Reduction of Hepatitis B Surface Antigen May Be More Significant in PEGylated Interferon-Alpha Therapy Combined with Nucleotide Analogues than Combined with Nucleoside Analogues in Chronic Hepatitis B Patients: A Propensity Score Matching Study

Background Nucleotide analogues (NTs) monotherapy may have a more significant effect on reducing hepatitis B surface antigen (HBsAg) than nucleoside analogues (NSs) due to their immunomodulatory function. However, this superiority remains unknown when combined with PEGylated interferon α (PegIFNα). Therefore, this study aimed to explore whether NTs have more significant antiviral effects than NSs in combination therapy with PegIFNα. Methods Chronic hepatitis B (CHB) patients treated with PegIFNα plus nucleos(t)ide analogues (NAs) were retrospectively recruited. Efficacy and the predictors of hepatitis B surface antigen (HBsAg) reduction >1 log10 IU/mL after 48 weeks were analyzed. Results A total of 95 patients were included and divided into the PegIFNα + NTs group and the PegIFNα + NSs group. Propensity score matching (PSM) was performed. The PegIFNα + NTs group had a greater reduction of HBsAg (−3.52 vs. −2.33 log10 IU/mL, P=0.032) and a higher proportion of patients with HBsAg reduction >1 log10 IU/mL (100.0% vs. 72.2%, P=0.003) even after PSM. However, HBsAg and hepatitis B e-antigen (HBeAg) loss rates, HBeAg seroconversion rates, degree of HBeAg and hepatitis B virus (HBV) DNA decline, HBV DNA undetectable rates, and alanine aminotransferase (ALT) normalization rates showed no significant differences. Subgroup analyses showed the difference in the reduction of HBsAg was particularly evident in HBeAg-positive and the “add-on” subgroups. PegIFNα plus NTs (OR = 36.667, 95% CI = 3.837–350.384) was an independent predictor for HBsAg reduction >1 log10 IU/mL after 48 weeks. Conclusion This study suggests that PegIFNα plus NTs may lead to more HBsAg reduction, especially in HBeAg-positive and “add-on” patients.


