IP10 and Anti-HBc can Predict Virological Relapse and HBsAg Loss in Chronic Hepatitis B Patients after Nucleos(t)ide Analog Discontinuation Open Access

Functional cure is defined as hepatitis B surface antigen (HBsAg) seroclearance or seroconversion, which has been considered as the ideal treatment endpoint for patients with chronic hepatitis B (CHB) [1]. However, intrahepatic covalently closed circular DNA could not be eliminated and functional cure could hardly be achieved by nucleos(t)ide analog (NA) treatment [2, 3]. In the pursuit of functional cure, the study of NA withdrawal is a new exploration. In some recent studies investigating the outcomes of NA discontinuation, a substantial amount of patients achieved sustained virological response, even few of them achieved HBsAg loss [2, 4]. Therefore, it is important to find predictors for patients who can remain virological response and patients who can get HBsAg loss after NA cessation.

In previous study, we have described virological relapse (VR) and HBsAg loss rates and evaluated the predictive value of hepatitis B core-related antigen (HBcrAg), hepatitis B virus (HBV), RNA and HBsAg after NA discontinuation [5]. Lower end of treatment (EOT) HBsAg level has the best predictive value for both VR and HBsAg loss after NA discontinuation, which was consistent with other studies [6‒8]. To further elucidate potential mechanism in off-treatment VR and HBsAg loss, we further analyzed serum interferon-inducible protein 10 (IP10) and hepatitis B core antibody (anti-HBc).

IP10 is a proinflammatory chemokine induced by interferon-γ. It may be a meaningful marker for liver inflammatory reaction and might be related to HBsAg and HBeAg seroconversion during NA treatment [9, 10]. Whereas, until now only one study has assessed IP10 in NA discontinuation setting. More investigation about this issue is warranted.

Anti-HBc is not a new HBV infection serum marker. Emerging data have revealed the anti-HBc level could predict either interferon or NA treatment response [11, 12]. Nevertheless, it is unclear whether anti-HBc can be used to monitor and predict VR and HBsAg loss after NA discontinuation. In view of previous studies showing a potential link between serum anti-HBc levels and immunity of host to the HBV, quantification of anti-HBc levels might have predictive ability for virological response after NA discontinuation because virus-specific immune is critical to off-treatment persistence [13]. In this prospective multicenter study, we investigated the dynamic changes of host immune status reflected by serological IP10 and anti-HBc after NA discontinuation to evaluate their predictive ability for off-treatment VR and HBsAg loss.

Overall, 139 non-cirrhosis CHB patients with initially hepatitis B e antigen (HBeAg) positive who discontinued NA were followed up in twelve hospitals in Beijing, Tianjin, and Hebei provinces in China from January 2017 to December 2020. The inclusion‐exclusion criteria and follow-up schedule have been outlined in our previous study [5]. Briefly, the NA cessation criteria in this study were referenced in the Chinese guideline of prevention and treatment for CHB [14], which was defined as normal alanine aminotransferase level, undetectable HBV DNA, and HBeAg seroconversion for at least 3 years and the overall NA treatment course more than 4 years. Included patients discontinued NA following up each 3 months for a total of 24 months or until clinical relapse (CR). Retreatment was started once CR occurred. VR was considered when HBV DNA was greater than 2,000 IU/mL. CR was considered when VR occurred and alanine aminotransferase was greater than 2 times of upper limit of normal. Consolidation treatment time was from the first time of HBeAg seroconversion to the time of NA discontinuation [5].

Methods of HBsAg, HBV DNA, and biochemical test have been described in our previous study [5]. Serum IP10 was detected by an enzyme-linked immunosorbent (ELISA) (Myhalic, Wuhan, China), following the operation instructions in the manual with an assay range from 5 to 100 pg/mL. When serum IP10 levels were tested exceeding the detection upper limit, samples would be retested by a 1:2 dilution and further a 1:4 dilution.

Serum anti-HBc was quantified by ELISA (Aoxing, Wuhan, China) with a range of 12.5–400 IU/L. When level of anti-HBc was above 400 IU/mL, the serum sample was retested by dilutions from 1:10 to 1:100,000.

Quantitative variables were expressed as numbers (percentages) or median (interquartile range). The differences between groups were analyzed by χ² test or Mann-Whitney U test. Generalized estimating equations model was conducted to compare the trajectory changes of seromarkers at serial time points between VR/non-VR as well as HBsAg loss/non-loss groups. Predictors for VR and HBsAg loss were assessed by Cox regression analysis. The area under receiver-operating characteristic (AUROC) curve was performed to assess the serum markers’ predictive ability of VR and HBsAg loss. Kaplan-Meier analyses were used and compared by the log-rank test to calculate the cumulative probabilities of VR and HBsAg loss. All statistical analysis was conducted by IBM SPSS software version 26.0. Graphical analyses and representation of data were conducted by GraphPad Prism 7.0 software.

