Abstract
Background: Despite improved survival outcomes with immune checkpoint inhibitors (ICIs), the treatment response of patients with metastatic melanoma remains highly variable. There is a growing need for reliable, accessible prognostic tools that incorporate clinical and inflammatory markers in order to stratify patients better and guide therapeutic decisions. Methods: We conducted a retrospective cohort study involving 73 patients with metastatic cutaneous melanoma treated with ICIs at a single tertiary center between 2017 and 2024. Baseline clinical and laboratory parameters were collected to calculate the Royal Marsden Hospital (RMH), Gustave Roussy Immune (GRIm) and MD Anderson Immune Checkpoint Inhibitor (MDA-ICI) scores. Overall survival (OS) and progression-free survival (PFS) were analyzed via Kaplan-Meier estimates and Cox proportional hazards models. Prognostic performance was assessed using Harrell's concordance index (C-index) and receiver operating characteristic (ROC) analysis. Results: The median follow-up was 35.9 months, and the median OS was 22.1 months. All three scores were significantly associated with OS in univariate analysis. In multivariate models, only the RMH (HR: 5.45, p < 0.001) and MDA-ICI (HR: 4.24, p = 0.015) scores remained independent predictors of OS. Harrell's C-index indicated strong discriminative ability for both RMH (0.742) and MDA-ICI (0.730) scores, whereas the GRIm score demonstrated lower predictive accuracy (0.615). Similarly, ROC curve analysis showed higher AUC values for RMH (0.732) and MDA-ICI (0.739) compared with GRIm (0.595). Conclusions: In this real-world cohort of metastatic melanoma patients treated with ICI, the RMH and MDA-ICI scores demonstrated favorable prognostic performance and outperformed the GRIm score in predicting overall survival. These findings support the clinical utility of RMH and MDA-ICI as practical, accessible tools for prognostic risk stratification in melanoma, though external validation in larger, multicenter cohorts is required.