Introduction
Chronic hepatitis B (CHB) is a global infectious disease.
Tere are currently about 70 million people infected with hepatitis B virus (HBV) in China, with more than 20 million CHB patients. Tese patients are at high risks of liver cirrhosis and hepatocellular carcinoma (HCC), especially in developing countries [1], presenting an immense medical burden [2]. Covalently closed circular DNA (cccDNA) persistence within hepatocytes is relevant for chronic HBV infection [3]. Hepatitis B surface antigen (HBsAg) is a surrogate marker for cccDNA transcriptional activity [3][4][5]. Te disappearance of HBsAg, accompanied by a sustained virological response, loss of hepatitis B e-antigen (HBeAg), recovery of alanine aminotransferase (ALT), and improvement of liver tissue lesions, is defned as functional cure. Tus, important guidelines consider sustained HBsAg disappearance after drug withdrawal as an ideal treatment endpoint [6,7].
However, HBsAg loss is not common with current standard antiviral strategies, including nucleos(t)ide analogues (NAs) and PEGylated interferon-alpha (PegIFNα). Reduced HBsAg level is often associated with better outcomes, including minimizing cirrhosis and HCC, and is conducive to HBsAg clearance. Terefore, it is often used as an efcacy indicator. NAs are economical and convenient but cannot directly act on cccDNA. Patients usually need to take long-term or even life-long medications, bringing unavoidable economic and psychological burdens and drug resistance problems. In contrast, PegIFNα can reduce HBsAg more thoroughly in a subset of patients [8]. Te low virologic response rate in PegIFNα monotherapy and poor reduction of HBsAg in NAs monotherapy shed some light on combination strategies.
Previous studies have proven that PegIFNα combined with NAs had better clinical efects than those of PegIFNα or NAs monotherapy [9][10][11], particularly in reducing HBsAg [12] and enhancing HBsAg loss rate [13]. Additionally, NAs can vary in efcacy. Nucleotide analogues (NTs), including tenofovir disoproxil fumarate (TDF), adefovir dipivoxil (ADV), and tenofovir alafenamide (TAF), are not only structurally but also functionally diferent from nucleoside analogues (NSs) like entecavir (ETV) and lamivudine (LAM). According to a small randomized controlled trial, the reduction in HBsAg was signifcantly higher in the TDF arm than in the ETV arm in NAs-naive patients [14]. Furthermore, switching from ETV to TDF or TAF signifcantly declines HBsAg [15,16]. Interestingly, NTs have also been found with an additional immunological efect in interferon lambda 3 (IFNλ3) induction compared to NSs [17]. Meanwhile, in some studies, TDF treatment was associated with a signifcantly lower risk of HCC than ETV [18,19]. Still, the comparison remains controversial [20]. Te HBsAg clearance rate could reach 9.1% after 48 weeks of therapy combining PegIFNα and TDF followed by TDF monotherapy until 72 weeks [9]. But the rate was only 0.8% when PegIFNα was combined with ETV for 48 weeks and followed up to even 96 weeks [11]. According to this indirect comparison, PegIFNα combined with TDF (which represents NTs) appears to reach a better HBsAg clearance rate than that of PegIFNα combined with ETV (which represents NSs) when the treatment durations are similar. However, the populations and the end-points were not totally consistent between the two studies, making comparison difcult. Tere is currently no study directly comparing the efcacy of these two combination therapies.
Terefore, comparing HBsAg reduction efcacy for PegIFNα therapy combined with NTs or NSs in CHB patients is valuable. Tus, we conducted a retrospective study using the data of CHB patients treated with a combination of PegIFNα plus diferent NAs. October 2011 and December 2018,  a total of 159 consecutive PegIFNα-naive CHB patients  who received PegIFNα for at least 48 weeks and combined with NAs during the course were retrospectively enrolled from two clinical centers, Huashan Hospital of Fudan University (Shanghai, China) and Shanghai Public Health Center of Fudan University (Shanghai, China). Chronic HBV infection was defned as HBsAg-positive and/or HBV DNA-positive for at least six months before enrollment. Te combination therapy strategies could be "add-on" (adding NAs on during the therapy of PegIFNα) and "NAs-experienced" (adding PegIFNα to NAs treatment which had been more than one year). Te NAs used back then were kept consistent with the previous type. In total, 64 patients were excluded, with four having underlying chronic hepatitis C, autoimmune hepatitis, HIV, or tumor, seven having used PegIFNα for more than 48 weeks when NAs were added to the therapeutic regimen, one combining nucleoside analogues with nucleotide analogues, six using the combination therapy for less than 12 weeks, and 46 having a PegIFNα therapy duration less than 48 weeks or incomplete data at an important timepoint. In this study, 95 patients were ultimately included, with one group including those who received PegIFNα combined with nucleoside analogues (ETV) (n � 18), and the other group including patients treated with PegIFNα combined with nucleotide analogues (TDF or ADV) (n � 77). Tis retrospective study was conducted under the approval of the Ethics Committee for Huashan Hospital of Fudan University and following the Declaration of Helsinki. Written informed consent was obtained for all patients included.

Clinical Data.
All patients' baseline clinical data and laboratory test results were recorded. Clinical data included demographic data, history of chronic hepatitis B, and treatment history (name, dose, time, and medication complications). Laboratory test results included blood routine, liver function, and hepatitis B-related indicators (HBsAg tilters, HBeAg titers, and HBV DNA levels). Te baseline was defned as the start of PegIFNα therapy. Te duration of PegIFNα therapy was at least 48 weeks, with combination therapy for a minimum of 12 weeks. Laboratory examination results at 0, 12, 24, 36, and 48 weeks and the medication changes during the treatment (complications, dose changes, and addition or withdrawal of NAs) were recorded in detail.

