HBsAg decline and clearance with peg-IFN therapy added to nucleos(t)ide analogues: an individual participant data meta-analysis of prospective trials (PROSPER)

Abstract

Background

Peg-interferon (peg-IFN) plays an increasingly important role in HBV cure strategies, either in combination with novel antivirals, as a lead-in or as consolidation treatment.

Objective

We aimed to provide estimates of hepatitis B surface antigen (HBsAg) decline and clearance that can be achieved with peg-IFN addition to nucleos(t)ide analogue (NA) therapy.

Design

This is a post hoc meta-analysis of individual participant data from eight clinical trials involving chronic hepatitis B patients on NA therapy who received peg-IFN add-on. The primary endpoint was HBsAg loss at end of follow-up (EOF, 6–12 months after end of peg-IFN). Secondary analyses focused on HBsAg decline.

Results

581 patients were included. At the start of peg-IFN therapy (SOT), 44% were hepatitis B envelope antigen (HBeAg) positive, mean HBsAg level was 3.03 log10 IU/mL (HBsAg<100: 12%; 100–1000: 28%; ≥1000: 60%), and planned duration of peg-IFN was 48 weeks in 496 patients (85%).

At EOF, 50 (8.6%) patients achieved HBsAg loss (HBsAg<100/100–1000/≥1000: 37.7/9.8/2.3%, p<0.001) Findings were consistent across ethnicities (Caucasian: 30.0/8.7/3.6%; Asian: 39.3/9.2/2.2%). In patients with SOT HBsAg≥1000 IU/mL, levels <1000 and <100 were achieved in 29.7% and 8.9% at 24 weeks and in 47.5% and 16.3% at 48 weeks of peg-IFN therapy, respectively.

Conclusion

Peg-IFN add-on results in HBsAg loss in 18% of patients with SOT HBsAg<1000 IU/mL, and in 38% if SOT HBsAg<100 IU/mL. Among patients with higher HBsAg levels, peg-IFN could be used to reduce HBsAg to below thresholds associated with response to novel compounds.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Previous (smaller) studies support the use of peg-interferon (IFN) to increase hepatitis B surface antigen (HBsAg) decline and clearance in patients on nucleos(t)ide analogue (NA) therapy, but robust estimates of response rates are lacking.

WHAT THIS STUDY ADDS

• This study provides benchmark estimates for HBsAg decline and clearance that can be achieved with peg-IFN add-on to NA therapy, based on individual participant data from eight prospective studies.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These data are important for clinicians and can be used as reference and benchmark for the design of trials aiming for HBV cure.

Introduction

Chronic hepatitis B (CHB) remains a major public health challenge with an estimated 254 million people infected worldwide, and over 1.3 million attributable deaths per year.¹ The current backbone of treatment is long-term viral suppression with nucleos(t)ide analogues (NAs), which reduces the risk of developing cirrhosis, hepatocellular carcinoma (HCC), hepatic decompensation and liver-related death.^{2 3} Nevertheless, a risk of adverse liver-related outcomes remains despite HBV DNA suppression, and this risk can be mitigated by achieving HBsAg loss.⁴ Unfortunately, hepatitis B surface antigen (HBsAg) loss is only rarely achieved with prolonged NA therapy, and alternative strategies are therefore being explored to increase HBsAg loss rates.

Examples of these novel therapies are RNA interference (RNAi) therapeutics, such as antisense oligonucleotides and small interfering RNAs (siRNAs), which have shown to increase HBsAg loss rates compared with NA monotherapy.^5–8 However, their effectiveness in achieving HBsAg loss is often limited to patients with HBsAg levels below 1000 IU/mL at start of therapy, and the durability of off-treatment responses is limited due to the occurrence of post-treatment HBsAg seroreversion. Based on results obtained with previous studies, it appears that achieving off-treatment sustained HBsAg loss may require regimens comprising antivirals and immunomodulators.^{4 7}

These observations have resulted in a regained interest in the use of peg-interferon (IFN) in novel cure regimens, either as part of a combination treatment regime to boost responses or as a lead-in therapy to reduce viral antigen levels before novel antivirals are introduced. Peg-IFN has been used as a treatment for CHB for several decades and has been shown to not only activate the host immune response by interacting with natural killer cells, T- and B cells, but also to directly inhibit the replication, synergistically reducing viral antigen load.⁹ Although previous small studies support the use of peg-IFN to increase HBsAg decline and clearance in patients on NA therapy, robust estimates of response rates are lacking. Moreover, recent studies focusing on novel targets often lack a peg-IFN add-on control arm, mainly because of the wide variety of side effects linked to peg-IFN. Nevertheless, robust estimates of response rates that can be obtained with peg-IFN add-on to NA therapy are required to accurately quantify the benefits of novel antivirals if these are combined with IFN or other immunomodulators.

