Factors Prognostic of Survival in Liver Transplant Recipients with Hepatitis B Virus Related Acute-on-Chronic Liver Failure

Objectives Factors prognostic of survival in liver transplant (LT) recipients with hepatitis B virus related acute-on-chronic liver failure (HBV-ACLF) remain unclear. This study evaluated risk factors for survival in LT recipients with HBV-ACLF and determined the scoring system optimal for assessing patient prognosis. Methods This retrospective study included 323 HBV-ACLF related patients undergoing LT, including 112, 146, and 65 patients with HBV-ACLF grades 1, 2, and 3, respectively. Overall survival (OS) was estimated by the Kaplan–Meier method, and factors associated with survival were analysed by multivariate Cox proportional hazards models. Pretransplant prognostic scoring systems were compared by receiver operating characteristic (ROC) curve analysis. Results The one-year survival rate was significantly lower in HBV-ACLF grade 3 (80.0%) than in grades 1 (93.8%) and 2 (91.8%) recipients (p=0.0063). Cox multivariate analysis showed that age >53 years (hazard ratio (HR) 3.731; 95% confidence interval (CI) 1.640–8.407), WBC count >8.6 × 109/L (HR 4.544; 95% CI 1.140–18.107), HBV-ACLF 3 (HR 2.729; 95% CI 1.050–7.096), and cold ischaemia time >8.5 hours (HR 2.867; 95% CI, 1.38–5.921) were independently prognostic of 1-year survival. Comparisons of pretransplant scoring systems showed that chronic liver failure-consortium ACLF score (CLIF-C ACLFs) was superior to COSSH-ACLF, MELD-Na, and MELD scores in predicting 1-year OS in these patients. Conclusions Age >53 years, WBC counts >8.6 × 109/L, HBV-ACLF grade 3, and cold ischaemia time >8.5 hours are independently prognostic of OS in LT recipients with HBV-ACLF. CLIF-C ACLFs is superior to other scoring methods in predicting 1-year OS in these patients.


Introduction
Hepatitis B virus related acute-on-chronic liver failure (HBV-ACLF) is a complex syndrome characterised by acute deterioration of liver function and high mortality rates, resulting in hepatic and extrahepatic organ failure due to chronic hepatitis B, regardless of the presence of cirrhosis [1,2]. Although some ACLF patients have autoimmune hepatitis, most patients with ACLF in the Asia-Pacific region have HBV-associated ACLF patients [3][4][5][6]. Short-term mortality rates of HBV-ACLF patients receiving comprehensive conservative management have been reported to be as high as 50% to 90% [2,7]. At present, liver transplantation (LT) is the most effective treatment for patients with HBV-ACLF [8][9][10][11][12]. Although factors prognostic of LT in patients with ACLF have been widely reported, few studies to date have focused on HBV-related ACLF and assessed the post-LT prognosis of patients with HBV-ACLF; moreover, those studies included relatively small numbers of patients and did not completely evaluate indicators [12,13]. No studies to date have reported reliable and uniform prognostic factors that can predict survival outcomes in LT recipients who developed HBV-ACLF.
Several prognostic scoring systems have been formulated to evaluate LT-free survival of patients with HBV-ACLF, including the chronic liver failure-consortium ACLF (CLIF-C ACLF) score [14], the Chinese group on the study of severe hepatitis B-ACLF (COSSH-ACLF) score [2], the COSSH-ACLF II score [1], the model for end-stage liver disease (MELD) score [15], and the MELD-sodium (MELD-Na) score [16]. Although patients in China are usually evaluated by determining their COSSH-ACLF II scores [1], the optimum scoring system for evaluating outcomes of LT recipients who developed HBV-ACLF has not been determined. e present study therefore evaluated the risk factors for survival in HBV-ACLF patients following LT and determined the scoring system optimal for assessing patient prognosis.

