Reduced-dose versus label-dose anti-PD-1 therapy in advanced cancer: a systematic review and meta-analysis of randomised and cohort studies

Background: This review assessed whether reduced-exposure anti-PD-1 therapy (a dose below label, or an interval beyond label) is non-inferior to label dosing for overall survival in advanced cancer, and compared progression-free survival, objective response and severe immune-related toxicity. Methods: Systematic review and meta-analysis of randomised and cohort studies, reported according to PRISMA 2020. Hazard ratios (HRs) and risk ratios (RRs) were pooled with restricted maximum-likelihood random-effects models and the Hartung–Knapp–Sidik–Jonkman adjustment. Risk of bias was assessed with RoB 2 and ROBINS-I, and certainty with GRADE. Non-inferiority was judged against a prespecified HR margin of 1.20. Results: Ten studies met the inclusion criteria; nine entered the primary analyses (one randomised trial at interim analysis, eight cohorts). For overall survival (7 studies, 2,932 patients), the pooled HR for reduced versus label dosing was 0.96 (95% CI 0.75–1.23; I² 0%), and 0.96 (0.73–1.25) after excluding the one cohort at critical risk of immortal-time bias. The upper bound exceeded the margin, and omitting the largest registry moved the HR to 1.07 (0.81–1.42). For progression-free survival the HR was 0.99 (0.63–1.56; 5 studies) with that cohort and 0.94 (0.56–1.58) without it. Objective response did not differ (RR 0.89, 0.44–1.78). Grade ≥3 immune-related adverse events were less frequent with reduced dosing (RR 0.61, 0.24–1.58; 3 studies), significantly so only under a common-effect model. Certainty was low for overall survival and very low for the other outcomes. Conclusion: Pembrolizumab doses 25–50% below label in non-small cell lung cancer were not associated with worse survival, but non-inferiority is not established and comparative data on ultra-low nivolumab doses are almost absent. Recommendation: Fully published randomised non-inferiority trials and prospective head-to-head studies of ultra-low doses in low- and middle-income countries are needed before reduced dosing becomes standard. Future studies should report outcomes by actual mg/kg exposure