Hepatitis B Virus (HBV) Core-Related Antigen During Nucleos(t)ide Analog Therapy Is Related to Intra-hepatic HBV Replication and Development of Hepatocellular Carcinoma

Abstract

Hepatitis B virus (HBV) infection is the major cause of hepatocellular carcinoma (HCC) and affects >350 million people worldwide. Recent advances in developing nucleos(t)ide analog (NA) therapy were effective in suppressing HBV replication and clinically well tolerated without serious adverse effects. Although long-term NA therapy yielded histologically confirmed improvement in inflammation and fibrosis stage [1] among patients chronic hepatitis B, patients who had advanced liver fibrosis and were receiving NA therapy still had a high incidence of HCC [2–5]. Therefore, NA therapy for chronic hepatitis B could not suppress the incidence of HCC satisfactorily; however, the clinical and virological characteristics of patients who developed HCC while undergoing NA therapy have not been elucidated fully.

The HBV core-related antigen (HBcrAg) assay system was developed recently, using a monoclonal antibody isolated from HBV core antigen (HBcAg)–immunized mice [6]. The HBcrAg assay system can detect free HBV e antigen (HBeAg), HBeAg-antibody complex, and precore/core (pre-C/C) proteins in HBV particles (HBcAg and p22cr) [7]. A recent European and Asian study suggested that HBcrAg might be a good virological marker to differentiate HBeAg-negative patients with active disease from those with inactive disease [8]. The HBV DNA load reportedly declined rapidly during NA therapy, while tests for HBcrAg continued to be positive for a longer period [9]. These results suggest the usefulness of monitoring HBcrAg in patients with chronic hepatitis B for the prediction of HCC development during NA therapy. However, clinical and virological studies regarding these issues have been limited [3, 10].

MATERIALS AND METHODS

Patients

We enrolled 109 patients with chronic hepatitis B who initiated NA therapy during 2001–2011 at the Graduate School of Medicine, Kanazawa University Hospital, Japan. All patients received a diagnosis of histologically confirmed chronic hepatitis B, tested positive for hepatitis B surface antigen (HBsAg), and had HBV DNA detected in serum by polymerase chain reaction (PCR)–based methods. All patients had received NA therapy for >2 years. Twelve patients (11%) received lamivudine (LAM), 25 (23%) had treatment switched from LAM to LAM plus adefovir (ADV), 17 (16%) had treatment switched from LAM to entecavir (ETV), and 55 (50%) received ETV. Details are described in the Supplementary Data.

HBV Testing

Serum samples were tested for HBsAg, HBeAg, and anti-HBe antibodies, using a chemiluminescence enzyme immunoassay (CLEIA; Abbot Japan, Tokyo, Japan). HBsAg was measured quantitatively using an Architect HBsAg-QT assay (Abbot). HBcrAg was measured by CLEIA, using a Lumipulse HBcrAg assay (Fujirebio, Tokyo, Japan). Details are described in the Supplementary Data.

Affymetrix GeneChip Analysis, Hierarchical Clustering, and Pathway Analysis of GeneChip Data

Thirteen liver tissue samples collected before treatment and 23 liver tissue samples obtained during treatment were analyzed using an Affymetrix GeneChip. GeneChip data analysis was performed using BRB-Array Tools (available at: http://linus.nci.nih.gov/BRB-ArrayTools.htm) as described previously [11]. The protocol was approved by the Institutional Review Board of Kanazawa University and accorded with the Declaration of Helsinki, good clinical practice guidelines, and local laws and regulations. Written informed consent was obtained from all patients involved in this study. Pathway analysis was performed using MetaCore (Thomson Reuters, New York, New York). Detailed procedures are described in the Supplementary Data.

Cell Lines

Hepatoma (HepAD38) cells were maintained in Dulbecco's modified Eagle's medium/F12 medium (Life Technologies) containing 10% fetal bovine serum, 100 U/mL penicillin, 100 µg/mL streptomycin, 400 µg/mL Geneticin, and 5 µg/mL insulin. HepAD38 cells were cultured in the absence of tetracycline. Primary human hepatocytes (PHHs) from PXB mice were obtained from PhoenixBio (Hiroshima, Japan) and maintained in dHCGM medium [12].

