Background:The geriatric nutritional risk index (GNRI) is an important determinant of overall survival (OS) in patients with stage I-III gastric cancer (GC) across all ages; however, its value as a determinant of disease-free survival (DFS) is unclear. Moreover, the prognostic values between the GNRI and prognostic nutritional index (PNI) remains unclear. Methods: We retrospectively evaluated the value of the GNRI and PNI as determinants of OS and DFS in patients with stage I-III GC who underwent curative-intent gastrectomy. Cox regression analysis was used for evaluating the determinants of survival outcomes. The discriminative capacity of the prognostic model was determined using the concordance index (C-index), and then C-indices of related models were compared. Results: Data from 450 patients were analyzed. The median patient age was 60 years (range: 26-92 years). In total, 276 (61.3%) patients had stage I cancer, 83 (18.4%) had stage II cancer, and 91 (20.2%) had stage III cancer. Multivariate Cox regression analysis revealed that age, type of gastrectomy (TOG), T stage, tumor-node-metastasis (TNM) stage, and GNRI were determinants of OS. These five covariates constituted the GNRI model for the OS. In addition, multivariate analysis revealed that age, TOG, TNM stage, and GNRI were determinants of DFS. These four covariates constituted the GNRI model for DFS. When constructing the PNI model for OS (comprising age, TOG, T stage, TNM stage, and PNI), and PNI model for DFS (including age, TOG, TNM stage, and PNI), the C-indices of the GNRI and PNI models were nearly equal for OS (0.818 and 0.818, respectively; p=0.909) and DFS (0.805 and 0.808, respectively; p=0.653). Using the GNRI models, nomograms for predicting OS and DFS were established. When validating the nomograms using calibration curves, the predicted survival closely matched the actual survival rate.
Conclusion:The GNRI and PNI were important determinants of both OS and DFS in patients with GC across all ages. In addition, the effects of the GNRI model on OS and DFS were similar to those of the PNI model.