HBsAg seroreversion after pegylated interferon alpha induced HBsAg seroclearance in patients with chronic hepatitis B: a real-world cohort study

Abstract

Background

The effect of different pegylated interferon alpha (Peg-IFNα) treatment strategies on the recurrence after Hepatitis B surface antigen (HBsAg) seroclearance has not yet been adequately elucidated.

Aims

This real-world study was conducted to compare the risk of HBsAg seroreversion after HBsAg seroclearance in hepatitis B e antigen (HBeAg)-negative patients with chronic hepatitis B (CHB) receiving Peg-IFNα-2b combination therapy with nucleos(t)ide analogues (NAs) and Peg-IFNα-2b monotherapy, and further to identify predictors by developing a risk prediction model.

Methods

Using a retrospective cohort design, 438 enrolled participants were partitioned to training and validation cohorts. The non-zero coefficient variables were identified via least absolute shrinkage and selection operator - Cox proportional hazards regression (Lasso-Cox), with subsequent multivariate Cox analysis establishing the predictive algorithm. Model performance underwent evaluation through calibration curves and time-dependent receiver operating characteristic (time-dependent ROC) analysis.

Results

The cumulative HBsAg seroreversion rates at 48 and 96 weeks were 15.7% and 18.9% in the combination group and 21.2% and 26.4% in the monotherapy group, respectively, with no statistically significant difference between groups.The final model incorporated two predictors: baseline HBsAg level (HR = 1.507, 95%CI: 1.081–2.101, P = 0.016), and end-of-treatment (EOT) hepatitis B surface antibody (HBsAb) level (HR = 0.973, 95%CI: 0.960–0.986, P < 0.001). The optimal cut-off value of HBsAb titer at EOT calculated by ROC curve was 32.28 IU/L. The calibration curves delineated favorable consistency for the nomogram model. Time-dependent ROC curves presented better discriminatory ability, with area under the curve (AUC) values of 0.781, 0.800, 0.820 and 0.827 at 12, 24, 48, and 96 weeks in the training set.

Conclusions

The observed durability of HBsAg negativity was not significantly improved with Peg-IFNα combination therapy compared with monotherapy. Baseline HBsAg and EOT-HBsAb levels may support individualized recurrence risk stratification.

Clinical trial

Not applicable.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12879-026-13989-x.

Introduction

Chronic hepatitis B virus (HBV) infection constitutes a substantial worldwide health burden, accounting for an estimated 1.1 million deaths annually. These mortalities principally stem from liver-related complications including hepatic decompensation and hepatocellular carcinoma (HCC) [1]. The therapeutic objectives in chronic hepatitis B (CHB) management center on achieving sustained viral control and preventing disease progression. Hepatitis B surface antigen (HBsAg) seroclearance, a hallmark of functional cure, represents the current therapeutic ideal given its association with favorable long-term clinical outcomes [2, 3].

Current antiviral strategies primarily rely on nucleos(t)ide analogs (NAs) and interferon-based therapies. Although NAs effectively inhibit reverse transcription of pregenomic RNA (pgRNA), they are unable to eliminate covalently closed circular DNA (cccDNA), the viral reservoir, leading to viral rebound upon treatment cessation [4–6]. Moreover, available evidences indicate inadequate HBsAg clearance rates with NA monotherapy [7–10]. In contrast, pegylated interferon-alpha (Peg-IFNα) exerts potential capacity to destabilize cccDNA reservoirs [11], thereby creating more opportunities for functional cure. Previous clinical studies have shown that Peg-IFNα-based therapy can raise HBsAg clearance rates more than NA monotherapy [12–16]. However, whether combination therapy provides incremental benefit over monotherapy in achieving HBsAg loss remains inconclusive [16–19]. Furthermore, it is still uncertain whether different Peg-IFNα-based treatment strategies influence the long-term durability of HBsAg seroclearance once functional cure has been achieved.

Accordingly, the present study aimed to compare the risk of HBsAg seroreversion after treatment-induced HBsAg seroclearance between patients treated with Peg-IFNα combination therapy in NAs-experienced patients and those treated with Peg-IFNα monotherapy, and subsequently to construct the clinically applicable prediction model to analyze predictors associated with HBsAg recurrence. We hope that this study will provide valuable real-world evidence to guide treatment strategy selection for CHB patients and clinical management for those who have already achieved HBsAg loss.