Therefore, the objective of this study was to perform an individual participant data meta-analysis of prospective clinical trials of peg-IFN add-on therapy to provide robust estimates of response rates that could be used as a reference and benchmark for future studies aiming to achieve HBsAg loss, and to identify predictors of response to guide patient selection and treatment duration.

Methods

Study design

This is an individual participant data meta-analysis of eight prospective studies of peg-IFN add-on to NA therapy.^10–17 Prospective studies were considered eligible if they met our inclusion criteria: (1) addition of peg-IFN to an NA therapy backbone, (2) duration of peg-IFN add-on for at least 24 weeks, (3) at least 24 weeks of follow-up (FU) after cessation of peg-IFN, and (4) cohort size of at least 30 patients (online supplemental figure S1). Patients were treated with either peg-IFN-α−2a or peg-IFN-α−2b dosed 180 µg/week10-15^{10–15 17} or 1.5 µg/kg/week.¹⁶ More details about the studies included in this analysis can be found in the original publications and are summarised in online supplemental table S1. Since all included studies^10–17 obtained ethical approval when conducted, including reuse of clinical data, no additional ethical approval was needed.

Patient and public involvement

Patients were not involved in the design of the current study.

Data acquisition

A standardised case report form was used for data capture, requesting individual patient data from the different studies. This form included baseline characteristics, laboratory data at various time points (ie, week 0, 12, 24, 48, 72 and 96) and patient outcomes, including HBsAg loss. All data cleaning, quality assessments and analyses were conducted in Rotterdam, the Netherlands. As described in the respective studies, various assays were used for HBsAg quantification (eg, Elecsys HBsAgII QuantII (Roche Diagnostics, Rotkreuz, Switzerland), and Architect (Abbott Laboratories, North Chicago, Illinois, USA)) and HBV DNA quantification (eg, Cobas TaqMan (Roche Diagnostics, Basel, Switzerland), Abbott RealTime HBV m2000 (Abbott Molecular, Illinois, USA) and ABI 7500 Real-Time PCR System (Applied Biosystems, Foster City, California, USA)). The lower limit of detection for quantitative HBsAg was 0.05 IU/mL, while for quantitative HBV DNA, it was either 10, 20 or 50 IU/mL, depending on the assay used.

Study endpoints

The primary endpoint was the proportion of patients with HBsAg loss (<0.05 IU/mL) at end of follow-up (EOF; 24–48 weeks after cessation of peg-IFN, depending on the study design). The secondary endpoint was HBsAg decline during peg-IFN therapy.

Statistical methods

Continuous variables were reported as mean with SD or median with IQR and compared by using Student’s t-test or Mann-Whitney U test. Categorical variables were shown as absolute number and frequencies (%) and compared with Pearson χ² statistics, Fisher’s exact test or McNemar’s test. Diagnostic accuracy was assessed using measures of discrimination (area under the receiver operating characteristic (AUROC) curve), and sensitivity, specificity, positive predictive values (PPV) and negative predictive values (NPV) for the outcome of interest. Baseline was defined as start of peg-IFN therapy (SOT).

We assessed the proportion of patients achieving HBsAg loss in the overall population, across SOT HBsAg strata (<100, 100–1000 and ≥1000 IU/mL) and different ethnicities (Caucasian vs Asian). Next, a multivariable logistic regression model was constructed based on factors associated with HBsAg loss in univariable analysis with a p value<0.10.

Furthermore, on-treatment HBsAg kinetics were assessed in the overall population and stratified by SOT HBsAg levels. Additionally, we studied the proportion of patients with SOT HBsAg≥1000 IU/mL that achieved a decline to <1000 IU/mL and <100 IU/mL after 24 and 48 weeks of peg-IFN therapy in the overall population, and across different SOT HBsAg strata (1000–2000, 2000–3000 and ≥3000 IU/mL) in patients allocated to 48 weeks of peg-IFN therapy.