Defnitions of Treatment
Response. Te primary endpoint was the reduction levels of HBsAg from the baseline at 48 weeks of treatment.
Serological responses after 48 weeks: (1) proportion of patients with HBsAg reduction >1 log 10 IU/mL from baseline; (2) HBsAg loss rate; (3) reduction levels of HBeAg from baseline at 48 weeks; (4) HBeAg loss rate and HBeAg seroconversion (HBeAg loss with the appearance of anti-HBe) rate. Virological responses after 48 weeks: (1) reduction of HBV DNA levels from baseline after 48 weeks; (2) HBV DNA undetectable rate (proportion of patients with DNA <500 IU/mL after 48 weeks according to the accuracy of the instrument at the time); (3) proportion of patients with HBsAg reduction >1 log 10 IU/mL from baseline, and HBV DNA were undetectable after 48 weeks. Biochemical response at 48 weeks was defned as ALT normalization rate (proportion of patients with baseline ALT >1 upper limit of normal (ULN) and normal ALT after 48 weeks, ULN � 40 U/ L).
HBsAg loss was defned as HBsAg <0.05 IU/mL. HBV DNA was measured using TaqMan fuorescence quantifcation, and the lower detection limit was 500 IU/mL. Routine biochemical and hematological tests were performed locally. Te normal upper limit of ALT was 40 IU/L. Data from laboratory assessments were collected at baseline and after 12, 24, 36, and 48 weeks of treatment.

Statistical Analysis.
HBsAg and HBV DNA levels and reduction levels were log (base 10) transformed.
Te Kolmogorov-Smirnov test was conducted for normality testing. Continuous variables are represented by the mean ± standard deviation (SD) and median (interquartile range (IQR)). Independent t-tests were used to compare continuous variables with normally distributed data (Z-score between ±1.96, calculated by skewness and kurtosis), while Mann-Whitney U tests were used to compare continuous variables with a skewed distribution. Te chi-squared test presented the categorical data as n (%). Diferences among groups were evaluated using one-way analysis of variance (ANOVA) if the variances were homogeneous, and the LSD-T test was used for intergroup comparison. Otherwise, the Kruskal-Wallis test (K-W test) for nonparametric statistics was conducted. Multivariate logistic regression analysis was applied to determine the predictors that afected HBsAg reduction >1 log 10 IU/mL from baseline at 48 weeks of treatment. To adjust for potential bias that could infuence the results, including sample size with excessive deviation, we applied a balanced study based on the propensity score matching (PSM) technique at a 1 : 1 ratio with a caliper of 0.2 separately between the PegIFNα + ETV group and the PegIFNα + ADV group or the PegIFNα + ETV group and the PegIFNα + TDF group. Age, HBsAg, and prior treatment duration of NAs combined with PegIFNα were imputed for PSM. Te balance of the variables between the groups was considered acceptable when the absolute value of the standard diference was less than 10%. Diferences were considered signifcant at a two-tailed P < 0.05. All statistical analyses were carried out using SPSS software version 24.0 (IBM, Armonk, NY, USA).

Baseline Characteristics.
A total of 95 cases were selected for efective analysis, including 18 patients who received a therapy combining PegIFNα with nucleoside analogues (PegIFNα + NSs) and 77 patients who received PegIFNα combined with nucleotide analogues (PegIFNα + NTs) ( Figure 1). In detail, the PegIFNα + NTs group included the PegIFNα + ETV group, the PegIFNα + NTs group included the PegIFNα + ADV group and the PegIFNα + TDF group. Tere were no signifcant diferences in baseline information between the two groups or among subgroups prior to PSM ( Table 1). PSM was performed, yielding 18 patients matched in each subgroup. After PSM, relative multivariate imbalance L1 was lower than before matching, indicating a better balance. No covariate exhibited a signifcant imbalance, and all the covariates reached a balance within 10%, so the balance of the variables between groups was considered acceptable after PSM. No statistically signifcant diferences were found among patients in each group after PSM (Table 1).

Serological Responses.
Before matching, the proportion of patients with an HBsAg reduction >1 log 10 IU/mL after 48 weeks of treatment was signifcantly higher in the PegIFNα + NTs group than in the PegIFNα + NSs group (98.7% vs. 72.2%, P � 0.001). Tis diference was still present after matching (100% vs. 72.2%, P � 0.003) ( Figure 3). Similarly, both the PegIFNα + ADV group and the PegIFNα + TDF group had a higher proportion of HBsAg reduction >1 log 10 IU/mL after 48 weeks than the PegIFNα + ETV group before and after PSM (Tables 2 and 3) ( Figure 3).