Materials and methods

Study design

This retrospective real-world analysis included CHB patients aged 20–60 years from Tongji Hospital (Huazhong University of Science and Technology, Wuhan, China), and Ezhou Central Hospital (Ezhou, China) between November 2019 and July 2025. The enrolled recipients were stratified into two cohorts based on treatment regimens: one receiving Peg-IFNα-2b combined with NA-experienced patients (≥ 6 months of oral medication), and the other receiving Peg-IFNα-2b monotherapy. For patients with combination therapy, the NA component consisted of contemporary first-line agents with a high barrier to resistance, including entecavir (ETV), tenofovir disoproxil fumarate (TDF), tenofovir alafenamide (TAF), and tenofovir amibufenamide (TMF). All participants were administered 180 µg of Peg-IFNα-2b via weekly subcutaneous injection, with a minimum treatment duration of 24 weeks. Most patients with good compliance received consolidation treatment with Peg-IFNα-2b after HBsAg clearance. Monitoring after treatment discontinuation usually required quarterly monitoring during the initial year, transitioning to biannual assessments thereafter. Inclusion criteria mandated confirmed HBsAg negativity (< 0.05 IU/mL) with undetectable HBV DNA (< 20 IU/mL) at treatment cessation; HBeAg-negative serostatus before Peg-IFNα-2b initiation. Exclusion criteria encompassed concurrent autoimmune liver diseases; hepatitis C, D virus (HCV, HDV) or human immunodeficiency virus (HIV) coinfection; and exposure history of immunosuppressive medications/corticosteroids during follow-up.

This research was approved by the Ethics Committee of Tongji Hospital (Approval No: TJ-IRB20220929), Huazhong University of Science and Technology, and it complies to the provisions of the Declaration of Helsinki.

Laboratory measurements

HBV serological markers underwent serological profiling at 3- to 6-month intervals during follow-up, with intensified surveillance frequency maintained throughout the initial 12-month period. Serum HBsAg and hepatitis B surface antibody (HBsAb) were quantified via electrochemiluminescence immunoassays (Elecsys HBsAg II Quant / Anti-HBs II; Roche Diagnostics, Mannheim, Germany). Serological criteria defined HBsAg negativity as < 0.05 IU/mL and HBsAb positivity as ≥ 10 IU/L. HBV DNA quantification employed the Cobas TaqMan assay (Roche Diagnostics, Branchburg, NJ), with a 20 IU/mL detection threshold.

Definitions and endpoints

The follow-up record was initiated upon cessation of Peg-IFNα therapy and terminated upon HBsAg seroreversion. Patients without HBsAg recurrence were subjected to continued observation to monitor HBV serological dynamics. Consolidation therapy duration was defined as the interval spanning from documented HBsAg loss to therapeutic discontinuation. Confirmed HBsAg seroclearance required ≥ 2 consecutive negative measurements with an interval of 3–6 months during Peg-IFNα therapy or ≥ 2 consecutive negative results within 3–6 months post-treatment withdrawal, including one assessment at treatment cessation. The primary endpoint was HBsAg seroreversion, which meant the reappearance of HBsAg during observation period. HBsAg seroconversion was defined by synchronous HBsAg negativity and HBsAb positivity at thresholds of < 0.05 IU/mL and ≥ 10 IU/L respectively.

Statistical analysis

Statistical summaries comprised median with interquartile ranges (IQR) for continuous variables and frequency counts (percentages) for categorical variables. Inter-group comparisons employed the Mann-Whitney U test for continuous variables and Pearson’s χ² test for categorical variables. Propensity score matching (PSM) was used to adjust the differences of clinical data between the groups. To evaluate whether the findings were affected by sample loss after 1:1 matching, stabilized inverse probability of treatment weighting (IPTW) was further performed as a sensitivity analysis using the full cohort. Multigroup categorical comparisons employed χ² or Fisher’s exact tests with post hoc adjustment using the Bonferroni method. Survival probabilities were estimated through Kaplan-Meier (K-M) analysis, with between-group differences assessed by log-rank testing. Predictive variables selection was conducted via least absolute shrinkage and selection operator - Cox proportional hazards regression (Lasso-Cox) regression, followed by multivariable Cox proportional hazards modeling. Model calibration was evaluated through calibration curves, and discriminatory power using time-dependent receiver operating characteristic (time-dependent ROC) curves. All analyses were executed in the R statistical environment (version 4.3.2), with a two-sided P value < 0.05 defining statistical significance.

