Abstract
Background and Aims
Humoral immunity is thought to play a key role in the immunopathogenesis of chronic hepatitis B (CHB). Peaks in anti-HBs immune complexes (HBsAg-ICs) have been associated with subsequent HBsAg seroclearance. Discontinuation of nucleos(t)ide analog (NA) therapy can promote HBsAg seroclearance but also carries a risk of hepatitis flare. We evaluated the relationship between serum HBsAg-IC levels and hepatitis flare following NA cessation in HBeAg-negative CHB.
Methods
In a multicentre prospective cohort of noncirrhotic HBeAg-negative CHB patients undergoing NA discontinuation, we conducted a nested case-control analysis (n = 38). Stored serum samples were tested for HBsAg-IC at end-of-treatment (EOT) and at serial timepoints off-treatment. Cases (n = 24) experienced a hepatitis flare (ALT >5× upper limit of normal); controls (n = 14) did not. Associations between HBsAg-IC levels and clinical outcomes were analyzed.
Results
Serum HBsAg-ICs were detectable in 97% of participants at EOT. High HBsAg-IC levels (≥40 IU/mL) at EOT were less frequent among those who developed hepatitis flare (8%) versus controls (57%, P = .002). Individuals treated with tenofovir (vs other NAs) had lower EOT HBsAg-IC levels. Longitudinal analysis revealed dynamic increases in HBsAg-IC levels during hepatitis flare, peaking with maximal ALT. Participants who experienced beneficial flares (≥1 log₁₀ HBsAg reduction) had higher peak off-treatment HBsAg-IC levels compared to those with nonbeneficial flares.
Conclusions
Serum HBsAg-ICs are detectable in patients with CHB, and low levels at EOT are associated with the risk of hepatitis flare after NA cessation. These findings support the role of humoral immunity in CHB immunopathogenesis.
Clinical Trials Registration
The study clinical trial ID is NCT02581033.
Keywords: chronic hepatitis B, immune complexes, nucleos(t)ide analog withdrawal, hepatitis flare, biomarker
Among HBeAg-negative participants with chronic hepatitis B who discontinued nucleos(t)ide analog therapy, lower levels of anti-HBs immune complexes (HBsAg-ICs) at end-of-treatment were associated with an increased risk of hepatitis flare. Increases in HBsAg-IC levels off-treatment occurred during hepatitis flares.
Hepatitis B virus (HBV) infection remains a significant global health challenge [1]. We remain far from achieving the World Health Organization elimination targets by 2030 [1].
Although nucleos(t)ide analog (NA) therapy is safe and effective in achieving viral suppression, functional cure (HBsAg loss) is rare, necessitating lifelong treatment for most patients [2–5]. Intentional discontinuation of NAs remains an effective strategy for achieving functional cure among carefully selected patients [2, 3, 6]. However, this approach carries a risk of severe hepatitis flares and, rarely, hepatic decompensation [6–9]. Although early data suggest that hepatitis B core-related antigen (HBcrAg) and HBV RNA may help predict NA-withdrawal flares, additional biomarkers are needed [10–13].
Hepatitis flares can occur spontaneously or in response to antiviral treatment withdrawal or immunosuppressive therapy. The underlying mechanisms of hepatitis flares remain incompletely understood, but accumulating evidence suggests that humoral immunity may play a central role [14–17]. While flares are generally undesirable due to the risk of decompensation, a subset may confer benefit. However, distinguishing beneficial from nonbeneficial flares remains a significant clinical challenge [17].
Humoral immunity is important in the pathogenesis of CHB, where anti-HBs induction by vaccination is protective against infection [18, 19]. In contrast, B-cell depleting therapies carry the risk of severe hepatitis B reactivation [20–22]. HBsAg-specific antibody responses are difficult to detect during chronic hepatitis B infection (CHB) and are often overlooked. The saturation of anti-HBs by the formation of HBsAg/anti-HBs immune complexes (HBsAg-ICs) hampers accurate measurement [23]. However, historical data have indicated that the presence and levels of circulating hepatitis B immune complexes may correlate with disease activity [24–26]. A deeper understanding of HBsAg-ICs could provide insights into HBV pathogenesis and guide flare risk stratification, as well as potentially inform the development of therapeutic vaccines.
To our knowledge, the role of HBsAg-IC as a predictive biomarker of outcomes following NA discontinuation has not been evaluated. We have recently optimized an immune complex assay for quantifying anti-HBs in the presence of HBsAg to characterize and quantify anti-HBV humoral immune responses [27]. Among an HBeAg-positive cohort infected with genotype A HBV, we observed that on-treatment spikes in HBsAg-IC were associated with subsequent functional cure after starting NA therapy and therefore may be a biomarker for a more effective B cell response against HBV [27]. Therefore, the clinical significance of HBsAg-IC in the pathophysiology and prediction of hepatitis flares is of considerable interest.
We aimed to evaluate end-of-treatment (EOT) HBsAg-IC levels as a predictive biomarker of outcomes following NA discontinuation in a well-characterized subset of a prospectively recruited cohort of individuals with noncirrhotic HBeAg-negative CHB [6]. We utilized this cohort as a model to evaluate the pathogenesis of flares in a prospective, longitudinal fashion. We also evaluated off-treatment HBsAg-IC kinetics among individuals who did and did not experience a hepatitis flare. We hypothesized that hepatitis flare would be associated with a concomitant peak in HBsAg-IC, reflective of the humoral immune response. By investigating the relationship between HBsAg-ICs and hepatitis flares, we hope to contribute to the development of more precise predictive models and therapeutic approaches for managing CHB.