Biochemical Responses.
Te proportion of patients with elevated baseline ALT who returned to normal levels at 48 weeks difered between the two groups. However, the diference was not statistically signifcant. In all, 33 patients (43.4%) in the PegIFNα + NTs group and nine patients (52.9%) in the PegIFNα + NSs group achieved a biochemical response of serum ALT level <40 IU/L at the end of therapy before PSM (P � 0.476) ( Table 2). After matching, 15 patients (42.9%) in the PegIFNα + NTs and nine patients (52.9%) in the PegIFNα + NSs group had biochemical responses, respectively (P � 0.494) (Figure 3). Biochemical responses did not vary substantially by subgroups (Tables 2  and 3).

Subgroup Analyses.
Patients were divided into subgroups based on HBeAg status or combination strategies. No signifcant diferences were found at baseline among patients treated with PegIFNα plus diferent oral drugs in HBeAgpositive, HBeAg-negative, or "add-on" patients (Tables S1-S3). Patients treated with PegIFNα + NSs had a lower baseline HBV DNA level in the "NAs-experienced" subgroup (Table S4). We found that there were more "addon" patients in the HBeAg-positive subgroup (66.7% vs. 17.6%, P � 0.001), and the ALT level was also higher (104 vs. 34 U/L, P � 0.001) than in the HBeAg-negative subgroup (Table S5). Patients in the "NAs-experienced" subgroup had a longer duration of antiviral therapy before adding on PegIFNα. Terefore, the levels of HBsAg, HBeAg, and HBV DNA were lower than the "add-on" subgroup at baseline (Table S6).
In the HBeAg-positive subgroup, the reduction of HBsAg (−3.62 vs. −2.43 log 10 IU/mL, P � 0.002) was signifcantly more and the proportion of patients with HBsAg reduction >1 log 10 IU/mL after 48 weeks was signifcantly higher (100.0% vs. 69.2%, P � 0.001) in the PegIFNα + NTs group (Table S7). Antiviral efects in HBeAg-negative patients seemed to have no signifcant diferences between the PegIFNα + NTs group and the PegIFNα + NSs group, although the sample size was too small for meaningful statistical analysis (Table S8). In the "add-on" subgroup, the reduction of HBsAg was signifcant more in the PegIFNα + NTs group than in the PegIFNα + NSs group (−3.89 vs. −2.27 log 10 IU/mL, P � 0.002). HBsAg reduction was signifcantly more both in the PegIFNα + TDF group (−3.85 vs. −2.27 log 10 IU/mL, P � 0.008) and in the PegIFNα + ADV group (−3.91 vs. −2.27 log 10 IU/mL, P � 0.003) than in the PegIFNα + ETV group. Te proportion of patients who achieved HBsAg reduction >1 log 10 IU/mL was higher in the PegIFNα + NTs group than in the PegIFNα + NSs group (100.0% vs. 75.0%, P � 0.019) (Table S9). In the "NAs-experienced" subgroup, no significant diferences in the reduction of the HBsAg after 48 weeks were observed between the PegIFNα + NTs group and the PegIFNα + NSs group. However, the proportion of patients with an HBsAg reduction >1 log 10 IU/mL at 48 weeks of treatment was signifcantly higher in the PegIFNα + NTs group than in the PegIFNα + NSs group (96.7% vs. 70.0%, P � 0.042), suggesting a signifcant difference in antiviral efcacy (Table S10). Because all the patients in the PegIFNα + NSs group had undetectable viral loads at baseline, so the HBV DNA level did not drop with treatment. Terefore, HBV DNA reduction levels were not comparable between two groups. Subgroup analyses were performed using data before PSM, given our sample size.

