The comparison of different antiviral therapies on the prognosis of hepatitis B virus-related hepatocellular carcinoma after curative treatments

Supplemental Digital Content is available in the text


Introduction
In Asia, hepatocellular carcinoma (HCC) commonly occurred in the underlying hepatitis B virus (HBV)-related liver disease. [1] Curative therapies like liver transplantation, hepatectomy, and radiofrequency ablation could improve the prognosis of HCC patients. With the improvement in surgical techniques and advancement in preoperative assessment, the 5-year survival rates after curative therapy has reached 50%. [2] However, tumor recurrence after curative therapy remains high with a 5-year recurrence rate >70%. [3,4] To date, no universally effective adjuvant treatment has been available to prevent HCC recurrence.
Chronic HBV infection is the main cause of HCC in Asia. The risk for HCC development is increased for patient with HBV infection. [5,6] Recent studies also showed that tumor recurrence after curative treatment of HCC was increased with the level of HBV-DNA and alanine aminotransferase (ALT). [2,7] Studies of large cohorts from China Hong Kong, China Taiwan, and Japan have confirmed that concomitant antiviral therapy with curative treatment reduced the recurrence of HCC. [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22] However, there was no consensus about which kind of oral antiviral treatment was the best option in the prevention of HBV-related HCC recurrence after curative treatment. The aim of this network meta-analysis was to sum up the current evidence about the efficacy of different nucleos(t)ide analogues on the prognosis of HBV-related HCC after curative treatment.

Methods
Our study was approved by the local institutional review board and was conducted in compliance with the Health Insurance Portability and Accountability Act of 1996. Our study protocol was received and approved by the Medical Ethics Committee of Sichuan University. Written informed consent was obtained from all participants.

Eligibility criteria for this review
This analysis was performed in accordance with the PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analyses) statement. [23] Studies were included if they fulfilled the following inclusion criteria: (1) Randomized trials or cohort design.
(2) HBV infection treated by nucleotide/nucleoside analogs after curative therapy. (3) HCC treated by curative surgical resection or ablation therapy. (4) Complete follow-up data about recurrence-free survival (RFS) or overall survival (OS). (5) Studies including nucleotide/nucleoside analogs therapy compared with placebo or no treatment after curative therapy of HCC.
Exclusion criteria were: (1) Studies involving in hepatitis C and/or D-related HCC.

Data sources and search strategy
We conducted a systematic search using the following electronic data-bases from inception to February 2019: PubMed, EMBASE, Cochrane Library databases, and Science Citation Index Expanded without language restriction. The search strategy was based on MeSH terms, combined with free text words. The detailed strategies are given in Figure 1. Reference lists of all identified papers (included studies and relevant reviews) were checked for additional studies suitable for inclusion. Our primary aim was to identify different effect of nucleos(t)ide analogues as antiviral therapy to postoperative survival of HBV-related HCC patients after radical hepatectomy.

Data collection and assessment of bias
Two investigators independently reviewed all the titles, abstracts, and manuscripts identified to determine if each study was eligible for inclusion in the meta-analysis, as recommended by the Cochrane Handbook for RCTs [24] and the Newcastle-Ottawa Scale (NOS) for observational studies. [25] Disagreements about eligibility were resolved by consensus with a third reviewer. A structured search using keywords ("hepatitis B virus," "chronic hepatitis B" or "hepatitis B" or "HBV" or "CHB"), ("hepatocellular carcinoma" or "liver cancer" or "primary liver carcinoma"), ("antiviral therapy" or "Nucleotide" or "Nucleoside" or "lamivudine" or "adefovir" or "entecavir" or "tenofovir" or "telbivudine"), and ("randomized trial" or "randomized" or "randomized controlled trial (RCT)") or ("retrospective" or "prospective" or "cohort") was performed. Data were extracted from each study with a predesigned review form as follows: study characteristics (publication year, publication type, journal, country, study design, disease population, inclusion criteria, and exclusion criteria); patients' baseline characteristics (age, gender, weight); HBV-related disease information (marrow function, liver function, viral information); tumor features (tumor size, tumor number, tumor encapsulation, satellite nodule, vascular invasion, tumor differentiation, tumor stages, and alpha fetoprotein (AFP); treatment information (follow-up duration, intervention drug, drug dose, tumor treatment type); and clinical outcomes (1-, 3-, and 5-year RFS or OS).

