Postoperative survival nomograms integrating preoperative eosinophil-to-lymphocyte ratio and clinicopathological factors in gastric cancer

Survival varies among patients with gastric cancer despite having similar pathological stages. We developed postoperative nomograms integrating routinely available preoperative inflammatory markers with clinicopathological factors while preserving continuous predictor information. All 405 patients from Shandong were used for model development and 1,000-resample bootstrap validation, and 108 patients from Tianjin formed an independent external validation cohort. Cox models for overall survival (OS) and disease-free survival (DFS) retained age, histological differentiation, nonlinear tumor size, nonlinear lymph-node ratio, and continuous eosinophil-to-lymphocyte ratio (ELR). In this retrospective cohort, higher ELR was inversely associated with the hazard of death (hazard ratio [HR], 0.690; 95% confidence interval [CI], 0.546–0.872) and DFS events (HR, 0.720; 95% CI, 0.578–0.898). Fixed-specification bootstrap-corrected C-indices were 0.772 for OS and 0.777 for DFS; external C-indices were 0.767 and 0.756. At 5 years, external areas under the curve (AUCs) were 0.847 for OS and 0.817 for DFS, exceeding pathological tumor-node-metastasis (pTNM) values by 0.085 and 0.102, respectively. External calibration showed underestimation of absolute event risk. These findings support the preliminary prognostic value of the nomograms, but recalibration and broader prospective validation are required before clinical use.