Graff JN, Tagawa S, Hoimes C, Gerritsen W, Vaishampayan UN, Elliott T, Hwang C, Tije AJT, Omlin AG, McDermott RS, De Wit R, Qiu P, Poehlein C, Kim J, Suttner L, Cristescu R, Marton MJ, Schloss C, de Bono JS, and Antonarakis ES. 61MO Biomarker analysis of men with enzalutamide (enza)-resistant metastatic castration-resistant prostate cancer (mCRPC) treated with pembrolizumab (pembro) + enza in KEYNOTE-199. Ann Oncol 2021; 32:S383.
Background: In KEYNOTE-199 (NCT02787005), pembro + enza had durable antitumor activity in enza-refractory mCRPC. We evaluated the association between prespecified biomarkers and clinical outcomes.
Methods: Cohorts 4 (C4; RECIST-measurable disease) and 5 (C5; nonmeasurable, bone-predominant disease) enrolled men with chemotherapy-naive mCRPC, irrespective of PD-L1 status, that progressed after initial response to enza. We evaluated TMB by whole exome sequencing (n = 64), PD-L1 combined positive score (CPS) by IHC (n = 124), and 18-gene T-cell–inflamed gene expression profile (TcellinfGEP) by NanoString (n = 51). Outcomes were DCR, PFS, PSA response, PSA progression, OS, and ORR per blinded independent review (C4 only). Significance of continuous biomarkers (CPS, TMB, GEP) was prespecified at 0.05 for 1-sided P values from logistic (ORR, DCR, PSA response) and Cox proportional hazard (PFS, OS, PSA progression) regression adjusted for ECOG PS.
Results: In C4, ORR was 10% (5/48) in pts with evaluable TMB data and 12% (10/81) in pts with CPS data. In C4 and C5, 16% (10/64) and 14% (17/124) of pts with TMB and CPS data, respectively, achieved a PSA response. TMB was significantly associated with DCR (P = 0.03) and trended toward an association with PSA response (P = 0.08). TMB (AUROC [95% CI]: 0.68 [0.51-0.86]), but not CPS (0.54 [0.41-0.67]) or TcellinfGEP (0.55 [0.37-0.74]), enriched for PSA response. TMB (P = 0.04), but not CPS (P = 0.57) or TcellinfGEP (P = 0.32), was significantly associated with PSA progression. There was 1 MSI-H pt (per Promega PCR assay); this pt achieved an objective and PSA response and had PFS >6 months. TMB, CPS, and TcellinfGEP were not associated with PFS or OS. There was a low prevalence of TMB ≥175 mut/exome (11%) and TcellinfGEP-high (≥−0.318; 16%).
Conclusions: In this biomarker analysis of KEYNOTE-199 C4-C5, PD-L1 CPS and TcellinfGEP were not significantly associated with clinical outcome. Despite the low prevalence of TMB ≥175 mut/exome, TMB was positively associated with outcomes of pembro + enza in pts with mCRPC. The sample sizes for the exploratory analyses were small, and results should be interpreted with caution.