Adverse event profiles of immune checkpoint inhibitor-based strategies for metastatic prostate cancer: a systematic review and network meta-analysis

Background Immune checkpoint inhibitor (ICI)-based combination strategies are increasingly evaluated in metastatic prostate cancer (mPCa), but their comparative safety profiles remain uncertain. We conducted a systematic review and network meta-analysis (NMA) to compare adverse event (AE) profiles of ICI-based strategies in mPCa. Methods PubMed, Embase, Cochrane Library, and Web of Science were searched. Ten randomized trials involving 8,197 participants and seven predefined treatment nodes were analyzed. Outcomes included any-grade AEs, grade ≥3 AEs, serious AEs, fatal AEs, any-cause AEs, and selected organ-specific immune-related AEs (diarrhea, rash, hypothyroidism, fatigue). Random-effects NMA estimated odds ratios (ORs) with 95% confidence intervals (CIs). SUCRA values were considered exploratory only. Transitivity was assessed qualitatively, and sensitivity analysis excluded metastatic hormone-sensitive prostate cancer (mHSPC) trials. Certainty was summarized using a structured qualitative assessment. Results All outcome networks were star-shaped, with no closed loops; therefore, loop inconsistency could not be formally assessed. In the endocrine-therapy subgroup, adding an ICI to an ARPI-based endocrine backbone significantly increased any-grade AEs (OR 1.33, 95% CI 1.19–1.47), grade ≥3 AEs (OR 2.85, 2.36–3.45), serious AEs (OR 5.33, 3.89–7.31), and fatal AEs (OR 8.03, 1.84–35.02), but not any-cause AEs (OR 1.04, 0.92–1.18). Across the NMA, CTLA-4-containing regimens and PD-(L)1 plus endocrine therapy showed less favorable signals for serious and fatal AEs, whereas PD-1 plus chemotherapy ranked relatively favorably in exploratory SUCRA analyses for several outcomes. However, fatal-event estimates were based on few events, several CIs were wide, and sensitivity analysis excluding mHSPC trials changed the precision and, for some comparisons, the direction or statistical significance of estimates. Conclusions In this exploratory NMA, ICI-based combination strategies in mPCa showed heterogeneous safety profiles. Adding ICIs to endocrine therapy was associated with increased AEs. PD-1 plus chemotherapy ranked relatively favorably in exploratory SUCRA analyses, but this should not be interpreted as confirmed safety superiority, and no regimen should be regarded as definitively safest. Certainty was limited by sparse, predominantly star-shaped networks, indirect comparisons, variable AE definitions, and imprecise estimates. Head-to-head randomized trials with standardized safety reporting are needed. Systematic Review Registration PROSPERO identifier CRD420261342375.