Statistical analysis
We performed a pairwise meta-analysis using STATA 14.0 software (Stata Corporation, College Station, TX). 1-, 3-, and 5year RFS rate and 1-, 3-, and 5-years OS were defined as the endpoint. Odds ratio (OR) with 95% confidence intervals (CIs) was calculated using the random-effects model or fixed-effects model for investigating treatment effects. [26,27] Z test was conducted to assess the significance of overall effect size. P value of less than .05 was considered statistically significant.
A network plot was produced to represent the overall information of the trials included in the analysis. Nodes size represents the number of trials for each treatment and lines thickness represents the number of available direct comparisons. [28] The contribution of each direct comparison to each network estimate was calculated according to the variance of the direct treatment effect and the network structure, later summarized in a contribution plot. [29] After constructing a heterogeneity matrix, the frequentist method was applied to the fitted meta-regression model. The model covariates as the basic parameters and assumed that heterogeneity is independent of the comparison between effect sizes from multi-arm studies. [30,31] A forest plot of the estimated summary effects, along with confidence intervals and corresponding predictive intervals (PrI) for all comparisons, summarizes the relative mean effects, the impact of heterogeneity, and predictions on each comparison in 1 plot. [32] To rank the treatments, we used the surface under the cumulative ranking probabilities (SUCRA); a SUCRA value of 100% is assigned to the best treatment and 0% for the worst treatment. [33] A comparison adjusted funnel plot was used to assess the presence of small-study effect. [34] Egger's test was used to assess the symmetry of the funnel plot. [35] To account for both the markedly effective rate and neurological deficits, we used multivariate methods to determine the dependency between outcomes. Clustering methods and 2dimensional plots were used to produce clusters of treatments. [36] Using the clusterank command, clustered ranking plots can be obtained using the STATA program. The markedly effective rate and neurological deficits became the data variable containing the SUCRA scores for all treatments in this network. The different colors correspond to the estimated clusters and were utilized for grouping the treatments according to their similarity for the outcomes.

Characteristics of the studies included
A total of 886 publications were identified in the initial search and 846 records were excluded based on the screening of titles, abstracts, or duplicate articles (Fig. 1). Full-text articles were retrieved for the 40 remaining publications and assessed for eligibility for inclusion. Of these 40 publications, 26 were excluded (12 publications in which nucleoside/nucleotide analogues were not classified separately, 2 publications with antiviral therapies before curative treatment, 2 publications with short follow-up time (less than 1 month), 10 publications not reporting recurrence rate or recurrence-free survival after antiviral therapy, 1 publication in which patients underwent transarterial chemoembolization (TACE)). Therefore, 14 remaining studies qualified for study inclusion. The 14 selected studies were published between 2000 and 2019 and involved a total of 2481 patients. [9][10][11]14,18,21,22,[37][38][39][40][41][42][43] The longest median follow-up duration was more than 120 months and the shortest median follow-up duration was 60 months. Study characteristics are summarized in Table 1. These studies were performed in 3 different countries (5 in Japan, [9,[40][41][42][43] 8 in China mainland, [10,11,14,18,21,22,37,38] and 1 in the Korea [39] ); all studies were published in full text. The type of study design included 1 case-control studies, [40] 11 retrospective observational studies, [9][10][11]14,22,[37][38][39][41][42][43] and 2 RCTs. [18,21] 3.2. Different effects of nucleos(t)ide analogues for HBVrelated HCC after curative resection in overall survivals Network connections of included trials were presented in Figure 2. This analysis includes 4 oral drugs for antiviral www.md-journal.com therapies, namely LAM, ETV, ADV, and LdT. It can be seen that LAM and ETV are the most studied treatments. As for ADV and LdT, despite the fact that it was included in 3 comparisons, its sample size was relatively large. Supplementary Figure 2, http:// links.lww.com/MD/E570 summarizes the contribution of direct comparisons in determining the network meta-analysis estimates for mixed and indirect evidence. Table 2 summarizes the results of the network meta-analysis for 1-, 3-, and 5-year overall survivals. In 1-year overall survival, patients with nucleos(t)ide analogues as postoperative antiviral therapy has no significantly improved overall survival benefit at 1-year follow-up compared to the patients who did not. Significant improvement of 3-year overall survival was found in the treatment of ETV (OR: 2.41, 95% CI: 1.55-3.73) versus control group. For 5-year overall survival, ADV (OR: 2.35, 95% CI: 1.17-4.73), LAM (OR: 2.08, 95% CI: 1.78-5.58), and ETV (OR: 2.14, 95% CI: 1.59-2.88) were found to be more beneficial than control group.
The induction treatment relative ranking of estimated cumulative probabilities of nucleos(t)ide analogues is demonstrated in Supplementary Figure 4 (Table 3). There is no significant publication bias from funnel plot in this network analysis ( Supplementary  Fig. 3, http://links.lww.com/MD/E571).

