Development and Internal Validation of a Novel Prognostic Score in Metastatic Colorectal Cancer: A Comparative Retrospective Cohort Study with the Glasgow Prognostic Score and Gustave Roussy Immune Score

Background: Reliable prognostic stratification is essential in metastatic colorectal cancer (mCRC), yet current inflammation-based scores do not fully integrate systemic inflammation, tumor burden, and clinical characteristics. Objective: This study aimed to develop a novel prognostic score integrating clinical, tumor burden, and inflammatory parameters in patients with mCRC and to compare its performance with the Glasgow Prognostic Score (GPS) and Gustave Roussy Immune Score (GRIm). Materials and Methods: This retrospective single-center study included 310 patients with mCRC treated between 2015 and 2025. GPS and GRIm were calculated. The novel prognostic score was constructed using seven equally weighted prognostic variables encompassing clinical characteristics, tumor burden, and laboratory parameters. Survival analyses were performed using Kaplan–Meier and Cox regression. Model performance was assessed using AUC, reclassification, calibration, and decision curve analysis. Results: Most patients had ECOG PS 0–1 (96.1%), left-sided tumors (76.1%) and synchronous metastases (79.0%). The median follow-up was 23.6 months. Both GPS and GRIm were independent prognostic factors for OS. GPS demonstrated superior discriminative performance compared with GRIm (AUC for 24-month OS: 0.697 vs. 0.620; p = 0.005). The novel score stratified median OS into low-, intermediate-, and high-risk groups as 40.9, 25.4, and 11.5 months, respectively (p < 0.001). The novel model outperformed GPS and GRIm (AUC: 0.768) and improved reclassification. Bootstrap-based internal validation demonstrated low optimism and good calibration. Conclusions: The novel prognostic score integrating clinical, inflammatory, and tumor burden parameters demonstrated superior prognostic performance compared with GPS and GRIm in mCRC. This score may serve as tool for individualized risk stratification.