CHB Patients hospitalized at Beijing Ditan Hospital’s Second Department of Hepatology between June 2021 and August 2022 made up the prospective cohort. According on their course of therapy, the patients were categorized into three groups: The Naïve group consisted of CHB patients who had never undergone antiviral medication, while patients who had only ever undergone first-line NAs antiviral medication made up the NA-treated group. Patients who had previously undergone pegylated interferon alfa (Peg-IFNα) therapy were enrolled in the Plateau group due to the failure of HBsAg to continuously decrease.

In the Naïve group, patients began to receive subcutaneous injection of Peg-IFNα-2a 180 μg/week, and selected first-line NAs antiviral drugs (entecavir [ETV] 0.5 mg/day or tenofovir disoproxil fumarate [TDF] 300 mg/day or tenofovir alafenamide fumarate [TAF] 25 mg/day) according to their clinical data and personal wishes for combined treatment. Peg-IFNα-2a 180 μg/week was administered subcutaneously to patients in NA-treated group on the basis of NAs antiviral drugs. Blood samples were taken from patients in Naïve group and NA-treated group at baseline, 4 week and 12-24 weeks after IFN treatment. The HBV virological and serological indices, biochemical indices, and phenotypes associated with T lymphocytes were identified. In the Plateau group, when HBsAg decreased to the plateau phase, the treatment of IFN was suspended and entered into intermittent treatment for 12-24 weeks, during which only ETV (0.5 mg/day) or TDF (300 mg/day) or TAF (25 mg/day) was used to maintain the antiviral efficacy, and after the interval, IFN was retreated on the basis of NAs. Blood samples were taken from patients in Plateau group at the baseline of plateau, during intermittent treatment of 12-24 weeks and during 12-24 weeks of IFN retreatment. The HBV virological and serological indices, biochemical indices, and phenotypes associated with T lymphocytes were identified. In addition, liver ultrasonography was performed every 12-24 weeks at baseline and during the follow-up period.

In addition, the comparison between the Plateau group and its corresponding control group (that is, patients who continued to receive IFN treatment when HBsAg droped to a plateau) could more accurately portray the advantages of IFN intermittent therapy and the relationship between the changes of T lymphocyte-related surface markers and clinical efficacy during this period. However, patients who HBsAg levels decreased < 1 lg IU/mL after 24 weeks of Peg-IFNα therapy should stop using IFN (3). Even if they remained receiving IFN therapy, patients whose levels of HBsAg plateaued found it challenging to experience a continuous fall in the level of HBsAg, according to our earlier studies (9, 12, 20). Therefore, we did not create a control group for the Plateau group in this study.

The Ethics Committee of Beijing Ditan Hospital, which is a part of Capital Medical University, granted approval for this study (Jing Di Lun Ke Zi 2018 NO. 023-01), and it was registered as a clinical trial (NCT04028856). It was designed and implemented according to Helsinki standards. All selected patients have signed informed consent.

Inclusion criteria (1): 18-65 years old; (2) HBsAg positive lasting more than 6 months (≥ 0.05 IU/mL). In addition, it was also necessary for Naïve group: (1) never received antiviral therapy; it was also necessary for NA-treated group: (1) so far and only take first-line oral antiviral therapy more than 6 months. For Plateau group, it also needed to be satisfied: (1) the level of HBsAg decreased to the plateau period after IFN antiviral treatment for more than 6 months. In order to be considered in a plateau period, the level of HBsAg must have dropped by 0.5 lg IU/mL from the preceding detection time point (more than 3 months) (12).

Exclusion criteria: (1) previously known viral hepatitis (except hepatitis B) and epstein-barr virus, cytomegalovirus, human immunodeficiency virus and other non-hepatitis virus infections; (2) complicated with other liver diseases, for instance metabolic-related fatty liver disease, alcoholic hepatitis, autoimmune hepatitis, hepatolenticular degeneration, decompensated cirrhosis and liver cancer; (3) mental illness; (4) severe heart, brain, lung, kidney or other system diseases; (5) usage of other liver-damaging drugs and/or chronic alcoholism; (6) usage of hormones and/or immunosuppressants.

