Immunogenicity and safety of HBAI20 Hepatitis B vaccine in non‐responders: Double‐blinded, randomised, controlled phase 2 trial

Abstract Background & Aims Approximately 5%–10% of the general population respond inadequately to licensed recombinant hepatitis B vaccines. We assessed the immunogenicity and safety of a new HBAI20 vaccine, consisting of a new AI20 adjuvant (20‐µg recombinant human IL‐2 attached to 20‐µg aluminium hydroxide) in combination with HBVaxPro®‐10 µg. Methods In a double‐blinded, randomised, controlled phase 2 trial, 18‐ to 59‐year‐old healthy non‐responders (titre <10 mIU/ml after three or more doses of hepatitis B vaccine) were assigned (3:1 ratio) to receive either HBAI20 vaccine or HBVaxPro®‐10 µg in a 0, 1 and 2‐month schedule. The primary outcome was seroprotection (titre ≥ 10 mIU/ml) measured 1‐3 months following the third vaccination. Results A total of 133 participants were randomised to receive either HBAI20 vaccine (n = 101) or HBVaxPro®‐10 µg (n = 32). In the modified intention‐to‐treat analysis, the seroprotection rate after the third vaccination was 92.0% (80/87) in the HBAI20 group and 79.3% (23/29) in the HBVaxPro®‐10‐µg group, P = .068. Using a generalised linear mixed model to adjust for stratification factors, a higher odds of seroprotection with HBAI20 vaccine was shown (adjusted odds ratio = 3.48, P = .028). Frequency of mild and moderate local adverse events was greater in the HBAI20 group than in the HBVaxPro®‐10 µg. Rates of severe local adverse events and systemic adverse events were low and similar in both groups. Conclusions In this group of hepatitis B vaccine non‐responders, the HBAI20 vaccine demonstrated a higher seroprotection rate when adjusting for stratification factors and a similar safety profile compared to the licensed recombinant HBVaxPro®‐10 µg.


| INTRODUC TI ON
Hepatitis B virus (HBV) infection is one of the most common and serious infectious diseases globally. It is estimated that one third of the world's population has been exposed to HBV, which has resulted in 257 million chronic HBV carriers worldwide. 1 The economic burden of HBV infection is substantial because of the high morbidity and mortality associated with cirrhosis, hepatocellular carcinoma and liver failure. 2 HBV is up to 100 times more infectious than HIV. 1 As such, the development of a hepatitis B vaccine has been a major breakthrough in the global effort to eradicate HBV infection. 3  Despite their proven immunogenicity and safety, currently available recombinant hepatitis B vaccines are unable to induce an adequate immune response in 5%-10% of the general adult population. 5 Persons lacking this antibody immune response (anti-HBs level < 10 mIU/ml measured at 1-3 months after completion of the hepatitis B vaccination schedule) are referred to as nonresponders. 5 The standard of care for non-responders consists of revaccination with currently licensed hepatitis B vaccines and is associated with a seroprotection rate of 50%-69% after three doses. [6][7][8][9][10] Thus, there is a need for a more immunogenic vaccine in defined groups such as healthy non-responders to recombinant hepatitis B vaccines.
Recently, various strategies have been developed to elicit pro- HBsAg. 12 Moreover, two single-blinded, randomised, controlled trials in healthy non-responders showed a higher immunogenic response with adjuvanted vaccines, one with Fendrix® and one with Heplisav-B®. 8,10 Among a total of 82 participants, the seroprotection rate was 97.5% in the group that received three doses of Fendrix® compared to 68.0% with three doses of Engerix-B®. 8 The observed higher seroprotection rate with Fendrix® was not only linked to the adjuvant 3-O-desacyl-4′-monophosphoryl lipid A (MPL) but also to the higher antigen content, ie, 40µg HBsAg in Fendrix® vs. 20 µg in Engerix-B®. 8 The study of Halperin et al 10 was conducted among 35 non-responders, and seroprotection was found in 88.9% and 66.7% after two doses with Heplisav-B® and Engerix-B®, respectively. At present, Fendrix® is only licensed for individuals with renal alteration, while Heplisav-B® was never licensed in Europe. Most pivotal studies for the license application of Fendrix® and Heplisav-B® have been performed on healthy vaccine naive adults and in the case of Fendrix® also on (pre-) dialysis patients. [13][14][15][16][17][18][19][20][21][22][23][24][25] More research on hepatitis B vaccines among the group of healthy non-responders is therefore warranted.
Since the available recombinant hepatitis B vaccines all have an aluminium-based adjuvant, we report for the first time the immunogenic properties of a cytokine-based adjuvant in a welldesigned double-blinded, randomised, controlled trial in healthy adult non-responders. In a phase 1 trial, the new AI20 adjuvanted (HBAI20) vaccine was shown to be safe, well-tolerated and immunogenic in healthy naive and non-responding adults. 26 The current study describes a phase 2 trial to evaluate in healthy nonresponders the immunogenicity and safety of the new HBAI20 hepatitis B vaccine compared to the licensed HBVaxPro®-10µg vaccine.
author had final responsibility for the decision to submit for publication.
Frequency of mild and moderate local adverse events was greater in the HBAI20 group than in the HBVaxPro®-10 µg. Rates of severe local adverse events and systemic adverse events were low and similar in both groups.

