Concurrent validity of GLIM criteria in nutritional assessment of surgical patients with colorectal Cancer

Introduction: Malnutrition affects outcomes in surgical treatment of patients with colorectal cancer (CRC). The Global Leadership Initiative on Malnutrition (GLIM) proposed a framework of diagnostic criteria, but it needs validation in different settings. The aim of this study was to evaluate the concurrent validity of GLIM criteria-across all 21 phenotypic-etiologic combinations-against a standard (modified Subjective Global Assessment, mSGA) in surgical patients with CRC. Methods: Prospective cohort of 105 adults scheduled for elective CRC surgery (June 2023-June 2024). Nutritional risk was screened with Nutritional Risk Screening 2002 (NRS-2002). Nutritional status was assessed by mSGA and by GLIM across all 21 combinations. Agreement and classification differences were tested using McNemar's test and Cohen's Kappa; sensitivity and specificity of GLIM (vs mSGA) were calculated. Bootstrap procedures supported inference. Results: = 0.505; 95% CI 0.368-0.656). Nine GLIM combinations did not differ significantly from mSGA by McNemar's test; among these, four combinations - P1EA, P12EA, P13EA, and P123EA - demonstrated significant agreement with mSGA (highest concordance). Across combinations, malnutrition prevalence ranged from 13.3 to 55.2%. Using aggregate GLIM, sensitivity was 96.97% and specificity 63.89% versus mSGA; the best single combination (P12EA - unintentional weight loss, low BMI, and reduced food intake or assimilation) yielded sensitivity 84.85% and specificity 83.33%. Discussion: In elective CRC surgery, several GLIM combinations- particularly P1EA, P12EA, P13EA, and P123EA -show meaningful agreement with mSGA and may offer a more straightforward, pragmatic approach to identify malnutrition preoperatively. Adoption of these combinations could streamline assessment and support timely perioperative nutritional interventions. Larger multicenter studies should confirm these findings and determine predictive value for clinical outcomes.