Nonlinear duration-efficacy association of neoadjuvant PD-1/PD-L1 blockade with chemoradiotherapy in pMMR/MSS locally advanced rectal cancer: a systematic review and meta-analysis.

BACKGROUND: Neoadjuvant PD-1/PD-L1 combined with chemoradiotherapy (CRT) has shown encouraging complete response (CR) rates in proficient mismatch repair (pMMR) or microsatellite-stable (MSS) locally advanced rectal cancer (LARC), yet immunotherapy duration varies widely across trials and its interaction with radiotherapy fractionation remains unclear. METHODS: PubMed, Embase, and Cochrane Library were searched through October 8, 2025 for prospective trials of PD-1/PD-L1 inhibitors plus CRT in pMMR/MSS LARC. Random-effects models pooled CR (cCR+pCR) rates, with subgroup analyses and study-level meta-regression conducted to explore sources of heterogeneity. A linear probit model and an interactive quadratic probit model were applied to assess associations between treatment duration and response (PROSPERO number CRD420251253156). RESULTS: Eighteen trials (972 patients; 10 LCRT-based and 8 SCRT-based) were included. Pooled CR and pCR rates were 44% (95% CI 0.38-0.49) and 40% (95% CI 0.36-0.45). CR was higher with SCRT plus ICI than LCRT plus ICI (50% vs 39%; P = 0.03), and pCR similarly favored SCRT (47% vs 34%; P < 0.01). Meta-regression identified radiotherapy fractionation and immunotherapy duration as study-level factors associated with CR. In analyses restricted to randomized controlled trials (RCTs), longer immunotherapy duration was positively associated with response. Exploratory quadratic models using all eligible studies suggested an inverted U-shaped pattern for LCRT (modeled peak at approximately 4.4 months) and a U-shaped pattern for SCRT (modeled trough at approximately 2.5 months). CONCLUSION: Among the included RCTs, longer immunotherapy duration was associated with higher response rates. The nonlinear patterns observed in the combined dataset were exploratory, may reflect heterogeneity between studies, and require prospective validation before informing treatment duration. SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261348026, identifier CRD420251253156.