Conclusions:
In this group of hepatitis B vaccine non-responders, the HBAI20 vaccine demonstrated a higher seroprotection rate when adjusting for stratification factors and a similar safety profile compared to the licensed recombinant HBVaxPro®-10 µg.

K E Y W O R D S
adjuvant, HBAI20, hepatitis B vaccine, immunogenicity, non-responder 2 | MATERIAL S AND ME THODS

| Study design
The study was a phase 2, double-blinded (participant and investigator), randomised, controlled, multicentre trial. The protocol (in accordance with the CONSORT statement) was approved by the local ethics committee and was conducted in accordance with the guidelines of the Declaration of Helsinki and its amendments and in accordance with good clinical practice and local laws.

| Participants
The participants were hepatitis B vaccine non-responders, ranging in age from 18 to 59 years. Non-responders were defined as subjects with documented three or more hepatitis B vaccinations and documentation of hepatitis B surface antibody (anti-HBs) level < 10 mIU/ ml measured within 1-3 months after the last vaccination.

| Randomisation and blinding
Participants were randomly assigned to the HBAI20 vaccine or the HBVaxPro®-10µg vaccine at a 3:1 ratio. Randomisation was performed by the ALEA screening and Enrolment Application Software (Formvision BV, Abcoude, the Netherlands) and with minimization of stratification factors: age (18-25, 26-35, 36-50 or 51-59 years), sex (male or female), hepatitis B vaccination history (one complete cycle or more than one cycle), and trial site (Maastricht UMC+, Antwerp University or Hospital East-Limburg). Blinding of participants and investigators was achieved by a label that concealed the volume difference between the two vaccines. The colour was similar in both vaccines. All investigators and participants were kept blinded for the respective anti-HBs results after the study vaccinations.

| Procedures
The investigational vaccine (HBAI20) consisted of a new adjuvant Participants were closely observed for 30 minutes after each vaccination for any immediate adverse events. On the day of vaccination and the subsequent 4 days, subjects were asked to record local adverse events (pain, impaired movement of injected arm, redness, swelling and induration) and systemic adverse events (fever, headache, fatigue, vomiting and diarrhoea) in individual diary cards.

| Outcomes
The primary endpoint of this study was the immunogenicity of the HBAI20 vaccine, tested in terms of proportion of participants that have attained seroprotection at 1-3 months after the third vaccination. Seroprotection rates measured at earlier time points and geometric mean concentrations (GMCs) were secondary endpoints.
Local and systemic adverse events were scored as absent, mild (no interference with daily activity), moderate (some interference with daily activity) or severe (prevented normal daily activities).
The size of redness, swelling and induration was scored as absent, mild (>5-25 mm), moderate (≥25-50 mm) or severe (>50 mm). Fever was defined as oral body temperature above 37.5°C and was scored as absent, mild (37.6-37.9°C), moderate (38.0-38.9°C) or severe (≥39.0°C). Vomiting was scored as absent, mild (once in 24 hours), moderate (twice in 24 hours) or severe (three or more times in 24 hours). Diarrhoea was scored as absent, mild (1-3 stools above normal), moderate (4-5 stools above normal) or severe (>5 stools above normal). Unsolicited symptoms and serious adverse events were recorded throughout the study period. All solicited local adverse events were considered as related to the study vaccine. The investigators assessed the relationship of all other reported adverse events to vaccination.