METHODS
Patients and Study Design
A nested case-control study was conducted to evaluate the association between EOT (baseline) and off-treatment HBsAg-IC levels and clinical outcomes among patients with CHB enrolled in a prospective study of NA discontinuation. Testing was performed on stored samples from individuals enrolled in a prospective multicenter cohort study, which evaluated outcomes until 96 weeks following NA cessation among HBeAg-negative adults with noncirrhotic chronic HBV monoinfection [13]. HBV genotype was determined using the methodology described by Thurnheer and colleagues [28], and retreatment criteria were as previously described [6].
Serum samples (stored at −80°C) from 38 patients were available for analysis. Participants were divided into cases [those who experienced hepatitis flare (peak ALT >5× upper limit of normal (ULN) by 96 weeks of follow-up)] and controls [those who did not experience any ALT flare (peak ALT <2× ULN by 96 weeks of follow-up)]. A prespecified sub-analysis was conducted among participants with EOT HBsAg ≥10 IU/mL based on our prior observation that those with EOT HBsAg <10 IU/mL tended to have low rates of virological reactivation and flare [6].
Specimens were available at EOT (the day that NA treatment was stopped) in all individuals and at longitudinal off-treatment timepoints. Among the hepatitis flare group, samples were available for testing at 5 off-treatment timepoints: (1) after virological relapse and before hepatitis flare (“pre flare”), (2) after hepatitis flare and before peak ALT (“early flare”), (3) maximum ALT (“peak flare”), (4) after peak flare and before ALT normalization (“post-peak flare”), and (5) after ALT normalization (“post flare”). For the control group, samples were available at 3 off-treatment timepoints: (1) following virological reactivation (“first detectable DNA”), (2) maximum HBV DNA (“peak DNA”), and (3) end of follow-up (“EOFU”).
Laboratory Testing for HBsAg-IC and Other HBV Markers
Serum HBsAg-IC and HBsAg levels were quantified using the Roche Elecsys® system. HBsAg-IC was initially isolated using an established polyethylene glycol precipitation method, which has been described in detail previously [24, 25]. We have previously validated the assay's sensitivity, specificity, and reproducibility [27]. HBsAg-IC was also expressed as a proportion of the total HBsAg level (HBsAg-IC/total HBsAg) at each timepoint.
Clinical Outcomes of Interest
We examined associations between EOT and off-treatment HBsAg-IC levels and clinical outcomes. In addition to hepatitis flare, the association between EOT HBsAg-IC level and rate of virological reactivation (detectable HBV DNA above the lower limit of quantification after stopping NA therapy), virological relapse (HBV DNA >2000 IU/mL), and HBsAg loss (undetectable quantitative HBsAg as well as negative qualitative HBsAg serology) was tested. Among the hepatitis flare group, EOT and off-treatment HBsAg-IC levels were compared between those who did and did not experience a beneficial flare, defined as a ≥1 log10 reduction in HBsAg level from the time of peak ALT to week 96 (end of follow-up). The ULN was set according to the local laboratory as previously described [6].
Statistical Analysis
Descriptive statistics were reported as median (IQR) for continuous variables and counts (percentages) for categorical variables. Comparison between cases and controls was performed using the Mann–Whitney U test for continuous variables and Fisher's exact for categorical variables. Correlation among biomarkers at each timepoint was assessed using Spearman's rank correlation coefficient. In the absence of established thresholds, a “high” HBsAg-IC level at EOT was defined as a value above the 75th percentile.
Off-treatment changes in HBsAg-IC levels were assessed by calculating the median (IQR) per-participant relative difference between each off-treatment timepoint and EOT [(log10 HBsAg-IC at timepoint − log10 HBsAg-IC at EOT)/log10 HBsAg-IC at EOT]. P-values of <.05 by a two-tailed test were considered statistically significant. Statistical analysis was performed using Stata software (version 17).
RESULTS
Participant Characteristics
Most participants were of Asian ethnicity with either genotype B (55%) or genotype C (21%) HBV (Table 1). Prior to discontinuation, the majority of patients were treated with entecavir (61%). The median HBsAg level was 637 IU/mL (193–1819) at EOT. There were no differences in baseline characteristics among those who experienced a hepatitis flare (cases; ALT >5× ULN, n = 24) versus those who did not (controls, n = 14) (Table 1). All patients experienced virological reactivation, and 33 (87%) developed virological relapse.
Table 1.
