The Strategy and Ecacy of Prophylaxis Against Hepatitis B Virus Recurrence After Liver Transplantation in the Era of High Potent Nucleos(t)ide Analogues: A Meta-Analysis

Background: High potent nucleos(t)ide analogues (HP NAs) with or without hepatitis B immunoglobulin (HBIG) is the standard prophylactic therapy for avoiding the HBV recurrence after liver transplantation (LT). While the clinical impact of HP NAs-based prophylaxis has not been well documented, the optimal treatment strategy is currently debated. This meta-analysis aims to evaluate clinical outcome of LT recipients under HP NAs-based regimens and investigate different prophylaxis schemes. Methods: We followed PRISMA statement to conduct this study. Two reviewers independently searched relevant literature via PubMed, EMBAES, MEDLINE, Web of Science, and Google Scholar. Studies were included if they observed HBV recurrence under HP NAs-based regimens in patients who received HBV-related LT. Primary and secondary outcome were HBV recurrence, HCC recurrence, all-cause and HBVrecurrence related mortality. Incidence with 95% condence intervals was aggregated and assess by xed effect model/random effect model. Subgroup analysis was applied to examine the impact of different treatment strategies. Results: 26 studies (n=2374) were included, with a pooled HBV recurrence rate of 0.92% (95% CI 0.47%-1.48%). Studies with and without HBV viremia or HDV superinfected patients demonstrated comparable HBV recurrence (p=0.25, p=0.69). Recurrence rate under indenite combination of HP NAs and HBIG was lower than that under HP NAs monotherapy (p=0.0003), and similar to that under NAs plus nite course of HBIG (p=0.53). Pooled HCC recurrence rate was 5.34% (95% CI 0.78%-12.48%). HBVrecurrence related mortality and all-cause mortality were 0.17% (95% CI 0.00%-1.51%) and 7.29% (95% CI 4.42%–10.71%) respectively. Conclusion: Our study suggests HP NAs-based regimens provide satisfactory HBV


Background
For patients with chronic hepatitis B virus (HBV) infection, the prognosis after liver transplantation (LT) has been improved markedly, largely attributable to prophylaxis against HBV recurrence [1][2][3] . The rst milestone of preventive therapy was long-term hepatitis B immune globulin (HBIG), which tremendously decreased the risk of graft infection (from 75% to 33%) and increased 3-year survival (from 54% to 83%) 4,5 . Despite of the advances, HBIG monotherapy was far from satisfactory. The e cacy was compromised among patients with high risk of HBV reactivation or those developed immune escape [6][7][8] . Other problems such as supply limitation and high cost were di cult to ignore 9 . The second breakthrough for HBV recurrence prophylaxis was antiviral agents. Lamivudine (LAM) and/or adefovir (ADV) proved synergistic activity with HBIG by further lowering the HBV recurrence rate to less than 10% [10][11][12] . Nevertheless, the emerging issues such as YMDD mutants and renal impairment were major obstacles to long-term application of these low potent nucleos(t)ide analogues (LP NAs) [13][14][15][16] .
A new era came with the development of high potent and high genetic barrier nucleos(t)ide analogues (HP NAs, i.e. entecavir, tenofovir disoproxil fumarate, tenofovir alafenamide). As the safety and e cacy being proved in registrational and real-world studies, HP NAs have been substituted for LP NAS as the preferred antiviral prophylaxis [17][18][19][20][21] . Although there have been numerous clinical studies relevant to this topic, the only systematic review focusing on HP NAs-based prophylaxis was issued in 2013 22 . In that review, Cholongitas E and Papatheodoridis GV included 519 patients treated by combination of HP NAs and HBIG, and they found HBV recurrence incidence under inde nite combination of HP NAs and HBIG was 1.0%, signi cantly lower than that under LAM plus HBIG. Since HP NAs-based regimens have become standard care for LT recipients in recent years [19][20][21] , the critical issues are not their rationale of usage anymore, but rather their clinical outcomes and treatment strategy. The HBV reinfection and recipients' mortality under HP NAs-based therapies is not clear. In terms of treatment strategy, there are still many questions need to be answer, for example what the risk factors of HBV recurrence for HP NAsbased therapies are, whether HP NAs monoprophylaxis is equally effective as combination of HP NAs and HBIG. Therefore, we conducted this meta-analysis to probe these two issues and hope to provide evidence for clinical practice.

Literature search strategy
We followed PRISMA (Preferred reporting items for systematic review and meta-analyses) statement to conduct this study (23). A comprehensive literature search was performed via PubMed, EMBAES, MEDLINE, Web of Science, and Google Scholar under the terms of "liver transplantation", "HBV recurrence", "entecavir" and "tenofovir". For example, the search algorithm in PubMed was: ((liver transplantation) AND ((HBV recurrence) OR HBV reactivation)) AND ((entecavir) OR tenofovir). Abstract books from the major international transplantation conferences and hepatology meetings were also reviewed. Literature search was last updated on February 2020.