Lastly, we assessed whether HBsAg levels at on-treatment week 12 could be used to identify patients eligible for cessation of peg-IFN therapy for futility. For this analysis, we performed a grid search of cut-off points aiming for a NPV of >95% based on HBsAg decline from baseline (no decline/0.25/0.50/0.75/1.00 log10 decline) or absolute HBsAg levels (10/100/1000 IU/mL). After identification of suitable cut-offs, performance was assessed across SOT HBsAg strata.

All analyses were performed using SPSS V.28.0 (SPSS, Chicago, Illinois, USA)¹⁸ and R V.4.2.3 (R Core Team 2022, Vienna, Austria).¹⁹ All statistical tests were two-sided and a p value below 0.05 was used for statistical significance.

Results

Baseline characteristics of included patients

In total, 581 patients were included, of whom 471 (81%) patients were male, 420 (72%) were Asian and 253 (44%) were hepatitis B envelope antigen (HBeAg) positive. Mean age was 43 (±11) years, mean SOT HBsAg was 3.03 (±0.94) log10 IU/mL (HBsAg<100 IU/mL: 69 (12%); 100–1000: 163 (28%); ≥1000: 342 (60%)), mean HBV DNA was 1.58 (±0.78) log10 IU/mL, with 517 (89%) patients having HBV DNA<100 IU/mL and median alanine aminotransferase (ALT) was 0.77 (0.57–1.09) × upper limit of normal (table 1). Peg-IFN add-on was scheduled for 48 weeks in 496 (85%) patients and 374 (64%) patients had 48 weeks of FU after peg-IFN cessation. 57 (10%) patients discontinued peg-IFN prior to the planned end-date, and these rates varied between 0% and 21% in the included studies (online supplemental table S1).

Table 1. Baseline characteristics of included patients.

	Total number patients (n=581)
Male sex, n (%)	471 (81.1)
Age, mean (±SD)	43.1 (±11.3)
Ethnicity, n (%)
Caucasian	121 (20.8)
Asian	420 (72.3)
Black	25 (4.3)
Other	15 (2.6)
Cirrhosis at baseline, n (%)	37 (6.4)
Type of peg-IFN, n (%)
Peg-IFN-alfa-2a	265 (45.6)
Peg-IFN-alfa-2b	316 (54.4)
Type of NA, n (%)
Entecavir	371 (64.1)
Tenofovir (TDF/TAF)	139 (24.0)
Lamivudine	10 (1.7)
Adefovir	8 (1.4)
Other*	51 (8.8)
Laboratory results at start of peg-IFN therapy
HBeAg, n (%)
Negative	328 (56.5)
Positive	253 (43.5)
ALT × ULN†, median (IQR)	0.77 (0.57–1.09)
HBV DNA (log10 IU/mL), mean (±SD)	1.58 (±0.78)
HBV DNA (IU/mL), median (IQR)	TND (TND-TND)
HBV DNA<2000 IU/mL, n (%)	550 (94.7)
HBV DNA<100 IU/mL, n (%)	517 (89.3)
HBsAg (log10 IU/mL), mean (±SD)	3.03 (±0.94)
HBsAg (IU/mL), n (%)
<100	69 (12.0)
100–1000	163 (28.4)
≥1000	342 (59.6)

^*For example, multiple NA’s at the same time.

^†ULN was adjusted for sex (female: 25, male: 35)

ALT, alanine aminotransferase; HBeAg, hepatitis B envelope antigen; HBsAg, hepatitis B surface antigen; NA, nucleos(t)ide analogues; peg-IFN, pegylated interferon; TAF, tenofovir alafenamide; TDF, tenofovir disoproxil; TND, target not detected; ULN, upper limit of normal.

