Ângulo de fase associado ao MELD-Na para predição de mortalidade em pacientes com cirrose descompensada

Abstract

The MELD-Na score is used to assess the severity of cirrhosis and prioritize liver transplant allocation, while the phase angle (PhA) is a nutritional and prognostic marker. Incorporating PhA into the MELD-Na score may improve its predictive performance. This study aimed to evaluate the prognostic impact of the combined use of MELD-Na and PhA in predicting mortality among patients with decompensated cirrhosis. This retrospective observational study used a database from four Brazilian tertiary referral centers. Adults and older adults diagnosed with decompensated cirrhosis (Child-Pugh classes B and C), either hospitalized or followed in the outpatient setting, were included if MELD-Na and bioelectrical impedance analysis (BIA) data were available for PhA assessment. Mortality occurring within one year after the BIA evaluation was considered the primary outcome. Six prognostic indices combining MELD-Na and PhA were developed, using both continuous PhA values and dichotomized PhA values based on five cutoff points previously reported in the literature. Logistic regression was used to calculate the probability of death based on MELD-Na and the proposed indices. Predictive performance was compared using the area under the receiver operating characteristic (ROC) curve (AUC) for the overall sample. Interactions between the continuous and categorical MELD-Na–PhA indices and clinical subgroups were also assessed, including MELD-Na (<15 and ≥15), presence of ascites/edema, liver transplantation, hepatocellular carcinoma, and malnutrition. A total of 452 patients were evaluated, with a median age of 54.5 (48.0–62.0) years; 73.0% were male, 43.6% had alcohol-related cirrhosis, and 71.9% were classified as Child-Pugh B. The median MELD-Na score was 17.0 (14.0–20.0), and the mean PhA was 5.2° ± 1.1 in men and 4.9° (4.3–5.6) in women. The one-year mortality rate was 21.7%. Comparison of the AUCs between MELD-Na alone and the combined MELD-Na–PhA model using a PhA cutoff of ≤5.05° demonstrated improved mortality prediction in the overall sample (AUC = 0.624 vs. 0.678; p < 0.05). No significant interactions were observed between the continuous or categorical MELD-Na–PhA indices and the clinical or nutritional variables evaluated (p > 0.05). In conclusion, the incorporation of PhA into the MELD-Na score improved the prediction of one-year mortality in the overall sample.