Exploration Of The Predictive Value Of Lipid Level On The Short-term Prognosis In Patients With Acute-on-chronic Liver Failure And The Development Of A Predictive Model

Background:Acute-on-chronic liver failure(ACLF)is an acute liver injury on the basis of chronic liver disease,and patients have a high mortality rate in a short period of time.Lipid levels correlate with the severity of liver disease and are useful in the prognostic assessment of cirrhosis and liver failure,but the performance of lipid levels varies in different studies.The predictive value of lipid levels in the short-term prognosis of ACLF deserves further exploration.Objective:This study was designed to explore the predictive value of lipid levels in the shortterm prognosis of ACLF and to develop a new short-term prognostic model based on lipids.Methods:The clinical data of all patients who met the inclusion and exclusion criteria and were hospitalized for ACLF in the First Affiliated Hospital of Nanchang University from January 2018 to December 2021 were retrospectively analyzed.1.The collected data were subjected to minimal absolute contraction and selection operator(LASSO),univariate logistic regression analysis,multicollinearity analysis and correlation analysis for preliminary screening of variables.The variables obtained after screening were subjected to multivariate logistic regression analysis to screen out independent predictors of death at 28 days and 90 days.2.Regression models were constructed for the blood lipid variables alone and also in combination with the MELD score,MELD-Na score,CLIF-C ACLF score,and COSSH-ACLF score,respectively(HDL-MELD,HDL-MELD-Na,HDL-CLIF-C ACLF,HDL-COSSH-ACLF).The area under the ROC curve(AUC),Nagelkerke R2,Brier Score,net reclassification index(NRI),and integrated discrimination improvement(IDI)of the lipid variables,the above four scores,and the four scoring models after combining the lipid variables were compared to assess the prognostic predictive ability of the lipid variables.3.A new prognostic prediction model based on the independent predictors obtained from multivariate logistic regression analysis.Results:1.266 patients with ACLF were included in this study,with 28-day and 90-day mortality rates of 25.94%(69/266)and 38.49%(102/266),respectively.After stepwise variable screening and analysis,independent predictors of 28-day mortality included age,overt hepatic encephalopathy at admission,high-density lipoprotein cholesterol(HDL),and international normalization ratio(INR),and independent predictors of 90-day mortality included age,albumin(ALB),serum creatinine(Cr),and INR.The results of secondary outcome analysis indicated that total cholesterol(TC)was an independent predictor of upper gastrointestinal bleeding.2.the De Long test for AUC showed that all other models had greater AUC values than HDL(all P values < 0.001).the COSSH-ACLF score had better predictive power than the MELD score and HDL-MELD(P values 0.013 and 0.017,respectively),and the model HDL-COSSH-ACLF had the same compared to the latter two(both P-values were 0.009).The AUC performance of the remaining models was not statistically different.Comparing the NRI values calculated using the equidistant assignment method with specified cutoff values,the addition of HDL increased the predictive power of the MELD score by 4.60%(NRI=0.0460,P=0.036),the MELD-Na score by8.44%(NRI=0.0844,P=0.047),and the COSSH-ACLF score by 14.43%(NRI=0.1443,p=0.015),but there was no statistically significant difference in the predictive power of the CLIF-C ACLF score(all p values >0.05).There were no statistically significant differences in the IDI for each model comparison when different cutoff values were considered.3.The new model for prognostic prediction of 28-day mortality based on age,overt hepatic encephalopathy at admission,HDL,and INR had good predictive ability(AUC of 0.766 and Hosmer-Lemeshow test P value of 0.99).Patients were divided into low-risk(score <118),and high-risk(score ≥118)groups according to model scores,corresponding to a 28-day mortality rate of 12.20%,and 39.50%,respectively,with significant differences(P<0.001).Internal Bootstrapping validation showed no statistical difference between the simulated results and the real situation(p-value 0.99).Conclusions:1.Lipid level was an independent predictor of short-term prognosis in patients with ACLF.The predictive value of lipid level for 28-day death in patients with ACLF was significantly lower than the MELD score,MELD-Na score,CLIF-C ACLF score,and COSSH-ACLF score,but its combination of the above scoring systems improved their predictive ability.2.The new model constructed with age,overt hepatic encephalopathy at admission,HDL and INR as predictors had good predictive ability for 28-day death in patients with ACLF.