Results

Clinical characteristics of enrolled patients

We totally registered 617 patients who achieved HBsAg clearance by Peg-IFNα-based therapy. The cohort comprised 392 NAs-experienced patients receiving combination therapy with Peg-IFNα, alongside 209 patients undergoing Peg-IFNα monotherapy. The cases enrollment protocol was delineated in Fig. 1. Following screening, 438 eligible subjects were ultimately included in the analytical dataset, stratified into treatment arms as follows: 303 in the combination therapy group versus 135 in the monotherapy group. In the combination therapy group, 94.1% (285/303) participants had received NAs orally treatment for at least 12 months before Peg-IFNα initiation; HBV DNA was still detectable at a low level in 14 of them and was undetectable in the remainders. Baseline HBsAg levels measured 1.84 (1.26–2.36) log₁₀IU/mL in the combination group and 1.53 (0.74–2.21) log₁₀IU/mL in the monotherapy group. The treatment duration for both cohorts were comparable (48 weeks). At treatment cessation, the proportions of HBsAg seroconversion were 64.7% in the combination therapy group versus 58.5% in the monotherapy group, respectively. Corresponding median HBsAb titers reached 46.09 (5.07-150.33) IU/L and 39.85 (2.00-128.62) IU/L. Univariate covariate analysis identified significant intergroup differences in gender, enrollment age, baseline HBsAg level, treatment time before HBsAg loss, and consolidation treatment duration (Table 1). Propensity score matching (PSM, 1:1 ratio) was performed to adjust the clinical parameters. A strict caliper threshold (0.2 times the standard deviation of propensity scores) was applied to ensure matching quality. The final matched cohort comprised 130 combination therapy cases and 130 monotherapy cases (Table 1). Post-matching standardized mean differences (SMD) for critical covariates were < 0.1, confirming adequate intergroup balance. We further performed stabilized IPTW as a sensitivity analysis using the full cohort, which improved covariate balance between the two groups, as shown in Supplementary Table 1.

Fig. 1.

Flowchart illustrating the case screening process. In combination group, Peg-IFNα-2b was added on NA-treated patients. In monotherapy group, Peg-IFNα-2b monotherapy was administered to initial patients

Table 1.

Clinical characteristics of the two groups before and after propensity score matching

Clinical parameters	Before PSM			After PSM			SMD
	Combination group	Monotherapy group	P-value	Combination group	Monotherapy group	P-value
	(n = 303)	(n = 135)		(n = 130)	(n = 130)
Gender, Male, n(%)	214 (70.6)	80 (59.3)	0.026	84 (64.6)	79 (60.8)	0.608	0.039
Age, years	42 (35, 48)	39 (34, 44)	0.024	38 (33, 45)	39 (34, 45)	0.511	0.063
Baseline HBsAg level, log10 IU/mL	1.84 (1.26, 2.36)	1.53 (0.74, 2.21)	0.002	1.56 (0.93, 2.08)	1.55 (0.84, 2.21)	0.981	0.008
Baseline ALT level, U/L	20.0 (14.0, 30.0)	21.0 (14.5, 29.0)	0.927	20.0 (14.0, 30.0)	21.0 (15.0, 29.0)	0.626	0.021
Baseline AST level, U/L	22.0 (18.0, 26.0)	21.0 (18.0, 26.0)	0.466	21.5 (18.0, 25.0)	21.0 (18.0, 26.0)	0.703	0.017
Treatment duration, weeks	48 (36, 48)	48 (36, 48)	0.155	48 (36, 48)	48 (36, 48)	0.827	0.032
Treatment time before HBsAg loss, weeks	24 (20, 36)	24 (12, 24)	0.001	24 (12, 24)	24 (12, 24)	0.631	0.008
EOT-HBsAb level, IU/L	46.09 (5.07, 150.33)	39.85 (2.00,128.62)	0.272	48.47 (4.39, 164.65)	39.92 (2.00,126.37)	0.295	0.084
EOT-HBsAb positivity, n(%)	196 (64.7)	79 (58.5)	0.260	81 (62.3)	77 (59.2)	0.703	0.031
Consolidation duration, weeks	24 (12, 24)	24 (12, 27)	0.054	24 (12, 24)	24 (12, 28)	0.909	0.049
Consolidation > 12 weeks, n(%)	186 (61.4)	98 (72.6)	0.031	95 (73.1)	93 (71.5)	0.890	0.015
EOT-ALT level, U/L	29.0 (18.5, 40.0)	32.0 (20.0, 47.0)	0.297	28.0 (18.0, 36.8)	33.0 (20.0, 46.0)	0.071	0.065
EOT-AST level, U/L	38.0 (28.0, 52.5)	41.0 (29.0, 61.5)	0.191	26.0 (28.0, 50.5)	40.5 (29.0, 61.0)	0.146	0.061
HBsAb positivity during follow-up, n(%)	170 (56.1)	66 (48.9)	0.195	62 (47.7)	64 (49.2)	0.901	0.015

ALT, Alanine aminotransferase; AST, Aspartate aminotransferase; EOT, end of treatment; HBsAb, hepatitis B surface antibody; IQR, interquartile range; PSM, propensity score matching; SMD, standardized mean difference