Quantification of HBV DNA and Covalently Closed Circular DNA (cccDNA) by Reverse-Transcription PCR (RT-PCR)

HBV DNA in the cells was extracted using a DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. HBV DNA in the medium was extracted using an SMI TEST EX R&D Kit (MLB, Nagoya, Japan) according to the manufacturer's protocol. HBV DNA was quantified by RT-PCR analysis, using the primer set 5′-ACTCACCAACCTCCTGTCCT-3′ and 5′-GACAAACGGGCAACATACCT-3′ and the probe 5′-FAM-TATCGCTGGATGTGTCTGCGGCGT-TAMRA-3′. Nucleic acids (50 ng) of the extracted DNA were treated for 60 minutes at 37°C with 10 U of Plasmid-safe DNase I (Epicentre, Madison, Wisconsin) and then treated for 30 minutes at 70°C for DNase inactivation. cccDNA was quantified by RT-PCR analysis, using the primer set 5′-CGTCTGTGCCTTCTCATCTGC-3′ and 5′-GCACAGCTTGGAGGCTTGAA-3′ and the probe 5′-FAM-CTGTAGGCATAAATTGGT-MGB-3′.

Metformin and ETV Treatment

PHHs were seeded at 2.0 × 10⁵ cells/well in 12-well plates. After 24 hours, the cells were infected with HBV (kindly provided by Dr Yasuhito Tanaka, Nagoya City University) at 5 genome equivalents/cell in the presence of 4% PEG8000 at 37°C for 12 hours. At 10 days after infection, the cells were treated with metformin, with or without ETV. At 48 hours after treatment, real-time detection PCR (RTD-PCR) and Western blotting were performed.

HepAD38 cells and HepAD38 cells overexpressing LRH1 (ie, HepAD38-LRH1 cells) were seeded at 2.0 × 10⁴ cells/well in 6-well plates. After 24 hours, the cells were treated with ETV, with or without metformin. The medium containing ETV of HepAD38 and HepAD38-LRH1 cells was refreshed every 3 day until 18 days. The medium containing ETV, with or without metformin of HepAD38 cells was refreshed every day until 6 days.

Quantitative RTD-PCR and Western blotting were performed as described in the Supplementary Data.

Statistical Analysis

The results are expressed as mean values ± standard deviations. At least 3 samples were tested in each assay. Significance was tested by 1-way analysis of variance with Bonferroni methods, and differences were considered statistically significant at P values of <.05. Cox proportional hazards regression analysis was used to assess clinical and virological variables that were significantly associated with the development of HCC.

RESULTS

Degree of Hepatic Fibrosis and HBcrAg Positivity Are Independent Significant Predictors of the Development of HCC Among Patients Receiving NA Therapy

One hundred and nine patients with chronic hepatitis B receiving NA therapy for >2 years were enrolled. The mean duration of NA therapy was 6.5 years (range, 2.1–12.7 years). During NA therapy, 36 patients (33%) developed HCC, and the annual incidence of HCC was 5.1% (2.3% in patients with fibrosis stages F1–F2 and 7.2% in patients with fibrosis stages F3–F4). The clinical and virological characteristics of the patients before treatment and at the end of the follow-up period are shown in Table 1. Comparison of the pretreatment clinical and virological characteristics between patients who developed and those who did not develop HCC showed that decreased HBsAg levels, an increased prevalence of anti-HBe positivity, a higher prevalence of genotype C, a higher frequency of α-fetoprotein positivity (≥10 ng/mL), a more severe level fibrosis stage, and a more severe inflammation grade were observed in the HCC group, compared with the non-HCC group. Long-term NA therapy improves clinical and virological features; therefore, we reevaluated the clinical and virological characteristics of the patients at the end of the follow-up period (at mean duration [±SD] of 6.9 ± 3.0 years in the HCC group and 6.3 ± 2.4 years in the non-HCC group). Higher age, lower amount of HBsAg, higher incidence of anti-HBe positivity, higher incidence of HBcrAg positivity, and a higher FIB-4 index were observed in the HCC group, compared with the non-HCC group (Table 1). Multivariate Cox regression analyses showed that, among the pretreatment variables, age (>56 years; hazard ratio [HR], 3.16) and liver fibrosis stage (F3–F4; HR, 3.41) were independently associated with the development of HCC, and, among the end of follow-up variables, age (>60 years; HR, 2.66), HBcrAg positivity (HR, 3.53), and FIB-4 index (>2.1; HR, 2.57) were independently associated with the development of HCC (Table 2).

