Reactogenicity, immunogenicity and breakthrough infections following heterologous or fractional second dose COVID-19 vaccination in adolescents (Com-COV3): A Randomised Controlled Trial

Background This was the first study to investigate the reactogenicity and immunogenicity of heterologous or fractional second dose COVID-19 vaccine regimens in adolescents. Methods A phase II, single-blind, multi-centre, randomised-controlled trial recruited across seven UK sites from September to November 2021, with follow-up visits to August 2022. Healthy 12-to-16 years olds were randomised (1:1:1) to either 30 µg BNT162b2 (BNT-30), 10 µg BNT162b2 (BNT-10), or NVX-CoV2373 (NVX), eight weeks after a first 30 µg dose of BNT162b2. The primary outcome was solicited systemic reactions in the week following vaccination. Secondary outcomes included immunogenicity and safety. ‘Breakthrough infection’ analyses were exploratory. Findings 148 participants were recruited (median age 14 years old, 62% female, 26% anti-nucleocapsid IgG seropositive pre-second dose); 132 participants received a second dose. Reactions were mostly mild-to-moderate, with lower rates in BNT-10 recipients. No vaccine-related serious adverse events occurred. Compared to BNT-30, at 28 days post-second dose anti-spike antibody responses were similar for NVX [adjusted geometric mean ratio (aGMR) 1.09 95% confidence interval (CI): 0.84, 1.42] and lower for BNT-10 [aGMR 0.78 (95% CI: 0.61, 0.99)]. For Omicron BA.1 and BA.2, the neutralising antibody titres for BNT-30 at day 28 were similar for BNT-10 [aGMR 1.0 (95% CI: 0.65, 1.54) and 1.02 (95% CI: 0.71, 1.48), respectively], but higher for NVX [aGMR 1.7 (95% CI: 1.07, 2.69) and 1.43 (95% CI: 0.96, 2.12), respectively]. Compared to BNT-30, cellular immune responses were greatest for NVX [aGMR 1.73 (95% CI: 0.94, 3.18)], and lowest for BNT-10 [aGMR 0.65 (95% CI: 0.37, 1.15)] at 14 days post-second dose. Cellular responses were similar across the study arms by day 236 post-second dose. Amongst SARS-CoV-2 infection naïve participants, NVX participants had an 89% reduction in risk of self-reported ‘breakthrough infection’ compared to BNT-30 [adjusted hazard ratio (aHR) 0.11 (95% CI: 0.01, 0.86)] up until day 132 after second dose. BNT-10 recipients were more likely to have a ‘breakthrough infection’ compared to BNT-30 [aHR 2.14 (95% CI: 1.02, 4.51)] up to day 132 and day 236 post-second dose. Antibody responses at 132 and 236 days after second dose were similar for all vaccine schedules. Interpretation Heterologous and fractional dose COVID-19 vaccine schedules in adolescents are safe, well-tolerated and immunogenic. The enhanced performance of the heterologous schedule using NVX-CoV2373 against the Omicron SARS-CoV-2 variant suggests this mRNA prime and protein-subunit boost schedule may provide a greater breadth of protection than the licensed homologous schedule. Funding National Institute for Health Research and Vaccine Task Force. Trial Registration International Standard Randomised Controlled Trial Number registry: 12348322.


