5-hour colcemid incubation, cells from both samples were harvested using routine cytogenetic methods. A commercially prepared directly fluorescein isothiocyanate-labeled PNA telomere probe (DAKO, Cat. # 5327) was used for this analysis. DAKO protocols were followed for the pre-treatment and hybridization steps. The slides were then stained with DAPI. In total, 255 interphase nuclei from each sample were individually analyzed for the number of telomeres/cell, fluorescence/cell, and fluorescence/telomere. A fixed exposure time is used for all cells. + +Telomere qPCR was done as described previously (Cheung et al.3). DNA was extracted using DNAeasy Kit (Qiagen, Cat. # 69504). qPCR was done on 10-ng DNA template using SYBR green Select master mix (Thermo Fisher, Cat. #4472908). Two sets of primer pairs were used in separate reactions were used to amplify the telomeric sequence and a common locus RPLP0, which was used as an internal control. The reactions were done in duplicates and the average delta C(t) was calculated for every sample by subtracting the average C(t) values of the telomeric reaction from those of the internal control reactions. The primers used are + +Telomere F: 5′-CGGTTTGTTTGGGTTTGGGTTTGGGTTTGGGTTTGGGTT-3′ + +Telomere R: 5′-GGCTTGCCTTACCCTTACCCTTACCCTTACCCTTACCC-3′ + +Control, locus: + +RPL0 F: 5′-CAGCAAGTGGGAAGGTGTAATCC-3′ + +RPL0 R: 5′-CCCATTCTATCATCAACGGGTACAA-3′ + +Eight ATRX-specific shRNA lentiviral vectors were purchased from Dharmacon as bacterial glycerol stocks (Cat. # V2LHS-092920 (#9), V2LHS-202920 (#20), V2LHS-22887 (#22), V3LHS-375686 (#37), V2THS-22887 (#87), V2THS-93289 (#30), V2THS-255090 (#13), and V2THS-201895 (#95)). Lentiviruses were made in HEK293T cells by co-transfecting the viral vectors with three packaging plasmids. The supernatants containing the viral particles after 48 and 72 h were collected, filtered, concentrated using ultracentrifugation, and titrated in HeLa cells using the GFP or RFP reporters in the vector. In addition, two non-silencing shRNA lentiviral particles were used as controls (Cat. # RHS4743 and RHS4346). Cells were transduced with equal number of viral particles and transduced cells were selected with flow cytometry or pouromycin after 48 h post transduction. + +Cells were trypsinized and harvested, pelleted and suspended in 1-ml propidium iodide solution (0.05 mg/ml propidium iodide (Sigma, Cat. # P4864), 0.1% sodium citrate, 0.1% Triton X-100). The cells were then treated with 10-ml 0.2 mg/ml of ribonuclease A (Colbiochem, Cat. #556746) for 30 min at room temperature, filtered through 40-μm nylon mesh, and analyzed by flow cytometry for the DNA content. + +Cells were transduced with a lentivirus expressing ATRX shRNA #9, ATRX shRNA #20, or scrambled control shRNA at multiplicity of infection 5 for 24 h. Then the cells were harvested, suspended into single cells in a pre-warmed medium, counted and seeded at 300 cells/well in six-well plates. The plates were shaken gently every 1–2 min for 10 min. Fresh medium was added every 4 days. After 2–3 weeks when colonies were visible, cells were washed, fixed in 1% paraformaldehyde for 30 min at room temperature and stained with 0.1% crystal violet. + +Guide RNAs (gRNAs) were designed targeting an early and conserved exon between isoforms of the hATRX and hDAXX genes. Bioinformatics analysis was performed to identify gRNAs with at least 3 bp of mismatch to other sites in the human genome. gRNAs were further validated in human K562 cells for cutting efficiency using targeted deep sequencing. + +Two validated hATRX (g2 and g5) and one hDAXX (g11) gRNAs were used in this study. The sequences of gRNAs are: + +hATRX.g2: 5′-CTGCACTGCTTGTGGACAACNGG-3′ + +hATRX.g5: 5′-CAATGAAGGGTGTCTATAAANGG-3′ + +hDAXX.g11: 5′-GGTTCTTCTGACAGTAACGANGG-3′ + +Cells were transiently transfected with PC526-Cas-9 plasmid (to express Cas-9; Addgene # 43945), either ZZ52.hATRX.g2 or ZZ52.hATRX.g5 (to express hATRX gRNA; the backbone plasmid Addgene # 43860) and pUB-GFP (to express GFP; Addgene # 11155) in a ratio of 7:7:1, respectively using Mirus 2020 (Mirus Biotechnology Cat. # 2020). After 48 h, the cells were sorted for GFP-positive cells. Half of the sorted cells were pelleted immediately and stored at −80 °C and the remaining cells were seeded in six-well plates and harvested at different time points. To test for the NHEJ efficiency after CRISPR targeting in IMR32 cells, we transfected the cells with the same mix of plasmids but added a control plasmid expressing gRNA targeting the hAAVS1 locus (Addgene # 41818) in a ratio of 4:4:6:1 for hAAVS1.gRNA: ATRX.gRNA: Cas-9: GFP expressing-plasmids, respectively. DNA was extracted from the cell pellets using DNAeasy kit (Qiagen, Cat. # 69504) and PCR was done to amplify the targeted region using primers to which MiSeq adaptors were annexed and the gel-purified PCR products were subjected to deep sequencing. + +The primers used are (adaptor sequences are underlines) + +hATRX F: TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTGGAATTTGCCAAGGTTGTCATGTGC + +hATRX R: GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGCTTTCTTCAAGACTGTGCCCTCAAAGGCC + +hAAVS1 F: TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGTGTCCCCGAGCTGGGACCA + +hAAVS1 R: GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGTAGGAAGGAGGAGGCCTAAGGATGGG + +The percentage of indels was determined using CRISPResso software as described before69. + +Alkaline comet assay was done using CometAssay Reagent Kit (Trivegen, Cat. # 4252-40-K) following the manufacturer’s protocol. In brief, cells were harvested and suspended in PBS buffer at a concentration of 1 × 105 cells/ml. Cells were combined with LMAgarose in a ratio 1:10 at 37 °C and 50-μl of the mix was spread over the CometSlide. The doxycycline-treated and untreated samples were loaded onto two different wells on the same slide. Slides were left at 4 °C for 30 min in the dark to allow agarose to polymerize. The slides were then incubated in a freshly made Alkaline Unwinding Solution (200 mM NaOH, 1 mM EDTA, pH > 13) overnight at 4 °C in the dark. Electrophoresis was done at 4 °C for 30 min at 21 volts in the Alkaline Electrophoresis Solution (200 mM NaOH, 1 mM EDTA, pH > 13). The slides were washed twice in dH2O, dried at 37 °C and stored in the dark. To analyze the samples, nuclei suspended in the agarose gel were stained by 100-μl of SYBR Gold (Thermo Fischer, Cat. #S 11494) diluted 1:30,000 in TE buffer (10 mM Tris-HCL pH 7.5, 1 mM EDTA). Then the slides were rinsed in water and dried completely at 37 °C. Slides were examined by a fluorescence microscopy for excitation/ emission: 496/522 and 10–15 images were taken per sample. Percentage of DNA in the tail was quantified in the images using ImagJ software by measuring fluorescence intensity in the tail/total fluorescence × 100. Mann–Whitney non-parametric statistical test was used to compare the scores in the samples with or without doxycycline. + +Cells were treated with colcemid for 5 h at 37 °C then the cells were harvested for cytogenetic analysis. SKY analysis was done using a commercially available probe from spectral imaging (Carlsbad, CA) following the manufacturer’s protocol. Fifty metaphase cells from every sample were scored for the presence of DNA fragmentations. Statistical analysis was done using Fischer exact test. + +The cells were seeded in 96-well plates at a density of 2500 cells per well over night. Next day (day 0), the medium was changed with a fresh medium with or without 2 μg/ml doxycycline. After 24 h of doxycycline treatment (day 1), the medium was changed again with fresh medium containing serially diluted hydroxyurea with or without doxycycline. At day 4, the cell viability was assessed using the CellTitre-Glo cell viability assay kit (Promega, Cat. # G7572) using PHERAstar FSX (BMG Labtech, San Diego, California). + +To make the doxycycline-inducible cell lines, we used pLenti4/TO/V5 DEST system (Invitrogene, Cat. # V480-20) following the manufacturer’s protocol. In brief, cells were transduced with the pLenti4/TO lentivirus to constitutively express the Tet repressor and the cells were selected with 10-μg/ml blasticidin-HCL (Thermo Fisher, R21001) in the medium. The cells were then transduced with the V5 DEST-MYCN or V5 DEST-control lentivirus and were selected with 10-μg/ml zeocin (Thermo Fisher, R25001) in the medium. The V5 DEST-MYCN virus was a gift from the Freeman lab at St. Jude Children’s Research Hospital. The cells were maintained in a selective medium containing 5-μg/ml blasticidin-HCL and 5-μg/ml zeocin. We also cloned MYCN into another dox-inducible system using Lenti-X Tet-One (Takara Cat# 631847) following manufacturer protocol. We confirmed the sequence of the construct by sanger sequencing. Cells were transduced with Lenti-X Tet-One lentiviruses containing MYCN or with the empty vector. Selection for the cells containing the construct was done using puromycin and the cells were maintained in puromycin-containing media. Induction of MYCN expression was done by adding doxycycline to the medium to a final concentration of 1–2 μg/ml. + +The cells were plated in six-well plates at a density of 50,000 cells per well over night. Then the medium over the cells was changed with a fresh medium with or without 1–2-μg/ml doxycycline. Three wells from each of the cell lines were harvested every day and counted for the total number of cells per well using a hemocytometer. + +Live cell time-lapse imaging experiments for SKNMMMYCN in the presence or absence of doxycycline were performed using a Nikon C2 confocal configured on a TE2000E2 microscope with a 20 × 0.8 NA Plan Apo lens (Nikon Instruments Inc., Melville, NY). A 640 nm DPSS lasers was used to simultaneously excite Annexin V-Alexa Fluor 647 and generate the DIC images. During imaging experiments, the cells were maintained at 37 °C and 5% CO2, and images were collected every 30 min for 67 h. Images were acquired and processed using NIS Elements software. + +SKNMMMYCN and SKNMMCONT cells were labeled with luciferase-YFP (yellow fluorescent protein) using a lentivirus. Two mixed cell pools were prepared in a 1:1 ratio: labeled SKNMMMYCN cells/unlabeled SKNMMCONT and labeled SKNMMCONT /unlabeled SKNMMMYCN. Cell mixes were injected in the para-adrenal region of the mice (four million cells/mouse) with ultrasound guidance under anesthesia as explained before. In this experiment, we used NOD.CB17-Prkdcscid/J female mice (Jackson laboratories, Stock # 001303). The mice were followed up for the tumor formation using ultrasound scanning every 2 weeks, xenogen signal every week and palpation every week. Mice were enrolled in the study when the tumor was at least 14 mm3 in size using the ultrasound scanning. Enrolled mice were randomly assigned to receive either doxycycline (2 mg/ml) in the drinking water or continue on regular water. + +The labeled/unlabeled cell mixes were also maintained in culture and the ratio of YFP-expressing cells in each cell mix was determined weekly using flow cytometric analysis. + +Lactic acid was measured using Lactate Assay Kit (Sigma Cat. # MAK064) following the manufacturer’s protocol. The absorbance at 570 nm (A570) was measured for a colorimetric detection of the lactate using the microplate reader SpectraMax-M5 (Molecular Devices, Sunnyvale, California). + +The analytical procedure applied was based on the one previously published28. In brief, cells were cultured in glutamine-free medium (10% fetal bovine serum (FBS), penicillin/streptomycin) supplemented with 4 mM of standard glutamine or [U-13C]-glutamine (CLM-1822-H); Cambridge Isotope Laboratories, Inc. for 5 h. Next, cells were trypsinized, washed with ice-cold saline solution, snap-frozen, and stored in liquid nitrogen for further processing. Extraction of hydrophilic metabolites was performed using two volumes of methanol/water solution (2:1 v/v, high-performance liquid chromatography (HPLC)-grade, spiked with 5 nM of norvaline as internal standard), homogenization and mixing with one volume of chloroform (HPLC-grade). Derivatization with 2% methoxylamine-HCl solution and GC-MS-grade MTBSTFA + 1% t-BDMCS reagent was performed as previously described. Samples were analyzed in split-less mode using Agilent 7890B GC system equipped with a 5977B MSD, Agilent VF5ms, + 1 m EZ, 60 m, 0.25, 0.25 μM column, helium carrier gas and Mass Hunter software package. Chromatographic gradient conditions: separation time 36 min; start at 80 °C, 1 min; ramp to 250 °C at 7oC/min; ramp to 300 °C at 50 °C/min and hold for 9 min. MSD settings: source 150 °C; quadrupole 150 °C; interface 300 °C; injector 250 °C; EI source 70 eV, EI pressure 1.8 × 10−5 torr. Signals were acquired using SIM mode conditions as described before (5) as well as using scanning mode (full scan range 50–500 m/z at ~ 2 scans/s (N = 3)). Enrichment of C13-labeled isotopomers of analyzed compounds was corrected by natural abundance using “standard glutamine” medium-incubated sample as reference. + +MYCN-induced and un-induced SKNMMMYCN cells were grown on either glutamine-free or glucose free completed medium supplemented with [U-13C]-glutamine (CLM-1822-H) or [U-13C] glucose (CLM-1396); Cambridge Isotope Laboratories, Inc. for 18 h. Metabolites were extracted from the cells following the protocol provided by Human Metabolome Technology Inc, (Boston, MA). Samples were filtered through 5-kDa cutoff filter (ULTRAFREE-MC-PLHCC, Human Metabolome Technologies, Yamagata, Japan) and sent to Human Metabolome Technologies Inc. for analysis. We also sent snap-freezed medium from the same samples for metabolomic analysis. + +Metabolites were measured in both Cation and Anion modes using Agilent Capillary electrophoresis–time of flight mass spectrometry (CE-TOFMS) system (Agilent Technologies Inc.) and Capillary: Fused silica capillary i.d. 50 μm Å ~ 80 cm with the following conditions: + +Cationic metabolites (cation mode): + +Run buffer: cation buffer solution (p/n: H3301-1001) + +Rinse buffer: cation buffer solution (p/n: H3301-1001) + +Sample injection: pressure injection 50 mbar, 10 sec + +CE voltage: positive, 27 kV + +MS ionization: ESI positive + +MS capillary voltage: 4000 V + +MS scan range: m/z 50–1000 + +Sheath liquid: HMT sheath liquid (p/n: H3301-1020) + +Anionic metabolites (anion mode): + +Run buffer: anion buffer solution (p/n: H3302-1021) + +Rinse buffer: anion buffer solution (p/n: H3302-1021) + +Sample injection: pressure injection 50 mbar, 25 sec + +CE voltage: positive, 30 kV + +MS ionization: ESI negative + +MS capillary voltage: 3500 V + +MS scan range: m/z 50–1000 + +Sheath liquid: HMT sheath liquid (p/n: H3301-1020) + +Data processing and analysis + +Data processing and analysis were done by Human Metabolome Technologies. In brief, MasterHands ver. 2.17.1.11 software was used to extract peaks detected in CE-TOFMS analysis in order to obtain m/z, migration time (MT), and peak area. Relative peak area was calculated as: + +Relative peak area = metabolite peak area/(internal standard peak area × sample amount) + +Signal-noise ratio (S/N) = 3 was set as a threshold for peak detection limit. + +Human Metabolome Technologies target library was used to identify putative metabolites and their isotopic ions on the basis of m/z and MT with tolerance ± 0.5 min in MT. + +Absolute quantification was performed in total amount of each detected metabolite. All the metabolite concentrations were calculated by normalizing the peak area of each metabolite with respect to the area of the internal standard and by using standard curves, which were obtained by single-point (100 μM) calibrations. + +Five milliliters of the medium were centrifuged at 500 × g for 5 min to remove suspended cells and the pH was measured using a pH meter (Mettler Toledo, Colombus, OH). + +MitoTracker Green FM (Thermofisher # M7514) or Tetramethylrhodamine ethyl ester perchlorate (TMRE, Thermofisher # T669) was added directly to the medium to a final concentration of 200 nM or 50 nM, respectively. After 20 min incubation at 37 °C, the medium was removed and the cells were washed twice in 1 × PBS and harvested. Fluorescent intensity was measured in at least 20,000 cells as suggested by the manufacturer. Data are presented as a geomean ±95% confidence interval. The comparison between the fluorescence intensity in cells with or without doxycycline was done using the two-tailed t test. + +Mitochondrial function and glycolytic flux were determined using Seahorse XFe 96 extracellular flux analyzer (Seahorse, Agilent) as oxygen consumption rate and extracellular acidification rate using “Mito-stress test” or “Glycolysis stress test” methods, respectively. Cells harboring inducible N-Myc overexpression vector were treated ±DOX for 2, 3, 4, or 5 days. Next, cells were plated onto 96-well plate using several strategies to verify experimental outcome. All plating approaches yielded consistently very similar results. In particular, cells were plated at two different concentrations (4 × 104 or 2.5 × 104 cells/well) in multiple wells (i) allowing to attach for 6 h or (ii) overnight to fibronectin-coated wells, or (iii) were attached readily after re-plating using Cell Tak cell adhesive (Corning). FBS-free Seahorse XF RPMI medium (without Phenol Red, Agilent) was supplemented with 2 mM glutamine, 1 mM sodium pyruvate and 2 g/L glucose (for “Mito-stress test”) or with 2 mM glutamine (for “Glycolysis stress test”). Measurements were performed at basal state and following serial injections of: oligomycin (1 μM), FCCP (0.5 μM) and rotenone/antymycin A (0.5 μM/1 μM) for “Mito-stress test” or of glucose (10 mM), oligomycin (1 μM), and 2-deoxy-glucose (20 mM) for “Glycolysis stress test”. + +The samples were fixed with 2.5% glutaraldehyde, 2% paraformaldehyde in 0.1 M sodium cacodylate buffer pH 7.4, and post fixed in 2% osmium tetroxide in 0.1 M cacodylate buffer with 0.3% potassium ferrocyanide for 1.5 h. After rinsing in the same buffer, the tissue was dehydrated through a series of graded ethanol to propylene oxide, infiltrated, and embedded in epoxy resin and polymerized at 70 °C overnight. Semithin sections (0.5 micron) were stained with toluidine blue for light microscope examination. Ultrathin sections (80 nm) were cut and imaged using an FEI Tecnai 200 Kv FEG Electron Microscope with an ATM XR41 Digital Camera. To score mitochondrial damage, 50 images from every sample were selected in which the nucleus and at least five mitochondria were visible at 5 K magnification. Images of these profiles were taken at 29 K with some portion of the nucleus visible, printed and all mitochondria profiles were scored blindly. Mitochondria were scored 0 = normal (aligned cristi and normal outer membrane), 1 = abnormal (swollen and disorganized cristi or disrupted outer membrane) or 2 = very abnormal (both swollen and disorganized cristi and severe outer membrane disruption). + +Cells were harvested and incubated on poly-l-lysin-treated (Sigma, Cat. # 8920) slides for 30 min at 37 °C. The cells were fixed in 4% paraformaldehyde in PBS at 4 °C overnight, blocked by normal donkey serum for 1 h at room temperature and incubated with mouse anti Phospho Ser139-histone H2AX antibody (Millipore, Cat. # 5-636) 1:10,000 overnight at 4 °C. The slides were then washed three times in PBS and incubated with donkey anti-mouse Cy3 IgG antibodies (Millipore, Cat. # AP192C) 1:500 for 1 h at room temperature. The slides were washed three times in PBS, counter-stained with DAPI for 10 min, washed three times in PBS and mounted with Prolong Gold Antifade Mountant (Thermo Fischer, Cat. # P36930). + +ROS was assayed using 2',7'-dichlorofluorescin diacetate (DCFDA) cellular ROS detection assay kit (Abcam # ab113851) or CellRox Green (ThermoFischer # C10444) following manufacturers’ protocol. For DCFDA cellular ROS detection assay kit, cells were plated overnight in 96-well plates (25,000 cells per well) in their complete medium with or without doxycycline. The fluorescence excitation/emission 485/535 was measured using the microplate reader SpectraMax-M5 (Molecular Devices, Sunnyvale, California). For CellRox Green, cells grow in 10-cm plates. CellRox Green was added to the medium 1:500 and incubated for 30 min then cells were harvested and analyzed using flow cytometry. + +The cells were plated in 96-well plates overnight at a density of 6000 cells per well in their complete medium with or without doxycycline. Glutathione concentration per well was measured using GSH-Glo Glutathione Assay Kit (Promega # V6911) according to the manufacturer’s protocol. The luminescence was measured using PHERAstar FSX (BMG Labtech, San Diego, California). + +Cell screening was done on MYCN-induced and un-induced SKNMMMYCN cells as previously described54 with the following modifications: to determine optimal cell plating density, SKNMMMYCN cells were plated on a flat-bottomed, white 96-well plate (Corning Cat#3917), at eight different cell densities ranging from 0 cells/well to 10,000 cells/well; 12 wells per cell intensity per plate. Twenty-four hours after plating, medium was changed with or without doxycycline. Medium was changed after that with or without doxycycline every 2 or 4 days. One plate from every testing condition was read every day for 11 days using CellTiter-Glo (‘CG’, Promega Cat#G7573) on PHERAstar FSX (BMG Labtech, San Diego, California). Cell density of 1250 cells per well and medium change once every 4 days was determined to have good signal to noise separation while maintaining logarithmic growth in both induced and un-induced cells. We tested different dimethyl sulfoxide (DMSO) concentrations for the selected cell density and confirmed minimal cell death with up to 0.197% DMSO. Positive control compound selection was performed with eight positive control compound candidates (doxorubicin HLC, staurosporine, etoposide, SN-38, bortezomib, cyclohexamide, panobinostat, and TAKA123) arrayed in single-point concentration and 1:3 dilution series. Based on successful cell killing Staurosporine was chosen as positive control for cell screenings. Assay validation in 384-well plates was performed with staurosporine. + +A schematic representation of the dose–response curve experiment is illustrated (Supplementary Data 14). Cells were plated in 96-well plates (Corning Cat#3917) at 1250 cells/100 μL; 12 cell plates per compound plate. Approximately 24 h following plating, medium was changed with fresh medium with or without doxycycline; six-cell plates per a compound plate each. Two days after MYCN induction, medium was changed again with fresh medium with or without doxycycline and cells were drugged with the compound plates and a positive control plates using a Biomek FX (Beckman Coulter) liquid handler equipped with a pin tool. The pin tool transferred 112 nL compound stock, resulting in 890-fold compound dilution. After 3 days of drugging, half of the plates were read. On day 4 post drugging, medium was changed with fresh medium again with or without doxycycline and fresh drugs were added to the medium using the Biomek FX. After 4 days of the second drugging, the other half of the plates were read. The compound plate contained compounds dissolved in DMSO arrayed in a 1:3-fold dilution series in columns 1–10 with eight compounds per plate (target concentration 10 mM or 1 mM). The positive control plate was empty from columns 1–10; in columns 11–12, it contained DMSO and the positive control. Two biological replicates were performed for each 8 days experiment, with three technical replicates per biological replicate. + +Normalization, fitting, and visualization of dose–response experiments were performed using custom code written in the R programing language (version 3.3.2, R Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.). + +Dose–response curve data analysis for cytotoxic compounds was done as previously described before54. Raw CellTiter-Glo (CTG) luminescence signals (RLU) were first log2 transformed, then normalized to the mean of the positive and negative controls on each plate. + +In order to detect drugs that reduced MYCN-induced cell death, traditional dose–response analysis was not appropriate because protective drugs would often protect at low concentrations, but eventually induce cytotoxicity at high concentration. To detect a protective effect, we needed to quantify the increase in CTG signal at low concentration before any decrease in signal at high concentration. To do so, the CTG RLUs from all wells were first log2 transformed for each plate, the value for each well was then normalized by subtracting out the mean log2 RLU for negative control wells. The normalized activity of each compound was then fit as a function of the log10 drug concentration using a smooth spline (smooth.spline function in R with default parameters). For each fit, the area under the curve (AUC) for the portion of the spline with RLU above the RLU of the lowest drug concentration tested was calculated. Higher values for this AUC metric indicate greater protection against loss of cell viability as measured by the CTG assay. + +SKNMMMYCN cells were transduced by a lentiviral vector (pLenti-C-mGFP-P2A-Puro) from Origene expressing either CUX2 (Cat# RC222063L4V) or GFP control construct (Cat# PS100093V). GFP-positive cells were sorted after 48 h and maintained in a selective media containing 1.5-μg puromycin per ml. + +We have made every effort to ensure reproducibility of our data by, when possible, repeating the experiments using independent samples, including positive and negative controls and using multiple approaches to confirm our observations. We stated the sample number for each experiment and how many times each experiment was repeated in the figure legends. However, the following experiments were done once: + +Fig. 1f–i: This was a unique case in which there was a discrepancy between COG FISH report and next-generation sequencing report. This sample was carefully studied as described in the results section. + +Fig. 3b: This is the first and only O-PDX developed from ATRX-mutant neuroblastoma, as explained in the text. + +Fig. 5h: This experiment was done to confirm the high-throughput screening and was done in a dose response. + +Fig. 5i: This experiment was done once. However, the ability of RA to reduce MYCN level is well-established in literature70. + +Supplementary Fig. 1B–S1E: Was done to all the clinical samples as mentioned in the material and methods. + +Supplementary Fig. 4H. This experiment was done once. However, this experiment is another evidence for the increase in replicative stress when ATRX is inactivated and MYCN is ectopically expressed in different cells as shown in Fig. 5. + +Supplementary Fig. 7C: This experiment was done once as another readout for increase ROS shown in Supplementary Fig. 7D. + +Further information on research design is available in the Nature Research Reporting Summary linked to this article. + + +Supplementary Information +Peer Review File +Description of Additional Supplementary Files +Supplementary Data 1 +Supplementary Data 2 +Supplementary Data 3 +Supplementary Data 4 +Supplementary Data 5 +Supplementary Data 6 +Supplementary Data 7 +Supplementary Data 8 +Supplementary Data 9 +Supplementary Data 10 +Supplementary Data 11 +Reporting Summary + + +Supplementary Information + +Peer Review File + +Description of Additional Supplementary Files + +Supplementary Data 1 + +Supplementary Data 2 + +Supplementary Data 3 + +Supplementary Data 4 + +Supplementary Data 5 + +Supplementary Data 6 + +Supplementary Data 7 + +Supplementary Data 8 + +Supplementary Data 9 + +Supplementary Data 10 + +Supplementary Data 11 + +Reporting Summary + +## Table 1 Distribution of ATRX mutations in neuroblastoma. + +| Classificationa | Grouping | N(%)b | ATRX mutation (%)c | p-value | +| INSS stage [rowspan=2] | Non-stage 4 | 312 (65.8) | 5 (1.6) | 0.0002 [rowspan=2] | +| Stage 4 | 162 (34.2) | 14 (8.6) | +| Risk Group [rowspan=2] | Low/intermediate | 375 (79.6) | 5 (1.3) | 4.4e-7 [rowspan=2] | +| High | 96 (20.4) | 14 (14.6) | +| Sex [rowspan=2] | Female | 253 (53.4) | 12 (4.7) | 0.3830 [rowspan=2] | +| Male | 221 (46.6) | 7 (3.2) | +| MYCN (FISH) [rowspan=2] | Not amplified | 431 (91.7) | 18 (4.2) | 1.0000 [rowspan=2] | +| Amplified | 39 (8.3) | 1 (2.6) | +| MYCN (NGS) [rowspan=2] | Not amplified | 447 (94.1) | 19 (4.3) | 0.6180 [rowspan=2] | +| Amplified | 28 (5.9) | 0 (0) | +| ALK [rowspan=2] | No mutation | 454 (95.6) | 17 (3.7) | 0.2026 [rowspan=2] | +| Mutation | 21 (4.4) | 2 (9.5) | +| Ploidy [rowspan=2] | Hyperdiploid | 277 (60.7) | 10 (3.6) | 0.6454 [rowspan=2] | +| Diploid | 179 (39.3) | 8 (4.5) | +| 11q LOH [rowspan=2] | No | 301 (91.8) | 5 (1.7) | 4.9e-4 [rowspan=2] | +| Yes | 27 (8.2) | 5 (18.5) | +| 1p LOH [rowspan=2] | No | 290 (88.4) | 9 (3.1) | 1.0000 [rowspan=2] | +| Yes | 38 (11.6) | 1 (2.6) | +| Histology [rowspan=2] | Favorable | 314 (69.8) | 1 (0.3) | 9.3e-9 [rowspan=2] | +| Unfavorable | 136 (30.2) | 17 (12.5) | +| Age at diagnosis [rowspan=2] | <18 months | 213 (44.9) | 0 (0) | 0.00001 [rowspan=2] | +| ≥18 months | 261 (55.1) | 19 (7.3) | +| Age at diagnosis [rowspan=4] | <18 months | 213 (44.9) | 0 (0) | 9.0e-6 [rowspan=4] | +| 18 mo-5 yrs | 76 (16.0) | 1 (1.3) | +| 5–12 yrs | 152 (32.1) | 15 (9.9) | +| >12 years | 33 (7.0) | 3 (9.1) | +| Grade [rowspan=2] | Differentiating | 43 (13.2) | 3 (7.0) | 0.7168 [rowspan=2] | +| Poorly/undifferentiated | 283 (86.8) | 15 (5.3) | +| MKI [rowspan=2] | Low/intermediate | 298 (90.3) | 18 (6.0) | 0.2356 [rowspan=2] | +| High | 32 (9.7) | 0 (0) | +| Race [rowspan=2] | Black | 55 (12.7) | 5 (9.1) | 0.0348 [rowspan=2] | +| other | 378 (87.3) | 12 (3.2) | + +aINSS international neuroblastoma staging system, MYCN amplification was assessed by FISH at each COG participating center and by next-generation sequencing (NGS) in our sequence analysis, MKI mitosis–karyorrhexis index. bN number of patients with percentage in parentheses. cThere were 19 ATRX mutations identified but for some tumors, the classifier was not applied so the total number of samples and the total number of ATRX-mutant samples is lower. + +## Table 2 Numbers of enrolled mice for neuroblastoma genetic mouse model survival study. + +| Strain | Enrolled | Did not reach the survival age | Censored | Reported in the survival analysis | +| MYCN | 45 | 8 | 1# | 36 | +| MYCN-ATRX | 59 | 13 | 2## | 44 | +| MYCN-ALK | 18 | 1 | 0 | 17 | +| MYCN-ALK-ATRX | 47 | 12 | 8* | 27 | + +#Found dead no tumors. ##Hunched and lethargic no tumors, large spleen. *Malocclusion, four mice with fight wounds, paralysis hind limbs, found dead no tumors, large thymus (dyspnea). \ No newline at end of file diff --git a/references_cache/PMID_32291317.md b/references_cache/PMID_32291317.md new file mode 100644 index 00000000000..75143031683 --- /dev/null +++ b/references_cache/PMID_32291317.md @@ -0,0 +1,222 @@ +--- +reference_id: PMID:32291317 +extractor_version: 1 +xml_extraction_version: 1 +title: Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma. +authors: +- Koneru B +- Lopez G +- Farooqi A +- Conkrite KL +- Nguyen TH +- Macha SJ +- Modi A +- Rokita JL +- Urias E +- Hindle A +- Davidson H +- Mccoy K +- Nance J +- Yazdani V +- Irwin MS +- Yang S +- Wheeler DA +- Maris JM +- Diskin SJ +- Reynolds CP +journal: Cancer Res +year: '2020' +doi: 10.1158/0008-5472.CAN-19-3068 +keywords: +- "Cell Line, Tumor" +- Child +- "Child, Preschool" +- Disease-Free Survival +- Female +- Follow-Up Studies +- "Gene Expression Regulation, Neoplastic" +- Humans +- Infant +- Male +- "Neoplasm Recurrence, Local" +- "Neuroblastoma/genetics, mortality, pathology" +- "RNA, Messenger/isolation & purification, metabolism" +- RNA-Seq +- "Telomerase/genetics, isolation & purification, metabolism" +- Telomere/metabolism +- Telomere Homeostasis +- Whole Genome Sequencing +- X-linked Nuclear Protein/genetics +- Xenograft Model Antitumor Assays +publication_types: +- Journal Article +- "Research Support, N.I.H., Extramural" +- "Research Support, Non-U.S. Gov't" +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# Telomere Maintenance Mechanisms Define Clinical Outcome in High-Risk Neuroblastoma. +**Authors:** Koneru B, Lopez G, Farooqi A, Conkrite KL, Nguyen TH, Macha SJ, Modi A, Rokita JL, Urias E, Hindle A, Davidson H, Mccoy K, Nance J, Yazdani V, Irwin MS, Yang S, Wheeler DA, Maris JM, Diskin SJ, Reynolds CP +**Journal:** Cancer Res (2020) +**DOI:** [10.1158/0008-5472.CAN-19-3068](https://doi.org/10.1158/0008-5472.CAN-19-3068) + +## Content + +Neuroblastoma is a childhood cancer with heterogeneous clinical outcomes. To comprehensively assess the impact of telomere maintenance mechanism (TMM) on clinical outcomes in high-risk neuroblastoma, we integrated the C-circle assay [a marker for alternative lengthening of telomeres (ALT)], TERT mRNA expression by RNA-sequencing, whole-genome/exome sequencing, and clinical covariates in 134 neuroblastoma patient samples at diagnosis. In addition, we assessed TMM in neuroblastoma cell lines (n = 104) and patient-derived xenografts (n = 28). ALT was identified in 23.4% of high-risk neuroblastoma tumors and genomic alterations in ATRX were detected in 60% of ALT tumors; 40% of ALT tumors lacked genomic alterations in known ALT-associated genes. Patients with high-risk neuroblastoma were classified into three subgroups (TERT-high, ALT+, and TERT-low/non-ALT) based on presence of C-circles and TERT mRNA expression (above or below median TERT expression). Event-free survival was similar among TERT-high, ALT+, or TERT-low/non-ALT patients. However, overall survival (OS) for TERT-low/non-ALT patients was significantly higher relative to TERT-high or ALT patients (log-rank test; P < 0.01) independent of current clinical and molecular prognostic markers. Consistent with the observed higher OS in patients with TERT-low/non-ALT tumors, continuous shortening of telomeres and decreasing viability occurred in low TERT-expressing, non-ALT patient-derived high-risk neuroblastoma cell lines. These findings demonstrate that assaying TMM with TERT mRNA expression and C-circles provides precise stratification of high-risk neuroblastoma into three subgroups with substantially different OS: a previously undescribed TERT-low/non-ALT cohort with superior OS (even after relapse) and two cohorts of patients with poor survival that have distinct molecular therapeutic targets. SIGNIFICANCE: These findings assess telomere maintenance mechanisms with TERT mRNA and the ALT DNA biomarker C-circles to stratify neuroblastoma into three groups, with distinct overall survival independent of currently used clinical risk classifiers. + +Unlimited proliferation of cancer cells requires that they maintain telomeres (1), which are typically maintained by activating telomerase (2). TERT encodes the catalytic component of telomerase (3,4). Cancer cells activate telomerase by acquiring TERT promoter mutations, TERT genomic alterations, or transcriptional dysregulation (5–8). TERT mRNA levels correlate with telomerase activity (9,10) and is often used to infer telomerase activity in cancer (5,9–11). Some cancers do not express TERT mRNA and use the recombination-mediated alternative lengthening of telomeres (ALT) mechanism (12). Telomeric repeats in the form of partially double-stranded DNA, termed C-circles, provide a highly specific marker for ALT activity (13). ALT is commonly associated with loss-of-function genomic alterations in ATRX or its binding partner DAXX (14–16), and less commonly with mutations in H3F3A (17) or SMARCAL1 (18). + +Neuroblastoma is a malignant childhood cancer arising from the developing sympathetic nervous system with clinical behavior that ranges from spontaneous regression to relentless and fatal tumor progression in spite of multimodal intensive therapy (19). High-risk neuroblastoma, as defined by stage, age, histology, and MYCN genomic amplification, has an overall survival rate ~ 50% with current intensive therapy (20–22). In addition to known oncogenic drivers, including MYCN amplification and ALK mutations (23,24), recent whole genome and exome sequencing studies demonstrated structural rearrangements near the TERT locus in ~25% of high-risk neuroblastomas (5,11) and largescale structural variations (SVs) in the chromatin remodeler gene ATRX in ~10% of high-risk neuroblastomas (24). + +Of these four most frequent genomic alterations identified in neuroblastomas, three converge on activating telomere maintenance mechanisms (TMMs) (5,24). Tumors with MYCN amplification or TERT rearrangements activate telomerase by inducing TERT mRNA expression (5,11), while loss of ATRX is implicated in the ALT phenotype (14–16). Genomic amplification of MYCN, TERT genomic rearrangements, and loss of ATRX are nearly always mutually exclusive events, suggesting a functional redundancy between these mechanistic pathways. Telomere maintenance has long been thought to be pivotal for high-risk neuroblastoma (25). However, neuroblastoma cell lines with high telomere content and continually shortening telomeres due to lack of a TMM (termed the ever-shorter telomere (EST) phenotype) have been described and analysis of primary patient tumors suggested that EST comprises >10% of high-risk patient tumors (26). + +A recent study of TMM in neuroblastoma reported that 3% of high-risk neuroblastoma tumors lacked evidence of TMM activation (27), but that study employed TERT expression by microarray and defined ALT by detecting ALT-associated promyelocytic leukemia bodies (APBs). Unlike the specific and sensitive C-circle assay (13), APBs can also be detected in non-ALT cells with long telomeres due to an increased frequency of telomere trimming events (28). Thus, the true frequency and clinical relevance of TMM-negative (EST phenotype) tumors in high-risk neuroblastoma remains to be defined. + +To comprehensively define TMMs and their clinical impact in high-risk neuroblastoma, we employed the ALT-specific C-circle assay, telomere content measurement by qPCR, and TERT mRNA expression measurement by RNA sequencing in 134 neuroblastoma primary tumors (110 high-risk and 24 non-high-risk). Genomic sequencing data for most of these tumors was available from the National Cancer Institute TARGET program (https://ocg.cancer.gov/programs/target;accession#phs000467 ) and was integrated with clinical risk stratification and outcome data. We also assessed TERT expression, C-circles, telomere content in patient-derived cell lines (n=103) and xenografts (n=28) established from high-risk neuroblastomas. + +Primary tumor samples prior to therapy were collected (snap-frozen) with written informed consent from children with neuroblastoma entered on to Children’s Oncology Group protocols. Pilot sections of tumors verified tumor content. RNA and DNA were extracted, RNA quality monitored by RIN values, and subjected to RNA and DNA sequencing by the NCI TARGET program (24). DNA and RNA were provided to TTUHSC for C-circle, telomere content, and RT-PCR for TERT mRNA expression. + +Structural variations from whole genomes (CGI) used a combination of alignment based and read-depth based approaches. Alignment based SVs are initially obtained through The CGI Cancer Pipeline 2.0. Read-depth breakpoints were obtained from segmentation profiles generated from tumor/blood matched log R ratios from every 2Kb window across the genome; ATRX variants were as previously described for the WES dataset (24,29). + +C-circle assay (13), amplified C-circles, detected using telomere qPCR or slot-blotting, (26,30) and telomere content assay by qPCR (26) were performed as previously described. + +TERT mRNA expression was measured by RNA sequencing, validated by RT-PCR, details in supplementary methods. TRAP assay was performed on whole cell lysate from selected cell lines and PDXs using either TRAPeze Telomerase Detection kit (EMD Millipore) or TeloTAGGG Telomerase PCR ELISA kit (Roche) as previously described (31). + +TRF analysis was performed as previously described (32). + +For constitutive overexpression, MYCN was cloned into pLenti-C-Myc-DDK-IRES-Puro (Origene; PS100069). + +All neuroblastoma patient-derived cell lines and patient-derived xenografts (PDXs) were obtained from the ALSF/Children’s Oncology Group (COG) Childhood Cancer Repository at Texas Tech University Health Sciences Center (www.CCcells.org). All cell line and PDX identities were confirmed using the AmpFLSTR Identifier Plus PCR Amplification kit (Applied Biosystems) at time of use for experimentation, verified against the STR database at www.CCCells.org. Neuroblastoma cell lines were used at passage numbers below 25, except when noted in results, were cultured in antibiotic-free medium, and routinely tested for lack of mycoplasma contamination. A total of 28 neuroblastoma PDXs were used in this study all employed at passage 2 in mice. PDXs were all verified to be free of human pathogens and Epstein-Barr virus by PCR. + +R (Version 3.3.2), SAS software (SAS Institute, Cary, NC, USA), IBM SPSS (V25) and GraphPad Prism (version 7.0a) were used for statistical analysis and data representation. Differences in TERT mRNA expression, telomere content, and telomerase activity between subgroups used the Wilcoxon rank-sum test. Associations of phenotype with clinical risk factors used Fisher’s exact test. Survival comparisons were by the Kaplan-Meier log-rank method using the R “survival” package (http://cran.r-project.org/web/packages/survival/index.html). Event-free survival (EFS) was calculated as time from initial diagnosis to the first occurrence of either relapse, progressive disease, secondary malignancy or death. Patients without events were censored at the time of last follow-up. Overall survival (OS) was defined as time from initial diagnosis until time of death from any cause, alive patients censored at last contact. Multivariate Cox regression was used to estimate the prognostic impact of TERT mRNA expression and C-circle presence while adjusting for ALK alterations, RAS/P53 alterations, ploidy, MYCN amplification, and 11q chromosome arm loss for both EFS and OS. Age and stage of the patients were not included in the final model, as all the patients in survival analysis were stage 4 high-risk patients with age at diagnosis > 18 months. + +See supplementary methods for additional details. + +To comprehensively assess telomere maintenance mechanisms and their clinical relevance in high-risk neuroblastoma, we integrated the C-circle assay (n=145), telomere content assay (n=145), RNA sequencing (RNA-seq; n=160), whole genome sequencing (WGS; n=136), whole exome sequencing (WES; n=222), and clinical co-variates (data availability and overlap is presented in Supplementary Fig. S1). All specimens were obtained at initial diagnosis with written informed consent at Children’s Oncology Group member institutions in accordance with recognized ethical guidelines (24). Most cases assessed in this study are stage 4 high-risk neuroblastoma patients over 18 months of age (n = 107), with 3 < 18-month-old high-risk patients and 24 patients being classified as low- or intermediate-risk. All samples analyzed had > 70% tumor content on pilot sections and high-quality nucleic acids. + +To investigate the prevalence of ALT in neuroblastoma, we quantified C-circles by qPCR for 134 primary tumors (110 high-risk (30 MYCN-amplified), 12 intermediate-risk, and 12 low-risk) selected solely based on DNA availability (Table 1; Supplementary Table S1). Of the 107 high-risk tumors (age > 18 months), 25 (23.4%; 95% CI: 15.3–31.4%) were positive for the ALT biomarker C-circles (Fig. 1A; Supplementary Table S1). For 23 of 25 ALT tumors WGS and/or WES data was available; ATRX genomic alterations were observed in 13 of 23 ALT tumors (57%) (10 focal deletions, 2 missense and 1 nonsense mutations; Table S1); 9 focal deletions and the 3 single nucleotide variants (SNVs) were previously reported (24), while one sample with an ATRX focal deletion (PATCFL) was missed in the previous WES study (Supplementary Fig. S2A) (24). None of the non-ALT tumors (C-circle negative) with WGS and/or WES (n=102) had ATRX genomic alterations. We assessed C-circles in 11 additional tumor samples previously reported to have ATRX genomic alterations including one sample (PASAJU) with a focal gain, and all of them were positive for C-circles (Supplementary Table S1) (24). + +Given that ATRX deletions commonly appear as intragenic in-frame fusions (15,33), we next explored the effect of ATRX variants on transcript levels using String Tie de novo assembly and quantification from RNA-seq and 42 different ATRX transcripts were predicted; we classified those transcripts into coding, likely coding, non-coding, exonic loss and in-frame fusion (IFF) (Supplementary Fig. S2B). Notably, aberrant transcripts matching IFFs were confirmed as the most expressed isoform in 7 of 10 ATRX structural variant (ATRXSV+) samples (PATGLU, PAPKXS, PANZVU, PANLET, PARACS, PATCFL, PASTCN and PANLET) while the 3 remaining (PARYNK, PAPKXS and PASRFS) expressed ATRX isoforms at very low levels (Supplementary Fig. S2C). + +In order to identify additional ATRX-altered cases that were missed or not overlapping with the WGS/WES dataset, we compared the total expression of IFF-ATRX isoforms against ATRX-coding isoforms (Supplementary Fig. S2C). Two additional cases (PAIFXV and PANBJH) showed overexpression of IFF-ATRX, and both cases were positive for C-circles. Considering the combined DNA and RNA variant detection approaches, 10 ALT cases had focal deletions in ATRX, 2 cases had higher expression of IFF-like transcripts, and 3 had SNVs (Supplementary Table S1; Fig. 1A and Supplementary Fig. S3). A total of 10/25 ALT tumors had no ATRX genomic alterations (Supplementary Table S1; Fig. 1A), which we validated as wild-type for ATRX using sanger sequencing. We did not detect genomic alterations in the other known ALT-associated genes (DAXX, H3F3A or SMARCAL1) in ATRX wild-type ALT tumors (14,17,18). Event-free survival (EFS) and overall survival (OS) were similar among ALT tumors based on ATRX status (Log-rank test; EFS, P = 0.741; OS, P = 0.346; Supplementary Fig. S4A and S4B). + +To further validate activation of ALT in neuroblastoma without ATRX genomic alterations, we screened 104 neuroblastoma patient-derived cell lines and 28 neuroblastoma patient-derived xenografts (PDXs) for C-circles using qPCR (Supplementary Table S2); 5 cell lines and 3 PDXs were C-circle positive. C-circle positivity was further confirmed by slot blotting of C-circle assay products onto a nylon membrane, detected using a DIG-labeled telomere probe (Supplementary Fig. S5A). ALT-associated promyelocytic leukemia bodies (APB’s) were also observed in ALT cell lines and PDXs (Supplementary Fig. S5B) (34). The ALT cell line SK-N-FI was previously shown to be wild-type for ATRX (26,35), and we identified 2 ALT PDXs (COG-N-589x, and COG-N-620x) that were ATRX wild-type (Supplementary Fig. S5C) (36). These ATRX wild-type ALT cell lines and PDXs were confirmed to express ATRX protein (Supplementary Fig. S5D and S5E). Additionally, none of the ATRX wild-type ALT models had mutations in DAXX, H3F3A, or SMARCAL1. Thus, ALT can occur in neuroblastoma with and without ATRX genomic alterations. + +ALT tumors were significantly associated with later disease onset (Supplementary Fig. S6A). Somatic mutation burden was significantly increased with patient age (Supplementary Fig. S6B) and in ALT tumors (Supplementary Fig. S6C). Consistent with prior reports, ALT tumors were exclusively non-MYCN-amplified (non-MYCN-Amp) (Fig. 1A; Supplementary Table S3) (35,37), and these cases uniformly showed significantly lower MYCN mRNA expression (Fig. S6D). Among the ALT cell lines and PDXs, two ALT cell lines had MYCN amplification (copy number assay by qPCR; 32 copies for COG-N-515 and 33 copies for COG-N-512), but MYCN mRNA expression in these MYCN-Amp ALT cell lines was lower than other non-ALT MYCN-Amp cell lines (Supplementary Fig. S6D). We did not find any associations between ALT and segmental chromosomal alterations, regardless of ATRX status (Fig. 1A). Frequency of ALK, RAS/MAPK, and TP53/CDKN2A genomic alterations were also similar among ALT and non-ALT tumors (Fig. 1A). + +In summary, ~23% of high-risk neuroblastoma tumors are ALT+. ALT activation can occur with and without ATRX alterations in neuroblastoma. ALT activation was detected exclusively in non-MYCN amplified tumors and in relatively older patients. + +As it is known that telomerase can be activated by TERT genomic rearrangements in neuroblastoma (5,11), we analyzed recurrent SVs affecting the TERT locus in 85 neuroblastoma samples that had overlapping WGS, C-circle, and TERT expression data (Fig. 1B). Of 59 high-risk tumors (age > 18 months), 12 (20.3%; 95% CI: 10.1–30.6%) harbored structural rearrangements near the TERT gene (Table S1; Fig. 1A), which is consistent with previous reports (5,11). Additionally, as reported previously (29), none of the tumors with WGS had TERT promoter mutations.. + +TERT expression quantified by RNA-seq versus qRT-PCR was strongly correlated in 19 primary tumors (Supplementary Fig. S7A; Spearman Rho = 0.72; P = 7.83×10−4). We assessed whether TERT mRNA expression could be used as an alternative to telomerase activity by using neuroblastoma cell lines for which TERT expression (by both qRT-PCR and RNA-seq) and telomerase activity (qTRAP by PCR ELISA) were simultaneously quantified (Supplementary Fig. S7B). TERT expression by RNA-seq strongly correlated with expression by qRT-PCR (n=23; Supplementary Fig. S7C; Spearman Rho = 0.92; P = 2.57×10−10) and with telomerase enzymatic activity (n=22; Supplementary Fig. S7D; Spearman Rho = 0.95; P = 6.226×10−12). Thus, RNA-seq can reliably quantify TERT mRNA expression and is indicative of telomerase activity. + +As expected, TERT structural variants (TERTSV+) tumors had the highest TERT mRNA expression, followed by MYCN-Amp tumors (Fig. 1C). ALT tumors expressed the least TERT of the high-risk subsets (Fig. 1A and 1C). Similar to patient tumors, ALT neuroblastoma cell lines and PDXs had low TERT expression (Supplementary Table S2) and were negative for telomerase activity (Supplementary Fig. S8A and S8B). The majority of high-risk tumors without TERTSV+, MYCN amplification, or ALT had lower TERT expression when compared to MYCN-Amp or TERTSV+ tumors (Wilcoxon P = 7.6×10−4 +Fig. 1C), however some had TERT expression as high as TERTSV+ tumors (Fig. 1C) suggesting the existence of unidentified mechanisms for activation of TERT in high-risk neuroblastoma. In our cohort, none of the 24 non-high-risk tumors had high TERT expression (Fig. 1C). + +Surprisingly, some MYCN-Amp tumors had TERT expression as low as seen in ALT tumors (Fig. 1C and Supplementary Fig. S9A). These MYCN-Amp samples had low TERT expression despite having high MYCN mRNA expression (Supplementary Fig. S9A). Similar to primary tumors, four MYCN-Amp cell lines had low TERT expression and were negative for telomerase activity (Table S2; Supplementary Fig. S8B). Knockdown of MYCN expression using MYCN shRNA in a MYCN-Amp neuroblastoma cell line (COG-N-452h) induced downregulation of TERT compared to non-targeted shRNA controls (Supplementary Fig. S9B; P < 0.05). Forced overexpression of MYCN using a lentiviral vector was undertaken in four non-MYCN-Amp neuroblastoma cell lines: two with telomerase activity (CHLA-171 and Felix-h) and two without telomerase activity (LA-N-6 and CHLA-132, Supplementary Fig. S9C). Forced MYCN overexpression upregulated TERT expression in one non-MYCN-Amp telomerase+ cell line (CHLA-171), but failed to enhance TERT expression in the telomerase-positive cell line with a TERT-rearrangement (Felix-h), or in the 2 cell lines negative for telomerase activity (Supplementary Fig. S9D, P < 0.005). This suggests that MYCN can upregulate TERT expression via an intact and non-repressed TERT promoter, but MYCN overexpression by itself is not sufficient to overcome the TERT repression in TERT non-expressing cells. + +In summary, TERTSV+ tumors, and the majority of MYCN-Amp tumors had high TERT mRNA expression. Some TERTwt and MYCNwt tumors had TERT expression as high as tumors with TERTSV+, indicating some other unidentified mechanisms for activation of TERT in neuroblastoma. Not all MYCN-Amp tumors had high TERT expression. Consistent with that observation, forced overexpression of MYCN in telomerase-negative cell lines did not induce TERT expression. + +We analyzed the outcome of 107 stage 4 high-risk patients with age at diagnosis > 18 months for which both TERT expression and ALT status were simultaneously assessed (Table S1). High TERT expression, defined as ≥ the median TERT expression of the high-risk cohort (n=107), was associated with poor overall survival (OS) (P < 0.001; Fig. 2A) and poor event-free survival (EFS) (P = 0.034; Supplementary Fig. S10A). The 5-year OS was 28% for the TERT high (TERT-H) group vs 62% for the TERT low (TERT-L) group. However, patients with ALT+ tumors showed no difference from non-ALT patients in OS (P = 0.99; Fig. 2B) or EFS (P = 0.32; Supplementary Fig. S10B). + +Because the OS curves based on ALT status intersect, we computed the Cox regression hazard ratios using time interaction term, and we observed a substantial increase in hazard ratio after 5 years for the ALT group (HR at >5 years: 14.1; P = 0.024). In non-ALT tumors, overall survival of high-risk patients was TERT expression-dependent, poor with higher TERT expression (Supplementary Fig. S11A and S11B). As ALT tumors are a subgroup of TERT-L tumors (all ALT tumors had TERT mRNA expression below the median; Supplementary Fig. S9A), we combined analysis of ALT with TERT expression, which allowed us to define three groups of patients (TERT-H, ALT, and TERT-L/non-ALT) that each showed distinct OS probabilities (log-rank test comparing three groups, P < 0.001; Fig. 2C). The 5-year OS for TERT-H patients was 28% vs 46% for ALT patients and 75% for TERT-L/non-ALT patients, while 10-year OS was 22% for TERT-H, 24% for ALT, and 75% for TERT-L/non-ALT (Table. 