Besifovir dipivoxil maleate versus other antivirals in reducing hepatocellular carcinoma in chronic hepatitis B

Abstract

Besifovir dipivoxil maleate (BSV) has potent antiviral efficacy against chronic hepatitis B (CHB). This study investigated the efficacy of BSV in reducing hepatocellular carcinoma (HCC) development compared to other antiviral therapy (AVT) agents. We conducted a retrospective cohort study on treatment-naïve patients with CHB who initiated an AVT between 2017 and 2022 with BSV (n = 486), entecavir (ETV) (n = 852), tenofovir alafenamide (TAF) (n = 801), or tenofovir disoproxyl fumarate (TDF) (n = 750). The incidence and hazard ratio (HR) of HCC were calculated. The incidence of HCC in BSV users (n = 6, 4.3 per 1000 person-years [PYs]) was similar to that in TAF users (n = 21, 9.2 per 1000 PYs, log-rank P = 0.086, HR = 2.191, 95% confidence interval [CI] 0.884–5.434), but significantly lower than that in ETV users (n = 38, 12.5 per 1000PYs, log-rank P = 0.026, HR = 2.627, 95% CI 1.103–6.255) and TDF users (n = 32, 12.3 per 1000PYs, log-rank P = 0.028, HR = 2.623, 95% CI 1.090–6.311). Similarly, compared to BSV users, the adjusted HRs for ETV, TAF, and TDF users were higher after stabilized inverse probability of treatment weighting (2.836, 2.784, and 3.294, respectively) and pairwise propensity score matching (3.200, 3.250, and 3.750, respectively) (all P < 0.05). BSV demonstrated comparable efficacy in HCC reduction compared to other AVTs.

Introduction

Chronic hepatitis B (CHB) infection contributes to the development of more than half of hepatocellular carcinoma (HCC) cases in the endemic areas¹. Direct mechanisms, such as hepatitis B virus (HBV) DNA integration into the host genome, expression of the viral regulatory or mutant proteins dysregulating cell proliferation control and hepatocyte transformation, and epigenetic changes affecting tumor suppressor genes, and indirect mechanism, such as chronic inflammation, fibrosis, and cirrhosis development, and other host and environmental factors, are all involved in HBV-related carcinogenesis^1,2. The current mainstay of CHB treatment is antiviral therapy (AVT) using the oral nucleos(t)ide analogs (NAs) with high-genetic barriers, among current guidelines, such as entecavir (ETV), tenofovir disoproxil fumarate (TDF), and tenofovir alafenamide (TAF)^3–6. Although NAs fail to eradicate HBV due to the limited activities on the reverse transcriptase domain of the viral polymerase⁷, NA treatments improve patients’ prognosis by suppressing viral replication, thereby attenuating liver-related complications and HCC development^8,9.

Besifovir dipivoxil maleate (BSV), an oral acyclic nucleotide phosphonate, is a novel potent AVT approved in May 2017 in South Korea¹⁰. BSV has provided a similar antiviral efficacy and a lower incidence of kidney and bone damage compared to those of TDF through the phase 3 trial¹¹. Moreover, BSV provided histologic improvement and a decrease in intrahepatic cccDNA in the treatment-naïve patients with CHB who underwent paired biopsy¹². Hence, BSV is recommended as the first-line AVT in treatment-naïve patients with CHB with or without cirrhosis in the Korean guideline⁶. However, data is limited regarding the influence of BSV on hepatocarcinogenesis. Yim et al. suggested that the decreased incidence of HCC after BSV therapy was through the decreased standardized incidence ratio calculated by the existing HCC prediction models¹³. Kim et al. presented that BSV offers comparable efficacy to TAF using national data (n = 41,949)¹⁴. Considering that the long-term viral suppression by ETV and TDF reduces the risk of hepatocarcinogenesis^15,16, BSV may also have comparable effects. However, a direct comparison analysis of hepatocarcinogenesis among the potent NAs, including BSV, has not been conducted.

Therefore, this study aimed to reveal the long-term risk reduction of HCC development with BSV use in treatment-naïve patients with CHB through a retrospective comparative analysis among the potent NAs.

