Safety and immunogenicity of a reduced dose of the BNT162b2 mRNA COVID-19 vaccine (REDU-VAC): a single blind, randomized, non-inferiority trial

21 COVID-19; fractional dose; reduced dose; mRNA vaccination; non-inferiority. 22 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 3

23

24 The use of fractional dose regimens of COVID-19 vaccines has the potential to accelerate vaccination rates 25 in low -income countries. Dose-finding studies of the mRNA vaccine BNT162b2 (Pfizer -BioNTech) have 26 suggested that a fractional dose induces comparable antibody responses to the full, licensed dose in 27 people below 55 years old. Here, we report the safety and immunogenicity of a fractional dose regimen 28 of the BNT162b2 vaccine. 29

30 REDU-VAC is a participant-blinded, randomised, phase 4, multicentre, non-inferiority study investigating 31 safety, reactogenicity and immunogenicity of BNT162b2. Adults aged between 18 and 55 years, without 32 uncontrolled co-morbidities, either previously infected or infection naïve, were eligible and recruited at 33 five sites across Belgium. Participants were randomly assigned to receive 20µg/20µg (fractional dose) or 34 30µg/30µg (full dose) of BNT162b2, administered intra-muscularly at a three-week interval. The primary 35 endpoint was the geometric mean ratio (GMR) of serum SARS -CoV-2 anti-RBD IgG titres at 28 days post 36 second dose between the reduced and the full dose regimen s. The reduced dose was considered non -37 inferior to the full dose if the lower limit of the two-sided 95% CI of the GMR was greater than 0.67. The 38 primary analysis was done on the per-protocol population, including infection naïve participants only. 39 Findings 40 Between April 19 and April 23, 2021, 145 partici pants were enrolled in the study and randomized, of 41 whom 141 were vaccinated and reached the primary endpoint. Participants were mostly female (69.5%), 42 of European origin (95%), with a mean age of 40 .4 years (SD 7 .9). At 28 days post second dose, the 43 geometric mean titre (GMT) of SARS-CoV-2 anti-RBD IgG of the reduced dose regimen (1,705 BAU/mL) 44 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 4 was not non -inferior to the full dose regimen (2 ,387 BAU/mL), with a GMR of 0.714 (two -sided 95% CI 45 0.540-0.944). No serious adverse events occurred. 46

47 While non-inferiority of the reduced dose regimen was not demonstrated, the SARS -CoV-2 anti-RBD IgG 48 titre was only moderately lower than that of the full dose regimen and, importantly, still markedly higher 49 than the reported antibody response to the licensed adenoviral vector vaccines. These data suggest that 50 reduced doses of the BNT162b2 mRNA vaccine may offer additional benefit as compared to the vaccines 51 currently in use in most low and middle -income countries , war ranting larger immunogenicity and 52 effectiveness trials. The trial is registered at ClinicalTrials.gov (NCT04852861). 53 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 5

54 Today, less than 15% of people in low-income countries (LIC) have been vaccinated against COVID-19 with 55 at least one dose, compared to 68% in high-income countries (1). Besides allowing the emergence of new 56 SARS-CoV-2 variants and their subsequent spread around the world, this blatant inequity in global vaccine 57 distribution is leaving hundreds of millions of people vulnerable to the deadly virus. According to the UN, 58 vaccine inequity will further deepen inequality and have a lasting impact on socio-economic recovery in 59 LIC (2). 60 The lack of access to COVID -19 vaccines is partly due to a phenomenon called ‘vaccine nationalism’, 61 wherein most vaccines are being reserved for wealthy countries, and vaccine-producing countries limit 62 exports to make sure that their own population is vaccinated first. Another important factor is the cost. 63 Under current pricing, LIC would have to increase health care expenditure by up to 50%, compared to less 64 than 1% for high-income countries, to vaccinate 70% of their population (3–5). The WHO, Gavi and CEPI 65 co-led program COVAX was set up to stimulate global equitable access to COVID -19 vaccines. Although 66 this program has led to the donation of over 1 billion vaccine doses to LIC to date, more doses, better 67 logistics, and political will are required to cover the needs of the Global South. 68 Fractionating doses as a dose -sparing strategy might help speed up vacci nation rates by effectively 69 reducing the cost per dose and thereby increasing the number of available doses. Based on proven safety 70 and immunogenicity, t he World Health Organization (WHO) has recommended fractional doses in the 71 past during outbreaks of yellow fever, polio and meningococcal disease in case of vaccine shortages in 72 resource-limited settings (6–9). 73 In terms of fractional vaccine doses for COVID-19 however, data is mostly limited to dose-finding studies. 74 In a phase 1/2 trial in healthy adults aged 18 -55 years, humoral and cellular responses to th e BNT162b2 75 mRNA vaccine (Comirnaty®, Pfizer-BioNTech) were found to be very similar between the full dose of 30µg 76 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 6 and fractional doses of 20µg and even 10µg (10). For the mRNA -1273 vaccine ( Spikevax®, Moderna), a 77 phase 2 trial showed comparable humoral responses (binding and neutralizing antibody titres) to a 50µg 78 dose as compared to the full dose of 100µg (11). As far as we know, there has only been one trial reporting 79 on clinical efficacy of fractional dosing in a small number of subjects. This multicentre trial observed 67%-80 97% efficacy of protection against symptomatic COVID-19 in a sub-group of participants who were primed 81 with a half dose and boosted with a full dose of the ChAdOx1 nCoV-19 (Vaxzevria®, AstraZeneca) vaccine 82 (12). In a recent paper, Więcek et al. used modelling data by Khoury et al. to predict vaccine efficacy of 83 fractional doses (13,14). Compared to the 95% efficacy provided by the full dose of 30µg, fractional doses 84 of 10µg and 20µg of BNT162b 2 are estimated to still be 80-90% effective in adults below 55 years old . 85 Together with the dose -sparing advantage, t he possibly fewer side effects caused by lower dosing, as 86 suggested by some clinical data, may help tip the balance in favour of implementing fractional dosing 87 strategies (15). 88 As more data is urgently needed to fill this knowledge gap and guide the potential use of fractional doses 89 of mRNA vaccines, we conducted a randomised controlled trial to determine whether immune responses 90 to a reduced dose of BNT162b2 (20µg) are non -inferior to the full, licensed dose (30µg). We found that 91 binding and neutralizing antibody responses to the reduced dose of 20µg were indeed inferior to the full 92 dose of 30µg of BNT162b2, but nevertheless still markedly higher than responses induced by other 93 approved COVID-19 vaccines with proven efficacy. 94

95 Study design 96 REDU-VAC is a participant-blinded, randomised, phase 4, multicentre, non-inferiority study investigating 97 safety, reactogenicity and immunogenicity of a fractional dose of the mRNA COVID-19 vaccine BNT162b2 98 (Pfizer-BioNTech). Recruitment occurred at five sites across Belgium of the ‘External Department of 99 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 7 Prevention and Protection at work ’ of Mensura. The trial was reviewed and approved by the Erasme 100 Hospital Ethics Committee (P2021/251) and the Federal Agency for Me dicines and Health Products 101 (EudraCT: 2021-002088-23A). The trial was registered at ClinicalTrials.gov (NCT04852861). The protocol is 102 available in S1 Appendix. 103 Two prime-boost schedules are compared consisting, respectively, of two doses of 20µg (reduced dose 104 schedule) and two doses of 30µg (full dose schedule) of BNT162b2. All participants were randomly 105 assigned to receive either the reduced or the full dose schedule. 106 Participants 107 Non-previously vaccinated adults aged 18-55 years old who were in good gener al health or having well-108 controlled co-morbidities, either infection naïve or previously infected, were eligible. Exclusion criteria 109 included history of severe adverse reactions associated with a vaccine or severe allergic reaction to any 110 component of the study intervention, acute severe febrile illness or acute infection , pregnancy and 111 breastfeeding. 112 Randomisation and masking 113 Participants were block -randomised (random block size between 2, 4, 6, and 8, ratio of 1:1:1:1) within 114 female/20µg, female/30µg, mal e/20µg, and male/30µg using the blockrand R package by the study 115 statistician (16). The age distribution in each group was then verified to be not different in the four groups 116 using a Kruskal-Wallis test. The obtained p-value was 0.54. 117 Participants and laboratory staff were masked to the administered vaccine dose (20µg versus 30µg). To 118 ensure participant blinding to the vaccine dose, randomisation lists were kept out of sight, vaccines were 119 prepared, and syringes were filled beforehand. 120 Procedures 121 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 8 Participants were informed of the study and asked to register online before study participation . Those 122 meeting all inclusion criteria were randomly assigned to one of the two study groups by the study 123 statistician and were invited to the baseline visit (day 0) by the study nurses . At the baseline visit, 124 participants provided informed consent before having blood drawn and being vaccinated. 125 Two different dosages of the same COVID-19 mRNA BNT162b2 vaccine were used in the study. BNT162b2 126 is a lipid nanoparticle –formulated, nucleoside-modified mRNA vaccine encoding a prefusion stabilized, 127 membrane-anchored SARS-CoV-2 full-length spike protein. The 20µg and 30µg doses were administered 128 via respectively a 0.2 mL and a 0.3 mL intra-muscular injection into the upper arm. 129 Vaccines were administered by trained study nurses at the different trial sites. Participants were observed 130 for a minimum of 15 minutes after vaccination. The interval between the two vaccine doses was three 131 weeks for all participants. Blood was drawn on days of vaccination (day 0 and day 21), four weeks post 132 second dose (day 49), and six months post first dose (month six). One week after administration of each 133 vaccine dose, participants were asked to record online any experienced local and systemic adverse events 134 as well as their severity (mild/moderate/severe) . Participants were also asked to record breakthrough 135 infections by registering online the date of a positive SARS-CoV-2 molecular test (on the condition that no 136 third vaccine dose had been received yet) , and in case of symptomatic infection, the duration of 137 symptoms. Breakthrough infections were followed-up until administration of a third vaccine dose. 138 SARS-CoV-2 anti -receptor binding domain (RBD) specific IgG concentrations were measured by ELISA 139 (reported as Binding Antibody Units [BAU]/mL) on days 0/21/49 and month 6. Neutralizing antibody titres 140 against SARS -CoV-2 Wuhan (2019-nCoV-Italy-INMI1, reference 169 008V -03893) (days 21/49) , the 141 B.1.617.2 Delta variant (83DJ-1) and the BA.1 Omicron variant (day 49) were measured with a live virus 142 neutralization assay (VNA, reported as 5 0% neutraliz ation, NT 50) (17). The VNA was only performed 143 against Delta and Omicron for samples with an NT50 (Wuhan) titre >50 and >400 respectively. Cellular 144 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 9 responses were measured at day 49 both with IFN-γ enzyme-linked immunosorbent spot assay (ELISpot) 145 as well as with multi-colour flow cytometry on a subsample of 45 randomly selected participants. Detailed 146