The peripheral venous blood of the patient was collected at the specified time point and put into the EDTA anticoagulation tube. Using lymphocyte separation solution, peripheral blood mononuclear lymphocytes (PBMC) were isolated by density gradient centrifugation. All samples were processed and analysed within 24 hours after sample collection.

PBMCs and direct coupling antibody were incubated together in dark environment at 4°C for 20 minutes. Then 2 mL of 1 × PBS were added to wash lymphocytes. After rotating 1200 rpm for 5 minutes, the supernatant was discarded. Finally, 200 mL of 1 × PBS was added to make the pre-test sample. The monoclonal antibodies were anti-human CD4-APC-Fire750 (Clone SK3), CD8-BV510 (Clone SK1), HLA-DR-PE-Cy7 (Clone L243), and CD38^-APC (Clone HB-7) (BioLegend, San Diego, CA, USA). FACSCanto flow cytometry (BD Biosciences, San Diego, CA, USA) and FlowJo software (Tree Star, Ashland, OR, USA) was used to gather and analyze the data, respectively.

Roche CobasTaqMan 96 (Roche, Pleasanton, CA, USA), an automatic real-time fluorescence quantitative polymerase chain reaction (PCR) detector, was utilized to detect serum HBV DNA load. Serum HBsAg and hepatitis B e antigen (HBeAg) levels were measured by Abbott Architect i2000 kits (Abbott Diagnostics, Abbott Park, IL, USA), and liver function indexes were detected by Hitachi 7600 automatic biochemical analyzer (Wako Pure Chemical Industries, Tokyo, Japan).

SPSS 21.0 (IBM Corporation, Chicago, IL, USA) and GraphPad5 (GraphPad Software, La Jolla, CA, USA) software were used for statistical analysis. The categorical variables were expressed by frequency and percentage, and the chi-square test was used to compare the differences between groups. The normality of continuous variables was evaluated by Shapiro-Wilk test. Continuous variables that were normally distributed were represented by mean ± standard deviation (SD), and continuous variables that were not normally distributed were represented by the median and inter-quartile range (median, Q1-Q3). According to the normality of the baseline data, One-way ANOVA test and Kruskal-Wallis test were respectively used to compare the multiple groups at baseline, and then Tukey’s multiple comparisons test and Dunn’s multiple comparisons test were utilized to conduct pairwise comparisons between the groups. to make pairwise comparisons between the groups. The linear mixed effect model (restricted estimation maximum likelihood) was used to analyze the changes of parameters at different time points in each group, and the multiplicity of P value was adjusted by Bonferroni method. Patients were taken as random effects, groups and measurement time points were taken as fixed effects. The standard model was selected by likelihood ratio test. It did not violate the standard model hypothesis. The adjusted model took age and sex as covariables to exclude potential confounding factors. The Spearman correlation test was used to get the P values and correlation coefficient. All meaningful analyses were double-tailed, and P < 0.05 was statistically significant.

This study comprised 150 patients with CHB in total, including 53 patients in Naïve group, 51 patients in NA-treated group and 36 patients in Plateau group. Table 1 displayed the initial demographic data of patients. Then, according to the wishes of the patients, 37 patients in the Naïve group were willing to start IFN treatment, collecting a total of 93 samples at 3 time points for follow-up detection and analysis, while 13 patients in the NA-treated group were willing to start using IFN on the basis of the original NA treatment, and a total of 33 samples were gathered at 3 different time points for follow-up analysis. 13 Plateau group patients were unable to complete the follow-up examination in our hospital, so a total of 56 samples were gathered from 23 patients at 3 time points for follow-up analysis.

In comparison to both the Naïve group and the NA-treated group, HBsAg amount in the Plateau group was much lower (P < 0.0001; P = 0.0122). Although the HBsAg level in the NA-treated group was marginally less than that in the Naïve group, there was no markedly significant variation (P = 0.0680). HBV DNA load was considerably higher in the Naïve group compared to the NA-treated group and Plateau group, respectively (P < 0.0001; P < 0.0001), but there was no remarkable variation in HBV DNA level between NA-treated and Plateau group (P = 0.9019). Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels in NA-treated group were notably lower than those in Naïve group and Plateau group respectively (ALT: P = 0.0015; P < 0.0001; AST: P = 0.0106; P = 0.0395), but there was no discernible variation in terms of ALT and AST levels between Naïve group and Plateau group (ALT: P = 0.3170; AST: P > 0.9999), as shown in Table 1 .