Results

Patient characteristics

According to the inclusion and exclusion criteria, 2,889 patients from the six tertiary hospitals were retrospectively recruited (Fig. 1). Of the study population, 486 (16.8%) patients started BSV, 852 (29.5%) started ETV, 801 (27.7%) started TAF, and 750 (26.0%) started TDF at the index date. The proportion of males (61.3% vs. 50.2%–55.3%), cirrhosis (38.5% vs. 31.1%–34.4%), and significant alcohol intake (11.9% vs. 5.5%–10.2%) were higher in the BSV group than the other groups (all P < 0.05). Patients in the ETV group were significantly older (median 54 years vs. 47–50 years), and had a higher proportion of hypertension (28.3% vs. 14.8%–17.1%), diabetes mellitus (19.5% vs. 9.9%–10.4%), chronic kidney disease (9.0% vs. 1.2%–1.9%), and negativity for HBeAg (64.2% vs. 44.1%–45.6%) (all P < 0.05) (Table 1).

Fig. 1.

Flowchart of patient selection. Abbreviation: CHB, chronic hepatitis B; AVT, antiviral therapy; HCC; hepatocellular carcinoma; BSV, besifovir dipivoxil maleate; ETV, entecavir; TAF, tenofovir alafenamide; TDF, tenofovir disoproxil fumarate.

Table 1.

Baseline characteristics and development of HCC in patients with chronic hepatitis B.

Variable (n = 2,889)	BSV (n = 486)	ETV (n = 852)	ASD*	TAF (n = 801)	ASD*	TDF (n = 750)	ASD*	P value	MASD**
Male sex	298 (61.3)	428 (50.2)	0.225	443 (55.3)	0.122	400 (53.3)	0.162	0.001	0.225
Age (year)	50.0 (42.0, 57.0)	54.0 (46.0, 61.0)	0.379	49.0 (40.0, 57.0)	0.063	47.0 (37.0, 54.0)	0.258	< 0.001	0.608
Body mass index (kg/m2)	24.0 (22.1, 26.0)	23.2 (21.5, 25.6)	0.182	23.5 (21.5, 25.6)	0.156	23.4 (21.4, 25.7)	0.152	0.013	0.182
Hypertension	72 (14.8)	241 (28.3)	0.332	125 (15.6)	0.022	128 (17.1)	0.062	< 0.001	0.332
Diabetes mellitus	48 (9.9)	166 (19.5)	0.274	83 (10.4)	0.016	74 (9.9)	0.000	< 0.001	0.274
Chronic kidney disease	6 (1.2)	77 (9.0)	0.359	31 (3.9)	0.168	14 (1.9)	0.051	< 0.001	0.359
Dyslipidemia	32 (6.6)	64 (7.5)	0.036	64 (8.0)	0.054	29 (3.9)	0.122	0.005	0.175
Cirrhosis	187 (38.5)	293 (34.4)	0.085	253 (31.6)	0.145	233 (31.1)	0.156	0.029	0.156
Significant alcohol intake	58 (11.9)	47 (5.5)	0.229	82 (10.2)	0.054	61 (8.1)	0.127	< 0.001	0.229
Platelet count (× 109/L)	181.0 (143.0, 218.0)	182.0 (138.5, 227.0)	0.028	179.0 (144.0, 219.0)	0.002	181.0 (135.5, 229.0)	0.052	0.998	0.052
AST (IU/L)	61.0 (41.0, 96.0)	43.0 (27.0, 76.0)	0.042	60.0 (39.0, 102.0)	0.036	55.0 (33.0, 103.8)	0.141	< 0.001	0.156
ALT (IU/L)	84.0 (44.0, 139.0)	43.5 (23.0, 89.0)	0.178	88.0 (39.0, 151.0)	0.015	67.5 (34.0, 133.8)	0.036	< 0.001	0.189
Gamma-GT (IU/L)	44.0 (26.0, 83.0)	42.0 (22.0, 87.5)	0.008	41.0 (26.0, 77.8)	0.011	43.0 (23.0, 94.0)	0.056	0.324	0.056
Serum albumin (g/dL)	4.3 (4.1, 4.5)	4.2 (3.8, 4.5)	0.364	4.3 (4.1, 4.5)	0.016	4.2 (3.9, 4.5)	0.303	< 0.001	0.364
Total bilirubin (mg/dL)	0.8 (0.7, 1.0)	0.8 (0.5, 1.1)	0.087	0.8 (0.6, 1.1)	0.064	0.8 (0.6, 1.1)	0.146	< 0.001	0.146
BUN (mg/dL)	13.4 (11.0, 15.5)	13.8 (11.2, 17.4)	0.306	13.0 (11.0, 15.6)	0.068	12.3 (9.9, 15.0)	0.172	< 0.001	0.392
Serum creatinine (mg/dL)	0.80 (0.70, 0.91)	0.79 (0.67, 0.94)	0.226	0.81 (0.68, 0.94)	0.090	0.79 (0.64, 0.91)	0.004	0.013	0.226
Prothrombin time (INR)	1.03 (0.97, 1.09)	1.03 (0.98, 1.11)	0.104	1.03 (0.97, 1.09)	0.004	1.03 (0.98, 1.11)	0.098	0.262	0.104
Positive for HBeAg	261 (54.5)	298 (35.8)	0.382	439 (55.9)	0.027	398 (54.4)	0.002	< 0.001	0.410
Log10 HBV DNA (IU/mL)	6.39 (5.23, 7.73)	5.17 (3.17, 6.75)	0.634	6.33 (4.95, 7.84)	0.072	6.00 (3.91, 7.52)	0.320	< 0.001	0.634
4—7 Log10 HBV DNA (IU/mL)	237 (48.8)	349 (41.0)	0.157	373 (46.6)	0.044	314 (41.9)	0.139	0.011	0.157
ALBI grade 2	94 (19.4)	234 (27.6)	0.279	161 (20.2)	0.038	208 (27.8)	0.278	< 0.001	0.279
ALBI grade 3	1 (0.2)	19 (2.2)		3 (0.4)		16 (2.1)
FIB-4	1.9 (1.3, 3.1)	2.0 (1.3, 3.6)	0.150	1.9 (1.2, 3.2)	0.012	1.8 (1.1, 3.5)	0.101	0.103	0.150
LS value (kPa) (n = 1,517)	8.1 (5.6, 12.1)	6.7 (4.9, 10.9)	0.078	7.4 (5.5, 11.5)	0.074	7.2 (5.1, 12.0)	0.035	0.003	0.101
CAP value ≥ 239 dB/m (n = 1,437)	135 (48.4)	182 (42.4)	0.120	163 (40.8)	0.154	135 (41.0)	0.148	0.196	0.154
CAP value ≥ 274 dB/m (n = 1,437)	61 (21.9)	81 (18.9)	0.074	76 (19.0)	0.071	63 (19.1)	0.067	0.757	0.074
Development of HCC	6 (1.2)	38 (4.5)	0.195	21 (2.6)	0.101	32 (4.3)	0.186	0.004	0.195
Follow-up duration (months)	35.6 (25.4, 42.9)	45.0 (28.0, 58.2)	0.550	32.6 (20.7, 48.9)	0.027	43.5 (25.1, 56.7)	0.453	< 0.001	0.550