Kaplan–Meier analysis showed no significant difference in the rate of seroconversion from HBeAg to anti-HBe or the rate of HBV DNA loss between the HCC and non-HCC groups (Figure 1); however, the rate of HBcrAg loss was significantly delayed in the HCC group (Figure 1). Moreover, a significantly higher cumulative incidence of HCC was observed in patients with advanced liver fibrosis (fibrosis stages F3–F4 before treatment; P = .002; Supplementary Data), a high FIB-4 index (>2.1 during treatment; P = .0016; Figure 1), and HBcrAg positivity (during treatment; P = .035; Figure 1). Interestingly, patients with a high FIB-4 index (>2.1) and HBcrAg positivity had an 8-fold higher incidence of HCC than patients who had a low FIB-4 index and were negative for HBcrAg (Figure 1). Thus, fibrosis stage and HBcrAg positivity were significantly associated with the development of HCC during NA therapy.

Figure 1.

Kaplan–Meier analysis of the improvement of virological markers and the incidence of hepatocellular carcinoma (HCC). Upper left, Cumulative anti–hepatitis B virus (HBV) e (HBe) seroconversion ratio between the HCC and non-HCC groups. Upper middle, Overall HBV DNA positivity among the HCC and non-HCC groups. Upper right, Overall HBV core-related antigen (HBcrAg) positivity among the HCC and non-HCC groups. Lower left, Cumulative incidence of HCC among patients with FIB-4 (defined as a score of >2.1) and those without FIB-4 (defined as a score of ≤2.1). Lower middle, Cumulative incidence of HCC among HBcrAg-positive patients and HBcrAg-negative patients. Lower right, Cumulative incidence of HCC among patients with FIB-4(>2.1) and HBcrAg positivity, those with FIB-4(>2.1) or HBcrAg positivity, and those without FIB-4(≤2.1) and HBcrAg negativity.

Open in new tabDownload slide

Twelve patients (11%) received LAM, 25 (23%) had treatment switched from LAM to LAM plus ADV, 17 (16%) had treatment switched from LAM to ETV, and 55 (50%) received ETV. Treatment regimen did not affect the development of HCC (data not shown).

Evaluation of HBV Replication in the Liver of Patients Receiving NA Therapy

To evaluate HBV replication in the liver of patients receiving NA therapy, cccDNA, HBV DNA, and HBV RNA levels were quantitated by TaqMan PCR (Supplementary Data). We first evaluated the relationship between cccDNA and HBV DNA levels in the liver. A significant correlation was observed between cccDNA and HBV DNA levels in the liver before treatment (r² = 0.95, P < .0001; Supplementary Data). However, this correlation was disturbed during treatment (r² = 0.18, P = .024; Supplementary Data). Next, we compared cccDNA, HBV DNA, and HBV RNA levels in the liver of 12 HBcrAg-negative patients and 16 HBcrAg-positive patients receiving NA therapy (Figure 2). Although there was no significant difference in cccDNA levels between HBcrAg-negative and HBcrAg-positive patients, HBV DNA levels were significantly higher in HBcrAg-positive patients than in HBcrAg-negative patients (Figure 2). Similarly, the expression of pregenomic RNA, pre-S/S RNA, and HBV x protein (HBx) RNA was significantly higher in HBcrAg-positive patients than in HBcrAg-negative patients (Figure 2). Interestingly, HBcrAg serum levels were significantly correlated with HBV DNA levels but not cccDNA levels in the liver. These results indicated that transcriptional activity from cccDNA was activated and that more HBV DNA replicates were present in the liver of HBcrAg-positive patients.