are described in the S2 appendix. 147 Previous infection status was established following a decision tree ( S1 Fig). Participants with a previous 148 laboratory-confirmed SARS-CoV-2 infection were considered previously infected irrespective of baseline 149 serology. All other participants with a baseline anti -RBD IgG titre <5 BAU/mL were considered infection 150 naïve, and those ≥ 30 BAU/mL were considered previously infected. Participants with a titre ≥ 5 and < 30 151 BAU/mL were further tested with a different, multiplexed immunoassay (Multi -SARS-CoV-2 152 Immunoassay) detecting four targets: RBD, spike subunit 1 (S1), spike subunit 2 (S2) and nucleocapsid (N) 153 (17). Participants with ≥3 out of 4 targets positive were considered previously infected. 154 Outcomes 155 The primary outcome was serum SARS-CoV-2 anti-RBD IgG concentration 28 days post second dose (day 156 49) in infection naïve participants. Secondary outcomes included reactogenicity, as measured by reported 157 local and systemic adverse events in the week following each vaccination, and safety, as measured by 158 suspected unexpected serious adverse reactio ns, serious adverse reactions, and adverse reactions with 159 grade equal or more than three over the entire study period. 160 Immunological secondary outcomes include SARS-CoV-2 anti-RBD IgG at day 0, day 21 and month 6; VNA 161 titres at days 21 (Wuhan) and 49 (Wuhan, Delta and Omicron); and SARS-CoV-2 specific cell frequencies 162 at day 49 in infection naïve participants only as well as in infection naïve and previously infected 163 participants combined. 164 Statistical analysis 165 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 10 The sample size was calculated assuming a true difference of geometric means of the primary outcome 166 on the log10 scale being 0 between the reduced and the full dose, and a standard deviation of GMT on the 167 log10 scale being 0 .27 (17). A minimum of 50 naïve participants per arm was necessary to achieve 90% 168 power at a two-sided 5% significance level. The geometric mean ratio (GMR) was then calculated as the 169 anti-logarithm of the difference between the mean on the log 10 scale of the primary outcome in the 170 reduced dose and that in the full dose (reference dose), after adjusting by a mi xed-effect linear model 171 using age, gender, and baseline titre of SARS-CoV-2 anti-RBD IgG as fixed factors and study sites as random 172 factor. To conclude on non-inferiority between the two groups, the WHO criterion of 0.67 was used (18), 173 i.e. the reduced dose was considered non -inferior when the lower limit of the two-sided 95% CI of the 174 GMR was greater than this cut-off. In the same way, GMRs were also calculated at day 49 for VNA titres 175 against Wuhan and Delta variants. 176 The proportion of participants with responses higher than the lower limit of detection were calculated for 177 SARS-CoV-2 anti-RBD IgG at days 0, 21 and 49 , and for VNA titres at days 21 (Wuhan) and 49 (Wuhan , 178 Delta, Omicron) with 95% CI calculated by the binomial exact method. They were compared between the 179 reduced and the full dose groups using the Fisher’s exact test. Data below the lower limit of detection 180 were given a value equal to half of the threshold before transformation. Comparisons of primary and 181 secondary outcomes were evaluated by linear mixed-effect model adjusting for age, gender, and baseline 182 titre of SARS -CoV-2 anti-RBD IgG (for days 21 and 49) or naïve/not naïve category (for day 0) as fixed 183 factors and study sites as random factor. Correlatio ns between SARS-CoV-2 anti-RBD IgG and VNA titres 184 (Wuhan), and between VNA against Wuhan and Delta variants were evaluated by Pearson correlation 185 coefficients. Concerning cellular responses, the proportion of participants with responses higher than 0 in 186 IFN-γ ELISpot and with a positive response in flow cytometry was computed at day 49 with 95% CI 187 calculated by the binomial exact method. Fisher’s exact test was used to compare reduced and full dose 188 groups. For IFN-γ ELISpot analysis, data equal to 0 were given a value of 1 before transformation, and for 189 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 11 flow cytometry, negative frequencies were given a value of 0.0001 before transformation. Comparisons 190 of GMTs were evaluated by linear mixed -effect model adjusting for age, gender, and baseline titre of 191 SARS-CoV-2 anti-RBD IgG as fixed factors and study sites as random factor. All statistical analyses were 192 done using R version 4.1.2. 193

194 Between April 19 and April 23, 2021, 152 participants were screened for eligibility in five sites across 195 Belgium among whom 145 were enrolled in the study and randomized (Fig 1). Four participants were 196 excluded from the study due to pregnancy , active infection, high -risk contact or medical reasons . 197 Participants were mostly female (69 .5%), of European origin (95%), with a mean age of 40.4 years (SD 198 7.9). Six participants (4.3%) were taking medication to treat a co-morbidity. Baseline characteristics were 199 well balanced across the two study arms (Table 1). Hundred twenty-four participants were considered 200 SARS-CoV-2 infection naïve and 17 were previously infected at baseline. Here, we focus on the results of 201 the per-protocol (PP) analysis including infection naïve participants only and present the results of the 202 intention-to-treat (ITT) analysis, including both naïve and previously infected participants, in 203 supplementary data. 204 Fig 1. Trial profile. 205 Table 1. Baseline characteristics by cohort and study arm. 206 Per-protocol Intention-to-treat 20 µg 30 µg 20 µg 30 µg Participants, n 60 64 70 71 Age, years Mean (SD) 40.4 (7.5) 41.0 (8.2) 39.8 (7.9) 41.0 (8.0) Median (range) 39.5 (23.0-54.0) 41.5 (24.0-55.0) 39.0 (23.0-54.0) 41.0 (24.0-55.0) Sex Female 43 (71.7%) 43 (67.2%) 49 (70.0%) 49 (69.0%) Male 17 (28.3%) 21 (32.8%) 21 (30.0%) 22 (31.0%) Ethnicity European 56 (93.3%) 63 (98.4%) 65 (92.9%) 69 (97.2%) All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 12 Latin 1 (1.7%) 0 1 (1.4%) 0 Asian 1 (1.7%) 0 1 (1.4%) 1 (1.4%) Sub-Saharan Africa 2 (3.3%) 0 2 (2.9%) 0 Northern-Africa 0 1 (1.6%) 1 (1.4%) 1 (1.4%) Comorbidities Cardiovascular 1 (1.7%) 2 (3.1%) 2 (2.9%) 2 (2.8%) Oncological 0 1 (1.6%) 0 1 (1.4%) Respiratory 0 1 (1.6%) 0 1 (1.4%) BMI <18.5 1 (1.7%) 1 (1.6%) 1 (1.4%) 1 (1.4%) ≥18.5 and <25 39 (65.0%) 36 (56.3%) 45 (64.3%) 40 (56.3%) ≥25 20 (33.3%) 27 (42.1%) 24 (34.3%) 30 (42.3%) Data are n (%) unless otherwise indicated. 207 At the primary endpoint, 28 days post second dose, the GMT of SARS-CoV-2 anti-RBD IgG in the PP cohort 208 was 1705 BAU/ml in the 20µg group and 2387 BAU/ml in the 30µg group (Table 2). The GMR of 0.714 209 (0.540-0.944, 95% CI) indicated that non-inferiority was not demonstrated given that the lower limit of 210 the 95% CI was inferior to the WHO recommended margin of 0 .67. Similarly, the ITT analysis did not 211 demonstrate non-inferiority of the reduced dose either, with GMTs of 1822 BAU/ml and 2381 BAU/ml in 212 the 20µg and 30µg arms, respectively, and a GMR of 0.765 (0.593-0.987, 95% CI) (S1 Table). 213 Table 2. Immune responses by study arm at 28 days post second vaccine dose (Day 49) and non -214 inferiority analysis in the per-protocol cohort. 215 216 Per-protocol 20 µg 30 µg GMR SARS-CoV-2-anti-RBD IgG n 60 64 - GMT in BAU/ml (95% CI) 1705 (1315-2211) 2387 (1899-3001) 0.714 (0.540-0.944) Neutralizing antibodies (Wuhan) n 60 64 - NT50 (95% CI) 160 (124-206) 216 (172-270) 0.740 (0.562-0.974) Neutralizing antibodies (Delta) n 56 63 - NT50 (95% CI) 39 (31-48) 40 (33-49) 0.962 (0.760-1.217) All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 13 Data are geometric mean titres (95% CI) at day 28 post second dose. GMRs (95% CI) were adjusted with a linear 217 mixed-effect model including gender, age and SARS -CoV-2 anti -RBD IgG titre at baseline as fixed variables and 218 location as random variable. In the figure, the dashed line indica tes the WHO recommended non-inferiority margin 219 of 0.67. GMT=geometric mean titre. GMR=geometric mean ratio. BAU=Binding antibody units. NT50=50% 220 neutralizing antibody titre. 221 We observed strong correlations between SARS-CoV-2 binding (anti-RBD IgG) and neut ralizing (Wuhan 222 NT50) antibody titres in both the reduced and the full dose arms (Fig 2A). As a result, GMRs for neutralizing 223 titres were similar to those described above for binding titres. Indeed, GMTs of neutralizing antibody titres 224 against Wuhan were 160 (124-206) and 216 (172 -270) in the 20µg and 30µg arms, respectively, with a 225 GMR of 0.740 (0.562-0.974) (Table 2). In the ITT analysis, GMTs were 216 (95% CI: 170-276) and 279 (224-226 347) in the 20µg and 30µg arms, respectively, with a GMR of 0.776 (0.592-1.017) ( S1 Table). In both 227 analyses, non-inferiority was not demonstrated. 228 Fig 2. Correlations between immune responses per study arm. Correlations were analysed at 28 days after 229 the second vaccine dose between SARS-CoV-2 anti-RBD IgG binding antibodies and SARS-CoV-2 Wuhan neutralizing 230 antibodies (A), and between SARS -CoV-2 Wuhan and Delta variant neutralizing antibodies (B). Pearson correlation 231 coefficients (95% CI) are given per study arm. Ellipses represent the 95% CI for the two study arms (purp le=20µg, 232 black=30µg), assuming multivariate normal distributions. NT50=50% neutralizing antibody titre, RBD=SARS -CoV-2 233 receptor binding domain, BAU=binding antibody units. 234 Neutralizing antibodies against Delta and Omicron were only tested for participants with an NT50 235 (Wuhan) >50 and >400, respectively. As expected, neutralizing capacity against these variants was much 236 lower compared to Wuhan. In the samples tested, just 21/56 (38%) and 25/63 (40%) naïve participants 237 had detectable Delta neutralizing titre s in the 20µg and 30µg arms, respectively (Table 3). For Omicron, 238 none of the naïve participants (0/8 and 0/11 for 20µg and 30µg, respectively) had detectable neutralizing 239 activity. Titres against Wuhan and Delta were strongly correlated in both study arms (Fig 2B). Correlation 240 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 14 between Wuhan and Omicron could not be determined due to the mostly undetectable titres. Non-241 inferiority of the reduced dose to the full dose was demonstrated for Delta neutralizing titres, in the PP 242 and ITT analyses, as the lower limit of the 95% CIs were superior to the WHO margin of 0.67 (Table 2 and 243 S1 Table). As all Omicron neutralizing titres were und etectable in naïve participants, regardless of the 244 study arm, a non-inferiority analysis was not performed. 245 Table 3. Humoral and cellular responses by study arm in the per -protocol cohort at the different time 246 points. 247 Per-protocol 20 µg 30 µg p-value SARS-CoV-2 anti-RBD IgG Day 0 n 60 64 Concentration (BAU/ml) 2.6 (2.3-2.9) 2.8 (2.5-3.0) p=0.421 > 5 BAU/ml 2 (3%, 0.4-12) 4 (6%, 1.7-15) p=0.681 Day 21 n 60 64 Concentration (BAU/ml) 117 (84-162) 223 (166-298) p 5 BAU/ml 59 (98%, 91-100) 64 (100%, 94-100) p=0.484 Day 49 n 60 64 Concentration (BAU/ml) 1705 (1315-2211) 2387 (1899-3001) p=0.019 > 5 BAU/ml 60 (100%, 94-100) 64 (100%, 94-100) p=1.000 Month 6 n 50 60 p=0.007 Concentration (BAU/ml) 157 (113-217) 246 (188-323) > 5 BAU/ml 50 (100%, 93-100) 60 (100%, 94-100) p=1.000 Neutralizing antibodies Wuhan - Day 21 n 60 64 NT50 27 (24-31) 29 (26-33) p=0.355 > 50 7 (12%, 5-23) 10 (16%, 8-27) p=0.606 Wuhan - Day 49 n 60 64 - NT50 160 (124-206) 216 (172-270) p=0.032 > 50 56 (93%, 84-98) 63 (98%, 92-100) p=1.000 Delta - Day 49 n 56 63 NT50 39 (31-48) 40 (33-49) p=0.743 > 50 21 25 p=0.852 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 15 (38%, 25-51) (40%, 28-53) Omicron - Day 49 n 8 11 NT50 25 (25-25) 25 (25-25) - > 50 0 (0%, 0-37) 0 (0%, 0-28) p=1.000 IFN-γ producing cells per million PBMCs SARS-CoV-2 S1-specific - Day 49 n 18 24 Mean number of cells/million 71.5 (33.0-155.1) 105.9 (60.0-187.0) p=0.327 > LOD 12 (67%, 41-87) 19 (79%, 58-93) p=0.812 SARS-CoV-2 S2-specific - Day 49 n 18 24 Mean number of cells/million 88.5 (34.0-230.3) 118.3 (50.9-274.8) p=0.509 > LOD 11 (61%, 36-83) 16 (67%, 45-84) p=1.000 Data are geometric means (95% CI) for continuous variables, and n (%, 95% CI) for binary values. The LLOQ for the 248 SARS-CoV-2 anti-RBD IgG was 5 BAU/mL, NT50 = 50 for the live virus neutralisation assay, and 54 and 66 cells/million 249 PBMCs for S1 and S2, respectively, for ELIspot. For continuous variables, p-values are reported using a linear mixed-250 effect model adjusted for gender, age and baseline infection status (for day 0 data) or baseline SARS-CoV-2 anti-RBD 251 IgG titre (for day 21/49 and month 6 data) as fixed variables and location as random variable. Fisher’s exact test was 252 used to report p -values for binary variables. BAU=Binding antibody units. NT50=50% neutralizing antibody titre. 253 PBMC=peripheral blood mononuclear cell. S1=SARS -CoV-2 spike protein subu nit 1. S2=SARS -CoV-2 spike protein 254 subunit 2. 255 In both the reduced and full dose study arms, all participants had seroconverted by 28 days post second 256 dose ( Table 3, Fig 3A). However, the GMTs of anti -RBD IgG were significantly lower in the 20µg arm 257 compared to the 30µg arm at the time of the second dose administration (day 21) and 28 days later (day 258 49). At six months post first dose anti-RBD IgG titres had waned significantly, but all participants remained 259 seropositive. The GMT of anti -RBD IgG was still significantly lower in the reduced dose arm at this time 260 point ( Table 3, Fig 3A ). In terms of neutralizing activity, 56/60 (93 %) versus 63/64 (98 %) of naïve 261 participants from the 20µg arm and the 30µg arm, respectively, had detectable neutralizing antibody titres 262 against Wuhan, with a significantly lower titre in the reduced dose versus full dose arm 28 days post 263 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 16 second dose (Table 3, Fig 3B). Neutralizing antibodies against Delta and Omicron were only measured at 264 day 49 and did not differ significantly between the two study arms (Table 3, Fig 3C/D). 265 Fig 3. Kinetics of immune responses per study arm of the per -protocol cohort. SARS-CoV-2 anti-RBD IgG 266 binding antibody data are shown on days 0/21/49 and month 6 post first dose (A), SARS -CoV-2 Wuhan neutralizing 267 antibodies on days 21/49 (B) and SARS-CoV-2 Delta (C) and Omicron (D) neutralizing antibodies on day 49 per study 268 arm (purple=20µg, black=30µg). In the case of SARS -CoV-2 anti-RBD IgG binding antibody at month 6, five samples 269 were excluded due to a detected breakthrough infection before month 6. Blue bars indicate geometric mean titres 270 with 95% CI. NT50=50% neutralizing antibody titre, RBD=SARS -CoV-2 receptor binding domain, BAU=binding 271 antibody units. 