Clinically interpretable nomogram combining body composition and clinicopathological features for one year survival prediction in advanced solid tumors

Immune checkpoint inhibitors (ICIs) have improved outcomes for patients with solid tumors, but reliable predictors of overall survival (OS) are limited. This retrospective study of 146 advanced solid tumor patients treated with ICIs aims to provide a nomogram to predict 1-year (1y) OS integrating body composition (BC) parameters with standard clinicopathological (CP) features. A two-stage approach was implemented: first random survival forest models were trained and tested to evaluate the prognostic value of (a) CP features alone, (b) BC metrics alone or as newly introduced BC scores, and (c) their combination. The best predictive performance (average cumulative AUC of 0.73 in test set) was achieved by combining 12 CP features with the BC score comprising intramuscolar adipose tissue content, visceral fat area index, and the visceral-to-subcutaneous fat area index ratio. Finally, a nomogram was developed with this feature set, offering a tool for personalized risk stratification and treatment planning (mean absolute error in calibration curve of 0.03 and overall AUC of 0.76). Integrating BC parameters with CP features substantially enhances 1y OS prediction in patients receiving ICIs.