Figure 2.

Quantitation of covalently closed circular DNA (cccDNA), hepatitis B virus (HBV) DNA, pregenomic RNA, pre-S/S RNA, and HBV x protein (HBx) RNA in the liver of HBV core-related antigen (HBcrAg)–positive patients (ie, those with an HBcrAg level of ≥3.0 U/mL) and HBcrAg-negative patients (ie, those with an HBcrAg level of <3.0 U/mL) receiving NA therapy. Abbreviations: NA, nucleos(t)ide analog; NS, not significant.

Open in new tabDownload slide

Gene Expression Profile in the Liver of Patients With Chronic Hepatitis B Receiving NA Therapy

To examine which signaling pathways were changed in the liver of HBcrAg-positive patients and HBcrAg-negative patients, hepatic expression profiling was performed. We first compared hepatic gene expression in 13 patients before and during treatment (Supplementary Data and Supplementary Data). Substantial changes in genes expression were observed (2030 genes, P < .001; 5239 genes, P < .01), and representative signaling pathways deduced from the 2030 genes are shown in Supplementary Data. The results indicated that long-term NA therapy substantially improved hepatic gene expression, which was compatible with the improvement of liver histological findings (Supplementary Data). To explore the differential signaling pathways related to the presence of HBcrAg, hepatic gene expression was examined in 11 HBcrAg-negative patients and 12 HBcrAg-positive patients. A relatively low number of differentially expressed genes was obtained (369 genes, P < .01; 1578 genes, P < .05), and we further analyzed the 1578 genes (1077 genes were upregulated and 501 genes downregulated in HBcrAg-positive livers). Viral replication–associated signaling, such as that occurring during protein translation, ubiquitin-proteasome pathway signaling, oxidative phosphorylation, and JNK signaling, were upregulated; in addition, cancer-related signaling, such as that occurring during cytoskeleton remodeling, transforming growth factor β signaling, WNT signaling, telomere signaling, Akt signaling, and Flt3 signaling, was upregulated. Conversely, immune-related signaling, such as that occurring during activation of natural killer cells and T cells, was downregulated (Supplementary Data). The results indicated that active HBV replication was present and that some oncogenic signaling pathways were evoked in HBcrAg-positive livers. For further analysis of the differentially expressed genes, network analysis of direct interactions among the genes (MetaCore) was performed as reported previously [11]. A direct-interactions algorithm created a network from the list of 1077 upregulated genes, and an integrated network including 9 representative genes formed the “hubs” of the network (Figure 3). Interestingly, 5 of the 9 genes (LRH1, PPARα, HNF4α, CEBPα, and SP1) reportedly increase HBV transcription and replication [13], and 4 of the 9 genes (p53, caspase 3, JNK1, and β-catenin) could be related to the host reaction due to HBV replication, mediating apoptosis, oxidative stress, and fibrosis signaling [11]. These results revealed that favorable conditions for HBV replication were preserved in HBcrAg-positive livers.

Figure 3.

Pathway analysis of 1077 upregulated genes (P < .05) in hepatitis B virus (HBV) core-related antigen (HBcrAg)–positive livers, using a direct interaction algorithm. Direct interaction analysis revealed 9 hub genes located at the center of each pathway. The genes with a red circle represent HBV transcription-related factors reported previously (LRH1, HNF4α, PPARα CEBPα, and SP1), and the genes with a green circle represent proapoptosis (caspase 3 and p53), oxidative stress (JNK), and WNT (β-catenin) signaling. This figure is available in black and white in print and in color online.