Introduction
Paediatric immunisation against SARS-CoV-2 has now been recommended in the majority of high-income countries, particularly in light of the sharp increase in infection rates in children and paediatric hospitalisations worldwide following the emergence of highly transmissible SARS-CoV-2 variants of concern (VOC). 1 An increased incidence of myocarditis has been observed following receipt of a second dose of an mRNA-based COVID-19 vaccine, particularly in male adolescents, with up to 105.9 cases per million reported. 2 This risk may be reduced by using another vaccine or fractional dose for the second vaccination (a mixed schedule), or variation to dose intervals. 3 Heterologous vaccination schedules in adults have been shown to be safe and immunogenic and have been implemented in Canada and northern Europe. 4 They have been approved by the World Health Organisation to enhance vaccination coverage, particularly where supplies are limited. 5 In the Com-COV2 trial a heterologous COVID-19 vaccine schedule utilising the Matrix-M adjuvanted recombinant nanoparticle spike protein vaccine NVX-CoV2373 (Novavax) as the second dose following BNT162b2 (Pfizer-BioNTech), was found to be less immunogenic in 50-to-70-year-olds than two doses of BNT162b2. The BNT/NXV-CoV2373 schedule was, however, still more immunogenic than two doses of the adenoviral-vectored ChAdOx1 n-CoV-19 (Oxford/AstraZeneca) vaccine, which is highly effective against hospitalisation and death. 4,6 J o u r n a l P r e -p r o o f 6 Fractional dose BNT162b2 schedules are now approved for 5-to-11-year-olds. A phase 2/3 study in participants aged 5-to-11-years showed a homologous two-dose regimen of 10μg BNT162b2 induced immune responses comparable to that of a two-dose regimen of 30μg BNT162b2 administered to 16-to-25-year-olds and was associated with a lower incidence of systemic adverse events. 7 No data have previously been published on the safety, immunogenicity, and efficacy of mixed and fractional COVID-19 vaccine schedules in an adolescent population. Here we report a randomised-controlled trial to determine the reactogenicity, immunogenicity, and number of SARS-CoV-2 infections in adolescents receiving 30μg BNT162b2 as a first dose and a second dose of either 30μg BNT162b2 (schedule hereafter referred to as BNT-30), 10μg BNT162b2 (schedule hereafter referred to as BNT-10), or NVX-CoV2373 (schedule hereafter referred to as NVX) given at least 8 weeks later.

Study Design
Com-COV3 is a single-blind, randomised, phase II, multi-centre study to determine reactogenicity and immunogenicity of mixed COVID-19 vaccine schedules in adolescents.
The study consists of two cohorts. This paper reports the results of the first cohort investigating the reactogenicity and immunogenicity of heterologous and fractional second dose COVID-19 vaccine schedules in adolescents. The second cohort will examine mixed third dose COVID-19 vaccine schedules in adolescents and will be reported separately.
Recruitment for the first cohort commenced on 27 th September 2021 and occurred across seven UK National Health Service and academic institutions. The study was approved by the South-Central Berkshire Research Ethics Committee (21/SC/0310) and the Medicines and 7 Healthcare products Regulatory Agency. The Consolidated Standards of Reporting Trials (CONSORT) guideline was followed. The study protocol is accessible at https://comcovstudy.org.uk/study-protocol.
We report here the safety, including adverse events (AEs) of special interest (AESIs) and serious AEs (SAEs), reactogenicity, immunogenicity, and breakthrough infections of mixed COVID-19 vaccine schedules following the second vaccination up until database lock on 24 th March 2023.

Changes to the planned protocol
Following a change to UK national immunisation policy on 29 th November 2021 recommending that all 12-to-15-year-olds should be offered a second dose of 30μg BNT162b2 , the study design was reviewed with the Trial Steering Committee. Recruitment to the study was stopped (thus the pre-planned sample size was not met), and the study amended to focus on the immune response to BNT162b2. After this date, enrolled participants who had not yet received their second vaccination were randomised to either BNT-30 or BNT-10. Participants were no longer randomised to the NVX study arm in order to prioritise those groups more likely to inform UK immunisation policy.

Participants
Adolescents aged 12-to-16-years inclusive, who were COVID-19 vaccine naïve or had received a single dose of 30μg BNT162b2, were eligible. 'High-risk' individuals who were advised to receive additional doses of BNT as part of the UK COVID-19 vaccination programme (e.g., confirmed, or suspected immunosuppressive condition or serious chronic illness) were excluded. Previous SARS-CoV-2 infection was not an exclusion criterion. (See https://comcovstudy.org.uk/study-protocol for protocol eligibility criteria).

Randomisation and Masking
Computer generated randomisation lists were prepared by the study statistician. Participants were randomised 1:1:1 at the time of their second vaccination to BNT-30, BNT-10, or NVX.
After 29 th November 2021, when UK national immunisation policy changed to offer all 12-to-15-year-olds a second dose of BNT, a protocol amendment was implemented and participants who had already received their first dose of BNT162b2 within the study were randomised 1:1 to receive 30µg BNT162b2 or 10µg BNT162b2 as a second dose.
