Preoperative prognostic nutritional index as a predictive factor for postoperative pneumonia in esophageal cancer patients undergoing esophagectomy

Chen ChenChenglin Li^*

• The Affiliated Huai'an No. 1 People's Hospital of Nanjing Medical University, Huai'an, China, Huai'an, China

Background: Postoperative pneumonia (POP) remains a serious complication following esophagectomy for esophageal cancer (EC) patients, contributing to increased morbidity, mortality, and healthcare costs. This study aimed to evaluate whether preoperative prognostic nutritional index (PNI) could be an independent predictor of POP in EC patients. Methods: This study included 200 EC patients who underwent esophagectomy between January 2021 to December 2022. Receiver operating characteristic (ROC) curve analysis was conducted to assess the predictive ability of preoperative PNI for POP. Univariate and multivariate logistic regression analyses were used to identify risk factors for POP among EC patients. A predictive nomogram model was conducted. The performance of the nomogram model was evaluated by the AUC curve, calibration curve and decision curve analysis (DCA). Results: Two hundred EC patients receiving esophagectomy were included finally, and 73 (36.5%) cases developed POP. ROC curve analysis showed that preoperative PNI predicted the occurrence of POP with an AUC value of 0.602 at a cut-off value of 49.6; the sensitivity, specificity, and Youden index was 64.38%, 63.78%, 0.2716, respectively. Univariate logistic regression analysis showed that male, aged ≥60 years old, TNM stage III, tumor location, hospital stay time >16 days, WBC counts >5.62×10⁹/L, neutrophil counts >3.52×10⁹/L, monocyte counts >0.40×10⁹/L, and preoperative PNI ≤ 49.6 were risk factors for POP. Multivariate logistic regression analysis indicated that tumor location, hospital stay time >16 days, WBC counts >5.62×10⁹/L, monocyte counts >0.40×10⁹/L, and preoperative PNI ≤ 49.6 were significant risk factors for POP among EC patients receiving esophagectomy. A nomogram model was established. The ROC curve incorporating PNI showed an excellent discrimination in detecting POP with an AUC value of 0.831 (95% CI: 0.772–0.890). The calibration curve suggested that the predicted results of this nomogram model exhibited a good concordance with the actual results. The DCA indicated that this nomogram model achieved net benefits for predicting POP. Conclusion: Preoperative PNI is a significant predictive factor for the occurrence of POP in EC patients. The nomogram model incorporating preoperative PNI shows good accuracy and clinical practicality in predicting the occurrence of POP among EC patients.