Different effects of nucleos(t)ide analogues for HBVrelated HCC in recurrence-free survivals
The evidence-based network is presented in Figure 3. This analysis includes 5 nucleos(t)ide analogues for antiviral therapies, namely LAM, ETV, ADV, TDF, and LdT. Studies between LAM and control group are the most studied treatments. Supplementary Figure 5, http://links.lww.com/MD/E573 summarizes the contribution of direct comparisons in determining the network meta-analysis estimates for mixed and indirect evidence.
In the results of network meta-analysis among all kinds of nucleos(t)ide analogues (Table 4), patients with ADV (OR = 1.92, 95% CI: 1.03-3.58) as postoperative antiviral therapy have     (Table 4). There is also no significant publication bias from funnel plot in the network analysis ( Supplementary Fig. 6, http://links.lww. com/MD/E574).

Quality assessment of trials and evidence grading
None of the eligible studies presented a severe risk of bias ( Supplementary Fig. 1, http://links.lww.com/MD/E569). Also, in the network meta-analysis, funnel plot analysis did not indicate any evident risk of publication bias ( Supplementary Fig. 3, http:// links.lww.com/MD/E571 and 6, http://links.lww.com/MD/ E574). These findings, coupled with the absence of inconsistency and the lack of violation of the transitivity assumption, allowed to grade as high the strength of evidence advocating nucleos(t)ide analogues including ETV, TDF, ADV, LAM, and LdT better treatment than patients with no antiviral therapy after curative liver resection. All the other treatment comparisons were characterized by a confidence interval crossing the null value. Accordingly, their strength of evidence was graded as moderate. Table 3 Ranking: probability from SUCRA of mean volume change, symptom score, and cosmetic score.

Overall survival
Recurrence-free survival Rank: probability of being the best treatment, of being the second best, the third best and so on, among the comparisons. ADV = Adefovir dipivoxil; ETV = Entecavir; LAM = Lamivudine; LdT = Telbivudine; SUCRA = surface under the cumulative ranking curve; TDF = Tenofovir disoproxil fumarate.

