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

Abstract

Cisplatin-based chemotherapy is the standard of care for patients with muscle-invasive urothelial carcinoma. Pathologic downstaging to pT0/pTis after neoadjuvant cisplatin-based chemotherapy is associated with improved survival, although molecular determinants of cisplatin response are incompletely understood. We performed whole-exome sequencing on pretreatment tumor and germline DNA from 50 patients with muscle-invasive urothelial carcinoma who received neoadjuvant cisplatin-based chemotherapy followed by cystectomy (25 pT0/pTis "responders," 25 pT2+ "nonresponders") to identify somatic mutations that occurred preferentially in responders. ERCC2, a nucleotide excision repair gene, was the only significantly mutated gene enriched in the cisplatin responders compared with nonresponders (q < 0.01). Expression of representative ERCC2 mutants in an ERCC2-deficient cell line failed to rescue cisplatin and UV sensitivity compared with wild-type ERCC2. The lack of normal ERCC2 function may contribute to cisplatin sensitivity in urothelial cancer, and somatic ERCC2 mutation status may inform cisplatin-containing regimen usage in muscle-invasive urothelial carcinoma.

Significance: Somatic ERCC2 mutations correlate with complete response to cisplatin-based chemosensitivity in muscle-invasive urothelial carcinoma, and clinically identified mutations lead to cisplatin sensitivity in vitro. Nucleotide excision repair pathway defects may drive exceptional response to conventional chemotherapy.

Conflict of interest statement

Conflict of Interest Statement: Dr. Rosenberg is a consultant for Boerhinger Ingelheim, Bristol Myers Squib, Oncogenex, Onyx, Johnson and Johnson, and Dendreon. Dr. Garraway and Dr. Wagle are equity holders in and consultants to Foundation Medicine. Dr. Garraway is a consultant to Novartis, Millenium/Takeda, and Boehringer Ingelheim, and a recipient of a sponsored research grant from Novartis. Drs. Rosenberg, Garraway, Van Allen, Mouw, Wagle, and D’Andrea have a patent pending for the relationship between somatic ERCC2 mutations and cisplatin response.

©2014 American Association for Cancer Research.

Figures

Figure 1. Study design, mutation rates, and aggregate significant somatic mutations

Panel A shows patients with muscle-invasive urothelial carcinoma cancer split into cases and controls based on their pathologic response to cisplatin-based neoadjuvant chemotherapy (TURBT: transurethral resection of bladder tumor). Nine cases could not complete sequencing due to technical reasons (failed sequencing or elevated contamination). Data in Panel B are arranged so that each column represents a tumor and each row represents a gene. The center panel is divided into responders (left and black) and non-responders (right and yellow). The mutation rates of responders are elevated compared to non-responders (top of Panel B). The alteration landscape (center of Panel B) of the aggregate cohort (n = 50 patients) demonstrates a set of statistically significant genes that are altered in urothelial carcinoma (TP53, RB1, KDM6A, ARID1A). The negative log of the q values for the significance level of mutated genes is shown (for all genes with q < 0.1) on the right side of Panel B. ERCC2 mutation status is also shown below the other genes, although ERCC2 was not significantly mutated across the combined cohort. Additional data regarding allelic fraction ranges for each case (bottom of Panel B), mutation rates (top of Panel B), and mutational frequency (left of Panel B) are also summarized in this figure.

Figure 2. Three tests examining selective enrichment of ERCC2 mutations in cisplatin-responder tumors

Panel A shows a plot of MutSigCV gene-level significance (−log10(MutSigCV p-value) and responder enrichment significance (−log10(Fisher’s exact test p-value)). The size of the point is proportional to the number of responder patients who harbor alterations in the gene. Genes with a responder enrichment p-value of < 0.01 are colored red; others are colored gray, and the dashed line denotes a p value of 0.01. Only ERCC2 reaches statistical significance in the responder cohort (P < 0.001; Fisher’s exact test). In Panel B, among genes with sufficient number of alterations for cohort comparisons (n = 9), only ERCC2 somatic mutations occur exclusively in the cisplatin responders, which is significant when accounting for the elevated mutation rate in responders compared to non-responders (P < 0.05, denoted by asterisk). Compared to unselected TCGA and Guo et al urothelial carcinoma cohorts, Panel C shows that ERCC2 somatic mutations are significantly enriched in the responder cohort (P < 0.01, denoted by asterisk).