Study Population.
All patients (N � 933) who underwent LT in the Second Xiangya Hospital, Central South University between January 2015 and June 2020 were retrospectively evaluated ( Figure 1). Patients not infected with HBV (N � 149) were excluded, as were patients aged <18 years, those with Rh incompatibility, and those who had undergone previous LT and combined organ transplantation (N � 16). HBV-infected patients who did not develop ACLF were also excluded from the study (N � 445). e 323 LT recipients with HBV-ACLF included in this study were divided into three grades based on the number of failed organs [2], with 112, 146, and 65 patients classified as having HBV-ACLF grades 1, 2, and 3, respectively. All patients were followed up for more than 1 year after LT. e study protocol conformed to the ethical guidelines of the Declaration of Helsinki and was approved by the institutional review board of the Second Xiangya Hospital, Central South University (No. 2019-050). A written informed consent was obtained from all the participants.

Operative Procedures and Immunosuppression erapy.
Starting from 1 January 2015, all organ donations were performed according to the Chinese protocol [17]. Hepatic allografts were obtained using a rapid technique for the procurement of donor livers, including flushing of the aortic and portal veins and cold storage, as described previously [18]. All hepatic allografts were perfused with and preserved in the University of Wisconsin (UW) solution. All operative procedures were performed using standard techniques or piggyback methods. Patients undergoing LT received ATG-Fresenius or basiliximab and methylprednisolone as part of the induction protocol. Standard posttransplant immunosuppressive treatment included tacrolimus and mycophenolate mofetil.

Statistical Analysis.
Continuous variables were reported as medians (ranges) and categorical variables as numbers (percentages (%)). Groups were compared using Kruskal-Wallis tests, unpaired Student's t-tests, and Pearson's chi-squared or Fisher's exact tests, as warranted. Overall survival (OS) was estimated by the Kaplan-Meier method. Receiver operating characteristic (ROC) curves were calculated for the MELD, MELD-Na, CLIF-C ACLF, COSSH-ACLF, and COSSH-ACLF preoperative prognostic scoring systems, and the areas under the ROC curves were compared using Z tests (Delong's method [19]). Risk factors significantly affecting patient survival were determined and included in a Cox proportional hazards regression model analysis. All statistical analyses were performed using SPSS 11.0 (IBM Corp., USA) or MedCalc 15.0 (MedCalc Software Ltd, Belgium) statistical software, with p values < 0.05 considered statistically significant.

General Characteristics of LT Recipients with HBV-ACLF.
A total of 323 HBV-ACLF related patients undergoing LT were eligible for inclusion in the study. eir 90-, 180-, 270-, and 360-day OS rates following LT were 93.8%, 91.6%, 90.7%, and 90.1%, respectively (Figure 2(a)). e one-year survival rate was significantly lower in recipients with HBV-ACLF grade 3 (80.0%) than in those with HBV-ACLF grades 1 (93/8%) and 2 (91.8%, p � 0.0063; Figure 2(b)). Table 1 describes the demographic and clinical characteristics of the study population. Recipients with HBV-ACLF 1, HBV-ACLF 2, and HBV-ACLF 3 were similar in age, gender distribution, the incidence of ascites, and serum concentrations of total bilirubin, urea, and sodium. However, significant differences were observed in body mass index (BMI), total creatinine concentration, international normalised ratio (INR), white blood cell (WBC), neutrophil and platelet counts, and haemoglobin concentrations, with all of these parameters being significantly worse in the HBV-ACLF grade 3 than in the other two groups. As expected, preoperative scores were significantly higher in the HBV-ACLF 3 than in the HBV-ACLF 1 and HBV-ACLF 2 groups (p < 0.001). e baseline characteristics and laboratory data immediately before liver procurement did not differ among donors to patients in the three HBV-ACLF groups.