Open in new tabDownload slide

In Vitro Model of HBV Replication in HBcrAg-Positive Patients Receiving NA Therapy

To explore the possibility that the upregulation of these HBV transcription-related factors (LRH1, PPARα, HNF4α, CEBPα, and SP1) in HBcrAg-positive livers might be directed by HBV proteins, we established HepG2 cells overexpressing each of HBV proteins by using a lentivirus expression system. Among HBV proteins (pre-C/C, pre-S/S, pol, and HBx), pre-C/C increased the expression of LRH1, PPARα, and HNF4α by >1.5-fold (Figure 4A). Among these transcriptional factors, LRH1 efficiently could activate core promoter [14], and we established HepAD38-LRH1 cells (Figure 4B). Using HepAD38-LRH1 cells, we monitored the levels of HBV DNA, pregenomic RNA, and cccDNA during ETV treatment over 18 days. In an ETV-free environment, the levels of HBV DNA, pregenomic RNA, and cccDNA increased over 18 days, and these levels were significantly higher in HepAD38-LRH1 cells than in HepAD38 cells. During ETV treatment, HBV replication was substantially repressed, but the levels of HBV DNA and pregenomic RNA were higher in HepAD38-LRH1 cells than in HepAD38 cells (Figure 4C). In contrast to HBV DNA and pregenomic RNA levels, the differences in cccDNA levels between the 2 cell types became negligible during long-term exposure to ETV (Figure 4C). The findings of in vitro analysis of HBV replication during long-term exposure to ETV resembled those for liver biopsy samples obtained from patients with chronic hepatitis B who were receiving NA therapy.

Figure 4.

Regulation of hepatitis B virus (HBV) transcription-related factors and HBV replication. A, Overexpression of HBV proteins (precore/core [pre-C/C], pre-S/S, pol, and HBV x [HBx]) and HBV transcription-related factors (LRH1, PPARA, and HNF4A) in HepG2 cells. B, Establishment of HepAD38 cells overexpressing LRH1 (ie, HepAD38-LRH1 cells). C, Time course (until 18 days) of HBV DNA, pregenomic RNA, and cccDNA levels in HepAD38-LRH1 cells and HepAD38 cells with or without entecavir (ETV). D, Effect of metformin (Met) on HBV transcription-related factors (LRH1, PPARA, and HNF4A) in HepAD38 cells and primary human hepatocytes (PHHs). The experiments were repeated 3 times (A, B and D) or performed in triplicate and repeated 3 times. *P < .05, **P < .01, and ***P < .001.

Open in new tabDownload slide

Metformin Suppresses HBV Replication in HepAD38 Cells and PHHs

These findings prompted us to suppress HBV replication by modulating the transcription factors preserved in HBcrAg-positive livers. We focused on metformin, an activator of AMPK and a key regulator of cellular metabolism. Western blotting showed that metformin activated AMPK (as determined by the ratio of p-AMPK to AMPK) with or without ETV in HepAD38 cells and PHHs (Figure 4D). In these conditions, the expression of LRH1, PPARα, and HNF4α was substantially repressed by metformin (Figure 4D).

Metformin effectively repressed viral protein synthesis, as observed by the sufficient suppression of HBsAg and HBcrAg levels in the culture medium of HepAD38 cells, while ETV had no effect on the production of HBsAg and HBcrAg (Figure 5A and Supplementary Data). Metformin repressed HBV DNA and cccDNA levels significantly at a concentration of 5 mM, but the extent of suppression at a concentration of 2.5 mM was less than that of ETV (Supplementary Data). The combination of metformin and ETV repressed HBV DNA and cccDNA more significantly than metformin or ETV alone (Figure 5B and Supplementary Data). Metformin substantially repressed the expression of pregenomic RNA in HepAD38 cells, while ETV slightly repressed pregenomic RNA expression (Figure 5C and Supplementary Data). The expression of LRH1, PPARα, and HNF4α increased in control cells in which HBV was replicating (Figure 5C), and the suppression of pregenomic RNA levels was in concordance with the suppression of LRH1, PPARα, and HNF4α levels by metformin (Figure 5C). Similar results were obtained by using PHHs, although neither ETV nor metformin had a suppressive effect on cccDNA levels (Figure 5D). Thus, metformin mainly repressed HBV transcription, while ETV essentially repressed HBV DNA levels, and the combination of metformin and ETV had an additive effect on the suppression of HBV DNA levels.

Figure 5.

