The Predictive Role of Preoperative Malnutrition Assessment in Postoperative Outcomes of Patients Undergoing Surgery Due to Gastrointestinal Cancer: A Cross-Sectional Observational Study

Background: Malnutrition affects patients undergoing surgery for gastrointestinal cancers and contributes to poor postoperative outcomes, including increased complication rates, longer hospital stays, and higher mortality. Despite the availability of several malnutrition screening tools and prognostic scores, their effectiveness in predicting postoperative outcomes remains unclear. This study aimed to compare the predictive accuracy of Patient-Generated Subjective Global Assessment (PG-SGA), Global Leadership Initiative on Malnutrition (GLIM), Geriatric Nutritional Risk Index (GNRI), and Controlling Nutritional Status (CONUT) score for postoperative outcomes in patients undergoing surgery for colorectal, hepato-pancreato-biliary and upper gastrointestinal cancers. Methods: A cross-sectional observational study from March 2022 to October 2023 was conducted in two university surgical departments, after registration on ClinicalTrials database (NCT05795374). Patient characteristics, preoperative nutritional status and postoperative outcomes were analyzed. Results: In total, 480 patients were enrolled. CONUT and GNRI demonstrated high specificity (over 90% and 80%, respectively) for predicting overall complications, major complications, prolonged hospital stay, mortality, and advanced disease stage across all cancer types. Notably, CONUT showed a specificity over 97% and GNRI over 89.7% for colorectal and upper gastrointestinal cancer patients, respectively, despite their lower sensitivity. On the contrary, PG-SGA and GLIM presented better sensitivity (up to 50%), but slightly lower specificity (up to 86.4%). Conclusions: CONUT and GNRI are valuable for ruling out non-at-risk patients for adverse postoperative outcomes, while PG-SGA and GLIM provide better sensitivity. A step-up approach-initial screening with PG-SGA and GLIM, followed by detailed evaluation with CONUT or GNRI- should be validated in future studies across diverse clinical settings.