Published online Aug 28, 2015. doi: 10.3748/wjg.v21.i32.9614
Peer-review started: April 4, 2015
First decision: April 23, 2015
Revised: May 21, 2015
Accepted: July 8, 2015
Article in press: July 8, 2015
Published online: August 28, 2015
Processing time: 147 Days and 17.3 Hours
AIM: To establish a new model for predicting survival in acute-on-chronic liver failure (ACLF) patients treated with an artificial liver support system.
METHODS: One hundred and eighty-one ACLF patients who were admitted to the hospital from January 1, 2012 to December 31, 2014 and were treated with an artificial liver support system were enrolled in this retrospective study, including a derivation cohort (n = 113) and a validation cohort (n = 68). Laboratory parameters at baseline were analyzed and correlated with clinical outcome. In addition to standard medical therapy, ACLF patients underwent plasma exchange (PE) or plasma bilirubin adsorption (PBA) combined with plasma exchange. For the derivation cohort, Kaplan-Meier methods were used to estimate survival curves, and Cox regression was used in survival analysis to generate a prognostic model. The performance of the new model was tested in the validation cohort using a receiver-operator curve.
RESULTS: The mean overall survival for the derivation cohort was 441 d (95%CI: 379-504 d), and the 90- and 270-d survival probabilities were 70.3% and 58.3%, respectively. The mean survival times of patients treated with PBA plus PE and patients treated with PE were 531 d (95%CI: 455-605 d) and 343 d (95%CI: 254-432 d), respectively, which were significantly different (P = 0.012). When variables with bivariate significance were selected for inclusion into the multivariate Cox regression model, number of complications, age, scores of the model for end-stage liver disease (MELD) and type of artificial liver support system were defined as independent risk factors for survival in ACLF patients. This new prognostic model could accurately discriminate the outcome of patients with different scores in this cohort (P < 0.001). The model also had the ability to assign a predicted survival probability for individual patients. In the validation cohort, the new model remained better than the MELD.
CONCLUSION: A novel model was constructed to predict prognosis and accurately discriminate survival in ACLF patients treated with an artificial liver support system.
Core tip: Liver failure has a high mortality. The current prognostic model to estimate the survival in acute-on-chronic liver failure (ACLF) patients treated with an artificial liver support system (ALSS) is not fully characterized. The aim of this study was to establish a new scoring model and to test its ability to predict the survival of ACLF patients treated with ALSS. This prognostic model accurately differentiated the outcome of ACLF patients with different risk scores and also had the ability to assign a predicted survival probability for individual patients.