Values are expressed as number (%) and median (interquartile range).

*ASD between the BSV group and each antiviral therapy group.

**MASD among all pairwise comparisons of the four groups.

Abbreviations: HCC, hepatocellular carcinoma; BSV, besifovir dipivoxil maleate; ETV, entecavir; TAF, tenofovir alafenamide; TDF, tenofovir disoproxil fumarate; ASD, absolute standardized difference; MASD, maximum ASD; AST, aspartate aminotransferase; ALT, alanine aminotransferase; Gamma-GT, gamma-glutamyl transferase; BUN, blood urea nitrogen; INR, international normalized ratio; HBeAg, hepatitis B e antigen; ALBI, Albumin-bilirubin,; FIB-4, fibrosis-4 index; LS, liver stiffness; CAP, controlled attenuation parameter.

Comparison of HCC development between groups before matching

Over a median follow-up period of 38.6 months (IQR, 23.9–53.4: 35.6, 45.0, 32.6, and 43.5 in the BSV, ETV, TAF, and TDF groups, respectively), HCC developed in 97 (3.4%) patients (Table 1). The cumulative 2-, 3-, and 5-year incidence rates were 1.5%, 3.1%, and 6.0%, respectively, with an incidence rate of 10.4 per 1000 person-years (PYs). In the subgroup with cirrhosis, HCC developed in 81 (8.4%) patients over a median follow-up period of 39.6 (IQR, 25.3–53.8) months, with the cumulative 2-, 3-, and 5-year incidence rate of 3.9%, 7.7%, and 14.6%, respectively.

Patients in the BSV group developed HCC with an incidence rate of 4.3 per 1000 PYs, which was statistically similar to that in the TAF group (9.2 per 1000 PYs, log-rank P = 0.086, crude HR = 2.191, P = 0.090), however, significantly lower than those in the ETV group (12.5 per 1000 PYs, log-rank P = 0.026, crude HR = 2.627, P = 0.029) and the TDF group (12.3 per 1000 PYs, log-rank P = 0.028, crude HR = 2.623, P = 0.031) (Table 2 and Fig. 2A). ETV, TAF, and TDF use (vs. BSV, HR = 2.938, 2.810, and 3.553, respectively) was significantly associated with an increased risk of HCC development, together with older age, hypertension, cirrhosis, higher level of gamma-glutamyl transferase, and baseline HBV DNA within 4–7 log₁₀ IU/mL (all P < 0.05) (Table S2 and Table 3).