HBsAg loss at end of follow-up

At EOF, 50 (8.6%) patients achieved HBsAg loss. Lower SOT HBsAg levels were associated with a higher probability of HBsAg loss; HBsAg loss was observed in 37.7% for patients with SOT HBsAg<100 IU/mL, 9.8% for 100–1000 IU/mL and 2.3% for ≥1000 IU/mL (p<0.001, figure 1A). HBsAg loss was achieved in 18.1% of patients with SOT HBsAg<1000 and in 11.8% of patients with SOT HBsAg<3000 IU/mL (online supplemental table S2). Lower HBsAg levels were consistently associated with higher rates of HBsAg loss regardless of patient ethnicity (Caucasian: 30.0/8.7/3.6%; Asian: 39.3/9.2/2.2%, figure 1B).

Figure 1. HBsAg loss rates at end of follow-up after peg-IFN therapy, categorised by HBsAg levels at start of therapy (A) overall (B) per ethnicity. EOF, end of follow-up; HBsAg, hepatitis B surface antigen; IFN, interferon.

In a multivariable logistic regression model, adjusting for HBeAg status and viral load at start of therapy, lower SOT HBsAg was the main predictor for HBsAg loss (adjusted odds ratio (aOR) 0.28; 95% CI 0.20 to 0.39, p<0.001; table 2). HBeAg status was not associated with HBsAg loss after adjusting for SOT HBsAg (aOR 1.19; 95% CI 0.51 to 2.76, p=0.68). In univariable analyses, ethnicity, age or ALT at baseline were not associated with HBsAg loss (p>0.05). In several sensitivity analyses, baseline HBsAg remained the main predictor for HBsAg loss, for example, when response was assessed at 24 weeks after peg-IFN cessation (aOR 0.25; 95% CI 0.17 to 0.35, p<0.001), when only patients who received 48 weeks of peg-IFN were included (aOR 0.29; 95% CI 0.21 to 0.40, p<0.001) and when analyses were limited to SOT HBeAg negative patients (aOR 0.30; 95% CI 0.21 to 0.43, p<0.001).

Table 2. Univariable and multivariable logistic regression models for HBsAg loss at end of follow-up.

	Univariate			Multivariate
	OR	95% CI	P value	aOR	95% CI	P value
Male sex	2.218	0.859 to 5.726	0.100
Age	1.008	0.982 to 1.034	0.556
Race
Caucasian	Reference	Reference	Reference
Asian	1.446	0.657 to 3.183	0.360
Other	1.145	0.289 to 4.543	0.847
Cirrhosis at baseline	0.933	0.276 to 3.153	0.911
Start of therapy
HBeAg positive	0.380	0.194 to 0.743	0.005	1.190	0.514 to 2.755	0.684
HBsAg (log10 IU/mL)	0.269	0.196 to 0.370	<0.001	0.279	0.201 to 0.387	<0.001
ALT (log10)	1.032	0.273 to 3.907	0.962
HBV DNA (log10 IU/mL)	0.117	0.014 to 1.011	0.051	0.230	0.018 to 3.010	0.262

ALT, alanine aminotransferase; (a)OR, (adjusted) Odds Ratio; HBeAg, hepatitis B envelope antigen; HBsAg, hepatitis B surface antigen.

HBsAg kinetics during peg-IFN add-on therapy

The HBsAg kinetics during peg-IFN add-on therapy are shown in figure 2 and online supplemental figure S2. Mean HBsAg decline after 24 and 48 weeks of peg-IFN was 0.61 (±0.96) and 0.95 (±1.18) log10 IU/mL. Among patients with HBsAg≥1000 IU/mL at SOT, HBsAg levels <1000 IU/mL and <100 IU/mL were achieved in 29.7% and 8.9% after 24 weeks of therapy, increasing to 47.5% and 16.3% after 48 weeks of peg-IFN therapy, respectively (figure 3).

Figure 2. HBsAg kinetics during peg-IFN therapy, categorised by baseline HBsAg, black dots = mean, vertical bar-lines = 95% Confidence Interval (CI), Continuous black line = <100, Dotted black line = 100-1000 IU/mL, Dashed black line = ≥1000 IU/mL. HBsAg, hepatitis B surface antigen; IFN, interferon.

Figure 3. Proportion of patients with start of treatment HBsAg ≥1000 IU/mL that achieved HBsAg (A) <1000 IU/mL (B) <100 IU/mL after 24 and 48 weeks of peg-IFN therapy, categorised by HBsAg levels at start of therapy (SOT). HBsAg, hepatitis B surface antigen; IFN, interferon.

