Globally, more than 296 million people are living with chronic hepatitis B (CHB) infection, and 820,000 deaths occur due to hepatitis B virus (HBV)-related illnesses annually [1]. Chronic HBV infection causes liver fibrosis and cirrhosis and progresses to hepatocellular carcinoma (HCC) in 25–40% of HBV carriers [2]. HBV infection is managed with nucleos( t)ide analogues (NAs), which are reverse transcriptase inhibitors [2]. These antiviral agents prevent progression to cirrhosis and reduce the risk of HCC by suppressing HBV replication [3].

Since the first U.S. Food and Drug Administration (FDA) approval of lamivudine in 1998, various additional NAs have been used in clinical fields, including telbivudine, adefovir, entecavir (ETV), and tenofovir. Recently, professional guidelines from the Asian-Pacific Association for the Study of the Liver, the American Association for the Study of Liver Diseases, and the European Association for the Study of the Liver have all recommended an oral antiviral with a high genetic barrier to resistance, such as ETV, tenofovir disoproxil fumarate (TDF), or tenofovir alafenamide (TAF), as the first-line therapy [4–6]. Similarly, the Korean Association for the Study of the Liver guideline also recommends ETV, TDF, and TAF as first-line therapies for HBV infection, together with besifovir dipivoxil maleate (BSV) [7].

There is interest in comparing NAs to establish the best options for those with CHB. Numerous studies, including randomized trials and cohort studies, have compared the efficacy and safety of high-genetic barrier NAs. The majority of these studies did not differ significantly in terms of virologic response (VR) and biochemical response (BR) [8–10]. However, previous meta-analyses identified differences between ETV and TDF in achieving complete VR, with favorable outcomes more commonly reported for TDF [11–14]. There have been fewer publications addressing TAF or BSV, but some studies have shown that these NAs have similar efficacies to those of ETV and TDF [15,16].

The aim of this study is to estimate the relative efficacies of ETV, TDF, TAF, and BSV in NA-naïve CHB patients in terms of VR, BR, serologic response (SR), virological breakthrough (VB), and hepatitis B surface antigen (HBsAg) loss. Since there have been only a few head-to-head randomized trials directly comparing these high-genetic barrier NAs for CHB patients, we compared these four NAs through a network meta-analysis (NMA) that included randomized trials and propensity score-matched (PSM) studies.

The current study analyzed the efficacies of high-genetic barrier NAs in treatment-naïve CHB patients with respect to VR, BR, SR, and VB. For VR, TDF was likely to achieve higher proportion of 48VR and 96 VR compared to ETV, while patients treated with ETV fared better than those receiving TDF with respect to 48BR, illustrating divergent outcomes in terms of therapeutic efficacy.

Numerous published studies, including RCTs and cohort studies, have reported the efficacy and safety of high-genetic barrier NAs, and the majority could not reveal statistical difference among the NAs in terms of VR and BR. Meta-analysis concerning efficacy of these drugs also been performed previously, and the results revealed differences between ETV and TDF in achieving complete VR, with most cases showing more favorable outcomes for patients receiving TDF [11–14]. Moreover, in recent years, several studies using nationwide data and/or hospital-based cohorts that compared the incidence of HCC in patients treated with ETV or TDF have been published, yielding disparate outcomes for both arms [30,32,34]. As for these results, there is debate whether difference exists on the outcome of treatment efficacy between TDF and ETV exists. Fewer studies addressing TAF or BSV have been published, but some of these have reported similar efficacies of TAF or BSV when compared with the previously approved drugs, ETV and TDF. In addition, several meta-analyses have been introduced regarding those two NAs [35,36].

This study demonstrated that TDF outperformed ETV in terms of VR at both 48 and 96 weeks when assessed by NMA and at 144 weeks when assessed by conventional meta-analysis. The results by NMA indicated superior results of ETV to TDF for the 48BR. A conventional meta-analysis was used to compare ETV and TDF with respect to BR, and the tendency for ETV to be superior persisted for 144 weeks. No difference between the drugs was noted with respect to SR. When comparing BSV or TAF to the other drugs, TAF performed better than ETV for the 48VR endpoint, but there was no significant difference in BR or SR. Furthermore, we performed subgroup analyses of the studies that only included compensated liver disease, and the results continued to indicate superior outcomes for TDF compared to ETV with respect to 48VR. Given the majority of CHB patients are diagnosed in the early stages of liver disease and are routinely monitored under an active surveillance system, subgroup analyses based on compensated liver disease represent types of patients seen in real-world clinical practice. Subgroup analyses of HBeAg-positive and -negative CHB also were performed. In the HBeAg-positive subgroup, TDF provided better outcomes than ETV, whereas no differences between the two drugs was found for the HBeAg-negative subgroup. The outcomes of the subgroup analyses may have reflected the greater reported prevalence of the YMDD mutation in HBeAg-positive patients than in HBeAg-negative patients, which may render ETV less efficient than TDF [37].