Table 2.

Comparisons of cumulative incidence of HCC and HR between treatment groups.

Treatment	HCC (%)	Cumulative incidence (%)			Incidence rate*	log-rank P value	Univariable Cox regression		E value (lower**)
		2-year	3-year	5-year			Crude HR (95% CI)	P value
All (n = 2,889)	97 (3.4)	1.5	3.1	6.2	10.4	0.135	-	-	-
BSV (n = 486)	6 (1.2)	1.2	1.2	1.8	4.3	-	Reference	-	-
ETV (n = 852)	38 (4.5)	1.2	3.6	6.8	12.5	0.026	2.627 (1.103, 6.255)	0.029	4.694 (1.441)
TAF (n = 801)	21 (2.6)	1.7	3.4	7.4	9.2	0.086	2.191 (0.884, 5.434)	0.090	NA
TDF (n = 750)	32 (4.3)	1.9	3.4	6.9	12.3	0.028	2.623 (1.090, 6.311)	0.031	4.686 (1.403)

*Incidence rate, per 1000 person-years.

**lower confidence limit.

Abbreviation: HCC, hepatocellular carcinoma; HR, hazard ratio; CI, confidence interval; BSV, besifovir dipivoxil maleate; ETV, entecavir; TAF, tenofovir alafenamide; TDF, tenofovir disoproxil fumarate.

Fig. 2.

Kaplan–Meier curves for the cumulative incidence of hepatocellular carcinoma of unweighted groups (A), after stabilized IPTW (B), after pairwise propensity score matching between BSV and ETV (C), BSV and TAF (D), and BSV and TDF (E). Abbreviations: IPTW, inverse probability treatment weighting, BSV, besifovir dipivoxil maleate; ETV, entecavir; TAF, tenofovir alafenamide; TDF, tenofovir disoproxil fumarate.

Table 3.

Multivariable Cox regression analyses.

Variables	Univariable P value	Mutivariable Cox regression
		P value	HR (95 CI)
BSV	-	-	Reference
ETV vs. BSV	0.029	0.026	2.938 (1.139, 7.575)
TAF vs. BSV	0.090	0.040	2.810 (1.048, 7.536)
TDF vs. BSV	0.031	0.009	3.553 (1.366, 9.242)
Age (year)	< 0.001	0.013	1.030 (1.006, 1.054)
Hypertension	< 0.001	0.018	1.722 (1.098, 2.699)
Cirrhosis	< 0.001	< 0.001	5.876 (3.087, 11.18)
Significant alcohol intake	0.002	0.084	1.699 (0.931, 3.101)
Platelet count (× 109/L)	< 0.001	0.053	0.996 (0.992, 1.000)
ALT (IU/L)	< 0.001	0.077	0.997 (0.994, 1.000)
Gamma-GT (IU/L)	0.019	0.03	1.002 (1.000, 1.004)
Serum creatinine (mg/dL)	0.388	0.236	0.667 (0.342, 1.303)
4—7 Log10 HBV DNA (IU/mL)†	< 0.001	0.030	1.631 (1.048, 2.537)

^†The proportional hazard assumption is not met. The hazard ratio is provided for descriptive purposes.

Abbreviation: HR, hazard ratio; CI, confidence interval; BSV, besifovir dipivoxil maleate; ETV, entecavir; BSV, besifovir dipivoxil maleate; ETV, entecavir; TAF, tenofovir alafenamide; TDF, tenofovir disoproxil fumarate; ALT, alanine aminotransferase; Gamma-GT; gamma-glutamyl transferase.

Comparisons after four-group stabilized inverse probability of treatment weighting (IPTW)

After stabilized IPTW, regardless of chronic kidney disease, and the baseline HBV DNA, the differences between the four groups were balanced (maximum absolute standardized mean differences (ASD) < 0.2) (Table S3). The incidence rate of HCC development was significantly lower in the weighted BSV group (3.4 per 1000 PYs) than those in the ETV, TAF, and TDF groups (11.6, 9.7, and 13.6 per 1000 PYs, respectively, log-rank P = 0.005) (Table 4 and Fig. 2B). IPTW-adjusted Cox regression analyses including the unbalanced variables showed that the risk of HCC development was independently higher in the ETV (aHR = 2.836, P = 0.023, E-value = 5.118), TAF (aHR = 2.784, P = 0.031, E-value 5.013), and the TDF groups (aHR = 3.294, P = 0.010, E-value = 6.044) than that in the BSV group (Table 4).