Metformin (Met) suppresses hepatitis B virus (HBV) replication in HepAD38 cells and primary human hepatocytes (PHHs). A, Effect of Met and entecavir (ETV) on HBV core-related antigen (HBcrAg) and HBV surface antigen (HBsAg) in the medium of HepAD38 cells. B, Effect of Met and ETV on HBV DNA and cccDNA levels in HepAD38 cells. C, Effect of Met and ETV on the expression of pregenomic RNA, LRH1, PPARA, and HNF4A in HepAD38 cells. D, Effect of Met and ETV on HBcrAg (medium), HBsAg (medium), pregenomic RNA, HBV x protein (HBx) RNA, HBV DNA, and covalently closed circular DNA (cccDNA) levels in PHHs. A–D, The experiments were performed in triplicate and repeated 3 times. *P < .05, **P < .01, and ***P < .001.

Open in new tabDownload slide

DISCUSSION

Although NA therapy is effective for suppressing HBV replication, there is still a high risk of HCC for certain patients with advanced liver fibrosis. A systemic review reported that HCC was diagnosed in 10.8% of NA-naive patients with cirrhosis and in 17.6% of LAM-resistant patients with cirrhosis [5], and the annual incidence of HCC in these patients was 3.2% and 6.8%, respectively. Recent reports from Japan showed that the annual incidence of HCC varied from 4.4% to 6.0% in patients with cirrhosis receiving LAM treatment [2, 4]. In this study, among the 109 patients who had chronic hepatitis B diagnosed on the basis of histological findings and were receiving NA therapy, 36 (33%) developed HCC over a period of 6.5 years. and the annual incidence of HCC was 5.1% (2.3% in patients with fibrosis stages F1–F2 and 7.2% in patients with fibrosis stages F3–F4). Our findings generally confirmed previous findings. The relatively high incidence of HCC in our study might be explained partially by the characteristics of patients with older age, more-severe liver histological findings, a high prevalence of genotype C, and a high rate of BCP mutation (Table 1).

The recently developed HBcrAg assay system can detect free HBeAg, HBeAg-antibody complex, and pre-C/C proteins in HBV particles (HBcAg and p22cr) [6, 7]. However, details about HBcrAg positivity in patients with chronic hepatitis B, especially those who are undergoing NA therapy, had not been evaluated previously. In our study, 27 of 38 patients (71%) who were negative for HBeAg and HBV DNA were positive for HBcrAg. Maasoumy et al speculated that active inflammation is present in HBcrAg-positive livers and that HBcrAg is released from injured hepatocytes [8].

These findings prompted us to evaluate the clinical value of HBcrAg for the prediction of the development of HCC during NA therapy. So far, 2 studies have evaluated the association between HBcrAg and HCC [3, 10]. Kumada et al first reported that HBcrAg levels before treatment were significantly associated with the development of HCC; however, in their study, the significance of HBcrAg during NA therapy was not evaluated. Hosaka et al reported that HBcrAg was a predictor of the posttreatment recurrence of HCC during NA therapy, but the relationship between HBcrAg and primary HCC was not addressed. In those respects, our study differs from theirs. Long-term NA therapy significantly improved virological markers; therefore, clinical markers for the prediction of HCC during NA therapy should be beneficial. In this study, anti-HBe seroconversion and loss of HBV DNA from serum were not significantly different between those with and those without HCC; however, the loss of HBcrAg was significantly delayed in the HCC group (Figure 1). Furthermore, we showed that the combination of fibrosis markers and the presence of HBcrAg would be useful clinical markers for predicting the development of HCC during NA therapy (Table 2).

