Genomic landscape of Ewing sarcoma defines an aggressive subtype with co-association of STAG2 and TP53 mutations

Authors

Franck Tirode

Didier Surdez

Xiaotu Ma

Matthew Parker

Marie Cécile Le Deley

Armita Bahrami

Zhaojie Zhang

Eve Lapouble

Sandrine Grossetête-Lalami

Michael Rusch

Stéphanie Reynaud

Thomas Rio-Frio

Erin Hedlund

Gang Wu

Xiang Chen

Gaelle Pierron

Odile Oberlin

Sakina Zaidi

Gordon Lemmon

Pankaj Gupta

Bhavin Vadodaria

John Easton

Marta Gut

Li Ding

Elaine R. Mardis

Richard K. Wilson

Sheila Shurtleff

Valérie Laurence

Jean Michon

Perrine Marec-Bérard

Ivo Gut

James Downing

Michael Dyer

Jinghui Zhang

Olivier Delattre

Doi

10.1158/2159-8290.CD-14-0622

PMID: 25223734 · DOI: 10.1158/2159-8290.CD-14-0622 · Journal: Cancer Discovery (2014)

TL;DR

Tirode and colleagues performed whole-genome sequencing of 112 Ewing sarcoma tumors with matched germline DNA and extended the analysis with targeted sequencing in 199 additional cases (299-patient combined clinical cohort). Beyond the obligate EWSR1–ETS fusion, Ewing sarcoma has a low background mutation rate (median 10 coding mutations/tumor). The most frequent somatic point mutations were in STAG2 (17% in discovery), CDKN2A deletions (12%), TP53 (7%), and epigenetic regulators EZH2, BCOR, ZMYM3 (2.7% each). STAG2 and CDKN2A alterations were mutually exclusive, while STAG2 and TP53 mutations were significantly co-associated and together defined an aggressive subtype with markedly worse overall survival. Subclonal STAG2 mutations at diagnosis could expand to dominate the clonal architecture at relapse.

Cohort & data

  • Discovery cohort: 112 Ewing sarcoma tumors with matched germline DNA, whole-genome sequenced on Illumina HiSeq2000 (median tumor depth 35X, germline 25X). 80% of tumors had >70% tumor purity.
  • Follow-up cohort: 199 additional French Ewing sarcoma patients treated under the EuroEwing99 protocol, profiled by targeted capture sequencing of STAG2 and TP53.
  • Combined clinical cohort: 299 patients with overall-survival data.
  • Cell line panel: 19 Ewing sarcoma cell lines (A673, RD-ES, SK-ES-1, SK-N-MC, MHH-ES1, TC-71, EW-1/3/7/16/18/22/23, STA-ET-1/3/8, MIC, ORS, POE).
  • Cancer type: Ewing sarcoma (ES).
  • Dataset: es_iocurie_2014. Raw data deposited in EGA under accessions EGAS00001000855 (Institut Curie) and EGAS00001000839 (St. Jude).
  • Assays / methods: whole-genome sequencing, targeted DNA sequencing, CREST for structural-variant calling, CONSERTING for copy-number analysis, BWA alignment to GRCh37-lite, genome-MuSiC SMG test, Sanger sequencing for CDKN2A confirmation, quantitative RT-PCR for CDKN2A status and fusion detection, and immunohistochemistry for STAG2 protein expression.