HBsAg recurrence on the two treatment regimens

In present study, 17.2% of patients (52/303) in the combination cohort experienced HBsAg recurrence during follow-up, while 21.5% (29/135) appeared in the monotherapy cohort. Within the combination group, HBsAg reversed into positive status in 5 of the 18 (27.8%) patients with < 12 months of NA therapy before Peg-IFNα add-on. HBsAg recurrence mainly occurred within 48 weeks of follow-up in both groups, while few recurrence events were documented after durable HBsAg negativity exceeding two years. Longitudinal analysis of K-M survival curves indicated that the cumulative recurrence rates at 12, 24, 48, and 96 weeks after treatment cessation were 5.0%, 9.6%, 15.7%, and 18.9%, respectively, in the combination therapy group. These outcomes did not demonstrate statistically significant superiority compared to those in the monotherapy group (8.1%, 14.8%, 21.2%, and 26.4%; P = 0.081. Fig. 2A). Propensity score-adjusted survival curves maintained consistent results in cumulative rates of HBsAg recurrence (P = 0.211, Fig. 2B). The IPTW-weighted Cox model with robust variance estimation also showed no significant difference in HBsAg seroreversion risk between the combination group and the monotherapy group (HR = 1.349, 95% CI 0.819–2.221, P = 0.240) (Supplementary Fig. 1).

Fig. 2.

Kaplan-Meier curves of HBsAg seroreversion in the combination group and monotherapy group before and after propensity score matching. A, Comparison analysis between the two unmatched groups; B, Comparison analysis between the two matched groups

We compared the clinical characteristics of patients with sustained HBsAg loss and HBsAg reversion in each treatment group, as shown in Table 2. In both combination therapy and monotherapy groups, patients with HBsAg reversion had significantly higher baseline HBsAg levels (2.23 and 1.96 log₁₀ IU/mL, respectively) compared to those without reversion (all P values < 0.05). The presence of HBsAb at the end of treatment (EOT) was a common characteristic among the majority of patients who did not experience HBsAg recurrence during follow-up, whereas detectable HBsAb at treatment completion was observed at lower percentage in those who developed HBsAg positivity for later (combination group: 71.7% vs. 30.8%, P < 0.001; monotherapy group: 69.8% vs. 17.2%, P < 0.001). The median time to HBsAg recurrence was comparable between the combination and monotherapy groups, both occurring at 24-week of follow-up. A notable difference was observed in the median duration of consolidation therapy between patients with and without HBsAg reversion in the combination group (12 weeks vs. 24 weeks, P < 0.001). On the contrary, no significant difference was found in the monotherapy group (24 weeks in both, P > 0.05).

Table 2.

Clinical characteristics in patients with sustained HBsAg loss and HBsAg reversion

Clinical parameters	Combination group			Monotherapy group
	Sustained HBsAg	HBsAg reversion	P-value	Sustained HBsAg	HBsAg reversion	P-value
	loss (n = 251)	(n = 52)		loss (n = 106)	(n = 29)
Gender, Male, n(%)	179 (71.3)	35 (67.3)	0.682	60 (56.6)	20 (69.0)	0.324
Age, years	42 (35, 48)	41 (33, 47)	0.214	39 (34, 44)	41 (34, 47)	0.472
Baseline HBsAg level, log10 IU/mL	1.71 (1.17, 2.29)	2.23 (1.70, 2.67)	< 0.001	1.28 (0.63, 2.15)	1.96 (1.47, 2.33)	0.035
Baseline ALT level, U/L	20.0 (14.0, 30.0)	19.5 (14.0, 27.3)	0.438	20.0 (14.0, 28.8)	24.0 (16.0, 30.0)	0.287
Baseline AST level, U/L	22.0 (18.5, 26.0)	21.0 (17.8, 25.5)	0.346	21.0 (18.0, 26.0)	22.0 (20.0, 27.0)	0.370
Treatment duration, weeks	48 (36, 48)	48 (48, 52)	0.011	48 (36, 48)	48 (37, 48)	0.119
Time to HBsAg loss, weeks	24 (12, 28)	36 (24, 36)	< 0.001	24 (12, 24)	24 (15, 36)	0.071
EOT-HBsAb level, IU/L	64.98 (8.35, 167.15)	4.24 (2.00, 22.97)	< 0.001	84.92 (5.42, 196.98)	2.01 (2.00, 5.52)	< 0.001
EOT-HBsAb positivity, n(%)	180 (71.7)	16 (30.8)	< 0.001	74 (69.8)	5 (17.2)	< 0.001
Consolidation duration, weeks	24 (12, 24)	12 (12, 24)	< 0.001	24 (12, 28)	24 (12, 24)	0.212
Consolidation duration, n(%)			< 0.001			0.206
≤ 12 weeks	85 (33.9)	32 (61.5)		28 (26.4)	9 (31.0)
12–24 weeks	110 (43.8)	17 (32.7)		46 (43.4)	16 (55.2)
> 24 weeks	56 (22.3)	3 (5.8)		32 (30.2)	4 (13.8)
EOT-ALT level, U/L	29.0 (19.0, 40.0)	27.0 (17.8, 40.0)	0.440	34.0 (19.3, 47.5)	27.0 (22.0, 42.0)	0.575
EOT-AST level, U/L	38.0 (28.0, 54.5)	36.5 (28.8, 47.3)	0.559	41.0 (30.0, 61.8)	43.0 (27.0, 59.0)	0.670
Follow-up duration, week, n(%)	96 (48, 132)	24 (12, 48)	< 0.001	48 (24, 96)	24 (12, 36)	< 0.001
HBsAb positivity during follow-up, n(%)	164 (65.3)	6 (11.5)	< 0.001	65 (61.3)	1 (3.4)	< 0.001