2). All patients with TERT-L/non-ALT tumors who died had RIN values > 6.5, ruling out low RNA integrity as a cause for low TERT expression by RNA sequencing. Consistent with previous studies, the presence of ALK and RAS/TP53 pathway alterations (Supplementary Fig. S12A and S12B) (27,38) and diploid DNA content (Supplementary Fig. S12C and S12D) were associated with poor outcome in these high-risk patients. + +Unlike OS, EFS was not substantially different among the three groups defined by TERT expression and ALT status (P = 0.137; Fig. 2D). Therefore, we assessed OS in only those patients who experienced an event, and observed that while TERT-H and ALT patients had abysmal OS following an event, patients in the TERT-L/non-ALT group had significantly higher OS than patients in the TERT-H or ALT groups (log-rank test comparing three groups: P < 0.001; Fig. 2E). These data suggest that the differences in OS likely represent differences in the success of salvage therapy following an event. + +Consistent with previous findings for high-risk patients, there was no difference in EFS (P = 0.68; Supplementary Fig. S13A) or OS (P = 0.46; Supplementary Fig. S13B) between MYCN-Amp and non-MYCN-Amp patients (39,40). Even though the majority of MYCN-Amp tumors had high TERT expression and belonged to the TERT-H group (n=22), some MYCN-Amp tumors had low TERT expression (n=7; Supplementary Fig. S9A) and were classified as TERT-L. Notably, MYCN-Amp patients in the TERT-H group had a lower 5-year OS (29%) relative to the 86% seen for patients with MYCN-Amp & TERT-L tumors (P = 0.013; Fig. 2F). This observation was further validated in the SEQC dataset (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi); MYCN-Amp tumors with low TERT expression had a significantly higher OS compared to MYCN-Amp tumors with high TERT expression (P = 0.0053; Supplementary Fig. S13C). Multivariate cox-regression analysis demonstrated that high TERT expression or ALT positivity predict poor OS but not EFS, independent of MYCN amplification, ploidy, ALK status, RAS/TP53 pathway alterations, or chromosome arm 11q deletion (Fig. 3A and 3B). + +As median cutoff may not be optimal for distinguishing TERT expressing and non-expressing tumors for biological classification, we also defined a TERT expression threshold that separates TERT positive and TERT negative tumors by fitting a bimodal normal distribution (Fig. S14 A and B), as described previously (27). Based on the latter TERT expression threshold, 16/107 (14.9%) high-risk tumors were classified as TERT expression low and negative for the C-circle ALT marker (TERT&ALT-). Overall survival but not Event-free survival was significantly different for ALT, TERT positive and TERT&ALT- tumors (log-rank test comparing three groups: P = 0.022; Fig. S14C and S14D), results comparable to using median TERT expression for the cutoff (Fig. 2C). + +Taken together, we classified high-risk neuroblastoma patients into 3 subgroups based on TERT expression and ALT status. Non-ALT patients with lower TERT expression were associated with superior overall survival when compared to patients with ALT phenotype or higher TERT expression, irrespective of other known risk factors. + +To investigate whether ALT is associated with high telomere content (TC) as described previously (14,15,26,41), we assessed relative TC using telomeric qPCR in 134 neuroblastoma samples for which both ALT status and TERT expression were known. As expected, ALT tumors had higher relative TC than TERT-H (Wilcoxon P = 1.87×10−11), TERT-L/non-ALT (Wilcoxon P = 9.5×10−4), or non-high-risk tumors (Wilcoxon P = 5.54×10−8; Fig. 4A). Consistent with primary patient tumors, ALT cell lines and PDXs had higher relative TC than non-ALT models (Supplementary Fig. S15A-C and Table S2; Wilcoxon P = 2.55×10−5). ALT neuroblastoma cell lines were also associated with high and heterogeneous telomere length relative to telomerase-positive cell lines, as detected by telomere restriction fragment (TRF) length analysis (mean TRF length >20kb; Supplementary Fig. S15C) and by telomere fluorescence in-situ hybridization on metaphase spreads (Supplementary Fig. S15D). + +TERT-L/non-ALT tumors and non-high-risk tumors were each significantly associated with a much higher relative TC when compared to TERT-H tumors (Wilcoxon P = 2.6×10−07; Supplementary Fig. 4A). As WGS data was available for 85/134 samples, we measured telomere DNA abundance (TTAGGG/CCCTAA canonical repeats) from short-read WGS for each tumor and from a blood sample from each patient, and demonstrated a significant correlation between tumor telomere content measured by qPCR and WGS (Supplementary Fig. S16A; Spearman Rho = 0.701; P = 5.87×10−14). We observed that 11/17 TERT-L/non-ALT tumors had increased telomere content (tumor/blood ratio > 1; Supplementary Fig. S16B) and of 82 non-ALT high-risk tumor samples with age > 18 months, 16 tumors had high TC, similar in range to that observed within ALT tumors (relative TC >5; corresponds to mean TRF length > 20 kb; Supplementary Table S1). This non-ALT high TC group meets the phenotypic criteria of the previously defined EST phenotype (26). + +In the 16 tumors with high TC, 12 had low TERT expression and 4 had high TERT expression (Supplementary Table S1). To validate that TERT-L&C-circle- tumors with high telomere content (TC>5) are ALT-negative, we assessed APBs in 34 primary patient tumors (12 tumors in the TERT-L&C-circle- group with TC>5, 11 tumors in TERT-L&C-circle- group with TC<5, and 10 C-circle+ tumors). APBs were detected in 3/12 TERT-L&C-circle- tumors with TC>5, while 0/11 tumors in TERT-L&C-circle- group with TC<5 were APB-positive. In C-circle positive tumors, 8/10 were positive for APBs (Supplementary Table S1, Supplementary Fig. S17). Thus, the C-circle assay missed detection of ALT in 3/11 low TERT tumors with high TC, while the APB assay did not detect ALT in 2/10 C-circle positive tumors. + +We compared clinical outcome between the 28/107 ALT tumors with ALT defined as either C-circle+ or APB+) and the 9/79 non-ALT tumors that had low TERT expression and TC>5 (Fig. 4B). OS among these two groups was significantly different, with the ALT cohort having a much worse outcome (P = 0.007; Fig. 4C). + +We also measured relative TC content, C-circles, and TERT mRNA in 104 neuroblastoma cell lines and 28 PDXs (Supplementary Table S2). We identified 5 neuroblastoma cell lines (LA-N-6, COG-N-291, CHLA-132, COG-N-509hnb, and COG-N-562hnb) and 2 PDXs (COG-N-518x and COG-N-649x) with low TERT expression that were ALT negative (Fig. 4D) and had relative TC>5 (Supplementary Table S2), i.e. consistent with the EST phenotype (26). We validated the lack of ALT activation in these models by confirming a low frequency of APBs (Supplementary Fig. S18A and S18B). All the EST cell lines and PDXs lacked telomerase activity by TRAP assay (Fig. 4E) and had high telomere length using TRF analysis in cell lines (mean TRF length > 20 kb; Fig. 4F), and telomere/centromere FISH in PDXs (Fig. 4G) and showed high and heterogeneous telomere lengths (Fig. 4G; Supplementary Fig. S18C) with a lower frequency of telomere-signal-free chromosomal ends than ALT cell lines (Wilcoxon P = 3.35×10−07; Supplementary Fig. S18C and S18D). Unlike ALT tumors, where ATRX/DAXX genomic alterations are common (14,15), none of the TERT-L/non-ALT tumor samples with TC>5 (n=14) had genomic alterations in ATRX, DAXX, H3F3A or SMARCAL1. Consistent with the patient samples, the TERT-L/non-ALT cell lines and PDXs with TC>5 expressed full-length wild-type ATRX and DAXX protein (Supplementary Fig. S19A and S19B). + +We were able to induce the ALT-specific marker C-circles in the EST cell line LA-N-6 (but not in the telomerase-positive cell line SK-N-BE (2)) by stable ATRX knockdown using shRNA (Supplementary Fig. S20A and S20B). Knockdown of ATRX in LA-N-6 also increased APBs in these cells (Supplementary Fig. S20C and S20D). This suggests that ALT repressors such as ATRX can block activation of ALT in TERT-L/non-ALT tumors. + +Thus, some neuroblastomas stratified clinically as high-risk have high TC despite no apparent activation of classical ALT markers or TERT expression, and as shown in Figure 3C, this phenotypic subset has a much more favorable OS probability than the ALT tumors. + +Roughly 15% of the high-risk neuroblastoma tumor samples were negative for TERT expression without any evidence of ALT activation (TERT- & ALT -; Supplementary Fig. S14 A). To investigate whether this phenotype was associated with yet another uncharacterized telomere maintenance mechanism, we measured the change in telomere length across multiple passages in 3 neuroblastoma TERT- & ALT- cell lines established from high-risk patients (LA-N-6, COG-N-346h, and COG-N-387h). All three cell lines had very low TERT expression, were negative for telomerase activity, and lacked ALT-specific markers (Supplementary Table S2); LA-N-6 is a non-MYCN-Amp neuroblastoma cell line with TC>5, COG-N-346h is a non-MYCN-Amp neuroblastoma cell line with TC<5, and COG-N-387h is a MYCN-Amp neuroblastoma cell line with TC<5 (Supplementary Table S2). All three cell lines showed continuous shortening of telomere length by TRF analysis across progressive passages (Fig. 5A); shortening of telomeres was verified as a reduction in telomere content by qPCR (Fig. 5B). All three TERT- & ALT- lines remained C-circle-negative (Supplementary Fig. S21A) and TERT-low across those same passages in culture (Supplementary Fig. S21B). + +We observed a substantial decline in cell viability in 2 of 3 EST cell lines following several passages in vitro (Fig. 5C). Interestingly, the decline in viability of LA-N-6, a cell line with very high telomere content, was minimal even after 40 passages (Fig. 5C). + +Taken together our data suggest that patients with TERT-L/non-ALT tumors derive a survival advantage from the failure of their tumors to maintain telomere length, but that some of these tumors may develop as yet unknown mechanisms to overcome this limitation. + +We report here a comprehensive analysis of telomere maintenance mechanisms (TMM) in high-risk neuroblastoma in the context of tumor genomics and clinical outcome. Consistent with previous studies, we confirmed using the ALT-specific C-circle assay that ALT is activated in 23% of high-risk neuroblastoma tumors (26). ALT patient tumors, cell lines, and PDXs expressed very low TERT mRNA, indicating ALT and telomerase activation occur in a mutually exclusive manner. + +ATRX genomic alterations have been associated with the ALT phenotype in neuroblastoma (15,27), but the relationship of ALT (using a specific marker) to ATRX genomic alterations has not been precisely defined. We observed ATRX structural variants, in frame fusions and single nucleotide variants in 60% of ALT tumors, but the remainder lacked any identifiable genomic lesions previously reported to be associated with the ALT phenotype. Consistent with previous reports that ATRX mutations are often enriched in older patients (15), the ALT patients in our cohort had a late disease onset irrespective of ATRX status. Patients with ALT phenotype had a poor overall survival, irrespective of ATRX status. + +High telomere content has been associated with ALT (15,37,41), but we confirmed a previous report that long telomeres are not exclusive to ALT tumors in high-risk neuroblastoma (26), as 16% of the ALT-negative high-risk tumors had total telomere content as high as that seen in the ALT-positive tumors. Thus, assessment of ALT status using high telomere content as measured by FISH, WGS/WES, or qPCR would over-estimate the frequency of ALT in neuroblastoma. + +Similar to previous reports, we identified TERT structural variants (TERTSV+) in ~20% of high-risk neuroblastoma patients, and all TERTSV+ tumors had highly upregulated TERT mRNA (5,11). Among tumors with MYCN genomic amplification (MYCN-Amp), a sub-population had TERT expression as low as that seen in ALT tumors. Using neuroblastoma cell lines, we showed that MYCN overexpression enhanced TERT mRNA expression in a TERT-expressing non-MYCN-Amp neuroblastoma cell line without TERTSV+, but failed to induce TERT activation in cell lines with TERTSV+ or in TERT non-expressing cell lines. Thus, MYCN overexpression by itself might not be always sufficient to transcriptionally activate TERT. + +As seen in a recent study (27), we observed that TERT activation in high-risk neuroblastoma was not limited to tumors with TERTSV+ and MYCN amplification, as some tumors without these alterations expressed TERT as highly as TERTSV+ and MYCN-Amp tumors. The prior study quantified TERT mRNA expression by microarray and detected ALT activation by scoring APBs on tumor tissue sections (27,42), and reported that only 3 of 90 high-risk tumors lacked a TMM (42). By employing RNA sequencing (validated by qRT-PCR) for TERT expression and the C-circle assay to detect ALT, we found that 12 to 26% (depending on the cutoff employed) of high-risk neuroblastoma tumors (including some MYCN-Amp tumors) had low TERT expression and lacked ALT activation, and those patients had a significantly better overall survival. Consistent with this latter observation high-risk neuroblastoma cell lines that were non-ALT, low TERT expressing, and lacked telomerase activity showed continuous shortening of telomeres with growth in vitro and 2 of 3 TERT-L/non-ALT cell lines spontaneously lost viability after multiple passages in culture. + +By combining analysis of ALT status (by C-circles) and TERT mRNA expression as measured by RNA-seq we were able to classify high-risk neuroblastoma patients into 3 subgroups based on TMM (TERT-H, ALT, and TERT-L/non-ALT). The 10-year overall survival (OS) in high-risk patients with TERT-H and ALT tumors was < 25% while the TERT-L/non-ALT group had a 75% 10-year OS, despite being classified as high-risk based on current risk stratification protocols (43). We also showed that OS for patients with MYCN amplification was significantly higher if the tumors had low TERT expression, which might explain the extreme dichotomy in the clinical course of MYCN-Amp patients (44). + +We observed that TERT high-expressing and ALT neuroblastomas have poor clinical outcomes comprising uniform cohorts of true high-risk patients. An important and novel observation was our demonstration that 12 to 26% of clinically high-risk neuroblastomas lacked a TMM and while those had EFS comparable to TERT-high and ALT patients, TMM-negative patients had a significantly higher long-term overall survival (75%). Based on our data, we hypothesize that TMM-negative tumors develop chemo-resistance and progress similarly to TMM-positive tumors, but due to a lack of TMM they continue to erode telomeres which eventually enables salvage therapy given after disease progression to eradicate the tumors. Thus, current standard of care for high-risk neuroblastoma patients that relies on dose-intensive cytotoxic chemotherapy in as short a time frame as possible may not be optimal for TMM-negative high-risk patients, and alternative strategies such as longer time frame low dose chemotherapy should be tested in future clinical trials. + +Employing the C-circle assay, considered to be specific for ALT cells and known to be a quantitative marker for ALT activity (13,45), enabled us to identify ALT in 23% of high-risk (age >18 months) neuroblastoma patients, and these patients had a low overall survival (Fig. 2). However, 3 tumors with high telomere content were positive for APBs but had a C-circle content below cutoff, which suggested that some ALT tumors may be missed by implementing only the C-circle assay as marker for detection of ALT. We analyzed clinical outcome for patients with low TERT expression and high telomere content with ALT defined as C-circle positive or APB positive and we observed an apparent higher overall survival for TMM-negative patients identified by employing both assays to define ALT-positivity (comparing Fig. 2c to Fig. 4c). APB analysis alone did not detect all ALT tumors as we identified 2 ATRX mutant C-circle positive tumors that were APB negative, indicating that APBs have limitations in detecting ALT tumors, which is consistant with a previous report (14). Additionally, APBs have the limitation of also being detectable in non-ALT cells (45). Cell lines with long telomeres resulting from forced telomerase overexpression have APBs (28), suggesting that APBs are associated with long telomeres rather than being specific for the ALT mechanism (45). Future studies in a large cohort of patients will be necessary to determine the optimal approach for TMM classification of tumors that have low TERT expression and a high telomere content. + +Using the C-circle assay, we observed that patients with ALT positive tumors had a dismal overall survival, with no difference in overall survival between ALT and TERT positive patients, which is consistent with previous reports that ATRX altered high risk tumors (all of which were ALT) were associated with very poor overall survival (15,33). Roderwieser and colleagues reported a substantially better overall survival for ALT neuroblastoma compared to TERT activated patients (42). We suggest that the difference in clinical outcome between our study and the prior study could be due to our use of the C-circle assay compared to use of APB’s to identify ALT, differences in risk and age groups (our study focused on HR samples in patients with age > 18 months), and length of clinical follow-up, as ALT tumors tend to have an indolent phenotype (15,46). Additionally, our study here reported higher frequency of ALT & TERT negative tumors, which may be due to use of different methods to assess TERT, different TERT expression cutoffs, inclusion of only high risk patients with age > 18 months, and differences in classification of low-TERT expressing MYCN amplified tumors. + +A strength of our study is the relatively large sample size of stage 4 high-risk neuroblastoma patients diagnosed at age > 18 months, representing patients receiving intense chemotherapy at high risk for a poor clinical outcome (43). Another strength of our study is that we determined TERT mRNA expression in the patient samples using RNA-sequencing, which may be more robust than microarray technology for quantifying low-abundance transcripts such as that of TERT, as suggested previously (47). A limitation of our study is that TERT expression is a continuous variable and cutoffs for high vs low expression using different approaches give comparable results. Our results need to be validated in a large cohort of patients, which will also allow optimizing the cutoff for TERT expression. While TERT mRNA expression could be affected by RNA integrity or tumor cell content, both of which were controlled for in this study, which is important to prevent classifying a tumor as TERT low inappropriately. Another limitation of our study and of previous studies that assessed telomere maintenance is that the patients studied were not from a uniformly treated group of patients receiving the most current standard-of-care therapies. Thus, it will be important to confirm our observations in a validation cohort of high-risk neuroblastoma patients receiving uniform therapy on current protocols. + +In summary, we demonstrated that quantifying TERT mRNA and C-circle abundance enables classifying high-risk neuroblastoma patients into 3 subgroups that have distinct clinical outcomes, independent of currently employed risk-stratification methods, including MYCN amplification. TMM-positive (TERT high-expressing and ALT) neuroblastomas are true high-risk tumors, each with distinct, potentially targetable molecular pathways required for maintaining telomeres. Therapeutic strategies targeting telomerase and ALT activation warrant investigation in TMM-positive neuroblastomas (30,48,49). An important and novel observation is our demonstration that clinically high-risk patients with TMM-negative tumors have an exceptionally high overall survival, which is likely due to the eventual success of therapy (combined with telomere erosion) even after disease progression. If future studies confirm the latter observation, stratification of patients based on TMM using robust markers (TERT mRNA expression and DNA C-circles) will improve the analysis of future clinical trials, and a reduction in the intensity of therapy may be attainable for TMM-negative patients. + +## Table 1. Patient characteristics of neuroblastoma patients at diagnosis based on risk stratification. + +| | Risk groups [colspan=4] | [colspan=2] | +| Characteristic | High Risk (n=110) [colspan=2] | Low/Intermediate Risk (n=24) [colspan=2] | All Patients [colspan=2] | +| Age at Diagnosis | N | Percent | n | Percent | n | Percent | +| < 18 months | 3 | 3% | 24 | 100% | 27 | 20% | +| >= 18 months | 107 | 97% | 0 | 0% | 107 | 80% | +| Gender | | | | | | | +| Male | 64 | 58% | 18 | 75% | 82 | 61% | +| Female | 46 | 42% | 6 | 25% | 52 | 39% | +| INSS stage | | | | | | | +| 2 or 3 | 1 | 1% | 4 | 17% | 5 | 4% | +| 4 | 108 | 98% | 1 | 4% | 109 | 81% | +| 4s | 1 | 1% | 19 | 79% | 20 | 15% | +| MYCN Status | | | | | | | +| Not Amplified | 79 | 72% | 24 | 100% | 103 | 77% | +| Amplified | 30 | 27% | 0 | 0% | 30 | 22% | +| Unknown | 1 | 1% | 0 | 0% | 1 | 1% | +| Histology | | | | | | | +| Favorable | 3 | 3% | 22 | 92% | 25 | 19% | +| Unfavorable | 99 | 90% | 1 | 4% | 100 | 75% | +| Unknown | 8 | 7% | 1 | 4% | 9 | 7% | +| Ploidy | | | | | | | +| Hyperdiploid | 54 | 49% | 20 | 83% | 74 | 55% | +| Diploid | 55 | 50% | 4 | 17% | 59 | 44% | +| Unknown | 1 | 1% | 0 | 0% | 1 | 1% | +| Grade | | | | | | | +| Differentiating | 7 | 6% | 0 | 0% | 7 | 5% | +| Undifferentiated or Poorly Differentiated | 86 | 78% | 23 | 96% | 109 | 81% | +| Unknown | 17 | 15% | 1 | 4% | 18 | 13% | + +## Table 2. EFS and OS probability estimates at 1 to 10 years from diagnosis in TERT-H, ALT, and TERT-L/non-ALT groups. + +| Time [rowspan=3] | TERT-H [colspan=4] | ALT [colspan=4] | TERT-L & non-ALT [colspan=4] | +| EFS [colspan=2] | OS [colspan=2] | EFS [colspan=2] | OS [colspan=2] | EFS [colspan=2] | OS [colspan=2] | +| Estimate | SE | Estimate | SE | Estimate | SE | Estimate | SE | Estimate | SE | Estimate | SE | +| 1 Year | 0.758 | 0.059 | 0.831 | 0.051 | 0.840 | 0.073 | 0.960 | 0.039 | 0.821 | 0.072 | 1.000 | 0.000 | +| 2 Years | 0.370 | 0.067 | 0.657 | 0.066 | 0.709 | 0.093 | 0.918 | 0.055 | 0.536 | 0.094 | 0.929 | 0.049 | +| 3 Years | 0.292 | 0.063 | 0.439 | 0.069 | 0.488 | 0.104 | 0.743 | 0.091 | 0.464 | 0.094 | 0.857 | 0.058 | +| 4 Years | 0.253 | 0.060 | 0.359 | 0.067 | 0.443 | 0.104 | 0.609 | 0.102 | 0.429 | 0.094 | 0.784 | 0.078 | +| 5 Years | 0.233 | 0.059 | 0.279 | 0.063 | 0.287 | 0.099 | 0.469 | 0.106 | 0.429 | 0.094 | 0.747 | 0.083 | +| 6 Years | 0.233 | 0.059 | 0.279 | 0.063 | 0.287 | 0.099 | 0.353 | 0.107 | 0.390 | 0.093 | 0.747 | 0.083 | +| 7 Years | 0.195 | 0.061 | 0.279 | 0.063 | 0.287 | 0.099 | 0.236 | 0.120 | 0.390 | 0.093 | 0.747 | 0.083 | +| 8 Years | 0.195 | 0.061 | 0.223 | 0.071 | 0.287 | 0.099 | 0.236 | 0.120 | 0.390 | 0.093 | 0.747 | 0.083 | +| 9 Years | 0.195 | 0.061 | 0.223 | 0.071 | 0.287 | 0.099 | 0.236 | 0.120 | 0.390 | 0.093 | 0.747 | 0.083 | +| 10 Years | 0.195 | 0.061 | 0.223 | 0.071 | 0.287 | 0.099 | 0.236 | 0.120 | 0.390 | 0.093 | 0.747 | 0.083 | \ No newline at end of file diff --git a/references_cache/PMID_33627664.md b/references_cache/PMID_33627664.md new file mode 100644 index 00000000000..63ed447f9b9 --- /dev/null +++ b/references_cache/PMID_33627664.md @@ -0,0 +1,254 @@ +--- +reference_id: PMID:33627664 +extractor_version: 1 +xml_extraction_version: 1 +title: Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome. +authors: +- Hartlieb SA +- Sieverling L +- Nadler-Holly M +- Ziehm M +- Toprak UH +- Herrmann C +- Ishaque N +- Okonechnikov K +- Gartlgruber M +- Park YG +- Wecht EM +- Savelyeva L +- Henrich KO +- Rosswog C +- Fischer M +- Hero B +- Jones DTW +- Pfaff E +- Witt O +- Pfister SM +- Volckmann R +- Koster J +- Kiesel K +- Rippe K +- Taschner-Mandl S +- Ambros P +- Brors B +- Selbach M +- Feuerbach L +- Westermann F +journal: Nat Commun +year: '2021' +doi: 10.1038/s41467-021-21247-8 +keywords: +- "Blotting, Western" +- Exons/genetics +- Flow Cytometry +- Humans +- Proteome/metabolism +- Retrospective Studies +- "Sequence Analysis, RNA/methods" +- "Telomere/genetics, metabolism" +- Telomere Homeostasis/genetics +- Whole Genome Sequencing/methods +- X-linked Nuclear Protein/genetics +publication_types: +- Journal Article +- "Research Support, Non-U.S. Gov't" +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# Alternative lengthening of telomeres in childhood neuroblastoma from genome to proteome. +**Authors:** Hartlieb SA, Sieverling L, Nadler-Holly M, Ziehm M, Toprak UH, Herrmann C, Ishaque N, Okonechnikov K, Gartlgruber M, Park YG, Wecht EM, Savelyeva L, Henrich KO, Rosswog C, Fischer M, Hero B, Jones DTW, Pfaff E, Witt O, Pfister SM, Volckmann R, Koster J, Kiesel K, Rippe K, Taschner-Mandl S, Ambros P, Brors B, Selbach M, Feuerbach L, Westermann F +**Journal:** Nat Commun (2021) +**DOI:** [10.1038/s41467-021-21247-8](https://doi.org/10.1038/s41467-021-21247-8) + +## Content + +Telomere maintenance by telomerase activation or alternative lengthening of telomeres (ALT) is a major determinant of poor outcome in neuroblastoma. Here, we screen for ALT in primary and relapsed neuroblastomas (n = 760) and characterize its features using multi-omics profiling. ALT-positive tumors are molecularly distinct from other neuroblastoma subtypes and enriched in a population-based clinical sequencing study cohort for relapsed cases. They display reduced ATRX/DAXX complex abundance, due to either ATRX mutations (55%) or low protein expression. The heterochromatic histone mark H3K9me3 recognized by ATRX is enriched at the telomeres of ALT-positive tumors. Notably, we find a high frequency of telomeric repeat loci with a neuroblastoma ALT-specific hotspot on chr1q42.2 and loss of the adjacent chromosomal segment forming a neo-telomere. ALT-positive neuroblastomas proliferate slowly, which is reflected by a protracted clinical course of disease. Nevertheless, children with an ALT-positive neuroblastoma have dismal outcome. + +Activation of a telomere maintenance mechanism (TMM) was found to be associated with poor outcome in neuroblastoma1, which is the most common extracranial solid tumor diagnosed in early childhood2–5. Telomerase activation is one way of maintaining telomere length and results from amplified MYCN (MYCN transcriptionally activates the TERT gene) or TERT rearrangement in neuroblastoma1,6. Activation of alternative lengthening of telomeres (ALT) presents an alternative route of telomere maintenance, which is seen in a substantial proportion of neuroblastoma tumors1,6–8. Mutations in the chromatin modifier alpha thalassemia/mental retardation syndrome X-linked (ATRX) are associated with elongated telomeres and ALT in neuroblastoma7,8. Furthermore, deletions of ATRX were associated with recurrent partial chromosomal losses in neuroblastomas without amplified MYCN9. ATRX and its complex partner death associated protein 6 (DAXX) incorporate histone variant H3.3 at pericentric and telomeric repeat sequences10–12. ALT activity is strongly associated with somatic ATRX mutations in different tumor entities13–16. Although loss of ATRX alone does not induce ALT, ATRX acts as an ALT suppressor and overexpression of ATRX in ATRX-mutant cells represses ALT markers17,18. Interestingly, knockdown of DAXX abolishes the ALT suppressive effect of ATRX overexpression in ATRX-mutant cells highlighting the importance of both complex members17,18. + +The clinical neuroblastoma risk stratification systems used by most national and international consortia (Society For Pediatric Oncology and Hematology-GPOH, International Society of Paediatric Oncology Europe Neuroblastoma Group–SIOPEN, Children’s Oncology Group-COG), as well as the International Neuroblastoma Risk Group (INRG) staging system, categorizes neuroblastomas mainly based on INSS stage, age, and MYCN status into low risk, intermediate risk, and high risk19. Thus in clinical practice, ALT-positive high-risk tumors are treated with the same treatment protocol as MYCN-amplified and TERT-rearranged tumors presuming similar biological features as these telomerase-activated tumors1. + +Focused studies on ALT in neuroblastoma are rare and previous observations were mostly based on genetic sequencing studies and ATRX mutation status as ALT marker6–8. Here, we aim to deepen the knowledge about the biology and clinical features of ALT-positive neuroblastomas by enriching ALT tumors in the study cohort independent of ATRX mutation status. Using a multi-omics profiling approach of ALT-positive neuroblastomas, we provide evidence that this subgroup is clinically and molecularly distinct. Clinically, ALT is associated with a protracted course of disease and poor outcome. Molecularly, ALT-positive tumors feature mutated ATRX and/or reduced protein abundance, heterochromatic telomeric chromatin, a slow proliferative capacity, and frequent neo-telomere formation. + +Using the presence of circular partially single stranded extrachromosomal telomeric repeat sequences (C-Circles) as ALT marker20, 9.2% of the neuroblastomas in a screening cohort of 720 tumors were classified as ALT-positive. C-Circle screening was further extended to a relapse cohort of tumors analyzed in the INFORM registry21 (n = 40). Importantly, 47.5% of the INFORM tumors were ALT-activated, indicating a strong enrichment of this molecular subgroup in relapsed cases (Fig. 1a, Tables 1–2). Clinically, ALT was significantly associated with INSS stage 4, high-risk disease, and an older age at diagnosis. However, about one third of the ALT-positive neuroblastomas had localized disease and were stratified at diagnosis as low or intermediate risk using a standard risk classification system. In the screening cohort, none of the 83 tested 4 S stage tumors, a clinically recognizable subtype associated with spontaneous regression seen in infants below one year of age, were ALT-activated. Amplified MYCN and C-Circle presence were mutually exclusive in the screening cohort, but not in the INFORM cohort (Tables 1–2, Supplementary Data 1–2).Fig. 1C-Circle screening of neuroblastomas.a ALT frequency detected using C-Circle presence as a marker in the screening cohort (n = 720, left) and INFORM cohort (n = 40, right). Top pies illustrate the composition of the cohorts concerning INSS tumor stage. b Schematic overview of the analyzed patient cohorts used for C-Circle screening and high-throughput analysis using whole genome sequencing (WGS), RNA-sequencing, whole proteome analysis, and ChIP-sequencing. Sample numbers are given in brackets. Number of C-Circle positive (CC+) and C-Circle negative (CC-) patients in the sub-cohorts are indicated.Table 1Clinical features of C-Circle positive tumors in the screening cohort (n = 720)a.Total cohortC-Circle positiveC-Circle negativeP valuegNo.f%fNo. f%fNo.f%fTotal720669.2%65490.8%Stage 1-3b41657.8%2537.9%39159.8%1.948E-06Stage 4b22130.7%4162.1%18027.5%Stage 4Sb8311.5%00.0%8312.7%0.000367LR/IRc44965.9%2436.9%42569.0%6.944E-07HRc23234.1%4163.1%19131.0%Male39755.1%3959.1%35854.7%0.519Female32344.9%2740.9%29645.3%Aged (mean in days)8712872669<2E-16MYCN ampe11416.2%00.0%11417.9%5.883E-06MYCN normale58883.8%66100.0%52282.1%Ganglioneuroma111.5%00.0%111.7%0.6115an refers to the number of tumors.bTumor staging was done according to the INSS criteria.cRisk stratification was based on the German Neuroblastoma Study NB2004. Risk stratification was not available for 39 tumors.dAge at diagnosis.eMYCN status refers to MYCN status at diagnosis, which is either amplified (amp, ≥8 copies) or normal. MYCN status was not available or inconclusive for 18 tumors.fAll values are given in absolute number of tumors (No.) and relative (%) to the total number of tumors in each group (total cohort, C-Circle positive, C-Circle negative).gP values were calculated using Fisher tests and two-sided Wilcoxon rank sum test (age).Table 2Clinical features of C-Circle positive tumors in the INFORM cohort (n = 40)a.Total cohortC-Circle positiveC-Circle negativeP valuegNo.f%fNo.f%fNo.f%fTotal401947.5%2152.5%Stage 1-3b410.0%421.1%00.0%0.041Stage 4b3485.0%1473.7%2095.2%Stage 4Sb25.0%15.3%14.8%1.000LR/IRc717.5%526.3%29.5%0.226HRc3382.5%1473.7%1990.5%Male3280.0%1368.4%1990.5%0.120Female820.0%631.6%29.5%Aged (mean in days)1645224311040.003MYCN ampe1127.5%210.5%942.9%0.029MYCN normale2870.0%1789.5%1047.6%an refers to the number of tumors.bTumor staging was done according to the INSS criteria.cRisk stratification was based on the German Neuroblastoma Study NB2004.dAge at diagnosis.eMYCN status refers to MYCN status at diagnosis, which is either amplified (amp, ≥8 copies) or normal. MYCN status was inconclusive for one tumor at diagnosis.fAll values are given in absolute number of tumors (No.) and relative (%) to the total number of tumors in each group (total cohort, C-Circle positive, C-Circle negative).gP values were calculated using Fisher tests and two-sided Wilcoxon rank sum test (age). + +a ALT frequency detected using C-Circle presence as a marker in the screening cohort (n = 720, left) and INFORM cohort (n = 40, right). Top pies illustrate the composition of the cohorts concerning INSS tumor stage. b Schematic overview of the analyzed patient cohorts used for C-Circle screening and high-throughput analysis using whole genome sequencing (WGS), RNA-sequencing, whole proteome analysis, and ChIP-sequencing. Sample numbers are given in brackets. Number of C-Circle positive (CC+) and C-Circle negative (CC-) patients in the sub-cohorts are indicated. + +Subsequently, neuroblastomas enriched for ALT were characterized using a multi-omics approach including high coverage whole genome sequencing (n = 165), RNA-sequencing (n = 144), whole proteome analysis (n = 34), and ChIP-sequencing (n = 27) (Fig. 1b, Supplementary Data 3, Methods) forming the discovery cohort. ALT tumors had a significantly higher telomere content (Fig. 2a). Telomere content and C-Circle intensity was comparable between ALT-positive tumors of the discovery and INFORM cohort (Supplementary Figs. 1–2). Overall, telomere content was positively correlated with C-Circle intensity (r = 0.76, P < 2.2e−16, Spearman, Supplementary Fig. 1c). Minimal to no TERT mRNA expression was observed for ALT-positive neuroblastomas (Fig. 2b), resulting from epigenetic silencing of the TERT locus by H3K27me3 (Supplementary Fig. 3a). In accordance with low TERT mRNA expression, ALT-positive tumors exhibited low telomerase activity (Supplementary Fig. 3b). Moreover, ALT was associated with increased expression of the polyadenylated telomeric long noncoding RNA TERRA (Fig. 2c, Supplementary Fig. 3c). Patients with ALT-positive tumors had a similar poor event-free survival compared to patients with MYCN-amplified tumors. Although the overall survival time of patients with ALT-positive tumors was significantly longer, they had a dismal outcome in the long perspective (Fig. 2d). No difference in event-free or overall survival was observed between patients with ALT-positive tumors stratified into high-risk and those predicted as low/intermediate risk (Fig. 2e), indicating that the current risk stratification is underestimating the poor prognosis of ALT-positive tumors. ALT-positive neuroblastomas with a higher telomere content had a significantly shorter event-free survival as compared to ALT-positive tumors with relatively lower telomere content (Supplementary Fig. 3d). There was no clear trend in telomere content gain or loss between matching primary/relapse samples of the INFORM cohort (Supplementary Fig. 4). For 3 of 7 ALT-positive relapse tumors with a matching primary, ALT activity was only observed in the relapse tumor. Two of these cases were MYCN amplified in the primary disease period and heterogeneous ALT/MNA in the relapse. Heterogeneity could be confirmed by FISH analysis (MYCN and telomere probe) of one relapsed tumor (Supplementary Fig. 5a). For the other heterogeneous relapse tumor, we found evidence of an ALT-positive subclone in FISH analysis of the primary tumor (Supplementary Fig. 5b). Noteworthy, the discovery cohort also contains three heterogeneous ALT/MNA cases, which are MYCN amplified and have a very high telomere content, but were otherwise C-Circle negative (Fig. 2a, Supplementary Figs. 1a and 5c).Fig. 2Characteristics and survival of ALT-positive neuroblastomas.a Relative telomere content (tumor/control) of neuroblastoma tumors in the discovery cohort grouped according to the genetic evidence of an active telomere maintenance mechanism (see methods for details): ALT activation (ALT, red), telomerase activation by amplified MYCN (MNA, yellow), structural variation affecting TERT or TERT promoter mutation (TERT, blue), and patients having no genetic evidence of an activated telomere maintenance mechanism (OTHER, gray). b +TERT mRNA expression in different subgroups. c Expression of the telomeric repeat-containing RNA TERRA in the different subgroups. a–c Boxplots indicate the median value (middle line), the 25th and 75th percentiles (box). The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Dots represent individual tumors. P values were calculated with two-sided Wilcoxon rank sum tests. d Event-free and overall survival probability compared between ALT-positive (n = 60) and MYCN-amplified (MNA, n = 34) neuroblastomas. e Event-free and overall survival probability of patients with ALT-positive tumors classified as low or intermediate risk (LR/IR, n = 20) compared to patients with ALT-positive tumors classified as high-risk (HR, n = 39) using the NB2004 risk stratification. d–e +P values were calculated using log rank tests. a–e +n describes the number of analyzed tumors. + +a Relative telomere content (tumor/control) of neuroblastoma tumors in the discovery cohort grouped according to the genetic evidence of an active telomere maintenance mechanism (see methods for details): ALT activation (ALT, red), telomerase activation by amplified MYCN (MNA, yellow), structural variation affecting TERT or TERT promoter mutation (TERT, blue), and patients having no genetic evidence of an activated telomere maintenance mechanism (OTHER, gray). b +TERT mRNA expression in different subgroups. c Expression of the telomeric repeat-containing RNA TERRA in the different subgroups. a–c Boxplots indicate the median value (middle line), the 25th and 75th percentiles (box). The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Dots represent individual tumors. P values were calculated with two-sided Wilcoxon rank sum tests. d Event-free and overall survival probability compared between ALT-positive (n = 60) and MYCN-amplified (MNA, n = 34) neuroblastomas. e Event-free and overall survival probability of patients with ALT-positive tumors classified as low or intermediate risk (LR/IR, n = 20) compared to patients with ALT-positive tumors classified as high-risk (HR, n = 39) using the NB2004 risk stratification. d–e +P values were calculated using log rank tests. a–e +n describes the number of analyzed tumors. + +The mutational landscape of ALT-positive neuroblastomas in the discovery cohort was very diverse with only few recurrent genetic events (Fig. 3a, Supplementary Fig. 6). ALT activation was mutually exclusive to TERT rearrangements (Fig. 3a). Loss-of-function mutations of ATRX were the most frequent events associated with ALT, but were only observed in 55% of ALT-positive tumors (Fig. 3a). ATRX mutations were mainly large deletions and single nucleotide variants (Fig. 3a–b, Supplementary Data 3–4). However, three patients had large intragenic ATRX duplications (Fig. 3a–b, Supplementary Data 3–4). Patient survival was not significantly different between ATRX-mutated and ATRX wild-type ALT-positive cases (Fig. 3c). Mutations in the ATRX complex members DAXX and H3F3A were extremely rare (Fig. 3a). Only one ATRX wild-type tumor exhibited a large inversion event affecting DAXX, associated with low DAXX mRNA expression (66.46 TPM; range discovery cohort 50-330 TPM). Another tumor had an A92S H3F3A mutation in addition to mutated ATRX. Somatic mutations in TP53 pathway genes (TP53, CREBBP, ATM, ATR, CDKN2A, and MDM2) were significantly enriched (P = 0.01) in ALT-positive tumors. ALT-positive tumors had the highest prevalence of CDK4 amplifications (3/4) (Fig. 3a), associated with increased CDK4 mRNA expression (Supplementary Fig. 7a), and showed higher CCND1 mRNA expression as compared to telomerase-activated cases (Supplementary Fig. 7b). Overall, many tumors had mutations in one or another gene associated with telomere maintenance according to the TelNet database22 (Fig. 3a, Supplementary Figs. 6 and 7c). Copy number loss of POLD3 and ATM was frequently observed in ALT-positive tumors (Supplementary Fig. 7c). However, both POLD3 and ATM are located on chr11q and loss of these genes was associated with loss of chr11q, which is frequently observed in ALT-positive (Supplementary Fig. 8) and other non-MYCN-amplified neuroblastomas23,24. Synaptic nuclear envelope protein 1 (SYNE1) was exclusively mutated in 7 of 60 ALT tumors (11.7%) (Fig. 3a, Supplementary Figs. 6, 9a). Furthermore, deletions in receptor-type tyrosine-protein phosphatase delta (PTPRD) were enriched in ALT-positive tumors (P = 2.46e-05; Fig. 3a, Supplementary Fig. 6). Structural variations affecting PTPRD were reported in neuroblastoma cohorts and low PTPRD expression was correlated with a poor prognosis25,26. However, structural variations of PTPRD were not correlated with reduced PTPRD mRNA expression in our discovery cohort (Supplementary Fig. 9b). Noteworthy, both genes are very large and thus the high mutation frequency in ALT neuroblastomas could also reflect the general increased mutational load found in older patients or a general increase in genomic instability (Supplementary Fig. 9c–e). The mutational landscape of relapsed neuroblastomas in the INFORM cohort was comparable to the discovery cohort (Supplementary Fig. 10a, Supplementary Data 2, 3). In one matched primary relapse pair, an ATRX duplication event was retained in the relapse, further indicating that duplications can lead to nonfunctional ATRX (Supplementary Fig. 4). Notably, canonical activating RAS pathway mutations (HRAS, NRAS, KRAS, BRAF, RAF1, CDK4, CCND1, NF1) were significantly more frequent in relapsed ALT-positive neuroblastomas in the INFORM cohort compared to the discovery cohort (P = 0.0013, Supplementary Fig. 10, Fig. 3a), supporting a specific impact of RAS pathway mutations in relapsed ALT-positive tumors.Fig. 3Mutational landscape of ALT-positive neuroblastomas.a Somatic alterations in the discovery cohort. Samples ordered based on relative telomere content. Exonic single nucleotide variations (SNVs), exonic small insertions and deletions (INDELs) and structural variations in exonic and intronic regions (SVs) in selected ALT-associated genes, genes associated with telomere biology according to the TelNet database22 [https://malone2.bioquant.uni-heidelberg.de/fmi/webd/TelNet] and TP53 pathway genes are illustrated. Telomere maintenance features, clinical features, chromosomal aberrations, and ALK/RAS pathway mutations (HRAS, NRAS, KRAS, BRAF, NF1 or ALK) are given in the top panel. b Exact genetic location of mutations affecting the ATRX gene. Bars represent structural variations (deletion, duplication) and small deletions, while SNVs are illustrated as lollipops and colored by the SNV type. c Event-free and overall survival rate of patients with an ATRX-mutated ALT-positive neuroblastoma (n = 33) compared to patients with an ATRX wild-type ALT-positive neuroblastoma (n = 27). P values were calculated using a log rank test. n describes the number of analyzed tumors. + +a Somatic alterations in the discovery cohort. Samples ordered based on relative telomere content. Exonic single nucleotide variations (SNVs), exonic small insertions and deletions (INDELs) and structural variations in exonic and intronic regions (SVs) in selected ALT-associated genes, genes associated with telomere biology according to the TelNet database22 [https://malone2.bioquant.uni-heidelberg.de/fmi/webd/TelNet] and TP53 pathway genes are illustrated. Telomere maintenance features, clinical features, chromosomal aberrations, and ALK/RAS pathway mutations (HRAS, NRAS, KRAS, BRAF, NF1 or ALK) are given in the top panel. b Exact genetic location of mutations affecting the ATRX gene. Bars represent structural variations (deletion, duplication) and small deletions, while SNVs are illustrated as lollipops and colored by the SNV type. c Event-free and overall survival rate of patients with an ATRX-mutated ALT-positive neuroblastoma (n = 33) compared to patients with an ATRX wild-type ALT-positive neuroblastoma (n = 27). P values were calculated using a log rank test. n describes the number of analyzed tumors. + +Using a mass spectrometry based whole proteome approach (n = 34) quantifying on average 6891 proteins per tumor, we found 470 proteins significantly different in ALT-positive tumors compared to the other subgroups (nominal P ≤ 0.01; fold change ≥2; Fig. 4a, Supplementary Data 5). Enrichment analysis of significantly different mRNAs and proteins for each subgroup in turn highlighted both the overall agreement of transcriptomics and proteomics as well as the distinct effects only significantly present in one or the other omics-level (Fig. 4b, Supplementary Fig. 11a, Supplementary Data 6). The Ras family proteins HRAS and NRAS, comprised in the term “prenylation”, were significantly upregulated in ALT-positive neuroblastomas (Fig. 4b, Supplementary Data 6). Among the top upregulated proteins in ALT-positive tumors were the cancer testis antigens P antigen family member 5 (PAGE5) and melanoma associated antigen 4 (MAGEA4), which may serve as potential targets for immunotherapies (Fig. 4a). Many pathways associated with proliferation were significantly downregulated on protein level in ALT-positive tumors compared to the telomerase-activated tumors including “DNA replication” and “chromosome”. Notably, the classical proliferation marker MKI6727, comprised in the term “chromosome”, was one of the top significantly downregulated proteins in ALT-positive neuroblastomas (Fig. 4a–b, Supplementary Fig. 11b). The term DNA replication includes among others the proteins PCNA, MCM2, MCM3, MCM4, MCM5, and MCM7, which are also indicative of the proliferative capacity of a cell28. Furthermore, the term “bromodomain” was significantly reduced in ALT-positive tumors, while strongly upregulated in MNA tumors. Amplified MYCN is known to be associated with increased occupancy of active promoter regions and enhancer invasion by MYCN and bromodomain proteins, leading to an increased transcriptional activity of many proliferation associated genes and downregulation of differentiation genes29. Moreover, ALT-positive tumors exhibit a significantly lower fraction of cycling cells compared to MYCN-amplified tumors (Supplementary Fig. 11c). Taken together, both protein expression data and cell cycle analysis support a low proliferative capacity of ALT-positive neuroblastomas. 96 of the 470 significantly different proteins were annotated in the TelNet database to be associated with telomere maintenance22 (Supplementary Data 5) (P = 0.03). The significantly downregulated proteins in ALT-positive neuroblastomas comprised ATRX, PCNA, EXO1, GNL3, KDM1A, RIF1, SCG2, TFAP2B, and GNL3L, all functionally linked to telomere maintenance22. Helicases and chromatin regulators, including ATRX, were less abundant at protein level in the ALT subgroup, but not at mRNA level (Fig. 4b). Importantly, ATRX itself was among the top candidates to exhibit lower protein abundance in ALT-positive tumors (Fig. 4a). Intriguingly, the reduced ATRX protein levels were independent of ATRX mRNA levels and mutation status (Fig. 5a). ATRX was among the most strongly downregulated proteins while the ATRX mRNA changes very little (Supplementary Fig. 11d), indicating that reduced ATRX protein abundance is a characteristic proteomic feature of all ALT-positive tumors (ATRX wild-type and mutated) that cannot be observed at the mRNA level. Analysis of exon-specific ATRX mRNA levels revealed that only the exons affected by deletion have reduced expression in the ATRX-deleted cases (Supplementary Fig. 12a). In summary, reduced ATRX protein level is a biomarker for ALT-positive neuroblastomas that is independent of mRNA levels and ATRX mutation. The decreased ATRX protein levels despite unchanged mRNA abundance could be due to reduced translation and/or increased degradation of ATRX in ALT-positive tumors. For example, it has been shown that unassembled subunits of multiprotein complexes (so-called orphans) are often degraded30. Downregulating one subunit can therefore induce degradation of other subunits, which partially explains divergence of protein- and mRNA-level changes in many biological contexts31,32. We therefore took a closer look at the ATRX binding partner DAXX. Interestingly, DAXX protein levels were specifically reduced in ALT-positive ATRX wild-type tumors (Fig. 5b). To investigate if reduced DAXX levels could explain ATRX downregulation at the protein level, we knocked-down DAXX in the ALT-negative neuroblastoma cell line NBL-S and observed that ATRX protein levels were indeed reduced after 96 h (Fig. 5c–d). Hence, the reduced ATRX protein levels in ATRX wild-type ALT positive tumors might result from downregulation of the DAXX protein. However, since DAXX mRNA levels do not differ significantly between tumors (Fig. 5b) and no recurrent mutation patterns in DAXX or ATRX/DAXX interacting proteins could be identified in ALT-positive ATRX wild-type tumors (Supplementary Figs. 12b, 13), the question what causes the downregulation of DAXX is still open. Hence, the mechanistic details behind ATRX/DAXX complex reduction remain to be uncovered.Fig. 4Proteomic analysis of ALT-positive neuroblastomas.a Label free quantification (LFQ) analysis comparing ALT tumors with MNA and TERT tumors as well as tumors without a genetic telomere maintenance mechanism (OTHER). P values were calculated using a two-sided Student’s t-test. Colors indicate P value and fold change categories as well as the association to telomere biology based on the TelNet database22 [https://malone2.bioquant.uni-heidelberg.de/fmi/webd/TelNet]. b Differences between neuroblastoma subtypes (ALT, TERT, MNA, OTHER). UniProt keywords ([https://www.uniprot.org/keywords/]; retrieved on 2016-11-02) with significant enrichments (BH adjusted P < 0.000005, at least 2.5-fold enriched, at least 3 protein/mRNAs per keyword) in at least one group with significantly more or less abundant mRNA or proteins comparing each group to all other samples in turn (BH adjusted P < 0.05, * marks significant enrichment). Positive and negative sqrt enrichment values for enrichments in more and less abundant mRNA or protein, respectively. Enrichment P values were calculated by hypergeometric test and multiple-testing adjusted by Benjamini–Hochberg method. Exact P values and individual proteins contributing to Uniprot keywords are given in Supplementary Data 6.Fig. 5ATRX/DAXX complex abundance.a ATRX protein and ATRX mRNA expression in ALT ATRX-mutated (dark pink) and wild-type (light pink) tumors compared to TERT (blue), MNA (yellow), and OTHER (gray). b Boxplots of DAXX protein and DAXX mRNA expression in ALT ATRX-mutated (dark pink) and wild-type (light pink) tumors compared to TERT (blue), MNA (yellow), and OTHER (gray). a, b Boxplots indicate the median value (middle line) and the 25th and 75th percentiles (box). The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Dots represent individual tumors. P values in boxplots were calculated with two-sided Wilcoxon rank sum tests. n describes the number of analyzed tumors. c, d ATRX protein expression in NBL-S cells (with moderate MYCN expression) after 96 h treatment with siRNAs against DAXX (black). ATRX siRNAs as well as two negative control siRNAs were used as controls (gray). Exemplary western blot (c) and quantification of three biological replicates (d) is shown. P values calculated using a two-sided Student’s t-test. Horizontal line shows median value. Dots represent biological replicates. Uncut images of all biological replicates are shown in the source data file. + +a Label free quantification (LFQ) analysis comparing ALT tumors with MNA and TERT tumors as well as tumors without a genetic telomere maintenance mechanism (OTHER). P values were calculated using a two-sided Student’s t-test. Colors indicate P value and fold change categories as well as the association to telomere biology based on the TelNet database22 [https://malone2.bioquant.uni-heidelberg.de/fmi/webd/TelNet]. b Differences between neuroblastoma subtypes (ALT, TERT, MNA, OTHER). UniProt keywords ([https://www.uniprot.org/keywords/]; retrieved on 2016-11-02) with significant enrichments (BH adjusted P < 0.000005, at least 2.5-fold enriched, at least 3 protein/mRNAs per keyword) in at least one group with significantly more or less abundant mRNA or proteins comparing each group to all other samples in turn (BH adjusted P < 0.05, * marks significant enrichment). Positive and negative sqrt enrichment values for enrichments in more and less abundant mRNA or protein, respectively. Enrichment P values were calculated by hypergeometric test and multiple-testing adjusted by Benjamini–Hochberg method. Exact P values and individual proteins contributing to Uniprot keywords are given in Supplementary Data 6. + +a ATRX protein and ATRX mRNA expression in ALT ATRX-mutated (dark pink) and wild-type (light pink) tumors compared to TERT (blue), MNA (yellow), and OTHER (gray). b Boxplots of DAXX protein and DAXX mRNA expression in ALT ATRX-mutated (dark pink) and wild-type (light pink) tumors compared to TERT (blue), MNA (yellow), and OTHER (gray). a, b Boxplots indicate the median value (middle line) and the 25th and 75th percentiles (box). The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Dots represent individual tumors. P values in boxplots were calculated with two-sided Wilcoxon rank sum tests. n describes the number of analyzed tumors. c, d ATRX protein expression in NBL-S cells (with moderate MYCN expression) after 96 h treatment with siRNAs against DAXX (black). ATRX siRNAs as well as two negative control siRNAs were used as controls (gray). Exemplary western blot (c) and quantification of three biological replicates (d) is shown. P values calculated using a two-sided Student’s t-test. Horizontal line shows median value. Dots represent biological replicates. Uncut images of all biological replicates are shown in the source data file. + +Many contradictory findings were reported on the telomeric chromatin landscape of ALT-positive cells33–35, but data on primary tumor samples was yet missing. Here, ALT-positive neuroblastoma tumors were found to be enriched for H3K9me3 at the telomeres (Fig. 6a). Overall levels of H3K9me3 at the telomeres of ALT-positive tumors were comparable to the H3K9me3 levels at SatII and SatIII alpha-satellite sequences (Supplementary Fig. 14a–b), which are known to be enriched for H3K9me335,36. In line with a heterochromatic enrichment, protein expression of the demethylase KDM1A responsible for removing methyl groups from H3K9me337 was significantly lower in ALT-positive neuroblastomas compared to TERT-activated tumors (Fig. 6b). SETDB1 protein activity is required for H3K9me3 deposition at the telomeres, while the histone methyltransferases SUV39H1 and SUV39H2 catalyze H3K9 trimethylation at heterochromatic sites38–40. Neither expression of SETDB1 nor SUV39H1/SUV39H2 could be associated with the increased telomeric H3K9me3 in ALT-positive neuroblastomas (Fig. 6c–d). H3K27ac at the telomeres was lower in ALT-positive tumors (Fig. 6a). However, this difference was not telomere specific and similarly observed at SatII and SatIII sequences indicating a general difference in H3K27ac (Supplementary Fig. 14c). Furthermore, ALT-positive tumors displayed reduced telomeric H3K27me3 (Fig. 6a, Supplementary Fig. 14d) and low protein and mRNA expression of the PRC2 component EZH2, responsible for H3K27me3 deposition (Fig. 6e). Low H3K36me3 was observed at the telomeres and satellite sequences of both ALT-positive and ALT-negative neuroblastomas (Fig. 6a, Supplementary Fig. 14e). Together, this indicates epigenetic dysregulation of repressive marks at the telomeres in ALT-positive tumors with excess of H3K9me3 and reduction of H3K27me3.Fig. 6Telomeric chromatin state of ALT-positive neuroblastomas.a Telomere repeats in ChIP of H3K9me3 (n = 26), H3K27ac (n = 25), H3K27me3 (n = 25) and H3K36me3 (n = 27) relative to telomere repeats in input of ALT-positive and ALT-negative tumors. b–e Protein and mRNA expression of KDM1A (b), SETDB1 (c), SUV39H1/H2 (d, only RNA), and EZH2 (e) in ALT tumors compared to tumors form the other groups (TERT, MNA, OTHER). a–e Colors indicate subgrouping based on presence and type of telomere maintenance mechanism (see legend). Boxplots indicate the median value (middle line) and the 25th and 75th percentiles (box). The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Dots represent individual tumors. P values were calculated using a two-sided Wilcoxon rank sum test. n describes the number of analyzed tumors. + +a Telomere repeats in ChIP of H3K9me3 (n = 26), H3K27ac (n = 25), H3K27me3 (n = 25) and H3K36me3 (n = 27) relative to telomere repeats in input of ALT-positive and ALT-negative tumors. b–e Protein and mRNA expression of KDM1A (b), SETDB1 (c), SUV39H1/H2 (d, only RNA), and EZH2 (e) in ALT tumors compared to tumors form the other groups (TERT, MNA, OTHER). a–e Colors indicate subgrouping based on presence and type of telomere maintenance mechanism (see legend). Boxplots indicate the median value (middle line) and the 25th and 75th percentiles (box). The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Dots represent individual tumors. P values were calculated using a two-sided Wilcoxon rank sum test. n describes the number of analyzed tumors. + +ALT-positive neuroblastomas exhibited an increased frequency of telomeric repeat loci compared to the other groups (Fig. 7a), with ALT-positive tumors displaying up to 19 such events per tumor. The telomeric repeat loci frequency was significantly higher in ATRX-mutated than ATRX wild-type ALT tumors. Telomeric repeat loci are characterized by a chromosomal junction site and the presence of telomeric repeat sequences either upstream or downstream of this junction site (one-sided) (Fig. 7b). In rare cases, two telomeric repeat loci in close proximity (10 kb) with opposite orientation can form a two-sided event (Supplementary Fig. 15a). In our discovery cohort, only five such two-sided events were observed (Supplementary Fig. 15b). Two-sided events can result either from two separate events in close proximity or from telomere sequences that are spanning the region between the two junction sites forming a true insertion. Evidence for a true insertion, meaning that mates of a read pair matched to both sides of the insertion could only be identified for two of these events (Supplementary Table 1). The presence of more than three telomeric repeat loci in a tumor was found to be associated with a significantly reduced overall survival in ALT-positive neuroblastomas (Fig. 7c). Hotspot regions for recurrent telomeric repeat loci were found on chromosomes 1q42.2, 18q23, and 19q13.43 (Fig. 7d). The increased occurrence of telomeric repeat loci in ALT-positive neuroblastomas and their enrichment on 1q42.2 was confirmed in the INFORM dataset (Supplementary Fig. 15c). 