Two-year free of complications during antiviral therapy predicts stable re-compensation in immediate-treatment HBV-related decompensated cirrhosis

Abstract Background and aims Disease progression could be altered or even reversed in decompensated patients with HBV-related cirrhosis once they initiate antiviral therapy. However, little is known about the stable re-compensation in these patients. Methods In this retrospective study, HBV-related liver cirrhosis patients were consecutively enrolled at the first decompensated event of ascites or variceal hemorrhage (VH), and divided into immediate-treatment, on-treatment and delayed/no treatment groups. Patients were followed up to at least presence of second decompensation event or to June 2021. Re-compensation was defined as patients who did not occur second (further) decompensation during follow-up. Results A total of 130 HBV-related decompensated cirrhotic patients were included with a median follow-up of 61.0 (41.6, 72.0) months. The cumulative incidence of re-compensation at year 6 was 39.0, 9.8 and 6.6 in immediate-treatment, on-treatment and delayed/no treatment group (p = 0.001). Among 87 patients in immediate-treatment group, thirty-seven (37/87, 42.5%) were recognized as stable re-compensation. Seventy percent (35/50) of second decompensated events occurred in the first 2 years. In patients free of 2-year decompensated complications, about 71.2% (37/52) maintained stable re-compensation. The cumulative incidence of death (and/or transplantation) and HCC in patients free of 2-year decompensated complications or not was 2.9 vs. 27.3% (HR 9.4, 95% CI 2.2–40.0, p = 0.002) and 12.6 vs. 37.7% (HR 4.5, 95% CI 1.5–13.3, p = 0.006), respectively. Conclusions In decompensated patients with HBV-related cirrhosis, about 40% in immediate-treatment group maintained stable re-compensation during 6 years of antiviral therapy. Two-year free of complications could predict stable re-compensation.


Introduction
It is estimated 296 million people globally infected with chronic hepatitis B virus (HBV), 820,000 people died from HBV-related causes in 2019 [1]. About 2.1-6.0% of CHB patients advanced to liver cirrhosis yearly [2], two distinct clinical stages of compensated and decompensated liver cirrhosis were classified according to development any of the following complications: variceal hemorrhage (VH), ascites, hepatic encephalopathy (HE) and jaundice [3]. The transition from compensated to decompensated cirrhosis is associated with a poor prognosis characterized by a median survival of 2-4 years, while it exceeds 12 years in compensated patients [4]. Antiviral therapy suppresses HBV DNA replication, reduce the risk of disease progression and significantly improve long-term clinical outcomes in CHB patients [5,6]. In patients with decompensated complications, oral antiviral therapy was also observed to improve virological, biochemical and clinical parameters remarkably [7][8][9][10][11][12], and 33.9% of treated patients was delisted for liver transplantation (LT) after longterm therapy [13]. The new definition of decompensated cirrhosis has been raised as more profound insight into the clinical course of decompensation complications [14]. Effective etiology control and treatment of complications provide another possible direction of decompensated cirrhosis, namely re-compensation.
There were limited data on re-compensation of decompensated HBV-cirrhosis. The Chinese Society of Hepatology addressed re-compensation as 'some early decompensated patients may no longer develop decompensated events for a long period time (at least 1 year) due to effective etiology control and treatment of complications' [15]. Two studies applied a similar definition above to evaluate factors related to re-compensation and prognosis of decompensated cirrhosis [16,17]. Restoration of cirrhosis status to Child-Pugh score of 5 points was also defined as re-compensation in a Korean study [18]. Recently, the BAVENO VII consensus proposed the concept of re-compensation and required its fulfillment of all the following criteria [19]: a) removal/suppression/cure of the primary etiology of cirrhosis; b) resolution of decompensation complications; c) stable improvement of liver function tests. Re-compensation represented a disease status transition, yet there was no solid evidence justifying the stability of re-compensation up to now.
In this study, we aimed to describe the clinical courses of those who initiated antiviral therapy at the first episode of complication in HBV-related decompensated cirrhosis and evaluate the stability of re-compensation.