Predictors Associated with HBsAg
Reduction >1 log 10 IU/ mL at 48 Weeks. All patients were divided into two groups according to whether or not they achieved HBsAg reduction >1 log 10 IU/mL after 48 weeks. Univariate analysis was performed to analyze the efect of clinical data and laboratory tests. Factors with a P value <0.1 or clinical signifcance were included in multivariate logistic regression analysis (forward: conditional method). As a result, we found that treatment with PegIFNα plus NTs (OR � 36.667, 95% CI � 33.837-350.384) ( Table 4) was an independent predictor contributing to HBsAg reduction >1 log 10 IU/mL at 48 weeks.

Discussion
To date, PegIFNα and NAs are important clinical frst-line anti-HBV drugs with diferent mechanisms and efects on innate and adaptive immunity. NAs are oral direct antiviral drugs that reduce the viral load by inhibiting HBV DNA polymerase and reverse transcriptase. At the same time, they cannot directly inhibit the transcriptional activity of cccDNA. Terefore, obtaining durable immunological control is difcult, and the clearance and seroconversion of HBsAg and HBeAg are not easily achievable. As a result, long-term medication is often required. PegIFNα can enhance innate immunity, trigger T cell-mediated immune responses, prevent HBV protein formation, and deplete the cccDNA pool [21], resulting in superior efectiveness to NAs in reducing HBsAg [8]. Nearly one-third of PegIFNα responders achieve HBsAg clearance. In addition, strong inhibition of viral replication by NAs can assist PegIFNα's immunomodulatory efect [22]. Hence, a combination strategy with PegIFNα plus NAs is theoretically feasible and an inevitable trend for future development. However, before a new generation of efective drugs is introduced and popularized, exploration of the combination strategy has become a major focus of current research.
Tere have been several studies on the efcacy of combination therapy, among which many have shown combination therapy to be superior to monotherapy in reducing HBsAg levels [9,23,24] and found that combination therapy could even signifcantly increase HBsAg loss   [24]. Furthermore, compared with NAs monotherapy, combination therapy resulted in a higher percentage of HBeAg loss (26% vs. 13%, at 96 weeks) [21] and a higher HBeAg seroconversion rate (15% vs. 5%, at 48 weeks) [25] as well. Terefore, it is evident that combination therapy has prominent advantages over monotherapy. However, combination therapy's baseline conditions, optimal treatment duration, and sustained response rate require further exploration. At the same time, it is unclear whether efcacy difers between nucleotide analogues and nucleoside analogues when combined with PegIFNα. Te two oral drugs are functionally diferent, especially in HBsAg reduction. Koike et al. found that TDF reduced signifcantly more HBsAg levels at week 24 (-0.147 vs. -0.027 log 10 IU/mL, P < 0.05) and 48 (-0.208 vs. -0.051 log 10 IU/mL, P < 0.05) in NAsnaive patients [14]. Furthermore, HBeAg-negative patients whose HBsAg had not been reduced during a 48-week ETV treatment had a signifcantly higher HBsAg reduction after switching to TDF or TAF than in the ETV continuation group [15]. HBV infection is a risk factor for hepatocarcinogenesis. Nevertheless controversial, previous research has shown that TDF treatment could be associated with a lower risk of HCC than ETV treatment. A large retrospective analysis in China found that over a median follow-up time of 3.6 years, 4.9% of ETV-treated patients developed HCC, while it occurred in only 0.6% of TDFtreated patients [19]. Similarly, a study in Korea consistently found that the annual incidence rate of HCC was signifcantly lower in the TDF group than in the ETV group (0.64 vs. 1.06 per 100 person-year) [18]. Notably, researchers have indicated that patients treated with nucleotide analogues, especially ADV, have higher serum IFNλ3 levels than those treated with nucleoside analogues [26,27]. Te ability of IFNλ3 to induce interferon-stimulated genes (ISGs) in Huh7 cell lines is stronger than that of interferon lambda 1/2 (IFN-λ1/λ2), and this ability is weaker but longer-lasting than that of IFNα [26]. ISGs can encode antiviral proteins via complex intracellular signaling pathways, implying that IFNλ3 may be more efective against viruses than IFNα. Recombinant IFNλ3 had been shown to reduce HBsAg levels in vitro and had an additive antiviral efect with IFNα [17], further regulating the secretion of cytokines and enhancing antiviral immune function [28]. Hence, we supposed that a combination of PegIFNα with nucleotide analogues could have a better efect on reducing HBsAg levels than with nucleoside analogues. According to Ahn et al., the HBsAg clearance rate could reach 9.1% after 48 weeks of therapy combining PegIFNα and TDF, followed by TDF monotherapy until 72 weeks. No patient achieved HBsAg clearance in the TDF monotherapy group [9]. Liem et al. found that when PegIFNα was combined with ETV for 48 weeks and followed up for 96 weeks, only 0.8% of patients achieved HBsAg loss. No patients in the ETV monotherapy group achieved HBsAg clearance [11]. On the contrary, there are meta-analyses showing that the diferences in HBsAg loss rates at the end of the combination therapy are not statistically signifcant among diferent NAs (ETV 11% vs. ADV 12% vs. LAM 9% vs. TDF 6%, P > 0.05) and have found similar results for the HBsAg seroconversion rate (5% vs. 5% vs. 9% vs. 4%, P > 0.05) [29]. A prospective follow-up study found HBsAg loss occurred similarly in PegIFN + ADV (18.6%) and PegIFN + TDF (11.7%) patients up to fve years after the end of a 48-week combination therapy. Tis study, however, did not provide the result when PegIFN combined with NSs [30]. Lin et al. recently found that the addition of TDF to Peg-IFNα-2b in HBeAg-positive CHB patients with a poor response after 12 weeks of Peg-IFNα-2b monotherapy reduced HBsAg signifcantly more than the addition of ETV to Peg-IFNα-2b (−1.799 log 10 IU/mL vs. −1.078 log 10 IU/mL, P � 0.0491) [31]. It was an important result as it compared the addition of TDF or ETV to Peg-IFNα-2b directly. However, considering the small sample size and the restrictive conditions for the selected population, it lacks universality, and a larger sample size study is required to verify the results. Terefore, whether PegIFNα combined with diferent NAs infuences HBsAg reduction and clearance is still unclear. Te loss rate of HBeAg after 48 weeks was similar between PegIFNα + TDF and PegIFNα + ETV (29.0% vs. 31.0%) [32]. Recent data from another study pointed out that PegIFNα combined with TDF could improve HBeAg responses in a short time. No advantages were found when PegIFNα was combined with LAM or ETV [33]. However, Lin et al. showed that the HBeAg loss rate was signifcantly higher in the TDF add-on group than in the ETV add-on group after 48 weeks (40% vs. 10%, P � 0.028) [31]. Interestingly, these studies suggested that PegIFNα combined with diferent NAs could have diferent efcacies, but direct evidence was required, and the mechanism underlying the diferences must be discussed. We conducted this retrospective study to provide this evidence based on these fndings. TAF has only been launched in recent years, and with insufcient studies discussing the efcacy of PegIFNα plus TAF, we did not include patients who received TAF in the current study. Meanwhile, patients in our cohort who used LAM were excluded according to the exclusion criteria, so ETV was the only nucleoside analogues analyzed.
To our knowledge, this study was the frst to retrospectively compare HBsAg level reduction efcacy for CHB patients treated with diferent NAs in PegIFNα combination therapy, no matter which combination strategy was adopted. Tis could help prove that the diference in reduction was due to the types of NAs. In order to minimize the impact of bias, PSM was performed to eliminate the inequality caused by excessive deviation of the general data and sample size. After PSM, the results showed that the HBsAg levels of the PegIFNα + NSs group decreased by an average of −2.33 log 10 IU/mL from baseline at 48 weeks, while it decreased signifcantly more in the PegIFNα + NTs group, by an average of −3.52 log 10 IU/ mL (P � 0.032). Te reductions of HBsAg in both groups were more than those in Lin et al.' study [31]. Tis could be because our study had a longer combination course and some patients had previously received NA treatment. Te proportion of patients achieving HBsAg reduction >1 log 10 IU/mL was signifcantly higher at 48 weeks in the PegIFNα + NTs group compared to the PegIFNα + NSs group (100% vs. 72.2%, P � 0.003). However, even after PSM adjustment, no signifcant diferences in the following indicators were found between the two groups: HBsAg loss rate, HBV DNA reduction levels, HBeAg reduction levels, HBeAg loss rate, HBeAg seroconversion rate, HBV