Predictors screening protocol and prognostic model construction

The dataset was randomly divided into training and validation subsets using a 7:3 split. Clinical variables related to HBsAg recurrence were initially screened using Lasso-Cox regression, with the distribution of clinical parameter coefficients illustrated in Fig. 3A. The optimal regularization parameter (λ) was identified through 10-fold cross-validation. Lasso regularization path analysis retained 2 features with non-zero coefficients at λ.1se = 0.047 (Fig. 3B). The screened variables were then subjected to the multivariable regression analysis.

Fig. 3.

Development of a predicted model for estimating HBsAg reversion risk. A, Lasso coefficient profiles for clinical parameters. The horizontal axis denoted the magnitude of log(λ) values in the lasso regression model, and the vertical axis represented the coefficient values of each feature in the model. Each curve showed the trajectory of the regression coefficient for each independent variable. B, The selection of optimal regularization parameter (λ) in Lasso regression using 10-fold cross-validation. The horizontal axis corresponded to the value of log(λ), and the vertical axis reflected the partial likelihood deviance. C, The determination of the optimal cut-off value of HBsAb level at treatment discontinuation by ROC curve. D, Nomogram model developed in the training set based on selected variables

In the training cohort, multivariable Cox regression analysis included two predictive indicators: baseline HBsAg level (HR = 1.507, 95%CI: 1.081–2.101, P = 0.016), and EOT HBsAb level (HR = 0.973, 95%CI: 0.960–0.986, P < 0.001), which were associated with the HBsAg recurrence (Table 3).

Table 3.

Multivariate Cox regression analysis of the selected variables in the training dataset

Clinical variables	HR	95%CI	P-value
Baseline HBsAg level (log10 IU/mL)	1.507	1.081–2.101	0.016
EOT-HBsAb level (IU/L)	0.973	0.960–0.986	< 0.001

ROC curve was employed to identified the optimal cut-off value of HBsAb level at EOT on HBsAg recurrence in training set. The results indicated that the area under the curve (AUC) of 0.797 (sensitivity: 0.868, specificity: 0.634), with a corresponding threshold value of 32.28 IU/L (Fig. 3C).

Based on this, we categorized the EOT HBsAb level and constructed a nomogram model incorporating baseline HBsAg level to estimate the probabilities of HBsAg reversion at 12, 24, 48, and 96-week follow-up (Fig. 3D). A higher baseline HBsAg level corresponded to a higher score, indicating an increased risk of subsequent HBsAg seroreversion. Likewise, an EOT-HBsAb level of ≤ 32.28 IU/L contributed a substantially higher score than those > 32.28 IU/L, suggesting that a lower level at treatment cessation was associated with poorer durability of HBsAg loss.

Performance evaluation for the predictive model

Predictive calibration of the nomogram was graphically evaluated in the training cohort using calibration curves (Fig. 4A-D), with subsequent validation performed in the validation set (Fig. 4E-H). At the four time points (12-week, 24-week, 48-week, and 96-week), the consistency between the predicted probability and the actual probability performed well. Meanwhile, time-dependent ROC curves were generated at the four time points to assess the discrimination ability of the model. The concordance index (C-index) values were 0.803 (95% CI: 0.758–0.848) in the training group and 0.767 (95% CI: 0.677–0.857) in the validation group. Temporal-distinction capability was further evidenced by time-dependent ROC analysis: in the training cohort, AUCs were 0.781 (12 weeks), 0.800 (24 weeks), 0.820 (48 weeks), and 0.827 (96 weeks); corresponding validation cohort AUCs reached 0.828, 0.856, 0.890 and 0.780, respectively (Fig. 4I-J).

