KEYSTEP-008: Pembrolizumab (pembro)-based combinations for microsatellite instability-high (MSI-H)/mismatch repair deficient (dMMR) metastatic colorectal cancer (mCRC).

137 Background: The open-label, phase 2 KEYSTEP-008 study (NCT04895722) evaluated pembro-based combinations vs pembro in participants (pts) with chemotherapy-refractory (cohort A) or previously untreated (cohort B) MSI-H/dMMR mCRC. Methods: Adults with stage IV colorectal adenocarcinoma, locally confirmed MSI-H/dMMR status and ECOG PS 0 or 1 were randomized 1:1 (cohort A) to coformulated quavonlimab (anti–CTLA-4; Qmab)/pembro (25 mg/400 mg) IV Q6W or pembro 400 mg IV Q6W or 1:1:1:1:1 (cohort B) to Qmab/pembro (25 mg/400 mg) IV Q6W, MK-4830 (anti-ILT4) 800 mg + pembro 200 mg IV Q3W, coformulated favezelimab (anti−LAG-3; fave)/pembro (800 mg/200 mg) IV Q3W, coformulated vibostolimab (anti-TIGIT; vibo)/pembro (200 mg/200 mg) IV Q3W, or pembro 400 mg IV Q6W. Pts received treatment for ~2 y. Primary end point: ORR per RECIST v1.1. Secondary end points: DOR and PFS per RECIST v1.1, OS and safety. Results: In cohort A, 121 pts were randomized to Qmab/pembro (n = 60) or pembro (n = 61). In cohort B, 181 pts were randomized to Qmab/pembro (n = 39), MK-4830 + pembro (n = 20), fave/pembro (n = 40), vibo/pembro (n = 41), or pembro (n = 41). Median (range) follow-up was 33.6 mo (19.3-46.0) for cohort A and 20.3 mo (10.9-37.5) for cohort B. In cohort A, ORR (95% CI) was 45% (32-58) with Qmab/pembro and 30% (19-43) with pembro. In cohort B, ORRs (95% CI) were 46% (30-63) with Qmab/pembro, 65% (41-85) with MK-4830 + pembro, 38% (23-54) with fave/pembro, 49% (33-65) with vibo/pembro, and 42% (26-58) with pembro. Additional results are in the table. Grade 3-4 TRAEs occurred in 12% with Qmab/pembro and 8% with pembro in cohort A; and 21% with Qmab/pembro, 0% with MK-4830 + pembro, 20% with fave/pembro, 27% with vibo/pembro, and 20% with pembro, in cohort B; 2 pts with vibo/pembro had grade 5 TRAEs. Conclusions: Pembro-based combinations evaluated in this study showed numerically higher ORR vs pembro alone, but the absolute improvement was modest, especially in the 1L setting. Despite manageable safety, these results do not support further clinical development of these combinations in pts with MSI-H/dMMR mCRC. Clinical trial information: NCT04895722 . Cohort AQmab/pembron = 60 Cohort APembron = 61 Cohort BQmab/pembron = 39 Cohort BMK-4830 + pembron = 20 Cohort BFave/pembron = 40 Cohort BVibo/pembron = 41 Cohort BPembron = 41 ORR (95% CI), % 45 (32-58) 30 (19-43) 46 (30-63) 65 (41-85) 38 (23-54) 49 (33-65) 42 (26-58) DOR, median (range), mo NR (8 to 42+) NR (6 to 44+) NR (4+ to 27+) NR (4+ to 31+) NR (4 to 27+) NR (2+ to 25+) NR (4+ to 34+) PFS, median (95% CI), mo 18 (4-NR) 4 (2-15) NR (12-NR) NR (2-NR) 8 (2-NR) 25 (8-NR) NR (4-NR) PFS HR(95% CI) 0.65 (0.41-1.02) - 0.78 (0.39-1.56) 0.77 (0.33-1.81) 1.18 (0.62-2.26) 0.78 (0.38-1.58) - OS, median (95% CI), mo NR (29-NR) 37 (14-NR) NR (NR-NR) NR (7-NR) NR (21-NR) NR (NR-NR) NR (21-NR) OS HR(95% CI) 0.70 (0.41-1.21) - 0.99 (0.43-2.29) 1.17 (0.45-3.07) 1.02 (0.45-2.32) 0.80 (0.33-1.95) - NR, not reached.