Abstract
Background: The gut microbiome plays a pivotal role in shaping systemic immunity and modulating anti-tumor responses. Preclinical and clinical studies have shown that higher gut microbial diversity and the presence of specific commensal taxa correlate with improved responses to immune checkpoint inhibitors (ICI) in melanoma. Conversely, broad-spectrum antibiotics can induce dysbiosis, reducing T cell activation and cytokine production, and have been linked to diminished ICI efficacy in several cancer types. Methods: We conducted a systematic review and meta-analysis of seven retrospective cohorts (total n = 5213) comparing overall survival in cutaneous melanoma (CM) patients who did or did not receive systemic antibiotics within six weeks before ICI initiation. From each study, we extracted hazard ratios (HRs) for death, antibiotic-to-ICI interval, ICI regimen (PD-1 monotherapy vs. PD-1 + CTLA-4 combination), cohort size, and country. Pooled log-HRs were estimated under fixed-effect and random-effects (REML) models. Statistical heterogeneity was quantified by Cochran's Q and I(2) statistics, and τ(2). We performed leave-one-out sensitivity analyses, generated a Baujat plot to identify influential studies, applied trim-and-fill to assess publication bias, and ran meta-regressions for regimen, antibiotic timing, sample size, and geography. Results: Under the fixed-effect model, antibiotic exposure corresponded to a pooled HR of 1.26 (95% CI 1.13-1.41; p < 0.001). The random-effects model yielded a pooled HR of 1.55 (95% CI 1.21-1.98; p = 0.0005) with substantial heterogeneity (Q = 25.1; I(2) = 76%). Prediction intervals (0.78-3.06) underscored between-study variability. Leave-one-out analyses produced HRs from 1.50 to 1.75, confirming robustness, and the Baujat plot highlighted two cohorts as primary heterogeneity drivers. Trim-and-fill adjusted the HR to 1.46 (95% CI 1.08-1.97). In subgroup analyses, combination therapy studies (k = 4) showed a pooled HR of ~1.9 (I(2) = 58%) versus ~1.3 (I(2) = 79%) for monotherapy. Meta-regression attributed the largest variance to the regimen (R(2) = 32%; β(monotherapy) = -0.35; p = 0.13). Conclusions: Pre-ICI antibiotic use in CM is consistently associated with a 26-55% increase in mortality risk, particularly with PD-1 + CTLA-4 combinations, reinforcing the mechanistic link between microbiome integrity and ICI success. Looking ahead, integrating prospective microbiome profiling into clinical trials will be critical to personalize ICI therapy, clarify causality, and identify microbial biomarkers for optimal treatment selection. Prospective, microbiome-integrated trials promise to refine melanoma immunotherapy by tailoring antibiotic stewardship and microbial interventions to enhance patient outcomes.