Table 4.

Cumulative incidence of HCC after IPTW and each 1:1 PSM.

IPTW	n	HCC (%)	Follow-up (PYs)	Incidence rate*	log-rank P value	Cox regression		E value (lower**)
						Adjusted HR† (95 CI)	P value†
BSV	312	3 (1.0)	891.6	3.4	0.005	Reference	-	-
ETV	726	31 (4.3)	2670.0	11.6		2.836 (1.151,6.986)	0.023	5.118 (1.569)
TAF	719	20 (2.8)	2051.5	9.7		2.784 (1.097,7.069)	0.031	5.013 (1.422)
TDF	610	29 (4.8)	2135.0	13.6		3.294 (1.336,8.121)	0.010	6.044 (2.007)
1:1 PSM	Pairs***	HCC (%)	Follow-up (PYs)	Incidence rate*	log-rank P value	Cox regression		E value (lower**)
						Adjusted HR (95 CI)	P value
BSV	486	6 (1.2)	1405.0	4.3	-	Reference	-	-
ETV	486	28 (5.8)	1835.1	15.3	0.011	3.200 (1.163,8.803)‡	0.024‡	5.853 (1.599)
TAF	486	17 (3.5)	1436.6	11.8	0.029	3.250 (1.060,9.967)	0.039	5.954 (1.311)
TDF	486	27 (5.6)	1724.5	15.7	0.012	3.750 (1.245,11.299)	0.019	6.961 (1.796)

*Incidence rate, per 1000 person-years.

**lower confidence limit.

***For every 1:1 PSM, 486 pairs are yielded.

^†IPTW-adjusted HR, adjusted with chronic kidney disease, and HBV DNA.

^‡PSM-adjusted HR, adjusted with serum albumin.

Abbreviations: IPTW, inverse probability treatment weighting; PSM, propensity-score matching; n, number; HCC, hepatocellular carcinoma; PYs, person-years; HR, hazard ratio; CI, confidence interval; BSV, besifovir dipivoxil maleate; ETV, entecavir; TAF, tenofovir alafenamide; TDF, tenofovir disoproxil fumarate.

Comparisons between the propensity score (PS)-matched groups

Because the baseline characteristics were not significantly different among the unmatched groups, each 1:1 PS matching yielded well-balanced pairs, with all BSV users (n = 486) included in the matching (all ASD < 0.2), except for serum albumin (ASD = 0.255) between the BSV and TAF groups (Table S4). Compared to the matched BSV group, the matched ETV, TAF, and TDF groups showed significantly higher incidence rates (4.3 vs. 15.3, 11.8, and 15.7 per 1000PYs, respectively, all P < 0.05) and a higher risk of HCC development (aHR and E-value, 3.200 and 5.853, 3.250 and 5.954, and 3.750 and 6.961, respectively, all P < 0.05) (Table 4 and Figs. 2C–E).

Subgroup analyses with baseline vibration-controlled transient elastography (VCTE) and cirrhosis

To further clarify the potential bias due to liver fibrosis, we conducted IPTW and PSM analyses on a subgroup of patients who had reliable baseline VCTE results (324 BSV, 431 ETV, 431 TAF, and 331 TDF users) (Table S5). ETV use (vs. BSV, HR = 3.828, P = 0.032) was associated with an increased risk of HCC development in the multivariable analysis, together with hypertension, cirrhosis, lower platelet count, baseline HBV DNA within 4–7 log₁₀ IU/mL, and higher liver stiffness (LS) values (all P < 0.05) (Table S6). After four-group IPTW, TAF and TDF users had a similar risk of HCC development: however, ETV users had a higher risk compared to BSV users (aHR = 4.404, P = 0.009, E-value = 8.277) (Tables S7, S8).

In patients with cirrhosis (n = 966), stepwise Cox regression analyses compared to the BSV group showed that the risk of HCC was significantly higher in the TDF group (aHR = 3.270, P = 0.018) (Table S9). There was also a trend toward a higher risk in the ETV and TAF groups (aHR = 2.746–2.748, P ≤ 0.051) (Table S9).