In paired analyses including patients allocated to 48 weeks of peg-IFN add-on with baseline HBsAg ≥1000 IU/mL, and available HBsAg measurements at week 24 and 48 (n=254), 48 weeks of peg-IFN was associated with a higher probability of achieving HBsAg<1000 IU/mL for patients with SOT 1000–2000 IU/mL (79.5% vs 65.4%, p<0.001), SOT HBsAg 2000–3000 IU/mL (48.6% vs 28.6%, p=0.003) and SOT HBsAg≥3000 IU/mL (25.5% vs 14.2%, p=0.008), figure 3A. Similarly, 48 weeks of peg-IFN was associated with a higher probability of achieving HBsAg<100 IU/mL for patients with SOT 1000–2000 IU/mL, but not in patients with higher SOT HBsAg levels in whom overall response rates were very low, figure 3B.

Prediction of HBsAg response using on-treatment HBsAg levels at week 12

At on-treatment week 12 (n=558 (96%)), both absolute HBsAg levels and HBsAg decline from baseline were associated with (failure to achieve) HBsAg loss. Less pronounced HBsAg decline was associated with a lower probability of achieving HBsAg loss (table 3). For instance, failure to achieve at least 0.5 log decline from baseline to week 12 was associated with a probability of HBsAg loss of only 3.0% in the overall population. Similarly, higher HBsAg levels at on-treatment week 12 were associated with a lower likelihood of achieving HBsAg loss. For example, in patients with an HBsAg≥100 IU/mL at on-treatment week 12, the probability of HBsAg loss during subsequent FU was only 1.5%.

Table 3. Predictive accuracy for HBsAg decline after 12 weeks of peg-IFN therapy and HBsAg loss at end of follow-up.

Week 12 stop criteria	Overall cohort (n=558)*				SOT HBsAg<1000 IU/mL (n=224)
HBsAg decline (log10) between week 0 and week 12	Percentage meeting stop criteria (%)	NPV	Percentage not meeting stop criteria (%)	PPV	Percentage meeting stop criteria (%)	NPV	Percentage not meeting stop criteria (%)	PPV
No decline	30.9	97.1	69.1	11.2	29.9	92.5	70.1	22.3
<0.25 log decline	71.8	97.2	28.2	23.6	63.4	93.0	36.6	36.6
<0.50 log decline	83.3	97.0	16.7	36.6	74.6	92.2	25.4	47.4
<0.75 log decline	87.8	95.5	12.2	38.2	80.4	88.9	19.6	45.5
<1.00 log decline	91.7	95.3	8.3	52.2	84.4	88.4	15.6	51.4

HBsAg levels (IU/mL) at week 12
≥10 IU/mL	91.0	95.9	9.0	54.0	79.0	92.1	21.0	55.3
≥100 IU/mL	81.5	98.5	18.5	39.8	57.1	96.9	42.9	37.5
≥1000 IU/mL	51.4	100	48.6	17.7	5.8	100	94.2	19.0

^*Baseline HBsAg was missing from two patients with available HBsAg levels at week 12, therefore HBsAg decline could not be calculated.

HBsAg, hepatitis B surface antigen; IFN, interferon; NPV, negative predictive value; PPV, positive predictive value; SOT, start of peg-IFN therapy.

Discussion

In this post hoc meta-analysis of individual participant data of 581 CHB patients treated with peg-IFN add-on therapy, we observed HBsAg loss in 9% of patients. Patients with lower SOT HBsAg levels were more likely to respond to treatment; HBsAg loss was observed in 38% of patients with SOT HBsAg<100 IU/mL. Among patients with SOT HBsAg≥1000 IU/mL, HBsAg loss rates were negligible, but peg-IFN add-on therapy successfully decreased HBsAg levels to below 1000 U/mL in 48% of patients. These findings provide important guidance for the use of peg-IFN add-on therapy in current clinical practice and can be used as a reference and benchmark for clinical studies evaluating HBsAg loss rates with novel compounds.