80.3% (212/264) of all t-type (TTAGGG) containing telomeric repeat loci showed at least 1 bp microhomology to the canonical telomeric repeat sequence, which is more than one would expect by chance. This enhanced microhomology at the junction sites of telomeric repeat loci indicates that telomeric sequences were added via a microhomology-dependent mechanism (Supplementary Fig. 15d-f). In general, 55.3% of all telomeric repeat loci were associated with a chromosomal copy number loss of the neighboring segment (Supplementary Fig. 15g). A smaller fraction of loci (4.9%) was associated with a copy number gain. However, 30.7% of the telomeric repeat loci were copy number neutral. Some of these copy number neutral events were associated with one or more structural variations, but for 77.2% of the copy number neutral telomeric repeat loci no other event could be detected (Supplementary Fig. 15g).Fig. 7Telomeric repeat loci.a Telomeric repeat loci frequency in ATRX-mutated (dark pink) and ATRX wild-type (light pink) ALT-positive tumors compared to TERT (blue), MNA (MNA), and tumors without genetic evidence of an activated telomere maintenance mechanism (OTHER, gray) in the discovery cohort. Boxplots indicate the median value (middle line) and the 25th and 75th percentiles (box). The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Dots represent individual tumors. n describes the number of analyzed tumors. P values were calculated with two-sided Wilcoxon rank sum tests. b Example of one telomeric repeat locus in tumor NBD119. Dark blue color indicates sequencing coverage and light blue color illustrates clipped sequences. Reads are shown in medium gray and clipped bases are color-coded. Non-telomeric ends of a discordant read pair are labeled in dark gray. c Overall survival probability of ALT-positive tumors harboring more than three telomeric repeat loci compared to ALT-positive tumors with three or less than three events. P value calculated using a log rank test. d Chromosomal location of telomeric repeat loci in the discovery cohort. e Exact genomic location of events contributing to the hotspot on chromosome 1q42.2 in the discovery and INFORM cohort and two ALT-positive cell lines. The 1q42.2 events of NBD158 and NBD137 ware added manually (see Methods). f Chr1 copy number profiles of tumors with a telomeric repeat locus on chr1q42.2. Junction sites and orientation of telomeric repeats are indicated in orange. Color-coding according to subgroup (# HET ALT/MNA). g Differential mRNA expression of genes located in the chr1q42.2-1qter deleted region comparing 1q42-1qter deleted to non-deleted tumors. Genes are colored by their mean expression normalized by exonic length or by their presence in the TelNet database22. *) Matched primary relapsed pair. + +a Telomeric repeat loci frequency in ATRX-mutated (dark pink) and ATRX wild-type (light pink) ALT-positive tumors compared to TERT (blue), MNA (MNA), and tumors without genetic evidence of an activated telomere maintenance mechanism (OTHER, gray) in the discovery cohort. Boxplots indicate the median value (middle line) and the 25th and 75th percentiles (box). The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Dots represent individual tumors. n describes the number of analyzed tumors. P values were calculated with two-sided Wilcoxon rank sum tests. b Example of one telomeric repeat locus in tumor NBD119. Dark blue color indicates sequencing coverage and light blue color illustrates clipped sequences. Reads are shown in medium gray and clipped bases are color-coded. Non-telomeric ends of a discordant read pair are labeled in dark gray. c Overall survival probability of ALT-positive tumors harboring more than three telomeric repeat loci compared to ALT-positive tumors with three or less than three events. P value calculated using a log rank test. d Chromosomal location of telomeric repeat loci in the discovery cohort. e Exact genomic location of events contributing to the hotspot on chromosome 1q42.2 in the discovery and INFORM cohort and two ALT-positive cell lines. The 1q42.2 events of NBD158 and NBD137 ware added manually (see Methods). f Chr1 copy number profiles of tumors with a telomeric repeat locus on chr1q42.2. Junction sites and orientation of telomeric repeats are indicated in orange. Color-coding according to subgroup (# HET ALT/MNA). g Differential mRNA expression of genes located in the chr1q42.2-1qter deleted region comparing 1q42-1qter deleted to non-deleted tumors. Genes are colored by their mean expression normalized by exonic length or by their presence in the TelNet database22. *) Matched primary relapsed pair. + +Telomeric repeat loci contributing to the hotspot on chr1q42.2 (230.370 mb – 231.433 mb, Fig. 7e) frequently coincided with loss of the adjacent segment of chr1q (Fig. 7f). Telomeric repeat loci at 1q42.2 showed a high degree of microhomology to the telomeric TTAGGG repeats (Supplementary Figs. 15e, 16a). Loss of chromosome 1q42.2-1qter was also observed in some tumors without the occurrence of telomeric repeats at this site. However, 1q42.2-1qter deletions were exclusively seen in ALT-positive tumors, most of which had an ATRX mutation (Supplementary Figs. 16b, 8). RNA expression of most genes in the deleted 1q region was lower in 1q42.2-1qter deleted compared to non-deleted tumors (Fig. 7g). Among the significantly reduced genes in 1q42.2-1qter deleted tumors were the genes ARID4B, EXO1, FH, CNST, ADSS, HNRNPU, and RYR2, which are known to be associated with telomere maintenance and biology22. Of note, expression of the 5′−3′ exonuclease, EXO1, located on chromosome 1q43 and involved in DNA double strand break repair41,42, telomere replication, and t-loop formation43, was not only reduced in 1q42.2-deleted tumors but generally reduced in ALT-positive neuroblastomas (Supplementary Fig. 16c). + +In summary, this in-depth analysis revealed that ALT-positive neuroblastomas are biologically and clinically distinct from tumors with telomerase activation. + +ALT in neuroblastoma was associated with characteristic ALT features, including high telomere content, low TERT expression, and increased TERRA expression, which is in line with previous findings on ALT-positive tumors44,45. ATRX mutations were identified in 55% of the ALT-positive neuroblastomas. In line with this observation, an ATRX mutation frequency of 60% was reported in a parallel study46. The most frequent type of ATRX mutations were large deletions. However, we could also identify focal intragenic ATRX duplication events. Focal duplication of ATRX was also observed in a matching primary/relapse pair indicating that these events are leading to loss of ATRX. Similar events resulted in loss of ATRX function in patients with the genetic disorder alpha thalassemia X-linked intellectual disability syndrome (ATRX-syndrome)47,48 further supporting a loss of ATRX function by partial duplications. + +Integrating proteomic profiling revealed reduced ATRX/DAXX protein complex abundance as a recurrent event in ALT-positive neuroblastoma, which could often not be explained by mutations in these genes. The observation that all five ALT-positive tumors with wild-type ATRX depicted reduced DAXX protein levels is intriguing. It is tempting to speculate that the ALT phenotype always results from loss of the ATRX/DAXX complex activity, which is caused either by ATRX mutations or by reduced ATRX/DAXX protein levels. Future studies will show if this is indeed the case in neuroblastoma and/or possibly other ALT-positive cancers. Nevertheless, we could identify a subgroup of ALT-positive neuroblastomas with wild-type ATRX, but low ATRX protein abundance. Reduced ATRX protein levels in ALT ATRX wild-type neuroblastoma could be explained by the reduced DAXX protein levels, which we specifically observed in this subgroup of tumors. In this scenario, reduced DAXX protein levels impair assembly of the ATRX/DAXX complex, which then causes degradation of orphan ATRX protein molecules. Consistent with this idea, we observed that knocking down DAXX reduced ATRX protein levels in neuroblastoma cells. However, the cause of reduced DAXX protein levels is not yet clear, especially since mRNA levels do not change significantly. Reduced DAXX protein levels in ALT-positive tumors may result from posttranscriptional events. It is known that DAXX is regulated via various posttranslational modifications including phosphorylation, SUMOylation, and ubiquitination49. Irrespective of the mechanistic details involved, our study highlights that proteomic data is closer to phenotypes than transcriptomic data, which is especially valuable in a clinical context32. + +A recent report stated that ATRX mutation and amplified MYCN are mutually exclusive50. C-Circle screening of a cohort of 720 neuroblastomas also found a mutual exclusivity of C-Circle presence and amplified MYCN. However, in the INFORM cohort we identified two cases with amplified MYCN and a positive C-Circle signal. Additionally, in our discovery cohort we classified three cases as ALT/MNA heterogeneous. These cases were negative in the C-Circle assay, but exhibited a very high telomere content. FISH analysis of some of these cases could also confirm that amplified MYCN and ultra-bright telomere spots are present in the same tumor (Supplementary Fig. 5). No ATRX mutations were detected in the MNA/ALT heterogeneous tumors (for NBI26 only lcWGS and WES data was available). Taken together, heterogeneous tumors with amplified MYCN and presence of ALT markers seem to occur, but are very rare. Further analysis and larger sets of these cases would be necessary to draw meaningful conclusions on the clonality of these events and the activity of the respective telomere maintenance mechanism. + +ChIP-sequencing analysis of ALT-positive neuroblastoma tumors showed that telomeres as well as SatII and SatIII sequences show lower H3K27ac compared to ALT-negative tumors. Furthermore, proteins comprised in the term “bromodomain”, including EP300 responsible for H3K27ac, were reduced in ALT tumors and enhanced in telomerase-activated tumors. Moreover, the telomeres of ALT-positive neuroblastoma tumors were enriched for H3K9me3. These observations confirm the results of a previous study on cell lines35 that only the telomeres of ALT-positive tumors are enriched for H3K9me3. In addition, we observed low H3K27me3 at the telomeres of ALT cases, which goes in line with a significantly lower protein and mRNA expression of EZH2 in ALT neuroblastomas compared to telomerase-activated tumors. + +ALT-positive neuroblastoma tumors exhibited a high rate of telomeric repeat loci. Since telomeric repeat loci were characterized by telomeric repeat sequences either upstream or downstream of a non-telomeric junction site, these events cannot have occurred from breakage of interstitial telomeric sequences (ITS). However, telomeric repeat loci frequently overlapped with breakpoints of copy number changes or structural variations. Because terminal chromosomal breaks are in need of telomeric repeats to protect the newly formed ends from degradation51–53, we propose that ALT-positive tumors are capable of adding telomeric repeat sequences to open ends of chromosomal breaks forming neo-telomeres (Fig. 8a). The high degree of microhomology to the telomeric repeat sequence at the junction sites indicates that telomeric repeats are added via a microhomology-dependent process like microhomology-mediated end-joining or non-homologous end-joining54. Further, we propose that the presence of microhomology at an open chromosomal break determines if telomeric sequences can be added to this site. Chromosomal loss of certain fragments might present a selection advantage leading to a selection of cells harboring the neo-telomere. We also identified a subset of copy number neutral telomeric repeat loci with no associated structural variation. This might be due to the fact that these events are subclonal and were thus not detected by the CNV/SV calling algorithm. Moreover, the detection limit of the used copy number algorithm is 50 kb and thus smaller copy number changes cannot be detected. Two-sided events, defined as two telomeric repeat loci in a 10 kb window with opposite orientation of the telomeric repeats, were very rare. These events may represent insertions of telomeric repeat sequences similar to previously described events55. Only two of five two-sided events exhibited evidence of a true insertion by mates of a read pair mapping to both sides of the insertion. However, for large insertions the used short read sequencing prevents the detection of supporting reads. Alternatively, two-sided events may result from neo-telomere formation on both sides of a breakpoint (Fig. 8b). A recurrent hotspot for telomeric repeat loci was identified on chr1q42, which was associated with 1q42.2-1qter loss forming a neo-telomere on chr1q42.2. Telomeric repeat loci were also associated with ALT and overlapped with chromosomal breakpoints in a pan-cancer cohort, but no enrichment was seen for telomeric repeat loci on 1q42.245, indicating that neo-telomere formation on chromosome 1q42.2 is neuroblastoma-specific. 1q42.2-1qter deletions retained in relapsed ALT cases indicate a selective advantage of the deletion. Furthermore, chr1q42.2-1qter deletions were associated with reduced expression of candidate genes involved in telomere biology, including EXO1. Interestingly, EXO1 expression is not only reduced in ALT-positive tumors with 1q42.2-1qter deletions, but in ALT tumors in general. This indicates a potential role of EXO1 as an ALT suppressor that may be silenced by loss of 1q42.2 or other means.Fig. 8Model of neo-telomere formation.Graphical abstract illustrating the hypothesis of neo-telomere formation in ALT-positive neuroblastomas. a The majority of telomeric repeat loci are one-sided. We hypothesize that ALT-positive cells are able to add telomeric sequences to open chromosomal breaks to protect them from degradation. I Microhomology to TTAGGG favors the formation of a neo-telomere at the open chromosomal break. Loss of some chromosomal regions might present a selection advantage for the cell and cells with a neo-telomere at the chromosomal breakpoint are selected. II Without microhomology, structural rearrangements involving other chromosomal arms (light gray) can represent an alternative route of protecting open chromosomal breaks. b Rare two-sided events can result from insertion of telomeric sequences I or from the formation of neo-telomeres on both sides of the breakpoint II. Small circles represent C-Circles. Chromosomes are shown in gray and telomeric sequences in blue. + +Graphical abstract illustrating the hypothesis of neo-telomere formation in ALT-positive neuroblastomas. a The majority of telomeric repeat loci are one-sided. We hypothesize that ALT-positive cells are able to add telomeric sequences to open chromosomal breaks to protect them from degradation. I Microhomology to TTAGGG favors the formation of a neo-telomere at the open chromosomal break. Loss of some chromosomal regions might present a selection advantage for the cell and cells with a neo-telomere at the chromosomal breakpoint are selected. II Without microhomology, structural rearrangements involving other chromosomal arms (light gray) can represent an alternative route of protecting open chromosomal breaks. b Rare two-sided events can result from insertion of telomeric sequences I or from the formation of neo-telomeres on both sides of the breakpoint II. Small circles represent C-Circles. Chromosomes are shown in gray and telomeric sequences in blue. + +ALT activity was associated with unfavorable outcome and a protracted clinical course of disease, which further substantiates previous observations56. Furthermore, ALT was enriched in a population-based clinical sequencing study cohort for relapsed neuroblastomas. Multiple factors may contribute to this enrichment in the INFORM cohort. The INFORM inclusion criteria might contribute to an underrepresentation of the most aggressive neuroblastomas, which are most likely MYCN-amplified or TERT-rearranged, since patients are required to have sufficient general condition and a life expectancy of at least three months to be included in the registry trial. Furthermore, we could show risk underestimation of a substantial proportion of ALT-positive tumors, which was initially diagnosed as low or intermediate risk, but nevertheless had poor outcome. The current standard treatment regimens targeting strongly proliferating tumors are probably not suited to treat slowly growing ALT tumors, which might lead to a selection and enrichment of ALT-positive cases or subclones in the relapse setting. Thus, ALT activation might present a therapy resistance mechanism. Analysis of matching primary/relapse pairs revealed a gain of ALT or outgrowth of an ALT-positive subclone in the relapse tumor. Taken together, a revised therapy concept based on a better molecular understanding of this subtype and considering ALT-specific vulnerabilities is urgently needed for these children. + +We retrospectively analyzed cohorts of primary and relapsed neuroblastoma tumors. ALT-activity was determined using C-Circle assays for 720 neuroblastoma tumors (screening cohort, Supplementary Data 1). High coverage whole genome sequencing (WGS) data was obtained for a subset of 165 tumors (discovery cohort). RNA-sequencing (n = 144), whole proteome analysis (n = 34), and ChIP-sequencing (n = 27) was done for subsets of the discovery cohort. A summary of the clinical information and available data for every patient in the discovery cohort is given in Supplementary Data 3. Patients were enrolled in the neuroblastoma clinical trials (NB2004-HR/NCT03042429, GPOH-NB2004/NCT00410631, GER-GPOH-NB97/NCT00017225, GER-NB95-S/NCT00002803, GER-NB90/NCT00002802, NB2016-registry) of the Society for Pediatric Oncology and Hematology (GPOH) between 1991 and 2018. All trials were approved by the Ethics Committee of the Medical Faculty, University of Cologne and collection and use of all tumor tissue material was approved. Tumors of nine Austrian neuroblastoma patients were analyzed as part of the screening and discovery cohort. Tumor and control DNA/RNA was provided by Prof. Dr. Peter Ambros and Dr. Sabine Taschner-Mandl from the St. Anna Kinderkrebsforschung at the Children’s Cancer Research Institute in Vienna, Austria. Austrian neuroblastoma patients were in part enrolled in the NB-A-8757 or the HR-NBL-1/SIOPEN58 trial (NCT01704716, EudraCT number 2006-001489-17). The copy number status of MYCN using FISH analysis was assessed as a routine diagnostic marker in all neuroblastoma clinical trials. Survival data for German patients of the discovery cohort was provided by Dr. Barbara Hero from the study office of the Neuroblastoma trial in Cologne. Additionally, a relapsed neuroblastoma cohort of patients enrolled in the registry trial INFORM21 was analyzed (n = 40). Whole exome sequencing and low coverage whole genome sequencing was done as part of the INFORM workflow21. For a subset of these (n = 31) high coverage whole genome sequencing was done. A detailed overview of all clinical parameters and available data of the INFORM cohort is given in Supplementary Data 2. All patients or their legal guardian signed a written informed consent. Genomic DNA was isolated from fresh frozen material using a phenol chloroform extraction and control DNA was isolated from whole blood using the NucleoSpin Blood DNA extraction kit (Macherey-Nagel). + +C-Circle assays were performed based on a previously described protocol20, which is briefly described in the following. 30 ng of genomic DNA were used for every tumor sample. Prior to amplification, tumor DNA was digested with RsaI (4U/ug) and HinfI (4U/ug) and RNAse (25 ng/μg) treated for 1 h at 37 °C. 0.2 mg/mL BSA, 0.1% Tween 20, 1 mM dATP, dGTP and dTTP, 2 x Φ29 buffer and 7.5 U Φ29 polymerase was added to the pretreated DNA and rolling circle amplification was done at 30 °C for 8 h, followed by an inactivation at 65 °C for 20 min. For every sample, a sample without polymerase was included as control and on every blot the neuroblastoma cell lines CHLA-90 (ALT-positive) and SK-N-BE2 (ALT-negative) were included as positive and negative assay controls. TeloTTAGGG kit (Roche) was used for hybridization and development according to the manufacturers’ instructions. All blots were quantified using the ImageJ gel quantification tool (ImageJ version 1.51). A C-Circle positive signal was defined as minimum 4-fold increase of the area under the curve relative to control without polymerase. In addition, the signal intensity had to be at least 20% of the CHLA-90 positive control. C-Circle status was assessed by quantification and optical inspection of all blots by two investigators. + +High coverage whole genome sequencing libraries were sequenced on the Illumina HiSeq 2000 (100 bp paired end) or on a patterned flowcell v.2.5 (150 bp paired end) with coverage of 30–60x for the tumor and whole blood control samples. Tumor samples exhibited a tumor cell content of ≥60%. Sequencing libraries for the Illumina HiSeq 2000 platform were generated using the NEB Next Ultra DNA kit (NEB) according to the manufacturers’ instructions. Libraries for sequencing on the patterned flowcell v2.5 were generated using the Truseq DNA Nano kit (Illumina) according to the manufacturers’ instructions. Libraries were size selected using SPRI beads (Beckman Coulter Genomics). One Touch Pipeline (OTP) service of the German Cancer Research Center (DKFZ)59 was used for alignment and variant calling. Whole genome sequencing data was aligned to the 1000 Genomes project phase 3 assembly with decoy and PhiX contigs reference genome using BWA-MEM version 0.7.8, 0.7.8-r2.08 or 0.7.15 with option -T 0. Sambamba version 0.6.5 was used for merging and duplication marking and bam files were filtered using Samtools version 0.1.19. The SNV calling workflow is available at [https://github.com/DKFZ-ODCF/SNVCallingWorkflow]. Copy number variations were assessed using ACEseq version 1.2.860, which was additionally used to assess the quality of the sequencing data. The ACEseq workflow is available at [https://github.com/DKFZ-ODCF/ACEseqWorkflow]. Genes were defined as amplified if the copy number was at least 8-fold higher than the total copy number of the tumor sample. Any copy number losses that were not homozygous deletions were defined as a deletion. Any loss of heterozygosity (LOH) that was not also a deletion (e.g., copy-neutral LOH) was defined as LOH. Small insertions and deletions were determined based on an OTP in-house workflow61, which is available at [https://github.com/DKFZ-ODCF/IndelCallingWorkflow]. SOPHIA version 1.0.16 or 1.2.16 was used for detection of structural variants. SV calling workflow is available at [https://github.com/DKFZ-ODCF/SophiaWorkflow]. Structural variants affecting ATRX and TERT were determined using SOPHIA in combination with a manual inspection in the Integrated Genomics Viewer (IGV)62,63. TERT promoter mutations (C228T and C250T) were specifically analyzed using less stringent criteria for mutation calling (mutated base in at least two reads, mutation frequency ≥20%). Candidate genes associated with telomere maintenance and biology were identified using the TelNet database22 [https://malone2.bioquant.uni-heidelberg.de/fmi/webd/TelNet]. Genes encoding ATRX and DAXX interaction partners were extracted from the BioGRID database [https://thebiogrid.org/]. + +The telomere content was calculated using the software tool TelomereHunter64 version 1.0.1. TelomereHunter v1.0.1 is available at [https://www.dkfz.de/en/applied-bioinformatics/telomerehunter/telomerehunter.html]. Telomeric reads containing six non-consecutive instances of the four most common telomeric repeat types (TTAGGG, TCAGGG, TGAGGG, and TTGGGG) were extracted. For the further analysis, only unmapped reads or reads with a very low alignment confidence (mapping quality lower than 8) were used. Telomere read count was normalized by all reads in the sample with a GC-content of 48–52% and furthermore telomere content of tumor samples was normalized to the matching blood control. For tumor samples without a matching control, the mean telomere content of all controls sequenced with the same protocol in the same sequencing center was used for normalization. + +All tumors were categorized into four groups according to the genetic evidence of an activated telomere maintenance mechanism (ALT, MNA, TERT) or the absence of such a mechanism (OTHER). Tumors being either C-Circle positive or having a log2 telomere content relative to the control sample >1 were classified as ALT. Tumors with an amplified MYCN gene were considered to use telomerase activation by MYCN (MNA). The TERT group comprises tumors with a structural variation affecting TERT or TERT promoter mutation. Tumors with a relative log2 telomere content above one and amplified MYCN were included in the heterogeneous group HET ALT/MNA despite being C-Circle negative. + +To find telomeric repeat loci we searched for tumor-specific discordant paired-end reads, where one end was an extracted telomere read and the other end was non-telomeric and uniquely mapped to a chromosome (mapping quality >30). In 1 kb regions containing at least four discordant reads in the tumor sample and none in the matching control, exact junction sites were defined by at least three split reads. The split reads had to contain at least one TTAGGG repeat. Regions with discordant read pairs in at least 15 control samples were excluded. Finally, the junction sites were visualized using the Integrative Genomics Viewer (IGV)62,63 to identify and remove remaining false positives. 1q42.2 hotspot region was manually inspected to add automatically excluded telomeric repeat loci (NBD158 and NBD137). The pipeline used to detect telomeric repeat loci is available via Github [https://github.com/linasieverling/TelomereRepeatLoci]. + +Using the Nucleo Spin RNA kit (Macherey Nagel) RNA was isolated from fresh frozen tissue according to the manufacturers’ instructions. mRNA was isolated using poly A selection with Dynabeads (Thermo Fisher). 100 bp paired sequencing was carried out using the Illumina Hiseq 2000 and Hiseq 4000. Alignment to the reference genome (1000 Genomes project phase 3 assembly with decoy and PhiX contigs) was done using STAR Version 2.5.2b65 with the parameters–sjdbOverhang 200–runThreadN 8–outSAMtype BAM Unsorted SortedByCoordinate–limitBAMsortRAM 100000000000–outBAMsortingThreadN=1–outSAMstrandField intronMotif–outSAMunmapped Within KeepPairs–outFilterMultimapNmax 1–outFilterMismatchNmax 5–outFilterMismatchNoverLmax 0.3–twopassMode Basic–twopass1readsN −1–genomeLoad NoSharedMemory–chimSegmentMin 15–chimScoreMin 1–chimScoreJunctionNonGTAG 0–chimJunctionOverhangMin 15–chimSegmentReadGapMax 3–alignSJstitchMismatchNmax 5 −1 5 5–alignIntronMax 1100000–alignMatesGapMax 1100000–alignSJDBoverhangMin 3–alignIntronMin 20–clip3pAdapterSeq AGATCGGAAGAGCACACGTCTGAACTCCAGTCA–readFilesCommand gunzip -c. Sambamba version 0.6.566 was used for merging and duplication marking. Bam files were filtered with SAM tools version 1.3.167. Read counts per gene were obtained with featurecounts version 1.5.168 by counting reads over exon features using GENCODE 19 as a gene model. Both reads of a paired fragment were used for counting and only unique alignments were considered by setting the quality threshold to 255. The read counting was performed in an unstranded manner. The workflow used for RNA-sequencing is available at [https://github.com/DKFZ-ODCF/RNAseqWorkflow]. For the calculation of TPM expression values, genes on chromosomes X, Y, MT and rRNA and tRNA genes were omitted when calculating total library abundance calculations to prevent library size estimation biases. Differential expression analysis was performed using DESeq2 version 1.18.169 to find differences between samples with chr1q42.2-1qter deletion and those without a specific deletion in this region. chr1q42.2-1qter deletion was defined as reduced copy number compared to the upstream neighboring segment. Low-count genes with less than 10 reads across the entire cohort were excluded. Gender and sequencing location were included as possible confounding variables. Filtered raw count data was used for the default analysis including size factor estimation, dispersion estimation, negative binomial generalized linear model fitting, and Wald statistics. Log fold change shrinkage was applied. The results were filtered afterwards to include only genes from 230 Mb to the end of chromosome 1. + +To determine TERRA expression, RNA-sequencing reads containing telomere repeats were extracted with TelomereHunter64 version 1.0.1 using a threshold of 14 telomere repeats per 100 bp read length and otherwise default settings. All extracted telomere reads were considered as TERRA reads independently of the mapping position. To normalize for different library sizes, the number of TERRA reads was divided by the total read count and multiplied by 1 million. (Note that RNA-sequencing was done using poly-A selection). + +The paired-end RNA-sequencing reads were aligned to human genome reference hg19 using STAR version 2.4.1 tool65 with restriction to only unique alignments. Gene expression counts per exon were computed using corresponding DEXseq tool70 version 1.24.3 based on adjusted GENCODE 19 annotation. The computed per exon counts from ATRX were extracted and normalized using RPKM measurement via exon length. Subset of formed exonic parts (EP 13, 16, 17, 23, 31, 37, 38) with no coverage variance across samples was excluded from the initial gene model annotation for the main heatmap generation. Genomic coordinates for the DEXseq ATRX exon model are given in Supplementary Table 4. + +Tumor samples were lysed with 2% SDS, 50 mM ammonium bicarbonate buffer and complete mini EDTA-free Protease Inhibitor Cocktail. Samples were homogenized at room temperature 15 s for 7 cycles with a 5 s pause using FastPrep-24™ 5 G Homogenizer and heated to 95 °C for 5 min. Freeze–thaw cycles were repeated seven times. Benzonase was added to each sample and the lysate was clarified by centrifugation at 16 rcf for 30 min at 4 °C. Protein concentration was then measured using the Bio-Rad DC Protein assay. From each sample, 50 µg of protein were taken in duplicate. Proteins were reduced with 10 mM DTT for 30 min and alkylated with 55 mM iodoacetamide for 30 min. Wessel–Flügge precipitation was performed as previously described71. The retrieved protein pellet was resuspended in 6 M Urea, 2 M Thiourea in 10 mM HEPES (pH 8). Proteins were then digested with LysC for 2 h; samples were diluted in 50 mM ammonium bicarbonate buffer to reach less than 2 M urea before trypsin was added for overnight digestion. The peptide solution was acidified and fractionated by SCX on StageTips as described previously72. Samples were then separated by reversed phase HPLC on a 2000 mm monolithic column and analyzed online by high-resolution mass spectrometry on a Q Exactive plus mass instrument as described previously73. The resulting raw files were analyzed using MaxQuant software version 1.5.5.1 with LFQ quantification and ‘match between runs’ function74. Data was searched against a Human Uniprot database (2014-10; [https://www.uniprot.org/]) and common contaminants74 at a false discovery rate (FDR) of 1% at both the peptide and protein level. The resulting text files were filtered to exclude reverse database hits, potential contaminants, and proteins only identified by site. Data was imputed by random draw from a normal distribution with default parameters: 0.3 width and 1.8 down shift. Proteins associated with telomere maintenance and biology were identified using the TelNet database22 [https://malone2.bioquant.uni-heidelberg.de/fmi/webd/TelNet]. + +NBL-S neuroblastoma cells were cultured in RPMI media supplemented with 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin. siRNAs were transfected in 6-well format using Lipofectamine®RNAiMAX (Life Technologies) according to the manufacturer’s instructions. NBL-S cells were transfected with siRNAs targeting DAXX or ATRX as control at a final concentration of 25 nM. siRNA sequences are given in Supplementary Table 2. NBL-S cells were harvested 96 h after transfection and cell pellets were lysed in M-PER lysis buffer (Thermo Fisher). 30 µg of protein lysate were run on NUPAGE 3–8% Tris-acetate gels (Thermo Fisher). Primary antibodies were incubated overnight at 4 °C. ATRX antibody was obtained from Sigma (HPA001906; rabbit polyclonal) and used at a dilution of 1:1000. DAXX antibody was obtained from Abcam (ab32140; rabbit monoclonal E94) and used at a dilution of 1:5000. Secondary anti-rabbit HRP (Dianova 111-035-144; goat polyclonal) coupled antibody was used at a dilution of 1:2000. Vinculin-HRP coupled antibody (clone 7F9, Santa Cruz sc73614) was used as loading control at a dilution of 1:1000. A list of all antibodies is provided in Supplementary Table 3. Experiments were done in three biological replicates. Western blot images were quantified using the gel analysis function of Image J (version 1.51). P values were calculated using a two-sided Student’s t-test with the function t.test in R. + +Tumor touch imprints or 5 µm frozen sections of primary tumors mounted on microscopy glass slides were fixed with 4% Rotifix (Roth) for 10 min at 4 °C and FISH was performed using a telomere PNA-Cy3 FISH probe (Dako/Life Technologies) according to the manufacturer’s instructions and a MYCN-digoxigenin (dig) probe (BAC clone RP11-355H10). Briefly, slides were pretreated with 0.05% pepsin in 0.01 N HCl for 1 min at 37 °C, followed by dehydration (70%, 96%, 100% EtOH series). Probes were applied, denaturation was carried out at 80 °C for 5 min and hybridization at room temperature for 3 h. After immersion in 1xPBS, slides were washed in standard saline citrate (SSC) containing 70% formamide at 42 °C for 15 min and twice with 0.1xSSC for 5 min. The MYCN-dig probe was detected using a sheep polyclonal anti-dig-FITC antibody (dilution 1:100; Roche 11207741910). Slides were washed in PNA Wash-buffer (Dako) and dehydrated in an EtOH series. Sections were counterstained and mounted with Vectashield containing DAPI (Vector laboratories) and covered with glass coverslips. Images were acquired using a Zeiss Axioplan2 microscope and the ISIS software version 5.7.4 (Metasystems). + +Amplification status of MYCN was determined as described previously75 with MYCN Vysis LSI N-MYC (2p24) SpectrumGreen/CEP2 SpectumOrange and Vysis LSI N-MYC Spectrum orange. Activation of ALT was determined by identification of ALT-associated promyelocytic leukemia (PML) nuclear bodies (APB) by immunofluorescence in combination with FISH using Alexa-488-(TTAGGG)n PNA probes as described previously6. Images were acquired using Leica DM 5550B fluorescent light microscope system at 63x resolution. Full z-stacks were taken at 0.5 µm and projected using Cytovision software (Version 7.6). + +Tumor touch imprints were fixed and labeled as previously described76. Telomere labeling was done using the Telomere PNA FISH Kit/Cy3 (Agilent). For MYCN FISH DNA from BAC clone RP11-355H10 was labeled with FITC-dUTP by nick translation. Fluorescent images were captured with a Leica DMRA2 microscope equipped with a Leica DFC360 FX camera and analyzed with the Leica CW 4000 FISH software. + +Subset of the tumors was part of a previous publication1. Telomerase activity was measured using the TeloTAGGG Telomerase PCR ELISAPLUS Kit (Sigma Aldrich), which is a PCR-based telomeric repeat amplification protocol (TRAP) enzyme-linked immunosorbent assay (ELISA) kit, according to the manufacturer’s instructions. Approximately 10 mg of tissue were used per tumor. + +FACS analysis of neuroblastoma tumors was done as part of a previously published study77. + +ChIP-sequencing was performed based on a combination of the previously published protocols78,79, which is briefly summarized in the following. Tumor tissue was cut in 30 µm slices and fixed with 1% formaldehyde for 10 min followed by quenching with glycine for 5 min. Approximately 30 mg of tissue were used for each IP. Tumor tissue was lysed by adding 120 µL lysis buffer (50 mM Tris-HCl (pH 8.0), 10 mM EDTA, 1% SDS, protease inhibitor), homogenized using a pestle, incubated on ice for 5 min and sonicated for 5 cycles in a Diagenode Bioruptor Pico (30 s ON, 30 s OFF). After centrifugation (10 min, 12,000 rpm, 4 °C), the supernatant was transferred into a sonication tube (Diagenode). The pellet was resuspended in 30 µL lysis buffer and centrifuged again. 150 µL of mRIPA-I buffer without SDS (10 mM Tris-HCl (pH 8.0), 1 mM EDTA (pH 8.0), 500 mM NaCl, 0.1% Na-Deoxycholate, protease inhibitor) were added to the combined supernatants and samples were sonicated for 40 cycles (30 s ON, 30 s OFF). 220 µL mRIPA-I buffer without SDS were added and centrifuged (10 min, 12000 × g, 4 °C). Supernatant was kept and pellet resuspended in another 220 µL mRIPA-I buffer without SDS. Centrifugation was repeated and supernatants combined. 75 µL/IP magnetic protein G beads (Dynabeads Invitrogen) were washed twice with binding/blocking buffer (PBS with 0.5% BSA and 0.5% Tween-20). 3 µg of antibody were added together with 300 µL of binding/blocking buffer and incubated for at least 1 h at room temperature while rotating. H3K36me3 (ab9050; rabbit polyclonal), K3K9me3 (ab8898; rabbit polyclonal), and H3K27ac (ab4729; rabbit polyclonal) antibodies were obtained from Abcam. H3K27me3 antibody was purchased from Active Motif (39155; rabbit polyclonal). All antibodies are listed in Supplementary Table 3. The beads were washed again with blocking/binding buffer and sonicated chromatin was added and incubated overnight at 4 °C while rotating. Beads were washed once with ice-cold RIPA buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA (pH 8.0), 140 mM NaCl, 0.1% SDS, 0.1% Na-Deoxycholate, protease inhibitor), five times with ice-cold RIPA buffer (no protease inhibitor), twice with ice-cold RIPA-500 (10 mM Tris-HCl (pH 8.0), 1 mM EDTA (pH 8.0), 500 mM NaCl, 0.1% SDS, 0.1% Na-Deoxycholate, 1% Triton X-100), twice with ice-cold LiCl wash buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA (pH 8.0), 250 mM LiCl, 0.5% NP-40, 0.5% Na-Deoxycholate), and once with ice-cold 1xTE buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA (pH 8.0)). Samples were eluted in 50 µL of direct elution buffer (10 mM Tris-HCl (pH 8.0), 5 mM EDTA (pH 8.0), 300 mM NaCl, 0.5% SDS). 5% of the sonicated material were used as input control and processed in parallel from this step onwards. Samples were RNAse A treated (0.2 µg/µL) for 30 min at 37 °C. 2.5 µL proteinase K (10 mg/mL) and 1 µL glycogen (20 mg/mL) were added and incubated for 2 h at 37 °C. Samples were inclubated at 65 °C overnight to reverse cross-linking. Purification was done using Agencourt Ampure XP beads (Beckman Coulter Genomics) at a 2.3 ratio. Library preparation was done using the NEBNext End Repair Module (NEB), NEBNext dA-Tailing Module (NEB), and NEBNext Quick Ligation Module (NEB). High-Fidelity PCR Master Mix (NEB) was used for PCR amplification and adapter addition. Libraries were size selected using AgencourtAmpure XP beads (Beckman Coulter). Sequencing was done using Illumina HiSeq2000 50 bp single end sequencing. + +ChIP-sequencing data was analyzed using a custom pipeline implemented in Snakemake. Using the Trimgalore tool version 0.4 [https://github.com/FelixKrueger/TrimGalore] reads were trimmed and subsequently data was aligned to the hg38 reference genome using Bowtie2 version 2.3. Unmapped reads were not excluded from the bam files during alignment. Bamfiles were processed using the deepTools2 suite80 version 3.0. Using bamCompare input files were subtracted. Signal files (bigwig) were obtained using the SES method for normalization. Peaks were called using MACS2 version 2.1 [https://github.com/taoliu/MACS] for H3K27ac or SICER81 version 1.1 for H3K9me3, H3K27me3, and H3K36me3. ChIP-sequencing workflow is available at [https://github.com/hdsu-bioquant/chipseq_telomeres/]. Telomeric reads containing at least 8 telomeric repeats per 100 bp read length were extracted in ChIP and input samples using TelomereHunter64 with the option -rt 8 –rl. Number of telomeric reads was normalized to the total number of mapped reads. Reads containing the canonical SatII (ATTCCATTCGATTCCATTCG) and SatIII (ATTCCATTCCATTCCATTCCATTCCATTCC) sequence35 were extracted using Telomere Hunter64 version 1.0.1. Number of satellite reads was normalized to the total number of mapped reads. Log2 enrichment of normalized telomeric/satellite reads in ChIP was calculated relative to input files. Samples with a fraction of reads in peaks below one were excluded from the analysis.\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$Tel\left( {log2} \right)enrichment = log2\left( {\frac{{telomeric\,reads\,in\,IP/total\,mapped\,reads}}{{telomeric\,reads\,in\,input/total\,mapped\,reads}}} \right)$$\end{document}Tellog2enrichment=log2telomericreadsinIP/totalmappedreadstelomericreadsininput/totalmappedreads\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$SatII\left( {log2} \right)enrichment = log2\left( {\frac{{SatII\,reads\,in\,IP/total\,mapped\,reads}}{{SatII\,reads\,in\,input/total\,mapped\,reads}}} \right)$$\end{document}SatIIlog2enrichment=log2SatIIreadsinIP/totalmappedreadsSatIIreadsininput/totalmappedreads\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$SatIII\left( {log2} \right)enrichment = log2\left( {\frac{{SatIII\,reads\,in\,IP/total\,mapped\,reads}}{{SatIII\,reads\,in\,input/total\,mapped\,reads}}} \right)$$\end{document}SatIIIlog2enrichment=log2SatIIIreadsinIP/totalmappedreadsSatIIIreadsininput/totalmappedreads + +Differences between various neuroblastoma subgroups in terms of telomere content, telomeric repeat loci, enrichment of ChIP signals, TPM, and LFQ expression values were tested with two-sided Wilcoxon rank-sum tests using R (R Foundation for Statistical Computing, version 3.5.1 or 3.6.1). All boxplots were generated using the function geom_boxplot() in the R package ggplot2 (version 3.2.0 or 3.3.1). Thus, all boxplots indicate the median value. The lower and upper hinge of the boxplots correspond to the 25th and 75th percentiles. The upper/lower whisker spans from the hinge to the largest/smallest value (values expanding a distance of 1.5 x inter quartile range are not considered). Additionally, individual samples are shown as jitter points. Dotplot showing quantification of ATRX protein levels after DAXX knockdown based on western blot images was done using the function geom_dotplot() in the R package ggplot2 (version 3.3.1). Horizontal line is representing the median value. P values for Western blot quantification were calculated using a two-sided unpaired Student’s t-test in R with the function t.test. P values for differential RNA expression analysis were calculated with Wald tests within DESeq2 version 1.18.169. P values for differential protein expression analysis were calculated with two-sided Student’s t-test statistics using Perseus software version 1.5.5.382. Correlation analysis was performed using ggpubr (version 0.3.0) or with the function cor in R. Enrichment analysis was tested by hypergeometric test on gene symbols of significantly different proteins and mRNA determined for this analysis by SAM-modified83 (S0 = 0.01) Welch t-test were performed using R. Multiple testing correction by Benjamini–Hochberg method was performed with the R function p.adjust. Survival curves were generated using Kaplan Meier analysis in the R package survminer (version 0.4.7). P values for survival analysis were calculated using log rank tests. The optimal cut point to assess the prognostic impact of the number of telomeric repeat loci and telomere content was determined using Maximally Selected Rank Statistics with the criteria that at least 20% of the cases have to be in each group using the R package survminer (version 0.4.7). Associations of C-Circle presence with clinical features like age, sex, stage, risk classification was calculated using Fisher tests. Enrichment of TP53 pathway mutations (TP53, CREBBP, ATM, ATR, CDKN2A, and MDM2) and PTPRD mutations was calculated using a Fisher test comparing the occurrence in ALT-positive compared to ALT-negative tumors. Enrichment of RAS pathway mutations (HRAS, KRAS, NRAS, BRAF, RAF1, NF1, CDK4, CCND1) was calculated using a Fisher test comparing their occurrence in ALT-positive tumors in the discovery cohort compared to ALT-positive tumors in the INFORM cohort. HET cases and relapse cases of the discovery cohort were excluded. A Fisher test was used to calculate the enrichment of proteins mentioned in the Telnet database in the differentially expressed proteins of ALT-positive tumors. Fisher tests were done using R (version 3.6.1). + +Further information on research design is available in the Nature Research Reporting Summary linked to this article. + +Supplementary Information + +Peer Review File + +Description of Additional Supplementary Files + +Supplementary Data 1 + +Supplementary Data 2 + +Supplementary Data 3 + +Supplementary Data 4 + +Supplementary Data 5 + +Supplementary Data 6 + +Supplementary Data 7 + +Supplementary Data 8 + +Supplementary Data 9 + +Reporting Summary + +## Table 1 Clinical features of C-Circle positive tumors in the screening cohort (n = 720)a. + +| | Total cohort [colspan=2] | C-Circle positive [colspan=2] | C-Circle negative [colspan=2] | P valueg | +| | No.f | %f | No. f | %f | No.f | %f | | +| Total | 720 | | 66 | 9.2% | 654 | 90.8% | | +| Stage 1-3b | 416 | 57.8% | 25 | 37.9% | 391 | 59.8% | 1.948E-06 | +| Stage 4b | 221 | 30.7% | 41 | 62.1% | 180 | 27.5% | | +| Stage 4Sb | 83 | 11.5% | 0 | 0.0% | 83 | 12.7% | 0.000367 | +| LR/IRc | 449 | 65.9% | 24 | 36.9% | 425 | 69.0% | 6.944E-07 | +| HRc | 232 | 34.1% | 41 | 63.1% | 191 | 31.0% | | +| Male | 397 | 55.1% | 39 | 59.1% | 358 | 54.7% | 0.519 | +| Female | 323 | 44.9% | 27 | 40.9% | 296 | 45.3% | | +| Aged (mean in days) | 871 | | 2872 | | 669 | | <2E-16 | +| MYCN ampe | 114 | 16.2% | 0 | 0.0% | 114 | 17.9% | 5.883E-06 | +| MYCN normale | 588 | 83.8% | 66 | 100.0% | 522 | 82.1% | | +| Ganglioneuroma | 11 | 1.5% | 0 | 0.0% | 11 | 1.7% | 0.6115 | + +an refers to the number of tumors. bTumor staging was done according to the INSS criteria. cRisk stratification was based on the German Neuroblastoma Study NB2004. Risk stratification was not available for 39 tumors. dAge at diagnosis. eMYCN status refers to MYCN status at diagnosis, which is either amplified (amp, ≥8 copies) or normal. MYCN status was not available or inconclusive for 18 tumors. fAll values are given in absolute number of tumors (No.) and relative (%) to the total number of tumors in each group (total cohort, C-Circle positive, C-Circle negative). gP values were calculated using Fisher tests and two-sided Wilcoxon rank sum test (age). + +## Table 2 Clinical features of C-Circle positive tumors in the INFORM cohort (n = 40)a. + +| | Total cohort [colspan=2] | C-Circle positive [colspan=2] | C-Circle negative [colspan=2] | P valueg | +| | No.f | %f | No.f | %f | No.f | %f | | +| Total | 40 | | 19 | 47.5% | 21 | 52.5% | | +| Stage 1-3b | 4 | 10.0% | 4 | 21.1% | 0 | 0.0% | 0.041 | +| Stage 4b | 34 | 85.0% | 14 | 73.7% | 20 | 95.2% | | +| Stage 4Sb | 2 | 5.0% | 1 | 5.3% | 1 | 4.8% | 1.000 | +| LR/IRc | 7 | 17.5% | 5 | 26.3% | 2 | 9.5% | 0.226 | +| HRc | 33 | 82.5% | 14 | 73.7% | 19 | 90.5% | | +| Male | 32 | 80.0% | 13 | 68.4% | 19 | 90.5% | 0.120 | +| Female | 8 | 20.0% | 6 | 31.6% | 2 | 9.5% | | +| Aged (mean in days) | 1645 | | 2243 | | 1104 | | 0.003 | +| MYCN ampe | 11 | 27.5% | 2 | 10.5% | 9 | 42.9% | 0.029 | +| MYCN normale | 28 | 70.0% | 17 | 89.5% | 10 | 47.6% | | + +an refers to the number of tumors. bTumor staging was done according to the INSS criteria. cRisk stratification was based on the German Neuroblastoma Study NB2004. dAge at diagnosis. eMYCN status refers to MYCN status at diagnosis, which is either amplified (amp, ≥8 copies) or normal. MYCN status was inconclusive for one tumor at diagnosis. fAll values are given in absolute number of tumors (No.) and relative (%) to the total number of tumors in each group (total cohort, C-Circle positive, C-Circle negative). gP values were calculated using Fisher tests and two-sided Wilcoxon rank sum test (age). \ No newline at end of file diff --git a/references_cache/PMID_35100718.md b/references_cache/PMID_35100718.md new file mode 100644 index 00000000000..3aa5770eda8 --- /dev/null +++ b/references_cache/PMID_35100718.md @@ -0,0 +1,49 @@ +--- +reference_id: PMID:35100718 +extractor_version: 1 +title: Telomerase Is a Prognostic Marker of Poor Outcome and a Therapeutic Target in Neuroblastoma. +authors: +- Roderwieser A +- Sand F +- Walter E +- Fischer J +- Gecht J +- Bartenhagen C +- Ackermann S +- Otte F +- Welte A +- Kahlert Y +- Lieberz D +- Hertwig F +- Reinhardt HC +- Simon T +- Peifer M +- Ortmann M +- Büttner R +- Hero B +- "O'Sullivan RJ" +- Berthold F +- Fischer M +journal: JCO Precis Oncol +year: '2019' +doi: 10.1200/PO.19.00072 +publication_types: +- Journal Article +content_type: abstract_only +full_text_attempted: true +--- + +# Telomerase Is a Prognostic Marker of Poor Outcome and a Therapeutic Target in Neuroblastoma. +**Authors:** Roderwieser A, Sand F, Walter E, Fischer J, Gecht J, Bartenhagen C, Ackermann S, Otte F, Welte A, Kahlert Y, Lieberz D, Hertwig F, Reinhardt HC, Simon T, Peifer M, Ortmann M, Büttner R, Hero B, O'Sullivan RJ, Berthold F, Fischer M +**Journal:** JCO Precis Oncol (2019) +**DOI:** [10.1200/PO.19.00072](https://doi.org/10.1200/PO.19.00072) + +## Content + +PURPOSE: Telomere maintenance is a hallmark of high-risk neuroblastoma; however, the contribution of telomerase and alternative lengthening of telomeres (ALT) to clinical phenotypes has remained unclear. We aimed to determine the clinical relevance of telomerase activation versus ALT as biomarkers in pretreatment neuroblastoma and to assess the potential value of telomerase as a therapeutic target. + +MATERIALS AND METHODS: The genomic status of TERT and MYCN was assessed in 457 pretreatment neuroblastomas by fluorescence in situ hybridization. ALT was examined in 273 of 457 tumors by detection of ALT-associated promyelocytic leukemia nuclear bodies, and TERT expression was determined by microarrays in 223 of these. Cytotoxic effects of telomerase-interacting compounds were analyzed in neuroblastoma cell lines in vitro and in vivo. + +RESULTS: We detected TERT rearrangements in 46 of 457 cases (10.1%), MYCN amplification in 93 of 457 cases (20.4%), and elevated TERT expression in tumors lacking TERT or MYCN alterations in 10 of 223 cases (4.5%). ALT activation was found in 49 of 273 cases (17.9%). All these alterations occurred almost mutually exclusively and were associated with unfavorable prognostic variables and adverse outcome. The presence of activated telomerase (ie, TERT rearrangements, MYCN amplification, or high TERT expression without these alterations) was associated with poorest overall survival and was an independent prognostic marker in multivariable analyses. We also found that the telomerase-interacting compound 6-thio-2'-deoxyguanosine effectively inhibited viability and proliferation of neuroblastoma cells bearing activated telomerase. Similarly, tumor growth was strongly impaired upon 6-thio-2'-deoxyguanosine treatment in telomerase-positive neuroblastoma xenografts in mice. + +CONCLUSION: Our data suggest telomerase activation and ALT define distinct neuroblastoma subgroups with adverse outcome and that telomerase may represent a promising therapeutic target in many high-risk neuroblastomas. \ No newline at end of file diff --git a/references_cache/PMID_35384159.md b/references_cache/PMID_35384159.md new file mode 100644 index 00000000000..24dd5357761 --- /dev/null +++ b/references_cache/PMID_35384159.md @@ -0,0 +1,385 @@ +--- +reference_id: PMID:35384159 +extractor_version: 1 +xml_extraction_version: 1 +title: Mutational spectrum of ATRX aberrations in neuroblastoma and associated patient and tumor characteristics. +authors: +- van Gerven MR +- Bozsaky E +- Matser YAH +- Vosseberg J +- Taschner-Mandl S +- Koster J +- Tytgat GAM +- Molenaar JJ +- van den Boogaard M +journal: Cancer Sci +year: '2022' +doi: 10.1111/cas.15363 +keywords: +- Chromatin +- DNA Helicases/genetics +- Humans +- Mutation +- "Neuroblastoma/genetics, pathology" +- Phenotype +- X-linked Nuclear Protein/genetics +publication_types: +- Journal Article +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# Mutational spectrum of ATRX aberrations in neuroblastoma and associated patient and tumor characteristics. +**Authors:** van Gerven MR, Bozsaky E, Matser YAH, Vosseberg J, Taschner-Mandl S, Koster J, Tytgat GAM, Molenaar JJ, van den Boogaard M +**Journal:** Cancer Sci (2022) +**DOI:** [10.1111/cas.15363](https://doi.org/10.1111/cas.15363) + +## Content + +Neuroblastoma is the most common extracranial solid tumor in children. The chromatin remodeler ATRX is frequently mutated in high-risk patients with a poor prognosis. Although many studies have reported ATRX aberrations and the associated clinical characteristics in neuroblastoma, a comprehensive overview is currently lacking. In this study, we extensively characterize the mutational spectrum of ATRX aberrations in neuroblastoma tumors reported in previous studies and present an overview of patient and tumor characteristics. We collected the data of a total of 127 neuroblastoma patients and three cell lines with ATRX aberrations originating from 20 papers. We subdivide the ATRX aberrations into nonsense, missense, and multiexon deletions (MEDs) and show that 68% of them are MEDs. Of these MEDs, 75% are predicted to be in-frame. Furthermore, we identify a missense mutational hotspot region in the helicase domain. We also confirm that all three ATRX mutation types are more often identified in patients diagnosed at an older age, but still approximately 40% of the patients are aged 5 years or younger at diagnosis. Surprisingly, we found that 11q deletions are enriched in neuroblastomas with ATRX deletions compared to a reference cohort, but not in neuroblastomas with ATRX point mutations. Taken together, our data emphasizes a distinct ATRX mutation spectrum in neuroblastoma, which should be considered when studying molecular phenotypes and therapeutic strategies. + +alternative lengthening of telomeres + +B‐cell acute lymphoblastic leukemia + +in‐frame deletion + +in‐frame fusion + +multiexon deletion + + +MYC‐N amplification + +segmental chromosomal aberration + +X chromosomal inactivation + +Neuroblastoma is the most common extracranial solid tumor in children and has an incidence of 10.2 cases per million under the age of 15 years. +1 +, +2 + It arises from progenitor cells during fetal adrenal neuroblast development. +3 + Clinical, genetic, and pathologic characteristics are used for risk stratification and treatment allocation. +4 + Although the 5‐year overall survival rate for patients has improved over recent years to 81%, for patients with high‐risk disease survival remains 50% despite multimodal treatment regimens. +1 + + + +High‐risk tumors can be subclassified in four categories based on genomic aberrations: MYCN amplifications (37%), TERT rearrangements (23%), ATRX deletions (11%), and having none of the previously mentioned aberrations. +5 +, +6 + All these genetic aberrations are independently associated with a poor clinical outcome for patients. Both MNA and TERT rearrangements induce elevated expression of TERT and thereby facilitate telomere maintenance, one of the hallmarks of cancer. +5 + In contrast, ATRX aberrations can induce ALT, which is a homologous recombination‐based telomere maintenance mechanism that occurs in 5–15% of cancer. +7 + The MNA/TERT rearrangements and ATRX aberrations are mutually exclusive, whereas MNA and TERT rearrangements sporadically co‐occur. +5 + + + + +ATRX, located on the X‐chromosome, is a gene that encodes for an ATP‐dependent helicase of the SWI/SNF family of chromatin remodelers. The coding sequence, excluding the introns, comprises 7479 base pairs and consists of 35 exons. ATRX forms a chromatin remodeling complex with DAXX, which performs the replication‐independent deposition of the histone variant H3.3 at pericentromeric and telomeric regions. +8 +, +9 + In addition, ATRX is both directly and indirectly involved in a plethora of nuclear processes, such as removing G‐quadruplexes, +10 + inhibiting macroH2A1 deposition, +11 + removing R‐loops, +12 + modifying histones, +13 + methylating DNA, +14 + and resolving replication fork stalling. +15 + + + +This wide spectrum of ATRX functions might be reflected by the large shift in the expression of hundreds of genes that has been observed in cell lines derived from neuroblastoma patients with large MEDs resulting in IFDs of the ATRX gene. +16 + These IFDs do not lead to an absence of protein, but rather result in a shorter protein product. In neuroblastoma, both ATRX MEDs and point mutations have been reported, and these aberrations are strongly associated with ALT development +5 +, +6 +, +16 +, +17 +, +18 +, +19 +, +20 +, +21 +, +22 +, +23 +, +24 +, +25 +, +26 +, +27 +, +28 +, +29 +, +30 +, +31 +, +32 +, +33 +, +34 +. ATRX aberrations are also present in various other pediatric +25 +, +31 +, +35 +, +36 +, +37 +, +38 +, +39 +, +40 +, +41 +, +42 +, +43 +, +44 +, +45 +, +46 +, +47 +, +48 +, +49 +, +50 +, +51 +, +52 +, +53 +, +54 +, +55 +, +56 + and adult cancers. +57 + + + +Although several studies investigating genomic alterations in neuroblastoma have reported ATRX aberrations, the number of patients was often limited and a comprehensive analysis of the spectrum of ATRX mutations and patient characteristics is currently lacking. Therefore, in this meta‐analysis we extensively characterized the mutational spectrum of ATRX aberrations in neuroblastoma tumors described in previously published works and compared these with other pediatric cancers. Furthermore, we present an overview of the patient characteristics and co‐occurring aberrations in this neuroblastoma subgroup. + +To obtain all reported pediatric cancers with known genetic ATRX aberrations, we carried out a thorough search in several databases. The last search was undertaken on April 30, 2020 and all identified papers were manually assessed for patients with known genetic ATRX aberrations. Twenty‐one tumors with ATRX deletions described by Brunner et al. +19 + and Abbasi et al. +17 + were reanalyzed for this study for breakpoints and deleted exons. From all identified papers with neuroblastoma patients with ATRX aberrations, we collected the specific genetic mutation and a selection of the tumor/patient data (Table S1). The ATRX aberrations were reported in our paper based on the human reference genome Hg19 and transcript variant 1 (NM_000489.6). As multiple published papers can use the same dataset, we carefully assessed the origin of each patient and in that manner removed duplicates. To create the overviews of the ATRX point mutations and MEDs we used the genomic positions for plotting. For the analysis of the predicted effect of the missense mutation we performed PROVEAN protein batch analysis +58 + and for every analysis on the patient and tumor data all available input data were used (Table S1) and missing data were removed per analysis. All the data were analyzed and plotted using R 4.0.2 +59 + and ggplot2_3.3.5. +60 + For further details see Document S1. + +After collecting all the data on ATRX aberrations from neuroblastoma reports, we identified 127 unique patients and three cell lines from a total of 20 papers and belonging to 16 cohorts. Apart from two tumors, that were stage 2 and 3, all tumors were classified as stage 4. For comparison we also included the ATRX aberrations identified in other pediatric cancers of 122 tumors and one cell line (Table S1). These data originated from 23 studies and comprised nine different tumor types, mainly osteosarcoma, high‐grade glioma, or B‐ALL (Figure 1A). Tumors with ATRX aberrations can be divided in three subgroups based on their mutation type: nonsense/frameshift mutations, missense mutations, or MEDs. We defined MEDs as deletions involving at least two complete exons; all reported smaller deletions induce frameshifts and were therefore included within the nonsense/frameshift category. We observed that 68% of the neuroblastoma samples contained MEDs, whereas the MEDs only comprised 8% in other pediatric cancers, with the majority belonging to osteosarcomas (Figure 1B and Table S1). This indicated that the MEDs are almost exclusively present in neuroblastoma. Furthermore, these data showed that point mutations were less frequent in neuroblastoma, only constituting 32% of the cases, while making up to 92% of the cases in other pediatric cancers (Figure 1B). Overall, this indicates that ATRX aberrations in neuroblastoma have a different distribution of mutation types, with a high frequency of MEDs. + +Patient and mutation type distribution in our compiled dataset. (A) Distribution in percentages of all patients within our dataset across the different pediatric cancer types. ACT, adrenocortical tumor; AML, acute myeloid leukemia; B‐ALL, B‐cell acute lymphoblastic leukemia; ETMR, embryonic tumor with multilayered rosettes; HGG, high‐grade glioma; LGG, low‐grade glioma; MB, medulloblastoma; NB, neuroblastoma; OS, osteosarcoma; T‐ALL, T‐cell acute lymphoblastic leukemia. *≤1%. (B) Distribution in percentages of the three ATRX mutation types in neuroblastoma (left) and all other pediatric cancers (right) + +To further analyze the ATRX mutational spectrum in neuroblastoma and other pediatric cancers, we determined the frequency and distribution of nonsense/frameshift mutations as well as missense mutations across the gene. + +For neuroblastoma, 23 nonsense/frameshift ATRX mutations were reported, whereas 59 cases were described for all other pediatric cancers (Table S1). These aberrations were randomly distributed across the ATRX gene, which was also the case for the other pediatric cancers (Figure 2A). Recurrent nonsense/frameshift ATRX mutations were very rare and none of the recurrent mutations occurred in more than two patients. The only exception was R2386* (Table S1), which was present in three osteosarcoma patients, including two siblings with a germline mutation. +50 + + + +Pediatric point mutation landscape within the ATRX gene. (A) Distribution of nonsense and frameshift mutations of all pediatric cancers within our dataset across the ATRX gene. (B) Distribution of missense mutations of all pediatric cancers within our dataset across the ATRX gene. In both panels, neuroblastoma mutations are displayed above the bar and below the bar mutations that were identified in other pediatric cancers are displayed. All mutations were plotted based on DNA positions. ACT, adrenocortical tumor; AML, acute myeloid leukemia; B‐ALL, B‐cell acute lymphoblastic leukemia; ETMR, embryonic tumor with multilayered rosettes; HGG, high‐grade glioma; LGG, low‐grade glioma; MB, medulloblastoma; NB, neuroblastoma; OS, osteosarcoma; T‐ALL, T‐cell acute lymphoblastic leukemia + +In addition, ATRX missense mutations were identified in 17 neuroblastoma tumors and one cell line and 54 other pediatric tumors (Table S1). When these mutations were mapped against the ATRX gene, we observed a mutational hotspot region for neuroblastoma in the helicase domain (56% of missense mutations), especially the C‐terminal helicase domain where nearly 45% of all reported missense mutations were located (Figure 2B). Missense mutations were significantly enriched in the helicase domain for neuroblastoma (binomial test, p = 8.633e‐06). Furthermore, 38% and 36% of all ATRX missense mutations for high‐graded glioma and osteosarcoma, respectively, were also restricted to the helicase domains. In summary, this strongly supports the presence of a mutational hotspot region within the ATRX helicase domain across pediatric cancers. + +To date no complete 3D structure of the ATRX protein exists, therefore we were unable to assess structural consequences of the missense mutations and were limited to mutation prediction software to indicate potential biological effects. Of the 18 missense mutations identified in neuroblastoma tumors, 13 were predicted to