The current treatment of choice for CHB is a drug with a high genetic barrier. While many studies have compared the efficacies of diverse NAs for CHB, to the best of our knowledge, no study has focused solely on high-genetic barrier NAs, and few studies have included BSV in the analyses. In contrast to previous meta-analyses, the majority of which did not include TAF or BSV in the comparison arms, our study included all four NAs known to have high barriers to HBV resistance, namely ETV, TDF, TAF, and BSV [7]. Moreover, our study demonstrated various outcomes that are commonly used in clinical practice to determine the efficacy of the drugs [5–7]. As a result, our study design strongly reflects the patient population in real-world practice and may be more useful in assisting clinicians when choosing NAs. Additionally, whereas prior systematic reviews included both treatment-experienced and treatment-naive patients, the current study restricted trial enrollment to treatment-naive patients, strengthening its novelty [12]. This allowed us to better reflect real-life clinical practice and to provide more relevant insights for clinicians managing such patients.

As stated earlier, there are discrepancies between the BR and VR results for ETV and TDF. However, the mechanisms underlying these discrepancies are not well established. Concerning this issue, Con et al. have proposed that the lipid-lowering effect of TDF may influence BR by disrupting the cellular membrane and causing liver enzyme leakage [36]. In addition, Chen et al. [11] have postulated that metabolites of the drugs might lead to such results. Another hypothesis, proposed by Chen et al. [11], posits that TDF’s ability to induce a more robust immune response, as evidenced by a higher rate of achieving 48VR, might inadvertently lead to greater hepatic cell damage, resulting in elevated liver enzyme levels. There may be other explanations of these outcomes. First, regarding BR, the background characteristics of the TDF and ETV groups in PSM studies might differ. TDF and ETV are unique drugs that were developed at different times and have different historical backgrounds. Even though PSM was performed, such differences may result in discrepancies in some areas that cannot be thoroughly calibrated by PSM, such as liver function, the presence of steatosis, and use of concomitant drugs. Second, even though lamivudine-resistant patients were excluded based on medical records, there is a small chance that such patients were included in PSM studies, potentially leading to bias. Although some studies have greater BR among patients treated with TAF than those receiving TDF [38,39], this tendency was not found in the current NMA. Since the present study only compared VR and BR over 144 weeks, outcomes for longer treatment durations should be compared to better understand the discrepancies between these outcomes.

Our study has several limitations. As stated earlier, current analyses only included treatment-naïve patients, which resulted in exclusion of initial phase 3 clinical trials of TAF and TDF that included both naïve and experienced patients [40,41]. As a result, only a few RCTs were available to evaluate the effectiveness of the four NAs using NMA. There was no RCT assessing TAF efficacy and only a few RCTs assessing BSV, resulting in insufficient data for both BSV and TAF. This could be attributed in part to the relatively recent introduction of these two drugs and could also be influenced by publication bias. In the same context, the rankogram highlighting BSV as the optimal choice for achieving a 48SR must be interpreted cautiously as it is based on limited evidence. Additional studies are imperative to enhance the level of evidence supporting the results obtained in our study. Second, because the included studies only examined short-term outcomes, no long-term outcomes such as HCC incidence, liver transplantation, or death could be addressed. Lack of analyses on safety issues, such as renal impairment and osteoporosis, was another limitation of our study. Furthermore, there were variations observed among studies in terms of the normal value for transaminase level and the detectable threshold of HBV DNA levels. These disparities introduce the potential for bias in the NMA. Last, the studies included in the NMA were all conducted in Asia-Pacific countries. Since genotypes B and C are predominant in these regions, it is uncertain whether the findings can be generalized to other regions where genotype A or D is predominant. While several observational studies outside the Asia-Pacific region have reported results consistent with our findings on VR, their inclusion in our NMA was precluded due to methodological constraints [42,43].

In conclusion, the present study found that TDF and TAF may be preferable to ETV in terms of VR, and ETV showed superior performance in terms of BR. Our analyses can serve as a useful reference for NA selection strategies for treatment-naïve CHB patients. Further analyses of high-quality RCTs and long-term outcomes are needed to better understand the efficacy of each of the NA drugs and to guide clinicians in selecting the appropriate option.