AVT switch

AVT switches were applied in 13 cases in the BSV group (3 to ETV, 6 to TAF, and 4 to TDF), 51 cases in the ETV group (3 to BSV, 26 to TAF, and 22 to TDF), 16 cases in the TAF group (6 to ETV and 10 to TDF), and 51 cases in the TDF group (1 to BSV, 12 to ETV, and 38 to TAF). Most switches from BSV occurred due to persistent low-level viremia (LLV) (1–2 log₁₀ IU/mL of HBV DNA after BSV use), except for one patient who switched to TDF due to pregnancy. No significant adverse drug reactions were reported in the BSV group. Among patients who switched AVTs, only one TDF user who had received treatment for 30.9 months developed HCC 31.3 months after switching to TAF. AVT was discontinued in three patients (2 ETV users and 1 TDF user) after achieving a functional cure, and none of these patients developed HCC during the follow-up period.

Decompensation and liver transplantation (LT)

During the follow-up periods, 9 non-BSV users experienced decompensation events such as ascites or varieal bleeding, of whom 2 patients after the HCC development. None of them were died or underwent LT. Three patients underwent LT, of whom 2 patients after the HCC development. One patients underwent emergent LT due to acute liver failure followed by cholangitis, and immediately died.

Discussion

To date, it has been known that BSV has similar clinical efficacy to ETV and TDF, including achieving undetectable HBV DNA levels, ALT normalization, and HBeAg seroconversion, with the added advantage of causing fewer side effects such as osteoporosis or renal dysfunction^6,11,17. Moreover, the similar efficacy and clinical outcome between BSV and TAF users was already reported by Kim et al.¹⁸ In the study, BSV showed similar biochemical and virological response rates compared to TAF, and the cumulative probabilities of HCC was not significantly different between the two AVT groups (P= 0.31)¹⁸. Yim et al. also reported that BSV users without cirrhosis have a decreased incidence of HCC than the predicted incidence calculated by existing HCC prediction models, indicating a further risk reduction¹³. Kim et al. showed that the incidence of HCC was statistically similar between the matched BSV and TDF groups (1.6 vs. 2.2 per 1000 PYs, P= 0.284) using the Health Insurance Review and Assessment Service in South Korea¹⁴. The present study incorporated a wide range of clinical variables, including LS measurements, allowing for a more precise comparison in real-world settings.

Considering that debates persist about whether TDF is significantly more advantageous than ETV in reducing the risk of hepatocarcinogenesis for patients with CHB, although both are beneficial^19,20,{Choi, 2019 #40} we conducted a multicenter large-volume retrospective cohort-based direct comparison between BSV and other potent NAs for the risk of hepatocarcinogenesis. Therefore, BSV users showed similar or significantly lower cumulative incidence and HR of HCC development compared to ETV, TAF, and TDF users, both before and after four-group IPTW and pairwise PSM analyses.

Our study had several important clinical implications. First, to the best of our knowledge, our study is the first to directly compare the benefits of BSV in the development of HCC with other potent AVTs in a large cohort (n = 2,889) with treatment-naïve CHB patients. ETV, TAF, and TDF use had significantly higher HR of hepatocarcinogenesis in the IPTW-weighted and pairwise PS-matched cohorts. Moreover, the calculated high E-values (5.013–6.961) suggested that any unmeasured confounders not addressed would need to increase the risk of HCC by at least 5–6 times to affect the findings of our study. Considering that fibrosis is a clinically significant factor, we conducted a subgroup analysis of patients who underwent TE, which is more accurate than Fibrosis-4 index. This analysis also demonstrated that BSV users have a similar or lower risk of developing HCC, especially compared to ETV users (HR 3.556–4.404, E value 6.571–8.277). Besides the use of AVT, cirrhosis, high LS value, and moderate levels of baseline HBV DNA were associated with HCC development. This is similar to recent studies suggesting a binomial on-treatment HCC risk associated with baseline HBV DNA levels^21,22.

Second, while the short follow-up period (median 38.6 months) makes it difficult to fully trust BSV’s superiority, the following mechanisms may account for our findings. BSV is an acyclic nucleotide phosphonate with a chemical structure similar to acyclic nucleoside phosphonates such as TDF and TAF. Current experimental evidence suggests that TDF and TAF attenuate liver fibrosis and hepatocarcinogenesis via various non-viral effects that ETV does not. Tenofovir prevents progression and promotes the regression of liver fibrosis by regulating inflammasome signaling pathways and the differentiation, activation, and proliferation of hepatic stellate cells^23,24. Moreover, TDF and TAF upregulate expression of the tumor suppressor p7 trans-regulated protein 3, thereby attenuating carcinogenesis by the Wnt/β-catenin signaling pathway²⁵. Although this experimental evidence is not directly available for BSV, it may share the additional advantages of TDF and TAF, such as interferon-λ3 induction in the gastrointestinal tract induced, immunomodulation of lipopolysaccharides-mediated cytokine production, and inhibition of Akt and mTOR pathways²⁶, which may result in the reported histologic improvement and decrease in intrahepatic cccDNA¹², as well as an attenuated risk of HCC development in our study.

