Influencing Factors On Prognosis Of Acute On Chronic Liver Failure Due To Withdrawal Of Nucleos(t)Ide Analogues Treatment In Patients With Chronic Hepatitis B

Background:Hepatitis B virus (HBV) infection is a worldwide epidemic, with epidemic strength greatly differing in different area. According to the World Health Organization, there were about 2 billion people in the world have been infected by HBV, among which 240 million were chronic HBV infected individuals and 650,000 died of liver failure, liver cirrhosis and hepatocellular carcinoma (HCC) caused by HBV infection each year. Among patients with liver cirrhosis and HCC in the world, 30% and 45% of them were caused by HBV infection respectively. Therefore, chronic HBV infection causes serious damages to human health. China is a high epidemic area of HBV infection. Hepatitis B sero-epidemiological survey in 2006 showed a HBsAg carrier rate 7.18% between age 1～59 years. According to this calculation, chronic HBV infected individuals in China are about 93 million, among which about 20 million patients are diagnosed as chronic hepatitis B (CHB). Among patients with liver cirrhosis and HCC in China,60% and 80% of them were caused by HBV infection respectively.Chronic HBV infection activates the immune response to damage the infected hepatocyte and thus causes the degeneration and necrosis of hepatocyte and initiates the hepatic fibrosis and finally leads to liver cirrhosis, liver failure and HCC. Nearly 1 million people died of HBV related end-stage liver disease every year, which seriously threatens human health. Therefore, anti-HBV therapy becomes a key tactics for the treatment of CHB. At present, anti-HBV medicines are classified into nucleos-(t)ide analogue (NAs) and interferon (IFN). IFN has limited therapy course, but shows relatively poor efficacy, and characterize as contraindications and side effects. Therefore, patients often choose NAs. Clinical studies have shown that NAs effectively inhabit HBV DNA replication and recover the liver function, thus it could consequently delay and even prevent liver cirrhosis and HCC development. The antiviral mechanism of NAs are directly inhibiting HBV DNA polymerase reverse transcriptase activity by competitively binding to virus DNA chain and thus terminate DNA chain extension and synthesis to inhibit virus DNA replication. However, NAs cannot eliminate the transcriptional template of HBV in the reproduction process-covalently closed circular DNA (cccDNA), which exists in hepatocyte persistently. Therefore, NAs treatment for anti-HBV is a long-term course. In the clinical practice, once NAs treatment withdraw, HBV DNA replication would restore and resulted in HBV DNA rebound, which will cause hepatitis flare, exacerbation, hepatic failure and even death. However, we still lack adequate studies on the clinical features and related independent influencing factors on prognosis of liver failure due to withdrawal of NAs treatment in patients with CHB.Objectives:To investigate the baseline clinical features, clinical outcome and independent influencing factors on prognosis of acute-on-chronic liver failure (ACLF) due to withdrawal of NAs in order to provide reference for clinical practices.Methods:1. Baseline DataPatients with hepatitis B related ACLF due to withdrawal of NAs treatment from Hepatology Unit, Nanfang Hospital affiliated to Southern Medical University between January 2011 and December 2014 were recruited. Diagnostic criteria and clinical classification were in accordance with Chinese guidelines for the diagnosis and treatment of liver failure formulated in 2012. Patients with following statues were excluded:co-infection of other hepatitis virus or HIV, biliary, alcoholic, autoimmune, metabolic liver diseases, drug-induced liver injury, liver tumors, liver transplantation, kidney disease or anticoagulant therapy. All patients received anti-HBV therapy on the basis of comprehensive internal medicine treatment during hospitalization.2. Observational index and end points of follow-upClinical information and inspection indexes of all patients on admission to hospital were collected. Clinical information included:whether having infection, hepatic encephalopathy (HE), liver cirrhosis, duration between drug withdrawal and ACLF onset, the antiviral agent and the duration of initial anti-HBV treatment. Inspection indexes included:alanine aminotransferase (ALT), aspartate transaminase (AST), total bilirubin (TBIL), albumin (ALB), serum