Study selection and data extraction
Studies which published in English and ful lled all of the following criteria were included: 1. Subject: adult patients received HBV-related LT. 2. Intervention: Entecavir or tenofovir disoproxil fumarate or tenofovir alafenamide (with or without HBIG). 3. Study design: RCT or observational cohort studies. We excluded the studies if they met any following criterion: 1. The number of patients who used HP NAs was less than 10. 2. Median/mean follow-up period was less than 12 months, since we expect the episode of HBV recurrence may not be captured during short period of follow-up. 3. HBV recurrence rate was not mentioned for HP NAs group. 4. Patients were exposed to LP NAs after LT. Two authors (Li MS, Hou ZH) conducted the literature search and screen independently, any disagreement was resolved by consensus or consulting to corresponding author.

Data Extraction
Data from manuscript and abstract were extracted manually (Li MS), in aspect of patient characteristics, therapeutic regimen and clinical outcome. To describe patient characteristics, the proportion of patients with pre-LT HCC, quanti able HBV DNA and HDV coinfection was pulled. Drugs, treatment duration and administrative route were recorded. In terms of clinical outcomes, events of HBV reactivation, HCC recurrence, HBV recurrence-related mortality and all-cause mortality were calculated. Parameters related to renal function were also extracted to review the impact of HP NAs on renal safety.

Assessment of quality (risk of bias)
Newcastle-Ottawa scale (NOS) and Cochrane Collaboration's tool for assessing risk of bias (RoB) were used to evaluate the study quality/risk of bias for observational studies and RCT 24, 25. In the NOS system, quality of observational study was rated as high, moderate or low based on total score i.e. score 7-9 indicated high study quality, 4-6 fair quality, and 0-3 low quality. According to RoB, the risk of bias of each RCT was graded as high, unclear or low by 6 domains.

Data analysis
The primary study outcome was HBV recurrence, which was de ned by original studies or as reappearance of HBsAg and/or HBV DNA if there was no de nition in original studies. The secondary outcome included HCC recurrence, HBV recurrence-related mortality and all-cause mortality. Heterogeneity across studies was assessed by I-square index, which indicated substantial heterogeneity with a value over 50% (26). Random effect model was applied in case of signi cant heterogeneity, and xed effect model was used in the absence or minor heterogeneity (27). Subgroup analysis was conduct by risk factors, HP NAs, combination approaches and administration of HBIG. We made sensitivity analysis to examine the robustness of primary nding. Meta-analysis was conducted with the R software program (version 3.6.2) and the R-package meta.

Studies and patient characteristics
A total of 419 citations were identi ed by initial search, 348 duplicated or irrelevant records were later removed. According to the inclusion/exclusion criteria, 45 literature was further excluded, and nal 26 studies (n= 2374) were included for our analysis (Figure 1).
The study and patient characteristics were summarized in Table 1. 24 observational studies and 2 RCT were included. 21 of them were presented in manuscript, and 5 in conference abstract. 18 studies were conducted in Asia, 3 in Europe, 3 in North America,1 in Australia, and 1 across continents. 15 studies enrolled patients with HBV viremia and 5 included patients with HDV superinfection. HCC was a major No observational studies scored under 4 or above 6, indicating all of them having fair quality. Both RCT had unclear risk of bias because the process of random sequence generation or blinding outcome assessment was not described.

