Article Text

Download PDFPDF
Original research
No difference in hepatocellular carcinoma risk between chronic hepatitis B patients treated with entecavir versus tenofovir
  1. Feng Su1,
  2. Kristin Berry2,
  3. George N Ioannou1,2,3
  1. 1 Division of Gastroenterology, University of Washington, Seattle, Washington, USA
  2. 2 Health Services Research and Development, Veterans Affairs Puget Sound Healthcare System, Seattle, Washington, USA
  3. 3 Division of Gastroenterology, Veterans Affairs Puget Sound Healthcare System, Seattle, Washington, USA
  1. Correspondence to Dr Feng Su, University of Washington, Seattle, WA 98195-0005, USA; fsu{at}medicine.washington.edu

Abstract

Objective Entecavir (ETV) and tenofovir disoproxil fumarate (TDF) are first-line agents for the treatment of chronic hepatitis B (CHB). Recent studies have challenged the assumption that these agents are equally effective at preventing hepatocellular carcinoma (HCC). We aimed to determine whether the risk of HCC and mortality differ in patients with CHB treated with ETV and TDF.

Design We performed a retrospective cohort study of Veterans Affairs patients with CHB in the USA who initiated treatment with ETV or TDF between the dates of Food and Drug Administration approval of these medications and 1 January 2017. Multivariable Cox proportional hazards regression was used to determine the association between antiviral therapy and HCC risk as well as the risk of death or liver transplantation. Propensity score adjustment and competing risks analysis were performed.

Results We identified 2193 ETV-treated and 1094 TDF-treated patients who were followed for a mean of 5.4 years. We found no difference in the risk of HCC in ETV-treated versus TDF-treated patients (adjusted HR (aHR) 1.00, 95% CI 0.76 to 1.32). Results were similar in propensity score adjusted and competing risks analysis, and in multiple sensitivity analyses. We also found no difference in the risk of death or liver transplantation (aHR 1.16, 95% CI 0.98 to 1.39).

Conclusions We found no difference in the risk of HCC between patients with CHB treated with ETV versus TDF. Our results support current guideline recommendations that both agents are appropriate first-line options for the treatment of CHB.

  • hepatitis B
  • hepatobiliary cancer
  • hepatocellular carcinoma
  • liver

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Significance of this study

What is already known on this subject?

  • The nucleo(s)tide analogues entecavir and tenofovir disoproxil fumarate are recommended as first-line agents for treatment of chronic hepatitis B infection. Both reduce the risk of long-term complications of hepatitis B including hepatocellular carcinoma.

  • Recent observational studies from Korea and Hong Kong suggest that tenofovir may be superior to entecavir in reducing the risk of hepatocellular carcinoma, but subsequent studies have produced mixed results.

  • Large observational studies of patients with chronic hepatitis B in non-endemic countries are still needed to determine whether findings of these recent studies generalise to other populations.

What are the new findings?

  • In a large cohort of patients with chronic hepatitis B treated with entecavir or tenofovir who received care through the US Veterans Affairs healthcare system, there was no association between antiviral regimen and the risk of hepatocellular carcinoma.

  • There was also no association between antiviral regimen and the risk of death or liver transplantation.

  • There was no association between antiviral regimen and the risk of hepatocellular carcinoma or death/liver transplant in patients with or without cirrhosis.

How might it impact on clinical practice in the forEseeable future?

  • These results support current guideline recommendations that consider both entecavir and tenofovir as first-line agents for the treatment of chronic hepatitis B infection.

