RESUMEN
Patients with acute-on-chronic hepatitis B liver failure (HBV-ACLF) show high morbidity and mortality. Independent prognostic predictors of short-term HBV-ACLF mortality include the Child-Turcotte-Pugh (CTP) score, the model for end-stage liver disease (MELD) score, other MELD-based indices and the dynamic changes in these indices. The aims of this study were to evaluate the existing prognostic scores in a large cohort of HBV-ACLF patients and create a new predictive model. We retrospectively reviewed 392 HBV-ACLF patients from December 2008 to November 2011 and evaluated their 3-month survival. The predictive accuracy of CTP, MELD and MELD-based indices and the dynamic changes in the MELD-related scores (Δ scoring systems) upon admission and after two weeks of treatment were compared using the area under the receiver operating characteristic (ROC) curve method. Life-threatening factors and a series of bio-clinical parameters were studied by univariate and multivariate analyses. Among the existing scores, MELD had the best predictive ability. However, our new regression model provided an area under the curve of 0.930±0.0161 (95% CI: 0.869 to 0.943), which was significantly larger than that obtained with the MELD score at admission and after two weeks of treatment as well as with the dynamic changes of the MELD score (0.819, 0.921, and 0.826, respectively) (Z=3.542, P=0.0004). In a large cohort of patients retrospectively reviewed for this study, our prognostic model was superior to the MELD score and is, therefore, a promising predictor of short-term survival in patients with HBV-ACLF.
RESUMEN
Patients with acute-on-chronic hepatitis B liver failure (HBV-ACLF) show high morbidity and mortality. Independent prognostic predictors of short-term HBV-ACLF mortality include the Child-Turcotte-Pugh (CTP) score, the model for end-stage liver disease (MELD) score, other MELD-based indices and the dynamic changes in these indices. The aims of this study were to evaluate the existing prognostic scores in a large cohort of HBV-ACLF patients and create a new predictive model. We retrospectively reviewed 392 HBV-ACLF patients from December 2008 to November 2011 and evaluated their 3-month survival. The predictive accuracy of CTP, MELD and MELD-based indices and the dynamic changes in the MELD-related scores (Δ scoring systems) upon admission and after two weeks of treatment were compared using the area under the receiver operating characteristic (ROC) curve method. Life-threatening factors and a series of bio-clinical parameters were studied by univariate and multivariate analyses. Among the existing scores, MELD had the best predictive ability. However, our new regression model provided an area under the curve of 0.930 ± 0.0161 (95% CI: 0.869 to 0.943), which was significantly larger than that obtained with the MELD score at admission and after two weeks of treatment as well as with the dynamic changes of the MELD score (0.819, 0.921, and 0.826, respectively) (Z=3.542, P=0.0004). In a large cohort of patients retrospectively reviewed for this study, our prognostic model was superior to the MELD score and is, therefore, a promising predictor of short-term survival in patients with HBV-ACLF.