| Laboratory assays
Haematological, biochemical, virological parameters and urinalysis were determined using commercially available laboratory methods.
Serum anti-HBs antibody levels were analysed with the Cobas 8000 chemiluminescence assay (Roche, Germany). Subjects with anti-HBs antibody levels ≥10 mIU/ml were considered to be seroprotected and deemed vaccine responders. 5

| Statistical analysis
Sample size calculation could not directly be based on phase 1 HBAI20 study data because the study population characteristics were different with respect to hepatitis B vaccination history (naive subjects vs. non-responders). Our aim was to include between 132 and 140 subjects in agreement with other phase 2 immunogenicity and safety studies for hepatitis B vaccines. 8,10,14,20 Sample size for this trial was based on superiority analysis.
All data of the outcome variable anti-HBs level were log10transformed. To overcome problems with 0 values, a value of +0.5 was added to all anti-HBs levels. Categorical data were expressed as frequencies (%) and continuous variables were expressed as mean ± standard deviation (SD). For the comparison of categorical variables, either chi-squared test or Fisher's exact test was used.
Student's t test or Kruskall-Wallis test was used to compare groups in terms of continuous outcomes.
Safety analysis was performed including all participants that received at least one vaccination ( Figure 1). The percentage of adverse events for the two vaccines was compared using a chi-squared test for the equality of proportions, or a Fisher's exact test in case of small samples. A modified intention-to-treat (mITT) approach was used for superiority analysis. In order to establish statistically significant superiority of the new vaccine, the one-sided 95% confidence interval (CI) for the odds ratio (OR) of seroprotection has to be entirely above one.
Since there was one missing seroprotection outcome at Visits 3 and 4, and two at Visit 2, multiple imputation was used to impute these values. The GMC values were imputed. Multiple imputation was done using the Statistical Analysis Software (SAS Institute Inc, NC, USA) procedure PROC MI. Considering the non-monotone pattern of missingness, the fully conditional specification (FCS) regression method was used to fill in the missing values. 28 Imputation was done separately for the two vaccine allocations, to avoid imposing any relationship between the outcome variables and this covariate.
The data were imputed five times. For each imputed dataset, a generalised linear mixed model outside the study protocol was used to estimate the vaccine effect at Visit 4, ie, 1-3 months following the last vaccination. The stratification factors vaccination history, age, and sex were included as fixed effects in the model, in addition to vaccine allocation (ie, treatment effect). Interactions between these stratification factors and vaccine allocation were also included in the model. If not significant, the interaction effect was removed from the model, indicating no difference between vaccine allocations within the stratification group. Trial site was included as a random effect. To assess the sensitivity of the results to the method of handling missing data, a sensitivity analysis based on extreme case analyses was done, ie, worst-case and best-case analysis. All reported P values for vaccine allocation were based on one-sided tests.

Independent data monitoring committee Clinical Trial Center
Maastricht oversaw the study.   One month after first vaccination, seroprotection percentages were respectively 64.0% vs. 69.0% (P = .761), and these numbers were 82.8%
The GMCs were higher in the HBAI20 group than in the HBVaxPro®-10µg group after vaccination with one (21.3 ± 1. Sixty-four of 117 subjects had received more than one hepatitis B vaccination cycle prior to inclusion in the present study and the seroprotection percentages were as follows: 66.0% vs. 64.7% (P = .575), 80.9% vs. 70.6% (P = .289) and 87.2% vs. 70.6% (P = .120) after respectively one, two and three vaccinations with HBAI20 and HBVaxPro®-10µg vaccine.
Using a generalised linear mixed model, adjusting for stratification factors, the primary endpoint was met, and a higher odds of seroprotection with the HBAI20 vaccine was shown (adjusted OR = 3.48, lower bound of 95% CI = 1.19, P = .028) ( Table 2). Since imputation of one missing value at visit 4, ie, 1-3 months after the third vaccination, resulted in a seroprotected score, results from 18/20 (90.0%) with pain at the injection site, P = .141).
There was a higher incidence of solicited local adverse events in the HBAI20 group than HBVaxPro®-10µg group (13.7% vs. 8.7%, P < .001), while the overall incidence of systemic adverse events was similar for both vaccines (4.0% vs. 3.4%, P = .198). The percentage of participants with mild, moderate and severe adverse events (local and systemic) in each group is illustrated in Figure 3.
The most frequently reported local adverse event in both the HBAI20 group (32.7%) and HBVaxPro®-10µg group (18.6%) was pain at the injection site, P < .001. Pain at the injection site resolved F I G U R E 2 Percentage of individuals in HBAI20 and HBVaxPro®-10µg group with hepatitis B surface antibody levels ≥10, ≥100 and ≥1,000 mIU/ml after one, two and three vaccinations. n = 86 for first HBAI20 vaccine and n = 87 for second and third HBAI20 vaccine. n = 29 for first, second and third HBVaxPro®-10µ-g vaccine within 4 days after vaccination for 94% and 97% of participants, respectively. There were no severe local adverse events in the HBVaxPro®-10µg group, and these were low in the HBAI20 group.