Introduction

Chronic hepatitis B (CHB) is a major global health concern with an estimated 240 million infected individuals worldwide.1 Complications of CHB, such as cirrhosis and hepatocellular carcinoma (HCC), lead to substantial morbidity and mortality. While complete eradication of hepatitis B virus (HBV) is not yet achievable, long-term viral suppression is possible with antiviral therapy. Benefits of antiviral treatment include normalisation of liver enzymes, improvements in liver histology and a reduction in the risk of HCC and other CHB-related complications.2 3

Current guidelines recommend the nucleo(s)tide analogues entecavir (ETV), tenofovir disoproxil fumarate (TDF) or tenofovir alafenamide as first-line therapy for CHB patients with an indication for treatment.1 4 Of the first-line agents, ETV and TDF are more commonly used as tenofovir alafenamide was introduced recently . There are limited data on the comparative effectiveness of ETV and TDF in preventing long-term complications of CHB, but both are known to have potent antiviral activity and a high genetic barrier to resistance.

Both ETV and TDF reduce HCC risk among patients with CHB.2 3 However, a recent cohort study of patients with CHB in Korea suggested that TDF may be associated with a lower risk of HCC compared with ETV.5 Although the initial publication was retracted, corrected results continued to show that TDF was superior to ETV in terms of HCC risk.6 Results of subsequent studies have been mixed. Three separate studies from Korea detected no difference in HCC risk between the two antiviral regimens,7–9 while a study in Hong Kong showed superiority of TDF compared with ETV.10

Notably, these studies were all conducted in countries where HBV is endemic. Studies of patients in non-endemic countries are still needed, as these patients may differ with respect to the natural history of HBV infection and the risk of HCC development. To date, there have been no large studies of predominantly non-Asian patients in non-endemic countries. Major shifts in prescribing patterns would be warranted if these findings were shown to be consistent across different patient populations. We, therefore, undertook this study to compare the incidence of HCC between patients treated with ETV versus TDF in a cohort of Veterans with CHB in the USA. Our secondary aim was to compare ETV and TDF with respect to the incidence of death or liver transplantation.

Methods

Data source

The Veterans Affairs (VA) Health Administration serves >9 million Veterans at 170 medical centres and 1063 outpatient clinics throughout the USA.11 It uses a nationally integrated electronic healthcare information network known as the Veterans Information Systems and Technology Architecture (VistA). The VA Corporate Data Warehouse (CDW) is a continually updated repository of data from VistA created for research purposes. We extracted healthcare data for our study population using the CDW.

Patient and public involvement

Patients and the public were not involved in the conceptualisation of this study.

Patient population

We conducted a retrospective cohort study of patients with CHB treated with ETV or TDF. We identified all patients with a positive hepatitis B surface antigen or viral load (HBV DNA) who initiated ETV or TDF before 1 January 2017 (n=4183). We excluded combination TDF regimens used for patients with HIV infection with or without HBV (eg, TDF/emtricitabine) because these medications are never used for treatment of HBV without HIV coinfection (introducing potential bias when comparing to TDF) and because we would not be able to distinguish the effects of TDF from those of other medications in the combination. We excluded 703 patients who initiated ETV before 29 March 2005 or TDF before 11 August 2008 (the dates these medications received Food and Drug Administration approval for HBV treatment), 9 who started ETV and TDF simultaneously, 109 who had HCC prior to treatment initiation and 75 who received liver transplantation prior to antiviral initiation, leaving 3287 patients in the analysis.

Outcomes

The primary outcome was HCC. The diagnosis of HCC was identified by International Classification of Diseases (ICD)-9 or ICD-10 codes documented at least twice (see online supplementary table 1). The ICD-9 code-based definition of HCC using VA records has a positive predictive value of 84%–94% compared with chart extraction12–14 and has been widely used in the literature.15–19 We evaluated death or liver transplantation as a secondary outcome. The date of death was identified from VA records, which are supplemented with the social security death index. Liver transplantation was determined by ICD9/10 diagnostic and procedure codes (see online supplementary table 1) and Current Procedural Terminology codes (47135, 47136).