272 Cellular responses elicited against a pool of SARS-CoV-2 spike protein subunit 1 (S1) and 2 (S2) peptides, 273 measured with IFN-γ ELISpot, were not significantly different either between the reduced and full dose 274 arms, either in the PP or the ITT cohort (Table 3, S2). The proportion of participants with detectable 275 responses was also not significantly different, with 67% (41-87) versus 79% (58-93) for S1 in the 20µg and 276 30 µg arms, respectively , and 61% (36 -83) versus 67% (45 -84) for S2 in t he 20µg and 30 µg arms, 277 respectively (Table 3), in the PP cohort. Similar findings were observed in the ITT cohort (S2 Table). These 278 data were corroborated by f low cytometry, where no differences were observed between the reduced 279 and full dose arms in the proportion of S1 or S2 specific CD4+ and CD8+ T cells expressing CD154 (CD4+ T 280 cells only), IFN-γ, IL-2, and TNF-α (S2 Fig). 281 A total of 18 breakthrough infections (BTI) were reported and all occurred between five to eight months 282 after the first dose, coinciding with a major infection wave in the country due to the Delta variant ( S3 283 Table). These BTIs occurred exclusively in infection naïve participants and were as likely to occur in the 284 20µg arm as in the 30µg arm (Fisher’s exact test, p=0.80). GMTs of binding and neutralizing antibodies at 285 28 days post second dose or six months post first dose were not significantly different between those 286 experiencing and not experiencing a BTI (Wilcoxon rank sum test, p>0.14). 287 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 17 No serious adverse events were reported during the study period. No difference in frequency or severity 288 of adverse events reported after each vaccine dose was observed between the 20µg and the 30µg arm 289 (S4 Table, S3 Fig). The only exception was the reported severity of nausea after the second dose which 290 was mostly moderate in the 20µg arm and mild in the 30µg arm. 291

292 The study results show that the administration of a reduced dose (2x20µg) of the mRNA vaccine BNT162b2 293 induces lower titres of SARS-CoV-2 anti-RBD IgG binding and SARS-CoV-2 Wuhan neutralizing antibodies 294 than the administration of the full vaccine dose (2x30µg). No such difference was observed for SARS-CoV-295 2 Delta neutralizing antibodies but this can likely be explained by the overall lower titres as compared to 296 Wuhan, with many participants having undetectable neutralizing antibody titres post vaccination. For 297 Omicron neutraliz ation, none of the naïve part icipants had detectable titres, precluding a proper 298 comparison between study arms but highlighting the lack of induction of neutralizing capacity against the 299 Omicron variant with two doses of BNT162b2. 300 The ratios of the GMTs across all analyses indicate that the investigated reduced dose regimen induced 301 around 25-30% lower antibody titres than the licensed full dose regimen. The magnitude of this reduction 302 and its potential impact on vaccine efficacy needs to be put into perspective , however. A recent non-303 inferiority trial compared immunogenicity of Pfizer-BioNTech’s BNT162b2 with Astra-Zeneca’s adenoviral 304 ChadOx1 nCoV-19 vaccine, amongst other vaccination regimens. At 28 days post second dose, the GMT 305 of anti-spike binding IgGs was 90% lower for ChadOx1 nCoV-19 than for BNT162b2 (19). Despite ChadOx1 306 nCoV-19’s markedly lower humoral responses, it still has been sho wn to have an efficacy against 307 symptomatic infection of about 70%, and it has been approved for use in 182 countries with more than 308 two billion doses administered , primarily in LIC (12,20). We therefore argue that the moderately lower 309 humoral response of the reduced dose regimen investigated in our study is still excellent , and likely 310 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 18 provides significant protection against COVID-19 with additional benefit as compared to the non-mRNA 311 vaccines currently in use in most low and middle -income countries (13,14). Larger immunogenicity and 312 effectiveness trials are warranted to support this notion. 313 In contrast to the humoral responses , we did not observe any significant differences in the cellular 314 immune responses to vaccination. We used both ELISpot and flow cytometry to quantify SARS-CoV-2 spike 315 protein subunits 1 and 2 specific cells, both showing very similar frequencies between study arms. This 316 was not entirely unexpected, considering that differences in cellular responses have been reported before 317 to be less pronounced than those in humoral responses (19). The equivalent cellular immune responses 318 observed in this study further support the notion that this reduced dosage likely induces potent immunity. 319 While acknowledging that this trial was not designed nor powered to study efficacy of protection, we did 320 not observe an obvious difference in incidence of breakthrough infections between both study arms. It is 321 important to consider, however, that these infections were most likely caused by the Delta variant of 322 SARS-CoV-2, around six months after vaccination, while neutralizing titres against the Delta variant at 28 323 days post second dose were already very low for both the reduced and full dose arms. In other words, a 324 possible difference in protection may have gone unnoticed due to a generally low incidence of infection 325 during the first months following vaccination. 326 Based on waning antibody data and the emergence of new variants, high-income countries are providing 327 third doses to the general population , and several countries have started administering fourth doses to 328 specific groups with comorbidities . In the meantime, primary vaccination coverage in LIC remai ns very 329 low. The COVAX program was founded in April 2020 with the specific aim to provide global equitable 330 access to COVID-19 vaccines. The program has delivered approx imately one billion vaccine doses to 144 331 participating countries, but current production capacity for COVID-19 vaccines does not cover the global 332 needs, thus delaying the end of the pandemic. We acknowledge that the use of fractional doses is 333 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 19 associated with a number of programmatic and operational challenges which may hamper its roll -out, 334 such as the possible need for special syringes, adjustments of diluent volumes, or exacerbated vaccine 335 hesitancy due to lower antibody levels (22,23). Nevertheless, we believe that these challenges are rather 336 limited and can still be solved relatively quickly for the most part. This is in contrast to m ore structural 337 strategies to speed up vacci nation rates, such as strengthening logistics and building local vaccine 338 production capacity, which are important but will arguably take more time to have an impact. 339 This study has several limitations, the first being the relatively limited sample size. A larger number of 340 participants would have resulted in smaller confidence intervals around the GMRs, which might have 341 impacted the conclusions on non-inferiority. Although males were underrepresented in this study, we do 342 not believe this is a major limitati on as immune responses to COVID -19 mRNA vaccination in healthy, 343 younger subjects are only minimally gender-dependent and importantly, there is no basis to assume that 344 fractional dosing would affect immune responses differently between males and females (24–26). 345 Secondly, the small proportion of previously infected participants in our study population (17/144, 12%) 346 does not allow for a separate sensitivity analysis in this group. With record high COVID -19 incidences 347 worldwide, the proportion of the population who experienced a past infection is rapidly growing, making 348 analyses including previously infected people ever more relevant. In addition, breakthrough infections 349 were not actively monitored by regular molecular testing. Therefore, we may have missed asymptomatic 350 infections, which are not re ported by the study participants. Thir dly, while protection from infection or 351 disease has been convincingly correlated with titres of binding and neutrali zing antibodies, as discussed 352 previously, it is not possible to determine with certainty that the moderately lower titres observed in our 353 study will translate to equally moderately lower efficacy of protection (14,27). Fourth, while the interval 354 of three weeks between both vaccine doses is recommended by the manufacturer, larger intervals are 355 commonly adopted and have been shown to induce higher immune responses as compared to the shorter 356 three week interval (28,29). Finally, a strength of t he study in terms of global health relevance, is the 357 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 20 relatively young age distribution of our cohort, ranging from 23 to 55 years old (median: 40). This makes 358 our data especially relevant for the context where it can be applied most, i.e. LIC where age dis tribution 359 is typically much lower than in high income countries (97% versus 81% of population below 65 years old, 360 respectively) (30). Nevertheless, further studies are needed including participants who are more reflective 361 of the target population in terms of genetic background and environment. 362 To our knowledge, this is the only non-inferiority study reporting on fractional dosing of a COVID-19 mRNA 363 vaccine so far . Our study shows moderately lower humoral responses and similar cellular immune 364 responses after two reduced doses of 20µg compared to two full doses of 30µg of the BNT162b2 mRNA 365 vaccine. Nevertheless, antibody responses to the reduced dose remain far supe rior to what has been 366 published in the literature so far for marketed adenoviral vector vaccines with proven efficacy of 367 protection against disease. Considering this and the important potential in accelerating vaccination rates 368 in LIC , our findings support the need for larger non -inferiority trial s on fractional dosing for primary 369 vaccination as well as for follow-up booster vaccinations. The relevance of our findings goes beyond the 370 current pandemic, as they highlight the potential of fractional dosing as a dose-sparing strategy in future 371 epidemics and vaccine shortages. 372 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 21 Author Contributions 373 Conceptualization: Maria E Goossens, Isabelle Desombere, Kevin K Ariën, Arnaud Marchant, 374 Pieter Pannus. 375 Data Curation: Kristof Y Neven, Stéphanie Depickère, Pieter Pannus, Sarah Houben. 376 Formal Analysis: Stéphanie Depickère, Pieter Pannus, Kristof Y Neven. 377 Funding Acquisition: Maria E Goossens, Katelijne Dierick 378 Investigation: Pieter Pannus, Kristof Y Neven, Leo Heyndrickx, Johan Michiels, Elisabeth Willems, 379 Stéphane De Craeye, Antoine Francotte, Félicie Chaumont, Véronique Olislaghers, Delphine 380 Kemlin, Alexandra Waegemans, Kevin K Ariën, Isabelle Desombere, Arnaud Marchant. 381 Methodology: Pieter Pannus, Maria E Goossens, Kevin K Ariën, Isabelle Desombere, Arnaud 382 Marchant. 383 Project Administration: Maria E Goossens, Pieter Pannus. 384 Resources: Kristof Y Neven, Pieter Pannus, Mathieu Verbrugge, Marie -Noëlle Schmickler, Maria 385 E Goossens, Kevin K Ariën, Isabelle Desombere, Arnaud Marchant. 386 Software: Stéphanie Depickère. 387 Supervision: Maria E Goossens, Pieter Pannus. 388 Validation: Pieter Pannus, Stéphanie Depickère, Kristof Y Neven, Kevin K Ariën, Isabelle 389 Desombere, Arnaud Marchant. 390 Visualization: Stéphanie Depickère, Pieter Pannus. 391 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 22 Writing – Original Draft Preparation: Pieter Pannus, Stéphanie Depickère. 392 Writing – Review & Editing: Pieter Pannus, Stéphanie Depickère, Delphine Kemlin, Sarah 393 Houben, Kristof Y Neven, Leo Heyndrickx, Véronique Olislagers, Alexandra Waegemans, Mathieu 394 Verbrugghe, Marie-Noëlle Schmickler, Steven Van Gucht, Katelijne Dierick, Arnaud Marchant , 395 Maria E Goossens, Isabelle Desombere, Kevin K Ariën. 396 Declaration of interests 397 The authors declare no conflicts of interest. 398

399 This study is funded by the Belgian Federal Government through Sciensano [COVID -19_SC004, COVID-400 19_SC059, COVID-19_SC095, COVID-19_SC096] and the Research Foundation - Flanders [grant number 401 FWO G0G4220N to KKA]. DK is a Clinical Master Specialist Applicant to a Ph.D. of the Fonds de la Recherche 402 Scientifique – FNRS. A.M. is Research Director of the FRS -FNRS, Belgium. We thank Martine Delaere, 403 Kristine Massez, Isabelle Maufort and Jody Serré for their dedicated work as study nurses. We also thank 404 Caroline Rodeghiero, Fabienne Jurion, Elfriede Heerwegh, Celien Van Oostveldt, Vincent Martens, Valérie 405 Acolty, Inès Vu Duc, Sara Cuman, Robin Van Naemen, Maria Lara Escandell, Sandra Coppens, Ann 406 Ceulemans and Koen Bartholomeeusen for their help in the laboratory , as well as Murat Gonen and 407 Jonathan Masala for their logistical support. We thank Piet Maes (Rega Institute, KU Leuven, Belgium) to 408 kindly provide the B.1.617.2 Delta variant (83DJ-1) isolate. Finally, we thank all study participants for their 409 availability, flexibility and dedication to the study. 410 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 23

