Background: Dostarlimab (TSR-042) is an investigational humanized anti-PD-1 monoclonal antibody that binds the PD-1 receptor, blocking the interaction with ligands PD-L1 and PD-L2. The ongoing GARNET trial (NCT02715284) is evaluating dostarlimab in pts with advanced solid tumors. Here we present safety and efficacy data from cohort F. Methods: Cohort F in the GARNET trial enrolled pts with dMMR or microsatellite instability high (MSI-H) non-endometrial solid tumors, the majority of which were GI in origin. Pts must have progressed following prior systemic therapy for advanced disease, assessed by independent central review (ICR). Pts received 500 mg dostarlimab Q3W for 4 cycles and 1000 mg Q6W thereafter. Objective response rate (ORR) and duration of response (DOR) were assessed by ICR per RECIST v1.1. Pts were included in the efficacy analysis if they received ≥1 dose of dostarlimab, had measurable disease at baseline, and ≥24 weeks of follow up. All pts who received ≥1 dose were included in the safety analysis. Results: 109 pts were included in the safety analysis, with 48 dMMR pts in the efficacy analysis. Of the 48 pts, 42 (88%) had GI tumors. Confirmed ORR in dMMR pts was 43.8% (95% CI: 29.5, 58.8), with a complete response rate of 8.3%. ORR was consistent across tumor type (Table). Median DOR was not reached. The probability of maintaining response at 12 months was 85.9%. Treatment-related adverse events (TRAEs) were reported in 55% of pts; 7.3% of pts experienced grade ≥3 TRAEs, including 1 each of adrenal insufficiency, rash, diarrhea, and fatigue. Treatment-related serious AEs (SAEs) were reported in 6 (5.5%) pts, and 2 pts (1.8%) discontinued dostarlimab due to a TRAE. No treatment-related deaths were reported. Conclusions: Dostarlimab demonstrated robust, durable antitumor activity in a cohort of dMMR solid tumor pts, the majority of whom had GI cancers, with an acceptable safety profile. Clinical trial information: NCT02715284