Fig. 4.

Calibration and discrimination performance of the nomogram model for predicting HBsAg recurrence. A-D, The calibration plots at 12, 24, 48 and 96 weeks in the training cohort. E-H, The calibration plots at the corresponding time points in the validation cohort. I, The time-dependent ROC curves in training group. J, The time-dependent ROC curves in validation group

Comparison of consolidation treatment between the two groups

The effect of consolidation treatment duration on HBsAg seroconversion was explored in combination and monotherapy cohorts. The analysis demonstrated significantly higher seroconversion rates at treatment discontinuation in patients receiving > 12 weeks of consolidation therapy compared to the ≤ 12-week subgroup. The similar results emerged in both treatment modalities: combination group: 47.0% (≤ 12weeks) vs. 70.1% (12-24weeks, P = 0.001) and 81.4% (> 24weeks, P < 0.001); monotherapy group: 29.7% (≤ 12weeks) vs. 62.9% (12-24weeks, P = 0.008) and 80.6% (> 24weeks, P < 0.001). It was also worth mentioning that the proportions of HBsAg seroconversion in patients with consolidation time > 24 weeks exceeding 80% in both groups; however, no statistically significant differences were observed between > 24-week and 12-24-week subgroups (the former: 81.4% vs. 70.1%; the latter: 80.6% vs. 62.9%. All P-values > 0.05) (Fig. 5A & B).

Fig. 5.

Consolidation treatment after HBsAg clearance in the two groups. A, The proportions of HBsAg seroconversion at different consolidation durations in the combination group. B, The proportions of HBsAg seroconversion at different consolidation durations in the monotherapy group. C, The cumulative rates of HBsAg recurrence at different consolidation durations in the combination group. D, The cumulative rates of HBsAg recurrence at different consolidation durations in the monotherapy group

We further conducted K-M analysis to evaluate the correlation between consolidation treatment time and HBsAg recurrence. Given the limited data in the monotherapy group, our analysis was restricted to the cumulative durability of HBsAg loss within 96 weeks. In combination group, patients with > 12 weeks of consolidation therapy manifested lower recurrence outcomes compared to those with ≤ 12-week consolidation scheme (9.7% vs. 24.5% at 48 weeks; 13.3% vs. 27.0% at 96 weeks; log-rank P < 0.001; Fig. 5C). Nevertheless, no significant discrepancy was observed in the monotherapy group (20.8% vs. 22.9% at 48 weeks; 26.8% vs. 27.2% at 96 weeks; log-rank P = 0.659; Fig. 5D).

Discussion

It is known that after HBV enters the hepatocytes, it converts its genome into a cccDNA form that persists in the nucleus as a stable microchromosome. The longevity of hepatocytes enables durable cccDNA residence in the nucleus, this simultaneously preserves its transcriptional activity, allowing viral replication reinitiation in functionally cured patients under receiving immunosuppression therapy [20]. The currently available therapeutic regimens (NAs and Peg-IFNα) are incapable of eradicating cccDNA from the nuclear chromosome of hepatocytes. Furthermore, residual integrated HBV DNA (iDNA) fragments persist as genomically anchored elements, evading existing antiviral clearance mechanisms. Therefore, it is not unexpected that the virus may re-emerge during the subsequent observation period after the attainment of HBsAg loss through antiviral therapy.

In this study, the cumulative HBsAg seroreversion rates were numerically lower in the combination group than in the monotherapy group, but the difference was not statistically significant. A similar observation was reported by Pan et al. in a 96-week prospective study of 420 patients who achieved functional cure, in which HBsAg seroreversion rates were comparable between the IFN monotherapy group and the sequential NA-IFN group in both the intention-to-treat analysis (24.83% vs. 23.08%, P = 0.70) and the per-protocol analysis (15.50% vs. 15.25%, P = 0.95) [21]. These external data lend further support to the view that, Peg-IFNα-based combination therapy may not provide a clear advantage over monotherapy in maintaining long-term response. In our cohort, the relatively small size of the monotherapy group was a potential limitation for detecting a modest between-group difference, although the result remained unchanged after matching. This should not be interpreted as evidence of equivalence between the two treatment strategies, because the study was not designed or powered as a noninferiority or equivalence analysis.

Previous studies have reported that Peg-IFNα and NAs exert distinct immunomodulatory effects on both innate and adaptive immunity, including effects on NK cells and CD8 + T lymphocytes. These findings provide a biological rationale for considering potential immune differences between Peg-IFNα-based combination therapy and monotherapy [22]. However, immune-cell phenotypes and functions were not assessed in the present study; therefore, whether these mechanisms contributed to post-seroclearance durability in our cohort cannot be determined. In this retrospective cohort, the durability of HBsAg clearance was not statistically significant between the two treatment groups. This observation may partly relate to the immune status of patients who had already achieved HBsAg clearance, but this explanation remains inferential. Larger prospective studies with longer follow-up and integrated immunological assessments are needed to clarify the mechanisms underlying durability after HBsAg clearance.