Baseline Characteristics and Outcomes
| Overall Cohort (n = 38) |
No Flare (n = 14) |
Hepatitis Flare (n = 24) |
P-Valuea | |
|---|---|---|---|---|
| Baseline characteristics | ||||
| Male | 23 (61%) | 7 (50%) | 16 (67%) | .499 |
| Age, median (IQR) y | 60 (51–62) | 60 (55–62) | 58 (51–63) | .579 |
| Treatment | ||||
| Entecavir | 23 (61%) | 9 (64%) | 14 (58%) | .535 |
| Tenofovir | 10 (26%) | 3 (21%) | 7 (29%) | … |
| Lamivudine | 2 (5%) | 0 (0%) | 2 (8%) | … |
| Adefovir/lamivudine | 3 (8%) | 2 (14%) | 1 (4%) | … |
| ALT, median (IQR) U/L | 27 (19–34) | 21 (19–30) | 28 (20–36) | .224 |
| Ethnicity | ||||
| Asian | 31 (82%) | 12 (86%) | 19 (79%) | .188 |
| Caucasian | 4 (11%) | 0 (0%) | 4 (17%) | … |
| Other | 3 (8%) | 2 (14%) | 1 (4%) | … |
| Genotype | ||||
| A | 2 (5%) | 1 (7%) | 1 (4%) | .292 |
| B | 21 (55%) | 6 (43%) | 15 (63%) | … |
| C | 8 (21%) | 5 (36%) | 3 (13%) | … |
| D | 5 (13%) | 1 (7%) | 4 (17%) | … |
| E | 1 (3%) | 0 (0%) | 1 (4%) | … |
| Not available | 1 (3%) | 1 (7%) | 0 (0%) | … |
| Liver stiffness measurement, median (IQR), kPa | 4.6 (3.5–5.4) | 4.6 (3.3–5.3) | 4.65 (3.65–5.4) | .785 |
| HBsAg level IU/mL | ||||
| Median (IQR) | 637 (193–1819) | 590 (8–11310) | 685 (251–1777) | .762 |
| <10 | 5 (13%) | 4 (29%) | 1 (4%) | .169 |
| 10–100 | 2 (5%) | 0 (0%) | 2 (8%) | … |
| 100–1000 | 14 (37%) | 4 (29%) | 10 (42%) | … |
| ≥1000 | 17 (45%) | 6 (43%) | 11 (46%) | … |
| Off-treatment outcomes | ||||
| Virological reactivation | 38 (100%) | 14 (100%) | 24 (100%) | n/a |
| Virological relapse | 33 (87%) | 9 (64%) | 24 (100%) | .004b |
| Maximum ALT (U/L), median (IQR) | 323 (61–924) | 49 (34–61) | 642 (378–1608) | <.001b |
| Maximum HBV DNA (log10 IU/mL), median (IQR) | 6.30 (4.73–7.65) | 4.13 (2.86–5.44) | 7.44 (6.30–8.32) | <.001b |
| HBsAg loss | 5 (13%) | 5 (35%) | 0 (0%) | .004b |
| NA recommencement | 21 (55%) | 2 (14%) | 19 (79%) | <.001b |
aComparison between cases (hepatitis flare; n = 24) and controls (no hepatitis flare; n = 14). Statistical test: Fisher's exact for categorical variables and for continuous variables.
b P < .05.
HBsAg-IC Level at EOT in the Overall Cohort
In the overall cohort, at EOT, HBsAg-IC was detectable in 37/38 (97%) participants, and the median HBsAg-IC level was 9.77 IU/mL (4.13–40.17). The 75th percentile for HBsAg-IC level at EOT was 40 IU/mL and was used as the threshold for a “high” HBsAg-IC level. EOT HBsAg-IC and HBsAg levels were positively correlated (r = 0.573, P = .002; Supplementary Figure 1). Participants with EOT HBsAg ≤10 IU/mL (n = 5) had a very low median EOT HBsAg-IC level (0.40 IU/mL, 0.29–0.96).
Participants Who Experienced Hepatitis Flare After Stopping NA Had Lower Levels of HBsAg-IC at EOT
To examine the relationship between EOT HBsAg-IC level and flare, we compared the EOT HBsAg-IC level and HBsAg-IC/HBsAg (%) between the flare cohort and controls. The median HBsAg-IC level (IU/mL) at EOT was numerically lower among those who experienced hepatitis flare compared with controls [8.20 (4.07–19.50) vs 43.49 (5.14–56.81); P = .080] (Figure 1A). Among participants with EOT HBsAg level ≥ 10 IU/mL, there was a statistically significant difference in median EOT HBsAg-IC level (IU/mL) between the hepatitis flare group and controls [8.68 (4.13–39.02) vs 47.27 (19.86–56.81), P = .004]. Participants in the flare cohort also had lower HBsAg-IC/HBsAg ratios (%) [1.88 (0.54–3.32) vs 4.18 (1.71–5.61), P = .046].
Figure 1.
Comparison of (A) EOT HBsAg-IC level and (B) proportion of participants with high EOT HBsAg-IC (≥40 IU/mL), among the flare cohort versus controls.
We found that participants who experienced hepatitis flare were less likely to have an EOT HBsAg-IC ≥ 40 IU/mL compared with those who did not [2/24 (8%) vs 8/14 (57%), P = .002] (Figure 1B). Those who experienced a hepatitis flare were also less likely to have an EOT HBsAg-IC/HBsAg (%) ≥ 4.5% (75th percentile) [3/24 (13%) vs 7/14 (50%), P = .021].
In contrast to HBsAg-IC levels, median EOT HBsAg levels (IU/mL) were similar among those who experienced hepatitis flare compared with controls [590 (8–11310) vs 685 (251–1777), P = .762]. An EOT HBsAg (IU/mL) threshold of <10 versus 10–100 versus 100–1000 versus ≥1000 was also tested, and there was no difference between the flare group and controls (Table 1). Additional EOT HBsAg (IU/mL) thresholds were also examined, including <100 versus ≥100 and <1000 versus ≥1000, and there was no difference between the flare group and controls (data not shown).
