The Risk of Hepatocellular Carcinoma in Entecavir versus Tenofovir Treated U.S. Cohort with Chronic Hepatitis B Virus

The comparative effectiveness of tenofovir (TDF) versus entecavir (ETV) in reducing the risk of hepatocellular carcinoma (HCC) in patients with chronic hepatitis B (HBV) remains unclear. Data from a retrospective Korean cohort study published by Choi et al. initially suggested lower than expected incidence of HCC in patients on long-term TDF¹. However, additional studies from Korea did not show a statistically significant difference in HCC incidence rate between TDF and ETV groups^2,3, and subsequent studies reported mixed results ranging from no association or a slight advantage for TDF⁴. Most of these studies examined Asian patients from Korea, Taiwan, and China.

The largest U.S. study examined a cohort including 2193 ETV-treated and 1094 TDF-treated patients in the national Department of Veterans Affairs (VA) database and reported no significant association between the type of HBV treatment of HCC risk. However, this study was limited by the possibility of HCC misclassification, incomplete measurement of exposure to medications and adjustment for confounders⁵.

We therefore re-examined the association between TDF vs. ETV and HCC risk in an updated VA cohort of patients. We extracted electronic data from the VA Corporate Data Warehouse (CDW) on patients 18 to 90 years of age with ≥ 1 positive hepatitis B serum surface antigen test between October 1, 1999 and December 31, 2018, who had at least one filled prescription for TDF or ETV. The primary exposure variable was treatment type (TDF vs. ETV) defined using 2 different treatment variables. We classified patients according to the first initiated treatment into TDF or ETV group, and created a time-updating variable of current use status. We also evaluated the cumulative duration of years based on daily prescription fill data where gaps did not count and overlapping prescriptions were added, and created a time updating variable of cumulative duration. The primary outcome was new cases of HCC recorded after the index date of first filled prescription of TDF or ETV through last visit recorded in the VA, or July 31, 2019. We used a multi-step process to identify incident HCC in CDW and the VA Central Cancer Registry (CCR). (Supplement). We adjusted for age, race, sex, baseline cirrhosis (ICD codes prior to or FIB-4 ≥2.67 within 1 year of index date)⁶, diabetes, alcohol use, hepatitis C coinfection, chronic kidney disease, previous HBV treatment, HBV e antigen, HBV DNA, ALT and AST levels.

We calculated incidence rates (IR) per 1000 person-years (PY) of follow up for HCC. We examined the risk for HCC by treatment using unadjusted and adjusted Cox proportional hazard regression analyses to generate hazard ratios (HR) and accompanying 95% confidence intervals (CI). Each analysis was also conducted after excluding patients who developed HCC in the first year after index date. We constructed a propensity score for receiving TDF vs. ETV using age, race, gender, cirrhosis, alcohol abuse, diabetes, chronic kidney disease, HBV DNA, and HBV e antigen. We adjusted for the propensity scores by: 1) adding the propensity probability in the model, 2) adding the logit of the propensity score in the model, 3) model Cox strata defined by quintiles of the propensity score and; 4) caliper matching of the treatment groups using the propensity score^7–9.

The study cohort comprised of 1,769 (47.4%) who received TDF and 1,966 (52.6%) who received ETV as their first HBV treatment. The mean age was 57.0 years, and 95.3% were men. Patients treated with TDF were less likely to be have HBVeAg, diabetes, chronic kidney disease or previous HBV treatment (Supplementary Table 1). After accounting for gaps and overlaps in treatment, patients were on TDF for 2.4 years (SD, 2.4) and ETV for 2.3 years (SD, 2.3). All patients on TDF were treated with 300 mg while 95.7% of ETV patients were treated with a daily 1 mg dose. During a follow up of 7230 PY in the TDF group, 84 patients developed HCC for an IR of 1.16% (95% CI: 0.93% - 1.44%) and 102 developed HCC during 8137 PY in the ETV group for an IR of 1.25% (95% CI: 1.02% - 1.52%). When the duration of follow up was calculated as time on HBV treatment, the IR of HCC while on TDF was 1.34% (95% CI: 1.02% - 1.73%) and ETV was 1.55% (95% CI: 1.21% - 1.95%) (Table 1).

