A personalized prognostic model based on preoperative body composition and nutritional parameters for gastric cancer patients receiving neoadjuvant chemotherapy

The long-term survival of patients with locally advanced gastric cancer (LAGC) undergoing neoadjuvant chemotherapy (NAC) remains suboptimal. In this retrospective study, we analyzed 403 NAC-LAGC patients followed by radical gastrectomy between January 2016 and December 2023. The cohort was randomly divided into a training set and a validation set in a 7:3 ratio. Variables with a univariable P value below 0.20 were first identified, and LASSO regression together with stepwise Cox proportional hazards regression was then applied to screen the candidate predictors (p < 0.10). This process yielded the final set of predictive factors based on multidimensional indicators related to nutritional status and body composition. Using the training set, we constructed separate nomograms for predicting overall survival, progression-free survival, and disease-free survival, and then developed a corresponding risk stratification model. Model performance was assessed with Kaplan-Meier survival analyses and the area under the receiver operating characteristic curve, and was further examined in the validation set. Through feature selection, we identified several independent prognostic predictors. Kaplan-Meier survival analyses confirmed that each variable was significantly associated with poor prognosis (p < 0.01). Based on these predictors, we first constructed individual nomograms to predict OS, PFS, and DFS, all of which achieved favorable discriminative performance with AUC values exceeding 0.800. To further enhance risk stratification, we subsequently developed a comprehensive prognostic risk stratification model (PRSM). The PRSM demonstrated robust and reliable predictive ability: in the training cohort, all AUCs were above 0.800 (p < 0.001), with a c-index of 0.836; in the validation cohort, AUCs similarly exceeded 0.800 (p < 0.001), with a c-index of 0.829. Decision curve analysis further indicated that, within an appropriate threshold range, the PRSM provided meaningful clinical net benefit for NAC-LAGC patients. In conclusion, we developed and validated PRSM that incorporates multidimensional predictors reflecting nutritional status and body composition to estimate long-term outcomes in NAC-LAGC patients. The model provides reliable risk stratification and may serve as a practical tool to support individualized nutritional optimization and postoperative management in clinical practice.