Discussion
Our meta-analysis demonstrated minimal risk of HBV reactivation under HP NAs-based regimens for LT recipients (0.92%). In subgroup analysis the risk of reactivation was similar between studies with and without HBV viremia/HDV coinfected patients, suggesting that these elements might no longer affect the prophylaxis signi cantly in the era of HP NAs 54 was supposed to present following challenges: HBV within tumor cells might be relatively inaccessible to drug because of the change in liver structure and blood supply. Beside the non-HCC tissues, closed circular DNA (ccc DNA) of HBV also existed in extrahepatic HCC micro-metastases and HCC cells, acting as potential reservoirs which couldn't be targeted by NAs. Furthermore, dysfunction of hepatic immune network induced by HCC might predispose to poorer response to antiviral treatment [56][57][58] . Although HCC was one of the major indications for LT, we found the HCC recurrence rate was not high (5.34%).
Identifying the small population with high risk of HCC recurrence might help to further decrease HCC and HBV recurrence. All-cause mortality in our study was low, comparing previous studies which applied HBIG or LP NAs-based prophylaxis 59,60 . Moreover, the HBV-are related mortality was much lower than allcause mortality. These results suggested effective antiviral prophylaxis have improved long-term clinical outcome for LT recipients notably.
In terms of the preventive strategy, whether HBIG is a necessity in the era of HP NAs remains controversial. Although HBIG has been the mainstay of prophylactic regimen, many investigators have been trying to reduce the usage of HBIG to overcome the disadvantages such as high cost and inconvenient administration. Although the preliminary research of lamivudine monoprophylaxis demonstrated 90% of prevention by week 24 61 , later studies with longer follow-up resulted much lower effect (59%-80%) mainly because of drug resistance [62][63][64] . HP NAs (i.e. ETV, TDF, TAF) was an optimal choice for HBV prophylaxis due to superiority at e cacy and genetic barrier (65)(66)(67)(68). Currently there were only two studies adopting HP NAs monoprophylaxis to over 10 patients 47,53 . These two studies both conducted long term follow-up (median follow-up period: 79m and 59m), while the HBV recurrence rate in Fung J's cohort (5.8%) was as twice as that in Muthiah MD's cohort (2.9%). One of the reasons for the disparity could be the HBV DNA level. In Fung J's trial, 26% of the patients had HBV DNA more than 4 log IU/ml at the time of transplant. And in Muthiah MD's cohort there were only 3% with more than 5 log copies/ml of HBV DNA. Subgroup analysis showed signi cant difference between HP NAs + HBIG and HP NAs monotherapy. These results suggested caution with HP NAs monoprophylaxis, especially for patients with high level of HBV DNA at the time of LT. On the other hand, similar HBV reactivation was observed between lifelong dual therapy (n=1607) and HBIG withdrawal (n=405). We noticed this nding was different from that at Cholongitas E's systematic review, in which inde nite combination of HP NAs and HBIG (n=303), but not nite combination(n=102), presented superior prevention to LAM plus HBIG 22 .
The inconsistence was reasonable considering Cholongitas E's study was publicized 7 years ago when there were fewer evidence for HP NAs-based therapies. There is no consensus at the timing to stop HBIG, and we didn't nd signi cant difference between early and late withdrawal (i.e. ≤6m vs >6m). In this regard, individualized decision should be made based on kinetics of anti-HBs 40, 44, 69 . We did not conduct strati cation by dosage because most of the studies applied low-dose of HBIG (≤ 1000 IU) one week/month post-LT.
Renal dysfunction is a major post-transplant morbidity, which negatively impacts the graft and patient survival 70 . The prevalence of renal failure in LT recipients was the second highest among non-renal solid organ transplant patients, the 5-year cumulative incidence was 18-22% 70,71 . Among the trials included in our analysis, there were 3 TDF/ETV-based studies reported deterioration in eGFR or creatinine clearance.
It was di cult to conclude the impact of HP NAs on renal function in this meta-analysis since we could not rule out other important factors such as calcineurin inhibitors and surgery operation. While HP NAs with good renal safety pro le are preferred since most of the patients require lifelong treatment. Minimal rates of renal function decline had been reported during long-term therapy with ETV and TDF in patients with CHB, while the nephrotoxic potential was higher for TDF 20, 72-74 . There had been several cases about TDF-associated Fanconi syndrome, including one LT recipient [75][76][77] . This issue might be mitigated by TAF, which has demonstrated a trend of improved renal function in CHB patients who were previously exposed to TDF 78,79 . In Gane E and Sabb S' studies, we saw the same trend of TAF in LT recipients 48,49 .
Our study has several strengths. To our knowledge this is the rst meta-analysis for HP NAs-based prophylaxis in post-LT recipients. Comprehensive subgroup analysis was applied to examine different treatment strategies. However, the strengths of this meta-analysis should be weighed against some limitations. First, there were limited evidence of certain regimens (e.g. TAF-based regimen, HP NAs monoprophylaxis). Second, there were only two RCT included in our analysis, and most of the original studies were observational with medium quality. Third, heterogeneity was moderate across studies which might attribute to varied sample size and preventive therapeutics. All of these issues may affect the robustness of the data and limit interpretation of the results.

Conclusions
our systematic and meta-analysis suggested HP NAs-based regimens are highly effective at preventing HBV recurrence, and bene cial to long-term outcome for LT recipients. HBV recurrence prophylaxis was not affected by viremia or HDV coinfection. In terms of HP NAs options, ETV-based, TAF-based and TDFbased therapies demonstrated equivalent e cacy. HP NAS plus short-term HBIG serves an alternative to inde nite combination, and monotherapy of HP NAs might be appropriate for patients with low HBV DNA level before LT. To reduce risk of renal impairment, HP NAs with an optimized renal safety pro le is preferred. Availability of data and material

List Of Abbreviations
The datasets supporting the conclusions of this article is included within the article and its additional les.
Competing interests MSL works for Gilead while concurrently enrolled in Ph.D.

Funding
This study was not supported by any commercial company or grants.

Authors' contributions
Deming Tan and Qian Lin designed the study and take responsibility for the entire process; Mingshu Li and Zhouhua Hou conducted literature search, data extraction and quality assessment; Guozhu Yao analyzed the data; Mingshu Li drafted the manuscript; All authors have read and approved the nal paper.         This is a list of supplementary les associated with this preprint. Click to download.