Table 1:

Incidence rates (IR) and hazard ratios (HR) for comparison of HBV treatment type (tenofovir vs. entecavir) measured two different ways on risk of hepatocellular cancer (HCC) by time period of HCC exclusion from treatment start date.

	# HCC	Person years	IR (%) (95% CI)	Unadjusted HR (95% CI)	P-value	Adjusted HR* (95% CI)	P-value
HCC any time after index
Overall	186	15366.9	1.21 (1.04 – 1.40)	N/A		N/A
Duration of Treatment †
Tenofovir	59	4396.37	1.34 (1.02 – 1.73)	0.91 (0.81 – 1.01)	0.0821	0.89 (0.79 – 0.99)	0.0380
Entecavir	72	4647.24	1.55 (1.21 – 1.95)	1.0
Current Treatment ‡
Tenofovir	84	7230.01	1.16 (0.93 – 1.44)	0.87 (0.62 – 1.23)	0.4357	0.86 (0.60 – 1.21)	0.3738
Entecavir	102	8136.85	1.25 (1.02 – 1.52)	1.0
Patients with HCC in 1st year removed
Overall	125	12062.5	1.04 (0.86 – 1.24)	N/A		N/A
Duration of Treatment †
Tenofovir	34	3275.81	1.04 (0.72 – 1.45)	0.90 (0.80 – 1.01)	0.0684	0.88 (0.79 – 0.99)	0.0345
Entecavir	47	3384.32	1.39 (1.02 – 1.85)	1.0
Current Treatment ‡
Tenofovir	53	5662.51	0.94 (0.70 – 1.22)	0.74 (0.48 – 1.16)	0.1869	0.73 (0.47 – 1.13)	0.1590
Entecavir	72	6400.03	1.13 (0.88 – 1.42)	1.0

^*Adjusted for age, race, sex, cirrhosis, alcohol abuse, chronic kidney disease, HBV DNA <2000, HBVeAg status, previous HBV treatment, HCV status

^†Cumulative duration of years

^‡Time updated current medication

Time periods of follow-up are classified at the patient level to entecavir, tenofovir or no drug. The sum of these are the total person years denominator. Similarly, a patient’s events (if any) are classified by the current exposure at the time of the event (again, entecavir, tenofovir or no drug). The sum of these counts form the numerator for the incidence rates.

When examining HBV treatment as time updating current medication variable, there was a small but not statistically significant reduction in the risk of developing HCC in patients treated with TDF (adjusted HR=0.86; 95%CI: 0.60–1.21). The HR values were similar regardless of the adjustment method (Table 1). However, when examining exposure to antiviral medications as time updating cumulative duration, there was a moderate reduction in HCC risk in patients receiving TDF compared with ETV (adjusted HR=0.89, 95% CI: 0.79 – 0.99). The association was similar after removing patients diagnosed with HCC in the first year from the analysis (adjusted HR=0.88; 95% CI: 0.79 – 0.99)). The reduced HCC risk for patients with TDF was also observed in all propensity score adjusted models but missed significance when matching by propensity score (Supplementary Figure 1).

Compared with the study by Su et al.⁵, we expanded the treated cohort to include one additional year of enrollment and follow up duration through 12/31/2018 (vs. 1/6/2018) resulting in a larger cohort (3,735 vs 2387). We reduced misclassification by verifying HCC diagnosis and excluded of 22% of cases. Our analysis considered treatment overlaps and gaps, and thus likely offered more accurate estimates of exposure.

In conclusion, our study indicated a small reduction (11%-12%) in HCC risk among patients treated with TDF when medication exposure was measured as duration of treatment. Most patients in our cohort likely acquired HBV as adults from injection drug use or sexual transmission and thus more closely represents the U.S. born patients. Whether this small difference justifies a practice change is not clear and needs to be considered based on further confirmation of the finding and in the context of other aspects of the treatment such as cost and side effects.

Supplementary Material

3

Supplementary Figure 1: Forrest plot of hazard ratios and 95% confidence interval for effect of tenofovir vs. entecavir on HCC risk using several adjustment methods. Treatment modeled as time updating duration of exposure to medication.

Abbreviations:

ETV

Entecavir

TDF

Tenofovir

HBV

Hepatitis B Virus

HCC

Hepatocellular carcinoma

U.S.

United States

CDW

Corporate Data Warehouse

PY

person-years

HR

hazard ratios

CI

confidence intervals