PD-L1 expression, Cancer Genome Atlas (TCGA) subtype, and mutational load as independent predictors of response to atezolizumab (atezo) in metastatic urothelial carcinoma (mUC; IMvigor210).

Abstract

104 Background: Biomarkers of clinical benefit of immunotherapies, such as non-synonymous mutations, mutational load, PD-L1 IHC and immune signatures, have not been systematically evaluated in mUC patients (pts) treated with anti–PD-L1 agents. Methods: Exploratory analyses of immune and genetic predictors of response to atezo were performed on archival tumor specimens from mUC pts (NCT02108652 cohort 2). PD-L1 expression in tumor samples was assessed by IHC (SP142 assay) and associated with response as determined by RECIST v1.1 (central review). Gene expression was quantified by RNA-seq (Illumina TruSeq RNA Access; n = 195); molecular subtypes were assigned as per TCGA. Mutations were detected by genomic profiling (FoundationOne 315-cancer gene panel; n = 150) to estimate overall load. Results: PD-L1 tumor-infiltrating immune cell (IC) status was associated with CD8+ T effector (Teff) gene expression levels (P < 0.0001). Notably, response to atezo was associated with high expression of CXCL9 (P = 0.0057) and CXCL10 (P = 0.0079). Evaluable samples were classified into luminal (n = 73) and basal (n = 122) TCGA subtypes. PD-L1 IC prevalence was enriched in basal (60%) vs luminal (23%) samples (P < 0.0001); elevated tumor cell PD-L1 was mostly restricted to the basal subtype (39% basal vs 4% luminal, P < 0.0001) and did not correlate with ORR. Teff expression was elevated in luminal cluster II and basal III/IV and not in luminal I. Responses occurred in all subtypes, but ORR was significantly higher in luminal II samples: 34% (P = 0.0017). While median mutational load was significantly increased in responders (12.4/Mb) vs non-responders (6.4/Mb; P < 0.0001) and correlated with both PFS (P = 0.003) and OS (P = 0.014), these associations were independent of TCGA subtype. Conclusions: PD-L1 expression on IC was associated with response to atezo. Intrinsic TCGA subtype and mutational load each added predictive value. Simultaneous classification by these characteristics in mUC pts treated with atezo further defines the drivers of immune responsiveness and may suggest potential combination strategies. Clinical trial information: NCT02108652.