Randomisation was performed using block randomisation. Block sizes of three and six were used before 29 th November 2021, and block sizes of two and four used thereafter.
Randomisation was stratified by study site and baseline anti-nucleocapsid IgG serostatus.
Participants were blinded to allocation until one-month after second dose. To maintain the blind, vaccines were prepared out of sight of the participant, and masking tape applied to the vaccine syringe. Laboratory staff were blinded to the study arm, staff involved in the study delivery were not. Statisticians were unblinded throughout the trial.

Procedures
Informed consent was obtained from participants aged 16 years or from their parents or guardians if younger than 16 years, and written assent obtained from participants aged 12-to-15-years. All participants and parents had the opportunity to read the Participant Information Sheet (customised versions available for parents, participants aged 16 years, and 12-to-15years to enhance comprehensibility) and to discuss trial participation with a member of the study team before signing the consent form. COVID-19 vaccine naïve participants attended a screening/enrolment visit and those eligible received a 30µg BNT162b2 first dose. Participants J o u r n a l P r e -p r o o f 9 who had received 30µg BNT162b2 in the community attended a screening visit eight weeks afterwards. All participants were randomised and vaccinated at least eight weeks after their first dose. Two vaccines were used in the study and administered by intramuscular injection in the upper arm. 30µg BNT162b2 was given as 0.3ml, 10µg BNT162b2 as 0.1ml, and NVXCoV2373 as 0.5ml injection.
At the time of this study, community testing for SARS-CoV-2 was free, widely available and conducted either in response to symptoms or as screening. Participants recorded in electronic diaries all SARS-CoV-2 infections (classified as AESIs) detected by community-based selftesting with either PCR or rapid antigen test, along with all solicited and unsolicited AEs

Statistical analysis
The primary outcome analysis was conducted on the safety analysis population and included participants who received a second vaccine in the study. The maximum severity for each J o u r n a l P r e -p r o o f 11 solicited systemic AE across seven days after second vaccination was derived for each participant and summarised by group. Analyses were conducted similarly for local reactogenicity.
Immunogenicity analyses were conducted on the modified intention-to-treat (mITT) populations at 28, 132, and 236 days following second vaccination overall and stratified by pre-second dose serostatus (defined by anti-nucleocapsid value pre-second dose or pre-first dose if missing). The mITT populations excluded participants who withdrew, had no blood sample, received a third dose in the community before their visit or self-reported a SARS-CoV-2 infection within 14 days after second vaccination. Immunogenicity outcomes were summarised using geometric mean concentrations (GMCs) and 95% confidence intervals (CIs). Adjusted geometric mean ratios (aGMRs) and 95% CIs were calculated comparing groups to BNT-30 as the reference, adjusting for study site in the linear regression models.
Immunogenicity secondary and sensitivity analyses were conducted on anti-spike antibodies 28 days after second vaccination. Secondary analyses were further adjusted for pre-second dose anti-spike antibodies and interval between two doses. Sensitivity analyses were conducted on the per-protocol population, excluding those randomised after 29 th November 2021, and excluding those who self-reported a SARS-CoV-2 infection within 28 days after second vaccination. Immunogenicity sensitivity analyses were conducted on anti-spike antibodies at 132 and 236 days after second vaccination excluding participants who were considered to have Of all 148 participants enrolled, the median age was 14 years (range 12-17), 62% were female, and six participants (4%) were from an ethnic minority group ( Table 2, Supplement). Fortyeight participants were randomised to BNT-30, 47 to BNT-10 and 37 to NVX. The median interval between two doses was 59 days (range 56-109) and was similar across the study arms.
Of 132 participants randomised, 65 received their first dose in the trial, of whom 19% were positive for SARS-CoV-2 anti-nucleocapsid IgG prior to their first dose. Seropositivity increased to 30% prior to second dose.