Associations Between EOT HBsAg-IC Level, NA Therapy, and Other Baseline Markers
Testing for associations between EOT HBsAg-IC level and baseline characteristics was evaluated among the overall cohort due to the small sample size. Tenofovir (vs other) treatment was associated with a lower median HBsAg-IC level at EOT [4.85 (4.13–8.68) vs 18.43 (4.40–50.89), P = .047]. No participant with a high EOT HBsAg-IC level (≥40 IU/mL) was treated with tenofovir [0/10 (0%) vs 18/28 (35.71%), P = .038]. While tenofovir treatment was associated with a lower EOT HBsAg-IC level, tenofovir (vs other) treatment was not independently associated with an increased risk of hepatitis flare [7/24 (29%) vs 3/14 (21%), P = .715]. Older participants (age ≥ 60 vs <60 years at EOT) had a lower median EOT HBsAg-IC (IU/mL) [6.16 (1.53–19.96) vs 22 (5.15–69.95), P = .028]. No other baseline characteristic, including genotype and ethnicity, was associated with EOT HBsAg-IC level (Supplementary Table 1).
Off-treatment HBsAg-IC Levels Increased During Hepatitis Flare
Off-treatment HBsAg-IC dynamics among the flare cohort were analyzed at 5 predetermined timepoints: “pre flare”, “early flare”, “peak flare”, “post-peak flare”, and “post flare” (defined in the methods). The median ALT and HBV DNA at the time of peak flare were 641 U/L (271–1443) and 6.04 log10 IU/mL (5.40–7.36) (Supplementary Figure 2 and Table 2).
Per-participant changes in biomarkers (log10 scale) were determined by calculating the relative change in biomarker at each off-treatment timepoint compared with EOT (baseline) (Figure 2A and Supplementary Table 3). There were no significant changes in HBsAg-IC compared with EOT at the pre-flare timepoint despite the occurrence of virological relapse [median HBV DNA 3.3 log IU/mL (2.06–4.88)]. However, significant relative increases in HBsAg-IC occurred during ALT flare. The median relative increase (%) in HBsAg-IC at early flare, peak flare and post-peak flare were 57.34% (21.78–127.11), 108.98% (64.17–160.62), and 64.07% (32.23–112.53), respectively. A moderate positive correlation was observed between peak off-treatment ALT (log₁₀ U/L) and peak off-treatment HBsAg-IC levels (log₁₀ IU/mL) (r = 0.449, P = .005) (Figure 3). Following resolution of ALT flare (“post flare” timepoint), HBsAg-IC levels were similar to EOT levels; the median relative change was 1.76% (−49.42 to 36.35).
Figure 2.
Median relative change (%) in HBsAg-IC and HBsAg level off-treatment compared with EOT among (A) flare cohort and (B) no-flare cohort. P-values reflect difference in median (IQR) relative change (%) from EOT in HBsAg-IC (log10 IU/mL) versus HBsAg (log10 IU/mL), Wilcoxon signed-rank test. Among the flare cohort (A), there were no significant changes in HBsAg-IC compared with EOT at the pre-flare timepoint despite the occurrence of virological relapse. Significant relative increases in HBsAg-IC occurred during ALT flare, peaking at time of maximal ALT. Following resolution of ALT flare, HBsAg-IC levels were similar to EOT levels. In contrast, there were minimal changes in HBsAg level during flare. Among the control group who did not experience hepatitis flare (B), there were no significant changes in HBsAg-IC or HBsAg level off-treatment.
Figure 3.
Correlation between off-treatment peak ALT and off-treatment peak HBsAg-IC within flare cohort.
Relative changes in off-treatment HBsAg level compared with EOT were similar to HBsAg-IC but less pronounced (Figure 2A). There was a significantly greater relative increase in HBsAg-IC level compared with HBsAg level at early flare (P = .002), peak flare (P < .001), and post-peak flare (P < .001) (Figure 2A and Supplementary Table 3). Per-participant off-treatment absolute change (%) in HBsAg-IC (log10 IU/mL) and HBsAg (log10 IU/mL) followed the same trend as relative changes (Supplementary Table 4). HBsAg-IC (log10 IU/mL) demonstrated moderate correlation with HBsAg (log10 IU/mL) at most timepoints (Supplementary Table 5).
Among the flare cohort (n = 24), 79% of flares resolved after re-treatment with NA therapy, and 21% resolved spontaneously. Retreatment did not influence HBsAg-IC dynamics at any off-treatment timepoint (Supplementary Figure 3). The relative changes (%) in HBsAg-IC level (log10 IU/mL) following hepatitis flare (“post flare” timepoint) were similar among those who were re-treated versus those who were not [−13.24 (−57.15 to 36.35) vs 18.69 (−20/31 to 20.50), P = .270].
No Significant Off-treatment Changes in HBsAg-IC Levels Were Observed in Patients Who Did Not Experience Hepatitis Flare
Analysis of off-treatment HBsAg-IC dynamics among the no-flare cohort was analyzed at 3 predetermined timepoints: “first DNA detectable”, “peak DNA”, and “end of follow-up” (defined in methods). All no-flare participants experienced virological reactivation, and 9 (64%) experienced virological relapse. The median peak HBV DNA (log10 IU/mL) was 2.64 (1.61–3.55). Compared with EOT, there were no significant off-treatment relative changes in median HBsAg-IC (log10 IU/mL) or HBsAg (log10 IU/mL) at DNA first detectable, peak DNA, or EOFU (Figure 2B).