Although treatment regimens did not materially affect the durability of HBsAg seroclearance in our cohort, heterogeneity in post-treatment outcomes remained evident at the individual level. This prompted us to further examine baseline factors related to the stability of HBsAg loss. Quantitative HBsAg has long been regarded as a useful biomarker for monitoring treatment response and predicting prognosis in CHB [23]. A multicenter long-term follow-up study of HBeAg-negative CHB patients treated with 48 weeks of Peg-IFNα, showed that lower baseline HBsAg levels were independently associated with subsequent HBsAg loss [24]. Recent evidence also suggests that baseline HBsAg level may be relevant to the durability of functional cure; Han et al. reported that patients with baseline HBsAg ≤ 100 IU/mL had a lower risk of relapse after achieving functional cure [25]. In HBeAg-negative CHB patients, cccDNA transcriptional activity is typically attenuated, whereas iDNA maintains stable genomic integration and drives sustained HBsAg production, thus emerging as the dominant HBsAg source [26]. Accordingly, a higher baseline HBsAg level may reflect a more persistent antigenic burden and a less stable off-treatment state. Consistent with this interpretation, we found that higher baseline HBsAg levels were significantly associated with HBsAg seroreversion during follow-up. This result suggests that baseline HBsAg level may be informative for evaluating the risk of HBsAg recurrence after HBsAg clearance, with further investigation warranted.

The presence of HBsAb is crucial for eliminating HBV infection [27]. It can neutralize circulating HBsAg and clearing infectious HBV particles in vivo, reflecting that the host’s immune system is capable of mounting an effective response to virus invasion. When patients acquired HBsAg clearance after antiviral treatment, the residual cccDNA and iDNA in hepatocyte nucleus implies that the virus has not been completely eradicated. From this perspective, the significance of HBsAb emerges distinctly. The persistent existence of HBsAb provides protection from viral rebound and ensures long-term serological and virological response. Our prior work demonstrated that HBsAb is closely related to the maintenance of HBsAg loss after Peg-IFNα therapy [28], adding to earlier evidence on the role of HBsAb in sustained seroclearance [29–31]. Recent studies support this analysis result. Guo et al. reported that EOT anti-HBs ≥ 1.3 log₁₀ IU/L independently identified patients at low risk of HBsAg reversion after Peg-IFNα discontinuation [32]. Another research conducted by Lu et al. observed a significantly higher cumulative recurrence rate in patients with HBsAb < 35.3 mIU/mL at Peg-IFNα cessation [33]. In this study, patients with higher HBsAb level at EOT were less likely to develop HBsAg reversion, which is in line with these observations and supports the relevance of EOT-HBsAb to the durability of HBsAg clearance.

However, not all patients who achieved HBsAg seroclearance are destined to subsequently attain HBsAg seroconversion. It is known that the function of B lymphocytes is of great importance to HBsAb production and secretion. The presence of baseline HBsAb-specific B cells in patients receiving interferon therapy represents a critical determinant for HBsAg seroconversion [34]. Whereas heightened HBsAg concentrations directly linked to the dysfunction involving both humoral and cellular specific immune function. Lowering HBsAg titers is anticipated to be available for the restoration of these specific immune responses, ultimately leading to a more efficient enhancement in the capacity to eliminate HBsAg [35]. Our observations discovered that patients with low HBsAg levels tended to develop HBsAb more readily after HBsAg loss. Besides, the administration of consolidation therapy with Peg-IFNα after HBsAg clearance also availed to increase the occurrence rate of HBsAg seroconversion, which has been bore out in previous research report [28]. Similarly, several studies also elucidated that consolidation treatment promoted the HBsAb levels in patients with HBsAg clearance [31, 36]. We could clearly observe that the proportion of HBsAg seroconversion at EOT in both combination and monotherapy groups were higher among patients completing > 12-week consolidation therapy compared to those with ≤ 12-week cohorts. Moreover, in the combination group, patients who received consolidation therapy > 12 weeks were at a lower risk to detect HBsAg positivity during follow-up. While a similar association was not observed in the monotherapy group. This difference should be interpreted cautiously, because the smaller sample size of the monotherapy group may have limited statistical power, and a significant result in one subgroup but not the other does not by itself indicates a true difference between treatment regimens. Therefore, this finding does not alter the overall conclusion that seroreversion risk did not differ significantly between Peg-IFNα combination therapy and Peg-IFNα monotherapy. Certainly, this also necessitated conducting a more comprehensive analysis within a broader sample size to ensure robust and reliable conclusion.

