Comprehensive genomic profiling (CGP) of clinical T2-4N0M0 muscle-invasive bladder cancer (MIBC) treated with neoadjuvant pembrolizumab or cisplatin-based chemotherapy before radical cystectomy (RC).

Abstract

653 Background: The identification of tumor biomarkers associated with pathological complete response (CR: ypT0N0) to neoadjuvant therapies is a primary goal of the ongoing clinical research in MIBC. Furthermore, evaluating the shifting spectrum of genomic alterations (GA) in matched pre-post therapy samples could orient novel therapeutic sequences. Methods: We retrospectively evaluated the clinical and genomic findings of patients (pts) with cT2-4N0M0 MIBC who received neoadjuvant pembrolizumab in the PURE-01 study, or standard cisplatin-based neoadjuvant chemotherapy (NACT), preceding RC at our center. We focused the analyses on the DNA GA via CGP assays, performed on tumor samples, to identify GA in 324 cancer-associated genes and genomic signatures, including TMB, using a hybrid capture-based comprehensive genomic profiling assay. TMB was categorized as low (<10 mutations [mut]/Mb), high (10-19), or very high (≥20). Germline status of GA was predicted using a validated somatic-germline computational method. Genomic signatures were analyzed via the principal Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways distribution according to the pathological response. The primary endpoint was to compare differences between ypT0N0 responders and nonresponders (NR: ypT≥2 and/or ypN1-3). All p-values were two-sided, and multiple hypothesis testing correction was performed using the Benjamini–Hochberg procedure to calculate the false discovery rate (FDR). Results: We identified TURBT samples from 129 pts included in PURE-01 and 27 patients treated with NAC, from 2017 to 2022. Mean GA/tumor of CR from PURE-01 was 28.8 vs 17.3 from NAC. No germline GA were found. In PURE-01, higher mean GA/tumor and higher mean TMB were the only GA associated with CR vs NR (p<0.001), whereas no differences were found at single gene and gene pathways level. In NAC cohort, no GA nor signatures were found to be significantly associated with CR. Despite the mean TMB of CR pts were similar between PURE-01 and NAC cohorts (16.4 vs 19.6 mut/Mb), mean TMB of NR in PURE-01 cohort was lower than that of NR from NAC cohort (9.3 vs 18.6 mut/Mb). In PURE-01 we analyzed 37 matched TURBT-RC samples: there were no significant differences in GA or pathways, with a nonsignificant decrease in mean TMB values in RC vs TURBT samples (6.73 vs 8.86 mut/Mb, p=0.16). Conclusions: In our study, no genomic biomarkers linked to NAC activity emerged at the CGP. Conversely, in PURE-01 TMB confirmed to be to most reliable DNA biomarker to separate CR vs NR. Interestingly, TMB values could also help predicting the lack of benefit from neoadjuvant pembrolizumab use, confirming the reliability of the 10 mut/Mb cutoff to allow the exclusion of predicted nonresponders from neoadjuvant pembrolizumab trials.