Beneficial Hepatitis Flare Was Associated With Higher EOT HBsAg-IC Levels and Higher Off-treatment HBsAg-IC Peaks
9/24 (38%) experienced a beneficial flare, defined as ≥1 log10 IU/mL reduction in HBsAg level from time of peak ALT to week 96, Supplementary Figure 4. These individuals had higher EOT HBsAg-IC levels [15.14 (7.71–29.17) vs 4.3 (1.78–10.96), P = .030] and higher off-treatment peaks in HBsAg-IC compared with those who experienced nonbeneficial flares [458.79 (312.60–984.02) vs 40.50 (6.11–54.00), P = .002] (Figure 4).
Figure 4.
EOT and peak off-treatment HBsAg-IC levels according to flare outcome: beneficial* versus nonbeneficial. *≥1 log10 decline in HBsAg level from time of peak ALT to end of follow-up (96 wks).
None of the participants in the flare group and a subset of the control group (n = 5) (35%) achieved HBsAg loss. The median EOT HBsAg (IU/mL) level among those who achieved HBsAg loss was 3.1 (2–8). Among the control cohort, those who experienced HBsAg loss had a lower EOT HBsAg-IC level compared with those who did not [.40 (.29–5.14) vs 47.72 (40.17–56.81), P = .039]. HBsAg-IC/HBsAg (%) was similar among those who experienced HBsAg loss compared with those who did not [4.98 (2.66–14.64) vs 3.85 (1.71–5.13), P = .463].
Due to low numbers, limited information was available regarding off-treatment dynamics for the 5 patients who achieved HBsAg loss (n = 5). None of the participants who achieved HBsAg loss experienced an off-treatment increase in HBsAg-IC level prior to HBsAg loss (Supplementary Figure 5). Two off-treatment timepoints were available prior to HBsAg loss. The median relative reduction in HBsAg-IC from EOT at the first off-treatment timepoint was 58.60% (−100 to −17.19), and all participants had negative HBsAg-IC at the second timepoint (timepoint prior to HBsAg loss) (Supplementary Figure 5).
DISCUSSION
We investigated the clinical significance of HBsAg-IC levels at EOT and during off-treatment follow-up among a predominantly Asian cohort of noncirrhotic patients with HBeAg-negative CHB who discontinued NA therapy. HBsAg-ICs appeared to contribute to the immunopathogenesis of hepatitis flares. Longitudinal analysis of HBsAg-IC levels revealed a dynamic increase in HBsAg-IC during hepatitis flares, peaking at the time of maximum ALT elevation and returning to baseline upon flare resolution. This pattern was not observed among individuals who did not experience a hepatitis flare, which may suggest that the rise in HBsAg-IC was not simply a consequence of viral replication or antigenemia, but instead reflect immune-mediated hepatocyte injury. Interestingly, HBsAg-IC increases preceded or coincided with peak ALT but did not rise significantly during early virological relapse. This temporal association strengthens the hypothesis that humoral immune responses—including formation of circulating antigen–antibody complexes—may contribute to the immunopathogenesis of hepatitis flares.
Predicting risk of hepatitis flare following NA cessation is clinically important due to the potential for hepatic decompensation. Our data show that EOT HBsAg-IC level may be a useful biomarker for identifying candidates in whom it is safer to offer NA discontinuation. We observed that patients who experienced NA-withdrawal flares had lower HBsAg-IC levels at EOT. Similarly, the HBsAg-IC ratio (HBsAg-IC/HBsAg %) was also inversely associated with hepatitis flare, suggesting that absolute HBsAg-IC levels are not simply a surrogate marker for HBsAg. In contrast, EOT HBsAg levels alone were not associated with risk of flare. This is consistent with the findings of the primary study (n = 110) as well as other large cohorts, which have shown that EOT HBsAg thresholds above 10–50 IU/mL have limited utility in predicting flare and hepatic decompensation [6, 29, 30].
Comparable observations have been made with other viral biomarkers [12]. Higher EOT HBcrAg levels have been associated with an increased risk of hepatitis flare, reflecting ongoing intrahepatic viral transcriptional activity despite NA suppression [10, 11, 31]. Similarly, detectable serum HBV RNA at EOT has been shown to predict virological relapse and flare following NA discontinuation [13]. Together, these findings suggest that HBsAg-IC levels provide complementary information to existing viral biomarkers. These markers are likely to be particularly informative in patients with intermediate EOT HBsAg levels, where the likelihood of HBsAg loss is lower and accurate flare prediction is most clinically relevant.
The association between higher levels of HBsAg-IC at EOT (on NA therapy) and a reduced risk of NA-withdrawal flare is a novel finding. We previously observed among the original study cohort that the magnitude of virological relapse off-treatment predicted subsequent hepatitis flare [6]. Therefore, we hypothesize that higher levels of circulating HBsAg-ICs during NA therapy reflect greater immune pressure on the virus, resulting in controlled viral rebound and reduced hepatocyte injury during immune responses. While hepatitis flares are generally undesirable, a subset of flares may convey beneficial immune activation. In our study, hepatitis flares were more likely to be beneficial (≥1 log10 HBsAg level decline) in individuals with higher EOT HBsAg-IC levels. This aligns with prior evidence indicating that humoral immune responses, including anti-HBs production, play a role in modulating HBV activity and promoting functional cure [14, 15, 27, 32]. While higher EOT HBsAg-IC levels were not associated with HBsAg loss, the event rate was small (n = 5) and all participants who achieved HBsAg loss had very low EOT HBsAg levels, which may have reduced the assay's ability to quantify HBsAg-IC.