We then analyzed the factors related to HBsAg seroreversion. Baseline HBsAg and EOT HBsAb were identified as independent predictors in the multivariable model. This is similar to the previous results by Han et al., in which baseline HBsAg ≤ 100 IU/mL together with EOT HBsAb ≥ 500 mIU/mL were associated with a lower relapse incidence [25]. Our study further explored risk stratification by incorporating baseline HBsAg as a continuous variable into the predictive model and categorizing EOT-HBsAb using a ROC-derived cutoff value. In this cohort, patients with EOT-HBsAb levels ≤ 32.28 IU/L showed a higher observed risk of HBsAg recurrence during follow-up. This cutoff value should be interpreted as an exploratory, cohort-derived stratification threshold rather than a biologically definitive or universally applicable boundary. It should be note that baseline HBsAg and EOT-HBsAb were handled differently because their clinical implications in the present study were not identical. EOT-HBsAb has been linked to the durability of HBsAg seroclearance, as also shown in our previous research, and was additionally categorized to provide clinically intuitive risk stratification [28]. In contrast, baseline HBsAg contributed to prediction primarily as a continuous quantitative marker in the multivariable model. Retaining it in continuous form was thus considered more appropriate than dichotomization, which could have obscured the graded nature of its association with seroreversion risk.

This study also has some limitations. First, this was a retrospective cohort study with a relatively limited number of patients receiving Peg-IFNα monotherapy; therefore, residual confounding cannot be fully excluded despite statistical adjustment. Second, the exact duration of prior NA therapy and sustained HBV DNA suppression could not be reliably retrieved for all patients in the combination group. Third, formal fibrosis staging by liver biopsy or transient elastography was unavailable. Baseline platelet count was not collected, precluding reliable calculation of baseline FIB-4 or APRI, and fibrosis regression after HBsAg seroclearance could not be systematically evaluated because longitudinal liver stiffness measurements or histological assessments were not available. Fourth, standardized data on BMI, hepatic steatosis, and metabolic syndrome-related variables were lacking, limiting further assessment of metabolic influences on HBsAg loss and post-clearance durability. Finally, although the prediction model showed acceptable internal performance, external validation is required before clinical application.

In conclusion, among HBeAg-negative patients who achieved HBsAg seroclearance, Peg-IFNα combination therapy did not demonstrate a statistically significant advantage over monotherapy in reducing HBsAg seroreversion. The clustering of recurrence events within 48 weeks after treatment cessation highlights the need for close follow-up during this period. Baseline HBsAg and EOT-HBsAb levels may help identify patients with different recurrence risks and support individualized post-seroclearance monitoring.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

cccDNA

Covalently closed circular DNA

CHB

Chronic hepatitis B

C-index

Concordance index

EOT

End of treatment

HBsAg

Hepatitis B surface antigen

ETV

Entecavir

HBsAb

Hepatitis B surface antibody

HBeAg

Hepatitis B envelope antigen

HBV

Hepatitis B virus

HCC

Hepatocellular carcinoma

iDNA

Integrated HBV DNA

IQR

Interquartile range

KM

Kaplan-Meier

Lasso-Cox regression

Least absolute shrinkage and selection operator - Cox proportional hazards regression

NAs

Nucleos(t)ide analogues

Peg-IFNα

Pegylated interferon alpha

PgRNA

Pregenomic RNA

PSM

Propensity score matching

SMD

Standardized mean differences

TAF

Tenofovir alafenamide

TDF

tenofovir disoproxil fumarate

TMF

Tenofovir amibufenamide

Time-dependent ROC

Time-dependent receiver operating characteristic

Author contributions

WC Z. performed the statistical analysis and drafted the manuscript. J.C. and WJ S. contributed to the research concept and guidance. NN X, FB T, WY Z. and Y.W. acquired and screened the clinical data. JX S. and QR R. were responsible for providing the research thoughts and revising the manuscript. All authors reviewed and approved the final manuscript.

Funding

No fundings for the study.

Data availability

Data supporting the study findings are available from the corresponding authors on reasonable request.

Declarations

Ethics approval and consent to participate

This retrospective study was approved by the Ethics Committee of Tongji Hospital, Huazhong University of Science and Technology (Approval No. TJ-IRB20220929) and was conducted in accordance with the Declaration of Helsinki. As part of the ethics review process, an application for waiver of informed consent was reviewed and approved by the Ethics Committee. Therefore, the requirement for informed consent to participate was waived for this study. The data were collected from existing medical records and were anonymized before analysis to protect patient privacy.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.