411 1. Global Dashboard for Vaccine Equity [Internet]. UNDP Data Futures Platform. [cited 2022 Jan 11]. 412 Available from: https://data.undp.org/vaccine-equity/ 413 2. COVID vaccines: Widening inequality and millions vulnerable [Internet]. UN News. 2021 [cited 414 2021 Dec 7]. Available from: https://news.un.org/en/story/2021/09/1100192 415 3. Global Dashboard for Vaccine Equity [Internet]. UNDP Data Futures Platform. [cited 2021 Dec 7]. 416 Available from: https://data.undp.org/vaccine-equity/ 417 4. Current health expenditure (CHE) [Internet]. [cited 2021 Dec 7]. Available from: 418 https://www.who.int/data/gho/data/themes/topics/indicator-groups/indicator-group-419 details/GHO/current-health-expenditure-(che) 420 5. COVID-19 Vaccine Market Dashboard [Internet]. [cited 2021 Dec 7]. Available from: 421 https://www.unicef.org/supply/covid-19-vaccine-market-dashboard 422 6. Mashunye TR, Ndwandwe DE, Dube KR, Shey M, Shelton M, Wiysonge CS. Fractional dose 423 compared with standard dose inactivated poliovirus vaccine in children: a systematic review and 424 meta-analysis. The Lancet Infectious Diseases. 2021 Aug 1;21(8):1161–74. 425 7. Juan-Giner A, Kimathi D, Grantz KH, Hamaluba M, Kazooba P, Njuguna P, et al. Immunogenicity and 426 safety of fractional doses of yellow fever vaccines: a randomised, double-blind, non-inferiority 427 trial. The Lancet. 2021 Jan 9;397(10269):119–27. 428 8. World Health Organization. WHO position on the use of fractional doses – June 2017, addendum 429 to vaccines and vaccination against yellow fever WHO: Position paper – June 2013. Vaccine. 2017 430 Oct 13;35(43):5751–2. 431 9. Resik S, Tejeda A, Lago PM, Diaz M, Carmenates A, Sarmiento L, et al. Randomized controlled 432 clinical trial of fractional doses of inactivated poliovirus vaccine administered intradermally by 433 needle-free device in Cuba. J Infect Dis. 2010 May 1;201(9):1344–52. 434 10. Sahin U, Muik A, Vogler I, Derhovanessian E, Kranz LM, Vormehr M, et al. BNT162b2 vaccine 435 induces neutralizing antibodies and poly-specific T cells in humans. Nature. 2021 436 Jul;595(7868):572–7. 437 11. Chu L, McPhee R, Huang W, Bennett H, Pajon R, Nestorova B, et al. A preliminary report of a 438 randomized controlled phase 2 trial of the safety and immunogenicity of mRNA-1273 SARS-CoV-2 439 vaccine. Vaccine. 2021 May 12;39(20):2791–9. 440 12. Voysey M, Clemens SAC, Madhi SA, Weckx LY, Folegatti PM, Aley PK, et al. Safety and efficacy of 441 the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an interim analysis of four 442 randomised controlled trials in Brazil, South Africa, and the UK. The Lancet. 2021 Jan 443 9;397(10269):99–111. 444 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 24 13. Więcek W, Ahuja A, Chaudhuri E, Kremer M, Gomes AS, Snyder CM, et al. Testing fractional doses 445 of COVID-19 vaccines. PNAS [Internet]. 2022 Feb 22 [cited 2022 Feb 9];119(8). Available from: 446 https://www.pnas.org/content/119/8/e2116932119 447 14. Khoury DS, Cromer D, Reynaldi A, Schlub TE, Wheatley AK, Juno JA, et al. Neutralizing antibody 448 levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection. Nat 449 Med. 2021 Jul;27(7):1205–11. 450 15. Ramasamy MN, Minassian AM, Ewer KJ, Flaxman AL, Folegatti PM, Owens DR, et al. Safety and 451 immunogenicity of ChAdOx1 nCoV-19 vaccine administered in a prime-boost regimen in young and 452 old adults (COV002): a single-blind, randomised, controlled, phase 2/3 trial. Lancet. 2021 Dec 453 19;396(10267):1979–93. 454 16. Snow G. blockrand: Randomization for Block Random Clinical Trials [Internet]. 2020 [cited 2021 455 Dec 10]. Available from: https://CRAN.R-project.org/package=blockrand 456 17. Pannus P, Neven KY, De Craeye S, Heyndrickx L, Vande Kerckhove S, Georges D, et al. Poor 457 antibody response to BioNTech/Pfizer COVID-19 vaccination in SARS-CoV-2 naïve residents of 458 nursing homes. Clin Infect Dis. 2021 Dec 3;ciab998. 459 18. World Health Organization. Guidelines on clinical evaluation of vaccines: regulatory expectations. 460 Proposed revision of WHO TRS 924, Annex 1. [Internet]. 2016 [cited 2022 Nov 1]. Available from: 461 https://www.who.int/biologicals/BS2287_Clinical_guidelines_final_LINE_NOs_20_July_2016.pdf 462 19. Liu X, Shaw RH, Stuart ASV, Greenland M, Aley PK, Andrews NJ, et al. Safety and immunogenicity of 463 heterologous versus homologous prime-boost schedules with an adenoviral vectored and mRNA 464 COVID-19 vaccine (Com-COV): a single-blind, randomised, non-inferiority trial. The Lancet. 2021 465 Sep 4;398(10303):856–69. 466 20. Two billion doses of AstraZeneca’s COVID-19 vaccine supplied to countries across the world less 467 than 12 months after first approval [Internet]. [cited 2022 Jan 13]. Available from: 468 https://www.astrazeneca.com/media-centre/press-releases/2021/two-billion-doses-of-469 astrazenecas-covid-19-vaccine-supplied-to-countries-across-the-world-less-than-12-months-after-470 first-approval.html 471 21. Keeton R, Tincho MB, Ngomti A, Baguma R, Benede N, Suzuki A, et al. T cell responses to SARS-472 CoV-2 spike cross-recognize Omicron. Nature. 2022 Jan 31; 473 22. Wilder-Smith A, Desai S, Cravioto A, Nohynek H, Hombach J. Caution before fractionating COVID-474 19 vaccines. Nat Med. 2021 Nov;27(11):1856–7. 475 23. Wolff J, Atuire C, Bhan A, Emanuel E, Faden R, Ghimire P, et al. Ethical and policy considerations for 476 COVID-19 vaccination modalities: delayed second dose, fractional dose, mixed vaccines. BMJ 477 Global Health. 2021 May 1;6(5):e005912. 478 24. Terpos E, Trougakos IP, Apostolakou F, Charitaki I, Sklirou AD, Mavrianou N, et al. Age-dependent 479 and gender-dependent antibody responses against SARS-CoV-2 in health workers and 480 octogenarians after vaccination with the BNT162b2 mRNA vaccine. American Journal of 481 Hematology. 2021;96(7):E257–9. 482 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 25 25. Wheeler SE, Shurin GV, Yost M, Anderson A, Pinto L, Wells A, et al. Differential Antibody Response 483 to mRNA COVID-19 Vaccines in Healthy Subjects. Microbiology Spectrum [Internet]. 2021 Aug 4 484 [cited 2022 Feb 14]; Available from: https://journals.asm.org/doi/abs/10.1128/Spectrum.00341-21 485 26. Pellini R, Venuti A, Pimpinelli F, Abril E, Blandino G, Campo F, et al. Initial observations on age, 486 gender, BMI and hypertension in antibody responses to SARS-CoV-2 BNT162b2 vaccine. 487 eClinicalMedicine [Internet]. 2021 Jun 1 [cited 2022 Feb 14];36. Available from: 488 https://www.thelancet.com/journals/eclinm/article/PIIS2589-5370%2821%2900208-489 X/fulltext#seccesectitle0015 490 27. Cromer D, Steain M, Reynaldi A, Schlub TE, Wheatley AK, Juno JA, et al. Neutralising antibody titres 491 as predictors of protection against SARS-CoV-2 variants and the impact of boosting: a meta-492 analysis. The Lancet Microbe. 2022 Jan;3(1):e52–61. 493 28. Payne RP, Longet S, Austin JA, Skelly DT, Dejnirattisai W, Adele S, et al. Immunogenicity of standard 494 and extended dosing intervals of BNT162b2 mRNA vaccine. Cell. 2021 Nov 11;184(23):5699-495 5714.e11. 496 29. Grunau B, Goldfarb DM, Asamoah-Boaheng M, Golding L, Kirkham TL, Demers PA, et al. 497 Immunogenicity of Extended mRNA SARS-CoV-2 Vaccine Dosing Intervals. JAMA. 2022 Jan 498 18;327(3):279–81. 499 30. Population ages 65 and above (% of total population) | Data [Internet]. [cited 2022 Jan 19]. 500 Available from: https://data.worldbank.org/indicator/SP.POP.65UP.TO.ZS 501 502 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 152 individuals screened for eligibility 145 randomly assigned 7 excluded 1 on immunosuppressive medication 6 could not attend first study visit 3 excluded 1 positive SARS-CoV-2 PCR, before 1st dose 1 medical reason, before first dose 1 pregnant, before second dose 72 in 30µg cohort 70 in 20µg cohort  60 naive  10 previously infected 71 in 30µg cohort  64 naive  7 previously infected 1 excluded 1 SARS-CoV-2 PCR positive high risk contact case, before first dose 73 in 20µg cohort Figure 1 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 30µg: R = 0.89 (0.82 to 0.93) 20µg: R = 0.84 (0.75 to 0.90) 10 100 1000 10000 100 1000 10000 SARS−CoV−2 anti−RBD IgG, BAU/mL NT50 (Wuhan) A 30µg: R = 0.77 (0.64 to 0.85) 20µg: R = 0.74 (0.59 to 0.84) 10 100 1000 10 100 1000 NT50 (Wuhan) NT50 (Delta) B Figure 2 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint day 0 day 21 day 49 month 6 p = 0.421 p < 0.001 p = 0.019 p = 0.016 1 10 100 1000 10000 SARS−CoV−2 anti−RBD IgG, BAU/mL A day 21 day 49 p = 0.355 p = 0.032 10 100 1000 10000 NT50 (Wuhan) B day 49 p = 0.743 10 100 1000 10000 NT50 (Delta) C day 4910 100 1000 10000 NT50 (Omicron) D Figure 3 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint SUPPLEMENTARY MATERIAL page S1 Fig. Decision tree to determine previous SARS-CoV-2 infection status at baseline. 2 S1 Table. Immune responses by study arm at 28 days post second vaccine dose (Day 49) and non-inferiority analysis in the intention-to-treat cohort. 3 S2 Table. Humoral and cellular responses at the different time points by study arm in the intention-to-treat cohort. 4-5 S2 Fig. Flow cytometry cellular data in the per-protocol cohort at day 49 (28 days after second dose). 6 S3 Table. Breakthrough infections. 7 S4 Table. Local and systemic adverse events in intention-to-treat and per-protocol cohorts. 8-11 S3 Fig. Adverse events. Reported local (A) and systemic (B) adverse events after the first and second vaccine dose, according to severity (mild/moderate/severe) and by study arm (20µg and 30µg) in the intention-to-treat cohort. 12 S1 Appendix: Study protocol 13-24 S2 Appendix: Methods 25 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 2 141 randomized participants Previous positive PCR/serology test results? Anti-RBD ELISA < 5 BAU/ml? Anti-RBD ELISA ≥ 30 BAU/ml? MIA ≥ 3 targets positive? Previously infected individuals (17) Naive individuals (124) No (129) yes (12) yes (3) yes (2) yes (118) No (11) No (8) No (6) S1 Fig. Decision tree to determine previous SARS-CoV-2 infection status at baseline. RBD=SARS- CoV-2 receptor binding domain, ELISA=Enzyme linked immunosorbent assay, BAU=binding antibody units, MIA=multiplex immunoassay. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 3 S1 Table. Immune responses by study arm at 28 days post second vaccine dose (Day 49) and non- inferiority analysis in the intention-to-treat cohort. Intention-to-treat 20 µg 30 µg GMR SARS-CoV-2 anti-RBD IgG n 70 71 - GMT in BAU/ml (95% CI) 1822 (1451-2289) 2381 (1937-2927) 0.765 (0.593-0.987) Neutralizing antibodies (Wuhan) n 70 71 - NT50 (95% CI) 216 (170-276) 279 (224-347) 0.776 (0.592-1.017) Neutralizing antibodies (Delta) n 66 70 - NT50 (95% CI) 50 (40-62) 52 (43-63) 0.965 (0.758-1.227) Data are geometric mean titres (95% CI) at day 28 post second dose. GMRs (95% CI) were adjusted with a linear mixed-effect model including gender, age and SARS-CoV-2 anti-RBD IgG titre at baseline as fixed variables and location as random variable. In the figure, the dashed line indicates the WHO recommended non-inferiority margin of 0.67. GMT=geometric mean titre. GMR=geometric mean ratio. BAU=Binding antibody units. NT50=50% neutralizing antibody titre. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 4 S2 Table. Humoral and cellular responses by study arm in the intention-to-treat cohort at the different time points. Intention-to-treat 20 µg 30 µg p-value SARS-CoV-2 anti-RBD IgG Day 0 n 70 71 Concentration (BAU/ml) 10.9 (9.1-12.9) 9.9 (8.3-11.8) p=0.310 > 5 BAU/ml 12 (17%, 9-28) 10 (14%, 7-24) p=0.649 Day 21 n 70 71 Concentration (BAU/ml) 173 (125-240) 296 (218-400) p=0.001 > 5 BAU/ml 69 (99% 92-100) 71 (100%, 95-100) p=0.497 Day 49 n 70 71 Concentration (BAU/ml) 1822 (1451-2289) 2381 (1937-2927) p=0.040 > 5 BAU/ml 70 (100%, 95-100) 71 (100%, 95-100) p=1.000 Month 6 n 59 67 Concentration (BAU/ml) 190 (143-254) 272 (212-349) p=0.019 > 5 BAU/ml 59 (100%, 94-100) 67 (100%, 95-100) p=1.000 Neutralizing antibodies Wuhan - Day 21 N 70 71 NT50 45 (36-56) 47 (39-58) p=0.677  50 17 (24%, 15-36) 16 (23%, 13-34) p=0.844 Wuhan - Day 49 N 70 71 NT50 216 (170-276) 279 (224-347) p=0.066 > 50 66 (94%, 86-98) 70 (99%, 92-100) p=0.209 Delta - Day 49 n 66 70 - NT50 50 (40-62) 52 (43-63) p=0.767 > 50 31 (47%, 35-60) 31 (44%, 32-57) p=0.863 Omicron - Day 49 n 18 17 NT50 37 (28-48) 40 (30-54) p=0.612 > 50 8 (44%, 22-69) 5 (29%, 10-56) p=0.489 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 5 S2 Table (continued). Humoral and cellular responses by study in the intention-to-treat cohort at the different time points . Intention-to-treat 20 µg 30 µg p-value IFN-γ producing cells per million PBMCs (ELISpot) SARS-CoV-2 S1-specific - Day 49 n 19 26 Mean number of cells/million 72.7 (35.8-147.8) 116.5 (67.3-201.6) p=0.219 > LOD 13 (68%, 43-87) 21 (81%, 61-93) p=0.818 SARS-CoV-2 S2-specific - Day 49 n 19 26 Mean number of cells/million 96.4 (39.8-233.1) 127.4 (58.1-279.1) p=0.494 > LOD 12 (63%, 38-84) 18 (69%, 48-86) p=1.000 SARS-CoV-2 specific T-cells (Flow cytometry) CD4+ (S1) – Day 49 n 19 26 % 0.086 (0.050-0.147) 0.105 (0.063-0.177) p=0.367 > 0.0001% 19 (100%, 82-100) 26 (100%, 87-100) p=1.000 CD8+ (S1) - Day 49 N 19 26 % 0.010 (0.001-0.113) 0.011 (0.001-0.116) p=0.920 > 0.0001% 14 (74%, 49-91) 16 (62%, 41-80) p=0.813 CD4+ (S2) - Day 49 N 19 26 % 0.092 (0.051-0.165) 0.133 (0.076-0.235) p=0.140 > 0.0001% 19 (100%, 82-100) 26 (100%, 87-100) p=1.000 CD8+ (S2) - Day 49 N 19 26 % 0.035 (0.004-0.339) 0.038 (0.004-0.345) p=0.931 > 0.0001% 13 (68%, 43-87) 18 (69%, 48-86) p=1.000 Data are geometric means (95% CI) for continuous variables, and n (%, 95% CI) for binary values. The LLOQ for the SARS-CoV-2 anti-RBD IgG titre was 5 BAU/mL, NT50=50 for the live virus neutralization assay, 54 and 66 cells/million PBMCs for S1 and S2, respectively, for ELIspot, and 0.0001% for flow cytometry. The mean in IFN-γ ELISpot was obtained from three replicate values. For continuous variables, p-values are reported using a linear mixed-effect model adjusted for gender, age and baseline infection status (for day 0 data) or baseline SARS-CoV-2 anti-RBD IgG titre (for day 21/49 and month 6 data) as fixed variables and location as random variable. Fisher’s exact test was used to report p-values for binary variables. BAU=Binding antibody units. NT50=50% neutralizing antibody titre. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 6 S2 Fig. Flow cytometry cellular data in the per-protocol cohort at day 49 (28 days after second dose). SARS-CoV- 2 spike protein subunit 1 and 2 (S1 and S2) specific T cell frequencies were measured in 42 infection naïve participants. A. Percentage of CD4+ and CD8+ T cells expressing CD154 (only in CD4), IFN-γ, IL-2, and TNF-α are depicted. Any: percentage of cells positive for at least on the activation markers. A circle represents one test subject; the GMTs (95% CI) are shown in blue. LLOQ was fixed to 0.0001 and represented by a grey dashed line. B. Percentages are given for CD4+ and CD8+ T cells stimulated with S1 Wuhan or S2 Wuhan expressing at least one of the activation markers (corresponding to “any”). For continuous variables, p-values are reported using a linear mixed-effect model adjusted for gender, age and baseline SARS-CoV-2 anti-RBD IgG titre as fixed variables and location as random variable. Fisher’s exact test was used to report p-values for binary variables. 20 µg 30 µg p-value CD4+ (S1) - Day 49 n 18 24 % 0.075 (0.047-0.120) 0.090 (0.063-0.129) p=0.447 > 0.0001% 18 (100%, 81-100) 24 (100%, 86-100) p=1.000 CD8+ (S1) - Day 49 n 18 24 % 0.007 (0.001-0.050) 0.005 (0.001-0.021) p=0.732 > 0.0001% 14 (78%, 52-94) 14 (58%, 37-78) 0.628 20 µg 30 µg p-value CD4+ (S2) - Day 49 n 18 24 % 0.085 (0.052-0.142) 0.130 (0.089-0.188) p=0.116 > 0.0001% 18 (100%, 81-100) 24 (100%, 86-100) p=1.000 CD8+ (S2) - Day 49 n 18 24 % 0.005 (0.001-0.037) 0.006 0.001-0.025) p=0.907 > 0.0001% 12 (67%, 41-87) 16 (67%, 45-84) p=1.000 A B All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 7 S3 Table. Breakthrough infections # Baseline status Time after 1st dose (months) Symptom duration (days) Anti-RBD IgG (BAU/mL) Day 49 NT50 (Wuhan) Day 49 NT50 (Delta) Day 49 Anti-RBD IgG (BAU/mL) Month 6 20µg study arm 1 Naïve 5.2 14 3886 283 50 7601 2 Naïve 6.3 7 525 25 N/A 23 3 Naïve 7.1 12 603 66 25 58 4 Naïve 6.2 14 2816 429 66 1732 5 Naïve - 0 616 75 25 2787 6 Naïve 6.3 6 1516 151 25 N/A 7 Naïve 7.1 3 2097 244 25 61 8 Naïve 5.4 8 797 126 25 1264 9 Naïve 6.9 5 694 82 25 131 10 Naïve 6.2 8 4324 504 79 N/A Average* (BTI) 6.3 (6.2-6.9) 8.0 (6.0-12.0) 1438 (763-2712) 150 (71-316) 36 (23-54) 57 (18-178) Average* (non-BTI) / / 1830 (1448-2311) 163 (129-205) 40 (33-48) 166 (126-218) 30µg study arm 11 Naïve 6.9 5 9722 566 107 769 12 Naïve 7.3 3 3160 320 100 214 13 Naïve 6.5 12 2744 184 25 422 14 Naïve 6.8 10 1578 174 25 224 15 Naïve 6.2 12 3609 350 67 434 16 Naïve 7.2 N/A 1455 123 25 139 17 Naïve 6.3 2 1554 141 25 149 18 Naïve 6.3 7 2496 200 25 197 Average* (BTI) 6.7 (6.3-7.0) 7.0 (4.0-11.0) 2686 (1593-4528) 227 (147-350) 40 (23-71) 251 (143-443) Average* (non-BTI) / / 2370 (1941-2894) 213 (176-257) 40 (34-48) 241 (194-298) Subjects with a breakthrough infection (BTI) after day 49 per study arm. All BTIs were confirmed with a positive molecular test, except for subject 5 who remained asymptomatic and had a strongly elevated anti-RBD IgG titre at month 6, revealing a BTI. Binding (anti-RBD IgG) antibody titres are given at day 49 and six months post first dose. Neutralizing (NT50) antibody titres are given at day 49 (28 days post second dose). Titres in bold red indicate a breakthrough infection between vaccination and month 6 and are excluded from GMT calculation (date of infection was considered as minimum two days before the date of the positive PCR). *Indicates the median (IQR) for ‘Time after 1st dose’ and ‘Symptom duration’, and geometric mean titre (95% CI) for ‘Anti-RBD IgG’ and ‘NT50’. BTI=breakthrough infection. BAU=Binding antibody units. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 8 Intention-to-treat Naive 20 µg 30 µg p-value 20 µg 30 µg p-value Participants, n 70 71 60 64 Total number of local adverse events mild 47 47 0.95* 41 42 0.97* moderate 46 44 39 41 severe 5 6 5 4 Total number of systemic adverse events mild 38 42 0.63* 29 39 0.23* moderate 87 76 74 68 severe 44 37 43 33 Local adverse events, dose 1 none 27 (39%) 29 (41%) 0.86 23 (38%) 26 (41%) 0.86 pain mild 23 (33%) 24 (34%) 1* 19 (32%) 22 (34%) 0.90* moderate 16 (23%) 15 (21%) 14 (23%) 14 (22%) severe 1 (1%) 2 (3%) 1 (2%) 1 (2%) total 40 (57%) 41 (59%) 1 34 (57%) 37 (58%) 1 redness mild 2 (3%) 1 (1%) 1* 2 (3%) 1 (2%) 1* moderate 1 (1%) 0 1 (2%) 0 total 3 (4%) 1 (1%) 0.37 3 (5%) 1 (2%) 0.35 swelling mild 4 (6%) 3 (4%) 1* 3 (5%) 2 (3%) 1* moderate 3 (4%) 2 (3%) 2 (3%) 2 (3%) total 7 (10%) 5 (7%) 0.56 5 (8%) 4 (6%) 0.74 swollen glands mild 1 (1%) 1 (1%) 1 1 (2%) 1 (2%) 1 Local adverse events, dose 2 none 30 (42%) 29 (41%) 0.87 24 (40%) 26 (41%) 1 pain mild 12 (17%) 13 (18%) 0.94* 12 (20%) 12 (19%) 0.53* moderate 21 (30%) 23 (32%) 17 (28%) 21 (33%) severe 3 (4%) 2 (3%) 3 (5%) 1 (2%) total 36 (51%) 38 (54%) 0.87 32 (53%) 34 (53%) 1 redness mild 1 (1%) 2 (3%) 1* 1 (2%) 2 (3%) 1* severe 0 1 (1%) 0 1 (2%) total 1 (1%) 3 (4%) 0.62 1 (2%) 3 (5%) 0.62 swelling mild 3 (4%) 3 (4%) 1* 2 (3%) 2 (3%) 1* moderate 3 (4%) 3 (4%) 3 (5%) 3 (5%) severe 0 1 (1%) 0 1 (2%) total 6 (9%) 7 (10%) 1 5 (8%) 6 (9%) 1 swollen glands mild 1 (1%) 0 1* 1 (2%) 0 1* moderate 2 (3%) 1 (1%) 2 (3%) 1 (2%) severe 1 (1%) 0 1 (2%) 0 total 4 (6%) 1 (1%) 0.21 4 (7%) 1 (2%) 0.20 S4 Table. Local and systemic adverse events in intention-to-treat and naïve only cohorts. Fisher’s exact test was used to report p-values. *test between the levels of the scale of the present symptoms (mild/moderate/severe). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 9 Intention-to-treat Naive 20 µg 30 µg p-value 20 µg 30 µg p-value Participants, n 70 71 60 64 Systemic adverse events, dose 1 none 39 (56%) 43 (61%) 0.61 34 (57%) 40 (63%) 0.58 chills mild 1 (1%) 0 1* 1 (2%) 0 1* moderate 2 (3%) 1 (1%) 2 (3%) 0 total 3 (4%) 1 (1%) 0.37 3 (5%) 0 0.11 diarrhea moderate 1 (1%) 0 1* 0 0 1* severe 1 (1%) 0 1 (2%) 0 total 2 (3%) 0 0.24 1 (2%) 0 0.48 fatigue mild 3 (4%) 4 (6%) 0.11* 1 (2%) 4 (6%) 0.16* moderate 12 (17%) 5 (7%) 10 (17%) 5 (8%) severe 3 (4%) 7 (10%) 3 (5%) 5 (8%) total 18 (26%) 16 (23%) 0.70 14 (23%) 14 (22%) 1 fever mild 0 1 (1%) 1 0 0 1 flu-like symptom mild 0 1 (1%) 1 0 1 (2%) 1 headache mild 2 (3%) 5 (7%) 0.49* 2 (3%) 4 (6%) 0.62* moderate 7 (10%) 7 (10%) 6 (10%) 5 (8%) severe 1 (1%) 0 1 (2%) 0 total 10 (14%) 12 (17%) 0.82 9 (15%) 9 (14%) 1 heating/sweating moderate 1 (1%) 0 1* 1 (2%) 0 1* severe 1 (1%) 0 1 (2%) 0 total 2 (3%) 0 0.24 2 (3%) 0 0.23 joint pain moderate 1 (1%) 0 0.50 1 (2%) 0 0.48 malaise mild 1 (1%) 0 1* 1 (2%) 0 1* moderate 0 1 (1%) 0 1 (2%) total 1 (1%) 1 (1%) 1 1 (2%) 1 (2%) 1 muscle pain mild 4 (6%) 2 (3%) 1* 3 (5%) 2 (3%) 1* moderate 6 (9%) 4 (6%) 4 (7%) 2 (3%) severe 2 (3%) 1 (1%) 2 (3%) 0 total 12 (17%) 7 (10%) 0.23 9 (15%) 4 (6%) 0.15 nausea mild 1 (1%) 2 (3%) 1* 1 (2%) 1 (2%) 1* moderate 2 (3%) 1 (1%) 2 (3%) 1 (2%) severe 0 1 (1%) 0 1 (2%) total 3 (4%) 4 (6%) 1 3 (5%) 3 (5%) 1 S4 Table (continued). Local and systemic adverse events in intention-to-treat and naïve only cohorts. Fisher’s exact test was used to report p-values. *test between the levels of the scale of the present symptoms (mild/moderate/severe). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 10 Intention-to-treat Naive 20 µg 30 µg p-value 20 µg 30 µg p-value Participants, n 70 71 60 64 Systemic adverse events, dose 2 none 21 (30%) 27 (38%) 0.38 17 (28%) 23 (36%) 0.44 ageusia moderate 1 (1%) 0 0.50 1 (2%) 0 0.48 anosmia moderate 1 (1%) 0 0.50 1 (2%) 0 0.48 chills mild 3 (4%) 2 (3%) 0.38* 2 (3%) 2 (3%) 0.51* moderate 2 (3%) 7 (10%) 2 (3%) 7 (11%) severe 2 (3%) 2 (3%) 2 (3%) 2 (3%) total 7 (10%) 11 (15%) 0.45 6 (10%) 11 (17%) 0.30 decreased appetite severe 0 1 (1%) 0.50 0 1 (2%) 1 diarrhea mild 0 1 (1%) 1* 0 1 (2%) 1* moderate 0 1 (1%) 0 1 (2%) severe 0 1 (1%) 0 1 (2%) total 0 3 (4%) 0.24 0 3 (5%) 0.24 fatigue mild 6 (9%) 6 (8%) 0.86* 5 (8%) 6 (9%) 0.65* moderate 15 (21%) 17 (24%) 13 (22%) 16 (25%) severe 16 (23%) 13 (18%) 16 (27%) 12 (19%) total 37 (53%) 36 (51%) 0.87 34 (57%) 34 (53%) 0.72 fever mild 4 (6%) 4 (6%) 1* 3 (5%) 4 (6%) 1* moderate 2 (3%) 2 (3%) 2 (3%) 2 (3%) total 6 (9%) 6 (8%) 1 5 (8%) 6 (9%) 1 flu-like symptom mild 1 (1%) 0 1* 1 (2%) 0 1* moderate 1 (1%) 0 0 0 total 2 (3%) 0 0.24 1 (2%) 0 0.48 headache mild 6 (9%) 4 (6%) 0.67* 6 (10%) 4 (6%) 0.61* moderate 17 (24%) 17 (24%) 14 (23%) 16 (25%) severe 6 (9%) 3 (4%) 6 (10%) 3 (5%) total 29 (41%) 24 (34%) 0.39 26 (43%) 23 (36%) 0.46 heating/sweating moderate 0 2 (3%) 0.33* 0 2 (3%) 0.33* severe 2 (3%) 0 2 (3%) 0 total 2 (3%) 2 (3%) 1 2 (3%) 2 (3%) 1 insomnia severe 0 1 (1%) 0.50 0 1 (2%) 1 joint pain mild 0 1 (1%) 0.20* 0 1 (2%) 0.20* moderate 3 (4%) 0 3 (5%) 0 severe 1 (1%) 1 (1%) 1 (2%) 1 (2%) total 4 (6%) 2 (3%) 0.44 4 (7%) 2 (3%) 0.43 S4 Table (continued). Local and systemic adverse events in intention-to-treat and naïve only cohorts. Fisher’s exact test was used to report p-values. *test between the levels of the scale of the present symptoms (mild/moderate/severe). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 11 Intention-to-treat Naive 20 µg 30 µg p-value 20 µg 30 µg p-value Participants, n 70 71 60 64 General adverse events, dose 2 (continued) malaise mild 1 (1%) 0 0.40* 0 0 1* moderate 1 (1%) 3 (4%) 1 (2%) 3 (5%) severe 1 (1%) 0 0 0 total 3 (4%) 3 (4%) 1 1 (2%) 3 (5%) 0.62 muscle pain mild 5 (7%) 5 (7%) 1* 3 (5%) 5 (8%) 0.81* moderate 7 (10%) 8 (11%) 7 (12%) 7 (11%) severe 5 (7%) 4 (6%) 5 (8%) 4 (6%) total 17 (24%) 17 (24%) 1 15 (25%) 16 (25%) 1 nausea mild 0 4 (6%) 0.01* 0 4 (6%) 0.05* moderate 4 (6%) 0 3 (5%) 0 severe 2 (3%) 1 (1%) 2 (3%) 1 (2%) total 6 (9%) 5 (7%) 0.76 5 (8%) 5 (8%) 1 palpitations moderate 1 (1%) 0 0.50 1 (2%) 0 0.48 redness severe 0 1 (1%) 0.50 0 1 (2%) 1 swelling severe 1 (1%) 0 0.50 1 (2%) 0 0.48 S4 Table (continued). Local and systemic adverse events in intention-to-treat and naïve only cohorts. Fisher’s exact test was used to report p-values. *test between the levels of the scale of the present symptoms (mild/moderate/severe). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 12 S3 Fig. Adverse events. Reported local (A) and systemic (B) adverse events after the first (in blue) and second (in orange) vaccine dose, according to severity (mild/moderate/severe) and by study arm (20µg and 30µg) in the intention-to-treat cohort. Occurrence number (x-axis) and percentage (numbers inside bars) per AE calculated on the cohort are given. A B All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 13 S1 Appendix: Study protocol All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Safety and Immunogenicity of a Reduced Dose of the BioNTech/Pfizer BNT162b2 Vaccine in a Healthy Population (REDU-VAC) A randomized multicenter interventional clinical COVID-19 vaccination trial 1. General information This phase IV dose -optimization study will be coordinated by the service Epidemiology of Infectious Diseases and Cancer center (contact: Mieke Goossens , Pieter Pannus ) of the scientific directorate of Epidemiology and public health, Sciensano . Laboratory analys es will be performed by the service Immune response (contact: Isabelle Desombere) and Viral diseases (contact: Cyril Barbezange, Isabelle Thomas) of the scientific directorate of Infectious diseases in humans of Sciensano. The study will be executed in collaboration with Mensura EDPB (contact: Marie-Noëlle Schmickler and Mathieu Verbrugghe), the Virology Unit of the Institute of Tropical Medicine Antwerp (contact: Kevin Ariën), the Institute for Medical Immunology of ULB, Campus Erasme, (contact: Arnaud Marchant) and Campus Gosselies (contact: Stanislas Goriely). 