have a deleterious effect by both the SIFT and PROVEAN analyses (Table S2) and two were reported as neutral by both algorithms. Overall, we can conclude that ATRX missense mutations are predominantly present in the helicase domain and that they are predicted to disturb protein function, whereas nonsense mutations are randomly distributed across the gene and could result in absence of ATRX protein production. + + +ATRX MEDs are observed in both neuroblastoma and osteosarcoma. Therefore, we undertook a detailed analysis of the ATRX MEDs identified in 77 neuroblastoma tumors and two cell lines and six osteosarcoma tumors and one cell line. For neuroblastoma, 29 unique MEDs were identified (Figures 3A and S1 and Table S1). Most commonly a deletion of exon 2–10 (28 of 79, 35.4%, MED#1) was observed, which was predicted to be in‐frame (called exon 1–11 IFF in Qadeer et al.) and to produce a mutant protein product. +16 + The deletion of exon 2–9 also occurred frequently (11 tumors and two cell lines, MED#2). Although this deletion is predicted to be out‐of‐frame, on the mRNA level alternative splicing with skipping of exon 10 was identified, thus resulting in the same mutant protein product as for a deletion of exon 2–106. The third most common deletion consists of exons 3–9 (2–10 IFF, five tumors and one cell line, MED#3), followed by several deletions that only occurred in two patients. The remaining 19 distinct deletions occurred only in a single case (Figure S1A). We also identified one MED that precisely covered the entire ATRX gene (Table S1, MED#29). We never observed a tumor with the same MED occurring in homozygosity. Of all deletions, 75% (59 of 79) was shown or predicted to be in‐frame and therefore could result in mutant protein production in most of the tumors. The minimal deleted region, the part that is lost in most cases, was exon 8 and 9 (Figure 3B). Upon close examination of all deletions, two distinct subgroups can be identified, namely the ones missing exon 8–9 and cases with smaller deletions that included loss of exon 11–12 (Figures 3A,B, S1, and S2). Interestingly, this is in sharp contrast to the missense mutations that are enriched in the helicase domain, while the deletions mostly involve the N‐terminal domains. This suggests a whole other type of functional consequence on the molecular level, as different protein‐interacting domains are affected and therefore could lead to different impairments of ATRX’s functions. + +Mutational landscape of recurrent multiexon deletions (MEDs) in neuroblastoma. (A) Overview of all recurrent MEDs in our dataset that occur in more than one patient. Numbers (#1–#9) indicate the unique MED identifier, and the number before every deletion indicates the number of patients with that specific aberration. Symbols after every deletion indicate whether the deletion is predicted to be in‐frame (Y) or not (N). Y* is predicted to be in‐frame because of exon skipping of exon 10. (B) Exon landscape plot showing the presence of each exon in all neuroblastoma MEDs in our dataset, including the deletions displayed in Figure S1 + + +Subsequently, we also wanted to examine deletions in other pediatric cancers and found that large deletions only occurred in osteosarcoma and B‐ALL (Table S1). However, in B‐ALL there were three out‐of‐frame MEDs, all different from the MEDs in neuroblastoma. In osteosarcoma there were seven reported MEDs, of which two were also reported in neuroblastoma, namely the predicted out‐of‐frame deletion of exon 2–15 (MED#6) and the predicted IFD of 2–13 (MED#5), of which the latter was the only recurrent aberration for osteosarcoma. The remaining four deletions were all predicted to be out‐of‐frame. In summary, we determined that MEDs were mostly observed in neuroblastoma and rarely occur in other pediatric cancers. Additionally, we found that 75% of ATRX deletions in neuroblastoma are predicted to be in‐frame and that there are two patterns of deletions, namely those lacking exons 8–9 or exons 11–12. + +There were several reports that showed that patients with neuroblastoma carrying ATRX aberrations are diagnosed at an older age. +20 +, +61 +, +62 + We wanted to investigate this for all three mutation subtypes separately. As a reference cohort for this analysis we used the cohort of Pugh et al. +28 + This cohort consisted of 240 stage 4 tumors, of which 24 had an ATRX aberration. After removing these samples, which were included in our ATRX subtypes, we had a reference cohort of 216 tumors. We took this cohort as it consisted of only stage 4 tumors, which was also almost the case for the ATRX tumors collected from published reports. Additionally, it was one of the very few studies reporting many of the SCAs observed in neuroblastoma. As none of the collected patients with ATRX aberrations belonged to a single cohort, the use of this reference cohort might introduce a systematic bias. We observed that there was a difference in the distribution of the age at diagnosis between the reference cohort and all three ATRX mutation subtypes (Mann–Whitney U‐test, deletion p = 2.6e‐13, nonsense p = 7.7e‐05 and missense p = 0.00053) (Figure 4A). Interestingly, this difference was found for all three subtypes, even for the patients with missense aberrations for which we had relatively few data points. + +Correlations between ATRX mutation subtypes with clinical characteristics in neuroblastoma. (A) Violin plot showing the distribution of age at diagnosis (years) for each of the mutation types. Red diamonds show medians: ATRX WT, 3.18 years; Deletion, 5 years; Nonsense/frameshift, 5.28 years; Missense, 11.7 years; p values (Mann–Whitney U‐test) are displayed. (B) Percentage of patients belonging to each age category for every mutation type. Fisher’s exact tests were carried out for each age category (18–60 months, 5–12 years, and >12 years) to compare the mutation types. Both p values and adjusted p values can be found in Table S3. (C) Distribution of gender in percentages within the ATRX aberrant subtypes. (D) Kaplan–Meier curves of event‐free survival (left) and overall survival (right) of the ATRX aberrant subtypes + +To confirm the different age distribution, we separated the patients with and without ATRX aberrations into three categories (18–60 months, 5–12 years, and >12 years) according to their age at diagnosis (Figure 4B; no patients were younger than 18 months of age in this cohort). We chose a cut‐off of 5 years as 90% of neuroblastoma patients are younger than 5 years old at diagnosis, +63 + and a cut‐off of 12 years because of the start of puberty. Eighty‐three percent of patients in the reference cohort (ATRX WT) were 5 years or younger, whereas less than 42% of patients were in this age group for the three ATRX mutation types. Compared to the reference cohort, the ATRX subtypes had a significantly different contribution of each age category (Table S3), except for the 5–12 years age category, where there was no significant difference between the reference cohort and the missense and nonsense subtype (Fisher’s exact test, p +adj = 0.88 and p +adj = 0.12, respectively). This could, however, be explained by the relatively low number of patients for these subtypes. Overall, this indicates that all three ATRX subtypes occurred more frequently in patients diagnosed at an older age, but still two‐fifth of the patients were diagnosed below 5 years of age. + +The ATRX gene is located on the X chromosome and therefore ATRX aberrations in neuroblastoma might affect male individuals more often than female individuals. To investigate this, we compared the male : female ratio in all three mutation types to the ratio reported in a large neuroblastoma cohort, which is 1.1:1. +64 + Interestingly, a higher male : female ratio than expected, with a ratio of 1.77:1 (Figure 4C), was detected in patients with ATRX MEDs (binominal test, p = 0.040). In contrast, the male : female ratio was not found to be skewed when ATRX was affected by missense or nonsense mutations (Figure 4C; binominal tests; p = 0.79 and p = 0.64, respectively). Our reference cohort had a relatively high male : female ratio of 1.6:1. However, male : female ratios of 1.2:1, +65 + 1.3:1, +66 + and 1.65:1 +67 + have also been reported in neuroblastoma cohorts and when we used these ratios the deletion subgroup was no longer significantly different. However, the ratio of 1.1:1 that we used for our analysis was derived from a large cohort across different continents and therefore seems more reliable. +64 + Thus, our analysis suggests that ATRX deletions, but not other ATRX aberrations, might occur more often in male than female patients. Nevertheless, more thorough investigation is necessary in the future. + +Patients with ATRX aberrant neuroblastoma are known to have poor event‐free and overall survival. +5 +, +61 + Here we assessed the event‐free and overall survival for each mutation type separately (Figure 4D). For event‐free survival, there was no difference between the three groups (Gehan–Breslow–Wilcoxon test, p = 0.098). However, a significantly better overall survival (Gehan–Breslow–Wilcoxon test, p = 0.022) of patients with missense mutations was observed as compared to the other two mutation types. Thus, patients with ATRX missense mutations might have a slightly better prognosis. + +Segmental chromosomal aberrations are common in neuroblastoma and several of these recurrent aberrations are associated with a poor prognosis, such as 17q gain, 11q deletion, 2p gain, and 1p deletion. +62 + To investigate the co‐occurrence of these SCAs in the three subgroups of ATRX aberrant tumors and one cell line (CHLA‐90), we plotted their frequencies compared to the reference cohort. This comparison showed no significant differences in the proportion of tumors with 17q gain, or 1p or 3p deletion (Figure 5 and Table S4). In contrast, 11q deletion (p +adj = 1.2694e‐10) and 2p gain (p +adj = 0.021) more often occurred in tumors with ATRX deletions compared to the other two ATRX mutation types or the reference cohort (Figure 5). This was most striking for the 11q deletion, as almost 90% of the patients with an ATRX deletion carried this SCA. 11q deletion and MNA are almost mutually exclusive, +68 + and the reference cohort contains approximately 36% MNA tumors; in contrast, MNA is rarely reported for ATRX aberrant tumors. Therefore, the observed enrichment of 11q deletions in tumors with ATRX deletions might result from a bias caused by MNA. To exclude this possibility, we carried out the analysis again after excluding all MNA tumors from the reference cohort. Even though the percentage of patients with 11q deletion increased in the reference cohort, there was still a significant difference for both 11q deletion (p +adj = 0.0004) and 2p gain (p +adj = 0.02) between the ATRX deleted tumors and the reference cohort (Figure S3 and Table S5). Thus, this was a strong indication that the co‐occurrence of 11q deletion occurred more frequently in the deletion subtype than in the other two ATRX mutation types. + +Co‐occurrence of ATRX aberrations with segmental chromosomal aberrations (SCAs). Percentage of patients with neuroblastoma‐specific SCAs for the ATRX mutation types and the ATRX WT reference cohort (including MYC‐N amplified tumors). Only significant p values after multiple testing correction are displayed (Fisher’s exact tests). del, deletion + +Additionally, it is important to consider the status of the X chromosome, as ATRX is localized on it. In male individuals, a single ATRX aberration will abolish WT ATRX expression, however, female individuals normally have two alleles. Even though the high prevalence of in‐frame MEDs might suggest a gain‐of‐function/dominant effect, this still must be experimentally confirmed. The X chromosomal status was reported for 24 of 47 female patients in our study. For one female patient (Table S1, AMC802T), both alleles contained an ATRX aberration; for another female patient (PASTKW), two ATRX aberrations were also reported and potentially inactivated both alleles (could also be on the same allele). From the remaining 22 female patients, seven tumors lost one X chromosome and therefore the detected ATRX aberrations will abolish WT ATRX expression. However, we also observed two girls and one boy in which the tumor acquired one X chromosome. Thus, not always both ATRX alleles are no affected by aberrations in all tumors; nevertheless, ALT has been reported in at least some of these tumors, suggesting a potential dominant effect. + +Finally, we also investigated the co‐occurrence of MNA, ALK, and TP53 mutations with ATRX aberrations. MYC‐N amplification was reported in only 4 of 121 ATRX aberrant tumors (two nonsense, one missense, and one MED tumor) (Table S1), which could be attributed to heterogeneous MNA as previously described. +69 + +ALK status and TP53 status were only reported in a subset of the ATRX aberrant tumors (71 and 46 tumors, respectively; Table S1). ALK mutations occurred in 14 of these tumors, including all ATRX mutation types. Interestingly, TP53 mutations were only identified in two samples, which are the cell lines SK‐N‐MM and CHLA‐90 carrying an ATRX IFD. Thus, neuroblastoma‐typical SCAs and mutations co‐occur with ATRX aberrations and specifically 11q deletions are more frequently observed within tumors with ATRX deletions. + + +ATRXaberrations are often associated with ALT and neuroblastomas with ATRX aberrations are more likely to have ALT than ATRX WT tumors. +61 + To investigate the frequency of ALT in the ATRX aberrant neuroblastomas from this study, we checked whether their ALT status was reported. The ALT status was assessed in a total of 44 tumors with ATRX aberrations, and a total of 43 samples were confirmed to have ALT (Table S1). The only ALT‐negative tumor contained a missense mutation at L407F, which was predicted as neutral and damaging by the PROVEAN and SIFT algorithms, respectively. The negative ALT status of this tumor might support a neutral role of this mutation in the tumor biology, although it could also be a false negative sample. Overall, these results confirm the strong association between the co‐occurrence of ALT with ATRX aberrations in neuroblastoma. + +The aim of this study was to characterize the mutational spectrum of ATRX aberrations in neuroblastoma tumors and to present an overview of the accompanying tumor and patient characteristics. In this manner we identified a mutational hotspot region for ATRX missense mutations within the helicase domain. Interestingly, we determined that nearly 70% of the ATRX aberrations in neuroblastoma are MEDs, compared to only approximately 10% in other pediatric cancers. Additionally, we found that 75% of these MEDs are predicted or known to be in‐frame and could result in protein production. Our separate comparison of the three mutational subtypes with ATRX WT tumors revealed that 11q deletion only co‐occurs more frequently with ATRX deletions. Finally, we identified that, even though there is a higher frequency of ATRX aberrations in patients diagnosed at an older age, still approximately 40% of patients are younger than 5 years of age at diagnosis. + +Although ATRX aberrations are frequently point mutations in most pediatric cancers, they represent a minority of the ATRX aberrations in neuroblastoma. We showed that in neuroblastoma the ATRX nonsense mutations were equally distributed across the ATRX gene. The effect of ATRX nonsense mutations in neuroblastoma has not been extensively studied. It is likely that these mutations lead to nonsense‐mediated mRNA decay or to an unstable protein prone to degradation. Unlike the random distribution of ATRX nonsense mutations, we identified the helicase domain as a missense mutational hotspot region in neuroblastoma. It is currently unknown whether missense mutations lead to reduced or absence of ATRX protein production in neuroblastoma. Interestingly, in patients with ATR‐X syndrome, a disease characterized by moderate intellectual disability and mild symptoms of α‐thalassemia, most of the missense mutations result in an unstable protein, more prone to degradation and resulting in reduced protein abundance. As the helicase domain is required for the DNA translocation activity of ATRX, +70 + defects in this domain could lead to disruption of histone variant H3.3 incorporation or other remodeling processes, such as removal of G‐quadruplexes or R‐loops. Therefore, ATRX point mutation models could illuminate the precise effects of the patient‐specific aberrations. + +Multiexon deletions of the ATRX gene are almost exclusively present in neuroblastoma and rare in other pediatric malignancies investigated in this study. We determined that 75% of these deletions are known or predicted to be in‐frame in neuroblastoma. Several studies have reported that these in‐frame ATRX deletions still result in protein expression. +16 +, +61 + This high percentage of IFDs indicates that there could be a gain‐of‐function effect of these aberrations. This effect might have been identified by Qadeer et al., who postulated that ATRX IFFs are still able to localize to chromatin with an alternative binding pattern. +16 + We also showed that the deletions can be divided into two subgroups, namely the more frequent larger deletions including exons 8–9, and the less common smaller deletions lacking exons 11–12. These smaller deletions, which so far have not been well studied, could have different molecular effects in neuroblastoma cells as compared to bigger deletions and therefore might need other therapies. The low frequency of these smaller deletions could be explained by these being much harder to detect in genome sequencing data, which is the most often used technique to identify such aberrations. Additionally, we determined that 25% of the deletions are predicted to be out‐of‐frame; nonetheless some could still produce in‐frame transcripts by using exon skipping, which has been reported for the second most common deletion of exon 2–9 (MED#2). +6 + + + +In our study we determined that there was a trend towards a gender bias in the group of patients with ATRX deletions, however, this was heavily dependent on the used reference ratio and therefore should be investigated on a larger cohort. In addition, it is often not reported whether female patients have one or two affected ATRX alleles. This is also complicated by whole X chromosome losses or gains and random XCI that occurs during each cell division in female cells. Interestingly, ATRX is involved in the maintenance of XCI +71 + and both ATRX alleles might be transcribed in female patients. Support for this comes from a study that generated a conditional heterozygous KO of ATRX in female cells of the 8‐ to 16‐cell state of developing mice that resulted in abrogation of XCI in extraembryonic tissues. +72 + Perturbed XCI has already been observed in several cancers including breast and colorectal cancer. +73 + Thus, also for neuroblastoma, it will be important to determine whether the ATRX aberrant allele is transcribed in all or a proportion of the tumor cells, which might be relevant for treatment strategies. + +It has previously been described that patients with neuroblastoma diagnosed at an older age more often had ATRX aberrations than younger patients. +20 +, +61 +, +62 + Our study confirmed this observation for each of the three mutation types independently in a cohort of 119 patients with ATRX aberrations. However, we also identified that almost 40% of the patients were younger than 5 years of age, which is much more frequent than described in earlier publications that contained fewer patients. +20 +, +61 + Furthermore, patients with ATRX aberrations have been reported to have a very poor overall and event‐free survival. +5 +, +61 + We observed this as well for the three mutation types regarding event‐free survival, but patients with a missense mutation had a better overall survival. An explanation could be that some of the observed missense mutations might be passenger variants that do not affect function, therefore patients might have been misclassified as ATRX aberrant. + +In neuroblastoma there are few recurring mutations, but almost all high‐risk patients carry SCAs, of which many have been associated with poor overall survival. +62 +, +74 + Here we showed that almost 90% of the tumors with an ATRX deletion also have a 11q deletion, which was more frequent than in the reference cohort. The association of ATRX aberrations and 11q deletions has very recently been reported, +61 +, +62 + but in our study this is only the case for tumors with ATRX deletions and not for ATRX point mutations. The minimal region of overlap lost in 11q deletions contains a lot of genes involved in apoptosis and in cell cycle progression. +68 + +ATRX aberrations are known to result in defective replication restart and concomitantly lead to increased DNA damage. +75 + Therefore, it could be evolutionary advantageous to acquire an 11q deletion when ATRX is aberrant to reduce the levels of DNA damage‐induced apoptosis, although it is also possible that the 11q deletion is already present before the ATRX deletion occurs and therefore provides the appropriate molecular environment. Why this co‐occurrence is more frequent for ATRX deletions than for point mutations and the order of the mutational events needs further investigation. We also show that ATRX deletions and 2p gain often co‐occur. Interestingly, in our dataset there were four tumors in which both an ATRX aberration and MNA were present. However, MNA and ATRX aberrations are reported to be clinically mutually exclusive +5 +, +62 + and elevated MYCN expression and ATRX loss were shown to be synthetically lethal in mouse models and cell lines. +61 + Nevertheless, for some tumors, heterogeneity of MNA and ATRX aberrations has been observed, +69 +, +76 + which could be a potential explanation for the patients with both aberrations in our dataset. Another possibility is that, in some instances, MNA might not result in high MYCN protein expression. Finally, TP53 mutations are very rare in our dataset, and are only present in two cell lines, where they might have been introduced in vitro. In contrast to glioma, where there is a strong co‐occurrence of TP53 and ATRX aberrations, +77 + we did not find such an association for neuroblastoma. + +The major limitation of this study is the lack of data of both patient and tumor characteristics for many of the categories that we investigated, such as overall survival and SCA status. We are especially lacking a lot of information for the less frequently observed SCAs (2p gain and 3p del) and for the mutational status of ALK and TP53. This lack of information might have influenced our outcomes and therefore our results should be re‐evaluated in the future. Most importantly, the X chromosomal or ATRX copy number status are frequently not reported, even though this might be important information for female patients. In the future, it would be of great value if more complete and more standardized clinical and tumor data would be reported and if sequencing data would be publicly available. + +In conclusion, this study shows that ATRX MEDs are the most frequent type of ATRX aberrations in neuroblastoma and that 75% of these are expected to produce in‐frame deleted protein products with a potential gain of function. We also identified a more frequent co‐occurrence of 11q deletions and ATRX deletions in neuroblastoma tumors. Overall, our study highlights that the different types of ATRX aberrations have distinct characteristics. It emphasizes the need for in vitro and in vivo models of these different aberrations, which might cause unique molecular phenotypes and require distinct treatments. + +The authors declare that there are no competing interests. + +Table S1 + +Click here for additional data file. + +Supplementary Material + +Click here for additional data file. \ No newline at end of file diff --git a/references_cache/PMID_36973454.md b/references_cache/PMID_36973454.md new file mode 100644 index 00000000000..fad29f2aa38 --- /dev/null +++ b/references_cache/PMID_36973454.md @@ -0,0 +1,1628 @@ +--- +reference_id: PMID:36973454 +extractor_version: 1 +xml_extraction_version: 1 +title: Neuroblastoma arises in early fetal development and its evolutionary duration predicts outcome. +authors: +- Körber V +- Stainczyk SA +- Kurilov R +- Henrich KO +- Hero B +- Brors B +- Westermann F +- Höfer T +journal: Nat Genet +year: '2023' +doi: 10.1038/s41588-023-01332-y +keywords: +- Infant +- Humans +- "Neuroblastoma/genetics, pathology" +- Prognosis +- Whole Genome Sequencing +publication_types: +- Journal Article +- "Research Support, Non-U.S. Gov't" +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# Neuroblastoma arises in early fetal development and its evolutionary duration predicts outcome. +**Authors:** Körber V, Stainczyk SA, Kurilov R, Henrich KO, Hero B, Brors B, Westermann F, Höfer T +**Journal:** Nat Genet (2023) +**DOI:** [10.1038/s41588-023-01332-y](https://doi.org/10.1038/s41588-023-01332-y) + +## Content + +Neuroblastoma, the most frequent solid tumor in infants, shows very diverse outcomes from spontaneous regression to fatal disease. When these different tumors originate and how they evolve are not known. Here we quantify the somatic evolution of neuroblastoma by deep whole-genome sequencing, molecular clock analysis and population-genetic modeling in a comprehensive cohort covering all subtypes. We find that tumors across the entire clinical spectrum begin to develop via aberrant mitoses as early as the first trimester of pregnancy. Neuroblastomas with favorable prognosis expand clonally after short evolution, whereas aggressive neuroblastomas show prolonged evolution during which they acquire telomere maintenance mechanisms. The initial aneuploidization events condition subsequent evolution, with aggressive neuroblastoma exhibiting early genomic instability. We find in the discovery cohort (n = 100), and validate in an independent cohort (n = 86), that the duration of evolution is an accurate predictor of outcome. Thus, insight into neuroblastoma evolution may prospectively guide treatment decisions. + +Cancers result from the accumulation of oncogenic mutations1. Insights into tumor evolution—and, particularly, the temporal order of driver mutations—are beginning to support diagnosis and treatment2. Initially, age–incidence curves were used to estimate the number of rate-limiting mutations in carcinogenesis3,4. Recently, the order of driver mutations has been inferred from genome sequencing data5–9, and mathematical approaches based on population genetics have been used to reconstruct the clonal evolution of cancers from the statistics of somatic variants2,10. Deep whole-genome sequencing (WGS) data are suited for such integrative analyses, because the allele frequencies of neutral somatic variants that hitchhike with driver mutations provide rich information on how drivers shape tumor growth7,11–13. + +Tumors of early childhood provide a paradigm for cancer evolution in the context of development. A key question is how driver mutations subvert the normal development of the tissue of origin14,15. The most common solid tumor in infants, neuroblastoma, arises in the sympathetic nervous system. A striking feature of neuroblastoma is the wide spectrum of clinical outcomes, ranging from low-risk cases requiring light or no treatment to high-risk disease that remains fatal for ~50% of patients16. Characteristic mutations, including MYCN amplification, gain of telomere maintenance mechanisms (TMMs) and gains (17q) or losses (1p, 11q) of chromosomal segments, have been associated with high-risk disease17. Nevertheless, the prospective stratification of patients into observation and different treatment groups remains a formidable challenge. In this Article, we study how neuroblastomas originate in development and evolve genetically and ask whether this understanding can provide insights into disease severity and outcomes. + +We assembled a discovery cohort of deep (~80×) WGS data of 100 neuroblastomas (Supplementary Table 1), covering all clinical stages of the disease according to the International Neuroblastoma Staging System (Fig. 1a). Sixty-seven samples were derived from initial diagnoses (including seven from metastases) and 33 from relapsed tumors. Median age at primary diagnosis was 0.7 years for stages 1, 2 and 4S, 3.5 years for stage 3 and 3.9 years for stage 4 (Extended Data Fig. 1a).Fig. 1Molecular subtypes and mutation spectrum of the discovery cohort.a, Clinical parameters and molecular characteristics. b, Tumor baseline ploidies and tumor cell content. Boxes show median, 25 and 75% percentiles and whiskers extend to the smallest and largest value within 1.5× interquartile range. Shown are n = 55 near-diploid, n = 33 near-triploid and n = 12 near-tetraploid tumors. c, Number of chromosomes harboring gains and losses ≥106 bp. Shown are mean and s.e. of the mean for n = 100 tumors. d, Copy number variants and small-scale mutations (SSNVs, small insertions/deletions, amplifications, homozygous deletions and structural rearrangements) in candidate driver genes. e, Exposures of mutational signatures (COSMIC v.3.1) per sample. Signatures SBS1, SBS5 and SBS40 were grouped into a single, clock-like mutations signature.Source data + +a, Clinical parameters and molecular characteristics. b, Tumor baseline ploidies and tumor cell content. Boxes show median, 25 and 75% percentiles and whiskers extend to the smallest and largest value within 1.5× interquartile range. Shown are n = 55 near-diploid, n = 33 near-triploid and n = 12 near-tetraploid tumors. c, Number of chromosomes harboring gains and losses ≥106 bp. Shown are mean and s.e. of the mean for n = 100 tumors. d, Copy number variants and small-scale mutations (SSNVs, small insertions/deletions, amplifications, homozygous deletions and structural rearrangements) in candidate driver genes. e, Exposures of mutational signatures (COSMIC v.3.1) per sample. Signatures SBS1, SBS5 and SBS40 were grouped into a single, clock-like mutations signature. + +Source data + +Median tumor purity was high (88%), which allowed for reliable estimation of tumor ploidy (Fig. 1b) as confirmed by direct measurement of DNA index (Extended Data Fig. 1b). Tumors had baseline ploidies between two and four, with 55, 33 and 12 samples being near-diploid, near-triploid and near-tetraploid, respectively. Each ploidy class contained tumors of all stages (Extended Data Fig. 1c). Relative to baseline ploidy, we detected in 96% of the tumors characteristic segmental (1q and 17q) or whole-chromosome (2, 7 and 17) gains, as well as segmental (1p, 11q) or whole-chromosome (11) losses (Extended Data Fig. 1d), all of which are probable drivers of neuroblastoma18–23. Based on tumor cell content and ploidy, the vast majority of these gains and losses were clonal and hence were present in the most recent common ancestor cell (MRCA) of the tumor (Fig. 1c). These data point to aneuploidy as an early feature of neuroblastoma. + +The majority of tumors (69%) combined chromosomal gains or losses with candidate driver mutations23 at smaller genomic scale (Fig. 1d and Supplementary Tables 2–5), including focal gene amplifications (for example, MYCN, CDK4, ALK), structural rearrangements (for example, in TERT and ATRX), large deletions (for example, CDKN2A, ATRX), small insertions/deletions (indels; for example, in ATRX and NF1) and somatic single-nucleotide variants (SSNVs which, among other genes, occurred in ALK and those coding for components of MAPK pathways: HRAS, KRAS, NRAS and BRAF). Most commonly, these additional oncogenic drivers support telomere maintenance by alternative lengthening of telomeres (ALT), MYCN amplification or rearrangement in the TERT locus, which, collectively, are a molecular predictor of poor outcome24–26. + +In contrast to driver mutations, neutral SSNVs and small indels are continuously accumulated and contain information on tumor evolution. Analysis of mutational signatures assigned the majority of SSNVs to clock-like signatures (SBS1, SBS5, SBS40). Overall the next most abundant signature was SBS18, associated with reactive oxygen species27 and frequently found in neuroblastoma28, followed by SBS3, associated with failure of homologous recombination in dividing cells (Fig. 1e). These signatures overall occurred at similar frequencies clonally and in subclones (Extended Data Fig. 1e). + +To understand when and how neuroblastomas evolve, we used neutral SSNVs as a molecular clock to time key events: (1) the emergence of the MRCA, from which the clonal sweep that defines the resected tumor emerges, and (2) the acquisition of clonal chromosomal gains (which may have occurred before or coincident with the MRCA). For each tumor, we computed the frequency distribution of somatic variants on all genomic segments of a given copy number. An example tumor, classified as near-triploid (Fig. 2a), had dominant copy numbers 2 and 3 and a smaller portion of segments at copy number 4—a typical configuration in neuroblastoma. On each copy number, we found a large clonal peak (Fig. 2b) comprising mutations occurring on a single chromosomal copy in each tumor cell. In addition, on copy numbers >2, we detected clonal mutations present on the two (and in some cases more) copies of a multiplied chromosome. Collectively, both single- and multiple-copy clonal SSNVs characterize the MRCA (Fig. 2c). Indeed, the density of clonal SSNVs was similar for the different copy numbers (Extended Data Fig. 2a), thus timing the MRCA by means of a molecular clock (Fig. 2d). To evaluate a potentially biasing effect of partial tumor sampling, which may erroneously cause a subset of subclonal mutations to be classified as clonal, we analyzed pairs of primary and relapse samples and found that the vast majority (85 ± 5%) of clonal SSNVs were present in both samples, indicating that sampling did not introduce a strong bias. Nevertheless, to estimate MRCA density conservatively we performed all subsequent computations with the measured densities corrected by a factor of 0.85. In the example tumor, the mean SSNV density of the MRCA was one SSNV per 5 million base pairs (bp).Fig. 2Timing of chromosomal gains using SSNVs.a, Copy number profile of a near-triploid tumor with ALT. Each segment of equal copy number is denoteded by a color in the bar at the bottom; these colors are used below to mark the segments in e,f. b, VAF distribution of SSNVs stratified by copy number for the tumor shown in a. c, Schematic introducing the nomenclature based on VAF of SSNVs. d,e, Densities of non-amplified (d) and amplified (e) clonal mutations on genomic segments of length ≥107 bp for the tumor shown in a. e, Dashed line indicates kernel-density estimate of the distribution of non-amplified clonal mutations shown in d. d,e, Chromosomal segments of equal copy number were combined into single genomic segments, and Holm-corrected, one-sided P values were computed based on negative binomial distribution (exact P values provided in Source Data). f, Mutation densities of ECA and MRCA, with 95% confidence bounds estimated by bootstrapping for the tumor shown in a. Horizontal lines represent mutation densities at gained segments, as in e, with 95% confidence bounds computed from χ2 distributions. Segments on which densities of amplified clonal mutations were significantly different from that at MRCA are marked by ** (adjusted P < 0.01); NS, not significant; test and P values as in e. e,f, Color coding for segments is in a.Source data + +a, Copy number profile of a near-triploid tumor with ALT. Each segment of equal copy number is denoteded by a color in the bar at the bottom; these colors are used below to mark the segments in e,f. b, VAF distribution of SSNVs stratified by copy number for the tumor shown in a. c, Schematic introducing the nomenclature based on VAF of SSNVs. d,e, Densities of non-amplified (d) and amplified (e) clonal mutations on genomic segments of length ≥107 bp for the tumor shown in a. e, Dashed line indicates kernel-density estimate of the distribution of non-amplified clonal mutations shown in d. d,e, Chromosomal segments of equal copy number were combined into single genomic segments, and Holm-corrected, one-sided P values were computed based on negative binomial distribution (exact P values provided in Source Data). f, Mutation densities of ECA and MRCA, with 95% confidence bounds estimated by bootstrapping for the tumor shown in a. Horizontal lines represent mutation densities at gained segments, as in e, with 95% confidence bounds computed from χ2 distributions. Segments on which densities of amplified clonal mutations were significantly different from that at MRCA are marked by ** (adjusted P < 0.01); NS, not significant; test and P values as in e. e,f, Color coding for segments is in a. + +Source data + +Next, we asked whether clonal chromosomal gains occurred coincident with, or ancestral to, the MRCA. To this end, we quantified SSNVs that were identical and clonal on two copies on trisomic and tetrasomic segments (termed amplified clonal SSNVs6; Fig. 2b). These mutations were acquired before a gain on the respective allele (Fig. 2c, dark green), and hence the number of these mutations correlates positively with the time at which the chromosomal gain occurred6,29. To make this timing measure independent of segment length, we use SSNV density. We compared these densities to those at the MRCA based on a negative binomial distribution of SSNVs across the genome (Methods). In the example tumor, nearly all gains had a mean density of one amplified clonal SSNV per 100 Mbp, which is significantly smaller (adjusted P < 0.01) than that of the MRCA (Fig. 2e). Hence, the molecular clock places these gains ancestral to the MRCA in an early common ancestor (ECA) of the tumor. The only exceptions were gains of Chr. 9 and 20q, which had a mutation density consistent with that of the MRCA and hence occurred later than the ECA. On those segments gained ancestral to the MRCA, the densities of amplified clonal SSNVs were statistically indistinguishable (Fig. 2e). Hence, near-triploidization occurred early and as a temporally confined event during the development of this tumor. Thereafter, further genetic evolution occurred before the clonal sweep from the MRCA commenced (Fig. 2f). + +We timed the MRCAs for all tumor samples, and the ECAs associated with chromosomal gains. First, we validated these timing approaches by comparison of mutation densities in samples taken following primary diagnosis with those of relapsed neuroblastomas. As expected, timing of the ECA, defining the putative early origin of tumorigenesis, was conserved in all sample types (Extended Data Fig. 2b). By contrast, the timing of the MRCA, defining the origin of a tumor sample, was significantly later in relapsed tumors, consistent with a bottleneck imposed by incomplete tumor resection or cytotoxic therapy (Supplementary Table 1) and, eventually, regrowth from a small number of surviving cells (Extended Data Fig. 2b,c). Interestingly, the MRCAs of metastases resected after initial diagnosis had SSNV densities indistinguishable from those of the primary tumors (Extended Data Fig. 2b,c), suggesting that metastases had originated around the time when the tumor started to grow from its MRCA. + +To infer the early evolution of tumors up to the MRCA, we focused on samples taken at initial diagnosis (primary tumors and metastases, n = 67 in the discovery cohort). Remarkably, mutation densities of the MRCA showed a bimodal distribution (Fig. 3a). This finding suggests that there are two classes of neuroblastoma, one where growth of the resected tumor commenced early (early-MRCA neuroblastoma) and another where it occurred later (late-MRCA neuroblastoma); for mutational signatures, see Extended Data Fig. 2d,e. Of note, although we estimated the mutation density of the MRCA conservatively (Methods), the dichotomy between early and late MRCA emerged robustly.Fig. 3Early- versus late-MRCA neuroblastoma.a, Bimodal distribution of mutation densities of the MRCA for the discovery cohort; a threshold of 0.05 SSNVs per Mb separates the two modes (dashed line). b,c, Top, cumulative SSNV densities of the ECA (dark green) and MRCA (light green) of primary tumors classified as either early-MRCA neuroblastoma (b, n = 20 tumors) or late-MRCA neuroblastoma (c, n = 26 tumors with ECA and n = 21 tumors without ECA). c, Left and right panels correspond to tumors with and without timeable ECA, respectively. b,c, Solid lines represent maximum-likelihood estimates, while shaded areas represent 95% confidence intervals obtained by nonparametric bootstrapping of chromosomal segments. Bottom, clonal chromosomal/segmental gains and losses implied in oncogenesis, ploidy, stage and presence of an acquired TMM. Gains were timed whenever possible as occurring in either the ECA (dark green) or MRCA (light green); gains that could not be timed are shown in gray. d, Mutation densities at the MRCA in 20 primary tumors classified as early MRCA, and at the ECA in 26 cases classified as late MRCA. Significance was tested using a two-sided Wilcoxon rank sum test/Mann–Whitney U-test (α = 0.05). Boxes show median, 25 and 75% percentiles and whiskers extend to the smallest and largest value within 1.5× interquartile range. N/A, not applicable.Source data + +a, Bimodal distribution of mutation densities of the MRCA for the discovery cohort; a threshold of 0.05 SSNVs per Mb separates the two modes (dashed line). b,c, Top, cumulative SSNV densities of the ECA (dark green) and MRCA (light green) of primary tumors classified as either early-MRCA neuroblastoma (b, n = 20 tumors) or late-MRCA neuroblastoma (c, n = 26 tumors with ECA and n = 21 tumors without ECA). c, Left and right panels correspond to tumors with and without timeable ECA, respectively. b,c, Solid lines represent maximum-likelihood estimates, while shaded areas represent 95% confidence intervals obtained by nonparametric bootstrapping of chromosomal segments. Bottom, clonal chromosomal/segmental gains and losses implied in oncogenesis, ploidy, stage and presence of an acquired TMM. Gains were timed whenever possible as occurring in either the ECA (dark green) or MRCA (light green); gains that could not be timed are shown in gray. d, Mutation densities at the MRCA in 20 primary tumors classified as early MRCA, and at the ECA in 26 cases classified as late MRCA. Significance was tested using a two-sided Wilcoxon rank sum test/Mann–Whitney U-test (α = 0.05). Boxes show median, 25 and 75% percentiles and whiskers extend to the smallest and largest value within 1.5× interquartile range. N/A, not applicable. + +Source data + +These data raise the question of whether late-MRCA tumors began to develop later or developed early and evolved for a longer period of time. To address this question in terms of the molecular clock, we analyzed SSNV densities on chromosomal/segmental gains in both tumor classes that defined an ECA. All 20 early-MRCA tumors had such gains (with 70% showing near-triploidization, 35% harboring segmental gains and 25% displaying both features; Fig. 3b). The respective mutation densities timing chromosomal gains and MRCA were statistically indistinguishable in the vast majority (90%) of cases (Extended Data Fig. 3a shows an example), suggesting that aneuploidy was not followed by acquisition of further clonal drivers in these tumors. In the remaining 10% of tumors, separate ECAs and MRCAs were timed but these were very close in regard to mutation density (0.04 SSNVs per Mb between ECA and MRCA; Extended Data Fig. 3b;), implying that they were temporally close events. Indeed, 95% of early-MRCA tumors lacked additional small-scale drivers while a single tumor harbored a MYCN amplification and a mutation in ALK. Hence early-MRCA tumors appear to be driven predominantly by aneuploidization. + +By contrast, in the majority (55%) of the 47 late-MRCA tumors, we distinguished two well-separated events, ECA and MRCA (Fig. 3c, left), with average distance 0.24 ± 0.1 SSNVs per Mb. In about one-third of these cases, a single early near-triplodization event defined an ECA (Fig. 2); whereas, in the remaining two-thirds, smaller-scale chromosomal gains defined the ECA (Extended Data Fig. 3c). Hence, the majority of late-MRCA tumors showed a signature of early gains followed by further genetic evolution to the MRCA. In the remaining 45% of late-MRCA cases, we could not reliably time a separate ECA (for example, if very short fragments were gained or if gains were found at copy number >4) (Fig. 3c (right) and Extended Data Fig. 3d). Hence, for these tumors we cannot time an early genetic event which, however, leaves open the possibility that small-scale mutations, chromosomal losses or high-level amplifications (four or more copies), neither of which can be timed reliably, preceded the MRCA of the tumor. Consistent with this hypothesis, the mutation densities of the MRCA in tumors without timeable ECA were nearly as high as in those with two distinguishable events (Extended Data Fig. 3e). Moreover, near-tetraploid tumors without timeable ECA showed evidence of sequential events. Here, we found a 2:0 allelic configuration in 11 of 12 near-tetraploid tumors with 1p loss, suggesting that 1p deletion preceded genome doubling. Overall, the ECA of late-MRCA tumors had mutation density indistinguishable from the MRCAs of early-MRCA tumors (Fig. 3d), suggesting that the majority of late-MRCA tumors acquired aneuploidy as early as the early-MRCA tumors. + +Finally, we analyzed the timing of gains in genome-wide events: near-triploidization of the genome or genome doubling. In all such tumors, the individual gains involved showed statistically indistinguishable timing, which is consistent with near-triploidization and genome doubling occurring as single catastrophic events (Extended Data Fig. 3f). + +In sum, we find that the MRCAs in primary neuroblastoma fell into two evolutionary groups: in early-MRCA tumors, near-triploidization of the genome and/or gains of chromosomal arms or whole chromosomes coincided with the MRCA. Similarly, in more than one-half of late-MRCA tumors, an early ECA was defined by whole-chromosomal or arm-level gains. Hence, these late-MRCA neuroblastomas originated at a time similar to early-MRCA examples but then showed prolonged genetic evolution. + +The early-MRCA class contained cases of stages 1, 2 and 4 S, typically having a prognosis, but only one case of stage 4 (Fig. 3a). Hence, we asked whether MRCA timing predicts outcome. Remarkably, early MRCA timing clearly identified cases with long event-free survival (Fig. 4a) and long overall survival (Fig. 4b). To further test this idea, we assembled an independent cohort of primary tumors and metastases, which was enriched for tumors with WGS at the customary coverage of 30× (n = 86; Supplementary Tables 6–10)22,25. We evaluated the timing of ECA and MRCA in each sample as described for the discovery cohort. The mutation density of the MRCA showed the same bimodal pattern as in the discovery cohort (Fig. 4c). Moreover, the three main patterns of ECA and MRCA occurrence were identical to the discovery cohort: coincidence of chromosomal/segmental gains and MRCA in the early-MRCA class (Fig. 4d; compare with Fig. 3b); in the late-MRCA class, chromosomal/segmental gains defining an early ECA that substantially preceded either a late MRCA (Fig. 4e, left; compare with Fig. 3c, left) or a late MRCA without timeable ECA (Fig. 4e, right; compare with Fig. 3c, right). Thus, the validation cohort corroborates the scenarios of ECA and MRCA timing found in the discovery cohort.Fig. 4Timing of the MRCA predicts outcome.a,b, Event-free (a) and overall survival (b) stratified by mutation density of the MRCA of primary tumors and metastases from the discovery cohort (n = 66; a single tumor lacking survival information was excluded from the analysis). Survival is shown for up to 10 years. P values were computed using the log-rank test; error band represents 95% confidence interval. c, Distribution of mutation densities of the MRCA for the validation cohort. The threshold (0.05 SSNVs per Mb, dashed line) is identical to that of the discovery cohort (compare with Fig. 3a). d,e, Mutation densities of ECA and MRCA for the validation cohort. Top, cumulative SSNV densities of the ECA (dark green) and MRCA (light green) of primary tumors classified as either early-MRCA neuroblastoma (d, n = 22) or late-MRCA neuroblastoma (e, n = 36 with ECA, n = 28 without ECA). e, Left and right panels correspond to tumors with and without timeable ECA, respectively. d,e, Solid lines represent maximum-likelihood estimates and shaded areas represent 95% confidence intervals obtained by nonparametric bootstrapping of chromosomal segments. Bottom, timing of pervasive chromosomal gains; segments compatible with both ECA and MRCA were classified as early, with subclonal gains excluded.Source data + +a,b, Event-free (a) and overall survival (b) stratified by mutation density of the MRCA of primary tumors and metastases from the discovery cohort (n = 66; a single tumor lacking survival information was excluded from the analysis). Survival is shown for up to 10 years. P values were computed using the log-rank test; error band represents 95% confidence interval. c, Distribution of mutation densities of the MRCA for the validation cohort. The threshold (0.05 SSNVs per Mb, dashed line) is identical to that of the discovery cohort (compare with Fig. 3a). d,e, Mutation densities of ECA and MRCA for the validation cohort. Top, cumulative SSNV densities of the ECA (dark green) and MRCA (light green) of primary tumors classified as either early-MRCA neuroblastoma (d, n = 22) or late-MRCA neuroblastoma (e, n = 36 with ECA, n = 28 without ECA). e, Left and right panels correspond to tumors with and without timeable ECA, respectively. d,e, Solid lines represent maximum-likelihood estimates and shaded areas represent 95% confidence intervals obtained by nonparametric bootstrapping of chromosomal segments. Bottom, timing of pervasive chromosomal gains; segments compatible with both ECA and MRCA were classified as early, with subclonal gains excluded. + +Source data + +In the validation cohort, MRCA timing was an accurate predictor of both event-free and overall survival (Fig. 5a,b). Merging the two cohorts (n = 152 primary tumors and metastases, excluding one case lacking survival data), we confirmed this result (Fig. 5c,d). To compare MRCA timing with other predictors of survival, we considered clinical variables used worldwide (stage, age), gain of a TMM (which improves on the clinically used criterion, MYCN amplification26) and a more recently proposed molecular predictor, the mutation status of the RAS/p53 pathway25. Early MRCA timing emerged as the most informative predictor of event-free survival (Fig. 5); overall survival was best explained by both MRCA timing and mutations in the RAS/p53 pathway (Extended Data Fig. 4). In sum, survival analyses suggest that extended evolution up to the founding cell of the primary tumor predicts unfavorable outcome.Fig. 5Survival analysis.a–d, Event-free (a,c) and overall survival (b,d) stratified by mutation density at the MRCA of primary tumors and metastases from the validation cohort (a,b, n = 86) and from both cohorts (c,d, n = 152). Survival is shown for up to 10 years; P values were computed using the log-rank test, and error band represents 95% confidence interval. e, Multivariate Cox regression analysis for event-free survival of both cohorts (n = 152), considering mutation densities at MRCA, acquired mechanisms of telomere maintenance, disease stage at diagnosis, age at diagnosis and functional mutations in the RAS/p53 pathway. Shown are mean hazard ratio, 95% confidence intervals and P values for each variable (two-sided Wald test; *, P < 0.05).Source data + +a–d, Event-free (a,c) and overall survival (b,d) stratified by mutation density at the MRCA of primary tumors and metastases from the validation cohort (a,b, n = 86) and from both cohorts (c,d, n = 152). Survival is shown for up to 10 years; P values were computed using the log-rank test, and error band represents 95% confidence interval. e, Multivariate Cox regression analysis for event-free survival of both cohorts (n = 152), considering mutation densities at MRCA, acquired mechanisms of telomere maintenance, disease stage at diagnosis, age at diagnosis and functional mutations in the RAS/p53 pathway. Shown are mean hazard ratio, 95% confidence intervals and P values for each variable (two-sided Wald test; *, P < 0.05). + +Source data + +Given that both early- and late-MRCA neuroblastomas begin to develop in the same time window (compare with Fig. 3d) but show markedly different durations of evolution and clinical outcome, we asked whether late-MRCA tumors have evolved further than early-MRCA examples simply by chance, or whether there are molecular factors that predispose to ongoing evolution in the late-MRCA class. Characteristic oncogenic events (chromosomal or segmental gains or losses, TMMs) and standard prognostic features (stage and age at diagnosis) showed strong separation between the two classes of neuroblastoma (Fig. 6a). Early-MRCA tumors had predominantly whole-chromosome aneuploidy (for example, chromosomes 17 and 7), whereas arm-level aneuploidy (including gains of 17q, 7q and 1q, as well as loss of 11q and 1p) was prevalent in the late-MRCA group. In late-MRCA tumors with timeable ECA, the vast majority of segmental gains that we could time were coincident with the ECA (Figs. 3c and 4e, dark green squares). Taken together, different processes underlie the early acquisition of aneuploidy in the two classes: mis-segregation of entire chromosomes in early-MRCA tumors vis-à-vis genomic instability in late-MRCA tumors (Fig. 6a).Fig. 6Early genomic instability is associated with unfavorable outcome.a, Enrichment of chromosomal aberrations and clinical parameters in early- and late-MRCA tumors predicting favorable and unfavorable outcome, respectively. Shown are odds ratios (centers) with 95% confidence intervals (error bars) between tumors with late and early MRCA for characteristics with significant enrichment (P < 0.05 according to two-sided Fisher’s exact test; exact P values are provided in Source data). The odds ratio for 7q gain is infinite and hence only the lower bound is displayed. b, Prevalence of TMMs across early- and late-MRCA neuroblastomas. c, Cumulative mutation densities of the MRCA stratified by TMM. Shown are 150 primary tumors and metastases of both cohorts, excluding three tumors with multiple TMM. Solid lines represent maximum-likelihood estimates and shaded areas represent 95% confidence intervals obtained by bootstrapping.Source data + +a, Enrichment of chromosomal aberrations and clinical parameters in early- and late-MRCA tumors predicting favorable and unfavorable outcome, respectively. Shown are odds ratios (centers) with 95% confidence intervals (error bars) between tumors with late and early MRCA for characteristics with significant enrichment (P < 0.05 according to two-sided Fisher’s exact test; exact P values are provided in Source data). The odds ratio for 7q gain is infinite and hence only the lower bound is displayed. b, Prevalence of TMMs across early- and late-MRCA neuroblastomas. c, Cumulative mutation densities of the MRCA stratified by TMM. Shown are 150 primary tumors and metastases of both cohorts, excluding three tumors with multiple TMM. Solid lines represent maximum-likelihood estimates and shaded areas represent 95% confidence intervals obtained by bootstrapping. + +Source data + +Acquired TMMs were strongly enriched in the late-MRCA class (Fig. 6a); 12% of early-MRCA tumors but 81% of late-MRCA tumors gained telomere maintenance—via MYCN amplification, ALT or TERT rearrangement—in the combined discovery and validation cohorts (Fig. 6b). Due to their small size, these structural rearrangements cannot be timed using mutation densities and hence may have been acquired before the emergence of arm-level aneuploidy, together with aneuploidy or subsequently. Early timing of arm-level aneuploidy in more than one-half of the late-MRCA cases (that is, tumors with an ECA; left-hand panels in Figs. 3c and 4e) suggests that telomere maintenance was gained during a secondary event between ECA and MRCA in these cases, which mainly included ALT and TERT rearrangement. In the remaining cases, where arm-level aneuploidy was timed at the MRCA and no ECA was identifiable (right-hand panels in Figs. 3c and 4e), telomere maintenance could also have been acquired in an oncogenic event preceding the MRCA. Interestingly, those cases with MYCN amplification fell predominantly within this latter group (Fisher’s exact test, P = 0.006055, odds ratio = 3.9 (1.4, 11.6)), suggesting that MYCN amplification tends to occur earlier than ALT or TERT rearrangement. Indeed, MYCN-amplified tumors generally had an earlier MRCA than tumors with ALT or TERT rearrangement (Fig. 6c). + +Collectively, these findings establish a link between genetic evolution of neuroblastoma and the observation that neuroblastomas with extensive arm-level aneuploidy tend to carry a poor prognosis30. The typically late-emerging MRCA in these cases indicates that the underlying genomic instability predisposes these tumors to prolonged evolution, including the acquisition of TMMs. + +Finally, we asked whether our genetic insights into neuroblastoma development could be synthesized into an integrative model of tumor evolution. To this end, we devised population-genetic models and quantified key parameters by fitting the models to our measured data, including the time-dependent incidences of ECA and MRCA and the variant allele frequency (VAF) distribution of somatic variants. In addition, we required the models to reproduce the overall incidence of the disease in the human population. For this reason, we focused on neuroblastomas with poor prognosis (‘high-risk’, enriched in the late-MRCA class), which occur in around one in 105 children (the frequency of low-risk cases, enriched in the early-MRCA class, is not reliably known due to incomplete diagnosis31–34). + +The models describe a proliferative population of putative cells of origin that are lost by either differentiation or cell death (Fig. 7). On average, µ SSNVs occur per cell division. A rare subset of mutations (including chromosomal gains/losses and smaller-scale events, such as SSNVs or localized MYCN amplification) will be oncogenic drivers and give rise to selected cell clones. As a minimal requirement for the evolution of a high-risk tumor, we accounted for two oncogenic events, defining ECA and MRCA, with mutation frequencies µ1 and µ2 per cell division. To determine these parameters, the model was fit to the experimental data using approximate Bayesian computation. Initially, we assumed that cells of origin are generated during fetal development and then remain available for neuroblastoma development. This simple model consistently overestimated the overall incidence of high-risk disease (Fig. 8a). By contrast, a model in which putative cells of origin were available for only a limited time window (Fig. 8b) accurately reproduced both the overall incidence and dynamics of the emergence of ECA and