Supplemental material

Baseline characteristics evaluated as potential confounders

We ascertained baseline characteristics and laboratory tests at the time of antiviral initiation (table 1, online supplementary table 1). HIV coinfection was determined using ICD-9/10 codes recorded at least twice before the date of first antiviral prescription. Hepatitis C virus (HCV) coinfection was defined by a positive HCV viral load prior to antiviral initiation. Laboratory tests reflect values closest to the treatment start date within the preceding 6 months. The Charlson Comorbidity Index (CCI) was calculated using relevant ICD-9/10 codes.20 Missing data are shown in online supplementary table 2. In multivariate models, for variables with missing data, a separate categorical variable was created whereby patients were categorised as ‘missing’ if they had missing data.

Table 1

Baseline characteristics of patients on entecavir and tenofovir

Statistical analysis

Association between antiviral therapy and HCC or death/liver transplantation

We used multivariate Cox proportional hazards regression to determine the association between antiviral therapy and HCC risk. Follow-up started on the date of first prescription of antiviral therapy and continued to 1 June 2018. Patients who did not develop HCC were censored at the date of death, liver transplantation or last follow-up. Patients who switched antiviral therapy or were escalated to combination therapy were analysed according to the initial medication they received in an intention-to-treat analysis. We adjusted for baseline characteristics measured at or shortly before the time of antiviral initiation. We did not adjust for viral suppression or alanine aminotransferase (ALT) normalisation after antiviral initiation because these are direct consequences of antiviral therapy and potential mediators of the effect of antiviral therapy on HCC risk. Adjusting for these factors could eliminate any ability to detect a difference between ETV and TDF in terms of HCC risk (if one were to truly exist). Since patients who died or underwent liver transplant are no longer at risk for HCC of their native liver, we also used competing risks analysis to evaluate the association between antiviral treatment and HCC, with death or liver transplantation considered a competing risk. We evaluated death or liver transplantation as a secondary outcome. Patients who did not die or require liver transplantation were censored at the date of last follow-up or 1 June 2018, whichever occurred first. The Kaplan-Meier method was used to generate cumulative incidence curves for HCC and death/liver transplant. We also evaluated death alone as an outcome and censored patients who received liver transplant at the time of transplantation. Subgroup analyses were performed in patients with and without cirrhosis.

Propensity score adjusted analysis

We additionally used propensity score (PS) adjustment to account for the nonrandom assignment of patients to ETV and TDF. We used an inverse probability weighting (IPW) method based on the estimated PS.21 Missing information due to censoring required specialised methods of estimation. The censoring process was assumed independent of outcome after adjusting for the confounders in a Weibull model. Conditional independence of treatment assignment was assumed by the inclusion of confounders in a logit model. Confounders adjusted for were age, race/ethnicity, sex, body mass index, diabetes, CCI, alcohol use disorder, substance use disorder, HIV, HCV, cirrhosis, decompensated cirrhosis, HBV viral load, hepatitis B e antigen (HBeAg) status, platelet count, aspartate aminotransferase/ALT, albumin, bilirubin, international normalised ratio and creatinine. After the application of IPW, sufficient overlap was assumed. Weights were generated using stteffects IPW in Stata V.15.22

Treatment as a time-varying exposure

In our primary analysis, we considered treatment a fixed exposure in an intention-to-treat analysis. To capture the effect of treatment duration and treatment changes over time, we performed a secondary analysis modelling treatment as a time-varying exposure. In this analysis, the risk of HCC or death/liver transplantation is compared during periods of time when a patient is on ETV versus periods of time when a patient is on TDF. For example, a patient who started ETV at time 0 then stopped treatment at time t was analysed under ETV from time 0 to time t. The HR describes risk at any given survival time based on duration of treatment. The comparison groups are any patients who have not experienced the event or been censored by that time. The duration of treatment (exposure) is then used to estimate the HR. The time-varying HR, therefore, compares the risk of HCC or death/transplant based on duration of treatment in that instant.