In order to determine the level of T lymphocyte activation under different treatment conditions, we analyzed the coexpression of HLA-DR and CD38, representing the key T lymphocyte activation indicator. We defined three subsets in activated HLA-DR⁺ CD4⁺ T and CD8⁺ T lymphocytes based on CD38 expression: HLA-DR⁺CD38^- subgroup, HLA-DR⁺CD38^dim subgroup and HLA-DR⁺CD38^hi subgroup ( Figure 1 ). The percentage of HLA-DR⁺CD38^- CD4⁺ T lymphocytes in Plateau group was much less than that in Naïve group and NA-treated group (P = 0.0075; P < 0.0001). In the NA-treated group, more HLA-DR⁺CD38^-CD4⁺ T lymphocytes were present than those in the Naïve group (P = 0.0289). HLA-DR⁺CD38^dim CD4⁺ T lymphocyte percentage in Plateau group was considerably greater than that in Naïve and NA-treated group (P < 0.0001; P < 0.0001), but no discernible variation was observed between Naïve and NA-treated group (P > 0.9999). The proportion of HLA-DR⁺CD38^hi CD4⁺ T lymphocytes in Plateau group was considerably greater than that in Naïve and NA-treated group (P < 0.0001; P < 0.0001), but there was no discernible variation in the percentage of HLA-DR⁺CD38^hi CD4⁺ T lymphocytes between Naïve group and NA-treated group (P > 0.9999). HLA-DR⁺CD38^dim CD8⁺ T lymphocyte percentage in Plateau group was noticeably less than that in Naïve group and NA-treated group, respectively (P = 0.0003; P < 0.0001), however there was also no discernible variation in HLA-DR⁺CD38^dim CD8⁺ T lymphocyte level between Naïve group and NA-treated group (P > 0.9999). The difference trend of HLA-DR⁺CD38^- CD8⁺ T lymphocytes and HLA-DR⁺CD38^hi CD8⁺ T lymphocytes in different groups was similar to that of CD4⁺ T lymphocytes. HLA-DR⁺CD38^- CD8⁺ T lymphocyte percentage in Plateau group was considerably lower than that in Naïve and NA-treated group (P < 0.0001; P < 0.0001). In NA-treated group, HLA-DR⁺CD38^- CD8⁺ T lymphocyte percentage was noticeably more than that in Naïve group (P = 0.0002). In comparison to the Plateau group, the HLA-DR⁺CD38^hi CD8⁺ T lymphocyte percentage was considerably less in Naïve group and NA-treated group (P < 0.0001; P < 0.0001). HLA-DR⁺CD38^hi CD8⁺ T lymphocyte proportion was not markedly different between Naïve group and NA-treated group (P > 0.9999), as shown in Table 1 . Our data initially showed that the phenomenon that the antiviral efficacy of patients with long-term use of IFN could not continue was related to the change of HLA-DR and CD38 coexpression subsets percentage on CD4⁺ and CD8⁺ T lymphocytes after long-term IFN treatment.

