A HDL-based cirrhosis risk score improves the prediction of short-term mortality in acutely decompensated cirrhosis

BACKGROUND AND AIMS: Liver failure is associated with severe lipid alterations including pronounced reductions of high-density lipoprotein cholesterol (HDL-C) levels that are also of prognostic value. In the present study we developed an optimized prognostic model based on HDL-C and other readily available blood parameters for survival prediction in patients with acutely decompensated (AD) cirrhosis. METHOD: We measured HDL-C in biobanked plasma samples of patients recruited from the large prospective CANONIC and PREDICT cohorts. Multivariable competing risk analysis was performed with death as the event of interest and liver transplantation (LT) as competing risk. Cox proportional hazards regression was used to construct a new prognostic model and its performance was evaluated using C-index and compared to other prognostic scores by Integrated Discrimination Index statistics test. RESULTS: We analyzed 1035 patients with AD/ACLF (median age 59 y; 70% male; ACLF at inclusion 20%; etiology alcohol 59%). Multivariable analysis yielded 6 independent prognostic variables associated with 90-day survival, i.e. age, HDL-C, creatinine, sodium, WBC, and INR, that were incorporated in a new prognostic model termed CLIF-C HDL score. The new model showed superior discrimination ability for prediction of 90-day mortality by C-index of 0.768 for CLIF-C HDL score vs. 0.735 for MELDNa (p<0.001) vs. 0.738 for MELD 3.0 (p<0.001). This superior performance of CLIF-C HDL score was confirmed in two external validation cohorts (Turin, n=338; Vienna, n=185). CONCLUSION: The new prognostic CLIF-C HDL score yields superior accuracy for prediction of short-term mortality in AD cirrhosis as compared to other prognostic scores.