This study was conducted at 38 sites in the United States and South Africa between January 2022 and May 2023, with a total of 1471 participants across all cohorts. In cohort 1, which was conducted in the United States and South Africa, 619 BNT162b2 vaccine-experienced participants were randomized 1:1:1 to receive either one dose of BNT162b2-BA.1, two doses of BNT162b2-BA.1, or one dose of BNT162b2 (Supplementary Figure S1). Overall, 99.7% of participants in cohort 1 received the study vaccination. In cohort 2, which was conducted in the United States, 640 BNT162b2 vaccine-experienced participants received a first study dose of either BNT162b2-BA.1 or BNT162b2 (Supplementary Figure S2). Overall, 70.8% of participants in cohort 2 received the optional second study dose of BNT162b2-BA.1. In cohort 3, which was conducted in the United States and South Africa, all 210 assigned COVID-19 vaccine-naive participants received the first BNT162b2-BA.1 dose, with 94.3% and 81.9% receiving the second BNT162b2-BA.1 and third BNT162b2 doses, respectively (Supplementary Figure S3). The most common reasons for study withdrawal across all cohorts were withdrawal by the participants and lost to follow-up.

Demographic characteristics in BNT162b2-experienced participants (cohorts 1 and 2) were generally similar across vaccine groups (Table 1). In cohort 1, the median age at first study vaccination was 39 years, the median time from the previous BNT162b2 dose to the first study vaccination was 6.1 months, and 88.2% of participants had evidence of previous SARS-CoV-2 infection at baseline. In cohort 2, the median age at first study vaccination was 43 years, the median time from the previous dose of BNT162b2 to the first study vaccination was 3.9 months, and 17.2% of participants had evidence of previous SARS-CoV-2 infection at baseline. In cohort 3 (COVID-19 vaccine-naive), the median age at first study vaccination was 28 years and 86.7% of participants had evidence of previous SARS-CoV-2 infection at baseline.

Demographic characteristics for the immunogenicity populations are presented in Supplementary Tables S4–S6.

In the immunogenicity population of BNT162b2 vaccine-experienced participants in cohort 1 with evidence of previous SARS-CoV-2 infection at baseline, neutralizing GMTs to Omicron BA.1 at 1 month after one or two doses of BNT162b2-BA.1 were higher compared with those 1 month after one dose of BNT162b2 (Table 2; Supplementary Figure S4). GMRs (two-sided 95% CI) compared with one dose of BNT162b2 were 2.87 (2.18, 3.78) and 2.64 (2.06, 3.38) after one and two doses of BNT162b2-BA.1, respectively. Differences in percentages of participants with seroresponse at 1 month after the last vaccination between the BNT162b2-BA.1 groups and the BNT162b2 group were 29.0% (95% CI, 17.2%, 40.3%) and 21.0% (95% CI, 8.3%, 33.2%) for one and two doses of BNT162b2-BA.1, respectively. Administration of a second dose of BNT162b2-BA.1 at 1 month after the first dose did not further induce higher neutralizing antibody responses compared with the neutralizing antibody responses after the first dose.

Although it is not feasible to compare titers across different assays, among those who received one or two doses of BNT162b2-BA.1 in cohort 1, GMFRs were lower for the ancestral strain than for the Omicron BA.1 strain; among those who received one dose of BNT162b2, GMFRs were similar between the Omicron BA.1 strain and ancestral strain (Supplementary Figure S4).

For the primary immunogenicity analysis in cohort 2 in BNT162b2 vaccine-experienced individuals without evidence of previous SARS-CoV-2 infection at baseline, a single dose of BNT162b2-BA.1 met the prespecified criterion for superiority with respect to GMR and prespecified criterion for noninferiority with respect to the difference in percentage of participants achieving seroresponse when compared with BNT162b2 (Fig. 2a). The GMR against Omicron BA.1 at 1 month after a single dose of BNT162b2-BA.1 to those after a single dose of BNT162b2 was 1.75 (two-sided 95% CI, 1.39, 2.22). Because the lower limit of the two-sided 95% CI for GMR was > 1, the superiority of BNT162b2-BA.1 to BNT162b2 for Omicron BA.1 was achieved. However, because the lower limit of the two-sided 95% CI for GMR was not > 1.5, the secondary immunogenicity endpoint of “super” superiority of BNT162b2-BA.1 to BNT162b2 was not achieved.