TA B L E 2 Results of the generalised linear mixed model
None of the participants experienced severe local adverse events for longer than 4 days. The frequencies of any and severe local adverse events are shown in Table 3.
The most frequently reported systemic adverse event was fatigue (10.6% vs. 9.7%, P = .655), which resolved within 4-day postvaccination in 96% and 91% of the participants in the HBAI20 group and HBVaxPro®-10µg group, respectively. The incidence of severe systemic adverse events was low and comparable in both groups.
In all cases, severe systemic adverse events resolved within 4 days.  Two serious adverse events were reported during the study period, one in the HBAI20 group (ie, hernia nuclei pulposi) and one in the HBVaxPro®-10 µg (ie, epileptic seizure). Both were considered to be unrelated to vaccination by the investigators.

| D ISCUSS I ON
Hepatitis B vaccine non-responders, ie, individuals with a deficient antibody response after three or more vaccinations, represent an unresolved problem in hepatitis B vaccination. 11 In order to reach hepatitis B elimination by 2030 as advocated by the World Health Organization, it is important to establish protective anti-HBs levels also in hepatitis B vaccine non-responders, who represent 5%-10% of the vaccinated adult population. 11,29,30 After adjusting the analysis for stratification factors, this phase 2 trial demonstrated that immunogenicity of the investigational HBAI20 vaccine was superior to that of the licensed HBVaxPro®-10µg vaccine, when given to healthy hepatitis B vaccine nonresponders at 0, 1 and 2-month schedule. Safety was comparable between both vaccines.
Seroprotection rates in our comparator arm HBVaxPro®-10 µg were higher compared to previous studies. 6 The primary endpoint of this study was met: compared to HBVaxPro®-10 µg, subjects receiving HBAI20 vaccine were about 3.5 times more likely to be seroprotected around 1-3 months following the last vaccination. The observed seroprotection percentage after three vaccinations with HBAI20 vaccine was 92%, which is higher than expected in this population of hepatitis B vaccine nonresponders. Moreover, seroprotection after two doses of HBAI20 vaccine was comparable to those attained after three doses of the in HBAI20 group and 19% in HBVaxPro®-10µg group, respectively, and was similar to those reported in previous studies with adjuvanted hepatitis B vaccines. 8,10 With respect to local adverse events, we could also show greater immunity, ie, seroprotection, in those participants with pain at the injection site in the HBAI20 group.
Further research is warranted on whether the presence of pain at the injection site is associated with more attraction of innate cells to the inoculum and subsequent higher seroprotection rate.
This study had some limitations. First, a formal sample size calculation could not be performed since the naive study population in phase 1 study differed significantly from the non-responder population in the current study. Therefore, this study was underpowered to find significant associations between HBAI20 vaccine and certain study endpoints in univariate analyses. However, univariate analyses do not account for heterogeneity between trial sites, and the baseline characteristics are almost never completely similar between both study groups, even after randomisation. By adjusting for stratification factors, the results from the multiple models avoid an unnecessary loss in power and are more reliable than those obtained from univariate analyses. In that respect, superiority of HBAI20 In conclusion, these results are suggestive of increased immunogenicity of the HBAI20 vaccine among healthy non-responders to three or more hepatitis B vaccinations, in addition to confirming that the HBAI20 vaccine could be safely administered. These results support a phase 3 study to assess the benefits of the HBAI20 vaccine in the unresolved group of healthy non-responders. It would be worth to investigate the benefits of the HBAI20 vaccine in other risk groups for a suboptimal immune response, ie, patients with advanced liver disease, haemodialysis patients, and patients with other immunodeficient conditions.

ACK N OWLED G EM ENTS
CyTuVax BV sponsored the study. The authors would like to thank the many participants and would like to acknowledge Mohannad

E TH I C S A PPROVA L
The protocol was approved by the local ethics committee and was conducted in accordance with the guidelines of the Declaration of Helsinki and its amendments and in accordance with good clinical practice and local laws.

CLI N I C A L TR I A L N U M B E R
ClinicalTrials.gov number, NCT03415672.

DATA AVA I L A B I L I T Y S TAT E M E N T
Proposals should be directed to the corresponding author; to gain access, data requestors will need to sign a data access agreement.
Only individual participant data that underlie the results reported in this article, after de-identification, will be shared.