Clinical Validation of Chemotherapy Response Biomarker ERCC2 in Muscle-invasive Urothelial Bladder Carcinoma

Neoadjuvant cisplatin-based chemotherapy (NACC) is a standard of care for muscle-invasive urothelial bladder carcinoma (MIBC), and complete responses at cystectomy are associated with improved overall survival. Somatic mutations in ERCC2, a member of the nucleotide excision repair pathway, correlated with pathologic response in a clinical discovery cohort and conferred cisplatin sensitivity in preclinical systems¹. Here we investigate this association in an independent validation patient cohort.

Methods

Samples were identified from patients with MIBC (T2+ disease) from two clinical trials of NACC^2,3 who completed chemotherapy and had available pre-chemotherapy tumor tissue. Tumor samples from this cohort previously underwent targeted gene panel sequencing² (which did not include ERCC2). We performed whole exome sequencing of tumor and matched germline DNA. Analysis and automated variant calls were performed as previously described¹. Quality control (QC) included minimum average read coverage, DNA fingerprinting, and contamination estimation. For this validation study, only ERCC2 results were exposed from the automated data output. All putative somatic ERCC2 mutations were manually reviewed⁴.

A two-sided Fisher’s exact test was used to test association between ERCC2 alteration and response, and p<0.05 was used as the threshold for significant results. Response was defined as pT0/pTis/pTa disease at cystectomy. Overall survival from time of cystectomy was estimated using Kaplan-Meier curves and association with ERCC2 alteration tested using a log-rank test in the current validation cohort and the original discovery cohort¹.

All patients were enrolled on IRB-approved protocols at their respective sites, and provided written informed consents.

Results

55/62 patients who received all 3 cycles of chemotherapy had adequate tissue for sequencing. After QC, 7 samples were excluded, leaving samples from 48 patients for analysis. 10 patients harbored nonsynonymous ERCC2 genetic alterations. ERCC2 was associated with response to therapy: 8/20 (40%) responders and 2/28 (7%) non-responders had a nonsynonymous ERCC2 genetic alteration (p=0.010, OR 8.3 [95% CI 1.4–91.4]). In follow-up, 17 patients had disease progression and 15 died; there was a statistically significant difference in overall survival among patients with ERCC2 alterations in both the current validation cohort (p = 0.03, Fig 1a) and in the discovery cohort¹ (p = 0.049, Fig 1b).

Fig 1.

(a) Overall Survival with and without somatic ERCC2 mutations in the current (Fox Chase Cancer Center) validation cohort. Kaplan Meier analysis of overall survival by the presence or absence of a somatic ERCC2 mutation. There is a statistically significant difference in survival (log-rank test p = 0.03)

(b) Overall Survival with and without somatic ERCC2 mutations in a previously reported¹ (Dana Farber Cancer Institute + Memorial Sloan Kettering Cancer Center combined) discovery cohort. Kaplan Meier analysis of overall survival by the presence or absence of a somatic ERCC2 mutation. There is a statistically significant difference in survival, log-rank test p = (0.049)

Discussion

In this study we validate the relationship between ERCC2 alteration and response to NACC in an independent MIBC cohort. ERCC2 is the helicase that unwinds DNA for repair via the nucleotide excision repair pathway, which is important for repair of platinum-induced DNA damage. This provides a biologically plausible mechanism for loss-of-function mutations leading to cisplatin sensitivity.

Preclinical studies have associated ERCC2 loss-of-function with cisplatin sensitivity⁵. While wild-type ERCC2 cloned into an ERCC2-null cell line rescued its cisplatin-sensitive phenotype, clinically-identified ERCC2 mutations introduced into the same line failed to correct the sensitivity¹. Taken together with the current study, the evidence strongly suggests that alterations in ERCC2 confer vulnerability to cisplatin chemotherapy.

ERCC2 as a biomarker has limitations; while it substantially increases the odds of a complete or near complete response and improved survival, it is not 100% specific. In non-responders, ERCC2 mutations are in the same (peri-) helicase regions as in responders. Furthermore, ERCC2 alterations are found only in 40% of responders, suggesting other factors affect NACC sensitivity. In the same cohort tested here, the three gene signature (ATM, RB1, FANCC; all genes associated with DNA repair) predicted platinum response and survival², and may be complementary. Further characterization of specific ERCC2 mutations to improve specificity and development of a more sensitive integrated panel of DNA repair biomarkers may inform clinical decision-making. Broadly, these findings inform the convergence of precision cancer medicine approaches with conventional chemotherapy.