2. Background COVID-19 vaccines are being rolled out in many countries all over the world. Promis ing efficacy data from phase three vaccination trials are being confirmed with real -world data from countries like Israel and the United Kingdom where large proportions of the population have already been vaccinated (1–3). Vaccine supply has proven an importa nt limiting factor for the speed of vaccination campaigns. Indeed, vaccine demand largely surpasses production capacity of the different manufacturers. In order to cope with this scarcity, different strategies have been proposed and implemented, including a delayed second dose and a single instead of dual dose for previously infected people. Another possible strategy is a reduction of the vaccine dose for those population groups who generally have better immunologic vaccine responses , which is the subject o f this clinical trial (4). Indeed, in this study we will investigate immune responses to a reduced dose of the BNT162b2 mRNA vaccine of BioNTech/Pfizer. Data from a dose -escalating phase 1 trial in healthy adults 18 to 85 years of age comparing two doses of 10µg, 20µg, 30µg a nd 100µg indicated that a dose of 30µg of BNT162b2 achieved the best immune response in participants of all ages (5). As a result, a phase2/3 trial was conducted and the vaccine was finally marketed at a dos age of 30µg (6). These immune responses were age dependent , however . While people aged 65-85 years (N=24) had markedly better responses with 30µg as compared to 20µg, this was not the case for people aged 18 -55 years (N=24). Indeed, SARS -CoV-2 specific binding and neutralizing antibody titers were even slightly higher in the 20µg group as compared to the 30µg g roup. We therefore propose to conduct a clinical trial investigating immune responses comparing a 20µg versus 30µg dose of BNT162b2 in a larger cohort of 150 subjects.. 3. Objectives and outcomes We aim to include 150 adults aged 18-55 years from five Mensura EDPB sites. These will be equally randomized in two study arms receiving two doses of: • Arm 1: 20µg BNT162b2 • Arm 2: 30µg BNT162b2 We aim to include up to 25 previously SARS -CoV-2 infected subjects per arm. The objectives of this study include assessing the immunogenicity (both humoral and cellular), safety and reactogenicity of a reduced vaccine dose. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 2 3.1. PRIMARY OBJECTIVE AND OUTCOME - The primary objective is to prove non -inferiority of immunogenicity of a reduced BNT162b2 vaccine dose (20µg) versus the reference vaccine dose (30µg). - The primary outcome is the geometric mean titer (GM T) of binding (IgG) antibodies specific to the receptor binding domain (RBD) of SARS -CoV-2 at four weeks after the second dose. Data from the literature and from our own PICOV -VAC data indicate strong correlation between titers of RBD binding antibodies and neutralizing antibodies. As methods to measure RBD binding antibodies can be implemented more rapidly than neutralizing antibody assays, we consider that RBD binding antibodies are a good surrogate for neutralizing antibodies and can therefore be used as a primary endpoint for this study. 3.2. SECONDARY OBJECTIVES AND OUTCOMES - GMT of RBD-specific binding antibodies at the time of the second dose, six months and one year after the first dose. - GMT of neutralizing antibody titers against wild type and variant SARS -CoV-2 viruses at four weeks after second dose administration . - Cellular immunity parameters ( i.e. Memory B -cell responses , T -cell responses, etc) at all clinically and biologically relevant time points. - Safety and reactogenicity of the different vaccine regimens as defined by the severity, duration and amount of adverse events experienced after each vaccine dose. 4. Methods This is a randomized interventional clinical trial in healthy subjects organized at five sites of Mensura EDPB. Inclusion criteria • Employed by Mensura EDBP (should be an employee at least until the end of the study) • Aged 18-55 years Exclusion criteria • Previously vaccinated against COVID -19 • Pregnant/breastfeeding women Discontinuation criteria All participants retain the right to end their participation in the study at any point in time. Participants who fail to provide the necessary information through the questionnaires, will be discontinued from the study, since any correlation analysis with biological measurements becomes impossible. The primary objective of the study is to prove non -inferiority of anti-RBD binding antibody titers four weeks after the second dose of the 20µg versus 30µg study arm. If the primary outcome of the study reveals an inferiority of the 20µg dose, participants in the 20µg dose arm will be offered a third, 30µg dose of the study vaccine. Although no correlate of protective immunity has been validated yet, a relationship between vaccine - induced SARS-CoV-2 antibodies and protection has been proposed (7). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 3 Sample size The primary analysis is a non -inferiority comparison at 28 days after the second dose, for the GMT of antibodies binding to the RBD of SARS -COV-2, comparing the reference dose ( 30µg BNT162b2) with the reduced dose (20µg BNT162b2). Currently available data from published and ongoing COVID -19 vaccination trials indicate that: • RBD-specific antibody titers (GMT , arbitrary units) 28 days after the second dose of BNT162b2 equals 2125. The standard deviation on the log scale (base 10) is 0.27 (PICOV-VAC trial). The following assumptions are made in the sample size calculations: • The non-inferiority margin is -0.15 absolute difference of GMT on log scale (base 10), between a reduced dose and regular dose vaccine schedule. • The standard deviation of GMT on the log10 scale is 0.27 for BNT162b2, based on the currently available data. • The true difference of GMT on log10 scale is 0. • A two-sided 2.5% family-wise error rate (at cohort level). Based on these assumptions, the comparison between both doses will need a minimum of 50 participants (infection naïve) per group to achieve 90% power. In order to be able to better evaluate reactogenicity however, we aim for a larger sample size. Trial randomization Subject randomization will be done using statistical software (R Studio). The system will randomly allocate the 150 participants across both study arms (20µg and 30µg) irrespective of their previous SARS-CoV-2 infection status . The recruitment and randomization process will aim to have a balanced representation of age and gender across the study arms. Therefore stratified randomization will be used based on age and gender. In addition, 20 participants will be randomly selected per study arm (40 in total) from which extra (heparinized) blood will be collected for in -depth cellular immunogenicity analyses (see table 1). This random selection will be done using the same statistical software (RStudio). Blinding/unblinding Everyone involved in the study will be blinded to the study arm allocat ion of each participant, except for the study nurses administering the vaccine. When the primary outcome of the study has been reached (28 days after the second dose), an interim analysis will be conducted to determine whether the reduced vaccine dose induces an inferior immune response as compared to the regular dose or not . In case of non -inferiority, there will be no need for unblinding before the end of the trial . In case of inferiority, study participants will be unblinded, informed about which study arm they were allocated to and offered a third dose (of 30µg) of BNT162b2. Study flow Participant information and consent procedure A research nurse checks the study eligibility criteria and informs the candidates on the possibility to participate in the study . Participant information and consent includes explaining that the national number will be recorded on site by the investigator for possible later data linkage. The participant information and consent will also include that a trusted third party (TTP) will receive and use the national number to link with administrative data. This data linkage is planned to obtain a more complete data set that will be used for the analysis of possible hospitalization and other medical activities billed to the health insurance. Participant information and consent includes explaining that the email and mobile number from study participants will be recorded in a separate and protected form in the database hosted by the All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 4 occupational physician. These data are only used to send emails and text messages to the participant for this study. After having had the possibility to think and discuss about their participation, the participant gives written informed consent and provides the research nurse with their email address and telephone number that they intend to use the next 12 months. Baseline activities After the participant has given written informed consent, the research nurse completes the Subject ID log with participant’s name, national number and study number. The participant national number is collected and kept on site by the research nurse in the Subject ID log. Each of the participating sites will transfer this local Subject ID log at study closure to the TTP for data linkage, following the procedure detailed in the authorization of the Information Security Committee. The research nurse cr eates a new participant in the study eCRF (hosted on LimeSurvey) and enters the date of consent, gender, year of birth, height, weight. They will also register whether the participant has had a past SARS -CoV-2 infection (i.e. positive nasopharyngeal swab, positive serology test ). Once the information is entered by the research nurse in the eCRF, the participant automatically receives an email (subject “Clinical Trial Message” and starting with “Dear Madam, Dear Sir”) with a link to the data entry screens for the patient-reported-outcomes (PROMs). In order to have a second identity check, the participant first enters year of birth and their study number. The research nurse can view the PROMs of his/her subjects and only exceptionally make corrections upon explicit request by the participant. Each such change will be logged and needs a jus tification. At all times the research nurse can indicate in the system the early end of study (and stop sending emails to the participant to complete the PROMs), e.g. in case of withdrawal of consent, if known to the research nurse. In case of withdrawal of consent, data already collected will be kept in the study data. Vaccination Depending on the randomization, participants will either receive two doses of 20µg or two doses of 30µg of the BioNTech/Pfizer SARS -CoV-2 mRNA vaccine, with a three week interval between both doses. Sampling visits An overview of the different sample types and volumes at each study visit is summarized in table 1 . Venous blood will be collected on the day of the first and second dose as well as four weeks after the second dose, and six months and one year after the first dose. Five mL of blood will be collected from all study participants at all study visits. From a random selection of 20 participants per study arm (40 in total), 36mL of heparinized blood will be collected on the day of the first dose as well as four weeks after the second dose and six months after the first dose . The sampling activity (date, sample types collected) is documented in the eCRF. Samples are handled as detailed in the sampling manual. One week after each vaccination dose, participants will be invited by email and text message by the research nurse to fill in a questionnaire collecting information on the reactogenicity of the vaccine. Monitoring of adverse reactions Suspected unexpected serious adverse reactions, serious adverse reactions, and adverse reactions with grade equal or more than 3 will be monitored for the duration of the study period. The study nurse will complete the reporting form (annex 4) and forward this to the study coordinator, who in turn will report this immediately to the Belgian Federal Agency for Medicines and Health Products. End of the trial The trial will end when the last study participant has been sampled one year after first dose administration. