Critérios do Global Leadership Initiative on Malnutrition (GLIM) para diagnóstico de desnutrição e predição de mortalidade em pacientes em lista de espera para transplante hepático: estudo de validação

Abstract

Malnutrition is a common finding and a mayor clinical relevance condition in patients with chronic liver disease. Despite this, nutritional assessment in these patients is difficult. Thus, identifying new tools that help in the early identification of malnourished patients is necessary. The Global Leadership Initiative on Malnutrition (GLIM) is a proposal that aims to operationalize the diagnosis of malnutrition. Although, is not yet validated for patients with chronic liver disease. The present study aimed to validate the GLIM criteria in patients with liver cirrhosis. This is a retrospective observational study that was carried out with adult patients on the waiting list for liver transplantation (LTx) consecutively evaluated between 2006 and 2021. Two types of validation were performed: concurrent, which compares malnutrition prevalence of different combinations of GLIM criteria with a semigold standard nutritional assessment tool (Subjective Global Assessment (SGA); and predictive, which assesses the ability of the GLIM criteria to predict mortality. Were performed 43 different combinations of GLIM criteria (that varied from one combination of each criterion to combinations of all criteria). The phenotypic criteria evaluated were involuntary weight loss, low body mass index and reduced muscle mass (mid-arm muscle circumference – MAMC). The etiologic criteria included were Model for End-Stage Liver Disease (MELD) and MELD adjusted for serum sodium (MELD-Na), Child-Pugh score, serum albumin and food intake/assimilation. Sensitivity (SE), specificity (SP), positive and negative predictive value between SGA and GLIM models and machine learning techniques were used to investigate the ability to diagnose malnutrition, considering SGA as pattern. The occurrence of death in one year was evaluated in survival analyzes with a significance level of 5%. Were included 419 patients with median age of 52.0 (46.0 – 59.0) years, 69.2% were male and the main LTx indication was alcoholic cirrhosis (55.4%). Malnutrition prevalence ranged from 2.2% to 58.2% between the GLIM models and 93% of the combinations has SE and/or SP < 50%. Combinations with MAMC or considering the presence of any phenotypic criteria associated with liver disease parameters were important for malnutrition predict in machine learning analysis. In terms of predictive validation, 15 GLIM models were predictors of mortality. However, only nine had HR≥2.0, mostly combinations that had MAMC as phenotypic criterion and MELD and MELD-Na as etiologic criterion. Combinations of criteria with MAMC and liver disease parameters were associated with the malnutrition diagnosis by SGA and were predictors of one-year mortality.