Image Biomarkers for Predicting Immunotherapy Response in Advanced Melanoma 9065

Abstract Description Background Immune checkpoint inhibitors (ICIs) targeting the negative regulatory checkpoints PD-1 and CTLA-4 have improved outcomes in advanced cutaneous melanoma, yet effective biomarkers to predict ICI response are lacking. Objective: To develop predictive biomarkers of ICI response in advanced melanoma patients treated with ICIs in the first-line setting. Materials and Methods This retrospective study included 157 advanced melanoma patients (mean age: 63.3 years; 65.6% male) who underwent pre-treatment CT imaging and received ICIs. AI algorithms were utilized to compute CT image body composition features, which are standardized across abdominal, chest, pelvic regions, and spinal vertebrae. Associations between CT/clinical features and durable progression-free survival (PFS≥12 months) were assessed using univariate analysis and ML models. Prediction performance was evaluated using the area under the receiver operating curve (AUROC) using stratified 10-fold cross-validation. Results Increased visceral fat density (p = 0.038) was negatively associated with PFS≥12, while increased subcutaneous fat (p = 0.001), muscle (p = 0.005), intermuscular fat (p = 0.007), and bone volumes (p = 0.001) were favorably associated with PFS≥12. An ML model combining three CT-derived features and two clinical features achieved a mean AUROC of 0.83 (95% CI: 0.72-0.94). Conclusion CT-derived body composition features are associated with improved response to ICI therapy in advanced cutaneous melanoma. Funding Sources This work was supported in part by Developmental Research funding through the University of Pittsburgh Melanoma and Skin Cancer SPORE NIH P50 CA254865 and NIH R01CA237277. Topic Categories Translational and Interventional Immunology (TI)