1L pembrolizumab (pembro) versus chemotherapy (chemo) for choice-of-carboplatin patients with advanced urothelial carcinoma (UC) in KEYNOTE-361.

Abstract

450 Background: 1L pembro is approved in advanced UC for cisplatin-ineligible pts with PD-L1 combined positive score (CPS) ≥10 and any platinum-ineligible pts regardless of CPS in the United States based on single-arm trial data. In the phase III KEYNOTE-361 study, 1L pembro + chemo did not statistically significantly improve PFS or OS vs chemo for pts with advanced UC; formal testing of 1L pembro vs chemo was not performed. We present an exploratory analysis of outcomes with pembro vs chemo for choice-of-carboplatin (carbo) pts in KEYNOTE-361 (NCT02853305). Methods: At randomization, choice of platinum agent (cisplatin or carbo) plus gemcitabine for each pt was selected based on investigator’s assessment of cisplatin ineligibility. ORR/DOR per RECIST v1.1 by blinded independent central review and OS were determined for all pts selected for carbo (“choice-of-carbo”) and also choice-of-carbo pts with CPS ≥10. Risk difference assessment for select AEs for pembro vs chemo was conducted in choice-of-carbo pts who received ≥1 dose study treatment. Results: As of Apr 29, 2020, the median (range) time from randomization to data cutoff in the full study cohort was 31.7 (22.0-42.3) mo. At randomization, renal impairment was the most common reason for choice of carbo by investigators (36% of all pts). 170 choice-of-carbo pts were randomized to the pembro arm, and 196 choice-of-carbo pts to the chemo arm. Median OS in this subgroup was 14.6 mo with pembro vs 12.3 mo with chemo (HR 0.83 [95% CI 0.65-1.06]). 18-mo OS rate was 42% with pembro vs 40% with chemo. ORR to pembro vs chemo was 27.6% vs 41.8%. Median (range) DOR with pembro vs chemo was not reached (NR) (3.2+-36.1+ mo) vs 6.3 (1.8+-33.8+) mo. 84/170 (49%) and 89/196 (45%) choice-of-carbo pts in the pembro and chemo arms, respectively, had CPS ≥10. In this subgroup, median OS was 15.6 mo with pembro vs 13.5 mo with chemo (HR 0.82 [95% CI 0.57-1.17]). 18-mo OS rate was 44% with pembro vs 43% with chemo. ORR to pembro vs chemo was 29.8% vs 46.1%. Median (range) DOR with pembro vs chemo was NR (4.2-36.1+ mo) vs 8.3 (2.1+-33.8+) mo. Among treated pts (N=166 for pembro, N=190 for chemo), 112 pts (68%) in the pembro arm and 163 pts (86%) in the chemo arm had grade 3-5 AEs of any cause. Pembro was associated with a higher risk of pruritus, while chemo was associated with a higher risk of decreased white blood cell, neutrophil, and platelet counts, nausea, thrombocytopenia, neutropenia, and anemia. Conclusions: Due to the trial design, this subset was not statistically tested and is exploratory. Median OS and 18-mo OS rates did not appear markedly different in the two arms; some parameters such as DOR favored pembro, although longer follow-up is needed to determine median DOR for pembro. The PD-L1 CPS ≥10 did not clearly enrich for responders to pembro in choice-of-carbo pts. Pembro was associated with a lower rate of grade 3-5 AEs of any cause than chemo. Clinical trial information: NCT02853305.