Univariate Analysis of Factors Associated with Prognosis in LT Recipients with HBV-ACLF.
e clinical characteristics of LT recipients with HBV-ACLF who did and did not survive for 1 year after LT were compared ( Table 2). Although gender distribution, BMI, incidence of ascites, total bilirubin, creatinine, sodium, and haemoglobin concentrations, INR, and platelet counts did not differ significantly in these two groups, significant differences were observed in patient age, serum urea concentration, WBC and neutrophil counts, and HBV-ACLF grade. Pretransplant scores, except for MELD score, were significantly higher in patients who died than in those who survived. Donors to patients in these two groups showed no significant differences in age, gender distribution, BMI, proportions with fatty liver, donation after cardiac death, cardiopulmonary resuscitation, and laboratory data. e rate of hepatitis B surface antigen (HBsAg) positivity was significantly lower and cold ischaemia time was significantly shorter in donors of patients who survived than in those who died.   recipients, as well as cold ischaemia time of donors, were risk factors significantly affecting the prognosis of LT recipients with HBV-ACLF (Table 2). Patients were followed up for 1 year after LT, with death being the final outcome. e ROC curve analysis showed the cut-off values for these indicators ( Figure 3), with the optimal cut-off values determined by calculating the maximum area under the ROC (AUROC) curve and the corresponding Youden index. ese cut-off values included age 53 years, serum urea concentration 6 mmol/L, WBC count 8.6 × 10 9 /L, neutrophil count 6.6 × 10 9 /L, and cold ischaemia time 8.5 hours. ese cut-off values, along with HBV-ACLF grade and HBsAg, were included in Cox univariate and multivariate analyses of factors significantly associated with patient survival 1 year after LT (Table 3). Cox multivariate analysis showed that age >53 years (hazard ratio (HR) 3.731; 95% confidence interval (CI) 1.640-8.407), WBC count >8.6 × 10 9 /L (HR 4.544; 95% CI 1.140-18.107), HBV-ACLF 3 (HR 2.729; 95% CI 1.050-7.096), and cold ischaemia time >8.5 hours (HR 2.867; 95% CI 1.38-5.921) were independently prognostic of 1-year survival in these LT recipients.

Comparison of Prognostic Scores in LT Recipients with
HBV-ACLF. All pretransplant scores were significantly higher in patients who survived than in those who died  Abbreviations: MELD, model for end-stage liver disease score; MELD-Na, MELD-sodium score; CLIF-C ACLF, chronic liver failure-consortium ACLF score; COSSH-ACLF, Chinese group on the study of severe hepatitis B-ACLF score; COSSH-ACLF II, Chinese group on the study of severe hepatitis B-ACLF II score. * p < 0.05 vs. HBV-ACLF 1; * * p < 0.05 vs. HBV-ACLF 2.
( Table 2). e five pretransplant scores were compared using ROC curves and Z tests to determine which optimally predicted survival after LT. e CLIF-C ACLF scoring system was significantly more accurate in predicting patient prognosis than the COSSH-ACLF IIs, COSSH-ACLF, MELD-Na, and MELD scores. e AUROCs of the CLIF-C ACLF score for predicting mortality at 90, 180, and 360 days in patients with HBV-ACLF after LT were 0.654, 0.715, and 0.734, respectively ( Figure 5).