Study design and patients
In this retrospective, single-center cohort study, a total of 1704 liver cirrhotic patients who occurred decompensated complications and were hospitalized at Beijing Friendship Hospital from Mar 2013 to May 2018, were consecutively screened according to the following criteria: (1) age !18 years; (2) hepatitis B surface (HBsAg) positive for more than six months; (3) clinical evidence for liver cirrhosis, including any of: a) esophageal gastric varices confirmed by endoscopy; b) features of cirrhotic portal hypertension confirmed by imaging assessment, including liver surface nodularity, splenomegaly or portal vein ! 1.3 cm; c) laboratory test showed PLT < 100 Â 10 9 /L and ALB < 35 g/L; (4) presence of ascites or VH as the first episode of decompensated complications, and clinical material about first event was available; (5) follow-up time duration from the first decompensated event was more than 12 months.
The exclusion criteria were as follows: first event presenting of HE, jaundice or other less common complications; coinfection with HCV or HIV; with alcoholic, autoimmune, genetic, drug-induced, severe nonalcoholic fatty-liver disease, cholestatic liver diseases or any other chronic liver diseases; history of hepatocellular carcinoma (HCC) before the first decompensation event or within three months of the first decompensation; underwent LT or death within three months at the first decompensation; history of severe diseases of the heart, lung, kidney, brain, blood system or other organs.
The study protocol was approved by the Ethics Committee of Beijing Friendship Hospital, Capital Medical University (2019-P2-247-02). This study was conducted in accordance with the principles of the Declaration of Helsinki.

Clinical assessment and follow-up
As initiation of antiviral therapy 3 months prior to, immediately and after the first decompensation resulted in different patterns of restoration of CTP and MELD scores and longterm prognosis in decompensated patients [13], all included patients were divided into three groups according to their time duration from initiating antiviral therapy to the first decompensation. On-treatment group was patients who received antiviral therapy at least three months before the first decompensation. Immediate-treatment group was patients who initiated nucleos(t)ide analogues (NUCs) immediately within three months at first decompensation event. Delayed or no treatment group was patients who initiated antiviral therapy at least three months after the first decompensation or did not receive antiviral therapy until death or loss to follow-up, for they had no health insurance and were unwilling to finance their own therapy.
Baseline and follow-up laboratory and clinical assessments were retrospectively collected every 3-6 months including demographics, laboratory, HBV DNA, serology, upper endoscopy and imaging assessment. These data were extracted from an electronic medical record system. The limitation of HBV DNA detection was 200 IU/mL. Model for end-stage liver disease (MELD) score and Child-Turcotte-Pugh (CTP) score were calculated [20,21]. Treatments for the first decompensated events such as diuretics, albumin infusion, and endoscopic esophageal varix ligation (EVL) were also documented. During follow-up, diuretics used at least for two weeks and portal pressure-lowering drugs (including carvedilol and propranolol) used at least for 1 year after the first onset decompensation, were also recorded. All patients were followed up to second (further) decompensated events, HCC, LT, death or up to Jun 2021. Patients were considered as loss of follow-up if they had only data of the first decompensated complication and could not be contacted after that, or those who were followed up less than 1 year without second decompensated events.
Since there is no gold standard definition of re-compensation till now, we regarded it clinically as restoring of decompensation to a long-term stable status without further decompensated complications and liver-related death. Therefore, re-compensation was defined as patients who did not occur any second (further) decompensation during follow-up in our study. Second (further) decompensation was defined as the occurrence of ascites, VH, HE or other complications after the first decompensated event during follow-up [22]. Hepatitis B flare was defined as ALT more than 5 times the upper limit of normal (40 U/L). Rebleeding was defined as recurrence of VH at least three months after the previous episode of bleeding. Recurrent ascites were identified as reappearance of ascites at least three months after the previous episode of ascites with imaging evidence of completely fading away at intervals. LT or death was considered as hard clinical endpoints.

Statistical analysis
Continuous variables were expressed as median (interquartile range). Categorical variables were reported as number of patients (proportion of patients). Comparisons of continuous variables were mainly performed using Mann-Whitney test. Group comparisons of categorical variables were performed using Pearson's chi-squared or Fisher's Exact. In addition, the cumulative incidence of re-compensation, HCC or hard clinical endpoint was calculated using Kaplan-Meier curves and group analysis was compared using log-rank test. In this study, p < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 23.0 software (SPSS Inc., Chicago, IL)and GraphPad Prism 7.0 (Graphpad Software Inc., La Jolla, CA). Sankey plot showing clinical courses of cirrhosis after first decompensation was performed using R package 'networkD3' in R x64 4.1.2.

