The study population recruited a total of 95 HBeAg-negative chronic hepatitis B patients from the third Affiliated Hospital of Sun Yat-Sen University, Guangzhou, China from March 2019 to May 2021. All procedures were conducted according to the principles of the Declaration of Helsinki, the protocol and consent forms were approved by the Research Ethics Committee of the Third Affiliated Hospital of Sun Yat-sen University. All participants provided written informed consent. Inclusion criteria included age between 18 and 65 years, NAs treated experience more than 1 year, treatment-naive for interferon, and low hepatitis B surface antigen levels: HBsAg < 1500 IU/ml. The exclusion criterion for this study included advanced liver cirrhosis, presence of hepatocellular carcinoma or other malignancies, other causes of liver disease or mixed causes, human immunodeficiency virus co-infection, and known serious psycho-pathological disorders.

All patients received peg-interferon α-2b (Pegberon, Amoytop Biotech Co., Ltd., China.) treatment based on the chronic administration of NAs for 48 weeks. The course could be shortened if a patient showed HBsAg seroclearance (HBsAg<0.05IU/ML) and lasted more than 12 weeks during the follow-up. The patients who achieved HBsAg seroclearance during the course of treatment was defined as cured group. Conversely, if the patient had detectable serum HBsAg at the end of treatment (EOT), the given patient was attributed to the uncured group. Synchronous serum samples were collected before and during Peg-IFNα treatment at 12-week intervals for 48 weeks. Blood sample (10 ml) from each patient was collected in the EDTA-treated tube. Plasma was collected after centrifugation, then peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll-Hypaque density gradient centrifugation. All samples were stored at −80°C and used for the present study.

Blood routine indexes were assessed using an automated blood cell analyzer (Sysmex XN-3000, Kobe, Japan). Biochemical indexes of liver function were detected by an automatic biochemical analyzer (HITACHI 7600, Japan). HBeAg, HBeAb, HBcAb, and HBsAb level testing were performed using Roche Cobas E601 automatic electrochemiluminescence assay (Roche Diagnostics, IN, Germany). HBsAg was measured with Elecsys HBsAg II Quant reagent kits (Roche Diagnostics, Indianapolis, IN, USA). Serum HBV DNA was assayed with a Roche COBAS AmpliPrep/COBAS TaqMan HBV Test v2.0 (Roche Molecular Diagnostics, Branchburg, NJ, USA). These analyses were performed at the laboratory department of the third Affiliated Hospital of Sun Yat-Sen University and all procedures were performed following the manufacturer’s instructions.

Serum samples of patients were collected by centrifugation of blood at 3200 rpm for 10 min and stored immediately at -80°C until use. Using FACS Calibur flow cytometer (BD, San Diego, USA), we detected 10 cytokines (IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12, IL-17 and TNF-α) with the cytokine detection kit (multiple microsphere flow immunofluorescence method; Changsha Weimi Bio-Tech Co., Ltd., China) by following manufacturer’s instructions.

Thawed PBMCs samples were taken from five patients (two in cured group and three in uncured group). A total of 15 blood samples were collected at 3 time points: day 0 (the day before), week 24, and week 48 of treatment. Samples were washed 3 times by 1x PBS with centrifugation at 400 g for 6 min and counted for cell numbers. 500,000 cells per tube were then stained for 30 min on ice with antibodies. For flow cytometric analyses, the following antibodies were used: CD3 (CD3-FITC; BioLegend, cat#300305), CD4 (CD4-APC-Cy7; BioLegend, cat#317418), CD8 (CD8-PerCP-Cy5.5; BioLegend, cat#344710), CD19 (CD19-APC; BioLegend, cat#302212), CD20 (CD20-PE; BioLegend, cat#302306), CD25 (CD25-APC; BioLegend, cat#302610), CD127 (CD127-PE; BioLegend, cat#351304), and PD-1 (PD-1-BV421; BioLegend, cat#329920). Cells stained by antibodies were resuspended in 1x PBS, and analyzed by flow cytometry using BD LSR Fortessa cell analyzer. Flow cytometry data were analyzed using FlowJo (version 10.4.0). Accordingly, B cells and Treg cells were selected as CD19⁺ CD20⁺ and CD3⁺ CD4 ⁺ CD25⁺ CD127⁻ subsets, independent CD3⁺ clustering of either CD4⁺ or CD8⁺ T cells with PD-1⁺ used for identifying PD-1 expression respectively from lymphocytes. Flow cytometry gating strategies for CD3+CD4+/CD3+CD8+ T cell populations see Supplementary Figure 2 .

Data were analyzed using SPSS (version 21.0 for Windows, Chicago, IL, USA). Continuous variables were shown as medians and interquartile ranges (IQR) unless stated otherwise. Mann-Whitney U test was used for the comparison among groups. The ROC curve was generated to analyze the predictive probability, and the area under the ROC curve (AUC) was calculated. All tests were two-tailed and P values of < 0.05 were considered to indicate a significant difference.