HBsAg-IC levels demonstrated a moderate correlation with HBsAg levels (r = 0.573). Our findings suggest that the association between HBsAg-IC and clinical outcomes was not merely a reflection of HBsAg levels. This is supported by several observations. The association between lower EOT HBsAg-IC levels and a higher risk of hepatitis flare is biologically plausible. In contrast, low EOT HBsAg levels have previously been linked to a reduced risk of NA-withdrawal flare [2, 29]. A higher HBsAg-IC:HBsAg ratio was associated with increased flare risk in our study, suggesting that anti-HBs activity may be the principal driver of this association. Finally, in the off-treatment analysis, relative increases in HBsAg-IC levels were consistently associated with hepatitis flare, whereas dynamic changes in HBsAg levels were not (Figure 2).
The mechanisms underlying the earlier occurrence of virological reactivation and hepatitis flare following tenofovir (vs entecavir) discontinuation remain unclear [2, 6, 33, 34]. Several pharmacodynamic hypotheses have been proposed, including differential impact on cccDNA transcriptional activity and interferon-λ3 production [6, 33, 35]. We found that tenofovir-treated patients had lower EOT HBsAg-IC levels, which may offer a novel explanation for this phenomenon. The observed association warrants validation in future studies.
To our knowledge, this is the first study to analyze HBsAg-IC levels in the setting of NA discontinuation. Despite the study's novelty, we acknowledge several limitations. The sample size was modest, and subgroup analyses, including off-treatment kinetics among those achieving HBsAg loss, were limited by the small numbers. Moreover, the median EOT HBsAg level for those who experienced HBsAg loss was very low (3.1 IU/mL), which may have reduced the assay's ability to detect HBsAg-IC. Therefore, further study with a large sample of participants who have experienced HBsAg loss is required. Testing for associations between EOT HBsAg-IC levels and baseline characteristics was performed in the overall cohort and therefore cannot determine causation. The case-control study design was important to identify differences between the flare and non-flare cohorts. However, the predictive utility of HBsAg-IC as a biomarker requires validation among a larger sample size and either a prospective cohort study or a randomized controlled trial. Finally, the generalizability of our findings to non-Asian populations or those infected with non-B or non-C genotypes is limited.
CONCLUSIONS
Our findings highlight the important role of humoral immunity in the pathogenesis of hepatitis flares and suggest that testing of serum HBsAg-IC level may aid in identifying suitable candidates for NA discontinuation. Lower HBsAg-IC levels at EOT were associated with increased flare risk and dynamic increases in HBsAg-IC off-treatment paralleled ALT peaks, suggesting immune-mediated liver inflammation during flares. Among individuals who experienced hepatitis flare, higher EOT HBsAg-IC levels and greater off-treatment HBsAg-IC elevation were associated with significant HBsAg decline, supporting the role of humoral immunity in effective anti-HBs responses. Further validation in larger cohorts is needed to confirm these results and establish HBsAg-IC as a clinically useful tool in CHB management.
Supplementary Material
Notes
Financial support . The clinical study was supported by the National Health and Medical Research Council Project Grant 1066536. A. J. T. received funding from the National Health and Medical Research Council of Australia (MRFF Practitioner Fellowship 1142976).
Ethical approval . The study was approved by the St Vincent's Hospital Ethics Committee (ID Number: 032/14) and at each participating site and was conducted in compliance with the principles of the 1975 Declaration of Helsinki and local regulatory requirements.
Data availability . Data are available from the corresponding author upon reasonable request, subject to ethics approval.
Informed consent . All participants provided written informed consent to participate in the study.
Contributor Information
S J Hume, St Vincent's Hospital, Melbourne, Victoria, Australia; Department of Medicine, The University of Melbourne, Victoria, Australia; Victorian Infectious Diseases Reference Laboratory, Royal Melbourne Hospital at the Peter Doherty Institute for Infection and Immunity, Melbourne, Victoria, Australia.
K Visvanathan, St Vincent's Hospital, Melbourne, Victoria, Australia; Department of Medicine, The University of Melbourne, Victoria, Australia.
K Cheng, Department of Medicine, The University of Melbourne, Victoria, Australia.
K Jackson, Victorian Infectious Diseases Reference Laboratory, Royal Melbourne Hospital at the Peter Doherty Institute for Infection and Immunity, Melbourne, Victoria, Australia.
S Vogrin, Department of Medicine, The University of Melbourne, Victoria, Australia.
S A L Hall, St Vincent's Hospital, Melbourne, Victoria, Australia; Department of Medicine, The University of Melbourne, Victoria, Australia.
G S Burns, Western Health, Melbourne, Victoria, Australia.
P V Desmond, St Vincent's Hospital, Melbourne, Victoria, Australia.
D Ratnam, Department of Medicine, School of Clinical Sciences at Monash Health, Monash University, Melbourne, Victoria, Australia.
