Stop vs continue PD-1/PD-L1 at 1 year shows similar “cure” and PFS, with OS signal favoring stopping

In the randomized Phase 3 NCT04157985 trial, patients with advanced solid tumors who had ≥SD after 1 year of standard-of-care PD-1/PD-L1 inhibitor therapy were randomized to stop vs continue immunotherapy (n=161; Stop 81, Continue 80; median age 71; tumor types: NSCLC 39.8%, dMMR 18.6%, RCC 11.8%, HNSCC 8.1%, melanoma 7.5%). With median follow-up 43.2 months, 5-year “cure rate” (no PD/death/new therapy including IO re-treatment) was 45% vs 38% (OR 2.26, p=0.22), median PFS 55.4 vs 31.2 months (HR 0.77, p=0.28), and median OS NR vs 53.5 months (HR 0.54, p=0.04) for stop vs continue; in the stop arm, 17 patients were retreated at progression with 20% achieving PR and post-retreatment median PFS 10.5 months. Safety was reported indirectly (7.7% discontinued for toxicity in the continue arm); the big clinical context is that many pivotal PD-1 regimens cap treatment at ~2 years (e.g., pembrolizumab up to 35 cycles in KEYNOTE-042), while the only prior randomized 1-year stop vs continue signal in NSCLC (CheckMate 153) favored continuing nivolumab (PFS HR 0.56; OS HR ~0.61 from randomization), making this cross-tumor OS advantage for stopping directionally surprising and very plausibly a power/imbalance artifact given early closure.