Although HBV DNA suppression with long-term NA therapy offers major clinical benefits through improvement of liver fibrosis and reduction in the risk of HCC, a persistent risk of adverse liver-related outcomes remains if HBsAg loss is not achieved. Given the higher rate of HBsAg loss achieved with de novo peg-IFN when compared with NA monotherapy, various studies have assessed the efficacy of adding peg-IFN to NA therapy to decrease HBsAg and potentially enhance HBsAg loss. These individual studies have yielded conflicting results, which is probably attributable to differences in cohort characteristics (eg, HBeAg status, ethnicity, HBsAg distribution) and limited sample size. To overcome these limitations, we collected all individual patient data from multiple previously published trials.^10–17 In the current meta-analysis, HBsAg loss was achieved in 9% of patients treated with peg-IFN add-on. Although this is much higher than what would be expected with NA monotherapy, it is comparable to results obtained with peg-IFN monotherapy (3–10%),^20–23 and thus relatively limited considering the costs of treatment and the side effects.

Importantly, lower SOT HBsAg levels were associated with higher response rates, with rates around 18% among patients with SOT HBsAg levels <1000 IU/mL and 38% in patients with SOT HBsAg<100 IU/mL. This appears to be in a similar or somewhat higher range than observed with NA withdrawal, being 11% at 2 years in those with end of treatment HBsAg<1000 IU/mL in the RETRACT-B cohort²⁴ to almost 30% in a prospective trial including mostly Caucasians.²⁵ The 18% response rate is quite similar to the response rates observed in patients with HBsAg<1000 IU/mL who were treated with novel RNAi therapeutics. For example, the recent phase 2b B-Clear study of bepirovirsen (antisense oligo nucleotide) reported HBsAg loss with undetectable HBV DNA at 24 weeks after end of study treatment in 16% of patients,⁸ and the preliminary report of the MARCH study reported 17% HBsAg loss at 24 weeks after withdrawal of combination therapy consisting of elebsiran (siRNA) with tobevibart (monoclonal antibody).²⁶ Interestingly, higher response rates were observed in studies in which RNAi-based strategies also included immunomodulators such as peg-IFN.^{5 7 26} Patients with HBsAg<1000 IU/mL treated with sequential bepirovirsen followed by peg-IFN therapy achieved response in 24–41%,⁵ and 21% HBsAg loss was reported for combination treatment of elebsiran/tobevibart with peg-IFN.²⁶ The PIRANGA study further supports the superiority of RNAi combined with an immunomodulator compared with RNAi alone; HBsAg loss rates were 7–14% for xalnesiran alone, 24% for the combination with ruzotolimod (Toll-Like Receptor 7 agonist) and 47% for xalnesiran with peg-IFN in patients with baseline HBsAg<1000 IU/mL.⁷ Importantly, erosion of HBsAg suppression was observed through subsequent FU, resulting in HBsAg seroreversion in a subset of patients.^{7 27}

However, head-to-head comparison of these findings is limited by differences in study cohort characteristics and NA cessation criteria. An important difference across cohorts is patient ethnicity distribution (online supplemental table S3). While our cohort included 72% Asian patients, this was 53% in B-CLEAR, between 65% and 78% in MARCH and over 85% in PIRANGA. Given these differences, we have provided outcomes stratified by ethnicity, but these data are not available for most reported studies. HBeAg status is also an important determinant of treatment outcome. The current cohort enrolled 44% HBeAg positive patients, whereas this varied between the aforementioned studies (online supplemental table S3). We therefore adjusted for HBeAg status in the different analyses. Another potentially relevant difference across cohorts is the HBV genotype distribution. HBV genotype is an important determinant of HBsAg loss after peg-IFN therapy^28–31 and NA withdrawal,³² but its influence on response to RNAi is not yet clear. Unfortunately, assessment of HBV genotype in NA suppressed patients is not possible, which is reflected in the absence of genotype data in recent trials and the current cohort.

Despite the relatively high response rate observed with peg-IFN add-on in patients with low HBsAg, it is important to note that novel compounds may have important advantages, particularly related to shorter treatment duration and superior tolerability. The current study highlights the high incidence of side effects during peg-IFN, which range from influenza-like symptoms to ALT flares, cytopenia or psychological effects (online supplemental table S1) and commonly results in dose reduction or discontinuation (in up to 21% in the included studies).