MRCA (Fig. 8c–e and Extended Data Fig. 5a).Fig. 7Population genetics models of tumor initiation and growth.a, Model scheme. Neuronal precursors divide and differentiate. Oncogenic events at ECA and MRCA cause the outgrowth of premalignant (ECA) and malignant (MRCA) clones. Cells divide at rate λ and differentiate at rate δ; oncogenic mutations reduce the loss rate by a factor 1/r (first event) and 1/s (second event). Neutral SSNVs are acquired at rate μλ in each cell while driver mutations are acquired at lower rates μ1λ (corresponding to the first oncogenic event and defining the ECA of the tumor) and μ2λ (corresponding to the second oncogenic event and defining the MRCA of the tumor). b, Population dynamics of normal neuroblasts (N), premalignant clones (harboring one oncogenic event, M1) and high-risk neuroblastomas (harboring two oncogenic events, NT). Rates are defined in a. c, The model outlined in a,b yields a probability distribution for the time point at which the MRCA emerges (PMRCA(t)); likewise, it also yields a conditional probability for the time point at which the ECA emerged, given the mutation density in the MRCA (PECA(t|no. of SSNVsMRCA)). The parameters of both probability distributions can be estimated from the measured SSNV counts at ECA and MRCA across the cohort. d, Model of mutation accumulation during neuroblastoma growth. Neutral SSNVs are continuously acquired at rate µeff, defined as the number of neutral mutations per effective division (where one effective division produces two surviving daughter cells). By fitting the model to the measured VAF distribution, an estimate for µeff is obtained on the level of individual tumors (subsetting on cases with sufficient data quality and excluding tumors with evidence for subclonal selection during tumor growth). e, Estimation of division rate in actual time. The time between gastrulation and diagnosis (tD) consists of a premalignant time span, up to the emergence of the MRCA (t1), and the expansion of the tumor thereafter (t2). Assuming exponential tumor growth and approximately 109 tumor cells at diagnosis, this yields an estimate for the division rate per tumor for the subset used to estimate µeff. To obtain an estimate for the population level, these estimates are subsequently averaged across the cohort.Fig. 8Evolutionary dynamics of neuroblastoma initiation.a, Assuming a constant putative population of origin following neuroblast expansion (left panel), the model fit does not capture the observed saturating incidence of high-risk neuroblastomas (right panel; shaded areas, 95% posterior probability of the fit; data, mutation densities of the MRCA from primary neuroblastomas with an acquired TMM, combining discovery and validation cohort; solid line, maximum-likelihood estimates of data with error bars representing s.d. estimated by bootstrapping). b, Transient putative population of origin. c,d, Model fits to ECA (c) and MRCA (d) with transient population of origin, as in b accounts for the experimental data [green shaded areas, 95% posterior probability bounds; vertical lines and shaded areas, mean and 95% CI of the estimated end of the first trimester (12 weeks p.c.) and of the time of birth (38 weeks p.c.); data, mutation densities of ECA (dark green, n = 47) and MRCA (light green, n = 95) from primary neuroblastomas (tumors/metastases) with acquired TMM; solid lines, maximum-likelihood estimates; error bars, s.d. estimated by bootstrapping]. e, Predicted transient expansion of putative cells of origin, agreeing with rapid proliferative phase of sympathetic neuroblasts (shaded area, 95% posterior probability; vertical line and shaded area, mean and 95% CI of the estimated end of the first trimester (12 weeks p.c.)). f, Estimated mutation rate per effective cell division, computed from primary tumors/metastases with MYCN amplification (amp.) (n = 11), TERT rearrangement (n = 2), ALT (n = 14) or no acquired TMMs (n = 15), using cases with highly accurate subclonal VAF distribution due to high tumor purity. Boxes represent median and 25 and 75% percentiles; whiskers extend to the smallest and largest values within 1.5× interquartile range. g,h, Estimated loss rate (relative to cell division, g) and cell division rate (h) in primary tumors and metastases analyzed in f (shown are n = 11 primary tumors with MYCN amplification, n = 2 primary tumors with TERT rearrangement, n = 14 primary tumors with ALT and n = 15 primary tumors without acquired telomere maintenance). Boxes represent median and 25 and 75% percentiles; whiskers extend to the smallest and largest values within 1.5× interquartile range; λT, division rate during tumor growth; δT, loss rate during tumor growth.Source data + +a, Model scheme. Neuronal precursors divide and differentiate. Oncogenic events at ECA and MRCA cause the outgrowth of premalignant (ECA) and malignant (MRCA) clones. Cells divide at rate λ and differentiate at rate δ; oncogenic mutations reduce the loss rate by a factor 1/r (first event) and 1/s (second event). Neutral SSNVs are acquired at rate μλ in each cell while driver mutations are acquired at lower rates μ1λ (corresponding to the first oncogenic event and defining the ECA of the tumor) and μ2λ (corresponding to the second oncogenic event and defining the MRCA of the tumor). b, Population dynamics of normal neuroblasts (N), premalignant clones (harboring one oncogenic event, M1) and high-risk neuroblastomas (harboring two oncogenic events, NT). Rates are defined in a. c, The model outlined in a,b yields a probability distribution for the time point at which the MRCA emerges (PMRCA(t)); likewise, it also yields a conditional probability for the time point at which the ECA emerged, given the mutation density in the MRCA (PECA(t|no. of SSNVsMRCA)). The parameters of both probability distributions can be estimated from the measured SSNV counts at ECA and MRCA across the cohort. d, Model of mutation accumulation during neuroblastoma growth. Neutral SSNVs are continuously acquired at rate µeff, defined as the number of neutral mutations per effective division (where one effective division produces two surviving daughter cells). By fitting the model to the measured VAF distribution, an estimate for µeff is obtained on the level of individual tumors (subsetting on cases with sufficient data quality and excluding tumors with evidence for subclonal selection during tumor growth). e, Estimation of division rate in actual time. The time between gastrulation and diagnosis (tD) consists of a premalignant time span, up to the emergence of the MRCA (t1), and the expansion of the tumor thereafter (t2). Assuming exponential tumor growth and approximately 109 tumor cells at diagnosis, this yields an estimate for the division rate per tumor for the subset used to estimate µeff. To obtain an estimate for the population level, these estimates are subsequently averaged across the cohort. + +a, Assuming a constant putative population of origin following neuroblast expansion (left panel), the model fit does not capture the observed saturating incidence of high-risk neuroblastomas (right panel; shaded areas, 95% posterior probability of the fit; data, mutation densities of the MRCA from primary neuroblastomas with an acquired TMM, combining discovery and validation cohort; solid line, maximum-likelihood estimates of data with error bars representing s.d. estimated by bootstrapping). b, Transient putative population of origin. c,d, Model fits to ECA (c) and MRCA (d) with transient population of origin, as in b accounts for the experimental data [green shaded areas, 95% posterior probability bounds; vertical lines and shaded areas, mean and 95% CI of the estimated end of the first trimester (12 weeks p.c.) and of the time of birth (38 weeks p.c.); data, mutation densities of ECA (dark green, n = 47) and MRCA (light green, n = 95) from primary neuroblastomas (tumors/metastases) with acquired TMM; solid lines, maximum-likelihood estimates; error bars, s.d. estimated by bootstrapping]. e, Predicted transient expansion of putative cells of origin, agreeing with rapid proliferative phase of sympathetic neuroblasts (shaded area, 95% posterior probability; vertical line and shaded area, mean and 95% CI of the estimated end of the first trimester (12 weeks p.c.)). f, Estimated mutation rate per effective cell division, computed from primary tumors/metastases with MYCN amplification (amp.) (n = 11), TERT rearrangement (n = 2), ALT (n = 14) or no acquired TMMs (n = 15), using cases with highly accurate subclonal VAF distribution due to high tumor purity. Boxes represent median and 25 and 75% percentiles; whiskers extend to the smallest and largest values within 1.5× interquartile range. g,h, Estimated loss rate (relative to cell division, g) and cell division rate (h) in primary tumors and metastases analyzed in f (shown are n = 11 primary tumors with MYCN amplification, n = 2 primary tumors with TERT rearrangement, n = 14 primary tumors with ALT and n = 15 primary tumors without acquired telomere maintenance). Boxes represent median and 25 and 75% percentiles; whiskers extend to the smallest and largest values within 1.5× interquartile range; λT, division rate during tumor growth; δT, loss rate during tumor growth. + +Source data + +The inferred rate of SSNV acquisition (µ) was 3.2 ± 0.4 SSNVs per day, consistent with measurements of somatic mutation rate in the developing central nervous system (5.1 (1.5, 9.0; 95% confidence interval) SSNVs per day)35. The model further inferred that oncogenic driver events occurred on average once per 1 million cell divisions (geometric mean of µ1 and µ2), which is consistent with a global estimate of driver mutation rate of 3.4 × 10–5 per division in the human genome13, given that only a subset of all drivers will cause neuroblastoma. We used SSNV rate to calibrate the molecular clock against real time, which placed the ECA within the first trimester of pregnancy (Fig. 8c), at which time rapidly dividing sympathetic neuroblasts are developing36. This first hit (ECA) sustains a subclone in which the MRCA emerges due to a further oncogenic event (Extended Data Fig. 5b) which, in the majority of cases, occurred within the first year of life (Fig. 8d). The model fit of Fig. 8 was based on all SSNVs, because our data (Extended Data Figs. 1e and 2d) suggest that mutational processes may not change markedly during neuroblastoma evolution. To test the robustness of our inference, we also performed all analyses with the subset of clock-like SSNVs as input. The inferred rate of clock-like SSNV acquisition (µ) was 2.3 (1.2, 2.3; 80% credible interval) per division (corresponding to 2.2 ± 0.3 SSNVs per day). All other inferred parameters remained practically unchanged (Extended Data Fig. 5c,d). Hence, confining the analysis to clock-like mutations corroborates the real-time calibration of ECA and MRCA. + +A further insight afforded by the population-genetic model is the extent of cell loss in the growing tumor. In general, only a subset of cell lineages will support growth by continued symmetric self-renewal of malignant cells whereas other lineages will terminate by either cell differentiation into nonproliferating states or cell death. We inferred the ratio of self-amplifying tumor cell divisions among all divisions from the subclonal tail of VAF distribution (Methods and ref. 37), finding that only ~10% of tumor cell divisions result in growth of late-MRCA neuroblastomas that acquired TMMs (Fig. 8f,g). This inference is consistent with the clinical observation of extensive cell death in such neuroblastomas. Moreover, the average cell division rate was lower in neuroblastomas with ALT than in those with MYCN amplification or TERT rearrangement (Fig. 8h), again in line with clinical observation; our estimated cell division rates agree quantitatively with those measured in neuroblastoma in vivo38. The fraction of self-renewing cell divisions supporting tumor growth was inferred from VAF distribution also for tumors without acquired TMMs, all falling within the early-MRCA class, yielding a higher value of ~30% (Fig. 8g). Hence, cell loss appears to be a stronger selective pressure for late-MRCA neuroblastomas, the majority of which gain TMMs, than for early-MRCA neuroblastomas, which rarely acquire such mechanisms. + +In this paper we timed genetic events in the evolution of neuroblastoma using the molecular clock of SSNV accumulation and, inferring the rate of SSNV acquisition from the distribution of VAFs, related this clock to real time by factoring in the age at diagnosis. In two-thirds of cases we find that chromosomal gains implicated in the pathogenesis of the disease occurred early, and typically within the first trimester of pregnancy. With respect to further evolution, these cases fall into two distinct classes: in the early-MRCA class the early chromosomal gain event also marked clonal outgrowth of the resected tumor whereas in the late-MRCA class the tumors evolved further before clonal outgrowth commenced. Remarkably, in our cohort MRCA class is an accurate predictor of clinical outcome. This is true regardless of whether we could time an early chromosomal gain, and implies that neuroblastomas with a longer evolutionary history are more aggressive. Because the strong association between MRCA timing and outcome was also present with 30× WGS, the utility of this predictor for patient stratification may be tested in clinical trials. + +Our real-time inference shows that neuroblastomas across the entire clinical spectrum acquired aneuploidy within the first trimester of pregnancy, when the adrenal medulla forms from sympathetic neuroblasts. Moreover, matching disease incidence with the population-genetic model suggests that the initial oncogenic event is limited to this time window. The transcriptomes of neuroblastomas most resemble those of sympathetic neuroblasts15,36. In this early window, neuroblasts are highly proliferative, which may make them vulnerable to aneuploidy. Finally, the observation that aneuploidy via near-triploidization is a temporally confined event is consistent with the long-standing hypothesis that this karyotype results from endoreduplication of the genome followed by tripolar cell division and selection of the fittest daughter cell39,40. + +The molecular nature of early aneuploidy is associated with whether the tumor continues to evolve: neuroblastomas with whole-chromosome aneuploidy typically did not evolve further and were overall associated with favorable outcomes; in contrast, most tumors with early genomic instability had unfavorable outcomes. Continued evolution of such tumors has also been noted in a recent study taking multiregion biopsies41, emphasizing the potential of spatially resolved genetic and transcriptomic analysis42,43. We did not detect specific drivers of genomic instability: in particular, p53 function was not impaired genetically. However, with prevalent 17q gains, reduced expression of TP53 (located on 17p) relative to driver genes expressed on 17q (for example, BIRC5, IGF2-BP1 and BRIP1) may favor genomic instability44,45. + +Accurate risk stratification in neuroblastoma remains a major concern. Our data suggest a link between diverse criteria—age at diagnosis, segmental versus whole-chromosome gains and losses and acquisition of TMM25,46—based on how neuroblastomas evolve. We find that a greater age at diagnosis is often linked to longer evolution of the tumor rather than later origin. Paradoxically, this implies that low-risk tumors reach detectable size earlier than high-risk. Indeed, we infer that low-risk tumors have a substantially lower fraction of tumor cell loss than high-risk tumors and hence should grow faster. Acquired TMMs should, consequently, provide a larger selective advantage in the high-risk, late-MRCA group where they indeed are enriched. Hence, late-MRCA tumors may grow more rapidly only after gaining telomere maintenance (similar to IDH-wild-type glioblastomas7) and hence are diagnosed later. Interestingly, we detected five neuroblastomas with amplified MYCN in the early-MRCA class, and, remarkably, all of these patients survived until the end of the observation periods (1,447–4,177 days), in contrast to the poor survival of patients with MYCN-amplified tumors in the late-MRCA group. Our findings suggest that MRCA timing may be worth considering as a parameter for patient stratification. + +Cohorts of primary and relapsed neuroblastoma tumors were retrospectively analyzed. Tumor material was collected as part of the diagnostic workflow of the German Neuroblastoma trial by the Society for Pediatric Oncology and Hematology and collected in the Neuroblastoma tumor bank. All trials were approved by the Ethics Committee of the Medical Faculty, University of Cologne, and collection and use of all tumor tissue material was approved (registry nos. NB97, NB2004 and NB2016). This study includes data of neuroblastoma tumors previously published in Hartlieb et al.24 and Peifer et al.26 The study by Hartlieb et al. also contains a subset of tumors from the St. Anna Kinderkrebsforschung at the Children’s Cancer Research Institute in Vienna, Austria, as well as tumors analyzed in the registry trial INFORM47. The study office of the Neuroblastoma trial in Cologne provided clinical annotations and survival information. All patients or their legal guardian approved the use of tumor material by signed informed consent. For analysis, all resected tumors were divided into four quadrants, all of which were evaluated histologically. MYCN status was assessed as a routine clinical marker for all tumors using fluorescence in situ hybridization. A cross-sectional slice of one quadrant was used for DNA extraction for WGS; the same quadrant was used for ploidy analysis, measuring the DNA index. DNA was isolated using phenol chloroform extraction from fresh-frozen tumor material. Control DNA was isolated from whole blood using the NucleoSpin Blood DNA extraction kit (Macherey-Nagel) according to the manufacturers’ instructions. Details of the included samples are provided in Supplementary Tables 1 and 6. + +WGS workflows for the previously published data are described in ref. 24. For additional samples, high-coverage, WGS was performed on a patterned flowcell v.2.5 (150-bp, paired-end) with coverage of about 80× for the tumor and whole-blood control samples. All tumors had a histological tumor cell content of ≥60%. Sequencing libraries were prepared using the Truseq DNA Nano kit (Illumina) according to the manufacturers’ instructions, and size selected using SPRI beads (Beckman Coulter Genomics). Alignment and variant calling was done using the One Touch Pipeline service of the German Cancer Research Center (DKFZ)48. Alignment was done using workflow v.1.2.73-1, available at Github (https://github.com/DKFZ-ODCF/AlignmentAndQCWorkflows). In brief, sequences were aligned to the 1000 Genomes project assembly with decoy and PhiX contigs using BWA-MEM v.0.7.15 with option ‘-T 0’. Merging and duplication marking were performed using Sambamba v.0.6.5, and bam files were filtered using Samtools v.0.1.19. Calling of SSNVs, somatic Indels, copy number variations and SVs was done using inhouse workflows, available at https://github.com/DKFZ-ODCF/SNVCallingWorkflow, https://github.com/DKFZ-ODCF/IndelCallingWorkflow, https://github.com/DKFZ-ODCF/ACEseqWorkflow and https://github.com/DKFZ-ODCF/SophiaWorkflow. + +Estimates of tumor cell content were manually adjusted in one case (NBE40) following visual inspection of VAF distribution. Structural variants were excluded in the present study if they had a minimal event score of <5; focal amplifications were defined as regions with copy number ≥10; homozygous deletions were defined as regions with copy number <0.9. Deletions of (parts of) chromosomal arm 1p were defined as p-terminal regions lost relative to 1q and, moreover, with copy number ≤1 in near-diploid, ≤2 in near-triploid or ≤3 in near-tetraploid tumors. In analogy, we annotated deletions of chromosome 11q if the copy number was ≤1 in near-diploid, ≤2 in near-triploid or ≤3 in near-tetraploid tumors, and if 11q was lost relative to 11p. For gains on chromosomes 1, 2, 7 and 17 we required the copy number to be higher than the (rounded) basal ploidy of the tumor. Partial gains on 1q, 2p, 7q and 17q were defined as regions on the respective chromosomal arm that were gained relative to the other chromosomal arm. + +Mutational signatures were learned de novo and thereafter decomposed into Cosmic mutational signatures v.3.1 (ref. 49) (http://cancer.sanger.ac.uk/cosmic/signatures) using SigProfilerExtractor (v.1.1.1)50. Only signatures contributing to ≥5% of the mutations in at least one sample and, in addition, identified in at least 10% of samples, were considered. The contributions of these signatures to each sample were then re-estimated using the R package mmsig v.0.0.0.9000 (ref. 51) (setting strandbias=F, bootstrap=F, cos_sim_threshold=0.01, force_include=c(“SBS1”, “SBS5”)). For visualization, signatures SBS1, SBS5 and SBS40 were combined into a single, clock-like mutational signature. + +Stratified analysis of clonal and subclonal mutations was performed by classifying mutations as subclonal if\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\mathop {\sum}\nolimits_{k = 0}^{n_{{{{\mathrm{var}}}}}} {\left( {\begin{array}{*{20}{c}} {n_{{{{\mathrm{var}}}}} + n_{{{{\mathrm{ref}}}}}} \\ k \end{array}} \right)} p^k\left( {1 - p} \right)^{n_{{{{\mathrm{var}}}}} + n_{{{{\mathrm{ref}}}}}} < 0.05;p = \frac{\rho }{{\rho {{{\mathrm{CN}}}} + 2\left( {1 - \rho } \right)}}$$\end{document}∑k=0nvarnvar+nrefkpk1−pnvar+nref<0.05;p=ρρCN+21−ρ, where ρ is estimated tumor cell content, nvar and nref are the number of variant and reference reads, respectively, and CN is copy number. + +We estimated mutation densities at partial and entire chromosomal gains, and at the MRCA of the tumor from the distribution of VAFs among clonal mutations. To this end we counted clonal mutations separately on each autosome, stratified by copy number (CN) state. Regions lacking a CN estimate were assigned to a specific CN state if the measured coverage ratio (CR) and B-allele frequency (BAF) matched the expected ratios within measurement error. Specifically, we required for each segment i, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\left[ {\frac{{{\mathrm{CN}}_i\rho + 2(1 - \rho )}}{{\pi \rho + 2(1 - \rho )}} - 0.1} \right] \le {{{\mathrm{CR}}}}_i \le \left[ {\frac{{{\mathrm{CN}}_i\rho + 2\left( {1 - \rho } \right)}}{{\pi \rho + 2\left( {1 - \rho } \right)}} + 0.1} \right]$$\end{document}CNiρ+2(1−ρ)πρ+2(1−ρ)−0.1≤CRi≤CNiρ+21−ρπρ+21−ρ+0.1 and \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\left[ {\frac{{b_i}}{{{{{\mathrm{CN}}}}_i}} - 0.05} \right] \le {{{\mathrm{BAF}}}}_i \le \left[ {\frac{{b_i}}{{{{{\mathrm{CN}}}}_i}} + 0.05} \right],$$\end{document}biCNi−0.05≤BAFi≤biCNi+0.05, where CNi is the copy number of segment i, ρ the tumor cell content, π the average tumor ploidy and b the number of B-alleles. States with copy number >4 or of size ≤107 bp were excluded from the analysis because the statistics become ambiguous for short pieces and high CN states. + +On each retained segment we estimated the number of clonal mutations, distinguishing clonal mutations present on a single allele (‘non-amplified clonal mutations’) from those present on multiple alleles (‘amplified clonal mutations’). To this end we applied a statistical framework52, distinguishing amplified clonal mutations acquired before a clonal gain and thus present on all A-alleles, or on all B-alleles at a given CN state, from non-amplified clonal mutations acquired either on the non-amplified allele or after clonal gain but before the MRCA and hence present on a single copy; the remaining mutations are subclonal. Accordingly, we expect to find clonal mutations at \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{VAFs}}}} \in \left\{ {\frac{1}{{{{{\mathrm{CN}}}}}},\frac{{{{{\mathrm{CN}}}} - b}}{{{{{\mathrm{CN}}}}}},\frac{b}{{{{{\mathrm{CN}}}}}}} \right\}$$\end{document}VAFs∈1CN,CN−bCN,bCN in a pure sample, and at \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{VAFs}}}}\, \in \left\{ {\frac{\rho }{\zeta },\frac{{({{{\mathrm{CN}}}} - b)\rho }}{\zeta },\frac{{b\rho }}{\zeta }} \right\}$$\end{document}VAFs∈ρζ,(CN−b)ρζ,bρζ in an impure sample with tumor cell content ρ, where:1\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\zeta = \rho {{{\mathrm{CN}}}} + 2\left( {1 - \rho } \right)$$\end{document}ζ=ρCN+21−ρis the average copy number of a given locus in the sample. Because measured VAFs are randomly distributed around their true values, we fit a binomial mixture model to the clonal mutations. To avoid misclassification of subclonal mutations as clonal, we neglected all variants with \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{VAF}}}} < \frac{\rho }{\zeta }$$\end{document}VAF<ρζ. By symmetry, this cutoff retains half of the non-amplified clonal mutations, introducing a subsequent correction factor of 2 (equation (4)). Defining the weights \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$w = (w_1,w_{{{{\mathrm{CN}}}} - b},w_b)$$\end{document}w=(w1,wCN−b,wb), for the non-amplified clonal and amplified clonal mutation peak, respectively, we computed the probability of measuring \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{var}},\,j}}}$$\end{document}nvar,j variant reads and \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{ref,j}}}}}$$\end{document}nref,j reference (nonvariant) reads at the position of the jth SSNV to:2\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P\left( {n_{{{{\mathrm{var,}}\,j}}},n_{{{{\mathrm{ref,}}\,j}}}|{{{\mathrm{CN}}}},\rho ,w} \right) = \mathop {\sum}\limits_{k \in \{ 1,{{{\mathrm{CN}}}} - b,b\} } {B\left( {n_{{{{\mathrm{var,}}\,j}}};n_{{{{\mathrm{var,}}\,j}}} + n_{{{{\mathrm{ref,}}\,j}}},\frac{{\rho k}}{\zeta }} \right)} P(k|w),$$\end{document}Pnvar,j,nref,j∣CN,ρ,w=∑k∈{1,CN−b,b}Bnvar,j;nvar,j+nref,j,ρkζP(k∣w),where \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$B\left( {n_{{{{\mathrm{var,}}\,j}}};n_{{{{\mathrm{var,}}i}}} + n_{{{{\mathrm{ref,}}j}}},\frac{{\rho k}}{\zeta }} \right)$$\end{document}Bnvar,j;nvar,i+nref,j,ρkζ is the binomial probability for drawing \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{var}},\,j}}}$$\end{document}nvar,j variant reads at sequencing depth \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{var}},\,j}}} + n_{{{{\mathrm{ref}},\,j}}}$$\end{document}nvar,j+nref,j from the peak comprising mutations that are clonal on k chromosomal copies. Defining a uniform prior probability, P(w), for the weights, we computed, up to normalization, the posterior probability as:3\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P(w|\{ n_{{{{\mathrm{var}},\,j}}},n_{{{{\mathrm{ref}},\,j}}}\} _j^N,{{{\mathrm{CN}}}},\rho ) \propto \mathop {\prod}\limits_{j = 1}^N P \left( {n_{{{{\mathrm{var}},\,j}}},n_{{{{\mathrm{ref}},\,j}}}|{{{\mathrm{CN}}}},\rho ,w} \right)P\left( w \right),$$\end{document}P(w∣{nvar,j,nref,j}jN,CN,ρ)∝∏j=1NPnvar,j,nref,j∣CN,ρ,wPw,where N is the total number of SSNVs under consideration. Clonal mutations were then assigned to distinct clonal peaks according to the weights at maximum a posteriori probability (MAP), yielding, on segment l, estimates for the number of clonal mutations on non-amplified chromosomes, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{1,l}$$\end{document}n1,l, on amplified b alleles, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{b,l}$$\end{document}nb,l (if b > 1) and on amplified a alleles (with copy number \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${\mathrm{CN}} - b$$\end{document}CN−b), \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{\mathrm{CN}} - b,l}$$\end{document}nCN−b,l, according to:4\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{1,l} = 2{{{\mathrm{MAP}}}}\left( {w_{1,l}} \right)N_l,n_{b,l} = 2{{{\mathrm{MAP}}}}\left( {w_{b,l}} \right)N_l,{{{\mathrm{and}}}}\,n_{{{{\mathrm{CN}}}} - b,l} = {{{\mathrm{MAP}}}}\left( {w_{{{{\mathrm{CN}}}} - b,l}} \right)N_l.$$\end{document}n1,l=2MAPw1,lNl,nb,l=2MAPwb,lNl,andnCN−b,l=MAPwCN−b,lNl. + +Mutation densities (SSNVs per bp) at the MRCA were estimated from the number of non-amplified clonal mutations and total size of the analyzed genome, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$g = \mathop {\sum}\nolimits_l {g_l}$$\end{document}g=∑lgl (ref. 52). The number of mutations per copy that were acquired up to the MRCA is5\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_l = \frac{{n_{1,l} + n_{{\mathrm{CN}} - b,l}({{{\mathrm{CN}}}}_l - b_l) + n_{b,l}b_l}}{{{{{\mathrm{CN}}}}_l}}.$$\end{document}nl=n1,l+nCN−b,l(CNl−bl)+nb,lblCNl. + +If tumor samples were well mixed, or tumors completely sampled, mutation densities at the MRCA could be directly estimated as nl/g. In practice, however, nl may consist of a set of true clonal mutations, acquired before tumor growth, and an additional set of mutations that appear as clonal in the tumor sample due to incomplete sampling. To correct for the latter, false-positive (FP), clonal mutations we compared primary and relapse samples from two such pairs available in our dataset (NBE11/NBE66, NBE51/NBE78). The fraction of conservative clonal mutations in the primary sample that remained undetected in the relapse sample was 15 ± 5% and was taken as the fraction of FP. With this correction, the mutation count, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$m_{{{{\mathrm{MRCA}}}}}$$\end{document}mMRCA, and mutation density, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\tilde m_{{{{\mathrm{MRCA}}}}}$$\end{document}m~MRCA, at MRCA were estimated as:6\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$m_{{{{\mathrm{MRCA}}}}} = \mathop {\sum}\limits_l {n_l} \left( {1 - {{{\mathrm{FP}}}}} \right)\,{{{\mathrm{and}}}}\,\tilde m_{{{{\mathrm{MRCA}}}}} = \frac{{m_{{{{\mathrm{MRCA}}}}}}}{g},$$\end{document}mMRCA=∑lnl1−FPandm~MRCA=mMRCAg,respectively. Lower and upper 95% confidence bounds for \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\tilde m_{{{{\mathrm{MRCA}}}}}$$\end{document}m~MRCA were estimated by bootstrapping the genomic segments 1,000 times. + +Next, we tested for each gained segment whether amplified clonal mutations were significantly less frequent than expected at the MRCA and, accordingly, assigned the segment to either the MRCA or an earlier time point (in the majority of cases the ECA; here we excluded a small number of segments in seven tumors with a higher density of amplified clonal mutations than the estimated mutation density at the MRCA, because such gains may be subclonal CN alterations erroneously classified as clonal). To this end, we modeled the number of mutations falling on each genomic segment with a negative binomial distribution, which accounts for overdispersion caused by heterogeneous mutation rates along the genome53. The probability that the gain of genomic segment l coincided with the MRCA is then computed as:7\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P\left( {n_{k,l;k \in \{ b,{{{\mathrm{CN}}}} - b\} }|g_l,g,m_{{{{\mathrm{MRCA}}}}}} \right) = \mathop {\sum}\limits_{r = 0}^{n_{k,l}} {\left( {\begin{array}{*{20}{c}} {m_{{{{\mathrm{MRCA}}}}} + r - 1} \\ {m_{{{{\mathrm{MRCA}}}}}} \end{array}} \right)} p^r\left( {1 - p} \right)^{m_{{{{\mathrm{MRCA}}}}}},$$\end{document}Pnk,l;k∈{b,CN−b}∣gl,g,mMRCA=∑r=0nk,lmMRCA+r−1mMRCApr1−pmMRCA,where: + + +\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$p = \frac{{m_{{{{\mathrm{MRCA}}}}}}}{{m_{{{{\mathrm{MRCA}}}}}\left( {1 + \frac{{g_l}}{g}} \right)}}.$$\end{document}p=mMRCAmMRCA1+glg. + + +Here, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{k,l;k \in \{ b,{{{\mathrm{CN}}}} - b\} }$$\end{document}nk,l;k∈{b,CN−b} is the number of amplified clonal mutations on segment l and gl is the respective segment size. We corrected the P values obtained with equation (7) for multiple testing using Holm correction (false discovery rate ≤ 0.01) and, accordingly, assigned each segment to either the MRCA or an earlier time point. + +Finally, we computed the mutation densities at ECAs from the segments with significance level α = 0.01 to:8\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\tilde m_{{{{\mathrm{ECA}}}}} = \frac{{\mathop {\sum}\nolimits_{l,p_{{{{\mathrm{adj}}}},l} \le 0.01} {n_{b,l}} + n_{{{{\mathrm{CN}}}} - b,l}}}{{\mathop {\sum}\nolimits_{l,p_{{{{\mathrm{adj}}}},l} \le 0.01} {g_{l,b}} + g_{l,{{{\mathrm{CN}}}} - b}}}.$$\end{document}m~ECA=∑l,padj,l≤0.01nb,l+nCN−b,l∑l,padj,l≤0.01gl,b+gl,CN−b. + +We then tested for each contributing segment whether its mutation load conformed to the ECA according to a negative binomial distribution (in analogy to equation (7)) and computed lower and upper 95% confidence bounds by bootstrapping, as before for the MRCA. + +We related mutation densities per haploid genome into weeks post conception (p.c.) by inferring SSNV rates per diploid genome and embryonic day (μλ) and mutation and division rates per day, using the measured VAF distributions and age at diagnosis as outlined in Real-time estimation of cell division rate (with similar results based on all SSNVs or only clock-like SSNVs). Because mutation calling was performed by comparing tumors against a matched blood control, mutation densities correlate with the time post gastrulation (at approximately 2 weeks p.c.). Thus, the mutation density per haploid genome, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\tilde m$$\end{document}m~, relates to the time p.c., t, according to \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\tilde m\left( t \right) = \frac{{\mu \lambda }}{d}\frac{1}{{3.3 \times 10^9}}(t - 14\,{{{\mathrm{days}}}})$$\end{document}m~t=μλd13.3×109(t−14days). The estimated time of birth was taken as 38 weeks after gastrulation (40 weeks p.c.). + +Survival analysis was performed using the R package survival v.3.1.12 (ref. 54). We detected a clear bimodality in the histogram of SSNV density at the MRCA in the discovery cohort and took the upper bin border just before the minimum (0.05 SSNVs per Mb) as threshold to split tumors into groups of early or late MRCA. The same value was found appropriate in the validation cohort. + +We modeled neuroblastoma initiation by sequential acquisition of driver mutations in two oncogenic events, assuming that both events are infrequent and occurring in early neuroblasts with low probabilities, μ1 and μ2, during cell divisions. We denote the selective advantage conferred by the driver mutations acquired during the two events by r and s. Specifically, we assume that one or more first events generate a precancerous cell population in which a second event creates the MRCA. On this basis, we compute the time-dependent probability of the emergence of the MRCA. Neuroblasts initially expand rapidly, but how this cell population behaves subsequently, because sympathetic neurons differentiate from these precursors, is not known precisely. Hence, two possible scenarios were considered: (1) an initial phase of exponential growth of the neuroblast population is followed by a subsequent phase of differentiation, modeled by exponential decay; and (2) an initial phase of exponential growth is followed by a subsequent phase of precursor homeostasis (for a related model of a two-step process of tumorigenesis, but in a homeostatic tissue, see ref. 55). The two phases are associated with distinct rates of cell division, λ1 and λ2, and loss, δ1 and δ2, where \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\lambda _1 > \delta _1$$\end{document}λ1>δ1 and \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\lambda _2 \le \delta _2$$\end{document}λ2≤δ2 (with equality in the case of precursor homeostasis). Thus the population dynamics of neural precursor cells, N(t), are described by:9\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$N\left( t \right) = \left\{ {\begin{array}{*{20}{l}} {{\mathrm{e}}^{\left( {\lambda _1 - \delta _1} \right)t},} \hfill & {0\le t \le T} \hfill \\ {{\mathrm{e}}^{\left( {\lambda _1 - \delta _1} \right)T}{\mathrm{e}}^{\left( {\lambda _2 - \delta _2} \right)(t - T)},} \hfill & {t > T,} \hfill \end{array}} \right.$$\end{document}Nt=eλ1−δ1t,0≤t≤Teλ1−δ1Teλ2−δ2(t−T),t>T,where T denotes the time point at which the cell population peaks. Cells undergoing the first oncogenic event are generated at rates:10\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\begin{array}{*{20}{l}} {\mu _1\lambda _1N(t){\mathrm{d}}t,} \hfill & {0 \le t \le T} \hfill \\ {\mu _1\lambda _2N(t){\mathrm{d}}t,} \hfill & {t > T.} \hfill \end{array}$$\end{document}μ1λ1N(t)dt,0≤t≤Tμ1λ2N(t)dt,t>T. + +For simplicity, we take the selective advantage associated with the first oncogenic event into account during the contraction (or homeostasis) phase and neglect it during the initial rapid expansion; this approximation is appropriate when the selective advantage of the first event is comparatively small, which is subsequently borne out by the parameter inference. Thus, M1 cells harboring the first mutation grow at rate λ1 − δ1 during tissue expansion and at rate \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\lambda _2 - \delta _2/r$$\end{document}λ2−δ2/r during tissue contraction or homeostasis. Hence, depending on the actual value of r > 1, a clone harboring the first event may slowly shrink or expand for t > T. We now ask whether in this clone the second oncogenic event takes place that defines the MRCA of the tumor. The second event occurs at rate:11\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\begin{array}{*{20}{l}} {\mu _2\lambda _1M_1(t){\mathrm{d}}t,} \hfill & {0 \le t \le T} \hfill \\ {\mu _2\lambda _2M_1(t){\mathrm{d}}t,} \hfill & {t > T.} \hfill \end{array}$$\end{document}μ2λ1M1(t)dt,0≤t≤Tμ2λ2M1(t)dt,t>T. + +Using the survival probability of the supercritical birth–death process, we have a cell undergoing the second oncogenic event survive with probability:12a\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\nu _{2,{{{\mathrm{E}}}}} = 1 - \frac{{\delta _1}}{{s\,\lambda _1}}$$\end{document}ν2,E=1−δ1sλ1during the expansion phase (E) and, provided that \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\frac{{\delta _2}}{{s\lambda _2}} < 1$$\end{document}δ2sλ2<1, with probability:12b\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\nu _{2,{{{\mathrm{D}}}}} = 1 - \frac{{\delta _2}}{{s\lambda _2}}$$\end{document}ν2,D=1−δ2sλ2during the decay or homeostatic phase (D). We are interested in the probability that at least one surviving cell underwent both oncogenic events, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P_{{{{\mathrm{MRCA}}}}}$$\end{document}PMRCA. There are three possible cases: (1) both oncogenic events occur during precursor expansion, associated with probability \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P_{{{{\mathrm{MRCA}}}},{{{\mathrm{1}}}}}$$\end{document}PMRCA,1; (2) the first oncogenic event occurs during precursor expansion, the second during precursor contraction or homeostasis, associated with probability \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P_{{{{\mathrm{MRCA}}}},{{{\mathrm{2}}}}}$$\end{document}PMRCA,2 ; and (3) both oncogenic events occur during precursor contraction or homeostasis, associated with probability \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P_{{{{\mathrm{MRCA}}}},{{{\mathrm{3}}}}}$$\end{document}PMRCA,3. For each case we assumed that the number of cells with only the first event, M1, is small compared with the number of normal cells. We have:13\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P_{{{{\mathrm{MRCA}}}}} = \left\{ {\begin{array}{*{20}{l}} {P_{{{{\mathrm{MRCA}}}},{{{\mathrm{1}}}}},t < T} \hfill \\ {1 - \left( {1 - P_{{{{\mathrm{MRCA}}}},{{{\mathrm{1}}}}}} \right)\left( {1 - P_{{{{\mathrm{MRCA}}}},{{{\mathrm{2}}}}}} \right)\left( {1 - P_{{{{\mathrm{MRCA}}}},{{{\mathrm{3}}}}}} \right),t \ge T} \hfill \end{array}} \right..$$\end{document}PMRCA=PMRCA,1,t T\end{array}$$\end{document}Pt1∣t2=ΘT−t1λ1δ21−1rt1+Θt1−Tλ2+λ1Tδ21−1/r−λ2eδ2(1−1/r)(T−t1)λ2+λ1Tδ21−1/r−λ2eδ2(1−1/r)(T−t2);t2>Twhere \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{\Theta }}\left( \cdot \right)$$\end{document}Θ⋅ is the Heavyside step function and, of course, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$0 \le t_1 < t_2$$\end{document}0≤t1 T.\end{array}$$\end{document}Pt1∣t2=ΘT−t1λ1t11−1/reλ21−1rt2−T+Θt1−Tλ1T1−1r+1eλ21−1rt2−T−eλ21−1rt2−t1λ1T1−1r+1eλ21−1rt2−T−1;t2>T. + +To relate the model to the measured data, we translate time into mutation counts. Assuming that SSNVs are accumulated at a constant rate, μ, per cell division, the expected mutation count per cell is Poisson-distributed with rate μλt. + +We estimated the following parameters: peak size of the neuroblast population (N), relative loss rates in the growth and decay phases (\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\delta _1/\lambda _1$$\end{document}δ1/λ1 and \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\delta _2/\lambda _2$$\end{document}δ2/λ2, respectively), rate of SSNVs (μ) and rates of first and second oncogenic events (μ1 and μ2) and their selective advantages (expressed as r and ν2), using equations (9–17) and Approximate Bayesian Computation with Sequential Monte Carlo sampling (ABC–SMC) as implemented in pyABC56. We used a population size of 1,000 parameter samples and prior probabilities as outlined in Supplementary Table 11. Evaluation was abrogated after 25 SMC generations, or if ε ≤ 0.05. We used mutation density estimates at MRCA and ECA of high-risk tumors (primary tumors and metastases) as determined by equations (6) and (8) as input data. Tetraploidization in tetraploid tumors was not included as ECA because there were probably earlier events, such as Chr. 17q gains. The experimental incidences were computed as \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$I_{{{{\mathrm{MRCA,ev,i}}}}} = {\sum} {\tilde m_{{{{\mathrm{MRCA,ev}}}}}} < i;i \in \tilde m_{{{{\mathrm{MRCA,ev}}}}}$$\end{document}IMRCA,ev,i=∑m~MRCA,ev 0.2/10^6} \end{array}} \right..$$\end{document}wi=10,ifm~MRCA≤0.2/1061,ifm~MRCA>0.2/106. + +Ninety-five per cent posterior probability bounds for the model fits were estimated by simulating the model at each sampled parameter set and cutting off 2.5% at each end of the simulated distribution. + +To assess the robustness of the model, we performed an additional parameter estimation on clock-like mutations only. To this end, we multiplied mutation densities at MRCA and ECA by the fraction of mutations generated by clock-like mutational signatures (SBS1, SBS5 and SBS40) and fit the model to the clock-like mutation densities. + +Denoting the rate at which tumor cells divide by λT and the loss rate by δT, we modeled the number of tumor cells, NT, with an exponential growth model, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$N_{\mathrm{T}}\left( t \right) = {\mathrm{e}}^{\left( {\lambda _{\mathrm{T}} - \delta _{\mathrm{T}}} \right)t}$$\end{document}NTt=eλT−δTt. We assumed that some SSNVs are already present in the founder cell of the tumor and denote their number by \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{clonal}}}}}$$\end{document}nclonal ; these mutations are clonally propagated to the entire tumor and will thus be found at frequency f = 1. Additional SSNVs are acquired during tumor growth, and we denote their number by \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{subclonal}}}}}$$\end{document}nsubclonal ; these mutations are present in a subset of the tumor only and will thus be found at f < 1. The VAF distribution of a neuroblastoma is hence a superposition of clonal and subclonal mutations accumulated before and during tumor growth, respectively. + +To model the number of subclonal mutations, we used a model of neutrally evolving tumors that accounts for the stochastic expansion of neutral subclones while assuming exponential expansion of the tumor mass overall57. This model assumes that neutral mutations are acquired at all times at rate \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\mu \lambda _{\mathrm{T}}N_{\mathrm{T}}\left( t \right)$$\end{document}μλTNTt and drift stochastically according to a supercritical birth–death process, where58:19\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$P_{1,i}\left( {\lambda _{\mathrm{T}},\delta _{\mathrm{T}},t} \right) = \left\{ {\begin{array}{*{20}{l}} {\alpha (t),i = 0} \hfill \\ {(1 - \alpha (t))\left( {1 - \beta \left( t \right)} \right)\beta \left( t \right)^{i - 1},i \ge 1} \hfill \end{array}} \right.$$\end{document}P1,iλT,δT,t=α(t),i=0(1−α(t))1−βtβti−1,i≥1where:\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\alpha \left( t \right) = \frac{{\delta _{\mathrm{T}}\left( {{\mathrm{e}}^{\left( {\lambda _{\mathrm{T}} - \delta _{\mathrm{T}}} \right)t} - 1} \right)}}{{\lambda _{\mathrm{T}}{\mathrm{e}}^{\left( {\lambda _{\mathrm{T}} - \delta _{\mathrm{T}}} \right)t} - \delta _{\mathrm{T}}}},\beta \left( t \right) = \frac{{\lambda _{\mathrm{T}}\left( {{\mathrm{e}}^{\left( {\lambda _{\mathrm{T}} - \delta _{\mathrm{T}}} \right)t} - 1} \right)}}{{\lambda _{\mathrm{T}}{\mathrm{e}}^{\left( {\lambda _{\mathrm{T}} - \delta _{\mathrm{T}}} \right)t} - \delta _{\mathrm{T}}}}.$$\end{document}αt=δTeλT−δTt−1λTeλT−δTt−δT,βt=λTeλT−δTt−1λTeλT−δTt−δT. + +Together, this yields the site frequency spectrum, S(i, μ), of subclonal mutations:20\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$S\left( {i,\mu } \right) = \mathop {\int}\limits_0^{t_{{{{\mathrm{end}}}}}} {P_{1,i}} (\lambda _{\mathrm{T}},\delta _{\mathrm{T}},t_{{{{\mathrm{end}}}}} - t)\mu \lambda _{\mathrm{T}}N_{\mathrm{T}}\left( t \right)dt.$$\end{document}Si,μ=∫0tendP1,i(λT,δT,tend−t)μλTNTtdt. + +The total number of subclonal mutations in a tumor is computed as;21\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{subclonal}}}}} = \mathop {\sum}\limits_{i = 1}^{N_{\mathrm{T}}\left( {t_{{{{\mathrm{end}}}}}} \right)} {S\left( {i,\mu } \right)} = \mathop {\sum}\limits_{i = 1}^{N_{\mathrm{T}}\left( {t_{{{{\mathrm{end}}}}}} \right)} {\mathop {\int}\limits_0^{t_{{{{\mathrm{end}}}}}} {P_{1,i}} } (\lambda _{\mathrm{T}},\delta _{\mathrm{T}},t_{{{{\mathrm{end}}}}} - t)\mu \lambda _{\mathrm{T}}N_{\mathrm{T}}\left( t \right)dt,$$\end{document}nsubclonal=∑i=1NTtendSi,μ=∑i=1NTtend∫0tendP1,i(λT,δT,tend−t)μλTNTtdt,where NT(tend) is the number of tumor cells at diagnosis. The number of mutations present in subclones of at least a and at most b cells is, in analogy to equation (21):22\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\mathop {\sum}\limits_{i = a}^b {S\left( {i,\mu } \right)} = \mathop {\sum}\limits_{i = a}^b {\mathop {\int}\limits_0^{t_{{{{\mathrm{end}}}}}} {P_{1,i}} } (\lambda _{\mathrm{T}},\delta _{\mathrm{T}},t_{{{{\mathrm{end}}}}} - t)\mu \lambda _{\mathrm{T}}N_{\mathrm{T}}\left( t \right)dt.$$\end{document}∑i=abSi,μ=∑i=ab∫0tendP1,i(λT,δT,tend−t)μλTNTtdt. + +Because clone sizes a and b are large, we approximate the sum in equation (22) by an integral, yielding:23\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\begin{array}{l}\mathop {\sum}\limits_{i = a}^b S \left( {i,\mu } \right) \approx \mathop {\int}\limits_a^b {\mathop {\int}\limits_0^{t_{{{{\mathrm{end}}}}}} {P_{1,i}} } (\lambda _{\mathrm{T}},\delta _{\mathrm{T}},t_{{{{\mathrm{end}}}}} - t)\mu \lambda _{\mathrm{T}}N_{\mathrm{T}}\left( t \right)dtdi \\\qquad\qquad\ = \mathop {\int}\limits_0^{t_{{{{\mathrm{end}}}}}} \mu \lambda _{\mathrm{T}}N_{\mathrm{T}}\left( t \right)\frac{{P_{1,b}\left( {\lambda _{\mathrm{T}},\delta _{\mathrm{T}},t_{{{{\mathrm{end}}}}} - t} \right) - P_{1,a}\left( {\lambda _{\mathrm{T}},\delta _{\mathrm{T}},t_{{{{\mathrm{end}}}}} - t} \right)}}{{\log \beta \left( {t_{{{{\mathrm{end}}}}} - t} \right)}}dt. \end{array}$$\end{document}∑i=abSi,μ≈∫ab∫0tendP1,i(λT,δT,tend−t)μλTNTtdtdi=∫0tendμλTNTtP1,bλT,δT,tend−t−P1,aλT,δT,tend−tlogβtend−tdt. + +To relate the model of mutation accumulation to the measured VAF distribution, we modeled mutations on each copy number state separately. To this end, we denote by \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{f,k}$$\end{document}nf,k the number of mutations at frequency f on genomic segments with copy number state k. Note that f reports the fraction of mutated cells among all tumor cells whereas VAFs report the fraction of mutated alleles in the tissue sample. The two quantities can be converted into each other using the following relation:24\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{VAF}}}} = \frac{{f\rho }}{\zeta }.$$\end{document}VAF=fρζ.where ζ is the average copy number of the sample as defined in equation (1). Clonal mutations are associated with frequency f = 1. The number of clonal mutations falling within genomic segments of copy number \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$k,n_{1,k}$$\end{document}k,n1,k, is expected to scale with gk, the fraction of the genome at copy number k, according to25\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{1,k} = n_{{{{\mathrm{clonal}}}}}\frac{{kg_k}}{{\mathop {\sum}\nolimits_k {kg_k} }}.$$\end{document}n1,k=nclonalkgk∑kkgk. + +To relate true with measured VAFs, we assumed that the latter are binomially distributed around the former according to\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$B\left( {\frac{{f\rho }}{\zeta },C_k} \right),$$\end{document}Bfρζ,Ck,where \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$C_k \propto {{{\mathrm{Pois}}}}(\widehat {C_k})$$\end{document}Ck∝Pois(Ck^) is the sequencing coverage on a segment with copy number k, and ρ is tumor cell content. + +Because we model mutation accumulation with neutral tumor expansion, we included all tumors with well-defined subclonal tails and absence of subclonal selection. To this end, we visually inspected the VAF histograms and excluded 29 cases with poorly resolved subclonal tails. To identify subclonal selection we ran Mobster12, excluding mutations on sex chromosomes and computing pseudoheterozygous VAFs, in the following termed \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\widetilde {{{{\mathrm{VAF}}}}}$$\end{document}VAF~ and defined as 50% of the mutated sample fraction, SF. The measured VAFs relate to SF according to \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{VAF}}}} = \frac{k}{\zeta }{{{\mathrm{SF}}}}$$\end{document}VAF=kζSF, where k is the number of alleles carrying the mutation. It follows that \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\widetilde {{{{\mathrm{VAF}}}}} = \frac{\zeta }{{2k}}{{{\mathrm{VAF}}}}$$\end{document}VAF~=ζ2kVAF. We excluded mutations with \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\widetilde {{{{\mathrm{VAF}}}}} < 0.1$$\end{document}VAF~<0.1 from the fit and ran the Mobster setting autosetup = “FAST”. This resulted in an additional exclusion of nine cases for which Mobster suggested subclonal selection. Thus we selected 62 tumors with well-resolved subclonal VAF histograms and no evidence of subclonal selection for parameter inference (Supplementary Table 1). + +We estimated \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{clonal}}}}}$$\end{document}nclonal, μ and \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\delta _{\mathrm{T}}/\lambda _{\mathrm{T}}$$\end{document}δT/λT using ABC–SMC as implemented in pyABC56. We used a population size of 1,000 parameter samples and prior distributions as outlined in Supplementary Table 12. Termination criteria were the same as above (Modeling neuroblastoma initiation). + +We stratified measured VAFs by copy number, excluding copy numbers >4 or present on, in total, <108 bp. For each copy number k, we merged mutations from the clonal VAF peaks constituted by amplified and non-amplified mutations (see Mutation timing for the definition of amplified and non-amplified clonal mutations). To this end, we first assessed the average coverage \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\widehat {C_k}$$\end{document}Ck^ from all mutations falling on segments with copy number k. Then, we classified mutations at copy number k as amplified clonal if \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\frac{{Q_{l - 1}^{0.95}}}{{\widehat {C_k}}} < {{{\mathrm{VAF}}}}_k \le \frac{{Q_l^{0.95}}}{{\widehat {C_k}}}$$\end{document}Ql−10.95Ck^ f$$\end{document}Fk,ev(f)=∑VAFk>f, where f runs from 0.05 to 1.00 in steps of size 0.05, and extrapolated the cumulative mutation counts to the whole genome with multiplication by \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\frac{{\mathop {\sum}\nolimits_k {g_k} }}{{g_k}}$$\end{document}∑kgkgk. At a sampled parameter set for \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{{{{\mathrm{clonal}}}}}$$\end{document}nclonal, μ and δT, the following steps were performed for each copy number state k included in the analysis:Sample for each clonal mutation a sequencing coverage Ck according to \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{Pois}}}}(\widehat {C_k})$$\end{document}Pois(Ck^).Sample for each clonal mutation a VAF according to equation (16).Determine \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{f,k;f \ne 1}$$\end{document}nf,k;f≠1 from equation (13), assuming a tumor size of 109 cells at diagnosis, and evaluating equation (13) in bins of size 0.05 at the lower limit:\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{f,k;f \ne 1} \approx \mathop {\sum}\limits_{i = f10^9}^{\left( {f + 0.05} \right)10^9} {S\left( {i,\mu } \right)} .$$\end{document}nf,k;f≠1≈∑i=f109f+0.05109Si,μ.Sample for each subclonal mutation a sequencing coverage Ck according to \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{Pois}}}}(\widehat {C_k})$$\end{document}Pois(Ck^) and a VAF according to equation (16).Compute the cumulative mutation counts, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$F_{k,{{{\mathrm{sim}}}}}{{{\mathrm{(f)}}}} = {\sum} {{{{\mathrm{VAF}}}}_k} > f$$\end{document}Fk,sim(f)=∑VAFk>f, where f runs from 0.05 to 1.00 in steps of size 0.05. + +Sample for each clonal mutation a sequencing coverage Ck according to \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{Pois}}}}(\widehat {C_k})$$\end{document}Pois(Ck^). + +Sample for each clonal mutation a VAF according to equation (16). + +Determine \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{f,k;f \ne 1}$$\end{document}nf,k;f≠1 from equation (13), assuming a tumor size of 109 cells at diagnosis, and evaluating equation (13) in bins of size 0.05 at the lower limit:\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$n_{f,k;f \ne 1} \approx \mathop {\sum}\limits_{i = f10^9}^{\left( {f + 0.05} \right)10^9} {S\left( {i,\mu } \right)} .$$\end{document}nf,k;f≠1≈∑i=f109f+0.05109Si,μ. + +Sample for each subclonal mutation a sequencing coverage Ck according to \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{\mathrm{Pois}}}}(\widehat {C_k})$$\end{document}Pois(Ck^) and a VAF according to equation (16). + +Compute the cumulative mutation counts, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$F_{k,{{{\mathrm{sim}}}}}{{{\mathrm{(f)}}}} = {\sum} {{{{\mathrm{VAF}}}}_k} > f$$\end{document}Fk,sim(f)=∑VAFk>f, where f runs from 0.05 to 1.00 in steps of size 0.05. + +The cost function is26\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$d = \mathop {\sum}\limits_k {\mathop {\sum}\limits_f {\left( {F_{k,{{{\mathrm{sim}}}}} - F_{k,{{{\mathrm{ev}}}}}} \right)^2} } \frac{{g_k}}{{\mathop {\sum}\nolimits_{k\prime } {g_{k\prime }} }}.$$\end{document}d=∑k∑fFk,sim−Fk,ev2gk∑k′gk′. + +In the previous sections we describe how we inferred the parameters for our models of neuroblastoma initiation and growth from mutation data. This yielded estimates for the rates at which cells are lost, as well as the rates at which neutral and oncogenic mutations are acquired in units of cell divisions. To convert these estimates to real time we estimated the actual cell division rate, using the age distribution at diagnosis and the approximate tumor size at diagnosis. + +We reasoned that the time span between gastrulation and tumor diagnosis (tD) consists of two phases: premalignancy up to the formation of the MRCA and malignant growth of the tumor thereafter. Hence,27\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$t_{\mathrm{D}} = \frac{{\tilde m_{{{{\mathrm{MRCA}}}}}}}{{\lambda \mu }} + \frac{{\log N_{\mathrm{T}}(t_{\mathrm{D}})}}{{\lambda \left( {1 - \frac{\delta }{{s\lambda }}} \right)}},$$\end{document}tD=m~MRCAλμ+logNT(tD)λ1−δsλ,where we assume exponential tumor growth until diagnosis. Thus the cell division rate, λ, can be expressed as28\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\lambda = \frac{1}{{t_{\mathrm{D}}}}\left( {\frac{{\tilde m_{{{{\mathrm{MRCA}}}}}}}{\mu } + \frac{{\log N_{\mathrm{T}}(t_{\mathrm{D}})}}{{1 - \frac{\delta }{{s\lambda }}}}} \right).$$\end{document}λ=1tDm~MRCAμ+logNT(tD)1−δsλ. + +To estimate λ with equation (28), we combined the clinical information with parameter estimates from our models as follows: we know tD (the age at diagnosis, A, plus approximately 250 days of embryogenesis after gastrulation) from the clinical data, and estimate that a tumor of a few cubic centimeters consists of the order of \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$N_{{{\mathrm{T}}}}(t_{\mathrm{D}}) = 10^9$$\end{document}NT(tD)=109 cells. As described below (Mutation timing and Modeling neuroblastoma initiation), we also have estimates for mutation density at the MRCA, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\tilde m_{{{{\mathrm{MRCA}}}}}$$\end{document}m~MRCA, and for the SSNV rate per actual cell division and haploid genome, μ. Finally, we obtained an estimate for the effective rate of acquisition of (neutral) SSNVs during tumor growth from the subclonal VAF histograms of 62 tumors (compare with Modeling mutation accumulation during tumor growth), which is related to δ and s as29\documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\mu _{{{{\mathrm{eff}}}}} = \frac{\mu }{{1 - \delta _{\mathrm{T}}/\lambda _{\mathrm{T}}}} = \frac{\mu }{{1 - \frac{\delta }{{s\lambda }}}}.$$\end{document}μeff=μ1−δT/λT=μ1−δsλ. + +Substituting equation (29) in equation (28), we obtain for each of the 62 tumors (labeled with the index i) an estimate for the division rate with mean, \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\left\langle {\lambda _i} \right\rangle = \frac{1}{{2\left\langle \mu \right\rangle (A_i + 250\,{{{\mathrm{days}}}})}}\left( {\left\langle {2\tilde m_{{{{\mathrm{MRCA,i}}}}}} \right\rangle + \log 10^9\left\langle {\mu _{{{{\mathrm{eff,i}}}}}} \right\rangle } \right)$$\end{document}λi=12μ(Ai+250days)2m~MRCA,i+log109μeff,i, and s.d. (standard deviation), \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\sigma \left( {\lambda _i} \right) = \frac{1}{{2\left\langle \mu \right\rangle (A_i + 250\,{{{\mathrm{days}}}})}}$$\end{document}σλi=12μ(Ai+250days) \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\left( \frac{{2\left\langle {\tilde m_{{{{\mathrm{MRCA,i}}}}}} \right\rangle + \log 10^9\left\langle {\mu _{{{{\mathrm{eff,i}}}}}} \right\rangle }}{{2\mu }}\sigma \right.