All of 139 patients completed the study and none of the patients experienced hepatic decompensation, cirrhosis, HCC, or died during follow-up. The cumulative probabilities of VR, CR, and HBsAg loss at 12 months after NA cessation were 38.8%, 15.1%, and 4.3%, respectively, and the corresponding cumulative probabilities were 50.4%, 24.5%, and 9.4% at 24 months, respectively. Patients’ characteristics have been described in our previous study [5]. As shown in the present study, EOT IP10 and anti-HBc were 29.2 (5.1–66.4) pg/mL and 193.6 (136.9–221.4) IU/mL, respectively (Table 1).

As shown in Table 1, the median levels of IP10 at 6, 12, 18, and 24 months after withdrawal were 26.7 (6.2–64.6) pg/mL, 31.7 (7.8–79.1) pg/mL, and 33.5 (10.2–82.8) pg/mL, respectively. The median levels of anti-HBc at 6, 12, 18, and 24 months after withdrawal were 192.5 (129.2–227.5) IU/mL, 180.6 (121.2–220.3) IU/mL, and 170.1 (108.8–218.3) IU/mL, respectively.

Significant differences of IP10 and anti-HBc levels between patients with VR and non-VR are displayed in Figure 1a, b. Patients with VR had lower IP10 levels at EOT and 6, 12, 18, and 24 months after NA discontinuation, respectively (p < 0.001). Higher anti-HBc levels were found in patents with VR at EOT and 6, 12, 18, and 24 months after NA cessation, respectively (p < 0.001). Results of generalized estimating equations showed that IP10 value of patients without VR was 49.358 times that of patients with VR (p < 0.001). The anti-HBc value of VR patients was 72.116 times higher than that of patients without VR (p < 0.001) (data not shown).

Significant differences of IP10 and anti-HBc levels between patients with and without HBsAg loss were found. As shown in Figure 1c, d, patients with HBsAg loss had higher IP10 levels at EOT and 6, 12, 18, and 24 months after NA cessation, respectively (p < 0.001). Lower anti-HBc levels were found in patents with HBsAg loss at EOT and 6, 12, 18, and 24 months after NA withdrawal, respectively (p < 0.001). Results of generalized estimating equations found that IP10 value of patients with HBsAg loss was 73.26 times higher than that of patients without HBsAg loss (p < 0.001). The anti-HBc value of patients without HBsAg loss was 100.245 times higher than that of patients with HBsAg loss (p < 0.001) (data not shown).

EOT HBsAg ≥2 log IU/mL, HBcrAg ≥4 log U/mL, and positive HBV RNA were related to higher risk of VR, which have been reported in our previous study [5]. In the present study, Cox regression analyses were conducted to assess the predictability of IP10 and anti-HBc for VR risk. EOT IP10 (hazard ratio [HR], 0.961; 95% confidence interval [CI], 0.913–0.981; p < 0.001) and EOT anti-HBc (HR, 1.147; 95% CI, 1.002–1.248; p < 0.001) were associated with VR in multivariable analysis (Table 2).

The AUROC was calculated to further evaluate the predictive value of EOT IP10 and anti-HBc for VR. The AUROC value of the EOT IP10 was 0.736 (95% CI, 0.654–0.818; p < 0.001), which was higher than EOT anti-HBc (0.676 [0.587–0.765], p < 0.001) (Fig. 2a, b). EOT IP10 of 26.99 pg/mL had the maximized Youden’s index with the sensitivity of 67.1% and specificity of 71%. EOT anti-HBc of 141.35 IU/mL had the maximized Youden’s index with the sensitivity of 85.7% and specificity of 43.5%.

Age ≥40 years and EOT HBsAg <2 log₁₀ IU/mL were independent predictors for HBsAg loss, which have been reported in our previous study [5]. The predictability of IP10 and anti-HBc for HBsAg loss was assessed by Cox regression in the present study. EOT IP10 (HR, 1.192; 95% CI, 1.001–1.34; p < 0.001) and EOT anti-HBc (HR, 0.975; 95% CI, 0.932–0.982; p < 0.001) were associated with HBsAg loss in multivariable analysis (Table 2).

The AUROC value of each variable was assessed to further evaluate the ability of EOT IP10 and anti-HBc to predict HBsAg loss. The AUROC value of the EOT anti-HBc was 0.828 (95% CI, 0.699–0.958; p < 0.001), which was higher than EOT IP10 (0.762 [0.606–0.917], p = 0.002) (Fig. 2c, d). EOT IP10 of 93.5 pg/mL had the maximized Youden’s index with the sensitivity of 61.5% and specificity of 92.1%. EOT anti-HBc of 78.42 IU/mL had the maximized Youden’s index with the sensitivity of 61.5% and specificity of 92.9%.

