Clinicopathologic and Genomic Characterization of PD-L1 Positive Urothelial Carcinomas.

INTRODUCTION Pembrolizumab was approved with an accompanying companion diagnostic (CDx) assay (PD-L1 DAKO 22C3) for urothelial carcinoma (UC). In this study, we further characterize the clinicopathologic and genomic features of UC that are PD-L1 positive. MATERIALS AND METHODS The cohort of this study consisted of a total of 528 consecutive UC patients with PD-L1 IHC and comprehensive genomic profiling (CGP). All PD-L1 IHC testing was performed using the DAKO 22C3 CDx assay for UC. PD-L1 positivity was determined at a combined positive score (CPS) ≥ 10. RESULTS 44.5% of 528 consecutive UC patients were PD-L1positive . A lower PD-L1 positivity rate was detected in primary (42.3%, 148/350) vs metastatic sites (48.9%, 87/178). PD-L1 positivity was dependent on the location of the metastatic sites with PD-L1 positivity rate. CGP revealed PD-L1positive patients had more frequent genomic alterations (GA) in TP53 (p = 0.006) and RB1 (p = 0.003) and less frequent GA in FGFR3 (p = 0.001) and MTAP (p = 0.028). The APOBEC mutational signature and TMB-H were more common in PD-L1positive patients. By testing UC patients with CGP, in addition to PD-L1 IHC, an additional 97 patients (18.4%) in the total cohort were eligible for immunotherapy based on TMB status. CONCLUSION PD-L1positive and PD-L1negative urothelial carcinomas are genomically different. Also, our study provides the framework for future clinical investigation with regards to specimen site selection for PD-L1 testing as well as candidate biomarker genomic alterations that may predict for better response or lack of response to immune checkpoint inhibitors. IMPLICATIONS FOR PRACTICE In this study, we found a higher prevalence of TP53 and RB1 alterations, and APOBEC mutational signatures in the PD-L1positive urothelial carcinoma disease subset and enrichment of FGFR3 alterations in the PD-L1negative disease subset. This data provides the basis for future investigation into the role of these genomic changes as positive and negative predictors of immunotherapy response. Also, we saw differences in PD-L1 positivity based on the collection site of the sample, which can provide a framework for future clinical trial design and could influence sample selection for PD-L1 testing in UC patients when multiple samples are available.