Observation versus maintenance PD-1 inhibitor therapy after clinical complete response in dMMR/MSI-H colorectal cancer managed with non-operative management: A multicentre cohort study.

3502 Background: Patients with mismatch repair–deficient (dMMR)/microsatellite instability–high (MSI-H) colorectal cancer (CRC) who achieve a clinical complete response (cCR) after PD-1 inhibitor therapy may undergo non-operative management (NOM) with active surveillance. We aimed to assess the necessity of maintenance immunotherapy after achieving cCR, with a focus on survival outcomes and the burden of immune-related toxicity. Methods: A multicentre observational cohort study was conducted at five tertiary hospitals in China between Jan 1, 2018, and March 21, 2025. Patients with dMMR/MSI-H CRC who achieved cCR after PD-1 inhibitor therapy and entered NOM were included. Patients were stratified into an observation group （discontinue PD-1 inhibitor after cCR) and a maintenance group (received ≥2 cycles of PD-1 inhibitors after cCR). Disease-free and overall survival were compared between groups using Kaplan–Meier methods with log-rank test, and local regrowth, distant metastasis, and immune-related adverse events were assessed. Results: Among 318 patients treated with PD-1 inhibitors, 195 (61·3%) achieved cCR. A total of 129 patients were analysed (observation n = 66; maintenance n = 63), including 58 with rectal cancer, 61 with colon cancer, and 10 with synchronous dual primary tumours; 14 had distant metastases at diagnosis. Clinical complete response was assessed with endoscopy and pelvic MRI/CT along with digital rectal examination for rectal cancer, and endoscopy with contrast-enhanced CT or PET/CT for colon cancer. Median PD-1 inhibitor exposure to achieve cCR was eight cycles in both groups; 69·8% achieved cCR within eight cycles. Median follow-up was 3·2 years (IQR 2·0–4·0). Local regrowth occurred in two patients in the observation group and none in the maintenance group, and no distant metastases were observed. The 3-year DFS was 96·4% (95% CI 91·6–100·0) in the observation group and 98·4% (95% CI 95·2–100·0) in the maintenance group (log-rank p = 0·55). The 3-year OS was 100·0% (95% CI 100·0–100·0) versus 98·4% (95% CI 95·2–100·0) (p = 0·34). Any-grade irAEs occurred in 42/66 (63·6%) versus 44/63 (69·8%); grade 3 irAEs were numerically higher with maintenance (7/63 [11·1%] vs 2/66 [3·0%]; p = 0·091). Conclusions: To our knowledge, this is the largest series of dMMR/MSI-H CRC patients achieving cCR after PD-1 inhibitor therapy and managed with NOM. Maintenance therapy after cCR did not provide a clear survival advantage but was associated with a higher burden of immune-related toxicity, supporting treatment discontinuation with close surveillance.