M T Levy, Liverpool Hospital, Sydney, New South Wales, Australia.
R Sawhney, Eastern Health, Melbourne, Victoria, Australia.
S I Strasser, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia.
M C Ngu, Concord Hospital, Sydney, New South Wales, Australia.
M Sinclair, Austin Health, Melbourne, Victoria, Australia.
C Meredith, Bankstown-Lidcombe Hospital, Sydney, New South Wales, Australia.
G Matthews, St Vincent's Hospital Sydney, Sydney, New South Wales, Australia.
Z Valaydon, Western Health, Melbourne, Victoria, Australia.
D Wong, Austin Health, Melbourne, Victoria, Australia.
M Littlejohn, Victorian Infectious Diseases Reference Laboratory, Royal Melbourne Hospital at the Peter Doherty Institute for Infection and Immunity, Melbourne, Victoria, Australia.
J A Holmes, St Vincent's Hospital, Melbourne, Victoria, Australia; Department of Medicine, The University of Melbourne, Victoria, Australia.
P A Revill, Victorian Infectious Diseases Reference Laboratory, Royal Melbourne Hospital at the Peter Doherty Institute for Infection and Immunity, Melbourne, Victoria, Australia.
A J Thompson, St Vincent's Hospital, Melbourne, Victoria, Australia; Department of Medicine, The University of Melbourne, Victoria, Australia.
N Warner, Victorian Infectious Diseases Reference Laboratory, Royal Melbourne Hospital at the Peter Doherty Institute for Infection and Immunity, Melbourne, Victoria, Australia.
Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.
References
- 1. World Health Organization . Global hepatitis report 2024: action for access in low- and middle-income countries. Geneva, Switzerland: World Health Organization, 2024. [Google Scholar]
- 2. Hirode G, Choi HSJ, Chen C-H, et al. Off-therapy response after nucleos(t)ide analogue withdrawal in patients with chronic hepatitis B: an international, multicenter, multiethnic cohort (RETRACT-B study). Gastroenterology 2022; 162:757–71. e4. [DOI] [PubMed] [Google Scholar]
- 3. Sonneveld MJ, Chiu S-M, Park JY, et al. Probability of HBsAg loss after nucleo(s)tide analogue withdrawal depends on HBV genotype and viral antigen levels. J Hepatol 2022; 76:1042–50. [DOI] [PubMed] [Google Scholar]
- 4. Hall S, Howell J, Visvanathan K, Thompson A. The yin and the yang of treatment for chronic hepatitis B—when to start, when to stop nucleos(t)ide analogue therapy. Viruses 2020; 12:934. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Hall SAL, Vogrin S, Wawryk O, et al. Discontinuation of nucleot(s)ide analogue therapy in HBeAg-negative chronic hepatitis B: a meta-analysis. Gut 2022; 71:1629–41. [DOI] [PubMed] [Google Scholar]
- 6. Hall SA, Burns GS, Anagnostou D, et al. Stopping nucleot(s)ide analogues in non-cirrhotic HBeAg-negative chronic hepatitis B patients: HBsAg loss at 96 weeks is associated with low baseline HBsAg levels. Aliment Pharmacol Ther 2022; 56:310–20. [DOI] [PubMed] [Google Scholar]
- 7. Chen CH, Hung C-H, Wang J-H, et al. Long-term incidence and predictors of hepatitis B surface antigen loss after discontinuing nucleoside analogues in noncirrhotic chronic hepatitis B patients. Clin Microbiol Infect 2018; 24:997–1003. [DOI] [PubMed] [Google Scholar]
- 8. Cornberg M, Wong VW-S, Locarnini S, et al. The role of quantitative hepatitis B surface antigen revisited. J Hepatol 2017; 66:398–411. [DOI] [PubMed] [Google Scholar]
- 9. Papatheodoridi M, Hadziyannis E, Berby F, et al. Predictors of hepatitis B surface antigen loss, relapse and retreatment after discontinuation of effective oral antiviral therapy in noncirrhotic HBeAg-negative chronic hepatitis B. J Viral Hepatitis 2020; 27:118–26. [DOI] [PubMed] [Google Scholar]
- 10. Hume SJ, Wong DK, Yuen M-F, et al. High end-of-treatment hepatitis B core-related antigen levels predict hepatitis flare after stopping nucleot(s)ide analogue therapy. Liver Int 2024; 44:2605–14. [DOI] [PubMed] [Google Scholar]
- 11. Brakenhoff SM, de Knegt RJ, van Campenhout MJH, et al. End-of-treatment HBsAg, HBcrAg and HBV RNA predict the risk of off-treatment ALT flares in chronic hepatitis B patients. J Microbiol Immunol Infect 2023; 56:31–9. [DOI] [PubMed] [Google Scholar]
- 12. Lim YS. New biomarkers of hepatitis B virus infection: HBV RNA and HBcrAg, new kids on the block? Clin Mol Hepatol 2022; 29:29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Thompson AJ, Jackson K, Bonanzinga S, et al. Baseline serum HBV RNA is associated with the risk of hepatitis flare after stopping nucleoside analog therapy in HBeAg-negative participants. Hepatol Commun 2023; 7:e0188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Wong D, Littlejohn M, Edwards R, et al. ALT flares during nucleotide analogue therapy are associated with HBsAg loss in genotype A HBeAg-positive chronic hepatitis B. Liver Int 2018; 38:1760–9. [DOI] [PubMed] [Google Scholar]
