Somatic ERCC2 mutations correlate with cisplatin sensitivity in muscle-invasive urothelial carcinoma

Authors

Van Allen EM

Mouw KW

Kim P

Iyer G

Wagle N

Al-Ahmadie H

Zhu C

Ostrovnaya I

Kryukov GV

O’Connor KW

Sfakianos J

Garcia-Grossman I

Kim J

Guancial EA

Bambury R

Bahl S

Gupta N

Farlow D

Qu A

Signoretti S

Barletta JA

Reuter V

Boehm J

Lawrence M

Getz G

Kantoff P

Bochner BH

Choueiri TK

Bajorin DF

Solit DB

Gabriel S

D’Andrea A

Garraway LA

Rosenberg JE

Doi

10.1158/2159-8290.CD-14-0623

PMID: 25096233 · DOI: 10.1158/2159-8290.CD-14-0623 · Journal: Cancer Discovery (2014)

TL;DR

Whole-exome sequencing of pre-treatment tumors from 50 patients with muscle-invasive urothelial carcinoma treated with neoadjuvant cisplatin-based chemotherapy followed by cystectomy (25 pT0/pTis “responders” vs. 25 pT2+ “non-responders”) identified somatic ERCC2 mutations as the only significantly enriched alteration in responders (9/25 responders vs. 0/25 non-responders; q = 0.007). All identified ERCC2 mutations clustered within helicase domains, failed to rescue cisplatin or UV sensitivity in an ERCC2-deficient cell line, and were associated with elevated background mutation rate, supporting a loss-of-function effect on nucleotide excision repair that confers cisplatin sensitivity.

Cohort & data

  • 50 patients with muscle-invasive urothelial carcinoma (BLCA) treated with neoadjuvant cisplatin-based chemotherapy followed by cystectomy (blca_dfarber_mskcc_2014).
  • 25 “responders” (no residual invasive disease, pT0/pTis) and 25 “non-responders” (residual ≥pT2 disease), accrued under Dana-Farber protocols 02-021 / 11-334 and MSKCC protocols 89-076 / 09-025.
  • Pre-treatment tumor and paired germline DNA sequenced by whole-exome-seq (SureSelect v2 Exome bait, Illumina HiSeq), mean target coverage 121× tumor / 130× germline; aligned to hg19 via Picard/Firehose pipeline.
  • Variants called with mutect (SNVs) and indelocator (indels), annotated with oncotator; cohort-level significance from mutsig (MutSigCV).
  • Orthogonal validation of candidate variants via Fluidigm Access Array amplicon resequencing on MiSeq (35/50 cases).
  • Comparison cohorts: unselected TCGA bladder cancer (n=130) (blca_tcga_pub) and a Chinese cohort (Guo et al., n=99).

Key findings

  • Median mutation rate was significantly higher in responders than non-responders: 9.7 vs 4.4 mutations/Mb (P = 0.0003; Mann-Whitney).
  • MutSigCV nominated four significantly altered genes across the full cohort consistent with prior bladder cancer studies: TP53, RB1, KDM6A, and ARID1A.
  • Nine non-synonymous somatic ERCC2 mutations were observed, all in cisplatin responders (9/25, 36%) and none in non-responders (P < 0.001; Fisher’s exact). After multiple-hypothesis correction, ERCC2 was the only gene significantly enriched in responders (q = 0.007; Benjamini-Hochberg).
  • The ERCC2 enrichment remained significant after adjusting for the higher overall mutation rate in responders (P = 0.04; binomial test).
  • ERCC2-mutant tumors had a significantly elevated median background mutation rate (15.5 vs 5.1 mutations/Mb in ERCC2-WT tumors; P = 0.01; Mann-Whitney), consistent with a DNA-repair defect.
  • The 36% responder ERCC2 mutation frequency was significantly higher than the ~12% rate in unselected TCGA bladder cancers (P < 0.001; binomial).
  • All identified ERCC2 mutations occurred at highly conserved residues within or adjacent to the helicase domains — mirroring the distribution of germline XP-D and XP/CS variants.
  • In an ERCC2-deficient XP-D patient-derived fibroblast line (GM08207), wild-type ERCC2 rescued cisplatin sensitivity (IC50 significantly higher than parent; P < 0.0001; ANOVA), but none of the five tested patient mutants rescued cisplatin or UV sensitivity.
  • Following cisplatin exposure, ERCC2-WT-complemented cells showed significantly fewer chromosomal aberrations than the ERCC2-deficient parent (P = 0.03; ANOVA), whereas the ERCC2-mutant cell lines showed no rescue of chromosomal stability (P > 0.5).
  • ERCC2 mutation status was not prognostic for overall survival in the TCGA bladder cohort (P = 0.77; log-rank), arguing against a confound from indolent disease.
  • In 7/9 ERCC2-mutant cases the variant allelic fraction was < 0.5, suggesting heterozygous mutation with retained WT allele — a haploinsufficient or dominant-negative model rather than classical biallelic loss.
  • In two responder tumors lacking ERCC2 mutations, truncating somatic mutations were observed in BRCA1 and BRCA2 (vs zero non-synonymous BRCA1/2 mutations in non-responders), consistent with the known platinum sensitivity of homologous-recombination-deficient tumors.
  • Survey of TCGA across 19 tumor types (n = 4,429) showed somatic ERCC2 mutations at low frequencies (< 4%) in 11 additional tumor types.