To examine the molecular mechanism underlying the presence of HBcrAg during NA therapy, we evaluated the HBV status in the liver (Supplementary Data and Figure 2). HBV DNA, pregenomic RNA, pre-S/S RNA, and HBx RNA levels were significantly higher in HBcrAg-positive livers, strongly suggesting that HBV actively replicates in HBcrAg-positive livers. Hepatic gene expression analysis showed that viral replication–associated signaling, such as that occurring during protein translation, ubiquitin-proteasome pathway signaling, oxidative phosphorylation, and JNK signaling, was upregulated together with cancer-related signaling in HBcrAg-positive livers. Interestingly, pathway analysis revealed that many transcription factors (LRH1, HNF4α, PPARα, CEBPα, and SP1) that could activate the HBV promoter [13] were upregulated in HBcrAg-positive livers (Figure 3). In vitro, HepAD38 cells overexpressing LRH1 (ie, HepAD38-LRH1 cells) supported active HBV replication, characterized by increased HBV DNA, pregenomic RNA, and cccDNA levels. ETV repressed HBV replication, but HepAD38-LRH1 cells still supported active HBV replication, in contrast to control HepAD38 cells, although long-term ETV treatment diminished the differences in cccDNA level between the 2 cell types. The mechanism of the different kinetics of cccDNA during ETV treatment could not be determined, but there might be different kinetics between cccDNA and HBV DNA in HBcrAg-positive livers during long-term ETV treatment. Conversely, we found that pre-C/C, as well as HBx, increased the expression of LRH1, PPARα, and HNF4α in HepG2 cells (Figure 4A). Although HBV core can regulate host gene expression [15], the regulation of these transcriptional factors by HBV core protein had not been reported previously. Recent report showed that HBV core protein enhanced activation of cyclic AMP response element (CRE) transcription via the CRE/CREB/CBP pathway [15]. Because CREB could bind the promoter lesion of LRH1, PPARα, and HNF4α [16, 17], it is possible that HBV core protein regulated these transcriptional factors. The results suggest the presence of positive feedback regulation between HBcrAg and HBV transcription-related factors that maintained active HBV replication even during NA treatment. Preserved HBV replication in HBcrAg-positive livers could contribute to an increased incidence of HCC.

These results prompted us to examine the effect of metformin, an AMPK activator, on HBV replication. Metformin successfully repressed the transcription of HBV RNA, the replication of HBV DNA, and HBV protein (ie, HBsAg and HBcrAg) synthesis in HepAD38 cells and PHHs (Figure 5A–D). Neither ETV nor metformin had a suppressive effect on cccDNA in PHHs (Figure 5D), probably because cell division was limited in PHHs. Metformin strongly repressed HBV protein synthesis, while ETV had little effect (Figure 5A and 5D). Importantly, the combination of metformin and ETV further inhibited HBV DNA replication (Figure 5B and 5D), suggesting that the addition of metformin to current NA therapy would be useful in further decreasing HBV replication in HBcrAg-positive patients. To date, one report showed that metformin inhibited HBV protein production and replication in HepG2 cells [18]. However, regulation of HBV-related transcriptional factors by metformin was not evaluated in the report.

In summary, we showed that HBcrAg positivity during NA therapy reflects HBV replication in the liver and could be predictive of the development of HCC. We also showed that HBV transcription was activated in HBcrAg-positive livers and that targeting HBV transcription factors could be a useful therapy to further repress HBV replication. As metformin reportedly has an antitumor effect on HCC and other cancers [19, 20], further clinical study could be performed to reduce the incidence of HCC development in patients with chronic hepatitis B undergoing NA therapy.

Notes

Acknowledgments. We thank Mina Nishiyama for technical assistance.

M. H. was responsible for the design of the study, interpretation of data, and drafting of the manuscript. T. S. was responsible for cellular experiments and the drafting of the manuscript. T. T. was responsible for the acquisition of clinical data. K. K. was responsible for the acquisition of clinical data. M. N. was responsible for the generation of gene expression data. N. O. was responsible for the acquisition of clinical data. T. S. was responsible for the acquisition of clinical data. H. O. was responsible for the generation of gene expression data. K. A. was responsible for the acquisition of clinical data. T. Y. was responsible for the acquisition of clinical data. Y. S. was responsible for the acquisition of clinical data. T. Y. was responsible for the acquisition of clinical data. E. M. was responsible for the acquisition of clinical data. Shuichi Kaneko was responsible for conceiving and designing the study.

Financial support. This work was supported by the JSPS Core-to-Core Program, B. Asia-Africa Science Platforms and Hepatitis Research Program, Japan Agency and Medical Research Development, AMED.

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References