Introduction
Delayed bleeding after gastric endoscopic submucosal dissection (ESD) is a serious complication in older adults. The BEST-J score, developed by Hatta et al. (Gut, 2021), is widely used for bleeding risk stratification but does not reflect systemic factors. The Prognostic Nutritional Index (PNI), calculated as [10 × serum albumin (g/dL)] + [0.005 × total lymphocyte count (/μL)], may complement this score. This study evaluated whether combining PNI with the BEST-J score improves risk stratification.
Aims & Methods
We retrospectively analyzed 252 patients who underwent gastric ESD between April 2016 and February 2025, including 127 older (≥75 years) and 125 younger (<75 years) individuals. Patients were classified into low (<40) and high (≥40) PNI groups. The association between delayed bleeding, PNI, and BEST-J score was analyzed using receiver operating characteristic (ROC) analysis. To evaluate internal validity, we conducted 5-fold cross-validation of a logistic regression model that added 1 point for PNI < 40 to the BEST-J score. A cutoff value of PNI < 40 was used, based on previously reported clinical thresholds reflecting impaired nutritional and immune status, including its predictive value for bleeding in various clinical settings (Zhao et al., 2024; Lin et al., 2021).
Results
The delayed bleeding rate was 4.7% in the older group and 0.8% in the younger group. Mean PNI was lower in bleeding cases (44.7) than in non-bleeding cases (48.5). Among older adults, the AUCs for PNI and BEST-J alone were 0.65 and 0.72, respectively. Adding 1 point for PNI < 40 improved the AUC to 0.82. Cross-validation confirmed good model reproducibility, with a mean AUC of 0.861 and standard deviation of 0.147, indicating stable predictive performance without overfitting. The inclusion of PNI < 40 enhanced model sensitivity for identifying bleeding cases without substantially compromising specificity, improving clinical applicability. These associations were most prominent in older patients, supporting age-specific application of the combined model.
Conclusion
The combination of BEST-J and PNI provides an improved and practical approach to risk stratification for delayed bleeding after gastric ESD in older adults. Cross-validation supports the model's internal validity and suggests applicability to broader clinical populations. Given the disproportionately higher bleeding risk in the older cohort, our findings underscore the importance of incorporating host-related factors such as nutrition into stratification strategies specifically in this age group.
References
1. Hatta W, Tsuji Y, Yoshio T, et al. Prediction model of bleeding after endoscopic submucosal dissection for early gastric cancer: BEST-J score. Gut. 2021;70(3):476–484. doi:10.1136/gutjnl-2019-319926.
2. Zhao Y, Chen X, Zhang M, et al. Association of Prognostic Nutritional Index with Post-Discharge Bleeding in Acute Coronary Syndrome Patients on Dual Antiplatelet Therapy. Front Cardiovasc Med. 2024;11:11995920.
3. Lin Y, Wang J, Chen Z, et al. Preoperative Prognostic Nutritional Index Predicts Intraoperative Transfusion and Surgical Outcomes in Hepatocellular Carcinoma Patients Undergoing Hepatectomy. Cancers (Basel). 2021;13(11):2508.
