Randomized phase II study of neoadjuvant (neoadj) anti-PD-1 dostarlimab (D) vs. D + anti-TIM-3 cobolimab (C) in high-risk resectable melanoma (mel) (NEO-MEL-T): Primary analysis.

Abstract

LBA9504 Background: Neoadj immunotherapy improves outcomes in high-risk resectable mel, although novel combinations that enhance efficacy while minimizing toxicity are needed. TIM-3 is highly expressed on dysfunctional CD8T cells and APCs within TME. Dual PD-1 and TIM-3 blockade promotes expansion and function of antigen-specific T cells, resulting in potent antitumor immunity. D and C are monoclonal antibodies targeting PD-1 and TIM-3. D+C has efficacy in PD-1 relapsed/refractory mel (Ribas, ASCO 2022) and NSCLC (Davar, SITC 2023). NEO-MEL-T is a randomized phase II study of neoadj D vs. D+C in patients (pts) with clinical stage III (AJCC IIIB-D) cutaneous mel. Methods: Pts aged >18 yo with clinical stage III mel were randomized (1:1) to neoadj D (500mg Q3W; Arm A) or D+C (500mg Q3W D + 300mg Q3W C; Arm B) for 2 cycles prior to surgery. Post-surgery, pts received further D (500mg Q3W x4; then 1000mg Q6W x6). Primary endpoint was major pathologic response (MPR) rate assessed by blinded pathologist. Secondary endpoints included safety, radiographic response rate (ORR), event-free survival (EFS), distant metastasis free survival (DMFS), and overall survival (OS). Target enrollment of 28 evaluable pts per arm (56 total) provided 80% power (1-sided α 0.05) to distinguish between null hypothesis of 28% MPR rate (historical neoadj PD-1) and alternative hypothesis of >50% in either arm. Primary analysis was a 1-sided z test. Results: Between 6/2020-11/2024, 57 pts were enrolled and randomized to either Arm A with neoadj D (n=30, 52.6%) or Arm B with neoadj D+C (n=27, 47.4%). Majority of pts were either stage IIIB (n=25, 43.9%) or IIIC (n=25, 43.9%), balanced across both arms. Median time to surgery was 51 days (range: 38-82), and all pts underwent curative surgery. Median follow-up time was 22 mos (range: 2-55). MPR rate was 33.3% (Arm A) and 51.9% (Arm B). MPR rate in Arm B was significantly greater than historical control (p=0.0029,1-sided z-test). Across both arms, median EFS was superior in MPR (unreached) vs. non-MPR (48 mos) (p=0.0380, log-rank test) pts, while median DMFS and OS have not been reached. 1-year EFS estimates were greater in Arm B (92%) compared to Arm A (82%), although this was not significant. The 1-year EFS in Arm B was significantly greater than historical adjuvant anti-PD-1 (p=0.0365, 1-sided z test). The proportion of pts with grade 3+ irAEs was similar in Arms A (16.7%) and B (14.9%) (p=0.4273). Conclusion: Neoadj D+C was safe and efficacious. The 1-year EFS of 92% in Arm B was significantly improved relative to adjuvant anti-PD-1, and this combination warrants further investigation. Clinical trial information: NCT04139902 . Arm A (D) (N=30) Arm B (D+C) (N=27) MPR (%; 95% CI) 10 (33;17-53) 14 (52;33-71) 1-year EFS (95% CI) 82 (61-92) 92 (71-98) 1-year EFS in MPR (95% CI) 100 (54-100) 100 (74-100) 1-year EFS in non-MPR (95% CI) 73 (47-88) 82% (45-95)