Generally, solicited systemic adverse reactions occurred more frequently in BNT-30 and NVX groups compared to BNT-10, while solicited local reactions were least common in the NVX group ( Figure 1). Most reactions were mild-to-moderate, with no clear difference in the frequency of severe reactions across the groups. There was no clear difference in the frequency or severity of reactions when stratified by serostatus (Figure 3, Supplement). These findings were reflected in paracetamol usage after second dose (Table 3, Supplement). J o u r n a l P r e -p r o o f 14 Two SAEs occurred: the first four months after second vaccination (toxic substance ingestion, considered unrelated to immunisation), the second seven months after vaccination (anorexia nervosa, also considered unrelated to immunisation) ( Table 4, Supplement). There were 65 AESIs, all of which were self-reported SARS-CoV-2 infections (Table 5, Supplement).
Unsolicited AEs are reported in Table 6, Supplement. Almost all participants had a troponin value below the LLOD after second vaccination ( Table 7 (Table 10, Supplement). When limited to seronegative participants, this was 81%, 89%, and 72% respectively.
There was a difference in the risk of self-reported 'breakthrough infections' across the study arms for participants randomised before 29 th November in the mITT population ( Figure 4A). The NVX and, especially, BNT-10 schedules studied here had favourable side-effect profiles compared to BNT-30, with the NVX reactogenicity findings consistent with studies in adults. 4,11 This was reassuring in this adolescent study given the increased reactogenicity previously observed in younger versus older COVID-19 vaccine recipients. 12 Although J o u r n a l P r e -p r o o f 19 increased reactogenicity has been reported in those with previous SARS-CoV-2 infection, we found no clear difference in reactogenicity when stratified by serostatus. 13 There was no evidence of myocardial inflammation in any study arm clinically or by troponin measurement, although population-based studies are required to definitively assess very rare side effects.
We found BNT-10 was immunogenic in adolescents. However, in participants who were infection naïve (i.e., seronegative to nucleocapsid antigen prior to the second dose), point estimates of binding antibody concentrations were lower than in BNT-30 participants. Furthermore, higher rates of self-reported breakthrough infections were observed in seronegative BNT-10 participants and, once participants experiencing breakthrough infections were excluded, lower antibody concentrations remained evident at day 132. This is consistent with a more rapid loss of protection from infection against VOCs observed following two doses of 10μg BNT162b2 in 5-to-11-year-olds than following 30μg BNT162b2 in 12-to-17-yearolds. 14 However, consideration of its use (or that of any COVID-19 vaccine regimen) in infection-naïve adolescents is now largely irrelevant. The current UK and global situation is that of near universal prior infection with SAR-CoV-2, 15 and in our study among participants previously infected with SARS-CoV-2 prior to their second COVID-19 immunisation, anti-SARS-CoV-2 antibodies were similar between BNT-10 and BNT-30 groups, with few breakthrough infections in either group and comparable antibody concentrations at day 236. 16 This is consistent with recent studies showing that "hybrid immunity" afforded through prior SARS-CoV-2 infection, provides enhanced protection against symptomatic re-infection. 17 These findings highlight the important role of antigen encounter in shaping the immune response to vaccination and support the potential use of fractional dosing when providing the third (or fourth) antigen exposure. This approach of reduced dosing for post-primary immunisation has been adopted by the mRNA vaccine manufacturer Moderna for its monovalent booster vaccines, albeit at a dose (50μg) that remains above that of the full dose BNT162b2 vaccine studied here, and for its bivalent Omicron containing variant vaccines (which contain 25μg for each of its two target strains). 18 The potential for 10µg BNT162b2 to be used as a third dose is currently being assessed in an extension to this Com-COV3 study. 19 In the meantime, the utility of homologous 10µg BNT162b2 schedules for primary immunisation in adolescents remains under investigation and will be further informed by an ongoing phase I adolescent study of homologous 10g BNT162b2. 20 Overall, the highest humoral (including neutralising antibody titres) and cellular immune responses, were observed in the NVX group; this pattern persisted out to day 132 even when participants with SARS-CoV-2 infections following vaccination were excluded from the analysis. Correspondingly, the lowest rate of self-reported and serologically confirmed infections were recorded in this group. Furthermore, when compared to the number of serologically confirmed infections, a lower number of self-reported infections were reported in the NVX group, suggesting these participants also experienced milder symptoms on infection.