In order to further explore the dynamic changes in HLA-DR and CD38 coexpression subsets on CD4⁺ and CD8⁺ T lymphocytes during IFN therapy, we followed up these patients who chose to start antiviral therapy with IFN in Naïve group and NA-treated group. Compared with those before IFN treatment, there was no noteworthy variation in the percentage of HLA-DR⁺CD38^- CD4⁺ T lymphocytes in Naïve group and NA-treated group after IFN treatment (Naïve group: 0 week vs. 4 week P > 0.9999; 0 week vs. 12-24 week P = 0.7210; NA-treated group: 0 week vs. 4 week P > 0.9999; 0 week vs. 12-24 week P > 0.9999). In both Naïve group and NA-treated group, HLA-DR⁺CD38^dim CD4⁺ T lymphocytes proportion increased considerably after 4 weeks of IFN treatment (P = 0.0020; P = 0.0010). HLA-DR⁺CD38^dim CD4⁺ T lymphocyte percentage in the Naïve group after 12-24 weeks of IFN treatment did not differ dramatically from that after 4 weeks of IFN treatment (P = 0.2020), but the HLA-DR⁺CD38^dim CD4⁺ T lymphocyte percentage in NA-treated group was dramatically less than that in IFN treatment for 4 weeks (P = 0.0090). The change trend in HLA-DR⁺CD38^hi CD4⁺ T lymphocyte percentage was consistent in Naïve group and NA-treated group during the treatment of IFN. At the beginning of treatment, HLA-DR⁺CD38^hi CD4⁺ T lymphocyte percentage in both groups increased compared with the baseline, but there was no discernible variation (P = 0.3500; P = 0.5110). In the late stage of IFN treatment, HLA-DR⁺CD38^hi CD4⁺ T lymphocyte percentage in both groups was dramatically higher than that at baseline (P < 0.0001; P < 0.0001). In Naïve group and NA-treated group, the proportion of HLA-DR⁺CD38^- CD8⁺ T lymphocytes decreased gradually with the increase of IFN treatment time (Naïve group: 0 week vs. 4 week P = 0.0170; 0 week vs. 12-24 week P < 0.0001; NA-treated group: 0 week vs. 4 week P = 0.0010; 0 week vs. 12-24 week P < 0.0001). In both Naïve group and NA-treated group, the proportion of HLA-DR⁺CD38^dim CD8⁺ T lymphocytes increased notably after 4 weeks of IFN treatment (P < 0.0001; P < 0.0001). After 12-24 weeks of IFN treatment, the percentage of HLA-DR⁺CD38^dim CD8⁺ T lymphocytes decreased notably compared with that of IFN treatment for 4 weeks in both groups (P < 0.0001; P < 0.0001). Moreover, with the increase of IFN treatment, the change trend of HLA-DR⁺CD38^hi CD8⁺ T lymphocyte percentage in Naïve group and NA-treated group was similar, which increased significantly with the duration of treatment (Naïve group: 0 week vs. 4 week P < 0.0001; 0 week vs. 12-24 week P < 0.0001; NA-treated group: 0 week vs. 4 week P < 0.0001; 0 week vs. 12-24 week P < 0.0001) ( Table 2 ).

The aggregate data showed that there was no significant change in HLA-DR⁺CD38^- on CD4⁺ T lymphocytes in both groups during the treatment of IFN, but the proportion of CD8⁺ T lymphocytes decreased gradually. Except that the percentage of HLA-DR⁺CD38^dim CD4⁺ T lymphocytes in Naïve group did not change significantly at the later stage of IFN treatment, the changes of other HLA-DR⁺CD38^dim subsets on CD4⁺ and CD8⁺ T lymphocytes in both groups increased markedly first and then decreased gradually. Meanwhile, the proportion of HLA-DR⁺CD38^hi on both CD4⁺ and CD8⁺ T lymphocytes increased significantly as the increase of IFN treatment time.

In Naïve group and NA-treated group, the HBsAg levels of all samples collected at 12-24 weeks after IFN treatment decreased more than 0.5 lg IU/mL compared with their respective treatment baselines. In Naïve group, compared with the baseline, HBsAg and HBV DNA levels declined considerably at 12-24 weeks after IFN treatment (P = 0.0010; P < 0.0001). At 12-24 weeks after IFN treatment, the level of HBsAg in NA-treated group declined markedly when compared to the baseline (P = 0.0270), but HBV DNA load had no significant change compared with the baseline (P = 0.3240). ALT and AST levels in Naïve group and NA-treated group were markedly higher than their respective baselines early in the course of IFN treatment (ALT: P = 0.0250; P = 0.0310; AST: P = 0.0080; P = 0.0420), but there was no discernible change compared with the baseline in the late stage of IFN therapy (ALT: P > 0.9999; P > 0.9999; AST: P = 0.8370; P > 0.9999), as shown in Table 2 .

Then, we used the longitudinal data collected from all patients in both groups to analyze the correlation between HBsAg and HLA-DR and CD38 coexpression subsets on CD4⁺ and CD8⁺ T lymphocytes ( Figure 2 ). The proportion in HLA-DR⁺CD38^hi CD4⁺ T lymphocytes and HLA-DR⁺CD38^hi CD8⁺ T lymphocytes was adversely linked with HBsAg level, respectively (R = -0.3298, P = 0.0025; R = -0.3495, P = 0.0013). HLA-DR⁺CD38^- and HLA-DR⁺CD38^dim subset proportions on CD4⁺ and CD8⁺ T lymphocytes were not linked with HBsAg level. Collectively, these findings suggested that when the patients did not enter the plateau phase, the proportion of HLA-DR⁺CD38^hi CD4⁺ T lymphocytes and HLA-DR⁺CD38^hi CD8⁺ T lymphocytes increased gradually with the decrease of HBsAg level during IFN treatment.