Baseline characteristics of included patients
A total of 187 HBV-related liver cirrhosis patients hospitalized due to the first decompensated events were carefully screened, and 57 of them were excluded for analysis, including 23 patients loss of follow-up (21 patients with only available data of first decompensation, 2 patients without second decompensation but followed up for only 4.0 and 5.7 months). Finally, 130 of them were included with a median follow-up duration of 61.0 (41.6, 72.0) months.
In delayed/no treatment group, 11 patients (11/15, 73.3%) received antiviral therapy at a median duration of 12.9 (5.9, 18.5) months after the first decompensation, 3 patients did not receive anti-HBV therapy until loss of follow-up at 56.2 (21.0, 61.8) months, and 1 patient died of decompensation and HCC at 59.3 months after the first event. In immediatetreatment group, first decompensated event was mainly ascites in 58 patients (66.7%) and VH in 29 patients (6 patients alone and 23 patients with ascites simultaneously), while VH was the main event in on-treatment group ( Baseline clinical characteristics about the first decompensation in three groups are shown in Table 1. In immediatetreatment group, 64 patients (73.6%) were male, with a median age of 54.0 years. Delayed/no treatment patients were younger and male patients were less in on-treatment group (p > 0.05). Median value of HBV DNA level was 5.9 and 5.8 Log10 IU/mL in immediate-treatment and delayed/no treatment group and it was undetectable in on-treatment group (p < 0.001). Immediate-treatment patients had a higher level of ALT, AST, TBIL and lower albumin (p < 0.05). And there were 68.7% and 50.0% of CTP class B patients in immediate-treatment group and delayed/no treatment group, mainly CTP class A patients (47.8%) in on-treatment group (p ¼ 0.010). Entecavir was the predominant treatment in three groups.

Long-term clinical courses of further decompensation after first decompensation
Patients occurred second/further decompensation accounted for 57.5% (50/87), 67.9% (19/28) and 93.3% (14/15) in immediate-treatment, on-treatment and delayed/no treatment group (Figure 1), and the 6-year cumulative incidence of death/LT was 12.1%, 25.2% and 10.0% (Supplementary Figure 1(A)), respectively. As there were only four cases of patients had no treatment, we were unable to compare the long-term clinical courses between treatment and without treatment. As shown in Supplementary Figure 1(B,C), though no significant difference was found in cumulative rate of death/LT (8.2 vs. 12.1%, p ¼ 0.517), second decompensation was significantly lower in on-treatment group than in immediate-treatment group (41.1 vs. 61.0%, p ¼ 0.016), which reflected that the earlier treatment the better. The second (further) decompensated complications occurred mainly in the first 2 years, which were 15 and 20 cases in the immediate-treatment group, were 6 and 5 cases in the on-treatment group, and were 5 and 8 cases in the delayed/no treatment group, respectively ( Figure 1).

Long-term re-compensation during antiviral therapy after first decompensation
The cumulative incidence of re-compensation at year 6 was 39.0, 9.8 and 6.6% in immediate-treatment, on-treatment and delayed/no treatment group (p ¼ 0.001) (Figure 2). The recompensation rate was almost similar between on-treatment and immediate-treatment group until a sudden decrease in the fifth year of follow-up in on-treatment group. In detail, four patients developed further decompensation until 5 years in on-treatment group, two of which had HCC simultaneously, suggesting HCC might be the confounder. The on-treatment group and delayed/no treatment group were not applicable to calculate yearly incidence due to few cases. In immediate-treatment group, the cumulative incidence of re-compensation from year 1 to year 6 was 82.8%, 59.8%, 51.7%, 47.5%, 41.2% and 39.0%, respectively ( Figure  2). And the Sankey plot showed that patients who maintained stable re-compensation at year 6 accounted for 50.0% (Figure 3). It showed a sharp decrease in the first two years and then relatively stable thereafter. This is consistent with the previous result that 70.0% of second decompensated events occurred within the first 2 years (Figure 4), indicating that 2 years maybe a potential watershed time point for stability of re-compensation in immediate-treatment group.