Sensitivity analyses

We conducted several sensitivity analyses. First, we limited the population to patients who initiated treatment after August 2008 in order to follow ETV-treated and TDF-treated patients during contemporaneous time periods. Second, we excluded HCC or death/liver transplantation within the first 180 days of antiviral initiation because (1) some tumours could have been present at the time of antiviral initiation and (2) outcomes that occurred shortly after antiviral initiation may not be attributable to the effect of the antiviral medication. We also excluded patients who developed outcomes within the first 360 days of antiviral initiation. Lastly, we excluded patients with prior treatment experience, HIV or HCV coinfection, decompensated cirrhosis, alcohol abuse or patients who switched therapy to more closely match the inclusion criteria used in prior studies.5 7

All analyses were performed using Stata MP V.15.1 (64-bit)22 except for competing risks for which R V.3.5.3 (64-bit) was used.23 24

Results

Baseline characteristics

We identified 2193 patients with HBV who initiated ETV and 1094 who initiated TDF before 1 January 2017. The majority of patients were male (96%) and mean age was 56.1 years (table 1). Most patients were non-Hispanic white (41.6%) or non-Hispanic black (37.1%) and 12.8% identified as ‘Other’. Demographic characteristics were similar in the two groups. ETV-treated and TDF-treated patients were similar with respect to the proportion with alcohol use disorder, cirrhosis, HCV coinfection and prior interferon or adefovir exposure. Compared with TDF-treated patients, ETV-treated patients had a higher CCI and a higher proportion had diabetes, HIV and prior lamivudine exposure. A greater proportion of ETV-treated patients were HBeAg positive and correspondingly median viral load was higher than in TDF-treated patients.

Antiviral therapy and HCC risk

Out of 3287 patients, 252 developed HCC during a mean follow-up of 5.4 years (incidence 1.44 per 100 patient-years (p-y)). Among ETV-treated patients, 167 developed HCC during a mean follow-up of 5.6 years (incidence 1.36 per 100 p-y), while 85 patients on TDF developed HCC during a mean follow-up of 4.7 years (incidence 1.64 per 100 p-y). Cumulative incidence curves reflected a slightly lower incidence of HCC in patients on ETV compared with those on TDF (figure 1A,B). In unadjusted analysis, ETV was associated with a lower risk of HCC than TDF but the confidence interval crossed one (crude HR 0.85, 95% CI 0.65 to 1.10 comparing ETV to TDF) (table 2). After adjustment for baseline confounders, there was no association between antiviral treatment and HCC risk (adjusted HR (aHR) 1.00, 95% CI 0.76 to 1.32). There was also no association after PS adjustment and in competing risks analysis (table 2).

Figure 1

(A) Cumulative incidence of HCC in patients on entecavir versus tenofovir. (B) Cumulative incidence of HCC in patients on entecavir versus tenofovir using inverse probability of treatment propensity score weights. (C) Cumulative incidence of death or liver transplantation among patients treated with entecavir or tenofovir. (D) Cumulative incidence of death or liver transplantation among patients treated with entecavir or tenofovir using inverse probability of treatment propensity score weights.

Table 2

Association between treatment with entecavir versus tenofovir and the risk of HCC

Antiviral therapy and risk of death or liver transplant

Cumulative incidence curves for death/liver transplantation are presented in figure 1C,D. Death or liver transplantation occurred in 784 patients (746 deaths and 38 liver transplantations) over a mean follow-up of 5.5 years (incidence 4.4 per 100 p-y). Among patients on ETV, 582 died or underwent liver transplant over a mean follow-up of 5.7 years (incidence 4.6 per 100 p-y), while 202 TDF-treated patients died or underwent liver transplant over a mean follow-up of 4.9 years (incidence of 3.8 per 100 p-y). In unadjusted analysis, ETV was associated with a higher risk of death/liver transplantation compared with TDF (crude HR 1.29, 95% CI 1.10 to 1.51) (table 3). After adjustment for baseline confounders, the association was attenuated (aHR 1.16, 95% CI 0.98 to 1.39). Results were similar after PS adjustment (table 3). When death was considered the endpoint and liver transplantation a censoring event, results were also similar (see online supplementary table 3 and supplementary figures 1A,B).