In order to further explore whether the persistent high proportion of HLA-DR⁺CD38^hi subset on CD4⁺ and CD8⁺ T lymphocytes played a role in the plateau phase following long-term treatment of IFN, we continued to analyze the data of patients in the Plateau group after IFN intermittent treatment and IFN retreatment. Our data showed that after 12-24 weeks of IFN intermittent therapy, the proportion of HLA-DR⁺CD38^- CD4⁺ T lymphocytes (P < 0.0001) and HLA-DR⁺CD38^dim CD4⁺ T lymphocytes (P < 0.0001) increased significantly compared with the plateau baseline, while the HLA-DR⁺CD38^hi CD4⁺ T lymphocyte percentage decreased significantly compared with the plateau baseline (P < 0.0001). After 12-24 weeks of IFN intermittent treatment, the corresponding subsets of CD8⁺ T lymphocytes also had a similar pattern. Compared with the plateau baseline, the proportion of HLA-DR⁺CD38^- CD8⁺ T lymphocytes (P < 0.0001) and HLA-DR⁺CD38^dim CD8⁺ T lymphocytes (P < 0.0001) increased significantly, while the proportion of HLA-DR⁺CD38^hi CD8⁺ T lymphocytes decreased markedly (P < 0.0001) after IFN intermittent treatment with 12-24 weeks. In comparison to IFN intermittent treatment, the proportion of HLA-DR⁺CD38^- CD4⁺ T lymphocytes and HLA-DR⁺CD38^dim CD4⁺ T lymphocytes decreased markedly after 12-24 weeks of IFN retreatment (P = 0.0010; P < 0.0001). However, compared with plateau baseline, there were no noteworthy variation in both the proportion of HLA-DR⁺CD38^- CD4⁺ T lymphocytes and HLA-DR⁺CD38^dim CD4⁺ T lymphocytes after IFN retreatment (P > 0.9999; P = 0.6840). After 12-24 weeks of IFN retreatment, HLA-DR⁺CD38^hi CD4⁺ T lymphocyte percentage was considerably more than that after IFN intermittent treatment (P = 0.0130), but dramatically lower than that at the plateau baseline (P = 0.0180). Likewise, HLA-DR⁺CD38^- and HLA-DR⁺CD38^dim subsets proportion on CD8⁺ T lymphocytes after 12-24 weeks of IFN retreatment was greatly lower than that after IFN intermittent treatment, respectively (P < 0.0001; P < 0.0001), but there was no discernible change compared with the plateau baseline (P > 0.9999; P > 0.9999). Furthermore, HLA-DR⁺CD38^hi CD8⁺ T lymphocyte proportion increased considerably following IFN retreatment compared to IFN intermittent treatment. (P = 0.0060), but considerably lower than that at plateau baseline (P = 0.0160) ( Table 3 ).

There was no significant variation in HBsAg level in Plateau group before and after IFN intermittent treatment (P > 0.9999). HBsAg level of each sample collected after IFN retreatment in Plateau group declined more than 0.5 lg IU/mL compared with that after IFN intermittent treatment, which was statistically noteworthy (P = 0.0030), and considerably lower than that at plateau baseline as well (P = 0.0060). During the IFN intermittent treatment and IFN retreatment, HBV DNA load in Plateau group had no significant change (0 week vs. 12-24 week IFN treatment interval P = 0.6950; 12-24 week IFN treatment interval vs. 12-24 week IFN retreatment P > 0.9999). In comparison to the baseline of plateau phase, ALT (P < 0.0001) and AST (P < 0.0001) levels in Plateau group decreased significantly after IFN intermittent treatment, while compared with those after IFN intermittent treatment, ALT (P < 0.0001) and AST levels(P < 0.0001) increased considerably in Plateau group after IFN retreatment, as shown in Table 3 .