creatinine (Cr), prothrombin time (PT), international normalized ratio (INR), count of white blood cell (WBC), count of platelet (PLT), HBV DNA load, HBeAg statue and alpha fetoprotein (AFP). End-point determined as patients died within 12 weeks after disease onset. For patients discharged themselves due to disease deterioration and did not return for follow-up, end-point determined by telephone follow-up or family visit.3. Score CalculationAs all subjects of this study were patients with HBV related liver failure, the model for end stage liver disease(MELD) scores equaled to 3.8×loge (serum bilirubin umol/Lx0.058)+11.2xloge (PT INR)+9.6×loge (serum creatinie umol/L×0.011)+6.4. MELD-Na scores=MELD scores+1.59x(135-Na+). If levels of serum Na+≥135mmol/L,135mmol/L would be taken for the calculation; if levels of serum Na+≤120mmol/L,120mmol/L would be taken for the calculation; if levels of serum Na+ was between 120mmol/L to 135mmol/L, exact values was taken. According to scoring criterion, Child-Turcotte-Pugh (CTP) scores were calculated by five indicators including hepatic encephalopathy, ascites, TBIL, albumin and PT.4.Statistical AnalysisAccording to homoscedasticity or not, the quantitative data was expressed as mean ±tandard deviation or median (range). One-way analysis of variance and LSD-t test were adopted for homoscedasticity to compare the intergroup difference. Kruskal-Wallis H test or Mann-Whitney U test was adopted for skewness data to compare the intergroup difference. Enumeration data was compared by chi-squared test or Fisher’s exact test. Categorical variables were compared by Kruskal-Wallis H test. Correlation between quantitative datum or categorical variables were analyzed by Pearson or Spearman correlation test as appropriate.While the survival rate was assessed by Kaplan-Meier method, the independent influencing factor on prognosis of liver failure was analyzed by Cox regression analysis. The performance of predicting prognosis was evaluated by the area under receiver operating characteristics curve (AUROC). The optimal predicting cutoff was chosen to obtain the maximal Youden index. All statistical analyses were performed using the Statistical Package for Social Science (SPSS version 19.0;SPSS Inc.Chicago,IL, USA). Statistical significance was taken as two-sided.P<0.05.Results:1. General dataA total of 67 patients were included in analyses, among which 58 patients were males and 9 patients were females. The mean age was 42.7±11.3 years with range 23-73 years. A total of 49 patients withdrew NAs treatment not according to recommendations of guidelines on prevention and treatment of chronic hepatitis B in China, and 55 patients did not implement periodic follow-up visit.A total of 41 patients died within 12 weeks with mortality 61.2%.2. Correlation between HBeAg status and prognosis2.1 Correlation between baseline HBeAg status of initial treatment and prognosisAmong 37 HBeAg positive patients,20 patients died within 12 weeks. On the contrary, among 30 HBeAg negative patients 21 patients died within 12 weeks. There was no significant statistical difference by Pearson Chi-Square test (P-0.295).2.2 Correlation between HBeAg status at NAs treatment withdrawal and prognosisAmong 29 HBeAg positive patients,17 patients died within 12 weeks after drug withdrawal. On the contrary, among 38 HBeAg negative patients,24 patients died within 12 weeks after drug withdrawal. There was no significant statistical difference by Pearson Chi-Square test (P=0.522).2.3 Correlation between HBeAg status at hepatitis flare and prognosisAmong 32 HBeAg positive patients,19 patients died within 12 weeks. On the contrary, among 35 HBeAg negative patients,22 patients died within 12 weeks. There was no significant statistical difference by Pearson Chi-Square test (.P=0.958).2.4 Correelation between HBeAg status at NAs treatment withdrawal and duration between treatment withdrawal and ACLF onsetThe duration was 6.0 (29.4,3.2) months in 29 patients with HBeAg positive, and 4.9 (8.0,2.5) months in 38 patients with HBeAg negative, there was no significant statistical difference by Mann-Whitney U test (P=0.239).3. Correlation among initial treatment duration,duration between treatment withdrawal and ACLF onset, NAs before and after treatment withdrawal and prognosis3.1 Correlation between initial treatment durationand duration between treatment withdrawal and ACLF onsetThe initial antiviral treatment duration range was 1-85 months, which was less