a GMRs for between-group comparisons of Omicron BA.1 neutralizing titers and b differences in percentages of participants with seroresponse to Omicron BA.1 at 1 month after the first study vaccination in BNT162b2-experienced participants who were without evidence of previous SARS-CoV-2 infection at baseline (cohort 2). Data are for participants who were without evidence of previous SARS-CoV-2 infection up to 1 month after the first study vaccination from the evaluable immunogenicity populations of the primary immunogenicity subset. The primary immunogenicity subset included a random sample of 175 participants in each vaccine group selected from the full expanded set (Supplementary Table S3). GMTs and two-sided 95% CIs were calculated by exponentiating the mean logarithm of the titers and the corresponding CIs (based on the Student t distribution). Assay results < LLOQ were set to 0.5 × LLOQ. GMRs and two-sided 95% CIs were calculated by exponentiating the mean differences in the logarithms of the titers and the corresponding CIs (based on the Student t distribution). Exact two-sided 95% CIs for percentages of participants with a seroresponse were calculated by the Clopper–Pearson method and the 95% CIs for difference in the percentage of participants with a seroresponse were calculated using the Miettinen and Nurminen method. The green dashed lines indicate (a) superiority and (b) noninferiority margins for the immunogenicity subset; the red dashed line indicates (a) “super” superiority. GMR geometric mean ratio, GMT geometric mean titer, LLOQ lower limit of quantitation

The percentages of participants in cohort 2 who were without evidence of previous SARS-CoV-2 infection at baseline and achieving seroresponse to Omicron BA.1 were 62.3% in the BNT162b2-BA.1 group and 39.3% in the BNT162b2 group 1 month after a single dose, a difference of 23.0% (two-sided 95% CI, 11.1%, 34.3%; Fig. 2b). As the lower limit of the two-sided 95% CI for the difference in the percentage of participants achieving seroresponse was greater than the prespecified –5% margin, noninferiority of the seroresponse of BNT162b2-BA.1 to BNT162b2 for Omicron BA.1 was achieved.

Findings in participants in cohort 2 who were without evidence of previous SARS-CoV-2 infection at baseline from the full expanded set were consistent with results from the primary immunogenicity subset. For Omicron BA.1, the GMR for the BNT162b2-BA.1 group to the BNT162b2 group was 1.96 (two-sided 95% CI, 1.62, 2.37). The difference in percentages of participants who achieved seroresponse to Omicron BA.1 between the BNT162b2-BA.1 and BNT162b2 groups was 21.4% (two-sided 95% CI, 12.0%, 30.4%).

In an exploratory assessment in BNT162b2-experienced participants in cohort 2, neutralizing GMTs to Omicron BA.1 among participants with or without evidence of previous SARS-CoV-2 infection at baseline were higher in participants with evidence of previous SARS-CoV-2 infection compared with those without in both vaccine groups and across all time points (Supplementary Figure S5A; Supplementary Figure S6A). GMTs to Omicron BA.1 at 1 month after a single dose of BNT162b2-BA.1 or BNT162b2 were similar among participants with evidence of previous SARS-CoV-2 infection at baseline, with a modest decline after 3 months (Supplementary Figure S6A). Among participants in cohort 2 who had evidence of previous SARS-CoV-2 infection at baseline, GMTs to Omicron BA.1 at 1 month after the second dose were higher compared with the GMTs before this dose, with a more substantial increase in the BNT162b2 group. Thereafter, a gradual decline in Omicron BA.1 GMTs at 3 and 6 months after the second dose was observed in both vaccine groups and Omicron BA.1 GMTs were generally similar at 6 months after the second dose in both the BNT162b2-BA.1 and BNT162b2 groups.

One month after the optional second dose of BNT162b2-BA.1 in participants in cohort 2 (who had received either BNT162b2-BA.1 or BNT162b2 as the first dose), GMTs to Omicron BA.1 and percentages achieving seroresponse in the BNT162b2 group among participants with evidence of previous SARS-CoV-2 infection at baseline were substantially increased versus those 1 month after the first dose and were greater than those in the BNT162b2-BA.1 group (who received BNT162b2-BA.1 at both doses; Supplementary Figures S6−S7). A gradual decline in GMTs and, consequently, percentages of participants achieving seroresponse were observed at 3 and 6 months after this second dose with BNT162b2-BA.1 in both participants with and participants without evidence of SARS-CoV-2 infection at baseline. However, GMTs remained above baseline for all groups.

Among BNT162b2 vaccine-experienced participants in cohort 2, GMFRs were lower for the ancestral strain than for the Omicron BA.1 strain across most time points and vaccine groups; the exception was for the first dose of BNT162b2, for which GMFRs were similar between the strains (Supplementary Figure S5).