Third, another consideration for the attenuated hepatocarcinogenesis in BSV users is the concurrent administration of L-carnitine supplementation due to L-carnitine depletion, a significant side effect observed in the phase IIb results²⁷. While this administration strategy may cause inconvenience for patients, it could potentially improve the outcomes of patients with CHB. The potential utility of L-carnitine for preventing hepatocarcinogenesis has been proposed in the literature. In an animal model, L-carnitine administration to Long-Evans Cinnamon rats prevented mitochondrial injury induced by chronic inflammation, resulting in significantly lower degrees of hepatic damage and inhibition of hepatocarcinogenesis²⁸. Moreover, a high-fat diet-induced steatohepatitis mouse model receiving L-carnitine showed substantially improved liver histology and a reduced incidence of liver tumors, with decreased activation of oncogenic signaling pathways^29,30. Clinically, patients in the highest quartile of serum L-carnitine concentration had a reduced risk of overall cancer (odds ratio = 0.67) compared to those with low L-carnitine concentration³¹. Therefore, L-carnitine supplementation might benefit BSV users more than patients with CHB who initiate other NAs, in addition to the antiviral efficacy of BSV Whether the combination of L-carnitine with other potent NAs also provides additional benefits requires further prospective studies.

Fourth, although most regimen changes from BSV occurred due to detectable HBV DNA, these BSV users maintained LLV without exacerbation or hepatocarcinogenesis. In in vitro drug susceptibility assays, the lamivudine-resistance mutations rtL180M (M) and rtM204V (V) were potentially associated with BSV resistance³². However, HBV mutants with primary mutations conferring resistance to adefovir or tenofovir were found to be susceptible to BSV³³. Therefore, BSV may serve as an alternative drug for patients with tenofovir-resistant HBV mutations that do not include the MV sequence³³. Despite debates, LLV during potent AVT with good adherence is not a predictive factor for HCC⁸. Our study also demonstrates that only one case of regimen change from TDF to ETV developed HCC, indicating that first-line BSV use has non-inferiority compared to the other AVTs in the reduction of HCC risk in treatment-naïve patients with active CHB.

Several limitations remained. First, the retrospective nature of our study had inevitable selection bias. Second, although patients were recruited during the same study period, we could not exclude the possibility that differences in follow-up durations between BSV or TAF users and ETV or TDF users may have influenced the incidence of HCC. Moreover, the limited events in BSV group (n = 6) may reduce statistical reliability. Notably, another recent study using Korean health insurance data also reported a similarly short median follow-up period (2.3–2.4 years)¹⁴. Since BSV is a relatively newly introduced agent, these data represent the most up-to-date evidence currently available. Third, the limited availability of BSV in South Korea mitigated the direct application of the present results to countries with different genotype dominance where BSV was currently unavailable. Fourth, the beneficial influence of BSV on other clinical outcomes, such as cirrhosis development, kidney function decline, and osteopenia, was not evaluated in the present study.

In conclusion, BSV showed comparable efficacy in reducing HCC development in patients with CHB compared to TAF, ETV, and TDF, indicating promising efficacy. Given the small number of events and the relatively short follow-up duration observed in BSV users, further long-term analysis is warranted.

Methods

Study population

This multicenter, retrospective cohort study recruited treatment-naïve patients with CHB who initiated AVT with ETV, TDF, TAF, or BSV, from six tertiary academic medical institutes in South Korea (Severance Hospital, Seoul National University Hospital, Korea University Anam Hospital, Korean University Ansan Hospital, Soonchunhyang University Bucheon Hospital, and Catholic University of Korea Bucheon St. Mary’s Hospital) between January 2017 and April 2022. Exclusion criteria were as follows: (1) prior experience of AVT or interferon-based treatment before 2017; (2) age < 19 or ≥ 80 years; (3) decompensated cirrhosis; (4) HCC development or liver transplantation before or within 12 months from AVT initiation; (5) follow-up duration < 12 months; (6) incomplete baseline data; (7) other viral hepatitis such as hepatitis C virus infection; (8) uncontrolled malignancy; and (9) other significant illnesses. This study conformed to the ethical guidelines of the Declaration of Helsinki (1975) and was approved by the Institutional Review Board of the Yonsei University College of Medicine, Seoul, South Korea. The written informed consent waived by the Institutional Review Board of the Yonsei University College of Medicine due to the retrospective nature of the study.