$$\end{document}2m~MRCA,i+log109μeff,i2μσ \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\left\langle {2\mu } \right\rangle + \sigma \left( {2\tilde m_{{{{\mathrm{MRCA,i}}}}}} \right) + \sigma \left( {\mu _{{{{\mathrm{eff,i}}}}}} \right) \Big)$$\end{document}2μ+σ2m~MRCA,i+σμeff,i, in actual time. Note that factor 2 accounts for the fact that μ and \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\tilde m_{{{{\mathrm{MRCA,i}}}}}$$\end{document}m~MRCA,i measure mutation rate and density, respectively, per haploid genome. From equation (29), we also get an estimate of \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\delta _{{\mathrm{T}}{{{\mathrm{,i}}}}}/\lambda _{{\mathrm{T}}{{{\mathrm{,i}}}}}$$\end{document}δT,i/λT,i with expectation \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\left\langle {\frac{{\delta _{{\mathrm{T}}{{{\mathrm{,i}}}}}}}{{\lambda _{{\mathrm{T}}{{{\mathrm{,i}}}}}}}} \right\rangle = 1 - \frac{{2\left\langle \mu \right\rangle }}{{\left\langle {\mu _{{{{\mathrm{eff,i}}}}}} \right\rangle }}$$\end{document}δT,iλT,i=1−2μμeff,i and s.d. \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$${{{{\sigma }}}}\left( {\delta _{{\mathrm{T}}{{{\mathrm{,i}}}}}/\lambda _{{\mathrm{T}}{{{\mathrm{,i}}}}}} \right) = \frac{{\sigma \left\{ {2\mu } \right\}}}{{\left\langle {\mu _{{{{\mathrm{eff,i}}}}}} \right\rangle }} + \frac{{2\left\langle \mu \right\rangle }}{{\left( {\left\langle {\mu _{{{{\mathrm{eff,i}}}}}} \right\rangle } \right)^2}}{{{{\sigma }}}}\left( {\mu _{{{{\mathrm{eff,i}}}}}} \right)$$\end{document}σδT,i/λT,i=σ2μμeff,i+2μμeff,i2σμeff,i. Fitting effective mutation rates to the VAF histograms of each of the selected 62 tumors individually hence yields tumor-specific division rates along with relative death rates, which we stratify by TMM. + +Finally, one also obtains an estimate for the daily mutation rate during tumor initiation by computing \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\mu \lambda _{{{{\mathrm{T}}}},{{{\mathrm{i}}}}}$$\end{document}μλT,i with associated uncertainty \documentclass[12pt]{minimal} + \usepackage{amsmath} + \usepackage{wasysym} + \usepackage{amsfonts} + \usepackage{amssymb} + \usepackage{amsbsy} + \usepackage{mathrsfs} + \usepackage{upgreek} + \setlength{\oddsidemargin}{-69pt} + \begin{document}$$\mu {{\Delta }}\lambda _{{{{\mathrm{T}}}},{{{\mathrm{i}}}}} + \lambda _{{{{\mathrm{T}}}},{{{\mathrm{i}}}}}\sigma (\mu )$$\end{document}μΔλT,i+λT,iσ(μ), which relates molecular clock to real time. For this purpose, we average across the inferences from the 44 primary tumors/metastasis, excluding 18 relapsed tumors among the 62 submitted for analysis. + +All statistical tests were computed with R (v.3.6.0 and v.4.0.0), and details (statistical tests and whether one- or two-sided, exact sample size, P values and test statistics) are specified in the respective figures and accompanying Source data. This is a retrospective analysis of tumor material that was collected as part of the diagnostic workflow of the German Neuroblastoma trial by the Society for Pediatric Oncology and Hematology and collected in the Neuroblastoma tumor bank. Hence, no statistical method was used to predetermine sample size. All analyzed tumors had a clear copy number profile, tumor cell content ≥25% and no hypermutation genotype; no data were excluded from the analyses. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment. + +Analysis was performed with R (v.3.6.0 and v.4.0.0) and python v.3.6.1. We used the following R packages: openxlsx v.4.1.5, ggsignif v.0.6.0, ggbeeswarm v.0.6.0, gridExtra v.2.3, RColorBrewer v.1.1-2, HDInterval v.0.2.2, cdata v.1.1.8, moments v.0.14, Hmisc v.4.4-0, scales v.1.1.1, bedr v.1.0.7, circlize v.0.4.10, ggplot2 v.3.3.2 (ref. 59), Bioconductor v.3.15 (ref. 60), ggbio v.1.34.0 (ref. 61), ggpubr v.0.4.0, pammtools v.0.2.2 (ref. 62), ComplexHeatmap v.2.5.1 (ref. 63), BSgenome.Hsapiens.UCSC.hg19 v.1.4.3, MASS v.7.3-51.6 (ref. 64), GenomicRanges v.1.38.0 (ref. 65), reshape2 v.1.4.4 (ref. 66), mixtools v.1.2.0 (ref. 67), dplyr v.1.0.0, survminer v.0.4.8, survival68 v.3.1-12, wesanderson v.0.3.6, cowplot v.1.1.1, mobster12 v.1.0.0, CNAqc v.1.0.0 and mmsig51 v.0.0.0.9000; and the python packages SigProfilerMatrixGenerator69 v.1.1.26, SigProfilerExtractor50 v.1.1.1 and pyABC56 v.0.9.13. + +Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. + +Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41588-023-01332-y. + + +Supplementary InformationSupplementary notes 1 and 2 and tables 11 and 12. +Reporting summary. +Peer review file. +Supplementary Table 1Patient characteristics and driver genes in discovery and validation cohorts. + + +Supplementary notes 1 and 2 and tables 11 and 12. + +Reporting summary. + +Peer review file. + +Patient characteristics and driver genes in discovery and validation cohorts. \ No newline at end of file diff --git a/references_cache/PMID_39635126.md b/references_cache/PMID_39635126.md new file mode 100644 index 00000000000..0aaab2cd576 --- /dev/null +++ b/references_cache/PMID_39635126.md @@ -0,0 +1,328 @@ +--- +reference_id: PMID:39635126 +extractor_version: 1 +xml_extraction_version: 1 +title: Copy-number dosage regulates telomere maintenance and disease-associated pathways in neuroblastoma. +authors: +- Burkert M +- Blanc E +- Thiessen N +- Weber C +- Toedling J +- Monti R +- Dombrowe VM +- Stella de Biase M +- Kaufmann TL +- Haase K +- Waszak SM +- Eggert A +- Beule D +- Schulte JH +- Ohler U +- Schwarz RF +journal: iScience +year: '2024' +doi: 10.1016/j.isci.2024.110918 +publication_types: +- Journal Article +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# Copy-number dosage regulates telomere maintenance and disease-associated pathways in neuroblastoma. +**Authors:** Burkert M, Blanc E, Thiessen N, Weber C, Toedling J, Monti R, Dombrowe VM, Stella de Biase M, Kaufmann TL, Haase K, Waszak SM, Eggert A, Beule D, Schulte JH, Ohler U, Schwarz RF +**Journal:** iScience (2024) +**DOI:** [10.1016/j.isci.2024.110918](https://doi.org/10.1016/j.isci.2024.110918) + +## Content + +Telomere maintenance in neuroblastoma is linked to poor outcome and caused by either telomerase reverse transcriptase (TERT) activation or through alternative lengthening of telomeres (ALT). In contrast to TERT activation, commonly caused by genomic rearrangements or MYCN amplification, ALT is less well understood. Alterations at the ATRX locus are key drivers of ALT but only present in ∼50% of ALT tumors. To identify potential new pathways to telomere maintenance, we investigate allele-specific gene dosage effects from whole genomes and transcriptomes in 115 primary neuroblastomas. We show that copy-number dosage deregulates telomere maintenance, genomic stability, and neuronal pathways and identify upregulation of variants of histone H3 and H2A as a potential alternative pathway to ALT. We investigate the interplay between TERT activation, overexpression and copy-number dosage and reveal loss of imprinting at the RTL1 gene associated with poor clinical outcome. These results highlight the importance of gene dosage in key oncogenic mechanisms in neuroblastoma. + +Neuroblastoma is the most common extracranial solid tumor in children accounting for 6–10% of malignancies1 and 9% of pediatric cancer deaths.2 Clinical manifestations range from high-risk cases with poor survival rates despite multimodal treatment to tumors that spontaneously regress without intervention.3 Incidence is highest in the first year of life and only 5% of diagnoses are made in patients older than ten years.1 Survival rates rapidly decrease for diagnosis made in children older than 1 year of age.2 + +Genetically, neuroblastoma is characterized by a low single-nucleotide variant (SNV) burden and only few recurrently mutated genes,4 but frequent somatic copy-number alterations (SCNAs).5,6,7 Amplification of the oncogenic transcription factor MYCN, often through extrachromosomal circular DNAs (ecDNA),8,9 is found in 20% of tumors and a key clinical indicator for high-risk disease and poor prognosis.3,10 In addition, recurrent segmental gains and losses, including 17q gains and losses of 1p and 11q6,11,12 are associated with unfavorable outcomes.13 Conversely, numerical alterations in chromosomes and whole-genome doubling (WGD) are features associated with better survival rates.3 In line with adult tumors,14 amplifications of e.g., MYCN and ALK and their downstream targets15,16 and larger segmental gains and losses correlate well with local RNA levels,17,18 which in turn predict patient survival.15,16,18 + +Telomere maintenance leading to replicative immortality19 is a common mechanism in high-risk neuroblastoma,20,21,22 while ineffective telomere maintenance may explain the spontaneous regressions observed in low-risk cases, particularly in stage 4S neuroblastomas.3,23 Canonical telomere maintenance (CTM) involves activation of the telomerase reverse transcriptase (TERT) gene either indirectly as a downstream effect of MYCN amplification, or directly through genomic rearrangements at the TERT locus.20,22 Alternative lengthening of telomeres (ALT) in tumors that lack TERT activation24 involves DNA recombination induced by breaks at telomeric sequences25 and is characterized by single stranded telomeric (CCCTAA)n sequences.26 Generally, ALT is associated with loss of function mutations in the ATRX and DAXX genes27 as well as missense mutations in H3F3A28 and has been found in 50% of all cancer types of the Pan-Cancer Analysis of Whole Genomes (PCAWG) cohort.29 Affected tumors show excess telomere length compared to normal tissue and other tumors, including those with activated TERT.29 In neuroblastoma ALT is associated with ATRX alterations,20,30,31 significantly enriched in relapse cases and associated with poor outcome independent of other risk markers.21,31,32 While previous studies have highlighted the molecular characteristics of telomere maintenance in neuroblastoma,20,30,31,33,34 +ATRX mutations were only found in 25% of high-risk and 50–60% of ALT-positive neuroblastomas,30,31,35 suggesting additional yet unrecognized mechanisms of ALT activation. Telomere maintenance is therefore a key phenotypic property of neuroblastoma cells and a prime example of phenotypic convergence in cancer evolution,36 where multiple somatic aberrations act individually or in concert to activate telomere maintenance pathways by modulating gene expression. + +To reveal such mechanisms, we here investigate the effect of genomic instability on total and allele-specific gene expression (ASE) and telomere maintenance in 115 primary neuroblastomas. We analyze whole genome sequencing (WGS) and RNA-seq from tumors and WGS of matched normals, characterize local genetic effects on gene expression variability, and examine the role of copy-number dosage in telomere maintenance and survival. + +We assembled a cohort of matched tumor WGS and RNA-seq and normal WGS from blood from 115 primary neuroblastoma samples, including 52 samples from the University Hospital of Cologne, previously reported in the study by Peifer et al.,20 and 63 new specimens from the GPOH-NB2004 clinical trial. All samples were jointly processed using unified pipelines to limit cohort-specific biases (Figure 1A; STAR methods) and stratified according to the GPOH-NB2004 clinical trial protocol37 into 66 high-risk, 6 medium-risk, and 43 low-risk tumors (Figure S1) and equipped with clinical annotations including age, sex, and survival times (Table S1).Figure 1Molecular profiles of 115 neuroblastoma tumors(A) Overview of integrated data processing of tumor WGS and RNA-seq as well as WGS of matched normal samples from 115 neuroblastomas to determine somatic alterations, telomere maintenance mechanisms and genetic effects on gene expression.(B) Overview of telomere maintenance mechanisms: canonical telomere maintenance is characterized by TERT activation. In neuroblastoma TERT activation is frequently induced by MYCN amplification or genomic-rearrangements at the TERT locus. ALT is characterized by excess telomere length and C-circles.(C) Molecular and clinical characterization of 115 neuroblastoma primary tumors (columns).ALT, Alternative lengthening of telomeres; MYCN-amp, MYCN amplification; TERT-re, TERT rearrangement; ATRX-mut, ATRX mutation; ALK-mut, ALK mutation; LOH, Loss of heterozygosity; WGD, Whole-genome doubling; CIN, Chromosome instability index; TMM = None indicates no evidence for a telomere maintenance mechanism detected. + +Molecular profiles of 115 neuroblastoma tumors + +(A) Overview of integrated data processing of tumor WGS and RNA-seq as well as WGS of matched normal samples from 115 neuroblastomas to determine somatic alterations, telomere maintenance mechanisms and genetic effects on gene expression. + +(B) Overview of telomere maintenance mechanisms: canonical telomere maintenance is characterized by TERT activation. In neuroblastoma TERT activation is frequently induced by MYCN amplification or genomic-rearrangements at the TERT locus. ALT is characterized by excess telomere length and C-circles. + +(C) Molecular and clinical characterization of 115 neuroblastoma primary tumors (columns). + +ALT, Alternative lengthening of telomeres; MYCN-amp, MYCN amplification; TERT-re, TERT rearrangement; ATRX-mut, ATRX mutation; ALK-mut, ALK mutation; LOH, Loss of heterozygosity; WGD, Whole-genome doubling; CIN, Chromosome instability index; TMM = None indicates no evidence for a telomere maintenance mechanism detected. + +Normal samples from blood were genotyped and phased at common germline variant sites (STAR methods). Total and ASE was quantified using phased variants and variant effects on gene expression in cis were quantified by genome-wide expression quantitative trait locus (eQTL) mapping (STAR methods).14 To explore the mutational landscape we determined somatic SNVs, structural variants (SVs), and allele-specific SCNAs from WGS (STAR methods). + +We first set out to determine the primary telomere maintenance mechanism (Figure 1B) and genetic alterations across all 115 tumors by examining somatic SNV, SV, SCNA, and expression data as well as WGS-based estimates of telomere length (STAR methods). We found MYCN amplifications in 23 tumors (20%), rearrangements affecting the TERT locus in 19 tumors (17%) and ATRX mutations in 12 tumors (10%), comprising 7 focal deletions, 4 missense or nonsense mutations and one tumor affected by a structural rearrangement (NBL54) (Figures 1C and S2). + +To determine the ALT status of tumors we estimated telomere lengths relative to the matched normal tissue by the abundance of telomeric repeat sequences from WGS (Figure S3A; STAR methods).38 We found 21 tumors to show increased telomere lengths, of which we assigned 20 to the ALT group, as one (NBL54) also harbored a TERT rearrangement and upregulation of TERT (Figures 1C and S3A). We validated our ALT classification by comparison against experimentally determined status of ALT-associated PML-nuclear bodies (APB)21 and the presence of circular partially single stranded extrachromosomal telomeric repeat sequences (C-circles)31 in subsets of 52 and 36 of analyzed tumor samples, respectively (Figures S3B and S3C). We found strong correspondence between our ALT classification and APBs (p = 5.47 × 10−9, one-sided Fisher’s exact test; sensitivity: 0.86; specificity: 0.97) as well as C-circles (p = 4.33 × 10−8, one-sided Fisher’s exact test; sensitivity: 0.91; specificity: 1.00). ALT was detected in 19 of 66 (29%) high-risk tumors, similar to the ALT prevalence estimate in an independent neuroblastoma cohort.35 Among ALT tumors 9/20 (45%) harbored ATRX alterations and a single tumor (NBL49) a H3F3A missense mutation (p.A48E). We did not find any DAXX alterations in ALT tumors of our cohort. While ATRX altered samples had significantly longer telomeres (p = 1.72 × 10−6, one sided Wilcoxon rank-sum test) (Figure S4), 10 out of 20 ALT samples (50%) did not show any mutation in previously described ALT-associated genes, pointing toward alternative activation of the ALT pathway. We found 8 tumors without MYCN amplification or TERT rearrangements to show high TERT expression (Figure S5; STAR methods), of which 4 were not classified as ALT and assigned to the TERT-high group (STAR methods). + +Except for three tumors, MYCN amplifications, TERT rearrangements, and long telomeres were mutually exclusive (Figure 1C), in support of convergence toward a common high-risk phenotype characterized by telomere maintenance.20,21,22 MYCN amplifications were also mutually exclusive to ATRX alterations, corroborating findings on incompatibility of these two molecular traits.39 Of 43 low risk tumors 40 (93%) showed neither increased telomere length (log ratio >0.5) nor elevated TERT expression (Z score > −0.10). Interestingly, active telomere maintenance was predicted in three low risk tumors (NBL09, NBL23, and CB2035), which all showed disease progression. Notably, we did not find any MYCN amplifications in ALT samples and only a single sample with both TERT rearrangement and long telomeres (NBL54). + +In summary, 43 tumors were classified as showing CTM characterized by TERT activation through MYCN amplification, TERT rearrangement or high TERT expression. 20 tumors showed ALT, one tumor exhibited a mixed phenotype (CTM and ALT), and in 51 tumors no evidence for any telomere maintenance mechanism was found (Figures 1B and 1C; STAR methods). 58 of 66 high-risk tumors (88%) were classified as CTM or ALT in contrast to 7 high-risk tumors (11%) without signs of telomere maintenance. + +We next determined allele-specific SCNAs and overall ploidy from WGS (STAR methods) and classified copy-number segments into states loss, shallow loss, neutral, weak gain, medium gain, strong gain, and focal amplification (Figure 2A) and into allelic imbalance states balance, weak imbalance, strong imbalance, amplification, and LOH (Figure S6; STAR methods). We further detected WGD events by phylogenetic copy number analysis as recently described.40 On average 50% of the genomic regions harbored SCNAs, 31% of genomic regions showed gains and losses relative to ploidy, and 44 tumors (38%) showed WGD (Figures 1C and S7). We identified gains in 17%, losses in 15%, and amplifications in <0.1% of genomic regions, with distinct hotspots on the cohort level (Figure 2A). We found WGDs to be overrepresented in tumors without telomere maintenance (26 of 51, expected 20, p = 0.03, Fisher-exact test), as opposed to ALT (2 of 20, expected 8, p = 0.01). Tumors with CTM in contrast did now show enrichment (16 of 43, expected 16, p = 1.0).Figure 2Genetic- and allelic dosage effects in gene regulation(A) Number of samples affected and copy-number state summarized in 5 Mb genomic bins on chromosomes 1–22.(B) Number of genes affected by copy-number imbalance and AEI for samples with (n = 23) and without (n = 92) MYCN-amplification (MNA).(C) Number of genes affected by copy-number imbalance and allelic expression imbalance (AEI) per sample. Gray line represents linear regression fit, with light gray ribbon showing the 95% confidence interval. Pearson correlation coefficient r and p-value P are shown. CN, copy-number. WGD, whole-genome doubling.(D) Quantification of genetic effects on the variance of total expression of expressed genes (n = 13,632).(E) Quantification of genetic effects on the variance of allele-specific expression (ASE) of ASE informative genes (n = 10,656).(F) Distribution of within-gene expression percentile per sample and gene by copy-number state: loss (n = 66,030), shallow loss (n = 233,696), neutral (n = 1,322,030), weak gain (n = 160,861), medium gain (n = 177,750), strong gain (n = 8,453), amplification (n = 383). (G) Proportion of RNA reads from major allele per sample and gene by copy-number balance state: balance (n = 1,224,053), weak imbalance (n = 549,177), strong imbalance (n = 165,790), amplification (n = 383).(H) Genome-wide gene-level copy-number dosage effects. Significant copy-number dosage effect genes (FDR <0.05, Benjamini-Hochberg) indicated in color scale, others in light gray.Boxplot midlines in (B, D, E, F, and G) mark median; upper and lower hinges extend to first and third quartile; upper and lower whiskers extend to the smallest and largest value max. 1.5 × IQR; p value of two-sided Wilcoxon test is shown between groups in (B, F, and G). + +Genetic- and allelic dosage effects in gene regulation + +(A) Number of samples affected and copy-number state summarized in 5 Mb genomic bins on chromosomes 1–22. + +(B) Number of genes affected by copy-number imbalance and AEI for samples with (n = 23) and without (n = 92) MYCN-amplification (MNA). + +(C) Number of genes affected by copy-number imbalance and allelic expression imbalance (AEI) per sample. Gray line represents linear regression fit, with light gray ribbon showing the 95% confidence interval. Pearson correlation coefficient r and p-value P are shown. CN, copy-number. WGD, whole-genome doubling. + +(D) Quantification of genetic effects on the variance of total expression of expressed genes (n = 13,632). + +(E) Quantification of genetic effects on the variance of allele-specific expression (ASE) of ASE informative genes (n = 10,656). + +(F) Distribution of within-gene expression percentile per sample and gene by copy-number state: loss (n = 66,030), shallow loss (n = 233,696), neutral (n = 1,322,030), weak gain (n = 160,861), medium gain (n = 177,750), strong gain (n = 8,453), amplification (n = 383). (G) Proportion of RNA reads from major allele per sample and gene by copy-number balance state: balance (n = 1,224,053), weak imbalance (n = 549,177), strong imbalance (n = 165,790), amplification (n = 383). + +(H) Genome-wide gene-level copy-number dosage effects. Significant copy-number dosage effect genes (FDR <0.05, Benjamini-Hochberg) indicated in color scale, others in light gray. + +Boxplot midlines in (B, D, E, F, and G) mark median; upper and lower hinges extend to first and third quartile; upper and lower whiskers extend to the smallest and largest value max. 1.5 × IQR; p value of two-sided Wilcoxon test is shown between groups in (B, F, and G). + +Next, we determined ASE in all 115 tumors. Briefly, read counts from RNA-seq were tallied up at heterozygous germline variants and aggregated to haplotype counts per gene using statistical phasing (STAR methods). In line with prior observations,41 we found lower genomic instability in MYCN-amplified tumors compared to non-amplified tumors, visible both in the number of copy-number-imbalanced genes (p = 3.7 × 105) and genes with ASE (p = 0.0023) (one-sided Wilcox rank-sum test) (Figure 2B). Interestingly, we also identified 4 out of 23 (17%) MYCN amplified tumors with significantly more allelically imbalanced genes compared to non-MYCN-amplified samples (37% of genes). All 4 tumors showed signs of WGD and overall high chromosomal instability (>80%) (Figure 2C; Table S1) and 3 out of 4 of these patients died from the disease. Increased genomic instability in MYCN amplified tumors thus might cf. an additional risk factor similar to earlier findings on chromosomal losses in MYCN amplified tumors.12 + +Focal amplifications were detected in 32 tumors and recurrently affected 35 genes, including COSMIC census genes42 +MYCN, ALK, CDK4, LRIG3, MDM2, and PTPRB (Figure S8). Genes amplified in three or more tumors were exclusively detected in MYCN co-amplified regions on 2p24, likely associated with ecDNA presence. LOH affected 5% of genomic regions, and 1p36 LOH was found in 26 tumors (22%) of which 18 also showed amplification of MYCN. Shallow losses of 1p36 without LOH were detected in 6 tumors (5%). LOH of 11q was found in 42 tumors (37%), out of which 17 affected ALT tumors, 8 affected tumors with TERT rearrangements and 3 MYCN-amplified tumors. We found 11q losses without LOH in 23 tumors (20%), of which 19 were classified as shallow losses. 11q loss was found in 18 of 19 ALT tumors (90%), in line with previous reports on frequent 11q losses in ALT.31 Gains of 17q were highly abundant and affected 104 tumors (90%). Interestingly, 10 out of 13 strong 17q gains (78%) affected ALT tumors, suggesting that relatively higher 17q copy-numbers may be linked to the ALT phenotype in neuroblastoma. SV analysis indicated substantial heterogeneity in SV burden between tumors (Figure S9A) and revealed frequent interchromosomal rearrangements (Figures S9B and S9E). We analyzed the frequency of somatic SVs in 500 kb segments along the genome and detected recurrent SV breakpoints at the MYCN, TERT, ATRX, 11q13, and 17q21 loci (Figures S9C–S9E), confirming previous findings on SV frequencies in a subset of tumors analyzed.20 + +To investigate the effect of SCNAs on patient survival systematically we associated allelic copy-number imbalances on the level of chromosome arms and in non-overlapping 5Mb bins with mortality (STAR Methods) and found expected associations at 1p and the MYCN locus as well as a yet undescribed association of 17p imbalance (Figures S10A–S10C). Five tumors of deceased patients harbored extreme copy-number imbalances (>0.9) due to loss of 17p (Figure S11A), pointing toward elevated risk conferred through chromosomal loss. However, also 10 out of 26 donors (38%) with tumors harboring imbalanced gains died from the disease. We compared survival probabilities using the Kaplan-Meier method and found that survival was significantly reduced for tumors with 17p imbalance (p = 5.2 × 10−4) (Figure S11B). Similarly, Cox proportional hazard regression showed that 17p imbalance is significantly associated with mortality (p = 1.44 × 10−5), independent of MYCN amplification (p = 4.32 × 10−6) (Figure S12). Notably, 17p LOH is frequent in neuroblastoma cell lines,43 but its occurrence in primary neuroblastoma is less well described. Interestingly, we did not find TP53 missense mutations or SVs, suggesting that 17p loss might act through downregulation of neuronal genes (Figures S11C and S11D; Table S2) or through a second hit in TP53 that occurred after the sampling time point. + +We next sought to identify the effects of genetic alterations and to quantify their contribution to gene regulation in neuroblastoma. We used linear models to predict both total gene expression and the ASE ratio per gene from its lead cis-eQTL variant, proximal SV breakpoints, copy-number status, and local mutational SNV burdens in promoter and gene body (see STAR methods and14). For ASE analysis, an average of 5,768 (2,691–7,544) expressed genes were considered per tumor. In keeping with the literature, we found SCNAs to have the strongest effect among all genetic factors on both ASE and total gene expression,44 explaining an estimated 30.3% and 8.0% of variance in ASE and total expression, respectively (Figures 2D and 2E), and demonstrating a clear allele-specific copy-number dosage effect on gene expression on average (Figures 2F and 2G). Lead germline cis-eQTL variants were the second largest genetic contributor explaining 1.6% of variance in ASE and 2.6% of variance in total gene expression. Despite emerging evidence of targeted cis-deregulation in neuroblastoma,20,33,45 overall somatic SVs and SNVs explain the least amount of variance in ASE and total expression with less than 1.0% and 1.2%, respectively, on average, in line with recent findings in adult tumors.14 + +Even though SCNAs exhibit a strong allelic dosage effect on gene expression, transcription levels of genes are subject to transcriptional adaptations and buffering.46,47 To investigate dosage sensitivity in our cohort systematically, we examined copy-number components in our linear models and found statistically significant copy-number effects that explain between 2.4% and 71.0% of observed variance in gene expression (Figures 2H and S13). We ranked all protein-coding genes by expression variance explained and tested for pathway enrichment using gene set enrichment analysis (GSEA, STAR methods). We found 69 Reactome pathways enriched (FDR <0.05) for copy-number dosage effects (Table S3), of which 25 remained after accounting for overlapping gene sets (Figure S14). Notably, dosage sensitive genes were enriched in pathways involved in cell cycle and DNA repair, and in regulation of tumor suppressor genes TP53, PTEN, and RUNX3. In contrast, conducting the same GSEA analysis on genes ranked by total copy-number alone did not yield any significant pathway enrichment. + +We queried TERT gene expression in all tumors and found both MYCN amplified and TERT-rearranged samples to have significantly higher TERT expression than those lacking both molecular features (Figure 3A), in line with previous observations.20,48 Comparison of TERT expression with telomere length estimates confirmed the existence of two distinct groups of high risk tumors: those with high TERT expression but short telomeres and those with low TERT expression but increased telomere length, indicative of ALT (Figure 3B). Event-free survival was significantly reduced for all of the three inferred telomere maintenance mechanisms compared to tumors without telomere maintenance (MYCN-amp: p = 1.01 × 10−6; TERT-re: 5.56 × 10−5; TERT-high: 1.16 × 10−7; ALT: p = 2.47 × 10−5; Cox proportional hazards regression) (Figure 3C). When considering overall survival, we found significant associations only with MYCN-amp and TERT status, but not with ALT (MYCN-amp: p = 1.17 × 10−5; TERT-re: p = 0.036; TERT-high: 3.16 × 10−5; ALT: p = 0.241; Cox proportional hazards regression) (Figure S15).Figure 3Copy-number dosage effects in canonical telomere maintenance(A) Comparison of TERT expression between tumors by molecular telomere maintenance characteristics. Horizontal bars indicate group mean. Two-sided Wilcoxon test p value is shown between groups.(B) Telomere length and TERT expression per sample.(C) Kaplan-Meier estimate of event free survival (EFS) by telomere maintenance mechanism in primary tumors.(D) Copy-number LogR and expression of TERT per sample. Linear regression line of samples with canonical telomere maintenance (MYCN-amp, TERT-re, TERT-high) in blue, regression line of other samples in gray. Gray ribbons indicate 95% confidence intervals. Pearson correlation coefficient r and p-value P are shown.(E) Local TERT copy-number gains in tumors NBL38 and CB2018 labeled in (D). Baseline (tumor LogR = 0) and TERT gene boundaries as magenta dashed line and box, respectively.(F) Copy-number gains induce higher TERT expression in tumors of activated TERT.Tumors with more than one molecular characteristic not shown in (A, B, C, and D). MYCN-amp, MYCN amplification; TERT-re, TERT rearrangement; TERT-high, high TERT expression; ALT: Alternative lengthening of telomeres. + +Copy-number dosage effects in canonical telomere maintenance + +(A) Comparison of TERT expression between tumors by molecular telomere maintenance characteristics. Horizontal bars indicate group mean. Two-sided Wilcoxon test p value is shown between groups. + +(B) Telomere length and TERT expression per sample. + +(C) Kaplan-Meier estimate of event free survival (EFS) by telomere maintenance mechanism in primary tumors. + +(D) Copy-number LogR and expression of TERT per sample. Linear regression line of samples with canonical telomere maintenance (MYCN-amp, TERT-re, TERT-high) in blue, regression line of other samples in gray. Gray ribbons indicate 95% confidence intervals. Pearson correlation coefficient r and p-value P are shown. + +(E) Local TERT copy-number gains in tumors NBL38 and CB2018 labeled in (D). Baseline (tumor LogR = 0) and TERT gene boundaries as magenta dashed line and box, respectively. + +(F) Copy-number gains induce higher TERT expression in tumors of activated TERT. + +Tumors with more than one molecular characteristic not shown in (A, B, C, and D). MYCN-amp, MYCN amplification; TERT-re, TERT rearrangement; TERT-high, high TERT expression; ALT: Alternative lengthening of telomeres. + +Copy-number dosage analysis of the TERT locus revealed that in CTM tumors (MYCN-amp, TERT-re, TERT-high) TERT dosage is significantly correlated with TERT expression (Figure 3D, Pearson’s r = 0.63, p = 8.5 × 10−6) as opposed to tumors with ALT or without telomere maintenance (r = 0.15, p = 0.22), independent of sample purity (p = 0.48, ANOVA F-statistic). TERT-re tumors harbored frequent TERT copy-number gains with breakpoints proximal to the TERT locus (Figure S16). Among three tumors with strongest TERT expression, two (NBL38, CB2018) harbored focal TERT gains (Figure 3E), indicating targeted upregulation of activated TERT by copy-number gains. None of the 3 TERT-high tumors showed TERT copy-number gains and thus these measures were not significantly correlated in this group either (Pearson’s r = −0.81, p = 0.19). Our findings show that SCNAs adjust the regulatory landscape of neuroblastoma toward dysregulation of key cancer pathways and that copy-number gains effectively upregulate TERT in tumors with CTM (Figure 3F), with highest telomerase expression found in tumors with both TERT activation and copy-number gains. + +To investigate if SCNAs are linked to increased telomere length in ALT tumors, we tested each chromosome arm for association between tumor DNA content and ALT using logistic regression, controlling for ATRX mutations (STAR methods). We found 11q losses (p = 4.83 × 10−7, ANOVA Chi-squared test) and 17q gains (p = 2.88 × 10−5, ANOVA Chi-squared test) to be significantly associated with ALT (Figure 4A), confirming previous observations of frequent 11q loss in ALT31 and revealing a yet undescribed association of 17q gain with ALT. We noticed that 11q loss co-occurs with strong 17q gains in 14 tumors and observed an overall negative correlation between DNA content of both chromosome arms across the cohort (r = −0.45, p = 2.01 × 10−7, Pearson’s correlation) (Figure 4B), suggesting a genomic rearrangement involving both chromosomes. Indeed, somatic SV analysis revealed 17q to 11q translocations in 19 tumors (Figure 4C), confirming that additional copies of chromosome arm 17q translocate to 11q in the aberrant tumor karyotype.49 Notably, 17q gains were identified in 105 of 115 tumors (91%) independent of ALT. However, ALT tumors were significantly enriched in the strongest 17q copy-number gains (Figure S17).Figure 4Copy-number alterations and upregulation of histone variant genes in alternative lengthening of telomeres(A) Association p-values of ALT and coverage LogR per chromosome arm. Significant observations in red (gain) and blue (loss), others in gray. Significance threshold (FWER 0.05) demarcated by gray dotted line.(B) 11q and 17q LogR per sample indicating ALT and ATRX status. Gray line represents linear regression fit, with light gray ribbon showing 95% confidence interval. Pearson correlation coefficient r and p-value P are shown.(C) Genome-wide somatic structural variant breakpoints. Breakpoints of frequent rearrangements between chromosome arms 11q and 17q highlighted by red arrows.(D) Correlation of gene expression and LogR of 17q and 11q. Differentially expressed genes in red (ALT up) and blue (ALT down).(E) STRING database protein interaction network of ATRX, H3F3B, H2AFJ and H3F3B and their first order interactions among differentially expressed genes in ALT. Red and blue indicate up- and downregulated genes respectively. ATRX is not differentially expressed (black). Histone variant genes and PRC2 complex member EED highlighted in yellow.(F) Proposed model of deregulated ATRX-interactions in ALT: 11q and 17q copy-number alterations upregulate histone variants as well as homologous recombination repair (HRR) pathway associated RAD51C and impairs PRC2 activity by EED downregulation. Reduced PRC2 activity results in H3K27me3 depletion. Mutant ATRX (ATRX mut.) impairs deposition of upregulated H3.3 and is associated with HRR-dependent elongation of telomeres in ALT. + +Copy-number alterations and upregulation of histone variant genes in alternative lengthening of telomeres + +(A) Association p-values of ALT and coverage LogR per chromosome arm. Significant observations in red (gain) and blue (loss), others in gray. Significance threshold (FWER 0.05) demarcated by gray dotted line. + +(B) 11q and 17q LogR per sample indicating ALT and ATRX status. Gray line represents linear regression fit, with light gray ribbon showing 95% confidence interval. Pearson correlation coefficient r and p-value P are shown. + +(C) Genome-wide somatic structural variant breakpoints. Breakpoints of frequent rearrangements between chromosome arms 11q and 17q highlighted by red arrows. + +(D) Correlation of gene expression and LogR of 17q and 11q. Differentially expressed genes in red (ALT up) and blue (ALT down). + +(E) STRING database protein interaction network of ATRX, H3F3B, H2AFJ and H3F3B and their first order interactions among differentially expressed genes in ALT. Red and blue indicate up- and downregulated genes respectively. ATRX is not differentially expressed (black). Histone variant genes and PRC2 complex member EED highlighted in yellow. + +(F) Proposed model of deregulated ATRX-interactions in ALT: 11q and 17q copy-number alterations upregulate histone variants as well as homologous recombination repair (HRR) pathway associated RAD51C and impairs PRC2 activity by EED downregulation. Reduced PRC2 activity results in H3K27me3 depletion. Mutant ATRX (ATRX mut.) impairs deposition of upregulated H3.3 and is associated with HRR-dependent elongation of telomeres in ALT. + +To pinpoint candidate genes contributing to ALT we tested for differential gene expression between ALT and non-ALT tumors, while controlling for MYCN amplification status, the presence of ATRX mutations and the sex of the patient (STAR methods, FDR 0.05). We found 408 up- and 224 downregulated genes (Figure S18; Table S4) and hypothesized that a subset of these genes might be driven by the ALT-associated SCNAs on 11q and 17q. Correlation between gene expression and DNA dosage of these chromosome arms revealed upregulated histone variant genes H3F3B (17q), H2AFJ (12p), and H3F3C (12p) among genes strongly affected by 17q and 11q dosage (Figure 4D). H3F3B (and its paralog H3F3A) encode the histone variant H3.3,50 which is altered by activating mutations in several pediatric tumor entities, including tumors of the central nervous system51,52 and up to 95% of chondroblastomas.53 Interestingly, activating H3.3 mutations triggered ALT in pediatric high-grade glioma regardless of ATRX mutation status,28 indicating that similarly, H3.3 upregulation may have functional implications in ALT neuroblastomas. H3F3C, which encodes for histone variant H3.5 is frequently mutated across different pediatric brain tumors, where alterations were found to be mutually exclusive to those in TP53 and associated with reduced genome stability.54 The H2AFJ gene encodes for histone variant H2A.J and is deregulated in melanoma,55 breast cancer,56 and colorectal cancer, where its upregulation is associated with poor survival.57 Taken together, these results suggest that copy-number alterations may deregulate histone variants contributing to epigenetic dysregulation and genome integrity in ALT neuroblastomas. + +The genetic effects model (STAR methods) predicted 41% and 60% of expression and ASE variance of H3F3B explained by local copy-number effects, indicating that expression of H3F3B is directly associated with 17q dosage (Figure S19). However, only 3% of H2AFJ and 2% of H3F3C expression variance is explained by local copy-number effects on 12p, suggesting upregulation through trans effects. + +To obtain a quantitative understanding how expression of the identified histone variant genes relates to ALT we predicted presence of ALT from the expression of H3F3B, H3F3C, and H2AFJ using logistic regression. We found expression of H3F3B and H2AFJ, but not H3F3C to be significantly associated with ALT in the presence of the two other genes (H3F3B: p = 0.001; H2AFJ: p = 0.008; H3F3C: p = 0.543; ANOVA), suggesting that expression of H3F3B and H2AFJ is independently associated with ALT. For an independent validation, we compared the expression levels of H3F3B and H2AFJ between 130 telomeric c-circle positive and negative neuroblastomas from Hartlieb et al.,31 and found significantly higher expression of H3F3B (p = 3.01 × 10−4, ANOVA) and H2AFJ (p = 0.02, ANOVA) in c-circle positive tumors, confirming their upregulation in ALT (Figure S20). + +Despite ATRX alterations being significantly associated with longer telomeres, we did not find ATRX to be differentially expressed between ALT and non-ALT (Table S4). We speculated that interaction partners of ATRX could be subject to deregulation in ALT tumors. To investigate this we obtained direct (first order) predicted protein interactions between ATRX, H3F3B, H2AFJ, H3F3C, and other proteins of differentially expressed genes in ALT affected by 11q or 17q dosage (STAR methods). The resulting network predicted high-confidence direct interactions between ATRX and differentially expressed histone 3.3 variant gene H3F3B, as well as RAD51C (Figure 4E). RAD51C is essential for homologous recombination repair,58 a process utilized in ALT for telomere extension. Notably, gene amplifications and pathogenic variants in RAD51C have been linked to breast and ovarian cancers.59,60 A network module containing H3F3B, H2AFJ, and H3F3C also included deregulated histone methylation factors EED and KMT2A. EED is part of the polycomb repressive complex 2 (PRC2), which modulates transcriptional repression by methylation of H3 histones,61,62 and we found EED to be downregulated in ALT tumors by 11q-dosage effects (Figures 4F and S21; Table S4). The PRC2 complex is frequently inactivated by EED loss in malignant peripheral nerve sheath tumors63 and adenosquamous lung tumors.64 Upregulation of H3.3 and H3.5 histones and concomitant downregulation of EED in ALT point toward a relative depletion of H3K27me3 as a consequence of higher H3 variant histone availability and impaired PRC2 activity (Figure 4F). In a related manner, PRC2 is inhibited by activating H3.3.pK27M mutations in pediatric gliomas65,66,67 or expression of PRC2 inhibitor EZHIP in ependymomas,68 both of which we did not find in our cohort (Figure S22). + +Recent studies have also demonstrated the binding of telomeric repeat-containing RNAs (TERRAs) to PRC2 and their role in H3K27 tri-methylation of telomeres.69 We thus investigated TERRA levels in our cohort (STAR methods) and found ALT tumors to show the highest TERRA expression with substantial variation between tumors (Figure S26). TERRAs were enriched in ALT compared to CTM (p = 0.0137, Wilcox rank-sum test) and MYCN-amp tumors (p = 0.0186), but not compared to tumors without detectable telomere maintenance (p = 0.0532, Wilcox rank-sum test) or TERT-re tumors (p = 0.0939, Wilcox rank-sum test) (Figures S26A and S26B). TERRA expression was significantly correlated with relative telomere length (Spearman’s rho: 0.3, p = 0.02) across the cohort (Figure S26C). + +To validate our findings on an independent dataset we investigated 11q loss and strong 17q gain in the neuroblastoma cohort presented by Gundem et al. 2023,70 using ATRX mutations as a surrogate for ALT. Relative losses of 11q (p = 0.0029, one-sided Wilcox rank-sum test) and relative gains of 17q were significantly associated with ATRX-mutated samples compared to other tumors (p = 0.0124, one-sided Wilcox rank-sum test) (Figure S27), confirming our initial associations of 11q loss and strong 17q gains in ALT in the discovery cohort. We further analyzed dosage effects of key deregulated genes H3F3B (on 17q) and EED (on 11q) in the cohort from Egolf et al. 201971 correlating copy-number with gene expression. We found significant associations of copy-number and expression for H3F3B (p = 5.6 × 10–11, Pearson correlation test) and EED (p = 3.8 × 10−7, Pearson correlation test) (Figure S28), confirming our initial findings of these copy-number dosage effects in the discovery cohort. + +Our findings implicate 11q loss and strong 17q gain in ALT neuroblastomas and show that these alterations deregulate ATRX interaction partners. They highlight histone variants as key components of ALT-deregulated ATRX protein interactions and indicate that activity of the PRC2 complex could be reduced due to attenuated EED expression resulting from 11q loss, providing additional evidence for histone-dependent chromatin deregulation by copy-number dosage in ALT neuroblastomas. + +Finally, we characterized genes by ASE frequency and average ASE ratio across tumors. The highest ASE ratio (0.96) and frequency (0.98) was found for the PEG10 gene, which is maternally imprinted in most tissues.72 Generally, imprinted genes73 including IGF2, DLK1, RTL1, and L3MBTL1 (Figure 5A) were enriched among the genes with strongest (p = 3.6 × 10−6) and most frequent ASE (p = 0.0059, both Wilcoxon rank-sum test) (Figure 5B), showing that expression imbalance recapitulates imprinting in neuroblastoma.Figure 5Bi-allelic expression of imprinted gene RTL1 in unfavorable tumors(A) AEI frequency and mean ASE ratio per gene. Imprinted genes in blue, others in gray.(B) AEI frequency (top) and ASE ratio (bottom) by imprinting status per gene: imprinted (n = 43), other (n = 13,007).(C) ASE-expression effect and differential expression (survival) per gene. Allelic regulated genes in color scale, others in light gray.(D) Gene expression (top) and ASE ratio (bottom) of the RTL1 gene by survival: Deceased/Relapse (n = 46), Other (n = 69).(E) Percent methylated CpGs in the genomic region −4 kb to +1 kb relative to RTL1 gene start. Samples are clustered by above (high) (n = 12) and below (low) (n = 11) median methylation level in the genomic window shown.(F) RTL1 expression by methylation clusters shown in (E).(G) Proposed model of RTL1 upregulation through loss of maternal imprinting. Incomplete imprinting on maternal allele results in bi-allelic expression and upregulation of RTL1. ICR: Imprinting control region.Boxplot midlines in (B, D, and F) mark median; upper and lower hinges extend to first and third quartile; upper and lower whiskers extend to the smallest and largest value max. 1.5 × IQR; p value of two-sided Wilcoxon test is shown between groups. + +Bi-allelic expression of imprinted gene RTL1 in unfavorable tumors + +(A) AEI frequency and mean ASE ratio per gene. Imprinted genes in blue, others in gray. + +(B) AEI frequency (top) and ASE ratio (bottom) by imprinting status per gene: imprinted (n = 43), other (n = 13,007). + +(C) ASE-expression effect and differential expression (survival) per gene. Allelic regulated genes in color scale, others in light gray. + +(D) Gene expression (top) and ASE ratio (bottom) of the RTL1 gene by survival: Deceased/Relapse (n = 46), Other (n = 69). + +(E) Percent methylated CpGs in the genomic region −4 kb to +1 kb relative to RTL1 gene start. Samples are clustered by above (high) (n = 12) and below (low) (n = 11) median methylation level in the genomic window shown. + +(F) RTL1 expression by methylation clusters shown in (E). + +(G) Proposed model of RTL1 upregulation through loss of maternal imprinting. Incomplete imprinting on maternal allele results in bi-allelic expression and upregulation of RTL1. ICR: Imprinting control region. + +Boxplot midlines in (B, D, and F) mark median; upper and lower hinges extend to first and third quartile; upper and lower whiskers extend to the smallest and largest value max. 1.5 × IQR; p value of two-sided Wilcoxon test is shown between groups. + +Since ASE can be caused by either up- or downregulation of gene expression on one parental haplotype, we systematically explored effect directionality by testing for association between ASE and total expression. 10,862 genes that were informative for ASE in at least 20 samples were considered, out of which 455 showed a significant (FDR <0.05) effect of ASE on total gene expression (STAR methods; Table S5). To narrow the search, we intersected these 455 genes with those differentially expressed between deceased and other patients, resulting in a final set of 107 candidate genes (Table S5). Among these, genes contained on the MYCN amplicon MYCN, NBAS, and DDX1 showed a positive ASE-expression effect due to strong upregulation by mono-allelic amplifications. In contrast, chromosome arm 1p (56%) and 17p (12%) were most frequent among all 76 genes with negative ASE-expression effect, indicating that loss of 1p and 17p underlies downregulation of these genes in tumors of deceased patients. + +Interestingly, a substantial negative ASE-expression association was found in the Retrotransposon Gag-like 1 (RTL1) gene, which was upregulated in tumors of deceased patients (Figures 5C and 5D). RTL1 is a maternally imprinted gene involved in placental/neonatal development74 and widely expressed in the nervous system.75 Upregulation of RTL1 confers selective growth advantage in hepatocarcinoma76 and promotes cell proliferation by regulating Wnt/β-Catenin signaling in melanoma.77 +RTL1 was one of 16 genes informative for survival time in a previous study of high-risk neuroblastomas, with stronger RTL1 expression associated with shorter survival.78 Our linear model revealed only a minor contribution of SCNAs and germline variants to ASE in RTL1 (Figure S23), suggesting that differences in allelic expression levels may result from methylation differences. Analyzing a subset of tumors using bisulfite sequencing (BS-seq) (STAR methods), we found that decreased methylation levels at CpGs upstream of RTL1 are associated with higher RTL1 expression (Figures 5E, 5F, and S24). Taken together these findings suggest that upregulation of RTL1 in neuroblastoma is induced by bi-allelic activation in tumors with unfavorable prognosis, likely due to loss of imprinting on the maternal allele (Figure 5G). + +We here systematically characterized the effects of copy-number dosage on neuroblastoma gene expression and demonstrated how copy-number gains interact with upregulated TERT to increase the efficacy of CTM. We found 11q loss and strong 17q gain as markers of ALT in addition to ATRX alterations, and revealed upregulation of histone variant genes H3F3B, H3F3C, and H2AFJ. + +Histone variants replace replication-dependent canonical histones in nucleosomes during the cell-cycle, affecting chromatin organization at telomeric,79 actively transcribed regions by replication-independent chromatin incorporation80,81,82 and interaction with chaperones and chromatin factors.83 +H3F3B resides on 17q, and our findings strongly suggest that H3F3B is directly upregulated by 17q gains, which have already been reported to exert oncogenic effects through increased gene dosage.43 In contrast, H3F3C and H2AFJ expression are associated with 11q loss and 17q gain, but neither of these reside on these chromosome arms, suggesting possible involvement of regulatory effects in trans. + +Possibly, this way copy-number alteration mediate histone replacement and chromatin re-organization in ALT, leading to decondensation and increased transcription.80,81 Dosage-dependent downregulation of the repressive PRC2/EED-EZH2 complex, which methylates the lysine 27 residue of H3 histones may contribute to this reprogramming, and we found EED, which is predicted to interact with all three histone variants, as differentially downregulated in ALT tumors by 11q loss. Similarly, PRC2 activity in pediatric high-grade glioma is impaired by H3.3K27M mutations altering EZH2 binding67 and resulting in depletion of H3K27 di- and tri-methylation.66 + +It is important to note that 17q gain is the single most prevalent cytogenetic feature across neuroblastoma, found in 91% (105/115) tumors of our cohort. However, we found the strongest 17q gains in the ALT group, even after controlling for covariates including tumor purity. 40–50% of variance in H3F3B gene expression was explained by copy-number differences and expression was again significantly higher in ALT tumors. + +ATRX stabilizes telomeres through depositing of H3.3 histones, thereby preventing replication-induced breaks conducive to ALT.79,84 In contrast, ATRX is not required to deposit H3.3 histones in actively transcribed regions.79 Consequently, H3.3 upregulation through H3F3B dosage in ALT tumors with defective ATRX may increase the prevalence of H3.3 in nucleosomes of active chromatin without its stabilizing effect at telomeres. Importantly, we found 11q loss and 17q gain to be associated with ALT independent of ATRX mutations. Because not all ALT tumors harbor ATRX alterations, deregulated histone variants may contribute to the ALT phenotype more directly. In high-grade gliomas ALT frequently occurs in H3.3G34R-mutant tumors independent of ATRX alterations,28 indicating a functional link between impaired H3.3 function and ALT. + +Additionally, loss of ATRX alone may not be sufficient to induce ALT,84 and ATRX mutations are likely not the only molecular feature responsible for this phenotype. However, in ATRX-wild type ALT-positive neuroblastomas, ATRX protein levels were found to be significantly decreased,31 suggesting that impaired ATRX activity could still underlie ALT in these cases. Furthermore, not all ALT-positive tumors showed 11q loss and strong 17q gain and these alterations were also present in a few ALT-negative tumors. Generally, ATRX can stabilize replication forks and is involved in the resolution of G-quadruplex DNA structures.85 Impaired ATRX activity could still cf. phenotypes independent of histone variant deregulation. Additional research with larger cohorts will be needed to further characterize the relationship between histone variants and ALT in neuroblastoma. + +In addition to upregulated histone variants we also identified 17q copy-number dependent upregulation of predicted ATRX interaction partner RAD51C, a recombinase of the HRR pathway. RAD51 is involved in C-circle-associated RAD52-independent ALT-mediated telomere synthesis.86 However, a previous study found that depletion of RAD51 and its homologs including RAD51C had no marked effect on telomere length, fragility, or APB formation in mouse embryonic fibroblasts lacking telomerase.87 This either suggests that RAD51C may not be essential for ALT or that, unlike the model, ALT-mediated telomere synthesis in neuroblastoma favors the RAD51-associated and RAD52-independent pathway, further supported by high prevalence of C-circles in ALT neuroblastomas,31 as well as the prognostic value of and chemotherapy resistance conferred by RAD51 expression in this disease.88 + +In tumors with a 17p LOH event, loss of function of TP53 due to a second hit could be responsible for the poor prognosis of these tumors, but no such second hit was found in our cohort and we did not observe a copy-number dosage effect on TP53 expression. Survival-associated 17p copy-number dosage effects were enriched for neuronal genes, suggesting a potential impairment of neuronal processes. However, the exact mechanism that underlies higher mortality of donors with 17p imbalance still needs to be investigated. + +Lastly, we identified RTL1 as a candidate marker for tumors with unfavorable prognosis due to loss of imprinting of the maternal allele, similar to earlier reports on loss of imprinting of the IGF2 gene in Wilms’ tumors.89 The transposon-derived RTL1 gene90 is part of a broader imprinted DLK1-DIO3 gene cluster with three paternally expressed genes DLK1, RTL1, and DIO3. DLK1 expression in neuroblastoma cell lines is associated with neuroendocrine lineage differentiation.91 Possibly, imprinting heterogeneity at the DLK1-DIO3 gene cluster reflects differentiation states in neuroblastoma progenitor cells, and incomplete imprinting may characterize the cell-of-origin of more aggressive or treatment-resistant tumors. + +Taken together, our analyses shed light on the complex interaction of genetic, epigenetic, and transcriptomic effects and how gene dosage interacts with other genetic and epigenetic factors to shape the regulatory landscape of neuroblastoma. In the future, the increase in size of such multi-omics datasets will enable a more complete understanding of the development of disease-relevant phenotypes and potentially convergent pathways. + +Neuroblastoma is a rare childhood cancer with an incidence as low as 11 cases per million children.92 As such, sample sizes of genomics cohorts, in particular with WGS data, remain limited. In addition, the systematic laboratory evaluation of telomere maintenance mechanisms are rarely routinely performed and often difficult to establish post-hoc. We here employ a combination of k-mer counts, TERT rearrangement and expression, and MYCN amplification status to determine telomere maintenance mechanisms. We validated this approach against the experimentally determined status of APB21 and C-circles31 in subsets of analyzed tumors with great success. Further experimental validation is crucial for a deeper understanding of the role of histone variants in maintaining ALT, particularly the histone H3.5 and H2A.J variants, whose connection to ALT remains unknown. However, this investigation was beyond the scope of the current study. + +Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Roland F Schwarz (roland.schwarz@iccb-cologne.org). + +This study did not generate new unique reagents. + + +•The raw data analyzed in this study is available from the European Genome-phenome Archive (https://www.ebi.ac.uk/ega/) under accession number EGAS00001001308,20 EGAS00001004022,8 EGAS0000100560493 and EGAS00001006983. These datasets are available under restricted access due to data privacy laws. Access to the EGA archive datasets is obtained by formal application to the corresponding Data Access Committee (DAC).•Processed data and analysis code is available in Zenodo under DOI https://doi.org/10.5281/zenodo.8373208.•Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. + + +The raw data analyzed in this study is available from the European Genome-phenome Archive (https://www.ebi.ac.uk/ega/) under accession number EGAS00001001308,20 EGAS00001004022,8 EGAS0000100560493 and EGAS00001006983. These datasets are available under restricted access due to data privacy laws. Access to the EGA archive datasets is obtained by formal application to the corresponding Data Access Committee (DAC). + +Processed data and analysis code is available in Zenodo under DOI https://doi.org/10.5281/zenodo.8373208. + +Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. + +This study was supported by funding from the Berlin Institute of Health project “TERMINATE-NB” (CRG04) to A.E. and U.O., as well as DFG research training group “GRK 1772: Computational Systems Biology” (project number 191312833) to M.B. and U.O. We thank the patients and their parents for granting access to the tumor specimens and clinical information that were analyzed in this study. R.F.S. is a Professor at the Cancer Research Center Cologne Essen (CCCE) funded by the Ministry of Culture and Science of the State of North Rhine-Westphalia. This work was partially funded by the German Ministry for Education and Research as BIFOLD—Berlin Institute for the Foundations of Learning and Data (ref. 01IS18025A and ref. 01IS18037A). The authors would like to thank the Helmholtz Association, Germany, for support and Martin Peifer for helpful comments and suggestions. Computation was performed on the HPC for Research cluster of the Berlin Institute of Health. + +M.B., A.E., J.S., S.M.W., D.B., R.F.S., and U.O. contributed to the study design and the collection and interpretation of the data. E.B. and N.T. performed quality control of sequencing data and aligned WGS and RNA-seq reads. M.B. analyzed WGS and RNA-seq data. M.B. performed allele-specific expression and allele-specific copy-number analysis. E.B., N.T., and M.B. analyzed somatic SNVs. J.T. and M.B. analyzed somatic SVs. R.M. and M.B. performed QTL mapping. M.B. conducted telomere length analysis. M.B. and C.W. conducted copy-number-survival analysis. M.B., R.F.S., and U.O. wrote the manuscript. J.S., R.F.S., and U.O. led the study design. + +The authors declare no competing interests. + + +REAGENT or RESOURCESOURCEIDENTIFIERDeposited dataWhole-genome sequencing data of primary neuroblastomaEuropean Genome-phenome Archive (EGA)EGAS00001001308Whole-genome- and RNA-sequencing data of primary neuroblastomaEuropean Genome-phenome Archive (EGA)EGAD00001005488Variant call files1000 Genomes Project (phase 3)93https://www.internationalgenome.org/category/phase-3/Mappability trackUCSC Genome Browser94wgEncodeCrgMapabilityAlign50merhttps://genome.ucsc.edu/Processed data and analysis code–https://doi.org/10.5281/zenodo.8373208Software and algorithmsTelseq v0.0.2Ding et al.38https://github.com/zd1/telseqBcftools v1.8–https://www.htslib.org/ASCAT v2.6Loo et al.95https://github.com/VanLoo-lab/ascat/HTseq v0.9.1Putri et al.96https://github.com/htseq/htseqDESeq2 v1.26.0Love et al.97https://github.com/htseq/htseqEnsembl variant effect predictor (VEP) v101.0McLaren et al.98https://www.ensembl.org/info/docs/tools/vep/index.htmlBWA-MEM v0.7.15Li and Durbin99https://github.com/lh3/bwaSTAR v2.5.3aDobin et al.100https://github.com/alexdobin/STAR/releasesGATK/Mutect2 v2.2McKenna et al.101https://gatk.broadinstitute.org/Novobreak v1.1.3Chong et al.102https://sourceforge.net/projects/novobreak/MEDICC2Kaufmann et al.40https://pypi.org/project/medicc2/0.5b3/fgsea v1.12.0Korotkevich et al.103https://bioconductor.org/GATK/ASEReadCounter v3.5.0McKenna et al.101https://gatk.broadinstitute.org/PEERStegle et al.104https://github.com/PMBio/peer/FastLMM v0.2.23Lippert et al.105https://github.com/fastlmm/FaST-LMMSTRING DB network viewer v11.0bSzklarczyk et al.106https://string-db.org/bcl2fastq v.2.19.0.316–https://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.htmlPiGx BS-seq pipelineWurmus et al.107https://github.com/BIMSBbioinfo/pigx_bsseqFastQC v0.11.9–https://github.com/s-andrews/FastQCTrimGalore v.0.6.2–https://github.com/FelixKrueger/TrimGalorebwa-meth v.0.7.17Pedersen et al.108https://github/com/brentp/bwa-meth/samblaster v.0.1.24Faust et al.109https://github.com/GregoryFaust/samblastermethylKit v1.15.4Akalin et al.110https://github.com/al2na/methylKit + + +This study is based on the analysis of sequencing data from tumor and blood samples of patients diagnosed with neuroblastoma between 1991 and 2016. Patients were registered and treated according to trial protocols of the German Society of Pediatric Oncology and Hematology (GPOH). Samples were collected at diagnosis from untreated patients. The study was conducted in accordance with the World Medical Association Declaration of Helsinki (2013) and good clinical practice. Informed consent was obtained from all patients or their guardians. Collection and use of patient specimens was approved by the institutional review boards of Charité Universitätsmedizin Berlin and of the Medical Faculty, University of Cologne. Collected specimens and clinical annotations were archived and made available by Charité-Universitätsmedizin Berlin or the National Neuroblastoma Biobank or or the National Neuroblastoma Biobank and Neuroblastoma Trial Registry (University Children’s Hospital Cologne) of the GPOH. + +Sequencing data were collected from tumor and blood samples of patients diagnosed with neuroblastoma enrolled and treated according to trial protocols of the German Society of Pediatric Oncology and Hematology (GPOH) in the multi-center study GPOH-NB2004.37 Analyzed DNA and RNA samples were obtained from primary tumors of at least 60% tumor cell content as evaluated by a pathologist. MYCN copy-number was determined by FISH. DNA was extracted from fresh-frozen tumor tissue and the corresponding matched normal using the Puregene Core Kit A (Qiagen) and NucleoSpin Blood DNA extraction kit (Macherey-Nagel), according to the manufacturers’ instructions. Libraries were prepared with the TruSeq DNA PCR-free sample preparation kit (Illumina). WGS of tumor-normal pairs was performed on the HiSeq X-Ten platform (Illumina, San Diego, USA), yielding paired-end reads of 2 × 150 bp length. RNA was isolated with Trizol according to the manufacturer’s protocol (Thermo Fisher). Purity was analyzed on a Nanodrop 2000 spectrometer and RNA integrity assessed on a Bioanalyzer 2100 or TapeStation4200 as per manufacturer’s instructions. Only samples with an RNA integrity number of 8 or above were included. Depletion of ribosomal RNA (rRNA) was performed by enzymatic digestion. Ribo-depletion of 95–99% was confirmed using RT-qPCR. The ribo-depleted RNA was used for generation of RNA sequencing libraries using the TrueSeq Stranded mRNA kit according to the manufacturer’s protocol (Illumina). Ribo-depleted RNA was sequenced on the HiSeq4000 platform (Illumina, San Diego, USA) yielding reads of 2 × 150 bp length. Additional sequencing data were obtained from the European Genome-phenome Archive under accession number EGAS00001001308 for a non-overlapping set of donors from a previous study on somatic structural rearrangements in neuroblastoma.20 After quality control 52 donors of this study were included, yielding a total of 115 donors with matched tumor RNA-seq, tumor WGS, and blood-derived normal WGS. + +Whole-genome bisulfite sequencing data was obtained for a subset of 23 tumors (Table S1). Libraries were prepared using the EpiTect Bisulfite and Illumina Truseq PCRFree DNA sequencing kit (V2.5) and sequenced on the Illumina Hiseq X platform yielding paired-end reads of 2 × 150 bp. + +Reads were aligned to the GRCh37. WGS reads were aligned with BWA-MEM 0.7.15.94 RNA-seq reads were aligned with STAR 2.5.3a.95 Samblaster 0.1.2496 was used to mark duplicates in alignment files. Quality control was performed using FastQC. Table S1 lists donors from which matched tumor and normal WGS as well as matched tumor RNA-seq was used in the analyses. + +Aligned tumor RNA-seq reads were counted using HTseq/htseq-count 0.9.1 on exons of protein coding genes according to Ensembl release 75 human gene annotations for the GRCh37 reference, summarizing counts on gene-level. We normalized gene expression for the purpose of between-sample comparisons in a given gene. To mitigate the effect of sequencing depths and batch effect introduced by different RNA library preparation- and sequencing methods between the two cohorts we normalized htseq counts by the following strategy: We first calculated library-size normalized DESeq2 variance stabilized counts from htseq counts. Then, we modeled the variance stabilized counts by cohort membership using a linear model for each gene and determined the residual for each gene and sample. If not indicated otherwise, this residual was used as the measure for gene expression in our analyses. + +Somatic SNVs were called by Mutect2 version 2.2 from the GATK software package.97 SNV calls were filtered using a panel of normals. Effects of SNVs were predicted using the Ensembl variant variation effect predictor version 10198 in offline mode with distance 100,000 bp. SNVs in categories missense, splice, stop, synonymous, 5′ UTR and 3′ UTR were summarized to gene level somatic mutation burden. Somatic SNVs annotated as promoter variants by the Ensembl variant effect predictor were considered separately. Splice, nonsense and missense variants for each gene were summarized based on the assigned consequence. + +SV were called using novobreak version 1.1.399 in pairs of matched tumor and normal WGS alignments. We only kept SV calls with QUAL ≥30, at least 5 high quality reads in support of each breakpoint in the tumor sample, 0 reads supporting each breakpoint in the normal sample, 5 or more discordant reads per breakpoint in the tumor sample and 3 or less discordant reads per breakpoint in the normal sample. The functional effects of SVs at the TERT locus have been established previously20,33 and for the detection of TERT SVs we relaxed the threshold on high quality reads in support of each breakpoint, requiring at least 2 of those reads to keep the SV call. Other thresholds were applied as described above. TERT rearrangement status was assigned to a sample positive for at least one somatic SV 100,000 kb upstream or downstream of TERT gene start and end coordinates (Ensembl/GRCh37) or annotated as TERT rearranged in.20 Additionally, the TERT locus was manually examined in tumor and normal WGS alignments for SVs that were missed by the variant calling procedure above, resulting in one additional assignment of positive TERT rearrangement status in tumor CB2064 (Figure S25). + +We used a targeted approach to identify ATRX exon deletions. To this end we determined read coverage at ATRX gene coordinates in 50 bp bins, normalized the read counts by the number of overall mapped reads and defined a tumor coverage ratio by si = log2(niT/niN), where niT and niN are normalized read counts in tumor and normal for bin i respectively. For each matched tumor/normal pair we then fit a two-component Gaussian mixture model to the signal and determine the mean and relative proportions of two hypothetical clusters, corresponding to read coverages of deleted and intact regions of the gene. Samples that harbored a signal mean difference of at least 1.5 units between the two clusters and in which the smaller cluster showed a proportion of 10% or more of the larger cluster were regarded as ATRX deleted. Tumors that showed either ATRX deletions as determined by this method, were positive for a somatic SV breakpoint inside ATRX gene boundaries or carried a somatic missense, nonsense or splice SNV were considered as mutant ATRX. + +Telomere lengths were estimated from WGS of normal and tumor samples by Telseq 0.0.238 with parameter -u (ignore read groups) and otherwise default settings. Briefly, the method estimates telomere length by counting WGS reads containing the telomere repeat sequence (TTAGGG)k, where k denotes the number of repeats of the 6-mer. Telseq uses default repeat length k = 7 and normalizes the resulting read count by GC content and a genome size factor. The authors calibrated the default parameters using telomere length measurements determined by Southern blot analysis of terminal restriction fragments. We summarized telomere lengths per sample by the log telomere length ratio log(TLR) = log(LT/LN), where LT and LN are the Telseq estimates for telomere length in tumor and normal WGS sample respectively. + +Tumors were clustered based on unsupervised Gaussian mixture modeling of TERT gene expression (n = 2 mixture components). A threshold (Z score > −0.1028) for high TERT expression was defined as the TERT expression Z score at which the probability of assignment of a tumor to the component of stronger TERT expression exceeded 95%, similarly as described in.21 We assigned the TERT-high attribute to tumors that exceeded the TERT expression threshold and for which neither MYCN amplification nor TERT rearrangements were detected. + +The telomere maintenance status (CTM, ALT, Mix, or None) was assigned based on the status of MYCN amplification, TERT rearrangement, TERT-high attribute and telomere length ratio as follows: Status canonical telomere maintenance (CTM) was assigned to tumors with either MYCN amplification, TERT rearrangement or that were classified as TERT-high. Status alternative lengthening of telomeres (ALT), was assigned to tumors with telomere length ratio log(LT/LN) > 0.5. Status Mix was assigned to all tumors that met criteria for both CTM and ALT. Status None was assigned to all other tumors, indicating a general lack of evidence of any telomere maintenance mechanism. + +Variant call files with 84,801,880 germline variants from the 1000 Genomes Project (phase 3)100 were downloaded and filtered for biallelic SNPs. SNPs from chromosomes 1–22 were filtered for ≥1% minor allele frequency (MAF) in the 1000 Genomes cohort. Only SNPs with a mappability score of 1 (unique 50mer, UCSC wgEncodeCrgMapabilityAlign50mer)101 were kept, resulting in 9,866,569 variant sites (SNP panel) for downstream analysis. We generated pileups at positions of the SNP panel from WGS alignments of blood-derived control samples by Bcftools 1.8 mpileup, excluding unmapped reads, or reads that were marked as optical duplicates or “not primary alignment”. The resulting pileups were used as input to the Bcftools 1.8 multiallelic-caller to call genotypes at the positions of the SNP panel. We only kept resulting genotypes with an allelic depth of 10 or more reads and a genotype quality of 20 or higher. The resulting individual variant files were merged and genotypes were phased by Eagle 2.4102 using the phased 1000 Genomes genotypes as reference. The resulting variant file, comprising phased genotypes of all individuals was defined as the genotype panel for further downstream analysis. + +Pileups of primary tumor WGS were generated by Bcftools 1.8 mpileup at SNP positions of the genotype panel established (STAR methods). Unmapped reads, or reads that were marked as optical duplicates or as “not primary alignment” were not considered in the pileup. For each of the SNPs the allelic depths were calculated from the pileups on normal and tumor alignments respectively. For SNPs with total depth of 10 or more reads in both tumor and normal alignments we determined the B-allele frequency (BAF) and the coverage log ratio (LogR). For a given pileup position the BAF is defined as the ratio between alternative allele nucleotide count and the number of total considered counts ai/(ri+ai), where ai and ri are the allelic depths of alternative and reference allele respectively. The LogR at SNP position i was defined as log2((dti/dni)/(∂t/∂n)), where dti is the total depth at SNP position i in the tumor sample, dni is the total depth at SNP position i in normal sample and ∂t and ∂n are mean depths at SNPs of tumor and normal sample respectively. + +The BAF of a heterozygous SNP position is informative for the proportion of aligned reads originating from the paternal and maternal allele. At a homozygous SNP the BAF is expected to be close or equal to 1, if the sample’s SNP genotype is homozygous alternative or close or equal to 0 if the genotype is homozygous reference. The BAF is calculated separately for alignments of normal and tumor, resulting in a normal BAF and a tumor BAF per SNP and sample. The LogR is a measure of total coverage difference between normal and tumor samples and is informative at any position, including homozygous and heterozygous SNPs. It is calculated for a pair of alignments (tumor and normal), resulting in a LogR value per SNP and sample. + +Allele-specific copy-number profiles were generated from tumor and normal BAFs and LogR values for each sample using ASCAT 2.6103 with a custom segmentation procedure. In ASCAT’s segmentation step the BAF and LogR values are converted into intervals of similar values. ASCAT’s original implementation of this segmentation considers both LogR and BAFs to obtain start and endpoints for segments. We found noisy coverage log ratios to introduce over-segmentation in some samples and therefore replaced the segmentation procedure with a custom implementation that only considers BAFs to determine start and endpoints of segments, but still estimates the segment’s coverage using the log coverage ratios. ASCAT’s output comprises copy-number segments with integer copy-numbers of major and minor alleles as well as estimates for tumor purity and ploidy. All copy-number segments were inspected manually for quality. For samples with estimated tumor purity less than 60% copy-number calling was rerun with adjusted purity and ploidy values that were manually selected after inspection of the goodness-of-fit plots and in agreement with pathology estimates of tumor purity. + +We associated the copy-number per chromosome arm with telomere length. For this purpose we derived a copy-number LogR per chromosome arm and tumor by an overlap length-weighted mean considering all copy-number segments of a given sample overlapping a chromosome arm. Samples were divided into two groups (ALT, non-ALT), excluding a single tumor (NBL54) with signs of both ALT and canonical telomere maintenance by TERT rearrangement (group Mix). We then used this binary (ALT, non-ALT) outcome as the response of a generalized linear model controlling for covariates MYCN amplification, ATRX alteration, age, sex, cohort, tumor purity, and tumor ploidy. The association p-value was determined by an analysis of variance (ANOVA) using a Chi-Squared test between two generalized linear models, of which the first modeled ALT by the covariates above and the second additionally included a term for the copy-number LogR. P-values were determined for each chromosome arm and corrected by the Bonferroni method. Chromosome arms below 0.05 FWER were considered significant. + +The WGD (whole-genome-doubling) status of the samples were estimated using the phylogenetic reconstruction tool MEDICC2.40 For single samples MEDICC2 calculates the minimum number of evolutionary events from a diploid genotype to the copy-number states of the sample by combining loss-of-heterogeneity events, whole-genome-doubling as well as chromosome-wide and focal losses and gains. By checking whether the shortest evolutionary path from the diploid to a sample copy-number profile contains a WGD we can estimate the WGD status. As a WGD event followed by multiple losses can also be modeled by multiple gains, MEDICC2 can be conservative in its WGD estimation. We employed a bootstrap method to improve the estimation of WGD events. For this we used 100 bootstrap copy-number profiles per sample that were created by randomly drawing 22 chromosomes with replacement from the original chromosomes. If 5% or more of the bootstrap runs exhibited a WGD the sample was marked as WGD positive. + +Copy-number states (CN state) were assigned to each segment based on ASCAT’s allele counts, ploidy estimates and LogR of coverage between tumor and normal WGS alignment. Copy-number (CN) states were assigned based on conditions in the following order: weak gain: CNtotal > round(ploidy); medium gain: CNtotal > 1.5 × round(ploidy); strong gain: CNtotal > 2.5 × round(ploidy); shallow loss: CNtotal < round(ploidy); loss: CNtotal < 0.5 × round(ploidy). Here, CNtotal = CNmajor + CNminor, where CNmajor and CNminor are the allele counts of major and minor allele respectively and round(ploidy) the ploidy estimate determined by ASCAT rounded to an integer value. Copy-number state amplification was assigned to focal alterations of segments smaller than 10 Mb with CNmajor ≥ 5 and LogRseg - LogRcontig > 0.7, where LogRseg is the mean LogR of the segment and LogRcontig the mean LogR of the segment’s chromosome (contig). Similarly we assigned a copy-number balance state (CN balance state) to each segment. For this purpose the copy-number ratio was determined as CNratio = CNmajor/(CNmajor + CNminor). Then, the CN balance state balance was assigned if CNminor > 0 and CNmajor = CNminor. CN balance state weak imbalance was defined as CNmajor > CNminor and CNratio ≤ ⅔, and state strong imbalance was defined as CNmajor > CNminor and CNratio > ⅔. CN balance state amplification was defined in the same way as for the copy-number state above. Copy-number segments were marked as LOH if CNminor = 0. We derived copy-number states for each chromosome arm and separately for cytoband 1p36, (which is frequently deleted in MYCN-amplified tumors) by summarizing the overlap of all segments per copy-number state in these broader regions and assigning the copy-number state of largest overlap. We applied the same procedure to assign copy-number states to genes by overlap between gene coordinates (Ensembl version 75). Gene amplification status was inferred from its copy-number state and gene-specific LogR measurements. Genes of copy-number state amplification or those with LogRgene > 2.5 were defined as amplified, where LogRgene is the mean LogR across all SNPs falling within the gene’s coordinates. + +Allele-specific RNA read counts were determined by GATK97 (version 3.5.0) ASEReadCounter from RNA-seq alignments at heterozygous SNPs.104 SNPs with less than 8 total or less than 2 allelic reads were removed. Additionally, only sites that qualified as bi-allelic according to a statistical test were retained: A binomial test on the minimum allele count = min(alt, ref), number of trials (alt + ref) and hypothesized probability of success sum(non_ref_alt)/sum(raw_depth) was applied, where ref and alt are the reference and alternative allele counts, and non_ref_alt and raw_depth the non-reference/non-alternative allele count and raw read depth per site respectively. Sites for which the null hypothesis was rejected (FDR 0.05, Benjamini-Hochberg) were classified as bi-allelic. The reference allele bias was estimated by averaging over the reference allele fraction ref/(ref + alt) of all ASE sites from balanced copy-number regions per sample. We used statistical phasing information (see genotyping and phasing methods) to summarize allelic counts at exonic heterozygous SNPs of the same haplotype per gene. Only genes with a total of 10 or more counts from both haplotypes were retained. The ASE ratio for a given gene was calculated as max(A, B)/(A + B), where A and B are haplotype counts of the arbitrary A and B allele respectively. Expression imbalances per gene and sample were assessed by a two-sided binomial test using A as the number of successes, (A + B) as the number of trials and 0.5 as the hypothesized probability of success. The p-value was adjusted for multiple testing using the Benjamini-Hochberg procedure. Allelic-expression imbalance (AEI) status was assigned to observations (gene-sample pairs) for which an expression imbalance was detected at FDR 0.05. + +For eQTL analysis the SNP genotypes called in 115 WGS samples of normal tissue were pooled and filtered. Only SNPs with a minor allele frequency of 5% and at least 10% genotyped samples in the cohort were retained. Htseq count105 was used to count reads from RNA-seq data of tumor samples in the union of all exons per gene based on the Ensembl 75 gene annotation. Raw RNA gene counts were normalized by library depth per sample and transformed to variance-stabilized counts by DESeq2.106 Only protein-coding genes on chromosomes 1–22 with at least 10 counts in 90% of the samples were considered. In total 13,903 genes were included in the analysis. Variance-stabilized counts were centered and strong outlier samples, defined as normalized count values exceeding 3 times the standard deviation of all normalized counts per gene, were removed. To estimate the expression variability between samples we applied probabilistic estimation of expression residuals (PEER)107 to derive 10 factors from the normalized counts. We took these factors as representatives for global expression differences that are likely not associated with cis-regulatory effects and incorporated them as covariates in the association test described below. Genotypes of SNPs in a cis-window of 500 kb upstream and downstream of annotated gene coordinates were associated with the gene’s quantitative trait. SNPs were associated with quantitative traits by FastLMM,108 version 0.2.23 in single SNP mode. FastLMM uses a linear mixed model in a regression of the number of alternative alleles on quantitative phenotypes controlling for given covariates. We combined gene- and sample-specific covariates individually in each test. The somatic gene copy-number was calculated as the average total copy-number in gene intervals and used as the only gene-specific covariate. Sex, cohort, tumor purity, tumor ploidy and the 10 PEER factors were incorporated as sample-specific covariates. Each association test was controlled by the matching set of sample and sample-gene-specific covariates for a given set of gene associations. + +We modeled both total expression and ASE by local genetic effects based on detected germline and somatic variation at the respective gene locus and additional covariates using linear regression. We predicted the ASE ratio from the lead eQTL variant (the heterozygosity status of the SNP with greatest effect size from eQTL mapping), the copy-number ratio and binary variables indicating the presence of a structural variation breakpoint overlapping with gene coordinates including +/− 100 kb flanking regions, somatic SNVs in the promoter, and at gene coordinates (including UTRs and introns) as determined by Ensembl variant effect predictor (VEP) version 101.0. Similarly, gene expression was modeled by the genotype (encoded as number of alternative alleles) of the SNP with greatest effect size from eQTL mapping, copy-number LogR, somatic structural variation and somatic SNVs in promoter and gene. Tumor purity and MYCN amplification status were used as additional covariates in models of both expression phenotypes. In the ASE model, the log sum of coverage at the ASE SNPs was used as an additional covariate. A linear model with up to 115 observations was fitted for each gene separately. Only genes with 20 or more complete observations (for effects/covariates and expression phenotype) were considered. The explained variance per genetic effect was determined by its relative contribution to the total sum of squares as given by ANOVA on the fitted model. Significant variance components were determined by ANOVA’s F-statistic and the resulting p-value was adjusted for multiple testing by the Bonferroni method for each effect. Significant effects per gene were defined as effects at 0.05 FDR. + +Pathway enrichment analysis of copy-number effects on gene expression was conducted using the fgsea R package110 version 1.12.0 with inbuilt Reactome pathway definitions. Genes were ranked based on the variance in expression explained by copy-number effects from the variance component analysis. Significant pathways were determined at FDR <0.01. Independently enriched pathways were determined by the collapsePathways function of the fgsea R package with default parameters. + +To associate allelic copy-number differences with survival we summarized copy-number ratio and LogR values in genomic regions. We calculated the copy-number ratio as CNratio = CNmajor/(CNmajor + CNminor), where CNmajor and CNminor, are major and minor allele counts as determined by allele-specific copy-number analysis respectively. CNratio and LogR values were summarized on the level of chromosome arms. Additionally we summarized copy-numbers in 5 Mb bins along the genome. The average value per region was defined as the mean value of CN segments overlapping the genomic region weighted by the length of overlap. 5 Mb bins overlapping amplifications were assigned the value of the amplified CN segment directly, dropping values of other segments overlapping the same bin. We used this strategy in order to maintain information about amplification of (focal) alterations in larger bins (e.g., 1 Mb focal amplification in 5 Mb bins). + +We associated the summarized copy-number ratio per region with patient survival. For each region we tested for the association of copy-number ratio to survival using a generalized linear regression on the binary response “deceased from disease” vs. other. The test was set up to control for covariates MYCN amplification, age, tumor stage 4, sex, tumor purity and tumor ploidy. The association p-value was determined by an analysis of variance (ANOVA) using a Chi-Squared test. The test was carried out between a generalized linear model (GLM) of the covariates above and a second model that included the copy-number ratio in addition to these covariates. Nominal p-values determined for each region were corrected by the Bonferroni method and regions below 0.05 FWER were considered significant. + +We used a Cox proportional hazard model109 to predict overall survival from the copy-number ratio of the chromosomal region identified in the regression analysis (STAR methods). In contrast to the binary survival outcome, here survival times are taken into account. Subsequent significant bins in the discovery model were merged and the average copy-number ratio was determined for the merged bins by the weighted average method as described above. Survival times were predicted by the covariates copy-number ratio, MYCN amplification status, age, tumor stage 4, sex, tumor purity and tumor ploidy. A survival function was estimated by the Kaplan-Meier method. Here, discretized states “balance” (copy-number ratio = 0.5) and “imbalance” (copy-number ratio >0.5) were used to split samples into two groups and to plot the corresponding survival curves. + +We analyzed gene expression differences between ALT and non-ALT tumors by linear regression using a similar methodology to the analysis that identified copy-number differences between these groups described above. We expect that this approach facilitates detection of expression differences mediated by the ALT-associated SCNAs identified. Expression values were modeled by linear combination of ALT status, MYCN amplification, ATRX alteration, age, sex, cohort, tumor purity and tumor ploidy. The p-value was derived from an ANOVA Chi-squared test for significance of the ALT status covariate and adjusted for multiple testing using the Benjamini Hochberg method. Genes with FDR <0.05 were considered as significantly different expressed in ALT tumors. + +DESeq2 1.26.0106 was used to perform differential expression analysis on HTseq/htseq-count 0.9.1105 gene counts between donors marked with survival status deceased from disease according to the clinical annotation file and other donors on variance stabilized raw RNA-seq counts. P-values and log-fold changes of differential expression were obtained controlling for sample covariates cohort, tumor purity, age and sex. Log-fold changes were shrunken using the apeglm method.111 Significant genes were determined at FDR <0.05 based on the Benjamini Hochberg-adjusted p-value from DESeq2. + +To determine genes underlying strong cis-regulatory control by activation or attenuation of gene expression from one of the two alleles, we performed a correlation analysis between ASE and gene expression. ASE ratios were filtered, so that only ratios from 10 or more RNA-seq read counts remained. Variance stabilized RNA-seq reads were matched with ASE ratios by sample and gene. We grouped observations by gene and only considered genes with at least 10 donors informative for ASE in that gene yielding a total of 10,862 genes with sufficient number of observations. Both ASE ratio and gene expression read counts were separately corrected by batch and tumor purity by fitting linear models per gene and obtaining residuals of expression and ASE ratio that were used in the subsequent analysis. Analogous to the ASE ratio, the tumor DNA ratio was defined as max(A,B)/(A + B), where A and B are phased and aggregated read counts of the tumor DNA alignment of expressed heterozygous SNPs gene for the two alleles respectively. We then modeled the ASE ratio by gene expression, cohort, tumor purity, total read count at heterozygous SNP (log) and normal DNA ratio using linear regression and determined a p-value for the gene expression term by ANOVA (F-statistic). Resulting p-values were corrected using the Benjamini-Hochberg procedure. Allelic dosage genes (AD genes) were defined as those genes at FDR <0.05. To identify a subset of AD genes with clinical relevance we matched Pearson’s r of ASE-expression correlation with adjusted p values from differential expression analysis. A subset of differentially expressed genes was defined by intersecting AD genes with genes significantly different expressed between deceased and not-deceased patients (FDR <0.05, Benjamini-Hochberg). + +Protein networks visualizations were created using the STRING DB network viewer (version 11.0b, https://version-11-0b.string-db.org/).112 Settings were adjusted such that line thickness between nodes indicate protein interaction confidence (strength of data support) and minimum required interaction score was set to “medium” (0.4). The default set of interaction sources was used (text mining, experiments, databases, coexpression, neighborhood, gene fusion, cooccurrence). + +The interaction graph in ALT associated protein interactions of ATRX and histone variant genes was subset to proteins of differentially expressed genes (ALT) (FDR <0.05) with correlation to LogR of 11q or 17q with abs(r) > 0.3, where r is Pearson’s correlation coefficient. Network nodes were colored based on up- (red) or down- (blue) regulated genes as identified in ALT differential expression analysis. Enrichment of biological processes (GO terms) in the protein network of 17p dosage effects were determined by the STRING network analysis tool. + +To quantify telomeric RNA expression for the subset of n = 63 tumors of the cohort for which total RNA libraries were available, we adapted the approach of Wang et al. 2015113 to paired-end sequencing data. Only considering RNA fragments, for which any of the two reads contained at least five perfect consecutive matches of the telomere repeat motif TTAGGG (UUAGG), we normalized these fragment counts by total fragments counts to derive FPM (fragments per million) of telomeric RNA per tumor sample. + +Reads were extracted with bcl2fastq (v.2.19.0.316) and processed with a developer version of the PiGx BS-seq pipeline.114 Quality control was performed with FastQC 0.11.9. Reads were trimmed with TrimGalore v.0.6.2 and aligned to the bisulfite-converted reference human genome GRCh37 using bwa-meth v.0.7.17.115 Duplicate reads were removed with samblaster v.0.1.24.96 Germline C/T SNPs detected in our cohort at MAF >0.01 were removed using bcftools 1.9. CpG DNA methylation calling was performed with methylKit v.1.15.4116 at a minimum 10x coverage. + +The GPOH-NB2004 clinical trial is identified by EUDRAC number EU-20661 and ClinicalTrials.gov ID NCT00526318. GPOH-NB2004 clinical trial information: https://www.gpoh.de/kinderkrebsinfo/content/e1676/e9032/e68518/e206421/index_ger.html. + +## Table + +| REAGENT or RESOURCE | SOURCE | IDENTIFIER | +| Deposited data [colspan=3] | +| [colspan=3] | +| Whole-genome sequencing data of primary neuroblastoma | European Genome-phenome Archive (EGA) | EGAS00001001308 | +| Whole-genome- and RNA-sequencing data of primary neuroblastoma | European Genome-phenome Archive (EGA) | EGAD00001005488 | +| Variant call files | 1000 Genomes Project (phase 3)93 | https://www.internationalgenome.org/category/phase-3/ | +| Mappability track | UCSC Genome Browser94 | wgEncodeCrgMapabilityAlign50mer https://genome.ucsc.edu/ | +| Processed data and analysis code | – | https://doi.org/10.5281/zenodo.8373208 | +| [colspan=3] | +| Software and algorithms [colspan=3] | +| [colspan=3] | +| Telseq v0.0.2 | Ding et al.38 | https://github.com/zd1/telseq | +| Bcftools v1.8 | – | https://www.htslib.org/ | +| ASCAT v2.6 | Loo et al.95 | https://github.com/VanLoo-lab/ascat/ | +| HTseq v0.9.1 | Putri et al.96 | https://github.com/htseq/htseq | +| DESeq2 v1.26.0 | Love et al.97 | https://github.com/htseq/htseq | +| Ensembl variant effect predictor (VEP) v101.0 | McLaren et al.98 | https://www.ensembl.org/info/docs/tools/vep/index.html | +| BWA-MEM v0.7.15 | Li and Durbin99 | https://github.com/lh3/bwa | +| STAR v2.5.3a | Dobin et al.100 | https://github.com/alexdobin/STAR/releases | +| GATK/Mutect2 v2.2 | McKenna et al.101 | https://gatk.broadinstitute.org/ | +| Novobreak v1.1.3 | Chong et al.102 | https://sourceforge.net/projects/novobreak/ | +| MEDICC2 | Kaufmann et al.40 | https://pypi.org/project/medicc2/0.5b3/ | +| fgsea v1.12.0 | Korotkevich et al.103 | https://bioconductor.org/ | +| GATK/ASEReadCounter v3.5.0 | McKenna et al.101 | https://gatk.broadinstitute.org/ | +| PEER | Stegle et al.104 | https://github.com/PMBio/peer/ | +| FastLMM v0.2.23 | Lippert et al.105 | https://github.com/fastlmm/FaST-LMM | +| STRING DB network viewer v11.0b | Szklarczyk et al.106 | https://string-db.org/ | +| bcl2fastq v.2.19.0.316 | – | https://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.html | +| PiGx BS-seq pipeline | Wurmus et al.107 | https://github.com/BIMSBbioinfo/pigx_bsseq | +| FastQC v0.11.9 | – | https://github.com/s-andrews/FastQC | +| TrimGalore v.0.6.2 | – | https://github.com/FelixKrueger/TrimGalore | +| bwa-meth v.0.7.17 | Pedersen et al.108 | https://github/com/brentp/bwa-meth/ | +| samblaster v.0.1.24 | Faust et al.109 | https://github.com/GregoryFaust/samblaster | +| methylKit v1.15.4 | Akalin et al.110 | https://github.com/al2na/methylKit | \ No newline at end of file diff --git a/references_cache/PMID_40638875.md b/references_cache/PMID_40638875.md new file mode 100644 index 00000000000..2a2a67ee228 --- /dev/null +++ b/references_cache/PMID_40638875.md @@ -0,0 +1,168 @@ +--- +reference_id: PMID:40638875 +extractor_version: 1 +xml_extraction_version: 1 +title: Clinical Significance of TERT Promoter Mutations in Neuroblastoma. +authors: +- Persaud N +- Gundem G +- Kushner BH +- Ladanyi M +- Shukla N +- Modak S +journal: JCO Precis Oncol +year: '2025' +doi: 10.1200/PO-25-00074 +keywords: +- Humans +- Telomerase/genetics +- "Neuroblastoma/genetics, mortality" +- "Promoter Regions, Genetic" +- Female +- Male +- Adolescent +- Adult +- Middle Aged +- Young Adult +- Aged +- Mutation +- Clinical Relevance +publication_types: +- Journal Article +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# Clinical Significance of TERT Promoter Mutations in Neuroblastoma. +**Authors:** Persaud N, Gundem G, Kushner BH, Ladanyi M, Shukla N, Modak S +**Journal:** JCO Precis Oncol (2025) +**DOI:** [10.1200/PO-25-00074](https://doi.org/10.1200/PO-25-00074) + +## Content + +PURPOSE: Telomere maintenance mechanisms including MYCN amplification (MYCN-A) (via upregulated telomerase reverse transcriptase [TERT] expression), TERT rearrangements (TERT-RA), and ATRX mutations confer neoplastic immortality in neuroblastoma (NB) cells. Clinical characterization of patients with NB harboring activating somatic TERT promoter point mutations (TERT-PM) in NB may improve stratification. + +METHODS: To identify TERT-PM and TERT-RA, tumors were profiled by the Memorial Sloan Kettering Cancer Center-Integrated Mutation Profiling of Actionable Cancer Targets targeted next-generation sequencing platform and whole-genome sequencing, respectively. Associated clinical outcomes were studied. + +RESULTS: TERT-PM and TERT-RA were detected in 17 of 603 (2.8%) and 31of 168 (18.4%) tumors from individual patients, respectively. The median age at diagnosis was 32 (range, 15-79) and 48 (range, 3-168) months for TERT-PM and TERT-RA, respectively. TERT-PM were located at canonical hotspots (C228T [16/17] and C250T [1/17]) and were concurrent with MYCN-A in 9 of 17 (53%). By contrast, MYCN-A occurred in 1 of 31 TERT-RA tumors. Most patients with TERT-PM had stage M (94%) and high-risk NB (HR-NB) (81%) at diagnosis. Twenty-eight patients with TERT-RA had HR-NB (90%). After risk appropriate therapy, complete response was achieved in 7 of 17 (41%) and 17 of 31 (55%) patients with TERT-PM and TERT-RA, respectively. The median progression-free (PFS) and overall survival (OS) were 9.7 ± 1.2 and 16.5 ± 2 months for patients with TERT-PM. Corresponding PFS/OS for patients with TERT-RA were 20 ± 6.4/56.8 ± 8.6 months (P = .015 for OS). Excluding MYCN-A status produced no significant survival difference (P > .1) between the two groups. CNS relapse occurred in 11 of 31 (35%) patients with TERT-RA versus 1 of 17 (6%) patients with TERT-PM. + +CONCLUSION: TERT-PM is rarer than TERT-RA but both are associated with HR-NB and poor prognosis. MYCN-A frequently co-occurs with TERT-PM and might represent an ultra-high-risk subset of patients. + +Neuroblastoma, the most common extracranial solid tumor of childhood, has a highly heterogenous clinical behavior presenting as locoregional or widely metastatic disease with spontaneous regression, cure with little or no cytotoxic therapy, or fatal outcomes despite aggressive multimodality treatment. + +Advances in molecular techniques have highlighted the critical role of telomere maintenance mechanisms that stabilize telomeres for unlimited replicative capacity and immortality in the pathogenesis of high-risk neuroblastoma (HR-NB)1–4. Telomere maintenance is mediated by telomerase-dependent (classical) and independent (alternate) pathways. The former includes activation of the telomerase reverse transcriptase (TERT) gene. Reported mechanisms for TERT activation in neuroblastoma include MYCN amplification (MYCN-A)5 (via upregulation of TERT, a direct transcriptional target of MYCN) or rearrangement of the TERT locus6. Whereas MYCN-A in NB was discovered decades ago, TERT-rearrangements (TERT-RA) have only recently been described in NB and are associated with tumors of aggressive phenotype in NB despite not co-occurring with MYCN-A5–7. Genomic aberrations of the ATRX gene are most frequently associated with activation of the alternate lengthening of telomeres (ALT) pathway8. Among HR-NB, the most common telomeric maintenance mechanisms – TERT-RA, ATRX mutation, and MYCN-A appear to be mutually exclusive. + +Activating somatic TERT promoter mutations (TERT-PM) have been described in several adult tumors including melanoma, glioblastoma, bladder cancer, hepatocellular carcinoma, thyroid and non-small cell lung cancer9, typically involve two hotspot mutations: C228T and C250T9, and are associated with a poor prognosis in several tumor types.10,11 These mutations appear to be functionally redundant in that they both generate a new consensus binding motifs for E-twenty-six (ETS) transcription factors and the subsequent constitutional activation and overexpression of TERT12–15. In NB, TERT-PM have been anecdotally investigated in cell lines and have previously reported to be absent in NB16 . We first noted TERT-PM in 9 patients in an investigation into clonal evolution of NB,17 but clinical impact was not described. We have identified additional patients with NB with TERT-PM from an expanded pool of NB tumor samples and we now report on clinical findings in comparison with those associated with TERT-RA. + +After obtaining approval from the Institutional Review Board of Memorial Sloan Kettering Cancer Center (MSK) records of patients with neuroblastoma who underwent targeted exome sequencing and/or whole genome sequencing between 1996–2023 were reviewed. No patients were excluded from analysis based on race, age, gender or ethnicity. Patients were enrolled on an institutional profiling protocol and informed consent was obtained (clinicaltrials.gov +NCT01775072). Patients harboring tumors with TERT-PM or TERT-RA were identified and included in analysis of clinical and biological data. Demographics are provided in Table 1. Staging was recorded using International Neuroblastoma Risk Group staging system and response using the International Neuroblastoma Response Criteria18. Progression-free survival (PFS) was defined as time from diagnosis to detection of progressive disease (PD). Overall survival (OS) time was censored at date of last contact. Survival was analyzed by the Kaplan-Meier method. Differences between groups were analyzed using chi-squared test. + +Targeted tumor sequencing was performed using the CLIA-certified MSK-IMPACT platform that utilizes established pipelines followed by manual review were to characterize germline and somatic mutations, CN variants, and if targeted, genomic rearrangements18. WGS analysis was performed using the Isabl platform19 and included data quality control, variant calling for somatically acquired mutations for each variant class, and variant classification. Tumor DNA was extracted from fresh frozen or OCT-embedded tissue biopsies and matched normal DNA from buffy coat using the DNeasy Blood & Tissue Kit (Qiagen). FFPE tissues were deparaffinized and DNA eluted. Genomic DNA was sheared, and sequencing libraries were prepared using the KAPA Hyper Prep Kit (Kapa Biosystems KK8504) with modifications. Libraries were subjected to a 0.5X selection using AMPure XP beads (Beckman Coulter) after post-ligation cleanup. Libraries were not amplified by PCR and were pooled equivolume for sequencing. Samples were run on a NovaSeq 6000 (Illumina; RRID:SCR_016387). Tumors were covered to an average of 80–90X, and normal covered at 42–52X20. Briefly, upon completion of each sequencing run, Isabl imports paired tumor-normal FASTQ files, executes alignment using BWA-mem21 (RRID:SCR_010910) as implemented by pcap-core v2.18.2 wrapper [https://github.com/cancerit/PCAP-core] and quality control using MultiQC (v1.9) (https://github.com/ewels/MultiQC) (RRID:SCR_014982), FastQC (v0.11.5) (https://github.com/s-andrews/FastQC) (RRID:SCR_014583) and Picard (v2.25.6) (https://github.com/broadinstitute/picard) (RRID:SCR_006525). Genome-wide median coverage was calculated using Mosdepth (RRID:SCR_018929)22. Estimation of tumor purity/ploidy and identification of copy number aberrations were performed using the Battenberg algorithm23. High-confidence mutations for single nucleotide variants (SNVs), insertions and deletions (indels) and structural variations were identified as those mutations called by at least two out of three algorithms run: 1) Caveman24 (RRID:SCR_017089), Mutect25 (RRID:SCR_000559) and Strelka26 (RRID:SCR_005109) for SNVs 2) cgpPindel27 (RRID:SCR_0107090), Mutect and Stelka for Indels 3) SvABA28 (RRID:SCR_022998), GRIDSS29 and BRASS (RRID:SCR_017091) [https://github.com/cancerit/BRASS] for SVs. High-confidence somatic SNV/indels were annotated for functional consequence using Ensembl Variant Effect Predictor (v92)30 (RRID:SCR_007931) and VAGrENT (v3.3.0) (RRID:SCR_005180). Mutations were further classified with regards to their putative role in cancer pathogenesis by assessing their clinical relevance using OncoKb (v1)31 (RRID:SCR_014782), COSMIC (v92)32 and ClinVar (2022_03_20)33 databases. Structural variants were annotated using a modified version [https://github.com/papaemmelab/AnnotSV] of AnnotSV [https://github.com/lgmgeo/AnnotSV]. Mutational signature analysis was performed with the MutationalPatterns package (v1.6.1)34 (RRID:SCR_024247) and using the COSMIC Mutational Signatures (v3.1) with the addition of Temozolomide signature from Kucab et al.35. Microsatellite-instability status was assessed using MSISensor36 (RRID:SCR_006418). Further details of the analysis pipeline have been reported previously37. + +TERT-PM were assessed using either IMPACT or WGS. However, since the former is not sensitive for detection of rearrangements, TERT-RA were evaluated only by WGS37. + +Clonal structure was analyzed using the union of high-confidence SNVs called in all WGS samples for a patient. DPClust (v0.2.2, https://github.com/Wedge-Oxford/dpclust) was used for calculation of cancer cell fraction corrected for purity and local CN, as well as clustering and assignment of mutations across samples with the exception of the Gibbs Sampling Dirichlet Process step that was optimized internally23. Clonal ordering was deduced using clonevol38 (v0.99.11, https://github.com/hdng/clonevol). Mutational signatures were computed in each cluster independently. Figures were generated with matplotlib (v3.1.0, https://matplotlib.org/). + +Tumor tissue was homogenized in 1mL TRIzol Reagent (Thermo Fisher Scientific) followed by phase separation with 200 μL chloroform. RNA was extracted from the aqueous phase using the miRNeasy Micro Kit on the QIAcube Connect (Qiagen) according to the manufacturer’s protocol with 350 μL input. Samples were eluted in 15 μL of RNase-free water. After RiboGreen quantification (Thermo Fischer Scientific) and quality control by Agilent BioAnalyzer (RRID:SCR_018043), 18ng- 1μg of total RNA with an RNA integrity number varying from 1 to 0.9 underwent ribosome depletion and library preparation using the TruSeq Stranded Total RNA LT Kit (Illumina) according to instructions provided by the manufacturer with 8 cycles of PCR. Samples were barcoded and run on a HiSeq 2500 in Rapid Mode or HiSeq4000 at PE100 or on a NovaSeq 6000 et PE150, using the HiSeq SBS Kit v2, HiSeq 3000/4000 SBS Kit or NovaSeq 6000 SP, S1 or S4 Reagent Kit (300 cycles (Illumina)). Sequencing was performed to achieve a median of 80 million plated reads per sample. Processing of RNA sequencing was performed as described previously37. + +MSK-IMPACT sequencing data is deposited to AACR Project GENIE (https://genie.cbioportal.org). All additional data generated in this study are not publicly available to protect patient privacy but are available upon reasonable request from the corresponding author. + +Of 603 NB tumors from individual patients profiled by MSK-IMPACT, 17 (2.8%) had TERT-PM. Of these 17 samples with TERT-PM, 13 (77%) were identified prior to relapse (at diagnosis or after neoadjuvant chemotherapy). At diagnosis, patients with TERT-PM were 16–79 (median 30) months old; 16 (94%) had metastatic disease, and 15 (88%) had HR-NB (Table 1). One patient with MYCN-non-amplified (MYCN-NA) stage L2 diagnosed at age 16 months was considered to have low-risk NB. One patient with intermediate-risk NB had MYCN-NA stage M disease diagnosed at age 15 months. Nine patients, all with stage M at diagnosis, had MYCN-A. + +All 15 patients with stage M HR-NB received appropriate neoadjuvant induction chemotherapy. The stage L2 patient presented with an incidentally discovered left adrenal mass and was followed without intervention. But 16 months after diagnosis, he had widespread PD received HR-NB therapy. However, 7 months after achieving remission PD occurred again. Salvage therapy achieved another remission but 22 months after diagnosis he developed myeloid neoplasm post-cytotoxic therapy which was successfully treated. He remains disease-free 56+ months post-diagnosis of NB and 30+ months post-diagnosis of leukemia. (Table 2) The patient with intermediate-risk NB at diagnosis received risk-appropriate chemotherapy plus crizotinib (for ALK mutation R1245C) but developed rapid osteomedullary PD and died of disease 7 months post-diagnosis. + +Of 168 tumors from individual patients with NB with tumors profiled by WGS, 31 (18.4%) had TERT-RA. At diagnosis, patients with TERT-RA were 3–168 (mean 48) months old and 28 (90%) had Stage M HR-NB. Three patients had intermediate-risk locoregional NB (all stage L2). Two of the latter, respectively diagnosed at 1 month and 4 years, developed metastatic PD, and died of disease 48 months and 14 months post-diagnosis. The third stage L2 patient was diagnosed at age 4 years, had PD which was treated with aggressive multi-modality HR-NB therapy, and remains disease-free 40+ months post-diagnosis (Supplemental Table 1). + +In tumors with TERT-PM, C288T and C250T hotspot mutations were observed in 94% (16/17) and 6% (1/17) of patients, respectively. MYCN-A was co-expressed in 9/17 (53%) and ALK mutations in 5/17 (29%) (Table 1). As expected, no tumor co-expressed TERT-PM and ATRX aberrations or TERT-RA. No patient had oncogenic germline aberrations in TERT, although 1 patient each with TERT-PM carried germline mutations in CHEK2 and PHOX2B. Of the 31 tumors with TERT-RA, only 1 co-expressed MYCN-A. No aberrations were noted in TERT-PM, ATRX or ALK in tumors with TERT-RA. + +TERT mRNA expression was assessed in 184 NB tumor samples. patients. TERT mRNA expression was significantly higher in MYCN-A tumors ( n=25) when compared to MYCN-NA (n=159). In the MYCN-NA cohort, mRNA expression was significantly higher in patients with TERT-RA (n=27) compared to wild type TERT (n=126) (p=0.0000006). There was no statistically significant difference in expression in TERT-PM compared to wild type TERT or TERT-RA though there appeared to be a trend towards higher expression in tumors harboring TERT-PM (Supplemental Figure 1). + +Five patients with TERT-PM had ≥2 separate tumors available for analysis. In 4 patients TERT-PM was shared clonally across all tumors while in 1 patient (H132374) it was present at 5% in the diagnostic tumor and became fully clonal in a lesion metastatic to the brain. Seven TERT-RA patients had ≥2 tumors evaluable. In 3 patients TERT-RA was shared by all tumors tested. In all 7 patients there were additional TERT-RA in later subclones. In 5/6 patients with diagnostic and relapse tumors the subclonal TERT-RA was specific to the relapse. (Supplemental Figure 2) These relapse-specific rearrangements result in increased copy number and expression of TERT consistent with an augmented addiction to TERT signaling17. + +Twelve patients of 17 with TERT-PM underwent attempted resection of primary tumor after induction chemotherapy. Gross total (R1) resection was possible in 11 (90%). Resection was not attempted in 4 due to disease progression while receiving induction chemotherapy. Initial management in 1 patient was watchful observation without attempt at resection. For the 17 patients with TERT-PM, responses to induction were CR in 7 (41.2%), PR in 3 (17.6%), SD in 2 (11.7%) and PD in 5 (29.4%). One patient died of sepsis. + +Of 31 patients with TERT-RA, 1 patient had PD while receiving induction chemotherapy, and resection was not attempted after partial response in another. Resection was attempted in the other 29 with gross total resection possible in 23 (79%). + +For the 31 patients with TERT-RA, responses to induction were CR in 17 (55%), PR in 10 (34%), SD in 2 (7%) and PD in 2 (3%). + +There was no significant difference in major responses (CR+PR) to induction between the two groups (p>0.05) though PD while receiving therapy was significantly higher in the TERT-PM group (p=0.03). + +Median PFS and OS of patients with TERT-PM was 9.7 ± 1.2 and 20 ± 6.4 months, respectively (Table 3). Corresponding survival probabilities for the TERT-RA group were 16.5 ± 1.1 and 56.8 ± 8.6 months (Figure 1). There was no difference in PFS between the 2 groups. OS was significantly better for TERT-RA than TERT-PM (p=0.01). When patients with MYCN-A were removed from the analysis, there was no statistically significant difference in survival between the 2 groups (p>0.1 for PFS and OS). PFS and OS were similar in patients with and without MYCN-A in the TERT-PM group (n=8 and 9, respectively). Patterns of relapse were similar among patients with TERT-PM and TERT-RA with one exception: central nervous system relapse occurred in 11/31 (35%) patients with TERT-RA (median follow-up 48 months) versus 1/17 (6%) patients with TERT-PM (median follow-up 16.5 months). Five patients with TERT-PM survive at a median follow up of 21 months. All received similar multimodality therapy to those who relapsed (Table 2). However, non-relapsed patients appeared to have highly chemosensitive disease, 4/5 achieving CR compared to only 2/15 patients who relapsed. Of the 603 patients with tumors undergoing testing by MSK-IMPACT, PFS and OS could be determined in a total of 520 including 17 with TERT-PM. For the 503 patients whose tumors did not harbor TERT-PM, median PFS and OS was 32.6 ± 1.2 and 145 ± 15.3 months respectively (p<0.001 for both when compared to those with TERT-PM). + +TERT-RA is one of the more common genetic aberrations in NB with some investigators reporting an incidence of 20–30% of HR-NB16. A similar incidence (16.7%) was encountered in our cohort of patients. Both MYCN, a known transcriptional activator of TERT and TERT-RA independently activate TERT gene transcription and converge on telomerase-dependent telomere maintenance mechanisms5–7,39. The presence of TERT-RA is associated with a poor prognosis in patients with NB with initial observation in small numbers of patients indicating survival of 20–40%6,7. + +TERT-PM first described in patients with melanoma40,41 have since been discovered in several other malignancies and in general are associated with a poor prognosis. However, little is known about TERT-PM in NB. In reports on tumor screens for presence of TERT-PM, one report observed their presence in 3/16 NB cell lines tested42 and another in 2/22 primary tumors43. A prior study reported on the absence of TERT-PM in NB tumors from 131 patients. The lack of TERT-PM in that study could be attributed to the relatively small number of samples tested for an aberration that is rare (2.8% in our cohort) and inclusion of a relatively few HR patients in the published study16. On the other hand, our study surveyed tumors from > 500 patients. In our cohort, TERT-PM mutations in NB occured in the canonical hotspots at a relative frequency typical for that reported in other malignancies i.e. C228T being far more frequent than C250Y (93% vs 7% in our cohort) and were mutually exclusive, i.e. no tumor harbored both mutations. As expected, ATRX aberrations were not observed in any tumor with TERT-PM. TERT-PM mutations were less common than TERT-RA (3% vs 18%) and there was mutual exclusivity between the two aberrations. In NB, TERT-RA has been reported to be mutually exclusive with MYCN-A6,7. Indeed, we encountered co-expression of MYCN-A and TERT-RA in only 1/31 tumors tested. On the other hand, TERT-PM were both observed in MYCN-A and MYCN-NA tumors. 