Associated factors with re-compensation
Predictors of 2-year re-compensation were analyzed in total patients in Supplementary Table 1, and prior or immediately treatment was independently related to re-compensation in multivariate analysis (p < 0.05). And in immediate-treatment group, male patients were less in re-compensated group (p ¼ 0.035). ALT (p ¼ 0.015) and TBIL (p ¼ 0.038) level in recompensated group was higher than that in second decompensated group (Supplementary Table 2). Whereas age, HBV DNA, APRI score, FIB-4 score and CTP class at baseline showed no difference (p > 0.05). And Supplementary Figure  4 shows the dynamic change of clinical parameters predominantly used in clinical practice in immediate-treatment group. ALT and ALB improved significantly in 6-12 months and gradually stabilized after that. The improvement of the re-compensation group was more obvious than that of the non-re-compensation group in immediate-treatment group.  Furthermore, the treatment related to re-compensation was also analyzed in immediate-treatment group. Patients in both second decompensated and re-compensated groups according to whether free of 2-year complications had high virological response greater than 95% (p > 0.05). Nine patients in immediate-treatment group had hepatitis flare, eight of them achieved re-compensation, one patient occurred VH at 5.5 months after antiviral therapy. Other therapeutics related to cirrhosis or portal hypertension, such as treatment with EVL, ALB infusion, oral diuretics or portal pressure-lowering drugs, there were no statistical differences in the second decompensated or re-compensated groups (p > 0.05). After initiation of antiviral therapy, usage of oral diuretics was less in re-compensation patients due to liver disease improvement (17.3 vs. 38.2%, Table 3).