Cumulative probabilities of VR were stratified by EOT IP10, EOT anti-HBc, or combined two parameters. Until 24 months after NA discontinuation, cumulative rates of VR in patients with EOT IP10 > 26.99 pg/mL and ≤26.99 pg/mL were 31.9% and 70.1%, respectively (HR 2.998, 95% CI 1.86–4.831, p < 0.001, Fig. 3a). Cumulative incidences of VR stratified by EOT anti-HBc ≤141.35 IU/mL and >141.35 IU/mL were 49.1% and 60.6%, respectively (HR 2.99, 95% CI 1.818–4.916, p < 0.001, Fig. 3b). When combined both of EOT IP10 and anti-HBc, cumulative probability of VR was 16.7% in patients with EOT IP10 > 26.99 pg/mL plus anti-HBc ≤141.35 IU/mL, which was significantly lower than 73.6% (HR 6.464, 95% CI: 3.631–11.51, p < 0.001, Fig. 3c).

Cumulative probabilities of HBsAg loss stratified by EOT IP10 and EOT anti-HBc are shown in Figure 3d, e. At 24 months after withdrawal, cumulative probabilities of HBsAg loss in patients with EOT IP10 > 93.5 pg/mL and ≤93.5 pg/mL were 46.2% and 4.7%, respectively (HR 10.94, 95% CI: 2.145–55.82, p < 0.001, Fig. 3d). Cumulative rates of HBsAg loss stratified by EOT anti-HBc ≤78.42 IU/mL and >78.42 IU/mL were 47.1% and 5%, respectively (HR 12.27, 95% CI 2.237–67.23, p < 0.001, Fig. 3e). Patients with EOT IP10 > 93.5 pg/mL plus anti-HBc ≤78.42 IU/mL had the highest 24-month cumulative HBsAg loss rate (53.8% vs. 4%, HR 16.83, 95% CI: 2.225–127.3, p < 0.001, Fig. 3f).

This study assessed the changes of serum IP10 and anti-HBc levels after NA discontinuation in HBeAg-positive non-cirrhosis CHB patients. Results showed that IP10 and anti-HBc were useful to distinguish patients with a lower risk of VR or higher chance to be HBsAg loss after NA cessation. Patients with EOT IP10 >26.99 pg/mL or EOT anti-HBc ≤141.35 IU/mL had lower risk of off-treatment VR. EOT IP10 >93.5 pg/mL or anti-HBc ≤78.42 IU/mL had good predictability for off-treatment HBsAg loss. Our results suggested that EOT IP10 and EOT anti-HBc levels could be used to guide in the decision of NA cessation.

IP10 can play potent biological effects in HBV by activating T-lymphocytes and natural killer cells [15‒17]. In previous studies, higher levels of serum IP10 which may reflect an activated immune system were related to favorable responses in CHB patients [18, 19]. In Papatheodoridi’s study [20], IP10 levels at NA discontinuation and month-1 were associated with the probability of off-treatment CR but not associated with VR. Whereas, significant difference of IP10 levels between patients with and without VR were found in our study (Fig. 1a). Lower levels of IP10 at EOT and 6 months, 12 months, and 24 months after withdrawal were independent predictors for VR by Cox regression analysis. We also found patients with EOT IP10 ≤26.99 pg/mL had significant higher cumulative rates of VR at 24 months after NA cessation (70.1% VS 31.9%, HR 2.998, p < 0.001, Fig. 3a). Correlation analysis of the levels of IP10 with off-treatment VR in our study indicated that the serum levels of IP10 reflect the immune system change and can serve as indicators suggesting the safety of NA cessation.

In some studies, IP10 was shown as a good predictor for HBsAg reduce in the period of NA treatment [18, 19, 21]. In Papatheodoridis’s study [18], the median levels of IP10 began increasing remarkably from the third year of treatment, which may influence HBsAg levels by extending treatment time over the fourth year. The results might be in line with the hypothesis that immune function may be resumed by extending effective NA treatment [22]. In a recent study [20], levels of serum IP10 at the first month after cessation of NA was found to be useful in prediction for functional cure in patients with baseline HBsAg levels of 100–1,000 IU/mL. In our study, patients with EOT IP10 >93.5 pg/mL had significant higher cumulative probabilities of HBsAg loss at 24 months after withdrawal (46.2% vs. 4.7%, HR 10.94, p < 0.001, Fig. 3d). Recently, alterations in the T‐cells phenotype were found to enhance cellular immunity and reduce T‐cell exhaustion, contributing to HBsAg loss [23, 24]. The higher IP10 level in our study might reflect an activated IFN system, which may be beneficial for HBsAg clearance. Therefore, detection of serum IP10 levels before NA treatment cessation may have clinical implications. Additional evaluation of the value of serum IP10 levels in CHB patients who achieved HBsAg loss after NA cessation will be interesting.