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 5 4.1. DATA MANAGMENT Safety Based on previously published phase 2 and 3 data of both vaccines that will be used in this study, very few to no serious adverse events are expected (6,8). Besides vaccination, n o safety issues linked with the other study procedures are expected either. As a consequence this is considered a low risk study. Data collection Laboratory data (samples) and epidemiologi cal data (questionnaire) will be collected as described above. Each participant will receive a unique identifier code at the start of the study that will be used both on the samples and the questionnaires through the whole study period. Laboratory data Blood samples will be collected as indicated in table 1. Serum will be isolated from serum gel blood collection tubes and will be used for: - SARS-CoV-2 specific binding antibody quantification: using an in-house anti-RBD IgG ELISA. - Neutralizing antibody capacity measurement against the wild type Wuhan strain and at least one variant of concern: using a virus neutralization assay . Peripheral Blood Mononuclear Cells (PBMC) will be isolated from heparinized blood on the day of blood draw and will be stored in liquid nitrogen for further analyses . These include: - Memory B-cell responses: using B -cell ELISpot of flow cytometry - T-cell responses: using intra -cellular cytokine staining Samples will be collected at the research sites by a research nurse of Mensura EDPB. Epidemiological data Socio-demographic characteristics and SARS -CoV-2 initial status (date of last PCR test) will be collected by the research nurse at the time of inclusion and will be pseudonymized (without the identity of the participant but with a unique participant code, s ee further) transferred to Sciensano after informed consent of the participant was obtained. Additional questionnaires will be filled in by the participants given time points (see Table 1), and will provide information on reactogenicity and severity of the adverse events. All questionnaires will be completed through a secured onl ine application (see further). Data flow and management Blood collection tubes as well as labelling stickers will be provided by Sciensano. For the entirety of the study, material w ill be provided in batch and a transport will be organized between Sciensano and the research sites. After collection, samples will be handled following the appropriate SOP. Epidemiological and laboratory data will be linked via a unique code assigned to e ach participant. This code will start with the first letters of the research site location followed by a three digit number (e.g. “BRU_001” for the first participant, “BRU_002” for the second one, and so forth). A sticker labelled with this code will be af fixed by the research nurse on each tube at the time of sampling, and the same corresponding code will be entered in each questionnaire, enabling the link for data analysis. For the planned days, the labels will also include an identifier of this point (e. g. “D00”, “D28” etc). This unique code must stay the same during all the duration of follow -up and a list of all participants and their assigned codes will be kept in a secure and protected way by the occupational physician. Questionnaires are filled in online through LimeSurvey. Before sampling takes place, an email will be sent to the participant containing a questionnaire link and the unique participant code. It will take around All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 6 five minutes for the participant to fill in t he questionnaire. If the questionnaire is not filled in yet at the moment of sampling, it will be filled in together with the research nurse. A confirmation email will be sent to the research nurse when a questionnaire is correctly received. LimeSurvey is a web application running on a server located within Sciensano’s datacenter. The application and its associated database are located on the same server. It means there are no data located outside Sciensano. Access to the administration of the LimeSurvey ap plication is restricted to a limited number of people involved in the administrative management of the survey, who are authenticated with a username and password. All files will be kept on the Sciensano SQL server with restricted access. Login to the shared server is password controlled. Each user/investigator or research nurse will receive a personal login name and password and will have a specific role which has predefined restrictions on what is allowed on the server. Furthermore, users will only be able to see data of subjects of their own site. Any activity in the software is traced and transparent via log files. Direct access to all study records, including source documents and the eCRF will be granted to authorized representatives from the Sponsor, h ost institution and the regulatory authorities to permit study-related monitoring, audits and inspections. The Data Manager will review the eCRFs to make certain that all items have been completed. Incorrect or inappropriate entries in the eCRFs will be returned to the research nurse for correction (data queries), if not already captured using automated data entry checks. Subject privacy must be respected at all times, in accordance to GDPR, GCP and all other applicable local regulations. The investigator/s tudy team should immediately notify the sponsor if he or she has been contacted by a regulatory agency concerning an upcoming inspection. The data capture software analysis, development and testing will be performed according to the procedures of the Spons or. Quality check of data management and data entry will be performed in accordance with the standard operating procedures of the Sponsor. Analysis Data analysis Analysis of all data collected within this study will be performed by Sciensano. Questionnaire responses will always be coded. All analyses will be performed in R (R Core Team (2018), available from https://www.R-project.org/), STATA or SAS. Meaningful laboratory results of each participant will be communi cated to the research nurse who will be able to communicate them to the participants. This includes the anti -SARS-CoV-2 antibody status at the primary endpoint and at the end of the study. As already mentioned, these tests are for research and not diagnost ic purposes, and depending on lab capacity, there will be a delay in the communication of the results. The global results of the study will be communicated to each participating center at the end of the study in the form of a report and/or presentations. Statistical analyses The analysis described here will be coordinated by the Sponsor. Any later analyses on the linked data will also be conducted by the Sponsor. All of the participants who m eet the eligibility criteria will be included in the main analysis . Descriptive analyses will be used to report participant characteristics. The level of anti body titers in each of the study arms will be analyzed with the Kruskal -Wallis test, while participant demographics (e.g. age, gender) will be taken into acco unt (co nfounders). Furthermore, risk factors and correlations with antibody detectability will be evaluated. P-values below 0.05 will be considered statistically significant. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 7 5. Ethics and Privacy Protection 5.1. INDEPENDENT ETHICS C OMMITTEE AND INFORME D CONSENT Prior to study start, this protocol will be reviewed and approved by a n Independent Ethics Committee (IEC). Written, dated and signed informed consent has to be obtained from each eligible subject prior to inclusion in the study. A sample Participant Information/Informed Consent Form has been prepared (see appendix). 5.2. DATA LINKAGE An electronic case report form (eCRF) will be used for data collection. Subject confidentiality will be maintained at all times, within the legal constraints. The exported participant reported outcome measures (PROMs) dataset will contain pseudonymized participant data. A separate instrument (“form”) with separate restricted access will contain direct identifiable subject data (subject’s email address and mobile phone number of the participant) such that emails and additional reminders by text message can be sent to the subject . Separately from the study database, the local i nvestigator will store on their Mensura EDPB facility a Subject ID log (e.g. in a spreadsheet with restricted access, GDPR compliant) containing the national number of the participant and the unique participant number for a possible later data linkage by a trusted third party (TTP), f.e. eHealth. The linkage will follow a procedure approved by the competent chambe r of the Information Security Committee. After the Ethics Committee has approved this protocol, the Sponsor will introduce a request to the competent chamber of the Information Security Committee for the linkage of the identifiable study data including participant reported outcome measures (PROMs), sickness fund (IMA/AIM) financial data (RIZIV/INAMI expenses). Only after the approval of this request, the parties will start data linkage. The principle investigator will guarantee the data protection. 6. Organization of the research project 6.1. TIMELINE The anticipated start date of the study is May 3rd 2021. The end of the trial is anticipated to be the beginning of June 2022. 6.2. STUDY PARTIES AND RE SPONSIBILITIES Study sponsor: Sciensano Study contact and principle investigator is Dr. M aria Goossens, Sciensano. The sponsor has in -house expertise in the lab tests. The spons or has in -house expertise with LimeS urvey for both the eCRFs and the collection of PROMs. Data will be capture d on a secured server. The biostatistician of the Sponsor will perform the study data analysis. The independent Ethic committee: ULB Erasme The current epidemic context justifies the rapid implementation of this study, with the aim of improving the care of the population and con tributing to a better control of the pandemic. Trusted Third Party (TTP): e -Health platform The national number of participants is stored on site in the subject ID log. The PROMs will be linked by the TTP using the national number with the invoices of medi cal acts in the IMA database, after approval by the competent chamber of the Information Security Committee. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 8 Participating sites and local investigators The local investigator , the occupational physician , will collect written informed consent from all participants. In order to facilitate the logistics of participant inclusion, a form to detail participants included will be provided to the investigators. This Subject ID log (appendix) is to be kept by the investigator on site and contains the national number . A copy of the study Subject ID log with the national number and study subject ID is to be provided to the TTP for data linkage. 6.3. RESOURCES - Purchase, preparation and transport of material for the study will be foreseen by Sciensano . - Mensura EDPB will organize the research sites through the occupational physician. - Lab analyses will be performed at Sciensano , ULB and ITG. - PBMC preparations will be done by Sciensano, ULB and ITG. - Epidemiological and laboratory data will be analyzed by the Sciensano researchers . - A budget of 203 600 € was estimated and approved . 7. Risk and benefits for participants 7.1. RISKS The risks for participants are small, they include side effects of the sampling procedure and vaccination. • A blood draw can s ometimes result in a local hematoma, and rarely in vagal discomfort . Epistaxis can occur in people taking anti -coagulants. • The most common side effects after BNT162b2 vaccination are usually mild or moderate and resolve within a few days. They include pain and swelling at the injection site, tiredness, headache, muscle and joint pain, chills and fever. They affect about 10% of vaccinated subjects (6). • Allergic reactions in response to the vaccine s do occur, which are severe (anaphylaxis) in very few cases. As for all vaccines, the BNT162b2 vaccine will be given under close supervision with appropriate medical treatment available. Participants will b e clearly informed about these risks, that are minimized due to the expertise of the persons in charge of collecting the samples. 7.2. BENEFITS In general, the study will provide insight in the possibility of reducing the vaccine dosage required to achieve acceptable immune responses. This might help alleviate the current vaccine scarcity and speed up vaccination campaigns. In particular, study participants will have the possibility to be vaccinated significantly earlier than foreseen by the Belgian national vaccination campaign, since they all belong to low priority groups. Secondly, those participants receiving a lower dose of the vaccine are expected to experience fewer and less severe side effects than they would with a full dose vaccination (5,9). Fina lly, if immune responses in the lower dose arm turn out to be inferior to the regular dose arm , these participants will be offered a third, regular dose to boost their immu nity. 7.3. CONFIDENTIALITY At the researcher level, sample results and questionnaires will be pseudonymized via an individual code attributed to each participant. None of the researchers who analyze the data will be involved in participant data collection, nor in the care of COVID -19 participants. Test results will be communicated by the laboratory to the occupational physician of Mensura EDPB or the research nurse by the individual All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 9 code attributed to each participant. At all times, the occupational physician and the research nurse are the only ones with access to the identity of the participants. 7.4. BIOLOGICAL SPECIMEN S All the samples, except the PBMCs, collected during the study will be stored in the biobank of Sciensano. Residues remaining after the analyses will be kept for a duration of maximum 10 years if the participant has given consent that the samples can be kept for further research. After 10 years, they will be destroyed. Specimens will be stored in the biobank “Biobank - Sciensano WD12 infectieziekten mens COVID-19” approved by the ‘Commissie vo or Medische Ethiek UZ Gent’ on 17 A pril 2020 (internal

