Abstract
BACKGROUND: A poor prognosis in patients with gastric cancer (GC) is independently linked to malnutrition and CT-defined low muscle mass. However, the combined effects on prognosis outcomes are not fully elucidated. This study systematically evaluated the synergistic effects of body composition parameters and nutritional indicators in predicting the prognosis of gastric cancer. METHODS: This retrospective study included 986 middle-aged and elderly patients with stage II/III GC who underwent surgical resection. Body composition parameters, including the skeletal muscle index (SMI), skeletal muscle density (SMD), and various adipose tissue indices, were evaluated by a single cross-sectional computed tomography image at the L3 level. The combined indices were defined as the product of the geriatric nutritional risk index (GNRI) and body composition parameters. Prognostic analyses were conducted using the Kaplan-Meier method. RESULTS: The 986 patients were divided into a training cohort (n = 690) and a validation cohort (n = 296) at a 7:3 ratio. The median (interquartile) age was 71 (67-75) years, and 623 (63.2%) patients were male. In the training cohort, the median values of the SMI, SMD, adipose tissue parameters, and their combinations with the GNRI differed significantly by sex. Patients were classified into sex-specific quartiles (Q1-Q4) on the basis of SMI × GNRI score. Both overall survival (OS) and disease-free survival (DFS) were significantly different across these quartiles (P < 0.001). Although all body composition parameters and their combinations with the GNRI were independent predictors of OS and DFS according to multivariate analysis, the combination of the SMI × GNRI demonstrated superior prognostic performance compared with other indices in the prediction of OS (c-statistics: 0.749, AICc: 3842.9) and DFS (c-statistics: 0.731, AICc: 4174.7). These results remained consistent across the stratified analyses. The validation cohort confirmed that the SMI × GNRI exhibited greater predictive and discriminative power than the other indices did. CONCLUSIONS: A robust and readily applicable tool for prognostic assessment, the SMI×GNRI index effectively predicts survival outcomes in middle-aged and elderly stage II/III GC patients. Further prospective studies are needed to validate its effectiveness across diverse populations and clinical settings.