Key findings

  • Low mutational burden. Median 319 SNVs per genome (range 13–1747) and a median of 10 coding variants per tumor; missense:silent ratio of 2.4. Background mutation rate 2.4×10⁻⁷ per base (range 8.0×10⁻⁹ to 1.4×10⁻⁶) PMID:25223734.
  • Age vs. mutation count. Positive correlation between age at diagnosis and SNV count (r² = 0.42, P = 2.7×10⁻⁵); patients >20 years had significantly more SNVs (P = 0.001, Mann-Whitney) PMID:25223734.
  • Copy-number landscape. Recurrent gains: whole chr 8 (47%), whole chr 12 (21%), chr 1q (18%). Recurrent losses: chr 16q (17%), CDKN2A on 9p (12%). Chr 1q gain and chr 16q loss were tightly co-associated (P = 10⁻⁸, Fisher) and each conferred shorter survival (P = 2×10⁻⁵ and P = 0.0037, log-rank) PMID:25223734.
  • Chromothripsis in 5/112 cases, including 3 EWSR1–ERG-fused cases with chr 21/22 chromothripsis PMID:25223734.
  • STAG2 is the most frequently mutated coding gene in the discovery cohort: 19/112 (17%) — 6 nonsense, 10 frameshift indels, 1 missense, 1 splice-site, 1 exon-22 duplication. STAG2 mutation correlated with elevated structural-variant count (P = 0.006, Mann-Whitney) but not SNV/indel count PMID:25223734.
  • TP53 mutated in 8/112 (7%), all but one missense and all reported in COSMIC PMID:25223734.
  • Recurrent epigenetic-regulator mutations. EZH2 (3 missense, all in the SET domain — Y646F, Y646H, A682G), BCOR (3 cases, 1 missense + 1 frameshift + 1 116-kb intragenic deletion), ZMYM3 (3 cases: 2 frameshift indels + 1 17-kb intragenic deletion); each 2.7%. Additional novel mutations in SETD2, KMT2D (MLL2), KMT2C (MLL3), and PRDM9. Epigenetic-regulator mutations seen in 17/112 (15.2%) tumors overall PMID:25223734.
  • MuSiC SMG test identified STAG2, TP53, and EZH2 as the only significantly mutated genes PMID:25223734.
  • Mutual exclusivity of STAG2 and CDKN2A. In primary tumors (Fig. 1) and a 19-cell-line panel — STAG2 mutated in 9/19 and CDKN2A deleted in 6/19, with no overlap. Combined tumor + cell line data: WT/WT 221, STAG2-only 46, CDKN2A-only 49, double-altered just 2 (Fisher P = 0.0076) PMID:25223734.
  • STAG2 and TP53 co-association in the 299-patient cohort. 41 patients (13.2%) STAG2-mutated; 16 (5.2%) TP53-mutated. STAG2 and TP53 mutations were significantly co-occurring (P = 2.4×10⁻⁴, Fisher) PMID:25223734.
  • Prognostic synergy. STAG2-mutant and TP53-mutant tumors had the worst overall survival; STAG2 or TP53 alone did not significantly reduce survival. Patients with neither mutation had the best survival PMID:25223734.
  • Subclonal STAG2 expansion at relapse. 7 diagnostic samples showed subclonal STAG2 mutations (mutant allele frequency <0.25 despite high tumor purity). In paired primary/relapse IHC of 21 cases, 3 showed loss of STAG2-positive cells at relapse; in two of these (SJEWS001303, SJEWS014721), loss-of-function STAG2 mutations expanded from undetectable/subclonal at diagnosis to high allelic fraction at relapse PMID:25223734.
  • All Ewing sarcoma cell lines harbored at least one STAG2, TP53, or CDKN2A lesion, and TP53 mutation frequency in cell lines was much higher than in primary tumors PMID:25223734.

Genes & alterations

  • STAG2 — loss-of-function mutations (nonsense, frameshift indels, splice-site, exon-22 duplication) in 17% of discovery WGS cohort and 13.2% of the 299-patient combined cohort; correlated with structural-variant burden; mutually exclusive with CDKN2A deletion; co-occurs with TP53 mutation and defines aggressive subtype; can be subclonal at diagnosis and expand at relapse PMID:25223734.
  • TP53 — 7% missense mutations in WGS cohort, 5.2% in extended cohort; mostly COSMIC-described; cooperates with STAG2 mutation for worst prognosis PMID:25223734.
  • CDKN2A — focal 9p21 deletions in 12% of tumors; mutually exclusive with STAG2 mutation; methylation is rare in this tumor type, so STAG2 may regulate CDKN2A epigenetically via CTCF-cohesin insulation PMID:25223734.
  • EZH2 — recurrent SET-domain missense mutations Y646F, Y646H, A682G (3/112), known gain-of-function alleles in B-cell lymphoma PMID:25223734.
  • BCOR — 3/112 cases: S1083I missense, M1259fs frameshift, and a 116-kb intragenic deletion PMID:25223734.
  • ZMYM3 — 3/112 cases: two L82fs frameshift indels and one 17-kb intragenic deletion PMID:25223734.
  • EWSR1 — obligate fusion partner (EWSR1–FLI1 in 101 cases, EWSR1–ERG in 9, EWSR1–ETV1 in 1); two novel somatic missense mutations also observed PMID:25223734.
  • FLI1, ERG, ETV1, FEV — ETS-family fusion partners of EWSR1 PMID:25223734.
  • TAF15, FUS — alternative TET-family N-terminal partners in rare cases PMID:25223734.
  • SETD2, KMT2D (MLL2), KMT2C (MLL3), PRDM9 — novel somatic mutations in chromatin/methyltransferase regulators PMID:25223734.
  • STAG1, CTCF — mechanistically discussed: STAG1 is the paralog of STAG2 in the cohesin complex; STAG2 interacts with CTCF to regulate chromatin insulation at loci including CDKN2A PMID:25223734.