EC 037 -2020/mf, appendix 3 ) with registration number assigned by the FAGG/AFMPS: BB200027. The PBMC samples will be stored in the Biobank from the laboratory where the preparation took place (Sciensano, ITG or ULB): • ITM Biobank approved by the ‘Commissie voor Medische Ethiek Universitair Ziekenhuis Antwerpen en de Universiteit van Antwerpen” on 8 April 2019 (internal reference: 19/13 /168) with registration number assigned by the FAGG/AFMPS: BB190041 • “Biobanque de l’Institut d’Immunologie Médicale (IMI)” approved by the ‘ le Comité d’Ethique hospital-facultaire Erasme -ULB’ on 4 May 2020 with registration number assigned by the FAGG/AFMPS: B2020/002 • Biobank - Sciensano WD12 infectieziekten mens COVID -19” approved by the ‘Commissie voor Medische Ethiek UZ Gent’ on 17 April 2020 (internal reference: EC 037 -2020/mf, appendix 3) with registration number assigned by the FAGG/AFMPS: BB200027. 7.5. INFORMED CONSENT Information on the study will be provided by Mensura EDPB and informed consent will be obtained from all participants. The informed consent can be found in appendix 1. 7.6. ETHICS COMMITTEE The study will be conducted in compliance with the principles of the Declaration of Helsinki (2008) and all of the applicable regulatory requirements. 7.7. PROTOCOL AMENDEMENTS Any substantial change, clarification, or addition to this protocol requires a writ ten protocol amendment, and this must be approved by the Sponsor before the change or addition can be considered effective. In addition, the IEC should be notified and formal approval by the IEC should be obtained as applicable, before the substantial amen dment can be implemented. The substantial amendment will reference the protocol by title and version date and must be signed by the principle investigator prior to initiating the change. Once approved, an amendment becomes an integral part of the protocol. 7.8. INSURANCE During their participation in the clinical investigation the participants will be insured as defined by legal requirements. An insurance with no fault responsibility has been foreseen by the sponsor in accordance with the Belgian law concerning experiments on humans, 7 May 2004. 8. Appendices 8.1. APPENDIX 1: INFORMATION BROCHURE AND INFORMED CONSENT 8.2. APPENDIX 2: QUESTIONNAIRE 8.3. APPENDIX 3: APPROVAL BIOBANK 8.4. APPENDIX 4: SUSAR FO RM All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 10 9. References 1. Dagan N, Barda N, Kepten E, Miron O, Perchik S, Katz MA, et al. BNT162b2 mRNA Covid -19 Vaccine in a Nationwide Mass Vaccination Setting. N Engl J Med. 2021 Feb 24; 2. Hall VJ, Foulkes S, Saei A, Andrews N, Oguti B, Charlett A, et al. Effectiveness of BNT162b2 mRNA Vaccine Against Infection and COVID -19 Vaccine Coverage in Healthcare Workers in England, Multicentre Prospective Cohort Study (the SIREN Study) [Internet]. Rocheste r, NY: Social Science Research Network; 2021 Feb [cited 2021 Mar 24]. Report No.: ID 3790399. Available from: https://papers.ssrn.com/abstract=3790399 3. Vasileiou E, Simpson CR, Robertson C, Shi T, Kerr S, Agrawal U, et al. Effectiveness of First Dose of COVID-19 Vaccines Against Hospital Admissions in Scotland: National Prospective Cohort Study of 5.4 Million People [Internet]. Rochester, NY: Social Science Research Network; 2021 Feb [cited 2021 Mar 24]. Report No.: ID 3789264. Available from: https://pa pers.ssrn.com/abstract=3789264 4. Flamaing J. Vaccination of older persons: a broader perspective. Aging Clin Exp Res. 2009 Oct;21(4–5):372–3; author reply 373-374. 5. Walsh EE, Frenck RW, Falsey AR, Kitchin N, Absalon J, Gurtman A, et al. Safety and Immunogenicity of Two RNA -Based Covid -19 Vaccine Candidates. N Engl J Med. 2020 Dec 17;383(25):2439–50. 6. Polack FP, Thomas SJ, Kitchin N, Absalon J, Gurtman A, Lockhart S, et al. Safety and Efficacy of the BNT162b2 mRNA Covid -19 Vaccine. N Engl J Med. 20 20 Dec 31;383(27):2603 –15. 7. Earle KA, Ambrosino DM, Fiore -Gartland A, Goldblatt D, Gilbert PB, Siber GR, et al. Evidence for antibody as a protective correlate for COVID -19 vaccines. medRxiv. 2021 Mar 20;2021.03.17.20200246. 8. Baden LR, El Sahly HM, Essink B, Kotloff K, Frey S, Novak R, et al. Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine. N Engl J Med. 2021 Feb 4;384(5):403 –16. 9. Jackson LA, Anderson EJ, Rouphael NG, Roberts PC, Makhene M, Coler RN, et al. An mRNA Vaccine against SARS -CoV-2 — Preliminary Report. N Engl J Med. 2020;12. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint Version 2021-04-23 11 10. Tables Table 1: Timetable of data and sample collection. Timepoint (days) Day of dose 1 Day of dose 2 4 weeks post dose 2 6 months post dose 1 1 year post dose 1 Informed consent X Questionnaire X X X 5 ml serum dry tube X X X X X 36 ml Heparin tubes* X X X * Only for a random selection of 40 study participants, 20 per study arm. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 25 S2 Appendix: Supplementary methods SARS-CoV-2 Specific Binding Antibodies Enzyme-linked immunosorbent assay Binding antibodies at baseline and after vaccination were assessed using an enzyme-linked immunosorbent assay (ELISA) for the quantitative detection of IgG-class antibodies to RBD (Receptor Binding Domain, Wuhan strain) (Wantai SARS-CoV-2 IgG ELISA (Quantitative); CE-marked; WS-1396; Beijing Wantai Biological Pharmacy Enterprise Co., Ltd, China). For quantification of antibodies, diluted serum samples (1/10, 1/100, 1/400, 1/1600 and 1/6400) were tested with an internal standard, calibrated against NIBSC 20/136 (First WHO International Standard Anti-SARS- CoV-2 Immunoglobulin), and an external positive control sample included on each plate. Diluted samples were incubated (37°C, 30 min.) with pre-coated micro wells and washed five times. Next, plates were incubated (37°C, 30 min) with horseradish peroxidase (HRP)-conjugated anti-human IgG antibodies and washed five times before adding a TMB and urea peroxide solution for 15 min (37°C, dark). After incubation, a stop solution (0.5 M H2SO4) was added and optical density (OD) was measured at 450 nm using a microplate reader. Net OD values were converted to arbitrary IgG units per ml by interpolation from a point-by-point plot fitted with the standard concentrations and net OD values (correlation coefficient R2≥0.9801), using GraphPad Prism version 9.0.0 for Windows (GraphPad Software, San Diego, California USA) and exported to Microsoft Excel. Antibody measurements were adjusted for any sample dilution, converted to international units per ml (IU/ml) and reported as such. Lower limit of quantification (LLQ) was 5 IU/ml. Clinical performance characteristics of the assay, evaluated in 69 PCR-confirmed COVID-19 patients (comprising mild and severe clinical outcomes, ≥ 15 days post onset of symptoms) and 167 pre-pandemic sera, resulted in a specificity of 100% (95% CI 97,75-100) at a sensitivity of 100% (95% CI 94,73-100) for a cut-off of 6 IU/ml. Multiplex Immunoassay (Luminex) Antibody responses at baseline were tested with an in house multiplex immunoassay (MIA). In this test, IgG antibodies to SARS-CoV-2 antigens RBD, S1, S2 and N (Wuhan strain) were measured simultaneously in one assay run. In short, purified antigens RBD (cat n° PX-COV-P046, ProteoGenix, Schiltigheim, France), S1 (cat n° PHA002, Sanyou Biopharmaceuticals, China), S2 (cat n° 40590- V08H1, Sino Biological, China) and N (cat n° PNA006, Sanyou Biopharmaceuticals, China) were coupled covalently to distinct color-coded activated carboxylated beads (Luminex, Austin, Texas, USA). Diluted serum samples (1/100, 1/400, 1/1600 and 1/6400, 1/25600) were measured with the international standard (NIBSC 20/136; first WHO International Standard Anti-SARS-CoV-2 Immunoglobulin), control sera and blanks included on each plate and MFI was converted to IU/mL by interpolation from a five-parameter logistic standard curve. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 26 SARS-CoV-2 Neutralizing Antibodies Serial dilutions of heat-inactivated serum (1/50-1/25600 in EMEM supplemented with 2mM L- glutamine, 100U/ml - 100μg/ml of Penicillin-Streptomycin and 2% fetal bovine serum) were incubated during 1h (37°C, 7% CO2) with 3xTCID100 of a wild type Wuhan strain (2019-nCoV-Italy- INMI1, reference 008V-03893), the B.1.617.2 Delta variant (83DJ-1) and the BA.1 Omicron variant of SARS-CoV-2, in parallel. Sample-virus mixtures and virus/cell controls were added to Vero cells (18.000 cells/well) in a 96-well plate and incubated for five days (37°C, 7% CO2). The cytopathic effect caused by viral growth was scored microscopically. The Reed-Muench method was used to calculate the neutralizing Ab titer that reduced the number of infected wells by 50% (NT50), which was used as a proxy for the neutralizing Ab concentration in the sample (1–3). SARS-CoV-2 specific cellular responses Enzyme-linked immunosorbent spot Spike-specific cellular responses were tested after vaccination using the Human IFN-γ ELISpot kit (3420-2H) from Mabtech (Stockholm, Sweden) following the recommended protocol. Plates (MAIPSWU10, Mabtech, Stockholm, Sweden) were activated for 15 sec with 50µl of 70% ethanol, and washed with distilled water. Plates were then coated with human IFN-γ antibody (15 µg/ml) overnight at 4°C, washed and blocked with 200µl of Roswell Park Memorial Institute (RPMI) containing 10% fetal bovine serum (FBS) for at least two hours. Next, triplicates of 250 000 PBMC were stimulated in the presence or absence of PepMix SARS-CoV-2 spike glycoprotein peptide pools (SUB1-SUB2, JPT, Berlin, Germany) at 1µg/ml and incubated for 20 hours in a 37°C humidified incubator with 5% CO2. After incubation, the plates were washed and incubated with the human biotinylated IFN-γ detection antibody (1µg/ml) for 2 hours, washed and the streptavidin– Horseradish Peroxidase (streptavidin-HRP) diluted at 1/750 in PBS-0,5% FBS was added for one hour. 3,3',5,5'-Tetramethylbenzidine substrate was added for minimum 10 min at room temperature. Wells were then washed with distilled water and air-dried. Spot were counted with an ELISpot reader (AID Autoimmun Diagnostika GmbH, Straßberg, Germany), mean values of triplicates were considered for S1 and S2 and expressed per million PBMCs after subtracting the mean of the triplicates of the unstimulated condition. The limit of detection is defined as the mean + 2 SD from naïve subjects at baseline (D0), corresponding to 54 and 66 cells per million PBMCs for S1 and S2, respectively. Data points equal to 0 were attributed the value 1 before transformation. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint 27 Flow cytometry Cells were stimulated in 96-well round-bottom plates with 1 × 106 PBMCs in RPMI 1640 medium (Lonza, Basel, Switzerland) supplemented with 10% heat-inactivated FBS (Sigma-Aldrich, Kawasaki, Japan), penicillin/streptomycin, amino acids and PepMix SARS-CoV-2 spike glycoprotein peptide pools (SUB1-SUB2, JPT, Berlin, Germany) in the presence of 1µg/mL purified anti-CD28 antibody (clone CD28.2, BD Biosciences, New Jersey, USA). Both peptide pools were used at 1µg/ml per peptide. Incubation was performed at 37°, 5% CO2 for 6 hours with 10µg/ml brefeldin A (Sigma- Aldrich, Kawasaki, Japan) added after 90 min. After stimulation, Live/Dead fixable red stain (ThermoFisher, Massachusetts, USA) was used to exclude dead cells and the staining of surface antigens was carried out for 20 min with the following fluorochrome-conjugated antibodies: CD3 BV711 (UCHT-1; BD), anti‐CD8 PeCy7 (RPA-T8; BD), CD4 HV450 (RPA-T4; BD). Fixation and permeabilization were performed with Cytofix/Cytoperm (BD) and intracellular staining was carried out for 30 min: IFN-γ FITC (RPA-T8; BD), IL2-APC (MQ1-17H12; BD), TNF-AF700 (Mab11; BD), CD154 APC-Cy7 (TRAP-1 ; BD). Cells stimulated with 1mg/ml Staphylococcus enterotoxin (SEB; Sigma-Aldrich) served as positive controls and unstimulated cells only contained anti-CD28. Samples were acquired on a BD LSRFortessa flow cytometer and analyzed with FlowJo v9. The proportion of cells producing cytokines was determined by subtracting the expression levels/production levels in the unstimulated wells, from the peptide stimulated wells.

1. Mariën J, Ceulemans A, Michiels J, Heyndrickx L, Kerkhof K, Foque N, et al. Evaluating SARS-CoV-2 spike and nucleocapsid proteins as targets for antibody detection in severe and mild COVID-19 cases using a Luminex bead-based assay. J Virol Methods. 2021 Feb;288:114025. 2. Selhorst P, Van Ierssel S, Michiels J, Mariën J, Bartholomeeusen K, Dirinck E, et al. Symptomatic SARS- CoV-2 reinfection of a health care worker in a Belgian nosocomial outbreak despite primary neutralizing antibody response. Clin Infect Dis. 2020 Dec 14;ciaa1850. 3. Ariën KK, Heyndrickx L, Michiels J, Vereecken K, Van Lent K, Coppens S, et al. Three doses of BNT162b2 vaccine confer neutralising antibody capacity against the SARS-CoV-2 Omicron variant. NPJ Vaccines. 2022 Mar 8;7(1):35. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprintthis version posted March 31, 2022. ; https://doi.org/10.1101/2022.03.25.22272599doi: medRxiv preprint