than 30 months (median 16.6 months with the interquartile range 24.5 months) in 49 patients. Duration range between treatment withdrawal and ACLF onset was 0.2-143 months, which was within 1 month in 8 patients, within 3 months in 21 patients, within 6 months in 39 patients and within 12 months (median 4.6 months with the interquartile range 6.3 months) in 52 patients. There was not statistical Spearman correlation between initial treatment duration and duration between treatment withdrawal and ACLF onset (P=0.365).3.2 Correlation between initial treatment duration, duration between treatment withdrawal and ACLF onset and prognosisThere was not statistical difference between mortality within 12 weeks and initial treatment duration (P=0.634). There was not statistical difference in survive rate between patients with initial treatment duration>30 months and<30 months (P=0.08). The duration between treatment withdrawal and ACLF onset was 4.9 (8.9, 1.8) months in Survival Group and 4.3 (9.5,2.2) months in Death Group, there was no significant statistical difference (P=0.954).3.3 Correlation between NAs category before and after drug withdrawal and prognosisThere was 21 patients,14 patients,5 patients,18 patients,9 patients received LAM, ADV, LDT, ETV and combination therapy (in which 8 patients with LAM+ADV and 1 patient with LDT+ADV), respectively. The mortality of patients who received ETV and non-ETV treatment was 72.2% and 57.1%, there was no significant statistical difference (P=0.209). After reoccurrence duo to drug withdrawal, all patients took NAs, there was 6 patients,1 patient,4 patients,41 patients,1 patient and 14 patients received LAM, ADV, LDT, ETV, TDF and combination therapy (in which 8 patients with LAM+ADV,4 patient with ETV+ADV and 2 patients with ETV+TDF), respectively. The mortality was 62.5% in 48 patients receiving ETV or TDF treatment and 57.9% in 19 patients receiving other antiviral therapy, there was also no significant statistical difference (P=0.619).4. Baseline Clinical characteristics and the influencing factors on prognosisTable 1 showed the baseline clinical characteristics. Compared with Survival Group, patients in Death Group was characterized with higher TBIL, INR, Cr, MELD score, HE morbidity and lower PLT at baseline. Table 2 showed the influencing factors on mortality. Cox regression uni-variable analysis showed that HBV DNA or AFP was not related to the prognosis (P=0.429 and 0.485), but TBIL, INR, Cr, MELD score, HE and infection were positively related to the mortality. Cox mult-variable analysis showed that only MELD score and HE were the independent influencing factors on survival. The higher MELD score and HE occurrence predicted higher mortality (HR is 1.085 and 3.754 respectively; both P<0.00\).5. Cutoff of baseline MELD score for prognosis predictionBaseline MELD score was related to the mortality. Compared with patients with MELD less than 25, the mortality of patients with MELD equal or more than 25 was higher with HR 10.131. MELD score predicted the mortality with AUROC 0.906, with Youden index 0.714, the optimal MELD 32, predicted mortality with sensitivity 82.9%, specificity 88.5%, positive predictive value(PPV) 91.9% and negative predictive value (NPV) 76.7%. While 30 patients with MELD less than 32 characterized with mortality 23.3%,37 patients with MELD equal or more than 32 characterized with mortality 91.9%(P<0.001). Compared with MELD less than 32, HR of MELD score equal or more than 32 was higher than 9.512 (P<0.001).Conclusions:1. There was high mortality rate reaching 61.2% within 12 weeks in patients with liver failure related to chronic hepatitis B due to NAs withdrawal prematurely.2. There was no correlation between prognosis of liver failure due to drug withdrawal prematurely and HBeAg status before initial treatment and after reoccurrence, initial antiviral treatment duration, time from drug withdrawal to reoccurrence as well as NAs category used in initial treatment or after reoccurrence.3. There was no relationship between the prognosis and HBV DNA or AFP. However, all of the TBIL, INR, Cr and MELD score, existing HE and infection were positively associated with the mortality rate of patients within 12 weeks. Furthermore, only MELD score and existed HE were positively associated with the mortality rate of patients within 12 weeks with Cox mult-variable analysis. That indicated patients with HE and MELD>32 should be performed the liver transplantation as soon as possible.