No difference in hepatocellular carcinoma risk in chronic hepatitis B patients treated with tenofovir vs entecavir: evidence from an updated meta-analysis

Abstract

Whether tenofovir disoproxil fumarate (TDF) is superior to entecavir in reducing hepatocellular carcinoma (HCC) risk among treatment-naïve chronic hepatitis B (CHB) patients remains controversial. We aimed to clarify this controversy. Several databases, including PubMed and Embase, were retrieved through November 2020. Cohort studies comparing the effectiveness of TDF and entecavir in reducing HCC incidence among treatment-naïve CHB patients were included if they reported multivariable-adjusted or propensity-score-matched risk estimates. A random-effects model was used to pool hazard ratios (HRs). Thirteen cohort studies, involving 4097 HCC cases and 80202 CHB patients, were included. Multivariable-adjusted meta-analysis revealed no significant difference in HCC incidence between TDF and entecavir groups (HR 0.86, 95% confidence interval 0.72–1.04), which was consistent with propensity-score-matched meta-analysis (HR 0.83, 95% confidence interval 0.66–1.03). Subgroup analysis showed that the observed similarity of TDF to entecavir for HCC prevention persisted in studies with follow-up length of ≥4 years but not in those with follow-up length of <4 years ( P interaction <0.01). In conclusion, TDF is similar to entecavir in reducing HCC incidence among treatment-naïve CHB patients. Heterogeneous results of included studies may result from their disparity in follow-up length. Our findings should be treated with caution and need to be further confirmed.

Use of tenofovir disoproxil fumarate is associated with a lower risk of hepatocellular carcinoma than entecavir in patients with chronic hepatitis B

“…The difference in the follow-up duration due to the asynchronous introduction of the two treatments may cause a bias in either direction. Previous meta-analyses by Tseng et al [33] and Li et al [40] also noticed this point, and they performed subgroup analyses based on follow-up duration. In studies with a follow-up duration of ETV that was longer than TDF by more than one year, TDF was significantly associated with a lower risk of HCC (aHR = 0.69; 95%CI: 0.61-0.79).…”

“…The results of the meta-analyses published so far are summarized in Table 1. Although the studies and patients included in the 11 meta-analyses were slightly different, most of the meta-analyses except for two (Tseng et al [33] and Yuan et al [34] ) reported that TDF was superior to ETV in preventing HBV-associated HCC when the adjusted HR was pooled. Most meta-analyses performed subgroup analyses and/or metaregression to address between-study heterogeneity.…”

“…While studies with a larger number of patients showed superiority of TDF in reducing the risk of HCC compared with ETV, studies with a smaller number of patients did not. Another meta-analysis by Tseng et al [33] addressed this issue by observing the differences in the study setting. TDF showed a similar risk for HCC among hospital-based cohort studies (HR = 1.03; 95%CI: 0.88-1.21), whereas TDF was associated with a lower HCC risk among studies based on the administrative claims database (aHR = 0.67; 95%CI: 0.59-0.76) [33] .…”

“…Another meta-analysis by Tseng et al [33] addressed this issue by observing the differences in the study setting. TDF showed a similar risk for HCC among hospital-based cohort studies (HR = 1.03; 95%CI: 0.88-1.21), whereas TDF was associated with a lower HCC risk among studies based on the administrative claims database (aHR = 0.67; 95%CI: 0.59-0.76) [33] . Tseng et al [33] explained these discrepant results between clinical cohorts and administrative database studies by the difference in residual confounding as the information contained in administrative databases is less precise and prone to errors because of inaccurate coding.…”

“…Determining the optimal sample size for a clinical study is critical to assure an adequate power to detect a clinical significance. To achieve a power of 0.8 for a type I error level of 0.05 with an assumed HR of 0.88 as reported in Tseng et al [33] , 36,364 patients would be required. Thus, studies that concluded that HCC risks were similar between the two drugs are subject to a type II error.…”