Figure 3. ERCC2 mutation mapping and distribution across tumor types

Panel A depicts a stick plot of ERCC2 showing the locations of somatic mutations in the responders compared to ERCC2 mutations observed in two separate unselected bladder cancer exome cohorts. The ERCC2 mutations cluster within or near conserved helicase motifs. Panel B illustrates the somatic ERCC2 mutation frequency in multiple tumor types from the Cancer Genome Atlas (TCGA). In Panel C, the structure of an archaebacterial ERCC2 (PDB code: 3CRV) with mutations identified in the responder cohort mapped to their equivalent position is illustrated. These locations are shown in the context of canonical germline ERCC2 mutations responsible for xeroderma pigmentosum (XP), xeroderma pigmentosum/Cockayne Syndrome (XP/CS), and trichothiodystrophy (TTD).

Figure 4. ERCC2 mutants fail to rescue cisplatin sensitivity of ERCC2-deficient cells

Panel A shows an immunoblot of ERCC2 expression in cell lines created by transfection of the ERCC2-deficient parent cell line (GM08207; Coriell Institute) with pLX304 (Addgene) encoding GFP (negative control), WT ERCC2, or a mutant ERCC2. The negative control ERCC2-deficient cell line (lane 1) expresses endogenous levels of inactive ERCC2 from the parent cell genome, whereas WT (lane 2) and mutant (lanes 3–7) ERCC2 cell lines show increased levels of ERCC2 expressed from the transfected gene. β-actin is shown as a loading control. Panel B shows the cisplatin sensitivity profiles of cell lines expressing WT or mutant ERCC2. Expression of WT ERCC2 in an ERCC2-deficient background rescues cisplatin sensitivity, whereas expression of the ERCC2 mutants fails to rescue cisplatin sensitivity. An IC50 was calculated from the survival data for each cell line and these values are shown in Panel C. The difference in IC50 between the parent (ERCC2-deficient) cell line and the cell line expressing WT ERCC2 was statistically significant, as was the difference between the WT ERCC2 cell line and each of the mutant ERCC2 cell lines (P < 0.0001; ANOVA). The difference between the ERCC2-deficient cell line and each of the mutant cell lines was not statistically significant.

Figure 5. ERCC2 mutants fail to rescue UV sensitivity of ERCC2-deficient cells

Panel A shows a representative colony formation assay for the ERCC2-deficient cell line (top) as well as the ERCC2-deficient line transfected with WT ERCC2 (middle) or one of the ERCC2 mutants (D609G, bottom) following increases doses of UV irradiation. Panel B shows clonogenic survival data for negative control, WT, and mutant ERCC2 cell lines. WT ERCC2 rescues UV sensitivity of the ERCC2-deficient cell line whereas the mutant ERCC2s fail to rescue UV sensitivity. In Panel C, UV IC50 values for cell lines are shown. The difference between the ERCC2-deficient cell line and the WT ERCC2 cell line was significant (P < 0.0001; ANOVA), whereas the difference between the ERCC2-deficient cell line and each of the ERCC2 mutant cell lines was not statistically significant (NS).

Figure 6. ERCC2 mutants fail to rescue genomic instability following cisplatin exposure

Representative mitotic spreads from an ERCC2-deficient cell line (Panel A), and the same ERC2-deficient cell line transfected with WT ERCC2 (Panel B) or one of the ERCC2 mutants (V242F, Panel C) following cisplatin exposure. Panel D shows chromosomal aberration data from ERCC2-deficient, WT ERCC2, and mutant ERCC2 cell lines. Rates of chromosomal aberrations following cisplatin exposure were significantly lower in the WT ERCC2 cell line than in the ERCC2-deficient line or the cell lines expressing mutant ERCC2 (P = 0.03; ANOVA)