Among BNT162b2-experienced participants in cohort 2 who received BNT162b2-BA.1 or BNT162b2, neutralizing immune responses against the ancestral strain were higher in participants who were baseline SARS-CoV-2 positive compared with those who were baseline SARS-CoV-2 negative in both vaccine groups across all time points (Supplementary Figure S6). The percentages of participants achieving seroresponse against the ancestral strain were numerically lower than against the Omicron BA.1 strain across all time points and regardless of baseline SARS-CoV-2 status and vaccine group (Supplementary Figure S7).

In COVID-19 vaccine-naive participants with or without evidence of previous SARS-CoV-2 infection, the GMR of Omicron BA.1-neutralizing titers 1 month after a second dose of BNT162b2-BA.1 compared with those 1 month after two doses of BNT162b2 in the historical control was 9.95 (two-sided 95% CI, 7.90, 12.53; Fig. 3a). Superiority of two doses of BNT162b2-BA.1 to two doses of BNT162b2 was met as the lower limit of the two-sided 95% CI for the GMR was > 1. “Super” superiority for GMR was also met as the lower limit of the two-sided 95% CI for the GMR was > 1.5.

a GMR for the between-group comparison of superiority and “super” superiority and b difference in percentages of participants with seroresponse for assessment of noninferiority for Omicron BA.1 at 1 month after the second study vaccination in COVID-19 vaccine-naive participants who were with or without evidence of previous SARS-CoV-2 infection at baseline (cohort 3). Data are for the evaluable immunogenicity population of the cohort 3 expanded set. The expanded set included all participants with or without evidence of baseline SARS-CoV-2 infection except those in the sentinel group (see Supplementary Table S3). Participants in the historical BNT162b2 comparator group were randomly selected (with matching baseline SARS-CoV-2 status) from the pivotal phase 2/3 efficacy trial [4]. GMTs and two-sided 95% CIs were calculated by exponentiating the mean logarithm of the titers and the corresponding CIs (based on the Student t distribution). Assay results < LLOQ were set to 0.5 × LLOQ. The GMR and two-sided 95% CI were calculated by exponentiating the mean difference in the logarithms of the titers and the corresponding CIs (based on the Student t distribution). Exact two-sided 95% CIs for percentages of participants with a seroresponse were calculated by the Clopper–Pearson method; the 95% CI for difference in the percentage of participants with a seroresponse was calculated using the Miettinen and Nurminen method. The green dashed lines indicate (a) superiority and (b) noninferiority margins for the immunogenicity subset; the red dashed line indicates (a) “super” superiority for the immunogenicity subset. GMR geometric mean ratio, GMT geometric mean titer, LLOQ lower limit of quantitation

Additionally, a similar percentage of COVID-19 vaccine-naive participants with or without evidence of previous SARS-CoV-2 infection who received two doses of BNT162b2-BA.1 or BNT162b2 achieved a seroresponse to Omicron BA.1 at 1 month after the second dose (Fig. 3b). The difference in percentage of participants achieving a seroresponse (BNT162b2-BA.1 group minus BNT162b2 group) was 5.4% (two-sided 95% CI, − 3.0%, 13.8%). Therefore, noninferiority of BNT162b2-BA.1 to BNT162b2 was achieved as the lower limit of the two-sided 95% CI was greater than − 5%.

Across all COVID-19 vaccine-naive participants, observed GMTs for Omicron BA.1 were substantially higher 1 month after the second dose of BNT162b2-BA.1 compared with those 1 month after the second dose of BNT162b2 (Supplementary Figure S8a). A decline of GMTs was observed in the BNT162b2-BA.1 group with lower values at 6 months compared with 1 month after the second vaccination. In participants who received a third dose of BNT162b2 (after two doses of BNT162b2-BA.1), GMTs 1 month after the third dose increased to similar levels as those observed 1 month after the second dose of BNT162b2-BA.1, followed by a decline at 6 months after the third dose. GMTs were similar 6 months after the second and third doses and substantially greater than baseline values.

Across all COVID-19 vaccine-naive participants, observed GMTs for the ancestral strain were substantially lower 1 month after the second dose of BNT162b2-BA.1 compared with those 1 month after the second dose of BNT162b2 (Supplementary Figure S8b). The observed GMFRs against the ancestral strain were lower in the BNT162b2-BA.1 group compared with the BNT162b2 group at 1 month after the second dose (7.8 vs. 48.7).

