New anthropometric and biochemical models for estimating appendicular skeletal muscle mass in male patients with cirrhosis

•The fluid retention common in cirrhosis (mainly ascites) impairs skeletal muscle mass estimation by available simple and accessible tools.•In the present study, we applied anthropometric and biochemical variables to design models for estimation of skeletal muscle mass and validated their applicability in diagnosing sarcopenia in cirrhosis.•Our models showed good accuracy, sensitivity, and specificity in predicting skeletal muscle mass, as well as an excellent accuracy in the prediction of sarcopenia in cirrhosis. The use of easily accessible methods to estimate skeletal muscle mass (SMM) in patients with cirrhosis is often limited by the presence of edema and ascites, precluding a reliable diagnosis of sarcopenia. The aim of this study was to design predictive models using variables derived from anthropometric and/or biochemical measures to estimate SMM; and to validate their applicability in diagnosing sarcopenia in patients with cirrhosis. Anthropometric and biochemical data were obtained from 124 male patients (18–76 y of age) with cirrhosis who also underwent dual-energy x-ray absorptiometry (DXA) and handgrip strength (HGS) assessments to identify low SMM and diagnose sarcopenia using reference cutoff values. Univariate analyses for variable selection were applied to generate predictive decision tree models for low SMM. Model accuracy for the prediction of low SMM and sarcopenia (when associated with HGS) was tested by comparison with reference cutoff values (appendicular SMM index, obtained by DXA) and clinical sarcopenia diagnoses. The prognostic value of the models for the prediction of sarcopenia and mortality at 104 wk of follow up was further tested using Kaplan–Meier graphics and Cox models. The models with anthropometric variables, alone and combined with biochemical variables, showed good accuracy (0.89 [0.83; 0.94] and 0.90 [0.84; 0.95], respectively) and sensitivity (0.72 [0.56; 0.85] and 0.74 [0.59; 0.86], respectively) and excellent specificity (0.96 [0.90; 0.99] and 0.97 [0.92; 0.99], respectively) in predicting SMM. Both models showed excellent accuracy (0.94 [0.89; 0.98], good sensitivity (0.68 [0.45; 0.86]), and excellent specificity (1.00 [0.96; 1.00]) in predicting sarcopenia. The models predicted mortality in patients with sarcopenia, with the likelihood of death sixfold greater relative to patients not predicted to have sarcopenia. Our simple and inexpensive models provided a practical and safe approach to diagnosing sarcopenia p