Discussion
Due to there were a few RCTs about antiviral therapy for hepatitis B virus-related hepatocellular carcinoma after curative treatment, we adopted different methods to prevent potential bias. Ultimately, this study included observational studies with high quality sores. Besides, RCT methodological quality assessment shows these 2 RCTs (Gang Huang et al and Jianhua Yin et al) [18,21] both have a low risk of bias mortality. The contribution plot demonstrates direct comparisons do not influence the entire network significantly (Gang Huang et al is below 10% in 1-, 3-, and 5-year recurrencefree survival). [21] Moreover, the symmetrically distributed funnel plot indicates low risk of publication bias. In the single-agent induction treatments for hepatitis B, using antiviral therapies had a significantly better recurrence-free survival than none antiviral therapy, which also conforms to the previous evidence. [16,17,19] In network meta-analysis, multiple treatment comparisons for a specific disease, which were not compared to each other, can be made simultaneously through a common comparator treatment. In our study, antiviral therapy showed a better prognosis than none antiviral therapy, but there is little differentiation between these 4 nucleos(t)ide analogues in overall survival. As for recurrence-free survival, ADV was more beneficial than ETV, LAM and control group. In Supplementary Figure 3 In 2015, the American Association for the Study of Liver (AASLD) adopted ETV and TDF as first-line treatment for hepatitis B. [44] However, given the necessity of long-term treatment in most patients with CHB, the importance of comparative data on the effectiveness and safety of nucleos(t) ide analogues is immense. In recent study, Murata and colleagues revealed that nucleotides, but not nucleosides, had the novel additional pharmacological effect of inducing IFN-l3, [45] which has potent efficacy on enhancement of HBV suppression. Furthermore, IFN-l3 also showed antitumor activity in murine models of cancer, including hepatoma. [46] Our results also found that ADV can obtain good outcomes in 1-, 3-, 5-year recurrencefree survival (64.7%, 64.7%, 70.4%), which is consistent with previous studies. As for the nucleoside analogues in the analysis (ETV, LdT, and LAM), ETV ranks higher than LAM, this probably due to the emergence of LAM-resistant HBV mutants. In the reported studies, [47,48] the rate of LAM resistance was about 23% in 1 year to 71% in 5 years.
Collectively, although nucleos(t)ide analogues showed a promising future to be an alternative to the current first-line treatment, more RCTs are required to confirm this suggestion based on the following meta-analysis results: (i) In the single-agent induction treatments for hepatitis B, using antiviral therapies had a significantly better survival benefit than none antiviral therapy. (ii) Nucleotide analogues like ADV and TDF were proved to have superior benefit for the prognosis of recurrence-free survival for HBV-related solitary HCC patients after curative liver resection.
However, some limitations of this network meta-analysis should be discussed. First of all, most of the included studies are observational studies, [9][10][11]14,22,[39][40][41][42][43]49] So the results must be interpreted with caution. Second, some extent of heterogeneity in direct comparisons existed. In this study, we conducted the pooled estimate neglecting some of these factors. Third, there are a few RCTs or observational studies concerned the TDF antiviral therapy after curative treatment of hepatitis B virus-related hepatocellular carcinoma. As known in recent study, [50][51][52] TDF are considered as the first-line treatment for hepatitis B. So more RCTs are required to conform this nucleotide analogue. Fourth, Table 4 Results of network meta-analysis for 1-, 3-, 5-year recurrence-free survival of 5 nucleos(t)ide analogue treatments for HBV-related HCC patients after curative liver resection.

Outcome
LdT TDF ADV ETV LAM

Control group
authors of some trials [9,10,[41][42][43] neglected to provide lucid and complete descriptions of critical information on methodology and findings, making it difficult to assess the extract data for meta-analysis.

Conclusions
This network meta-analysis indicates that patients with postoperative nucleos(t)ide analogues antiviral therapy had better survival benefit than those without antiviral therapy for HBVrelated HCC patients after the curative treatment. Additionally, nucleotide analogues postoperative therapy like ADV and TDF has better impact on early and late recurrence-free survival of patients after curative treatment than those undertaking nucleoside analogues. While TDF has been listed as first-line treatment for hepatitis B, it might also have considerable clinical implications for preventing the recurrence of HCC in patients with CHB after curative treatment. Further studies for the efficacy of TDF to the prognosis of HBV-related HCC patients are needed to ensure the replicability of our findings.