Dynamic prognostic models for colorectal cancer with liver metastases

Current prognostic models for colorectal liver metastases (CRLM) primarily incorporate clinicopathologic features assessed at a single time point, resulting in a statistic risk assessment for individuals. Tumor progression is a dynamic process and patients’ data are continuously collected during follow-up visits. In a retrospective prognostic study of 976 patients with CRLM, dynamic prediction models incorporating multiple longitudinal markers consistently outperformed models based on static clinicopathologic and baseline laboratory data in predicting progression-free and overall survival, with improved areas under the receiver operating characteristic curve and Brier scores across multiple time points.