Most local reactions were mild or moderate in severity (Fig. 4). Among BNT162b2-experienced participants (cohorts 1 and 2) and COVID-19 vaccine-naive participants (cohort 3), pain at the injection site was the most common prompted local reaction across BNT162b2-BA.1 and BNT162b2 groups. Systemic events were also predominantly mild or moderate in severity (Fig. 5). The most common prompted systemic events across vaccine groups and doses were fatigue and headache among BNT162b2-experienced (cohorts 1 and 2) and COVID-19 vaccine-naive participants (cohort 3). No grade 4 local reactions or systemic events were reported. For all local reactions and systemic events across all cohorts, the median day of onset for most local reactions and systemic events was day 1 or 2 after vaccination, and all resolved within a median duration of 1–2 days after onset.

Local reactions within 7 days of vaccination in a BNT162b2-experienced participants (cohorts 1 and 2) and in b COVID-19 vaccine-naive participants (cohort 3). Data are for the safety populations (defined in Supplementary Table S3) of (a) cohort 1 and 2 and (b) cohort 3. Severity grading is provided in Supplementary Table S2. The numbers above the bars are the percentage of participants with that specific local reaction overall. D1 dose 1, D2 dose 2, BNT-BA.1 BNT162b2-BA.1 vaccine, BNT BNT162b2 vaccine

Systemic events within 7 days of vaccination in a BNT162b2-experienced participants (cohorts 1 and 2) and in b COVID-19 vaccine-naive participants (cohort 3). Data are for the safety populations (defined in Supplementary Table S3) of a cohort 1 and 2 and b cohort 3. Severity grading is provided in Supplementary Table S2. The numbers above the bars are the percentage of participants with that specific systemic event overall. D1 dose 1, D2 dose 2, BNT-BA.1 BNT162b2-BA.1 vaccine, BNT BNT162b2 vaccine

Among BNT162b2-experienced participants (cohorts 1 and 2), the frequency of AEs from the first vaccination through 1 month after the last vaccination ranged from 2.3%–6.1% among BNT162b2-BA.1 recipients and from 2.0%−3.7% among BNT162b2 recipients (Supplementary Tables S7−S8). Corresponding related AEs within 1 month of the last vaccination ranged from 0–2.5% among BNT162b2-BA.1 recipients and 0.5%–1.5% among BNT162b2 recipients. There were no AEs leading to withdrawal. Frequencies of SAEs throughout the study were generally similar between all groups (0.9%−1.6%); all were assessed as not related to study vaccination by the investigator.

Overall, 6.7% of COVID-19 vaccine-naive participants (cohort 3) experienced an AE from the first BNT162b2-BA.1 dose through 1 month after the second dose, 0.5% experienced an AE assessed as related to study vaccination by the investigator, and 0.5% experienced an AE leading to study withdrawal (Supplementary Table S9). Throughout the study, 3.8% of participants reported SAEs, none of which were assessed as related to study vaccination.

No cases of myocarditis or pericarditis (protocol-specified AESIs), or anaphylaxis or Bell’s palsy were reported. Lymphadenopathy occurred in six BNT162b2-experienced participants (cohort 2) who received BNT162b2-BA.1; all were assessed as related to vaccination and resolved within 2–8 days.

In cohort 1, severe AEs reported from the first study vaccination to 6 months after the last study vaccination included dilated cardiomyopathy in a participant in the two-dose BNT162b2-BA.1 group, which was assessed as not related to study intervention by the investigator. Life-threatening cardiac disorders reported in the two-dose BNT162b2-BA.1 group included acute myocardial infarction, congestive cardiac failure, and dilated cardiomyopathy (1 participant reported both congestive cardiac failure and dilated cardiomyopathy, while another participant reported acute myocardial infarction); all were assessed as unrelated to study vaccine.

Cases of confirmed COVID-19 after study vaccination were reported in 3.4%−5.3% and 20.6%−25.2% of BNT162b2-experienced participants in cohorts 1 and 2, respectively (Supplementary Table S10). This discrepancy observed between cohort 1 and cohort 2 is likely attributable to marked differences in baseline SARS-CoV-2 status (88.2% vs. 17.2% had evidence of previous SARS-CoV-2 infection at baseline) and geographic distribution, with the majority of cohort 1 participants recruited from South Africa and cohort 2 participants recruited only from the United States (Table 1). There was one reported case of severe COVID-19 in a participant (cohort 1) who received two doses of BNT162b2-BA.1. The participant did not require hospitalization and there was no diagnosis of significant renal, hepatic, or neurologic dysfunction. The participant was noted to have an oxygen saturation of 93% on room air. Confirmed cases of COVID-19 occurred in 2.9% of participants who had not previously received COVID-19 vaccination after the study vaccination (cohort 3). Among the determinate sequencing results for COVID-19 cases, all identified cases were of the Omicron lineage.