Imaging Biomarkers for Immunotherapy Response in Advanced Lung Cancer 9139

Description Background Immunotherapy has revolutionized advanced lung cancer treatment, but predicting response remains challenging. Objective: To identify image biomarkers associated with immunotherapy response in patients with advanced lung cancer. Materials and Methods A retrospective cohort of 210 advanced lung cancer patients treated with pembrolizumab, atezolizumab, or nivolumab was analyzed. Pretreatment chest CT scans were processed using AI algorithms to extract features related to body composition (e.g., visceral adipose tissue, skeletal muscle volume, bone density), lung characteristics (e.g., lung and airway volumes, emphysema), and circulatory parameters (e.g., blood vessel volumes). Univariate and multivariate Cox proportional hazards models were used to evaluate the relationship between these image biomarkers and immunotherapy outcomes. Results Significant predictors of favorable response to immunotherapy included higher skeletal muscle volume (HR: 0.24, p = 0.0286), higher bone density (HR: 1.28, p = 0.0288), and lower vein volume (HR: 0.45, p < 0.0001). Additionally, elevated visceral adipose tissue ratio (HR: 1.72, p = 0.0051) and circulatory features such as blood vessel volume under 5 mm² (HR: 0.02, p = 0.0235) demonstrated strong predictive associations. A composite model integrating these features showed a mean time-dependent AUC of 0.78. Conclusion CT-derived image biomarkers provide valuable insights into predicting immunotherapy response in advanced lung cancer. Funding Sources N/A Topic Categories Computational and Systems Immunology (COMP)