- 15. Chang M-L, Liaw Y-F. Hepatitis B flare in hepatitis B e antigen-negative patients: a complicated cascade of innate and adaptive immune responses. Int J Mol Sci 2022; 23:1552. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Jeng W-J, Liu Y-C, Peng C-W, et al. Highly significant differences in HBsAg kinetics among patients with two types of hepatitis B flare, with and without retreatment. J Antimicrob Chemother 2022; 77:205–12. [DOI] [PubMed] [Google Scholar]
- 17. Liaw YF. Hepatitis B flare after cessation of nucleos(t)ide analogue therapy in HBeAg-negative chronic hepatitis B: to retreat or not to retreat. Hepatology 2021; 73:843–52. [DOI] [PubMed] [Google Scholar]
- 18. Chang JJ, Lewin SR. Immunopathogenesis of hepatitis B virus infection. Immunol Cell Biol 2007; 85:16–23. [DOI] [PubMed] [Google Scholar]
- 19. Jung M-C, Pape GR. Immunology of hepatitis B infection. Lancet Infect Dis 2002; 2:43–50. [DOI] [PubMed] [Google Scholar]
- 20. Hoofnagle JH. Reactivation of hepatitis B. Hepatology 2009; 49(S5):S156–65. [DOI] [PubMed] [Google Scholar]
- 21. Shi Y, Zheng M. Hepatitis B virus persistence and reactivation. BMJ 2020; 370:m2200. doi: 10.1136/bmj.m2200. [DOI] [PubMed] [Google Scholar]
- 22. Wang Y, Han S-HB. Hepatitis B reactivation: a review of clinical guidelines. J Clin Gastroenterol 2021; 55:393–9. [DOI] [PubMed] [Google Scholar]
- 23. Chisari FV, Ferrari C. Hepatitis B virus immunopathogenesis. Ann Rev Immunol 1995; 13:29–60. [DOI] [PubMed] [Google Scholar]
- 24. Madalinski K, Burczynska B, Heermann KH, et al. Analysis of viral proteins in circulating immune complexes from chronic carriers of hepatitis B virus. Clin Exp Immunol 1991; 84:493. [PMC free article] [PubMed] [Google Scholar]
- 25. Anh-Tuan N, Novak E. Detection and quantitation of hepatitis B surface antigen immune complexes (HBsAg-ICs) by an antigen-specific method. II. Circulating immune complexes (CICs) in patients with hepatitis B and asymptomatic HBsAg carriers. J Immunol Methods 1980; 35:307–18. [DOI] [PubMed] [Google Scholar]
- 26. Cai Y, Yin W. The multiple functions of B cells in chronic HBV infection. Front Immunol 2020; 11:582292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Xu H, Locarnini S, Wong D, et al. Role of anti-HBs in functional cure of HBeAg + chronic hepatitis B patients infected with HBV genotype A. J Hepatol 2022; 76:34–45. [DOI] [PubMed] [Google Scholar]
- 28. Thurnheer MC, Edwards R, Schulz TR, et al. Genotypic profiles of hepatitis B in African immigrants and their clinical relevance. J Med Virol 2017; 89:1000–7. [DOI] [PubMed] [Google Scholar]
- 29. Sonneveld MJ, Park JY, Kaewdech A, et al. Prediction of sustained response after nucleo(s)tide analogue cessation using HBsAg and HBcrAg levels: a multicenter study (CREATE). Clin Gastroenterol Hepatol 2022; 20:e784-e793. [DOI] [PubMed] [Google Scholar]
- 30. Hirode G, Hansen BE, Chen CH, et al. Incidence of hepatic decompensation after nucleos(t)ide analog withdrawal: results from a large, international, multiethnic cohort of patients with chronic hepatitis B (RETRACT-B study). Off J Am College Gastroenterol 2023; 118:1601–8. [DOI] [PubMed] [Google Scholar]
- 31. Tseng T-N, Jeng WJ, Hu TH, et al. Combined baseline HBcrAg and end-of-treatment HBsAg predict HBV relapse after entecavir or tenofovir cessation. J Antimicrob Chemother 2023; 78:436–9. [DOI] [PubMed] [Google Scholar]
- 32. Chang M-L, Liaw Y-F. Hepatitis B flares in chronic hepatitis B: pathogenesis, natural course, and management. J Hepatol 2014; 61:1407–17. [DOI] [PubMed] [Google Scholar]
- 33. Su T-H, Yang HC, Tseng TC, et al. Distinct relapse rates and risk predictors after discontinuing tenofovir and entecavir therapy. J Infect Dis 2018; 217:1193–201. [DOI] [PubMed] [Google Scholar]
- 34. Höner zu Siederdissen C, Hui AJ, Sukeepaisarnjaroen W, et al. Contrasting timing of virological relapse after discontinuation of tenofovir or entecavir in hepatitis B e antigen–negative patients. J Infect Dis 2018; 218:1480–4. [DOI] [PubMed] [Google Scholar]
- 35. Murata K, Asano M, Matsumoto A, et al. Induction of IFN-λ3 as an additional effect of nucleotide, not nucleoside, analogues: a new potential target for HBV infection. Gut 2018; 67:362–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.