The majority of breakthrough infections for the NVX group occurred later in the study compared to BNT-30 and BNT-10 groups, consistent with the gradual rise in antibody responses seen from day 132 to day 236. NVX-CoV2373 has already been shown to be highly immunogenic in adult populations and to provide protection against VOCs. 21 Our findings also concur with the preliminary results from the adolescent PREVENT-19 Phase 3 trial in which NVX-CoV2373 demonstrated protective efficacy of 79.5% against SARS-CoV-2 and 82% vaccine efficacy against the Delta variant. 22,23 It has been postulated that the success of the vaccine's performance may be attributable to the presence of the novel Matrix M adjuvant, previously shown to enhance immunogenicity. 24 This is the first study to examine immune responses to mixed COVID-19 vaccine schedules in adolescents, and revealed striking differences in responses elicited by NVX compared with J o u r n a l P r e -p r o o f 21 50-to-70-year-olds. 4 Specifically, in the adolescent seronegative NVX group, higher anti-spike antibodies were observed compared to BNT-30 (aGMR 1.33), whereas in seronegative adults, this schedule induced concentrations that failed non-inferiority compared to BNT-30 (aGMR 0.53). 4 Cellular immune responses to NVX also differed, with a relatively low frequency of SARS-CoV-2 specific T-cells in adults (29 SFC/10 6 , aGMR 0.6 compared to BNT-30), versus adolescents (121 SFC/10 6 , aGMR 1.73). 4 An age-related decline in immunogenicity ('immunosenescence') for COVID-19 and non-COVID-19 vaccines is well recognised, 25,26,27 and is further evidenced by geometric mean NAb titres against the ancestral strain being 4-fold higher following BNT-30 in seronegative adolescents (Com-COV3) versus adults (Com-COV2) and the observation that two doses of 10μg BNT162b2 in children elicits an immune response similar to BNT-30 in young adults. 6,7 Nevertheless, the relatively better immune response to NVX compared with BNT-30 in adolescents versus adults suggests that this agerelated waning was more pronounced for NVX than BNT-30.
Previous SARS-CoV-2 infection may be expected to be associated with higher baseline, and hence post-vaccination spike-specific antibody levels. However, we found markedly greater binding and neutralising antibody responses in seronegative compared to seropositive NVX recipients. This unexpected finding in our NVX group should however be interpreted cautiously because of the small sample size. Two NVX recipients who tested SARS-CoV-2 positive in the week following their second vaccination were excluded from the analysis to reduce the risk of confounding bias.
The limitations of this study include the use of a pragmatic approach, with no formal sample size calculations. The change to UK national immunisation policy which necessitated a change to the study design resulted in fewer participants recruited than originally planned, limiting the J o u r n a l P r e -p r o o f 22 reliability of the conclusions which can be drawn from these data. The age and ethnicity of the study population affects the generalisability of the findings because of the limited representativeness when compared to the general population. The differences in self-reported infection observed in this study need to be interpreted with caution as the study was not powered to assess efficacy, and BNT-10 participants may have been more likely than other groups to self-test once unblinded at day 28 following vaccination. However, the pattern of infection across groups remained consistent for serologically defined infections which were not influenced by testing behaviour. Also, the formulation of BNT-10 used (a one-third dose of the adult preparation, administered in a 0.1 ml volume) differs from the licensed formulation of this vaccine (administered as a 0.2 ml dose). 28 Further clarity on the relevance of this will be provided in the second cohort of the Com-COV3 study, investigating 'third dose' COVID-19 schedules, in which both preparations of the 10μg dose are being compared directly.
Considering the varied vaccine volumes used in the study, there is also a risk that participants may have been inadvertently unblinded at the time of vaccination. However, to maintain the blind, vaccines were prepared out of sight and the syringes covered with masking tape, thereby minimising this risk.
In summary, this study shows that heterologous and fractional dose COVID-19 vaccine schedules studied are well-tolerated and immunogenic in adolescents. BNT-10 demonstrated a highly favourable reactogenic profile, however it elicited the lowest peak immune response.
Although BNT-10 neutralising activity against Omicron BA.1 and BA.2 variants was comparable to BNT-30, BNT-10 participants were more than twice as likely to have a self- The study protocol is provided in the Supplement. Individual participant data will be made available when the trial is complete, upon requests directed to the corresponding author; after approval of a proposal, data can be shared through a secure online platform.