Supplemental material

Table 3

Association between treatment with entecavir versus tenofovir and the risk of death or liver transplantation

Antiviral therapy as a time-varying exposure

When treatment was modelled as a time-varying exposure, there was no association between antiviral therapy and HCC risk (aHR 1.18, 95% CI 0.84 to 1.67) (see online supplementary table 4). ETV was associated with a higher risk of death/liver transplantation than TDF (HR 1.40, 95% CI 1.12 to 1.75) in unadjusted analysis, but the association was attenuated and no longer significant in adjusted analysis (aHR 1.19, 95% CI 0.95 to 1.50). Results were similar when death was evaluated as an isolated outcome (see online supplementary table 4).

Subgroup and sensitivity analyses

In subgroup analysis, there was no association between antiviral therapy and HCC risk in patients with or without cirrhosis (table 4). There was also no association between antiviral therapy and death/liver transplantation in patients with or without cirrhosis.

Table 4

Subgroup analyses evaluating the association between treatment with entecavir versus tenofovir in patients with or without cirrhosis

In sensitivity analysis, after excluding patients who started antivirals before August 2008, we found no association between antiviral regimen and HCC risk (aHR 0.99, 95% CI 0.74 to 1.32) (table 2). Results were similar after excluding patients diagnosed with HCC in the first 180 days after antiviral initiation (aHR 0.98, 95% CI 0.71 to 1.35). Analyses adjusted for PS and accounting for competing risks produced similar results (table 2). There was no association between antiviral therapy and HCC risk in other sensitivity analyses (see online supplementary table 5).

There was no association between antiviral therapy and death/liver transplant after excluding patients who initiated therapy before August 2008 (aHR 1.09, 95% CI 0.91 to 1.31), or after excluding patients who died or received liver transplantation in the first 180 days (aHR 1.21, 95% CI 0.99 to 1.48) (table 3). However, when excluding patients who died or underwent liver transplant within the first 360 days of antiviral initiation or patients who switched therapy, ETV was associated with a higher risk of death/liver transplant (aHR 1.27, 95% CI 1.02 to 1.58 and aHR 1.32, 95% CI 1.09 to 1.60, respectively) (see online supplementary table 6). There was no association between antiviral regimen and death/liver transplant in other sensitivity analyses (see online supplementary table 6).

Discussion

Current HBV treatment guidelines do not specify a preference for a particular first line nucleo(s)tide analogue in most patients with an indication for treatment.1 4 The decision to use ETV versus TDF is often guided by factors such as prior antiviral exposure, comorbid conditions and cost rather than evidence of superiority of one agent over the other. In our analysis of a large, ethnically diverse cohort of patients in the USA with HBV infection, we found no difference in the risk of HCC between patients on ETV and TDF.

Our results differ from those of a large, national cohort study of CHB patients in Korea that showed a lower risk of HCC in TDF-treated than ETV-treated patients (aHR 0.68, 95% CI 0.59 to 0.77).5 The initial publication contained errors that led to retraction of the original article, but the association remained significant in corrected results.6 A recent study from Hong Kong also showed that TDF was associated with a lower risk of HCC than ETV (aHR 0.36, 95% CI 0.16 to 0.80).10 However, only 8 HCCs occurred in the TDF group and there was a large disparity in the number of patients receiving ETV (28 041) and TDF (1309), raising questions about the comparability of the two groups.25