AVT initiation and data collection

AVT was initiated in accordance with the practice guidelines of the Korean Association for the Study of the Liver^6,34 (Table S1). An index date was defined as the date AVT was initiated. Cirrhosis was histologically or clinically diagnosed as follows: 1) platelet count of < 150 × 10³/μL or imaging findings suggestive of cirrhosis, including a blunted, nodular surface accompanied by splenomegaly (> 12 cm); or 2) clinical signs of portal hypertension³⁵. ETV and TDF were administered at 0.5 mg and 300 mg daily, respectively, and the doses were appropriately reduced according to the patient’s kidney function. TAF was administered at 25 mg daily as a fixed dose. BSV was administered at 150 mg daily as a fixed dose with supplementary L–carnitine at 600 mg daily²⁷.

Patients with CHB routinely visited the outpatient clinic at 3–6 month intervals, and information on their laboratory data (such as blood chemistry test, viral marker, and serum levels of HBV DNA), abdominal image study (such as abdominal ultrasonography), change in their AVT from the index date until loss of follow–up or HCC development were collected^6,8. LS value and controlled attenuation parameter (CAP) was measured using VCTE (FibroScan®, EchoSens, Paris, France) in a standard way³⁶. Only LS and CAP values with at least 10 valid measurements, a success rate of at least 60%, and an interquartile range (IQR) to median ratio of < 30% were considered reliable. Patients with CAP values greater than 239 dB/m and 274 dB/m were considered to have mild and moderate steatosis, respectively³⁷.

The primary outcome was HCC development between the index date and April 2023. HCC was diagnosed based on histological evidence or by dynamic computed tomography and/or magnetic resonance imaging findings (nodule > 1 cm with arterial hypervascularity and portal-/delayed-phase washout)^38–40..

Statistical analysis

Baseline characteristics and the primary outcome were described as medians (IQRs) for continuous variables and numbers (percentages) for categorical variables. Kruskal–Wallis test was used for continuous variables, and Chi-squared test or Fisher’s exact test was used for categorical variables to assess statistical differences among the treatment groups. The cumulative incidence of HCC was evaluated using the Kaplan–Meier method, and were compared using the log-rank test. Cox proportional hazards models were used to assess the association among the treatment groups and outcome. The proportional hazards assumption was examined using the Schoenfeld residuals tests. To account for possible factors that could have influenced the attending physician’s choice of AVT, stabilized IPTW was adopted to create a “pseudo-sample” that is balanced in the distribution of baseline characteristics among the treatment groups. Furthermore, we conducted a sensitivity analysis using PS matching with a 1:1 optimal matching algorithm for pairwise comparisons. Cox proportional hazards models were performed, including imbalanced confounders (ASDs ≥ 0.2) after weighting for doubly robust estimation. An E-value with their corresponding 95% confidence interval lower limits was calculated to evaluate the robustness of the primary outcome to potential unmeasured confounders²³. Subgroup analysis of patients with VCTE results were performed before and after the IPTW analysis. For further details, see the Supplementary information. All statistical analyses were performed using R (version 4.3.3; http://cran.r-project.org/). Two-sided P values of < 0.05 were considered to indicate statistical significance.

Author contributions

JS Lee, Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Validation, Visualization, Writing – original draft, Writing – review & editing; SW Lee, Conceptualization, Funding acquisition, Investigation, Project administration, Resources, Writing – original draft, Writing – review & editing HL Lee, Data curation, Investigation, Resources, Writing – review & editing; JJ Yoo, Data curation, Investigation, Resources, Writing – review & editing; YS Seo, Conceptualization, Funding acquisition, Investigation, Project administration, Resources, Writing – review & editing; SJ Yu, Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Writing – review & editing; HJ Yim, Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Writing – review & editing; YK Jung, Data curation, Investigation, Resources, Writing – review & editing; J Moon, Data curation, Formal analysis, Methodology, Validation, Visualization; HW Lee, Resources, Writing – review & editing; MN Kim, Resources, Writing – review & editing; BK Kim, Resources, Writing – review & editing; JY Park, Resources, Writing – review & editing; DY Kim, Resources, Writing – review & editing; SH Ahn, Resources, Writing – review & editing; SG Kim, Conceptualization, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing, Funding acquisition; SU Kim, Conceptualization, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing, Funding acquisition.

Funding

This study was supported by ILDONG Pharmaceutical, Inc. The funders played no role in the study design, data collection, analysis, interpretation, or the writing of the manuscript.

Data availability

The data that support the findings of this study are available from the co-corresponding authors upon reasonable request.

Declaration

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-13456-8.