Acknowledgements:
The study is funded by the UK Government through the National Institute for Health Research BNT-30: BNT162b2 30µg; BNT-10: BNT162b2 10µg; NVX: NVX-CoV2373. The severity presented is the participant's highest severity across 7 days following vaccination for each solicited adverse event. Fever: Mild: 38·0°C to <38·5°C; moderate: 38·5°C to <39°C; severe: ≥39·0°C. Feverish: Selfreported feeling of feverishness. For systemic symptoms, grading was classified as: Mildeasily tolerated with no limitation on normal activity; Moderatesome limitation of daily activity; Severeunable to perform normal daily activity. There were two self-reported SARS-CoV-2 infections in days 0-7 after second vaccination both occurring in the NVX-CoV2373 study arm. The first participant self-reported 6 days after second vaccination had Figure 2. Immune responses at day 28 (humoral), and day 14 (cellular) after the second vaccination, by study arm and pre-second dose serostatus in the day 28 modified intention-totreat populations.
BNT-30: BNT162b2 30µg; BNT-10: BNT162b2 10µg; NVX: NVX-CoV2373; CI: confidence interval. J o u r n a l P r e -p r o o f 29 Data presented are the geometric means, adjusted geometric mean ratios and their corresponding 95% confidence intervals. The boxes indicate the adjusted geometric mean ratio and the horizontal lines indicate the corresponding 95% confidence intervals. The geometric mean ratios between BNT-30 and either BNT-10 or NVX are adjusted for study site as a fixed effect. The vertical dotted line refers to an adjusted geometric mean ratio of one and indicates the line of no difference. A confidence interval that lies completely to one side and not intersecting the line of no difference indicates a significant difference in the geometric mean concentrations between the study arm and the reference BNT-30 study arm.
J o u r n a l P r e -p r o o f 30 Figure 3. Neutralising activity against Omicron BA.1 and BA.2 variants by study arm and serostatus pre-second dose at 28 days after the second vaccination in the day 28 modified intention-to-treat population.
BNT-30: BNT162b2 30µg; BNT-10: BNT162b2 10µg; NVX: NVX-CoV2373; CI: confidence interval. Data presented are the geometric means, adjusted geometric mean ratios and their corresponding 95% confidence intervals. The boxes indicate the adjusted geometric mean ratio and the horizontal lines indicate the corresponding 95% confidence intervals. The geometric mean ratios between BNT-30 and either BNT-10 or NVX are adjusted for study site as a fixed effect. The vertical dotted line refers to an adjusted geometric mean ratio of one and indicates the line of no difference. A confidence interval that lies completely to one side and not intersecting the line of no difference indicates a significant difference in the geometric mean concentrations between the study arm and the reference BNT-30 study arm. J o u r n a l P r e -p r o o f 32 A: participants randomised before 29 th November 2021; B: seronegative participants randomised before 29 th November 2021; C: seropositive participants randomised before 29 th November 2021; D: participants randomised to BNT-30 or BNT-10 during the recruitment period; E: seronegative participants randomised to BNT-30 or BNT-10 during the recruitment period; F: seropositive participants randomised to BNT-30 or BNT-10 during the recruitment period.
BNT-30: BNT162b2 30µg; BNT-10: BNT162b2 10µg; NVX: NVX-CoV2373. Self-reported SARS-CoV-2 infections occurring from >14 days following second dose were considered an event. Participants were censored at the date of either: self-reported SARS-CoV-2 infection within 14 days of second dose inclusive, third dose vaccination in the community, withdrawal, day 236 visit, or 236 days after second vaccination if day 236 visit was missed and no infection was self-reported, whichever came first. A vertical step down on the curve indicates a self-reported SARS-CoV-2 infection and a tick mark on the curve indicates a censored event. Participants were randomised 1:1:1 at the time of their second vaccination to BNT-30, BNT-10, or NVX. After 29th November 2021, when UK national immunisation policy changed to offer all 12-to-15-year-olds a second dose of BNT, recruitment stopped and participants who had already received their first dose of BNT within the study were randomised 1:1 to receive 30µg BNT162b2 or 10µg BNT162b2 as a second dose.