Seventy-eighty out of nighty-five eligible patients with complete serum cytokines were included in this analysis. The demographic and clinical characteristics of patients were shown in Table 1 . In all patients, 30 individuals (38.5%) achieved HBsAg seroclearance. The mean age in the cured group was 40.6 (36-45) years old, which was younger than that of the uncured group. There was a comparable gender distribution in both groups with which the majority were male 67 (85.9%): 26 (86.7%) in the cured group and 41 (85.4%) in the uncured group. Baseline HBsAg was significantly lower in the cured group (311.6 (39.0-485.3) than that in the uncured group (460.1 (185.2-741.3), P =0.031). No significant differences in the levels of white blood cell (WBC) (P = 0.077), neutrophil (P = 0.132), hemoglobin (P = 0.124), platelets (P = 0.204), albumin (P = 0.224), ALT (P = 0.305), AST (P = 0.486) and total bilirubin(TBIL)(P = 0.461) were found between two groups.

The kinetics of cytokine expression in two groups were shown in Figure 1 . The expressions of IL-1β, IL-2, IL-4, IL-10 and IL-12 showed similar overall upward trends throughout the course of treatment in all patients, with no significant difference (IL-1β: P0W = 0.307, P24W = 0.212, P48W = 0.082; IL-2: P0W = 0.979, P24W = 0.303, P48W = 0.124; IL-4: P0W = 0.938, P24W = 0.781, P48W = 0.118; IL-10: P0W = 0.845, P24W = 0.906, P48W = 0.579; IL-12: P0W = 0.508, P24W = 0.589, P48W = 0.459) ( Figures 1A, B, C, G, H ). The levels of IL-17 decreased at 24 weeks then increased at 48 weeks in the cured group while constantly elevated in the uncured group (IL-17: P0W = 0.497, P24W = 0.16, P48W = 0.126); ( Figure 1I ). IL-8 exhibited a downward trend before 24 weeks then increased at 48 weeks in both groups (IL-8: P0W = 0.704, P24W = 0.369, P48W = 0.479) ( Figure 1F ). The constantly elevated IL-6 expression was found in all patients before 24 weeks. Furthermore, IL-6 expression increased further until 48 weeks in the uncured group while it was decreased at 48 weeks in the cured group (IL-6: P0W = 0.244, P24W = 0.637, P48W = 0.246) ( Figure 1E ). In contrast to the persistently slow increase in the uncured group, TNF-α displayed a significant increase in the cured group from 24 weeks to 48 weeks (TNF-α: P0W = 0.618, P24W = 0.558, P48W = 0.059) ( Figure 1J ). IL-5 was notably increased in cured group during the whole course of treatment, while it decreased in the uncured patients at 24 weeks and then increased. Interestingly, the difference between the cured and the uncured groups was statistically significant (IL-5: P0W = 0.596, P24W = 0.033, P48W = 0.027) ( Figure 1D ).

We also investigated the relationship between cytokines expression and HBsAg seroclearance. Serum levels of IL-5 were associated with HBsAg seroconversion between both groups at 24 and 48 weeks of treatment ( Figure 1D ). We thus performed ROC curve analysis. The results of the diagnostic accuracy of IL-5 expression to predict the rate for HBsAg seroconversion were shown in Figure 2 . The AUCs presented by IL-5 expression at 24 and 48 weeks were 0.644 (95% CI, 0.515-0.773) with P < 0.05 and 0.649 (95% CI, 0.517-0.781) with P < 0.05 respectively.

As shown in Figure 3A , the proportion of CD4⁺ T cells were increased from the beginning of treatment to 24 weeks in the cured patients, then decreased to the basal level at 48 weeks, however, the uncured group showed an opposite trend, a decrease from 0 to 24 weeks after treatment, which was followed by a slow increase from 24 to 48 weeks. The similar patterns found in the CD8⁺ T cells were shown in Figure 3B . The proportion of CD8⁺ T cells changed less at 24 weeks after treatment in the cured group, and then increased from 24 to 48 weeks. Conversely, the uncured patients showed an elevation of CD8⁺ T cells within 24 weeks after therapy, which was decreased a little from 24 to 48 weeks.

The effects of B cell subsets among CHB patients following combination treatment were also examined and results were shown in Figure 4 . In both groups, B cell subsets showed similar increases until 24 weeks of treatment, then further increased in the cured group while decreased in the uncured group at 48 weeks.

To explore the mechanisms underlying immunomodulation during combination treatment, we measured Treg population ( Figure 5 ). The proportion of Treg cells in the uncured group was persistently elevated after treatment. In the cured group, this proportion was increased until 24 weeks then decreased at 48 weeks.

In the end, we examined PD-1 expressions in T cells from available patients ( Figure 6A ). As a result, the expressions reached to similar levels in both groups at 24 weeks, however, at 48 weeks, PD-1 expressions were higher from both CD4⁺ T cells ( Figure 6B ) and CD8⁺ T cells ( Figure 6C ) in the uncured group compared to that in the cured group, especially obvious from CD8⁺ subsets ( Figure 6C ).