It is learned that humoral immunity has an important role on clearing HBV infection. Anti-HBc is generated by B lymphocytes which is specific for HBcrAg. Results of previous studies showed that patients with better response to PegIFN or NA therapy (e.g., HBeAg seroconversion or HBsAg loss) had higher anti-HBc levels [11, 12, 25]. In few studies, higher EOT anti-HBc levels were associated with lower risk of relapse and higher probabilities of functional cure after NA discontinuation [26, 27]. On the other hand, recent studies found that higher level of anti-HBc was related to more inflammatory activity and advanced fibrosis in histology [28, 29]. Anti-HBc level decreased during NA treatment [11, 12, 25]. A remarkable increase of anti-HBc level was shown in CR patients as compared to non-CR patients after treatment cessation [26]. In our study, significant difference of anti-HBc levels was found in patients with VR or HBsAg loss (Fig. 1b, d). Patients with EOT anti-HBc ≤78.42 IU/mL had significant higher cumulative probabilities of HBsAg loss at 24 months after withdrawal (47.1% VS 5%, HR 12.27, p < 0.001, Fig. 3e). Although anti-HBc has been widely used in clinical practice for more than 3 decades, few studies have focused on the role of anti-HBc level in CHB patients who stopped NA treatment, and it is not clear whether EOT anti-HBc level is predictive for VR or HBsAg loss after NA cessation. Based on previous research which found the potential association between serum levels of anti-HBc and host immunity against HBV, we hypothesize that measurement of anti-HBc may be used in guiding NA cessation, in so much as virus-specific immune activity holds the key for sustained viral remission.

The accurate mechanism of off-treatment virological response and HBsAg loss is still not known. Cytokines may inhibit HBV replication directly and determine the primary host immune response pattern indirectly. IP10 induces the chemotaxis of CXC3 receptor-expressing T1 cells to enhance the T1 response and increase the IFN-γ secretion, all of which eventually induces and maintains the chronic inflammation reaction [30]. Results of one study [31] found that B lymphocytes also had critical impact on regulating immune response against HBV. B lymphocytes not only generate neutralizing antibodies against HBV but also generate some cytokines that can suppress virus replication and regulate the activity of CD4 and CD8 T cells responses. Serum anti-HBc was produced by HBcrAg-specific B lymphocytes, which might reflect the adaptive immune response to viral control.

Here are some strengths in our study. First, as a prospective multicenter study, we collected comprehensive data making the results to be more credible. Second, there are few studies investigating value of IP10 and anti-HBc in NA discontinuation. By contrast, we not only supplied EOT IP10 and anti-HBc levels but also collected serial results at every 3 months after NA discontinuation, providing the consecutive data for analysis. Third, our study adopting the Chinese NA discontinuation criteria in which longer NA treatment duration and consolidation time was restricted, so as to make more data from different NA cessation criteria. Nevertheless, this study also has some limitations. First, as only initially HBeAg-positive patients were involved, whether the results can be applied to initially HBeAg-negative patients is not clear. Second, only Asian patients were included in our study, so similar results cannot be simply presumed for Western CHB patients.

In conclusion, EOT IP10 and EOT anti-HBc levels could predict VR and HBsAg loss after NA cessation. High EOT IP10 or low EOT anti-HBc levels were related to both lower risk of VR and higher possibility of HBsAg loss. EOT IP10 and EOT anti-HBc alone or in combination could be used to select suitable patients who can discontinue NA.

We are grateful to Dr. Xiajie Wen for her help in statistical analysis methods.

The study was approved by the Institutional Review Board of Peking University People’s Hospital (2017PHB001‐01). Written informed consent was obtained from all patients.

The authors have no conflicts of interest to declare.

The work was supported by the Beijing Municipal Science and Technology Commission of Major Projects (No. D161100002716002), National Major Science and Technology Project of China (No. 2017ZX10302201-004-001, No. 2017ZX10203202-003), National Natural Science Foundation of China (No. 82000557), and the Peking University People’s Hospital Research and Development Funds (RDY2020-12).

Conceptualization: B.F. and L.W. Methodology: X.Y.D., M.H.L., and S.J.Z. Formal analysis: Y.D.X. and J.Liu. Investigation: B.F., M.H.L., Y.J.G., and Y.M.N. Resources: Y.D.X., M.H.L., X.J.O., S.J.Z., Y.J.G., X.Y.X., Y.Y., A.L.M., J.Li, Y.M.N., H.W.Z., L.W., and B.F. Writing (original draft preparation): Y.D.X. Writing (review and editing): L.W. and B.F. Funding acquisition: Y.D.X. and B.F.

Data are not publicly available due to ethical reasons. Further inquiries can be directed to the corresponding author.