A Novel Precautionary Model To Predict Occurrence Risk Of Acute-on-Chronic Liver Failure

Acute-on-chronic liver failure (ACLF) is characterized by acute hepatic insultmanifesting as jaundice and coagulopathy, complicated by ascites and/or encephalopathy inpatients with previously diagnosed or undiagnosed chronic liver disease. In2009Asia-Pacific Association for the Study of Liver (APASL) released the first internationalguidance of ACLF, termed ACLF as the acute onset within24weeks of chronic liverdisease to liver failure based on diagnostic criteria. Total bilirubin (TBIL)≥5x ULN,Prothrombin activity (PTA)<40%or INR≥1.5. In United States and Europe, theincidence of ACLF is rare, occurring mainly in the chronic alcoholic or fatty liver diseasedand autoimmune liver diseased patients with non-viral cirrhosis; while in China, theincidence is much higher occurring in the population of chronic hepatitis B virus (HBV)infection and Chronic liver disease including chronic asymptomatic carrier state, chronichepatitis B and decompensated liver cirrhosis. In clinic, we diagnose hepatitis B-relatedacute on chronic liver failure as chronic severe hepatitis B in China.The ACLF onset mechanism is complex. When the disease progresses to liver failure,short-term mortality is as high as50%to70%, and medical treatment is often ineffective,and then liver transplantation is the only effective treatment method. ACLF onsets abruptlyand progresses faster compared with other chronic hepatitis, but much slower thandrug-induced acute liver failure in United States and Europe, therefore, there is a periodfrom onset to liver failure, called acute-on-chronic pre-liver failure (Pre-ACLF). Althoughthe ACLF onset mechanisms are not yet clear, it can be identified that there is an essentialdifference between ACLF and ordinary chronic hepatitis B from pathogenesis to diseaseprogression, so the study about early warning of ACLF has some theoretical basis. However,the research on the ACLF warning is still in infancy. Directions of previous studies concentrated more in the prevention of complications of liver failure and prognosis, andthere are many difficulties in collecting study cases to perform ACLF early warningresearch. There are significant meanings to build an Acute on chronic liver failure warningmodel in clinic, to forecast the risk of ACLF, to assist clinical doctors to diagnose, toconduct an early intervention treatment on patients with a high risk of liver failure, toprevent and block liver failure occur, and to reduce the mortality of patients with liverfailure.This study retrospectively enrolled the cases that patients with chronic severe hepatitisB who were diagnosed ACLF from2003to2009in our hospital and got70cases in total.We also collected82cases with the severe chronic hepatitis B cases as control. The case’sbasic information and all the relevant auxiliary test results after the onset of each time pointwere recorded.20cases from each of the two groups were retained as model performanceassessments, and the rest are analyzed from the routine clinical indicators. Then we analyzethe progression differences between Pre-ACLF and CHB-S to discover the trends ofcharacteristic change and the value of early warning of liver failure. We also studied thequantitative of serum HBV core antibody and the qualitative of self-antibodies to evaluatewhether the two antibodies show the effectiveness of warning liver failure. Finally, weassessed each indictor and established the model to warn the ACLF with the Logisticregression analysis to predict the incidence of liver failure in the early phase, and we usedthe20cases of each group to assess the model’s warning performance.Main Research Results1. Data analysis according to50cases ACLF patients and62cases of patients withCHB-S showed that there was no significant difference in gender, predisposing factors,jaundice time, with or without liver cirrhosis, and HBeAg positive rate. However, severechronic hepatitis B patients showed a larger range of age distribution (12-72years old) thanACLF patients (29-69years old). When calculate the rate of two groups cases clinicalconventional indexes, the maximum rise rate of PT, INR in ACLF group were significanthigher than severe chronic hepatitis B group (P=6.386×10-11,1.905×10-11) and the biggestincrease rate of TBIL is significant higher than that in CHB-S group (P=0.0065) 2. Anti-HBc IgM quantitative levels show significant differences between the coursesof ACLF and CHB-S group in three time points CHB-S group is significantly higher thanACLF group (before TBIL peak point P=0.008, TBIL peak points P=0.005, TBILdecrease time point P=0.008), indicating that quantitative serum anti-HBc IgM level inpatients with CHB-S is higher than that in ACLF patients. Quantitative anti-HBc IgM levelin ACLF patients concentrated in below0.2S/CO, while anti-HBc IgM quantitative level inCHB-S patients had wider distribution. There is no significant change in anti-HBc IgMquantitative levels at three time points in two groups (ACLF group P=0.442, CHB-S groupP=0.371), and anti-HBc IgM level in the course of the TBIL previous peak time, peak timepoints, and after peak time points is not related ALT, TBIL, which means that there is nosignificant change in the process of development of hepatitis anti-HBc IgM levels.3. Antibodies qualitative tests showed that, when using IgG fluorescence antibody, thepositive rate of two groups cases in ANA, AMA, SMA, anti-LKM-1show no significantdifferences. When applying IgG+IgA+IgM fluorescence antibody, the positive rate ofANA and anti-LKM-1in ACLF group were significantly higher than that in CHB-S group(P=0.046and0.004).4. The rates of clinical conventional index including gender, age are analyzed withLogistic regression, variables are excluded with likelihood ratio, and the earlyprecautionary model is set as P=1/[1+e^-(-2.414+7.687×PT rise rate)]. The predict accuracyin this model is87%.5.20cases from both Pre-ACLF and CHB-S patients are estimated with earlyprecautionary model to verify the efficiency of this model, and the overall accuracy was95%, while the effective prediction rate of ACLF group was63%.From these results, we draw conclusions as below1. The maximum rise rate of PT and TBIL has the ability to predict whether ACLFwill occur or not.2. Because of the limited cases alloted, we have not found the characteristic changestrend of anti-HBc IgM quantitative level, so anti-HBc IgM do not have the capability topredict whether ACLF will occur or not. 3. Because of the limited cases allotted and test method needs to be improved, we alsothink that the positive of ANA and anti-LKM-1IgM antibody do not have the capability topredict ACLF occur or not.4. Set up the early precautionary model of whether ACLF occur or not initially,P=1/[1+e^-(-2.414+7.687×PT rise rate)], and the model’s predict accuracy is rather higher.