In contrast, other studies of patients with CHB have not demonstrated any difference between ETV and TDF with respect to HCC risk. A multi-centre study of 2879 patients from Korea found no association between antiviral therapy and HCC risk (aHR 0.98, 95% CI 0.75 to 1.27).7 Two single-centre studies from Korea also concluded that there was no difference in HCC risk between patients receiving TDF and ETV.8 9 It is unclear why results of these studies differ from the national cohort study. Each study did have slightly different inclusion criteria. While the national cohort study included patients with decompensated cirrhosis,5 the others did not.7–9 Other subtle differences include the fact that outcomes were variably excluded 6 months or 1 year after antiviral initiation in different studies, patients were assigned to a treatment group only if they received treatment for a specified duration in some but not all studies, and the national cohort study performed a ‘per-protocol’ analysis in that patients who switched therapy were censored 6 months afterwards. It has also been noted that the Kaplan-Meier curve in the national cohort study suggests that no HCCs developed in TDF-treated patients after approximately 2 years of follow-up whereas HCCs continued to occur in the ETV-treated group, which is somewhat implausible.25

All aforementioned studies are from countries where HBV is endemic. Repeating these studies in non-endemic countries is crucial to determine the generalisability of the results. Our study included a large sample of non-Asian patients with CHB, most of whom likely had a different natural history of HBV infection than patients in endemic countries. Patients in our study also differ with respect to important factors that influence HCC risk, such as demographic characteristics and comorbidities. Given the dearth of data on the effects of antiviral therapy in non-Asian patients with HBV in non-endemic countries, our results are an important addition to the existing body of literature. Other ‘Western’ studies to examine this issue include an analysis of the Chronic Hepatitis Cohort Study cohort, which included 407 TDF-treated and 415 ETV-treated patients (of whom >60% were Asian) and found no association between antiviral regimen and HCC risk.26 The direction of the association differed in Asian (aHR 0.70, 95% CI 0.29 to 1.68 comparing TDF to ETV) and non-Asian patients (aHR 1.87, 95% CI 0.60 to 5.87), but there was no statistically significant interaction between race/ethnicity and antiviral treatment. A study based on the Real-World Evidence from the Asia-Pacific Rim Liver Consortium for HBV international consortium included 997 patients from the USA and also found no association between antiviral therapy and HCC risk, however, 88.2% of these patients were Asian.27 A prior observational study of Caucasian patients in a non-endemic country found no association between antiviral therapy and HCC risk,28 but the primary aim was not to compare antiviral regimens and as such it may not have been sufficiently powered for this purpose.5 29

Collectively, our results and those of other studies show that there is insufficient evidence to conclude that TDF is superior to ETV at reducing HCC risk. Moreover, a plausible biological mechanism to explain such a difference is lacking. One proposed hypothesis is that ETV may have carcinogenic properties based on studies in mice,5 but the doses of ETV used in these studies far exceed the typical dose used for treatment.30 Furthermore, these results have not been replicated in humans. Indeed, a cohort study of patients with CHB in Hong Kong failed to show any difference between ETV and other antivirals with respect to the risk of extrahepatic malignancies.31 Recent studies also suggest that nucleotide analogues (adefovir, TDF) may differ from nucleoside analogues (lamivudine, ETV) in their ability to modulate lipoprotein saccharide-mediated cytokine production and to induce interferon-λ3, which may impact their antiviral and antitumour effects.32 33 Further studies are needed to determine the clinical importance of these mechanisms. A more conceivable explanation is that TDF and ETV may differ slightly in effectiveness of viral suppression. Prior studies have suggested that among HBeAg-positive patients, TDF may be more effective at viral suppression than ETV.34 In the hospital-based cohort of the study by Choi et al, a greater proportion of TDF-treated patients achieved virological response, but virological response was not independently associated with HCC risk in multivariable analysis.5

