Neoadjuvant therapy in resectable cutaneous melanoma: a systematic review and meta-analysis of 35 trial arms

Neoadjuvant systemic therapy is used for resectable stage III and oligometastatic stage IV cutaneous melanoma and is under active evaluation in high-risk node-negative stage II disease, but response and toxicity estimates vary. To pool pathologic complete response (pCR), major pathologic response (MPR), and grade 3 or higher adverse events and assess toxicity-response associations. PubMed, EMBASE, and the Cochrane Library were searched for prospective neoadjuvant cutaneous melanoma studies from 1998 through 2026. Arm-level event rates were pooled using Freeman-Tukey transformation and DerSimonian-Laird random-effects models. Heterogeneity was assessed with I2. Correlation and weighted regression evaluated associations between toxicity and response. Thirty-five trial arms were included. Pooled pCR was 32.2% [95% confidence interval (CI): 23.4-41.5%; 249/738; I2 = 83.5%], MPR was 58.4% (95% CI: 51.9-64.8%; 528/873; I2 = 66.4%), and grade 3 or higher toxicity was 12.4% (95% CI: 6.2-20.2%; 158/887; I2 = 87.1%). Sensitivity analyses yielded similar estimates. pCR was not significantly correlated with serious toxicity (ρ = 0.36; P = 0.18), whereas MPR showed a moderate correlation (ρ = 0.53; P = 0.023). Serious toxicity was concentrated in cytotoxic T-lymphocyte antigen 4 (CTLA-4)-containing combinations. Neoadjuvant therapy achieved pCR in 3 in 10 patients and MPR in more than half, with grade 3 or higher toxicity in about 1 in 8. MPR, but not pCR, correlated with serious toxicity, indicating that pathologic efficacy and serious toxicity are not tightly coupled across the regimens analyzed, and that the excess toxicity observed was attributable to CTLA-4 blockade rather than to the number of checkpoints targeted. Time-dependent survival endpoints were not analyzed.