Development of a prognostic model for major postoperative complications in gastric cancer patients undergoing laparoscopic gastrectomy

Abstract

Early screening for gastric cancer (GC) patients undergoing laparoscopic gastrectomy (LG) at high risk for major postoperative complications is necessary for effective interventions. In this study, univariate analysis and LASSO regularized logistic regression were performed to screen predictors of major postoperative complications, and a predictive nomogram was established. The performance of the nomogram was validated with receiver operating curves, calibration curves, decision curve analysis and SHapley Additive exPlanations (SHAP). In this study, a total of 410 GC patients undergoing LG were included, of whom 48 patients (11.71%) developed major postoperative complications. Multivariate analysis showed that age [Odds Ratio (OR) = 1.043], gender (OR = 0.328) and preoperative skeletal muscle index (SMI) (OR = 0.848) were independent risk factors for major postoperative complications. These variables were included in the nomogram. The concordance index of the model was 0.855 in the training cohort and 0.919 in the temporal validation. Among all features, SMI exhibited the highest global average absolute SHAP value. This individualized prognostic nomogram, established based on age, gender and SMI to predict the risk of major postoperative complications in GC patients undergoing LG, exhibited satisfactory predictive performance for major postoperative complications. The selected predictors were readily accessible in routine clinical practice and could be used as a reference for clinical decision-making. SMI showed superior predictive performance compared with conventional BMI, and emerged as the most powerful preoperative predictive indicator in this model.