PG-SGA, GLIM, and CONUT in Hepatopancreatobiliary Cancer: Complementary Tools for Different Clinical Questions

Background/Objectives: Malnutrition is among the most consequential yet inconsistently addressed comorbidities in hepatopancreatobiliary (HPB) cancer, arising at the intersection of cachexia, sarcopenia, systemic inflammation, hepatic dysfunction, and biliary obstruction. The Patient-Generated Subjective Global Assessment (PG-SGA), the Global Leadership Initiative on Malnutrition (GLIM) framework, and the Controlling Nutritional Status (CONUT) score are often used interchangeably, but they serve different purposes: nutritional assessment, malnutrition diagnosis, and prognostic risk stratification, respectively. This narrative review compares what the three tools measure, summarizes the HPB-specific evidence for each, and proposes a framework for matching the tool to the clinical question. Methods: A structured narrative review was performed. PubMed, Scopus, and Google Scholar were searched for studies applying PG-SGA, GLIM, or CONUT in pancreatic cancer, hepatocellular carcinoma, or biliary tract cancer. HPB-specific evidence was prioritized and the evidence supporting each tool–tumor combination was classified by directness and study design. Results: Applied to the same populations, the tools report different prevalences and classify different patients. In the few HPB cohorts that directly compare tools, PG-SGA classified the largest proportion of patients as malnourished and is the only one of the three designed to record nutrition-impact symptoms and to guide intervention, but it is effort-dependent and may underdetect occult sarcopenia in cirrhosis. GLIM yields a criteria-based diagnosis with severity grading whose prevalence and prognostic value depend on how the muscle mass criterion is operationalized, for example from staging computed tomography. CONUT, computed from three routine laboratory values, is the least resource-intensive and is associated with survival across HPB tumor types in predominantly retrospective cohorts, but its components are confounded by hepatic dysfunction, cholestasis, and inflammation. Conclusions: PG-SGA, GLIM, and CONUT are complementary rather than interchangeable. Because direct comparative evidence in HPB cancer is limited, we propose a framework for matching each tool to the clinical question, tumor subtype, treatment phase and available resources, which requires prospective validation.