Discussion
is study analysed the survival of HBV-ACLF patients following LT at a single centre in China from January 2015 to June 2020. Of the 933 patients who underwent LT, 323 (34.6%) had HBV-ACLF, with these patients having a 1-year survival rate of 90.1%. LT was shown to be safe and effective, with good outcomes in patients with HBV-ACLF. e present study found that several risk factors, involving both recipients and donors, were associated with recipient prognosis. Recipient factors included age, serum urea concentration, WBC and neutrophil counts, pretransplant scores, and HBV-ACLF grade, whereas donor factors included HBsAg positivity and cold ischaemia time. Cox multivariate and Kaplan-Meier survival analyses showed that recipient age, WBC counts, and HBV-ACLF grade, along with donor cold ischaemia time, were significantly prognostic of survival in LT recipients with HBV-ACLF. In addition, a comparison of pretransplant scores found that the CLIF-C ACLF scoring system was more accurate than the other scoring systems in evaluating survival outcomes in these patients.
Several previous studies have evaluated factors associated with prognosis of HBV-ACLF patients following LT. For example, a study of 290 patients with ACLF after LT found that recipient WBC count, the ratio of alanine aminotransferase (ALT) to aspartate aminotransferase (AST), and the number of organs that failed were prognostic of  survival [13]. Another study, of 78 patients, reported that patients with initial COSSH-ACLF grade 3 and with grades 2 and 3 on days 3-7 after diagnosis had poorer prognosis [12]. In these studies, HBV-ACLF was diagnosed using the Asian Pacific association for the study of the liver (APASL) criteria [6], which did not include donor characteristics. In addition, the risk of death after LT was found to be higher in patients with ACLF 3 who required mechanical ventilation at LT and who received marginal organs [20]. at study, however, excluded patients diagnosed with HBV-ACLF.
In the present study, characteristics of both recipients and donors were evaluated to determine factors prognostic of survival in HBV-ACLF patients who underwent LT. Consistent with previous findings, the present study found that recipient age was an important prognostic factor in liver transplant patients [21,22], in that the one-year survival rate was significantly lower in HBV-ACLF patients aged >53 than ≤53 years. In addition, pretransplant infection has been reported to affect the posttransplant fatality rate in ACLF patients who underwent LT [23]. e present study found that HBV-ACLF patients with WBC counts over 8.6 × 10 9 /L had poorer outcomes after LT than those with WBC counts ≤8.6 × 10 9 /L. e severity of disease in ACLF patients undergoing LT is a key factor affecting prognosis [11,20]. e present study found that the 1-year OS rate was significantly lower in patients with ACLF grade 3 than in those with ACLF grades 1 and 2. However, the 1-year OS rate in patients with ACLF grade 3 was 80% after LT, significantly higher than in patients with ACLF grade 3 who did not undergo LT   Canadian Journal of Gastroenterology and Hepatology [1,2,12,24,25]. ese findings suggested that patients with HBV-ACLF be assessed for LT, with indications determined on an emergency basis, particularly for patients with a severe early course.
Although univariate analysis showed that receiving grafts from HBsAg-positive donors was a risk factor for HBV-ACLF patients, this factor was not statistically significant on multivariate analysis and was not independently prognostic of survival. is finding is consistent with results showing that receiving grafts from HBV positive donors did not increase complication or mortality rates after LT [26]. Additionally, mortality rates were found to be significantly higher in HBV-ACLF patients with cold ischaemia times >8.5 hours than ≤8.5 hours throughout the entire observation period, suggesting that longer cold ischaemia time has a major impact on the prognosis of HBV-ACLF patients after LT. Similar to our findings, previous studies have reported that the CLIF-C ACLF scoring system was superior to the classical MELD and MELD-Na scores in evaluating the prognosis of patients with ACLF [14,25]. In contrast, other studies have suggested that the COSSH-ACLF scoring system was better than the CLIF-C ACLF system in evaluating the outcomes of patients with HBV-ACLF [1,2]. All of these studies, however, evaluated outcomes in ACLF patients who did not undergo LT. Based on these findings, the present study used the COSSH-ACLF scoring system to select and classify HBV-CALF recipients. Compared with the APASL and EASL-ACLF scoring systems, the COSSH-ACLF scoring system identified nearly 20% more patients with HBV-ACLF, thus increasing their opportunity to get timely intensive management and earlier LT [1,2]. e present study showed that the CLIF-C ACLF scoring system was superior to the four other pretransplant scoring systems  in evaluating patient prognosis after LT, including survival rates at 90, 180, and 360 days. is study had several limitations, including its retrospective design and its inclusion of a relatively small number of patients who underwent LT at a single transplant centre. Prospective, multicentre studies with a larger sample size are needed to optimise the ability to predict the prognosis of LT in patients with HBV-ACLF.

Conclusions
Overall, this study evaluated risk factors affecting the prognosis of HBV-ACLF patients undergoing LT, finding that recipient age, WBC count, HBV-ACLF grade, and cold ischaemia time were significantly prognostic in this patient cohort. Pretransplant CLIF-C ACLF score was predictive of survival outcomes after LT in patients with HBV-ACLF.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon reasonable request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
JL and ZZ contributed in study design, data collection, analysis, and writing of manuscript. JY contributed in data collection, analysis of manuscript. QL, WH, GC, and ZS contributed in critical revision of manuscript. All authors approved the final version of the manuscript. ZZ and JY contributed equally to this study. Prof. JL is the guarantor of the article.