50% of our patients had co-expression of TERT-PM and MYCN-A. TERT-PM appear to arise early in the development of the disease; all 4 patients in whom multiple samples were available, had TERT-PM noted before and after relapse. + +We first observed TERT-PM in 9 patients17 but did not previously study their clinical implications. In this current expanded cohort that includes samples from the prior study, we noted TERT-PM in a small proportion of NB tumors with an incidence of 2.8%. However, with the increasing use of NGS in assessment of NB tumors, it is possible that further tumors with TERT-PM will be discovered. TERT-PM, similar to TERT-RA appear to confer an aggressive clinical phenotype. In our cohort of 17 patients, TERT-PM were invariably associated with HR-NB (15/17). Furthermore, the 2 patients that were deemed to have low or intermediate risk disease at diagnosis, had rapid disease progression. Response to induction therapy appeared to be similar in both TERT-RA and TERT-PM, but patients with the latter had a higher rate of PD while receiving chemotherapy (even without concurrent MYCN-A). Both TERT-PM and TERT-RA were associated with a dismal prognosis with median long-term PFS and OS < 5 years. The prognosis appeared to be even worse in those patients with tumors co-expressing TERT-PM and MYCN-A, though the individual contribution of each genetic aberration could not be determined. However in patients harboring tumors with TERT-PM, robust response to induction chemotherapy appeared to imbue patients with a better prognosis. Four of five patients who did not experience relapse had CR to induction chemotherapy compared to 2/15 who relapsed. + +CNS relapse is rare in NB occuring in 5–10% of patients with NB44,45 with a higher incidence in MYCN-A tumors compared to MYCN-NA (11% of MYCN-A vs 3%)46,47. Our group previously reported that MYCN-NA CNS metastases show focal gains at 5p15.33, a region corresponding to the TERT locus and have higher higher TERT mRNA expression, suggesting a role of TERT in CNS relapse of NB. However, we had not hitherto performed genomic TERT studies48. It is therefore not surprising that TERT aberrations would be associated with CNS metastases: we report CNS metastases in 38% of patients with TERT-RA tumors. Although CNS metastases occurred in only 1/16 patients with TERT-PM, the median follow up of this cohort is relatively short. TERT-PM have been detected at increased frequency in thyroid malignancies metastatic to the CNS, 14/15 in one study49. + +NB is a disease well known for its biologic and clinical heterogeneity, and patients have benefited from improved understanding of the genetic and/or molecular properties of individual tumors. Despite the reported independent prognostic impact of genetic aberrations impacting telomere-maintenance mechanisms, neither ATRX nor TERT aberrations are included in the current stratification and management strategy of HR-NB. Although in published reports, TERT-RA have been reported exclusively in patients deemed to have high-risk NB, TERT-PM can be observed even in those patients considered to have low or intermediate risk disease as in 2 of the patients in our cohort. In one of these patients, administration of high-dose multimodality therapy resulted in remission after relapse and he continues to be in remission 44 months after initiation of salvage therapy. Finally, attempts are currently being made to target TERT. The FDA has recently approved a telomerase inhibitor, imetelstat, for MDS50 which has already undergone phase I evaluation in children51. Another telomerase-interacting compound, 6-thio-2’-deoxyguanosine (6-thio-dG)52, has been tested preclinically in NB models. Several other therapeutic approaches targeting TERT and related molecules are being investigated for a variety of cancers53. Therefore, routine analysis for TERT-PM and TERT-RA may arguably help refine risk stratification and identify patients with NB who may be eligible for trials of novel TERT-targeted agents. + +## Table 1. Patient and disease characteristics. + +| | | TERT-PM | TERT-RA | +| N | | 17 | 31 | +| Sex M:F | | 10:7 | 16:15 | +| Median (±SD) Age at diagnosis (months) | | 32±17 | 48±26 | +| Site of primary | Adrenal | 12 | 21 | +| | Central abdominal | 4 | 8 | +| | Mediastinal | 1 | 2 | +| Stage | M (4) | 16 | 28 | +| | L2 (3) | 0 | 2 | +| | L1 (1) | 1 | 1 | +| MYCN status | Amplified | 9 | 1 | +| | Non-amplified | 8 | 31 | +| ALK status | Mut | 5 | 0 | +| | WT | 12 | 31 | +| INRG Risk | High | 15 | 28 | +| | Intermediate | 1 | 3 | +| | Low | 1 | 0 | + +Abbreviations: TERT-PM, TERT promoter mutation; TERT-RA, TERT rearrangement; INRG, International Neuroblastoma Risk Group; Mut, mutant; WT, wild type + +## Table 2. Clinical characteristics, genetic findings, treatment, response and outcome for 17 patients with TERT promoter mutations. + +| Demographics [colspan=3] | Clinical Characteristics [colspan=3] | Genetic findings [colspan=4] | Initial Treatment and Response [colspan=6] | Relapse/Current status [colspan=4] | +| Pt.# | Gender | Age at dx (months) | Stage | Primary Site | INRC risk | Timing of sequencing | ALK | MYCN | TERT-PM | HDC Induction | Surgery | HSCT | XRT | Response to induction | Anti GD2-Ab | Relapse status | Time to relapse/months | Status | Survi-val | +| 1 | M | 15 | M | Adrenal | Int | pre-relapse | R1245 | NA | C228T | N | R1 | N* | N | PD | N | Yes | 3.6 | DOD | 6.9 | +| 2 | M | 15 | M | Adrenal | High | Relapse | WT | A | C228T | Y | R2 | N* | Y | SD | Y | Yes | 5.1 | DOD | 11.9 | +| 3 | M | 16 | L2 | Adrenal | Low | pre-relapse | WT | NA | C228T | N | None | N | N | PR | N | Yes | 18.6 | NED | 63.3+ | +| 4 | F | 17 | M | Adrenal | High | pre-relapse | WT | A | C228T | Y | R1 | N | Y | CR | Y | No | 33.5 | NED | 33.5+ | +| 5 | F | 19 | M | Adrenal | High | pre-relapse | WT | A | C228T | Y | R1 | N | Y | CR | Y | Yes | 9.5 | NED | 11+ | +| 6 | M | 20 | M | Adrenal | High | Relapse | F1174 | A | C228T | Y | None* | N | Y* | PD | Y | Yes | 8.1 | AWD | 9.1+ | +| 7 | F | 21 | M | Adrenal | High | pre-relapse | WT | NA | C228T | Y | R1 | N | Y | CR | Y | No | 21.7 | NED | 21.7+ | +| 8 | F | 27 | M | Abdominal | High | Both | WT | NA | C228T | Y | R1 | N | Y | PR | Y | No | 9.1 | NED | 9.1+ | +| 9 | M | 30 | M | Adrenal | High | pre-relapse | WT | A | C250T | Y | R1 | N* | N* | PR | N* | No | 6.6 | Toxic death | 6.6 | +| 10 | M | 31 | M | Adrenal | High | pre-relapse | WT | A | C228T | Y | R1 | Y | Y | CR | Y | Yes | 12.3 | DOD | 25.5 | +| 11 | F | 34 | M | Abdominal | High | pre-relapse | F1174 | NA | C228T | Y | None* | N* | N* | PD | Y** | Yes | 4.2 | DOD | 9.2 | +| 12 | F | 34 | M | Adrenal | High | Relapse | F1174 | A | C228T | Y | R1 | N* | N* | CR | Y** | Yes | 8.4 | DOD | 17.4 | +| 13 | M | 35 | M | Adrenal | High | Relapse | WT | A | C228T | Y | R1 | Y | Y | CR | Y | Yes | 20.6 | DOD | 31.9 | +| 14 | M | 36 | M | Adrenal | High | pre-relapse | WT | NA | C228T | Y | R1 | N | Y | CR | Y | No | 60.1 | NED | 60.1+ | +| 15 | F | 46 | M | Abdominal | High | pre-relapse | F1174 | NA | C228T | Y | None* | N* | N* | PD | Y** | Yes | 5.0 | DOD | 9.3 | +| 16 | M | 55 | M | Adrenal | High | Both | WT | A | C228T | Y | None* | N* | N* | SD | Y** | Yes | 9.7 | DOD | 10.9 | +| 17 | M | 79 | M | Abdominal | High | pre-relapse | WT | NA | C228T | Y | R1 | N* | N* | PD | N* | Yes | 7.7 | DOD | 20.0 | + +Abbreviations: dx, diagnosis; Int, intermediate; WT, wild type; R1, gross total resection; R2, incomplete resection; HDC, high-dose chemotherapy; HSCT, hematopoietic stem cell transplantation; XRT, radiotherapy; AWD, alive with disease; NED, no evaluable disease; DOD, died of disease * relapse before planned therapy. + +## Table 3. Outcomes of risk appropriate therapy for TERT-PM and TERT-RA + +| Outcomes | | TERT-PM (n=17) | TERT-RA (n=31) | +| Resectability after induction | R1 | 11 | 23 | +| | Incomplete | 1 | 6 | +| | Not attempted | 5 | 2 | +| Response to induction | CR | 7 | 17 | +| | PR | 3 | 10 | +| | SD | 2 | 2 | +| | PD | 5 | 2* | +| Median PFS (±SEM) months | | 9.7 ± 1.2 | 16.5 ± 2 | +| Median OS (±SEM) months | | 20 ± 6.4 | 56.8 ± 8.6* | +| | | | | +| Median PFS (±SEM) months in MYCN-NA | | 7.7 ± 8.2 | | +| Median PFS (±SEM) months in MYCN-A | | 9.7 ± 0.9 | | +| Median OS (±SEM) months in MYCN-NA | | 20 ± 13.8 | | +| Median OS (±SEM) months in MYCN-A | | 25.5 ± 7.8 | | +| Median PFS excluding MYCN-A (±SEM) months | | 7.7 ± 8.3 | 16.3 ± 1.1 | +| Median OS excluding MYCN-A (±SEM) months | | 20 ± 13.8 | 48.8 ± 6.9 | \ No newline at end of file diff --git a/references_cache/PMID_42558321.md b/references_cache/PMID_42558321.md new file mode 100644 index 00000000000..5a48ba498f3 --- /dev/null +++ b/references_cache/PMID_42558321.md @@ -0,0 +1,181 @@ +--- +reference_id: PMID:42558321 +extractor_version: 1 +xml_extraction_version: 1 +title: "ATRX large deletions and truncating mutations are associated with primary induction chemotherapy resistance in high-risk neuroblastoma: a multicenter propensity-weighted cohort study." +authors: +- Sun J +- Qian B +- Lu L +- Zhou J +journal: Front Oncol +year: '2026' +doi: 10.3389/fonc.2026.1857869 +publication_types: +- Journal Article +content_type: full_text_xml +full_text_attempted: true +full_text_provider: pmc +oa_status: green +--- + +# ATRX large deletions and truncating mutations are associated with primary induction chemotherapy resistance in high-risk neuroblastoma: a multicenter propensity-weighted cohort study. +**Authors:** Sun J, Qian B, Lu L, Zhou J +**Journal:** Front Oncol (2026) +**DOI:** [10.3389/fonc.2026.1857869](https://doi.org/10.3389/fonc.2026.1857869) + +## Content + +BACKGROUND: ATRX alterations are adverse prognostic factors in high-risk neuroblastoma patients. However, the precise impact of distinct ATRX mutation subtypes-particularly multiexon deletions versus missense mutations-on response to standard-of-care multiagent induction chemotherapy remains poorly defined. + +METHODS: This multicenter retrospective cohort study evaluated newly diagnosed patients with high-risk neuroblastoma who underwent comprehensive molecular profiling and received platinum-containing multiagent induction chemotherapy between 2017 and 2022.center Following 1:2 subtype-based stratified random sampling, patients were categorized into the ATRX wild-type (WT), large deletion/truncating (LDT), and missense mutation subgroups. Multiple imputation by chained equations (MICE) and inverse probability of treatment weighting (IPTW) were applied to control for baseline confounders. The primary endpoint was the objective response rate (ORR), which was assessed by a blinded independent central review. Secondary endpoints included event-free survival (EFS), overall survival (OS), and cumulative incidence of progression (CIP). + +RESULTS: The final weighted pseudocohort comprised 138 patients (WT, n = 92; LDT, n = 33; missense, n = 13). Compared with the WT group (71.0%), the LDT group exhibited a substantially lower ORR (36.0%; P < 0.001), with a primary induction-refractory rate of 64.0%. Patients with missense mutations demonstrated an intermediate ORR (60.0%, P = 0.285 vs. WT). At a median follow-up of 42.5 months, the LDT group had a lower 3-year EFS rate than the WT group (18.5% vs. 52.4%; P < 0.001) and a higher 3-year CIP (72.6% vs. 35.2%; subdistribution hazard ratio, 2.88; P < 0.001). The association between the LDT subtype and poor induction response persisted across clinical strata and was independent of 11q deletion or MYCN amplification status. + +CONCLUSIONS: Large ATRX deletions and truncating mutations identify an ultrahigh-risk molecular subtype of neuroblastoma characterized by primary resistance to conventional induction chemotherapy and accelerated progression. Detailed ATRX structural-variant subtyping may facilitate the early identification of patients at high risk of primary induction failure and supports the prospective evaluation of alternative or molecularly targeted induction strategies. + +Neuroblastoma is the most prevalent extracranial solid tumor in pediatric populations worldwide, representing approximately 8–10% of childhood malignancies and contributing to approximately 15% of pediatric cancer-related fatalities (1). Despite intensive multimodal therapy, the disease is characterized by remarkable clinical heterogeneity, with outcomes ranging from spontaneous regression in infants with favorable biology to relentless disease progression. Approximately half of neuroblastoma cases are classified as high risk according to the International Neuroblastoma Risk Group (INRG) classification system, which integrates clinical, histopathological, and molecular parameters for pretreatment risk stratification (2). Despite significant advances in multimodal therapeutic approaches, such as induction chemotherapy, surgical resection, myeloablative consolidation with autologous stem cell transplantation, radiotherapy, and maintenance immunotherapy with anti-GD2 monoclonal antibodies, long-term survival for patients with high-risk neuroblastoma remains suboptimal, with contemporary studies reporting a 5-year survival rate of approximately 50% (3). + +Platinum-based compounds, including cisplatin and carboplatin, constitute essential components of all contemporary induction regimens for high-risk neuroblastoma, with alkylator and platinum doublets serving as the therapeutic backbone across multiple international cooperative group protocols (4). A recent systematic review of induction regimens published between 1995 and 2024 reported a median study-level end-induction response rate of 84.4%, yet the response rates have not improved substantially over the past three decades despite iterative refinements in dose intensity and schedule (4). Furthermore, the end-induction response has been established as a powerful independent predictor of the ultimate outcome, with patients who fail to achieve an objective response to frontline therapy facing a substantially elevated risk of subsequent relapse and death (5). While several recurrent genomic alterations, including MYCN amplification and 11q deletion, have been strongly associated with adverse outcomes, the molecular determinants governing differential responses to multiagent induction chemotherapy remain incompletely characterized (2). + +Chromatin remodeling protein alpha-thalassemia/mental retardation syndrome X-linked (ATRX) has emerged as a recurrently altered gene in high-risk neuroblastoma, with mutations occurring in a recognizable patient subgroup characterized by older age at diagnosis, mutual exclusivity with MYCN amplification, and a chemotherapy-resistant, slowly progressive disease course (6, 7). Critically, recent studies have demonstrated that the ATRX mutational spectrum in neuroblastoma is uniquely enriched for multiexon deletions, with approximately 70% of ATRX aberrations comprising such structural variants, and an estimated 75% of these deletions are predicted to be in-frame alterations that produce internally truncated protein products (8, 9). This distinctive mutational pattern, which appears nearly exclusive to neuroblastoma compared with other pediatric malignancies, suggests potential functional divergence between mutation subtypes (9). Preliminary evidence indicates that patients harboring ATRX missense mutations may exhibit slightly better overall survival than those with other ATRX mutation types and that 11q deletions cooccur more frequently with multiexon deletions than with other ATRX mutation types or wild-type tumors do (8, 9). Moreover, ATRX aberrations are strongly correlated with alternative lengthening of the telomere pathway, and both ATRX aberrations and TERT rearrangements are enriched in 11q-deleted high-risk tumors (10). + +Despite these advances in delineating the ATRX mutational landscape, a fundamental gap persists in understanding whether structurally distinct ATRX mutations are differentially associated with response to standard-of-care multiagent induction chemotherapy. Prior studies have largely aggregated all ATRX alterations into a single analytic category, potentially obscuring subtype-specific therapeutic vulnerabilities that could inform risk-adapted treatment strategies. Therefore, this multicenter retrospective cohort study was designed to evaluate whether ATRX mutation subtypes, defined according to the structural consequences of the underlying genomic alterations, were associated with differential objective response rates to multiagent induction chemotherapy, event-free survival, overall survival, and cumulative incidence of progression in children and adolescents with high-risk neuroblastoma. We hypothesized that, compared with missense mutations and wild-type ATRX status, large deletions and truncating mutations, particularly multiexon in-frame deletions, would be associated with primary resistance to conventional induction chemotherapy and poor clinical outcomes. + +This multicenter retrospective cohort study was conducted using data extracted from the electronic medical records (EMR) and molecular genetics databases of The Affiliated Taizhou People’s Hospital of Nanjing Medical University (Taizhou, China) and Children’s Hospital of Nanjing Medical University (Nanjing, China) between January 2017 and December 2022. To minimize selection bias, comprehensive molecular profiling, including next-generation sequencing (NGS) and chromosomal microarray analysis (CMA) encompassing the ATRX gene, was implemented as a routine baseline screening protocol for all newly diagnosed high-risk neuroblastoma patients across participating centers starting in 2017. Given the relative rarity of ATRX mutations in high-risk neuroblastoma, a subtype-based stratified random sampling strategy was employed to guarantee adequate statistical power across mutation subgroups while maintaining the clinical feasibility of a blinded independent central review (BICR) for imaging. Specifically, all eligible patients harboring pathogenic ATRX mutations were exhaustively identified from the institutional databases; subsequently, a concurrent cohort of ATRX wild-type (WT) patients was established by randomly selecting cases at a 1:2 ratio (mutant to WT) from the available WT pool using computer-generated random numbers. The study protocol was reviewed and approved by the Institutional Review Board of The Affiliated Taizhou People’s Hospital of Nanjing Medical University (IRB Approval No: KY-2023-045-01) and the respective ethics committees of all the participating centers. A waiver of informed consent was granted by the Institutional Review Board of The Affiliated Taizhou People’s Hospital of Nanjing Medical University because of the retrospective, noninterventional nature of the data analysis, and all procedures were conducted in strict accordance with the Declaration of Helsinki and local data privacy regulations. + +The study cohort was identified through a systematic query of the EMR systems and molecular genetics databases of the participating institutions using structured data filters, including relevant diagnostic codes for neuroblastoma and specific molecular testing logs. All potentially eligible patients diagnosed between the prespecified study dates were initially screened. Potential participants underwent a rigorous three-stage validation process: automated identification by clinical data warehouses on the basis of initial age and diagnostic criteria; manual review of medical, imaging, and genetic records by two independent pediatric oncologists using standardized electronic case report forms; and final adjudication by a senior pediatric oncologist in cases of diagnostic, molecular, or eligibility uncertainty. Any discrepancies encountered during the screening process were resolved through consensus discussion. + +The inclusion criteria included the following: (1) a definitive diagnosis of neuroblastoma confirmed by histopathology or bone marrow cytology (11); (2) a classification as “high-risk” disease according to the International Neuroblastoma Risk Group (INRG) staging system (12); (3) age strictly between 18 months and 18 years at the time of diagnosis (13), encompassing both pediatric and adolescent populations; (4) initiation of a protocol-based multiagent induction chemotherapy regimen, generally planned for 6–8 cycles and including cisplatin or carboplatin as one component, with either an evaluable end-of-induction assessment or a documented clinical event before completion of induction (14); (5) the availability of complete NGS and CMA or fluorescence in situ hybridization (FISH) data clearly defining the ATRX mutation status and critical segmental chromosomal aberrations (specifically 11q deletion), with the strict prerequisite that all molecular data were derived from primary biopsy or surgical resection specimens obtained at initial diagnosis (prechemotherapy) to rigorously exclude acquired mutations induced by therapeutic selection pressure; and (6) the availability of complete clinical baseline data and comprehensive long-term follow-up records encompassing either the first systemic imaging evaluation following induction therapy or documentation of an early clinical endpoint (e.g., disease progression or toxicity-related death). The exclusion criteria included any of the following: (1) receipt of a nonprotocol or unclassifiable induction regimen, or insufficient treatment records to reconstruct the administered regimen and treatment exposure; (2) missing genetic sequencing data or sequencing depth insufficient for accurate mutation profiling and subtyping; (3) loss to follow-up prior to the first postinduction disease evaluation without any documented clinical endpoint; and (4) concurrent primary malignant neoplasms, advanced severe congenital anomalies, or other life-limiting conditions expected to confound treatment efficacy and survival assessments. + +During the study period, patients received protocol-based multiagent induction chemotherapy according to the high-risk neuroblastoma treatment protocols adopted at the participating centers. The induction backbone generally consisted of alternating cyclophosphamide, doxorubicin, and vincristine (CAV) cycles and platinum–etoposide cycles, administered for 6–8 cycles before the end-of-induction response assessment. Cisplatin was the principal platinum agent, whereas carboplatin was used in selected patients according to the applicable institutional protocol or when cisplatin substitution was clinically required. Drug doses were calculated according to body surface area and were adjusted for body weight, organ function, and treatment-related toxicity in accordance with the relevant protocol. + +For each patient, the induction regimen, platinum agent, number of completed cycles, cumulative platinum dose, dose reductions, treatment delays, and reasons for premature discontinuation were extracted from the medical records. Because cisplatin or carboplatin was administered as part of a multiagent regimen rather than as monotherapy, treatment response was evaluated as response to the overall induction chemotherapy regimen and was not attributed specifically to the platinum component. + +Patients were stratified into three distinct molecular subgroups on the basis of the structural consequences of the identified ATRX alterations derived from next-generation sequencing breakpoint analysis, with variant pathogenicity rigorously evaluated in accordance with the American College of Medical Genetics and Genomics (ACMG) guidelines (15). The WT group comprised patients with no detectable pathogenic or likely pathogenic ATRX variants. The large deletion/truncated protein group included patients harboring massive structural variations, predominantly multiexon in-frame deletions (IFDs), as well as nonsense and frameshift mutations. In neuroblastoma, these characteristic multiexon IFDs frequently escape nonsense-mediated decay and are translated into abnormal, partially active truncated proteins rather than resulting in a complete loss of protein expression (16). The missense mutation group consisted of patients with pathogenic single-nucleotide variants that specifically disrupt critical functional regions, notably the ATRX-DNMT3-DNMT3L (ADD) domain or the helicase domain, thereby selectively impairing chromatin remodeling capabilities (17). + +Baseline demographic, clinical, and tumor-specific variables, including age at diagnosis, sex, primary tumor location, baseline serum lactate dehydrogenase (LDH) levels, histological categorization according to the International Neuroblastoma Pathology Classification (INPC) (11), and baseline functional status assessed via the Lansky play-performance scale (18), were extracted from the electronic medical records. Key prognostic molecular and staging variables incorporated into the analysis included MYCN amplification status, 11q deletion status, and International Neuroblastoma Risk Group (INRG) stage (12). To rigorously control for the heterogeneity of pharmacokinetic exposure and chemotherapy adherence inherent in multicenter real-world data, the induction phase “dose exposure and compliance” was introduced as a core categorical adjustment variable. This metric was strictly dichotomized into “adequate exposure,” defined as the successful administration of at least 80% of the planned cumulative platinum dose with a total treatment delay of 4 weeks or less secondary to adverse events, and “suboptimal exposure,” defined as the receipt of less than 80% of the intended dose or a cumulative treatment delay exceeding 4 weeks due to severe myelosuppression or organ toxicity. Furthermore, the actual median cumulative platinum dose, calculated in milligrams per square meter of body surface area (mg/m²), was extracted as a continuous variable to verify dose equivalence across subgroups. + +The primary efficacy endpoint was the objective response rate (ORR), defined as the proportion of patients achieving a complete response (CR) or partial response (PR) at the end-of-induction assessment. Response assessments were performed according to the revised International Neuroblastoma Response Criteria (INRC) (19). Imaging studies were independently reviewed by two senior pediatric radiologists who were blinded to ATRX status and clinical outcomes. Bone marrow and other nonimaging disease assessments were reviewed using the corresponding pathology and clinical records. Disagreements were resolved by consensus or adjudication by a third reviewer. + +Secondary endpoints included the primary induction-refractory rate, event-free survival (EFS), cumulative incidence of progression (CIP), and overall survival (OS). Primary induction-refractory disease was defined as MR, SD, or PD at the end-of-induction assessment, together with documented disease progression necessitating premature discontinuation or modification of the planned induction regimen. EFS was calculated from initiation of induction chemotherapy to the first occurrence of disease progression, relapse, a second malignant neoplasm, or death from any cause. OS was calculated from initiation of induction chemotherapy to death from any cause. For the CIP analysis, death without documented disease progression was treated as a competing event. + +Given the inherent rarity of large ATRX deletions and pathogenic variants in the target population, this study consolidated all pathogenic ATRX alterations into a pan-ATRX mutant (Mut) group, with exploratory subgroup analyses predefined on the basis of the predicted functional impact of the specific mutations. The sample size estimation was prospectively based on the primary efficacy endpoint of the objective response rate (ORR). On the basis of the outcomes reported in the HR-NBL1.5 trial, the anticipated ORR for high-risk neuroblastoma patients receiving standard platinum-based induction therapy was 70%, which served as the baseline expectation for the WT control group (14). Assuming an overall ATRX mutation frequency of 10% to 15%, a query of the participating multicenter databases was projected to yield approximately 45 eligible patients with ATRX mutations. Utilizing the predefined 1:2 subtype-based stratified random sampling strategy, approximately 90 WT patients were subsequently selected, resulting in a targeted total analytic cohort of 135 patients. Statistical power analyses indicated that detecting a clinically significant reduction in the ORR from 70% in the WT group to an anticipated 45% in the Mut group (corresponding to an approximate odds ratio of 0.35) with a two-sided alpha level of 0.05 would yield a statistical power of approximately 78%. Although this power was marginally below the conventional 80% threshold—a recognized constraint inherent to rare molecular subgroup analyses—the study was deemed adequately powered for initial real-world evaluation. To mitigate potential type II errors, statistical interpretation prioritized the estimation of effect sizes and their corresponding 95% confidence intervals over an exclusive reliance on null-hypothesis significance testing. + +All the statistical analyses were executed using R software (version 4.5.1, R Foundation for Statistical Computing), with a two-sided P value of less than 0.05 predefined as the threshold for statistical significance. To address baseline covariate missingness (<20%), multiple imputation using chained equations (MICE) was performed, generating five imputed datasets that were subsequently pooled for downstream weighting and analyses in strict accordance with Rubin’s rules. To account for baseline imbalances, an inverse probability of treatment weighting (IPTW) approach was implemented using generalized propensity scores derived from multinomial logistic regression. The weighting model incorporated critical baseline confounders, specifically age at diagnosis, MYCN amplification status, 11q deletion status, and the previously defined chemotherapy dose exposure and compliance categorical metric. Covariate balance across the three molecular subgroups was rigorously verified by calculating the standardized mean difference (SMD), with an SMD of less than 0.1 considered indicative of optimal equilibrium. Concurrently, the variance inflation factor (VIF) was evaluated both before and after IPTW adjustment to assess multicollinearity. Differences in the ORR and primary induction-refractory rates across the weighted cohorts were assessed using weighted chi-square tests. Survival outcomes, including EFS and OS, were estimated using the weighted Kaplan–Meier method and statistically compared utilizing weighted log-rank tests. To accurately assess the absolute risk of disease progression while accounting for competing events, the Fine–Gray competing risk regression model was employed to calculate and compare the CIP, wherein non-tumor-specific mortality (e.g., severe chemotherapy-related toxicity) was strictly treated as a competing risk event. To establish the independent prognostic value of the ATRX mutation subtypes, a multivariable Cox proportional hazards regression model was constructed within the original (unweighted) cohort, simultaneously adjusting for ATRX genotype, MYCN amplification, 11q deletion, age, and chemotherapy exposure. This model included an interaction term between ATRX and 11q deletion status to assess the independence of the prognostic effects of ATRX. The proportional hazards (PH) assumption for the multivariable Cox models was rigorously verified using Schoenfeld residuals and graphical diagnostics. Furthermore, to overcome the statistical instability (quasicomplete separation) caused by the extreme mutual exclusivity between ATRX mutations and MYCN amplification, an independent, focused subgroup analysis was conducted exclusively within the MYCN-nonamplified (MYCN-NA) pure subset. Within this refined cohort, multivariable logistic regression was utilized to additionally adjust for 11q deletion status, thereby isolating the pure biological effect of ATRX alterations from this highly collinear genomic event. Finally, forest plots were generated to visualize the consistency of the resistance phenotype across various clinical strata, particularly when the truncating and IFD subgroups were compared. The robustness of the core findings was validated through prespecified sensitivity analyses: (1) weight trimming, which involved the truncation of extreme propensity score weights at the 1st and 99th percentiles; and (2) a perfect compliance cohort analysis, which restricted the evaluation exclusively to patients who achieved adequate exposure, thereby reducing potential confounding from suboptimal chemotherapy delivery in the response assessment. + +A total of 138 eligible patients with high-risk neuroblastoma were included in the final analytic cohort, comprising 92 ATRX WT, 33 large deletion/truncating (inclusive of in-frame deletions), and 13 missense cases, successfully established via the predefined 1:2 subtype-based stratified random sampling strategy (Figure 1). In the original unweighted cohort, pronounced biological and demographic heterogeneity was observed among the three subgroups. Compared with the WT group, both mutant subgroups presented significantly older median ages at diagnosis (WT: 34.5; large deletion/truncating: 78.0; missense: 72.0 months; p < 0.001), substantially higher 11q deletion frequencies (23.9% vs. 72.7% and 53.8%; p < 0.001), and a near-complete absence of MYCN amplification (45.7% vs. 0.0% and 7.7%; p < 0.001). Following multiple imputation and IPTW, optimal equilibrium was achieved across all evaluated baseline covariates, including the Lansky Play Performance Scale, INPC categorization, and categorical dose exposure compliance, with all maximum standardized mean differences (Max SMDs) < 0.100 and VIFs consistently within safe thresholds (Table 1). Furthermore, the nonparametric comparison of cumulative platinum dose showed no statistically significant difference in delivered platinum exposure across the weighted WT, large deletion/truncating, and missense groups (395.0, 401.0, and 396.0 mg/m², respectively; p = 0.768), supporting the comparability of measured treatment exposure across the three groups. + +Flow diagram of subtype-based stratified random sampling, data imputation, and weighting. + +Efficacy outcomes, confirmed by the BICR according to the revised INRC guidelines, were formally compared across the molecular subgroups (Figure 2). In the IPTW-weighted pseudocohort, the large deletion/truncating group had a significantly lower ORR than the WT group (36.0% vs. 71.0%; P < 0.001) and a substantially higher primary induction-refractory rate (64.0% vs. 29.0%; P < 0.001). The missense mutation group demonstrated an intermediate ORR of 60.0%, which did not significantly differ from that of the WT cohort (P = 0.285) (Table 2). To rigorously eliminate the confounding quasicomplete separation induced by MYCN amplification, an independent analysis was conducted within the MYCN-NA subcohort (n = 95). Within this pure subset, the large deletion/truncating variant remained robustly associated with a diminished ORR (36.4%) relative to that of the WT group (70.0%). Subsequent multivariable logistic regression adjusted for 11q deletion status showed that large deletion/truncating ATRX alterations remained independently associated with a lower ORR in the MYCN-NA population (adjusted odds ratio = 0.28, 95% CI: 0.11–0.69; P = 0.006) (Table 2). + +Waterfall plot of the best objective tumor response confirmed by blinded independent central review (BICR). The vertical axis represents the maximum percentage change in the sum of the longest diameters of the target lesions from baseline. The bars are color-coded according to the ATRX mutation subgroup (WT, large deletion/truncating, and missense). The dashed reference lines indicate the thresholds for partial response (-30%) and progressive disease (+20%) as defined by the revised INRC guidelines. + +Survival outcomes were systematically evaluated within the IPTW-weighted pseudocohort. At a median follow-up of 42.5 months, compared with both the WT and missense cohorts, the large deletion/truncating group demonstrated significantly inferior survival trajectories. Specifically, the 3-year EFS rates were 18.5% for the large deletion/truncating group, 52.4% for the WT group, and 46.2% for the missense group (weighted log-rank P < 0.001). Consistent with the EFS findings, the 3-year OS rates were notably lower in patients harboring large deletions or truncating variants (28.4%) than in those in the WT (68.5%) and missense (61.0%) groups (weighted log-rank P < 0.001) (Figure 3). To rigorously account for the competing risk of non-tumor-specific mortality, particularly severe chemotherapy-induced toxicity prior to documented tumor progression, a Fine-Gray competing risk regression analysis was performed. This analysis revealed that the 3-year CIP was markedly greater in the large deletion/truncating group (72.6%) than in the WT group (35.2%; Gray’s test P < 0.001), with an estimated subdistribution hazard ratio (sHR) of 2.88 (95% CI: 1.92–4.32; P < 0.001) (Figure 4). + +Kaplan–Meier curves for event-free survival (EFS) and overall survival (OS) stratified by ATRX mutation status in the IPTW-weighted cohort. (A) Event-free survival and (B) overall survival. The solid lines represent the survival estimates for the wild-type (WT), large deletion/truncating, and missense groups. Shaded areas denote the corresponding 95% confidence intervals. Tick marks indicate censored patients. P values were calculated using the weighted log-rank test. + +Cumulative incidence of progression (CIP) curves adjusted by the fine–gray competing risk model in the IPTW-weighted cohort. The curves illustrate the absolute probability of disease progression over time across the three molecular subgroups. Nontumor-specific mortality (e.g., lethal treatment-related toxicity) was strictly treated as a competing risk event. P values were derived from Gray’s test. The subdistribution hazard ratio (sHR) was calculated to quantify the relative risk of progression for the large deletion/truncating group vs. the WT group. + +Subgroup analyses demonstrated that the association between large ATRX deletion/truncating variants and an inferior ORR remained consistent across all predefined clinical strata (Figure 5). Specifically, a negative impact was observed regardless of age at diagnosis (< 5 years: OR 0.26; ≥ 5 years: OR 0.29), metastatic burden (bone marrow/CNS: OR 0.25; other sites: OR 0.31), or platinum agent included in the induction regimen (cisplatin-based: OR 0.27; carboplatin-based: OR 0.29), with all interaction P values > 0.05. Notably, an exploratory internal subgroup analysis within the large deletion/truncating cohort revealed that the deleterious effect on treatment response was equivalent between multiexon IFDs (OR 0.27, 95% CI: 0.11–0.65) and nonsense/frameshift variants (OR 0.29, 95% CI: 0.12–0.72), supporting the functional consolidation of these molecular subtypes (P for interaction = 0.812). + +Forest plot of the association between large deletion/truncating ATRX mutations and the objective response rate to induction chemotherapy across predefined subgroups. The plot displays odds ratios (ORs) and 95% confidence intervals (CIs) for comparisons between the large deletion/truncating group and the WT group. In addition to clinical strata, including age, metastatic site, and the platinum agent included in the induction regimen, the plot includes an internal molecular comparison between multiexon IFDs and nonsense/frameshift variants. The vertical dashed line represents an OR of 1.0. P values for interactions indicate the consistency of the mutation effect across all evaluated subgroups. + +Two sensitivity analyses further confirmed the robustness of the primary findings (Supplementary Table 1). After weight trimming was applied to the IPTW model (excluding the 1st and 99th percentiles), the ORR remained significantly lower in the large deletion/truncating group than in the WT group (36.4% vs. 70.8%, P < 0.001). Similarly, in the “perfect compliance” cohort (n = 118), the ORR disparity persisted (37.5% in the large deletion/truncating group vs. 71.8% in the WT group, P < 0.001), indicating that the observed association was not explained by measured differences in chemotherapy delivery or extreme propensity weights. + +To determine the independent prognostic significance of the ATRX mutation subtypes, univariate and multivariable Cox proportional hazards regression analyses were performed using the original unweighted cohort (n = 138). In the multivariable model adjusted for MYCN amplification, 11q deletion, age at diagnosis, and chemotherapy dose exposure, large ATRX deletion/truncating variants were confirmed as independent risk factors for both inferior EFS (adjusted hazard ratio [aHR] = 3.42, 95% CI: 2.15–5.44, p < 0.001) and OS (aHR = 3.12, 95% CI: 1.88–5.18, p < 0.001) (Table 3). Notably, the interaction term between ATRX status and 11q deletion was not statistically significant (p = 0.428 for EFS; p = 0.512 for OS), indicating that the adverse prognostic impact of ATRX structural alterations is independent of 11q genomic status. Conversely, ATRX missense mutations were not significantly independently associated with survival outcomes (EFS: aHR = 1.24, p = 0.412; OS: aHR = 1.18, p = 0.584). The proportional hazards assumption was satisfied for all multivariable models on the basis of Schoenfeld residual testing. + +This multicenter retrospective cohort study delineated the prognostic and therapeutic implications of ATRX mutation subtypes in children and adolescents with high-risk neuroblastoma treated with platinum-containing multiagent induction chemotherapy. The principal finding was that the clinical associations of ATRX alterations varied according to mutation subtype: patients harboring large deletions or truncating mutations, including characteristic multiexon in-frame deletions, had substantially lower objective response rates, higher primary induction-refractory rates, and inferior event-free and overall survival compared with patients with ATRX wild-type tumors. In contrast, patients with ATRX missense mutations demonstrated intermediate outcomes that did not differ significantly from those of the wild-type cohort. These associations persisted after adjustment for age, MYCN amplification, 11q deletion status, and measured treatment exposure and were supported by sensitivity analyses restricted to patients with adequate treatment delivery. Because platinum compounds were administered together with other cytotoxic agents, these findings should be interpreted as evidence of resistance to the overall induction regimen rather than resistance specifically attributable to platinum agents. + +Previous investigations have established that ATRX alterations define a distinct high-risk neuroblastoma subtype characterized by older age at diagnosis, mutual exclusivity with MYCN amplification, and an indolent yet inexorably progressive disease course associated with poor long-term survival. Kurihara and colleagues initially reported that ATRX-altered tumors without MYCN amplification detected beyond 18 months of age had poor prognosis and chemoresistance (7). More recently, a systematic review by Gonzales-Céspedes and colleagues revealed that ATRX and TERT aberrations constitute key biomarkers for accurate prognostic stratification in high-risk neuroblastoma (3). Lorenzi and colleagues further characterized the ATRX-mutant subgroup as presenting with chemotherapy-resistant, slowly progressive disease and demonstrated that ATRX alterations upregulate inflammatory pathways and mediate macrophage infiltration, revealing previously unrecognized immunogenic properties of these tumors (6). However, the literature has produced conflicting results regarding the precise magnitude of the adverse prognostic effect attributable to ATRX mutations. This ambiguity stems largely from the historical practice of aggregating all ATRX mutations into a single analytic category without consideration of their diverse structural and functional consequences. Our study addresses this inconsistency by performing fine-grained molecular subtyping on the basis of the predicted functional impact of each ATRX alteration. A recent dissertation by van Gerven systematically characterized the ATRX mutational landscape in neuroblastoma and reported that patients with ATRX missense mutations demonstrated slightly better overall survival than those with other mutation types did, a finding that aligns closely with our observation that missense mutations confer a neutral prognostic effect (9). By stratifying patients into large deletion/truncating, missense, and wild-type subgroups, our analysis reconciles the apparent prognostic heterogeneity reported in prior studies, highlighting that structural alterations involving extensive genomic loss constitute the predominant drivers of the aggressive clinical phenotype in this context. + +The biological basis for the poor response to induction chemotherapy observed in the large deletion/truncating subgroup may reside in the unique molecular consequences of multiexon in-frame deletions that predominate in neuroblastoma. Unlike typical loss-of-function mutations that trigger nonsense-mediated mRNA decay and complete protein ablation, these characteristic multiexon deletions frequently escape nonsense-mediated decay and translate into internally deleted but partially functional truncated proteins. Recent genomic profiling studies have shown that 70% of ATRX aberrations in neuroblastoma are multiexon deletions, of which 75% are predicted to be in-frame and produce shortened protein products, a pattern that appears almost exclusive to neuroblastoma compared with other pediatric cancers (8, 9). Furthermore, these distinct aberrations produce divergent gene expression signatures; for instance, ATRX exon 2–10 multiexon deletion models exhibit reduced expression of genes involved in ribosome biogenesis and metabolism, confirming the existence of functionally distinct ATRX subtypes (9, 20). These aberrantly truncated proteins may exert dominant-negative effects that disrupt normal chromatin remodeling functions while retaining partial activity, thereby generating a unique cellular state characterized by defective DNA damage response signaling, altered replication fork dynamics, and aberrant telomere maintenance (21). ATRX plays a central role in depositing the histone variant H3.3 at repetitive genomic regions, including telomeres and pericentric heterochromatin, and its deficiency leads to telomere dysfunction, which activates the alternative lengthening of telomeres (ALT) pathway (17). Hartlieb and colleagues demonstrated that ALT-positive tumors display reduced ATRX/DAXX complex abundance and are molecularly distinct from other neuroblastoma subtypes (20), whereas Djos and colleagues corroborated a uniform correlation between ATRX aberrations and the ALT biomarker C-circles (10). Several agents used during induction, including platinum compounds, exert cytotoxic effects through DNA damage and replication stress. In ATRX-deficient cells, chronic replicative stress, ALT-mediated genomic instability, and a characteristically slow-cycling biological state may collectively reduce sensitivity to DNA-damaging multiagent chemotherapy. This altered cell cycle dynamics and the potential upregulation of damage tolerance pathways enable tumor cells to survive despite intensive induction therapy. Although ATRX deficiency may alter sensitivity to DNA-damaging agents, George and colleagues demonstrated that ATRX-mutant neuroblastoma cells harbor specific vulnerabilities in the DNA damage response network and exhibit selective sensitivity to PARP and ATM inhibition (22).while A recent study by Lorenzi and colleagues further revealed that ATRX in-frame deletions and loss-of-function mutations elicit differential chromatin accessibility responses, underscoring the functional divergence between mutation subtypes and their potential impact on therapeutic response (23). + +A critical methodological challenge in assessing the independent prognostic value of ATRX alterations lies in disentangling their effects from those of tightly linked genomic covariates. Our data confirmed that ATRX structural alterations cooccur at a high frequency with 11q deletion, with 72.7% of the large deletion/truncating group harboring concurrent 11q loss compared with 23.9% of wild-type patients. This enrichment pattern has been well documented, as 11q deletions are known to cooccur more frequently with ATRX multiexon deletions than with point mutations or wild-type tumors (8, 9). To rigorously address this collinearity, our multivariable Cox regression model included an interaction term between ATRX status and 11q deletion. The absence of a statistically significant interaction indicated that the adverse prognostic impact of ATRX structural alterations operates independently of 11q genomic status, establishing that ATRX disruption is an autonomous driver of poor outcome rather than a mere passenger event in 11q-deleted tumors. Furthermore, the near-complete mutual exclusivity between ATRX mutations and MYCN amplification observed in our cohort mirrors the incompatibility previously characterized in preclinical models. Zeineldin and colleagues demonstrated that elevated MYCN expression and ATRX mutations are incompatible in neuroblastoma, leading to metabolic reprogramming and lethal replicative stress (24). By restricting the analysis to the MYCN-nonamplified subcohort, we reduced confounding related to MYCN amplification and found that LDT ATRX alterations remained independently associated with a lower ORR to induction chemotherapy. + +The clinical implications of these findings are potentially important. Contemporary induction regimens for high-risk neuroblastoma use platinum compounds as components of multiagent chemotherapy. In our cohort, patients harboring large ATRX deletions or truncating mutations responded poorly to these conventional induction regimens, with an objective response rate of 36% and a primary induction-refractory rate of 64%. These findings support the incorporation of detailed ATRX mutation subtyping into baseline molecular profiling to facilitate the early identification of patients at high risk of inadequate induction response. However, because all patients received combination chemotherapy, the present study does not establish that omission or substitution of the platinum component would improve outcomes. Preclinical evidence suggests that ATRX-mutant neuroblastoma may have therapeutic vulnerabilities involving PARP, ATM, ATR, or WEE1-related pathways (21, 22, 25). Future prospective studies should therefore evaluate molecularly adapted or early response-guided induction strategies and determine whether modifying individual cytotoxic components or incorporating targeted agents improves outcomes in patients with ATRX structural alterations. + +Several limitations of this study warrant careful consideration. First, despite the implementation of multiple imputation, inverse probability of treatment weighting, and rigorous sensitivity analyses, the retrospective design inherently precludes the complete elimination of unmeasured residual confounding. Second, although our stratified random sampling strategy successfully generated a balanced analytic cohort, the absolute number of patients in the missense mutation group remained modest, raising the possibility of a type II error with respect to the neutral prognostic effect attributed to this subgroup. Validation in larger international collaborative cohorts is essential to confirm the generalizability of these findings. Third, the absence of paired in vitro or in vivo functional validation experiments within this study precludes the definitive mechanistic attribution of the observed resistance phenotype to specific molecular pathways. Fourth, although central reannotation of original raw sequencing data was not performed, the use of standardized ACMG criteria for variant interpretation and BICR for imaging endpoints significantly mitigated this concern. Fifth, the long-term follow-up period, while adequate for assessing induction response and early survival outcomes, may not fully capture the chronic but progressive disease trajectory characteristic of ATRX-mutant neuroblastoma, and extended surveillance is warranted. Sixth, although we adjusted for the cumulative platinum dose, treatment delays, and categorical treatment exposure, detailed agent-specific relative dose intensity was not uniformly available. Moreover, because platinum compounds were administered exclusively as components of multiagent induction regimens, the present study could not separate their contribution from that of the accompanying cytotoxic agents. Consequently, the observed phenotype should be interpreted as resistance to conventional induction chemotherapy rather than as platinum-specific resistance. Finally, the study population was derived from institutions with an established molecular profiling infrastructure, which may limit its applicability to settings with more limited genomic testing capabilities. + +This study showed that large ATRX deletions and truncating mutations, including multiexon in-frame deletions, identify an ultrahigh-risk molecular subtype of neuroblastoma characterized by poor response to conventional multiagent induction chemotherapy, accelerated disease progression, and markedly inferior survival outcomes. These associations were independent of age, MYCN amplification, 11q deletion status, and measured treatment exposure. Detailed ATRX mutation subtyping may therefore improve early risk stratification and facilitate the prospective evaluation of molecularly adapted or response-guided induction strategies. Because platinum compounds were administered as components of combination regimens, further treatment-specific and functional studies are required to determine whether the resistance phenotype is selectively attributable to platinum agents. + +## Table 1 Comparison of baseline clinical and treatment characteristics among the three molecular subgroups before and after inverse probability of treatment weighting (IPTW). + +| Characteristic [rowspan=2] | Unweighted Cohort [colspan=5] | Weighted Cohort (Post-IPTW) [colspan=6] | +| WT (n = 92) | Large Deletion/Truncating (n = 33) | Missense (n = 13) | Statistic | p value | WT | Large Deletion/Truncating | Missense | Max SMD | Statistic | p value | +| Age at diagnosis (months), median (IQR) | 34.5 (22.0-48.5) | 78.0 (62.0-98.0) | 72.0 (58.0-95.0) | H = 41.52 | < 0.001 | 48.0 (32.0-68.0) | 48.5 (33.5-67.0) | 49.0 (33.0-66.0) | 0.025 | H = 0.12 | 0.941 | +| Sex, n (%) [colspan=12] | +| Male | 52 (56.5) | 20 (60.6) | 8 (61.5) | χ² = 0.28 | 0.869 | 58.20% | 57.80% | 58.00% | 0.012 | χ² = 0.01 | 0.995 | +| Female | 40 (43.5) | 13 (39.4) | 5 (38.5) | | | 41.80% | 42.20% | 42.00% | | | | +| Primary tumor location, n (%) [colspan=12] | +| Adrenal | 63 (68.5) | 22 (66.7) | 8 (61.5) | χ² = 0.32 | 0.852 | 66.80% | 67.20% | 67.00% | 0.014 | χ² = 0.01 | 0.995 | +| Extra-adrenal | 29 (31.5) | 11 (33.3) | 5 (38.5) | | | 33.20% | 32.80% | 33.00% | | | | +| Baseline LDH (U/L), median (IQR) | 850 (510-1280) | 795 (485-1160) | 810 (475-1180) | H = 0.85 | 0.653 | 825 (490-1210) | 828 (500-1230) | 830 (505-1240) | 0.015 | H = 0.09 | 0.955 | +| INPC categorization, n (%) [colspan=12] | +| Favorable | 15 (16.3) | 4 (12.1) | 2 (15.4) | χ² = 0.36 | 0.835 | 15.40% | 15.80% | 15.50% | 0.011 | χ² = 0.01 | 0.995 | +| Unfavorable | 77 (83.7) | 29 (87.9) | 11 (84.6) | | | 84.60% | 84.20% | 84.50% | | | | +| Lansky play-performance scale, median (IQR) | 80 (70-90) | 80 (70-90) | 80 (70-90) | H = 0.15 | 0.927 | 80 (70-90) | 80 (70-90) | 80 (70-90) | 0 | H = 0.00 | 1 | +| MYCN amplification, n (%) [colspan=12] | +| Yes | 42 (45.7) | 0 (0.0) | 1 (7.7) | χ² = 25.85 | < 0.001 | 30.50% | 30.20% | 30.80% | 0.015 | χ² = 0.01 | 0.995 | +| No | 50 (54.3) | 33 (100.0) | 12 (92.3) | | | 69.50% | 69.80% | 69.20% | | | | +| 11q deletion, n (%) [colspan=12] | +| Yes | 22 (23.9) | 24 (72.7) | 7 (53.8) | χ² = 24.12 | < 0.001 | 38.60% | 39.20% | 38.90% | 0.014 | χ² = 0.01 | 0.995 | +| No | 70 (76.1) | 9 (27.3) | 6 (46.2) | | | 61.40% | 60.80% | 61.10% | | | | +| INRG stage, n (%) [colspan=12] | +| L2 | 6 (6.5) | 3 (9.1) | 1 (7.7) | χ² = 0.31 | 0.856 | 7.30% | 7.50% | 7.40% | 0.009 | χ² = 0.00 | 1 | +| M | 86 (93.5) | 30 (90.9) | 12 (92.3) | | | 92.70% | 92.50% | 92.60% | | | | +| Dose exposure and compliance, n (%) [colspan=12] | +| Adequate exposure | 79 (85.9) | 28 (84.8) | 11 (84.6) | χ²= 0.04 | 0.978 | 85.80% | 86.20% | 86.00% | 0.011 | χ² = 0.01 | 0.995 | +| Suboptimal exposure | 13 (14.1) | 5 (15.2) | 2 (15.4) | | | 14.20% | 13.80% | 14.00% | | | | +| Cumulative platinum dose (mg/m²), median (IQR) | 395.0 (380.0-420.0) | 401.0 (375.0-418.0) | 396.0 (380.0-415.0) | H = 0.62 | 0.733 | 395.0 (380.0-415.0) | 401.0 (378.0-418.0) | 392.0 (375.0-410.0) | 0.024 | H = 0.54 | 0.768 | + +WT, wild-type; IQR, interquartile range; Max SMD, maximum standardized mean difference among all pairwise comparisons; LDH, lactate dehydrogenase; INPC, International Neuroblastoma Pathology Classification; INRG, International Neuroblastoma Risk Group. The “Statistic” column displays the Kruskal–Wallis H statistic (H) for continuous variables and the chi-square value (χ²) derived from Pearson’s chi-square test (or Fisher’s exact test, as appropriate) for categorical variables. In the weighted cohort, the frequencies are presented as weighted percentages to reflect the adjusted pseudopopulation distribution. A Max SMD value of < 0.100 indicates optimal covariate balance across all three subgroups. + +## Table 2 Associations between ATRX mutation status and response to induction chemotherapy in the IPTW-weighted cohort and the MYCN-nonamplified subcohort. + +| Clinical Response (Revised INRC) | WT | Large Deletion/Truncating | Missense | Statistic | P value | +| IPTW-Weighted Cohort [colspan=6] | +| Complete Response (CR) | 24.50% | 9.50% | 15.00% | | | +| Partial Response (PR) | 46.50% | 26.50% | 45.00% | | | +| Minor Response (MR) | 8.00% | 15.00% | 12.00% | | | +| Stable Disease (SD) | 13.00% | 25.00% | 18.00% | | | +| Progressive Disease (PD) | 8.00% | 24.00% | 10.00% | | | +| Objective Response Rate (ORR) | 71.00% | 36.00% | 60.00% | χ² = 15.82 | < 0.001 | +| WT vs. Large Deletion/Truncating | | | | χ² = 14.65 | < 0.001 | +| WT vs. Missense | | | | χ² = 1.14 | 0.285 | +| Primary Induction-Refractory Rate* | 29.00% | 64.00% | 40.00% | χ² = 14.95 | < 0.001 | +| MYCN-NA Subcohort (Unweighted, n=95) | (n = 50) | (n = 33) | (n = 12) | | | +| Complete Response (CR), n (%) | 12 (24.0) | 3 (9.1) | 2 (16.7) | | | +| Partial Response (PR), n (%) | 23 (46.0) | 9 (27.3) | 5 (41.7) | | | +| Minor Response (MR), n (%) | 4 (8.0) | 5 (15.2) | 1 (8.3) | | | +| Stable Disease (SD), n (%) | 7 (14.0) | 8 (24.2) | 3 (25.0) | | | +| Progressive Disease (PD), n (%) | 4 (8.0) | 8 (24.2) | 1 (8.3) | | | +| Objective Response Rate (ORR) | 35 (70.0) | 12 (36.4) | 7 (58.3) | χ² = 10.51 | 0.005 | +| WT vs. Large Deletion/Truncating | | | | χ² = 9.87 | 0.002 | +| WT vs. Missense | | | | χ² = 0.62 | 0.431 | + +WT, wild-type; INRC, International Neuroblastoma Response Criteria; IPTW, inverse probability of treatment weighting; MYCN-NA, MYCN-nonamplified. In the IPTW-weighted cohort, the data are presented as weighted percentages. The “Statistic” column displays the chi-square value (χ²) derived from the weighted chi-square test (for the IPTW cohort) or Pearson’s chi-square/Fisher’s exact test (for the MYCN-NA subcohort). The objective response rate (ORR) encompasses CR and PR. *Primary induction-refractory disease comprised MR, SD, or PD at the end-of-induction assessment, together with documented disease progression necessitating premature discontinuation or modification of the planned induction regimen. + +## Table 3 Univariate and multivariate Cox regression analyses of prognostic factors for EFS and OS in the original cohort (n = 138). + +| Variable [rowspan=2] | Univariate EFS [colspan=2] | Multivariable EFS [colspan=2] | Multivariable OS [colspan=2] | +| HR (95% CI) | p value | aHR (95% CI) | p value | aHR (95% CI) | p value | +| ATRX Status [colspan=7] | +| Wild-type | Ref. | -- | Ref. | -- | Ref. | -- | +| Large Deletion/Truncating | 4.15 (2.85-6.04) | < 0.001 | 3.42 (2.15-5.44) | < 0.001 | 3.12 (1.88-5.18) | < 0.001 | +| Missense | 1.35 (0.72-2.54) | 0.352 | 1.24 (0.65-2.36) | 0.412 | 1.18 (0.58-2.42) | 0.584 | +| MYCN status [colspan=7] | +| Nonamplified | Ref. | -- | Ref. | -- | Ref. | -- | +| Amplified | 5.12 (3.48-7.52) | < 0.001 | 4.45 (2.95-6.72) | < 0.001 | 4.05 (2.55-6.42) | < 0.001 | +| 11q deletion [colspan=7] | +| No | Ref. | -- | Ref. | -- | Ref. | -- | +| Yes | 2.58 (1.75-3.81) | < 0.001 | 1.42 (0.92-2.18) | 0.114 | 1.35 (0.85-2.14) | 0.205 | +| Age at diagnosis [colspan=7] | +| < 5 years | Ref. | -- | Ref. | -- | Ref. | -- | +| ≥ 5 years | 2.84 (1.92-4.20) | < 0.001 | 1.55 (0.98-2.45) | 0.062 | 1.48 (0.88-2.48) | 0.142 | +| Dose exposure [colspan=7] | +| Adequate | Ref. | -- | Ref. | -- | Ref. | -- | +| Suboptimal | 1.95 (1.25-3.04) | 0.003 | 1.28 (0.82-2.02) | 0.284 | 1.22 (0.75-1.98) | 0.426 | + +EFS, event-free survival; OS, overall survival; HR, hazard ratio; aHR, adjusted hazard ratio; CI, confidence interval; Ref., reference category. The multivariable model also included an interaction term between ATRX status and 11q deletion (interaction p = 0.428 for EFS, p = 0.512 for OS), which did not reach statistical significance. Chemotherapy dose exposure was categorized on the basis of the 80% dose and 4-week delay criteria. \ No newline at end of file