DNA undetectable rate, and ALT normalization rate. Te observation endpoint of this study was the 48th week of treatment, and subsequent follow-up had not yet been carried out, resulting in difculty achieving HBsAg clearance, especially for antiviral treatment-naive patients. Te ability to maintain steady HBsAg clearance after combination therapy cannot be confrmed. Another reason for the signifcant diferences in decline levels, but not in HBsAg loss rates, maybe the small sample size. Based on the results of our study, we believe that NTs may signifcantly reduce more HBsAg than NSs when combined with PegIFNα. Tis reduction will contribute to achieving HBsAg clearance and even a functional cure. In our study, the proportion of patients who simultaneously reached HBV DNA below the lower detection limit and HBsAg reduction >1 log 10 IU/mL from baseline at 48 weeks difered between the PegIFNα + ETV group and the PegIFNα + TDF group after PSM (100.0% vs. 72.2%, P � 0.045). Tis result exemplifes the dual efectiveness of PegIFNα combination therapy with TDF over combination therapy with ETV in inhibiting viral replication and reducing HBsAg levels simultaneously. Furthermore, the multivariate logistic regression also showed that treatment with PegIFNα plus NTs was an independent predictor for HBsAg decline >1 log 10 IU/mL at 48 weeks, suggesting that the combination of PegIFNα and NTs can fasten HBsAg decline. Combination strategies have been studied, including "De novo," "NAs-experienced," "add-on," and "switch-to." Several studies have shown that the "NAs-experienced" strategy seemed the best. Te "switch-to" strategy was particularly efective and improved the HBsAg clearance rate [13,29,34]. Tis maybe because the direct antiviral activity of NAs can lead to virological suppression, which can further improve the immunomodulatory efect of PegIFNα, thereby maximizing the advantages of combination therapy. Since patients with diferent combination strategies and HBeAg status were enrolled, subgroup analyses were performed using data before PSM to determine whether the antiviral efects were diferent between "add-on" and "NAsexperienced" subgroups as well as between HBeAgpositive and HBeAg-negative subgroups. We found that in HBeAg-positive patients, the reduction of HBsAg was signifcantly more in the PegIFNα + NTs group. Possible mechanism may be the additional immunomodulatory effects of NTs combined with PegIFNα, which deregulate the immunosuppression caused by HBeAg [35,36], resulting in a better clinical efcacy. However, further research is needed to investigate the diferent efects of NTs and NSs on the immune system. Regrettably, the number of HBeAgnegative patients was relatively small and was prone to bias. Terefore, no statistical analysis of this subpopulation was conducted, and further studies are warranted to confrm our fndings. More signifcant reduction of HBsAg in the PegIFNα + NTs group was also observed in the "add-on" subgroup. Besides, patients who achieved a reduction in HBsAg >1 log 10 IU/m were signifcantly more in the PegIFNα + NTs group in HBeAg-positive, "add-on," and "NAs-experienced" patients. We, therefore, infer that our fndings were generally consistent across subgroups.
Limitations of our study include that it is a retrospective study with small sample size and short therapy duration without a long-term follow-up. Furthermore, the combination strategy was not precisely uniform, although the combination therapy duration was guaranteed at least 24 weeks. However, the prior treatment duration and drugs before combination for NAs-experienced patients, the weeks of adding-on NAs for "add-on" patients, and the total weeks of combination at baseline before and after PSM were not statistically diferent. In addition, the results were partially observable in subgroup analyses. So, the following analysis was considered reliable. However, further randomized controlled trials are required for verifcation.

Conclusion
In conclusion, reduction of HBsAg might be more pronounced in PegIFNα therapy combined with NTs than NSs, especially in HBeAg-positive patients and patients using "add-on" strategies. Tis fnding will be benefcial for promoting further HBsAg clearance and functional cure. In addition, this fnding can be used to make clinical decisions. Terefore, similar fndings and mechanisms should be investigated further [37].

Consent
Informed consent was obtained from all patients.