Genes & alterations

  • ERCC2 — nine non-synonymous somatic mutations in 9/25 (36%) cisplatin responders, all clustered in helicase domains; functional assays (cisplatin/UV survival, chromosomal aberration) show loss-of-function; proposed predictive biomarker for cisplatin sensitivity in muscle-invasive urothelial carcinoma.
  • TP53 — significantly mutated across the full cohort (MutSigCV), consistent with prior bladder cancer studies; not differentially enriched between responders and non-responders.
  • RB1 — significantly mutated across the full cohort (MutSigCV).
  • KDM6A — significantly mutated across the full cohort (MutSigCV).
  • ARID1A — significantly mutated across the full cohort (MutSigCV).
  • BRCA1 — truncating somatic mutation in one ERCC2-WT cisplatin responder; absent in non-responders.
  • BRCA2 — truncating somatic mutation in one ERCC2-WT cisplatin responder; absent in non-responders.

Clinical implications

  • Somatic ERCC2 mutation status is proposed as a candidate predictive biomarker to select muscle-invasive urothelial carcinoma patients most likely to benefit from neoadjuvant cisplatin-based chemotherapy, given that 100% of ERCC2-mutant tumors in this cohort responded.
  • The authors explicitly state that the data “should not yet be used to justify avoiding cisplatin-based treatment in ERCC2 WT patients” — prospective validation is required.
  • Because ~50% of bladder cancer patients are not cisplatin-eligible due to comorbidity, the authors suggest carboplatin-based regimens may warrant study in ERCC2-mutant tumors that are cisplatin-ineligible.
  • ERCC2 mutation does not appear prognostic in the absence of cisplatin therapy (no OS difference in TCGA cohort), so the signal is genuinely predictive rather than reflecting indolent biology.

Limitations & open questions

  • Small cohort (n = 50) with extreme-phenotype case-control design; signal requires independent prospective validation.
  • Multiple cisplatin-based regimens were used (cisplatin was the only common agent across all patients); contribution of non-cisplatin agents to response is unresolved.
  • 7/9 ERCC2 mutations were heterozygous (allelic fraction < 0.5); the mechanism (haploinsufficiency vs dominant-negative, as described for the yeast Rad3 homolog) is not yet defined.
  • A majority of responders had no recurrent genomic determinant of cisplatin response identified by WES; epigenetic or expression-based DNA-repair alterations may mediate sensitivity in those cases.
  • Larger cohorts may reveal ERCC2-mutant non-responders; intratumoral heterogeneity in post-chemotherapy resistant tumors was not examined.
  • Generalizability of the ERCC2-cisplatin association to other tumor types harboring ERCC2 mutations (observed at low frequency in 11 of 19 TCGA tumor types) remains untested.

Citations from this paper used in the wiki

  • “ERCC2, a nucleotide excision repair gene, was the only significantly mutated gene enriched in the cisplatin responders compared with non-responders (q < 0.01).” (Abstract)
  • “The median mutation rate was 9.7 mutations per megabase (mutations/Mb) for responders and 4.4 mutations/Mb for non-responders (P = 0.0003; Mann-Whitney test).” (Results, p. 3)
  • “All ERCC2 non-synonymous somatic mutations occurred in the cisplatin sensitive tumors (P < 0.001; Fisher’s exact test). ERCC2 remained significantly enriched in responders following false discovery analysis … (q = 0.007; Benjamini-Hochberg).” (Results, p. 4)
  • “When compared to these unselected populations, ERCC2 mutations were significantly enriched in the cisplatin responder cohort (36% of cases; P < 0.001; binomial test).” (Results, p. 4)
  • “ERCC2 mutation status does not appear to be prognostic, as it had no impact on overall survival in the TCGA cohort (P = 0.77; Log-Rank).” (Results, p. 4)
  • “Expression of wild-type (WT) ERCC2 rescued cisplatin sensitivity of the ERCC2-deficient cell line, whereas none of the ERCC2 mutants rescued cisplatin sensitivity.” (Results, pp. 4-5)
  • “In seven (78%) of the ERCC2-mutant cases, the ERCC2 mutation allelic fraction was < 0.5 … suggesting that WT ERCC2 remains present at one allele.” (Discussion, p. 6)
  • “In two responder tumors that did not have ERCC2 mutations, somatic nonsense (truncating) mutations were detected in the homologous recombination DNA repair genes BRCA1 and BRCA2.” (Results, p. 5)

This page was processed by crosslinker on 2026-05-11.