Dynamic evaluation based on acute-on-chronic liver failure predicts survival of patients after liver transplantation: a cohort study.

Abstract

Dynamic evaluation of critically ill patients is the key to predicting their outcomes. Most scores based on Model for End-stage Liver Disease (MELD) and acute-on-chronic liver failure (ACLF) utilize point-in-time assessment. This study mainly aimed to investigate the impact of dynamic clinical course change on post-liver transplantation survival. This study included 637 adults (overall cohort) with benign end-stage liver diseases. We compared the MELD scores and our ACLF-based dynamic evaluation scores. Patients enrolled or transplanted with ACLF3 were defined as ACLF-3 cohort (n = 158). The primary outcome was 1-year mortality. ΔMELD and ΔCLIF-OF represented the respective dynamic changes in liver transplant function. Discrimination was assessed using the area under the curve (AUC). Cox regression analysis identified independent risk factors for specific organ failure and 1-year mortality. Patients were grouped as deterioration group, stable group, and improvement group. The deterioration group (ΔCLIF-OF ≥ 2) was more likely to receive national liver allocation (P=0.012) but experienced longer cold ischemia time (P=0.006) than other groups. The AUCs for ΔCLIF-OF were 0.752 for the entire cohort and 0.767 for ACLF-3 cohorts, both superior to ΔMELD (P<0.001 for both). Compared to the improvement group, the 1-year mortality hazard ratios (HR) of deterioration group were 12.57 (6.72-23.48) for overall cohort and 7.00 (3.73-13.09) for ACLF-3 cohort. Extrahepatic organs subscore change (HR=1.783 (1.266-2.512) for neurologic; 1.653 (1.205-2.269) for circulation; 1.906 (1.324-2.743) for respiration; 1.473 (1.097-1.976) for renal) were key to transplantation outcomes in the ACLF-3 cohort. CLIF-OF at LT (HR=1.193), ΔCLIF-OF (HR=1.354), and cold ischemia time (HR=1.077) were independent risk factors of mortality for the overall cohort, while ΔCLIF-OF (HR=1.384) was the only independent risk factor for the ACLF-3 cohort. Non-ACLF3 patients showed a higher survival rate than patients with ACLF-3 in all groups (P=0.002 for I, P=0.005 for S and P=0.001 for D). This was the first ACLF-based dynamic evaluation study. ΔCLIF-OF was a more powerful predictor of post-LT mortality than ΔMELD. Extrahepatic organ failures were core risk factors for ACLF-3 patients. CLIF-OF at LT, ΔCLIF-OF, and cold ischemia time were independent risk factors for post-LT mortality. Patients with a worse baseline condition and deteriorating clinical course had the worst prognosis. Dynamic evaluation was important in risk stratification and recipient selection.