With respect to our secondary outcome, ETV-treated patients had a higher risk of death/liver transplantation than TDF-treated patients in unadjusted analysis. We suspect this is because ETV-treated patients had more comorbidities, as reflected by a higher mean CCI score and a higher prevalence of diabetes and HIV coinfection (table 1). In other studies, ETV-treated patients tended to be older with more advanced liver disease and comorbidities than TDF-treated patients.10 27 Clinicians may favour ETV in patients who are perceived to be sicker, perhaps because TDF has been linked to renal dysfunction and decreased bone density. A ‘patient warehousing’ phenomenon has also been proposed, whereby the sickest patients received ETV when it was first approved, leaving a healthier pool of patients eligible for TDF when it became available later.27 Adjusting for baseline confounders including the CCI score substantially attenuated the association between ETV and death/liver transplantation in our study. Additionally, when the cumulative incidence curves for death/liver transplant were stratified by CCI score, the curves for TDF-treated and ETV-treated patients essentially overlapped (see online supplementary figure 2). Of note, in sensitivity analyses excluding patients who switched therapy and excluding outcomes in the first 360 days after antiviral initiation, there was a statistically significant association between antiviral regimen and death/liver transplantation. However, these analyses were not PS adjusted and, when modelling treatment as a time-varying exposure, we did not observe a significant association between antiviral therapy and death/liver transplantation.

Supplemental material

Notably, the absolute incidence of death/liver transplantation was higher in our cohort than in recent studies from Korea5 7–9 and in a prior study of Caucasian patients with CHB in Europe.35 Patients in our study differ substantially from patients in other studies with respect to characteristics that would be expected to impact mortality. Our goal was to determine whether the results of prior studies of ETV versus TDF generalise to a real world, representative population of patients with CHB in the USA. As such, we elected to include patients with HCV or HIV coinfection, treatment-experienced patients and patients with alcohol use disorder, substance use disorder, diabetes and did not exclude patients based on other comorbid conditions. Our observed mortality rate is similar to that reported in VA-based studies of patients with CHB36 and HCV,37 and on the same order of magnitude as a US-based cohort of non-VA patients with CHB.38

Several limitations should be considered when interpreting our results. First, we did not account for adherence to antiviral therapy in our primary analysis. Instead, we prespecified an intention-to-treat analysis. If our patients had poor adherence to therapy, a true difference between ETV and TDF in terms of HCC risk could have been diluted. To some extent, however, the time-varying exposure analysis would be expected to account for differences in antiviral adherence. Second, we cannot guarantee that the same type or intensity of follow-up occurred in both treatment groups, which means that ascertainment of HCC could have differed in the two groups. Third, given the observational nature of our study, we cannot exclude the possibility of unmeasured confounding. However, the availability of detailed information in the national VA CDW allowed us to ascertain all baseline variables we considered to be relevant confounders. Lastly, men comprised 96% of our study population, and therefore, it is conceivable that results may be different among women.

In summary, in a large cohort of patients with CHB in the USA, there was no difference in the risk of HCC in patients treated with ETV versus TDF. Given the conflicting results of our study and several recent observational studies, there may be sufficient clinical equipoise to justify prospective, randomised trials to definitively determine whether first-line antiviral agents differ meaningfully with respect to their effectiveness in preventing HCC and other long-term CHB-related complications. However, such a trial may be practically difficult as it would require many years of follow-up and large sample sizes to achieve adequate statistical power. In the absence of randomised studies, observational studies must be relied on to inform practice decisions. Our results support current guideline recommendations that consider both ETV and TDF as first-line agents for treatment of CHB.

References

Footnotes

  • Twitter @FengSu_MD

  • Correction notice This article has been corrected since it published Online First. The author order has been amended.

  • Contributors FS conceptualisation, methodology, visualisation, writing of original draft, critical editing and review of final draft. KB formal analysis, methodology, visualisation, review of final draft. GNI supervision, conceptualisation, methodology, critical editing and review of final draft.

  • Funding The study was funded by a VA CSR&D grant I01CX001156 to GNI. FS was supported by NIH T32 grant 5T32DK007742-22.

  • Disclaimer The funding source played no role in study design, collection, analysis or interpretation of data. The contents do not represent the views of the US Department of Veterans Affairs or the US Government.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Patient consent for publication Not required.

  • Ethics approval The study was approved by the Institutional Review Board of the VA Puget Sound Healthcare System.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are deideintified participant data, available from KB (kberry@alum.mit.edu) and GNI (georgei@medicine.washington.edu) on reasonable request.