Clinical implications

  • Combined STAG2 + TP53 mutation status defines a high-risk Ewing sarcoma subtype with significantly inferior overall survival in 299 patients; either alone is not significantly prognostic in this cohort PMID:25223734.
  • Chr 1q gain and chr 16q loss are independently prognostic, conferring shorter survival; they almost always co-occur via the unbalanced t(1;16) derivative chromosome PMID:25223734.
  • Higher SNV/indel burden (top tertile) and outlier-high SV counts associate with shorter survival, the latter dramatically so (P = 0.003, log-rank) PMID:25223734.
  • Subclonal STAG2 mutations at diagnosis can drive relapse, supporting STAG2 immunostaining or mutation tracking as a potential clonal-evolution biomarker PMID:25223734.
  • Mutual exclusivity of STAG2 and CDKN2A suggests partial functional redundancy and may inform stratification for cohesin- or CDK4/6-targeted therapeutic strategies in Ewing sarcoma PMID:25223734.

Limitations & open questions

  • The authors note their findings on the CDKN2A/STAG2 mutual exclusion differ from the contemporaneous Brohl et al. report; the discrepancy is attributed to differences in technique (WGS vs exome) and cohort size, and requires further investigation PMID:25223734.
  • The positive correlation between STAG2 mutation and structural-variant burden is confounded by the STAG2–TP53 co-association: when only one of the two is mutated, the SV correlation is no longer significant PMID:25223734.
  • Whether STAG2 directly regulates the CDKN2A locus via cohesin–CTCF insulation in Ewing sarcoma cells remains hypothesized but not experimentally proven PMID:25223734.
  • The clinical/therapeutic actionability of cohesin loss in Ewing sarcoma — whether STAG2-deficient subclones expand at relapse because of selection by chemotherapy or simply outgrowth dynamics — is unresolved PMID:25223734.
  • The unbalanced t(1;16) translocation was inferred but never directly resolved by WGS, attributed to centromeric-repeat breakpoints PMID:25223734.

Citations from this paper used in the wiki

  • “the most common somatic mutations were detected in STAG2 (17%), CDKN2A (12%), TP53 (7%), EZH2, BCOR, and ZMYM3 (2.7% each). Strikingly, STAG2 mutations and CDKN2A deletions were mutually exclusive” — Abstract.
  • “In an expanded cohort of 299 patients with clinical data, we discovered that STAG2 and TP53 mutations are often concurrent and are associated with poor outcome” — Abstract.
  • “In our cohort, STAG2 and TP53 mutations were significantly co-associated (P = 2.4×10⁻⁴, Fisher exact test)” — Results: STAG2/TP53 co-association.
  • “Patients with neither STAG2 nor TP53 mutations had the highest probability of survival, and patients whose tumors carried mutations in both genes had the worst outcome” — Results.
  • “When compiling all our data (299 tumors and 19 cell lines), the overlap between STAG2 and CDKN2A genetic lesions was much lower than expected by chance (Fisher’s test: 0.0076, STAG2/CDKN2A, WT/WT: 221, WT/Mut: 49, Mut/WT: 46, Mut/Mut: 2)” — Results.
  • “loss of function STAG2 mutations were detected at relapse with high allelic fractions but were either not detected (SJEWS001303) or detected at a subclonal level at diagnosis (SJEWS014721)” — Results: subclonal expansion.
  • “All three EZH2 mutations were missense mutations within the SET domain (Y646F, Y646H, and A682G according to NM_004456)” — Results.
  • “Data are available in the European Genome-phenome Archive with the study accession number: EGAS00001000855 (Institut Curie cohort) and EGAS00001000839 (St. Jude cohort)” — Methods.

This page was processed by crosslinker on 2026-05-12.