Besides its use as add-on therapy to a NA backbone, there is emerging evidence supporting the use of peg-IFN in combination with novel antivirals to boost on-treatment responses and increase off-treatment durability of HBsAg loss. Addition of peg-IFN has been shown to be effective in increasing off-treatment sustained response rates and delaying viral relapses after therapy withdrawal,³³ and in increasing response rates to siRNA therapy^5–7 as well as nucleic acid polymers.³⁴ However, careful assessment of any synergistic effects on HBsAg loss rates requires the inclusion of a peg-IFN only control arm in trials,³⁵ which exposes patients to a treatment that is both poorly tolerated and has limited efficacy. The current study now provides robust estimates for HBsAg loss rates that can be achieved with peg-IFN add-on to NA therapy across SOT HBsAg strata and ethnicities. These data could potentially be used as a reference, limiting the need for peg-IFN control arms in trials involving NA-treated patients. As response to peg-IFN is determined by many factors, including ethnicity and baseline HBsAg, we performed stratified analyses to provide detailed information on response rates in clinically relevant subgroups of patients. Nevertheless, some potentially relevant factors, such as HBV genotype, duration of NA treatment and HBV DNA and ALT levels at start of NA therapy, were not consistently available and could therefore not be stratified for.

In addition to the potential role for boosting responses to novel antivirals, there may also be a role for peg-IFN lead-on therapy to reduce HBsAg levels before novel antivirals are introduced. This concept is based on recent observations that HBsAg loss is predominantly observed in RNAi-treated patients with SOT HBsAg levels <1000 IU/mL.^5–7 In the current study, peg-IFN add-on to NA therapy significantly reduced HBsAg levels in the overall population. Among patients with SOT HBsAg levels ≥1000 IU/mL, 30% of patients achieved levels <1000 IU/mL after 24 weeks of peg-IFN add-on therapy, which increased to 48% with prolongation of treatment to 48 weeks. These findings suggest that peg-IFN lead-on could be used to decrease HBsAg levels before initiation of novel antivirals, and our study provides estimates for the guidance of trials investigating such an approach. It remains, however, to be elucidated whether patients achieving treatment induced HBsAg reductions have similar outcomes when compared with patients who have achieved these reductions without antiviral therapy.

Furthermore, when considering the addition of peg-IFN to certain treatment regimes, it is essential to carefully weigh the benefits and the risks. Peg-IFN therapy should be limited to patients with a high likelihood of response, as also suggested by international guidelines.^36–38 In order to optimise therapy response and limit exposure to side effects, treatment discontinuation should be considered if the likelihood of success appears to be low. In the past, several stopping rules have been proposed based on treatment response, in line with current recommendations for peg-IFN monotherapy.^{12 13 29 36 39–41} Our study supports the concept that on-treatment response evaluation might be beneficial to identify patients who are unlikely to benefit from continuing treatment. Based on our analysis, both failure to achieve significant on-treatment decline as well as higher absolute levels are predictive of failure to achieve subsequent HBsAg loss, in line with previous studies.^{13 31 36} Based on our dataset, a HBsAg level ≥100 IU/mL at on-treatment week 12 appears to be an excellent stopping rule for patients treated with peg-IFN add-on therapy, as it has a NPV of >95% in both the overall population as well as those patients with a baseline HBsAg<1000 IU/mL and is easy to implement in clinical decision-making.

Although this large study included individual patient data from multiple trials on the efficacy of peg-IFN add-on in CHB, it does have certain limitations. First, we limited this analysis to high quality prospective studies, and therefore not every single study on peg-IFN add-on was included.⁴² Another limitation is that information about HBV genotype was often missing. Finally, previous studies have shown that HBsAg loss rates can increase with extended FU due to longer lasting (post-treatment) peg-IFN effects or eventually decrease because of HBsAg seroreversion.^{43 44} Since the post-treatment FU was limited to 48 weeks in most studies included in this analysis, the durability of response beyond this time frame could not be studied.

In conclusion, the addition of peg-IFN to NA-therapy results in HBsAg loss in 18% of patients with SOT HBsAg<1000 IU/mL, and in 38% of those with SOT HBsAg<100 IU/mL. Among patients with higher HBsAg levels, peg-IFN could be used to reduce HBsAg to below thresholds associated with response to novel antivirals, although 48 weeks of therapy seems required to optimise response rates. These findings can potentially be used as reference and benchmark for the design of trials aiming for HBV cure.

Data availability statement

No data are available.