Discussion
In this retrospective study, we provided a comprehensive description of long-term clinical course and evaluated the stability of re-compensation of patients with HBV-related decompensated cirrhosis who initiated antiviral therapy at the first episode of complication. The second (further) decompensated complications occurred mainly in the first 2 years. About 40% of patients maintained 6-year stable recompensation after initiation of antiviral therapy in the immediate-treatment group. In patients who developed second decompensation, 70% occurred within the first 2 years. Two-year free of complications can predict stable recompensation and had favorable long-term prognosis. This was the first study described stable re-compensation in HBVrelated decompensated cirrhosis after initiation of antiviral therapy and could be valuable to clarify the definition of recompensation.
The rates of re-compensation varied according to different definitions in different studies. In an Australia study to evaluate four patterns of acute decompensation (AD) in hospitalized cirrhotic patients mainly caused by alcoholic liver  disease, there were 26.7% (12/45) of stable decompensated cirrhosis (SDC) patients achieved re-compensation (defined as no need for diuretic or HE treatment as well as no further decompensating events during long-term follow-up) [17]. Recently, a Korean study investigated determinants of recompensation in patients with HBV-related decompensated cirrhosis starting antiviral therapy mainly within 3 months of recognition, with re-compensation defined as restoration of a Child-Pugh score to 5 points, the re-compensation rates were 57.2% and 66.7% in the derivation and validation cohorts, respectively [18]. While in our study, the re-compensation rates decreased from 82.8% at year 1 to 41.2% at year 5. The difference is not surprising since achievement of Child-Pugh score of 5 points did not guarantee no further decompensation events later. Therefore, free of decompensation events in our study is more convincing of stable re-compensation and more than 40% of patients achieved 5-year stable re-compensation who initiated antiviral therapy at the first episode of decompensation complication.
On the other side, in patients who developed further decompensation, 70.0% (35/50) of second events occurred within 2 years since the initiation of antiviral therapy. Patients who developed second events within 2 years had significantly higher risk of mortality (including LT) or HCC compared with those who were free of second events. This revealed that patients had two different patterns of longterm prognosis who initiated antiviral therapy at the first episode of AD. Regarding the clinical courses of AD, the CANONIC and PREDICT studies provided deeper insights and promoted new conceptions [14,23,24]. The CANONIC study recognized the sub-population of acute-on-chronic liver failure (ACLF) with a very high 28-d mortality rate in acute decompensated patients [23]. The PREDICT study subdivided an onset of AD into three clinical courses with different 3and 12-month mortality: pre-ACLF (developing ACLF within 3 months with high mortality risk), unstable decompensated cirrhosis (UDC, requiring frequent hospitalizations but a lower mortality risk) and SDC (rarely require hospitalization and have much lower 1-year mortality risk) [24]. However, these two studies evaluated relatively short-term clinical outcomes of patients. Our long-term data showed that in patients free of two-year decompensated complications, about 71.2% (37/52) maintained clinically stability without second events during follow-up. SDC was not precise enough to describe this sub-population, while stable re-compensation may be more appropriate as these patients had no further decompensated events and favorable prognosis can be expected as well.
Though BAVENO VII put forward a relatively clear criteria of re-compensation, the timeframe to evaluate and predict stable re-compensation still need to be testified [19]. According to the Chinese guideline on liver cirrhosis, at least 1-year timeframe was suggested to ensure a stable state of re-compensation. A Korean retrospective study using longterm health insurance review data illustrated that incidence of HCC and mortality sharply decreased after 1-year treatment [25]. Though it is a population-based cohort study that verified antiviral therapy effects, it is an indirect information about timeframe evaluating prognosis and many detailed clinically data were not available. A multicenter, prospective cohort study showed improvement in liver function over time during 5 years of antiviral treatment [13]. In our study, there were 30% (15/50) and 70% (35/50) of second events occurred in the first year and the second years respectively, which indicated that 1 year after antiviral treatment may not be long enough. Therefore, we suggested 2 years after antiviral therapy as a timeframe to evaluate stable re-compensation.
Moreover, we found that patients who achieved re-compensation were more likely to have a higher level of ALT and TBIL at baseline and then decreased dramatically through effective etiology control. A retrospective study indicated that baseline hepatitis flare was associated with better outcomes in HBV-related decompensated cirrhosis patients [26]. In our study, among nine hepatitis B flare patients, eight of them achieved re-compensation. Higher ALT and TBIL reflect more active inflammation in the liver, which is the right target of antiviral treatment. Other factors such as ascites or VH, CTP scores, specific treatment for complications at baseline showed no differences between further decompensation and re-compensation patients.
Furthermore, anti-HBV therapy significantly modifies the natural course of decompensated cirrhosis and improves their long-term prognosis [13]. However, in our study, ontreatment patients seemed a lower re-compensated rate than immediate-treatment patients at the sixth year after the first decompensation, detailed looking through revealed that HCC occurrence might interfere with the risk of second decompensation. This is consistent with that HCC would increase the risk of further decompensation and death or LT [27]. Meanwhile, the duration of antiviral therapy was longer in on-treatment group, and showed significantly lower rate of second decompensation compared with immediate-treatment group. Long-term prognosis between antiviral treatment and without treatment was failed to compare due to few cases in this study. But we recommend that HBV-related cirrhotic patients should initiate antiviral therapy as early as possible.
However, our study still had several limitations. First, all patients were screened from hospitalized population, outpatients with mild decompensated complications might be missed and led to selection bias, but we believe our selection would underestimate the re-compensation rate. Second, since this is a retrospective study, specific treatments of portal hypertension were not well controlled during follow-up, and related information may not be well documented in the medical record. Third, four patients did not receive anti-HBV therapy due to health insurance or unwillingness until loss of follow-up or death, which may magnify 3 months delay effect on survival analysis. Decompensated patients should start antiviral therapy as early as possible. Finally, this is a single-center study and the sample size was relatively small, and a large prospective multi-center cohort study could verify the characters of stable re-compensation in future.
In conclusion, this study showed that in decompensated HBV-cirrhosis, about 40% of patients maintained stable recompensation since initiation of antiviral therapy. Two-year free of complications can predict stable re-compensation and had favorable long-term prognosis.

Ethical approval
The study protocol was approved by the Ethics Committee of Beijing Friendship Hospital, Capital Medical University (2019-P2-247-02). This study was conducted in accordance with the principles of the Declaration of Helsinki.

Data availability statement
No additional data are available.