Luis Calzadilla-Bertot1, Eduardo Vilar-Gomez2,3, Vincent Wai-Sun Wong4, Manuel Romero-Gomez3, Rocio Aller-de la Fuente5, Grace Lai-Hung Wong4, Marlen Castellanos6, Mohammed Eslam7, Archita P Desai2, Gary P Jeffrey1,8, Jacob George7, Naga Chalasani2, Leon A Adams1,8. 1. Medical School, Faculty of Health and Medical Sciences, The University of Western Australia, Nedlands, WA, Australia. 2. Division of Gastroenterology and Hepatology, Indiana University School of Medicine, Indianapolis, IN. 3. Unit for the Clinical Management of Digestive Diseases, Centro para la Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD), Virgen del Rocio University Hospital, University of Seville, Seville, Spain. 4. Institute of Digestive Disease, The Chinese University of Hong Kong, Hong Kong, China. 5. Department of Digestive Disease, Institute of Endocrinology and Nutrition, University of Valladolid, Valladolid, Spain. 6. Department of Hepatology, National Institute of Gastroenterology, Havana, Cuba. 7. Storr Liver Centre, Westmead Institute for Medical Research, Westmead Hospital and University of Sydney, Sydney, NSW, Australia. 8. Department of Hepatology, Sir Charles Gairdner Hospital, Nedlands, WA, Australia.
Abstract
BACKGROUND AND AIMS: Nonalcoholic fatty liver disease (NAFLD) is an increasingly important cause of liver cirrhosis and subsequent complications. We retrospectively developed and validated a model to predict hepatic decompensation in patients with NAFLD and cirrhosis and compared this with currently available models. APPROACH AND RESULTS: Baseline variables from an international cohort of 299 patients with biopsy-proven NAFLD with compensated cirrhosis were examined to construct a model using competing risk multivariate regression and Akaike/Bayesian information criteria. Validation was performed in 244 patients with biopsy-proven NAFLD cirrhosis from the United States. Prognostic accuracy was compared with the NAFLD fibrosis score (NFS), fibrosis-4 (FIB-4), Model for End-Stage Liver Disease (MELD), Child-Turcotte-Pugh (CTP), and albumin-bilirubin (ALBI)-FIB-4 score using time-dependent area under the curve (tAUC) analysis. During a median follow-up of 5.6 years (range 2.4-14.1) and 5.4 years (range 1.5-13.8), hepatic decompensation occurred in 81 and 132 patients in the derivation and validation cohorts, respectively. In the derivation cohort, independent predictors of hepatic decompensation (Aspartate aminotransferase/alanine aminotransferase ratio, Bilirubin, International normalized ratio, type 2 Diabetes, and Esophageal varices) were combined into the ABIDE model. Patients with a score ≥4.1 compared with those with a score <4.1 had a higher risk of decompensation (subhazard ratio, 6.7; 95% confidence interval [CI], 4.0-11.2; P < 0.001), a greater 5-year cumulative incidence (37% vs. 6%, P < 0.001), and shorter mean duration to decompensation (3.8 vs 6.7 years, P < 0.001). The accuracy of the ABIDE model at 5 years was good in the derivation (tAUC, 0.80; 95% CI, 0.73-0.84) and validation cohorts (0.78; 95% CI, 0.74-0.81) and was significantly more accurate than the NFS (0.72), FIB-4 (0.74), MELD (0.69), CTP (0.72), and ALBI-FIB-4 (0.73) (all P < 0.001). CONCLUSIONS: In patients with NAFLD and compensated cirrhosis, ABIDE, a predictive model of routine clinical measures, predicts future hepatic decompensation.
BACKGROUND AND AIMS: Nonalcoholic fatty liver disease (NAFLD) is an increasingly important cause of liver cirrhosis and subsequent complications. We retrospectively developed and validated a model to predict hepatic decompensation in patients with NAFLD and cirrhosis and compared this with currently available models. APPROACH AND RESULTS: Baseline variables from an international cohort of 299 patients with biopsy-proven NAFLD with compensated cirrhosis were examined to construct a model using competing risk multivariate regression and Akaike/Bayesian information criteria. Validation was performed in 244 patients with biopsy-proven NAFLD cirrhosis from the United States. Prognostic accuracy was compared with the NAFLD fibrosis score (NFS), fibrosis-4 (FIB-4), Model for End-Stage Liver Disease (MELD), Child-Turcotte-Pugh (CTP), and albumin-bilirubin (ALBI)-FIB-4 score using time-dependent area under the curve (tAUC) analysis. During a median follow-up of 5.6 years (range 2.4-14.1) and 5.4 years (range 1.5-13.8), hepatic decompensation occurred in 81 and 132 patients in the derivation and validation cohorts, respectively. In the derivation cohort, independent predictors of hepatic decompensation (Aspartate aminotransferase/alanine aminotransferase ratio, Bilirubin, International normalized ratio, type 2 Diabetes, and Esophageal varices) were combined into the ABIDE model. Patients with a score ≥4.1 compared with those with a score <4.1 had a higher risk of decompensation (subhazard ratio, 6.7; 95% confidence interval [CI], 4.0-11.2; P < 0.001), a greater 5-year cumulative incidence (37% vs. 6%, P < 0.001), and shorter mean duration to decompensation (3.8 vs 6.7 years, P < 0.001). The accuracy of the ABIDE model at 5 years was good in the derivation (tAUC, 0.80; 95% CI, 0.73-0.84) and validation cohorts (0.78; 95% CI, 0.74-0.81) and was significantly more accurate than the NFS (0.72), FIB-4 (0.74), MELD (0.69), CTP (0.72), and ALBI-FIB-4 (0.73) (all P < 0.001). CONCLUSIONS: In patients with NAFLD and compensated cirrhosis, ABIDE, a predictive model of routine clinical measures, predicts future hepatic decompensation.