Immune response and safety of DS-5670d, an mRNA COVID-19 vaccine targeting Omicron XBB.1.5, in a phase 3 controlled trial

This was a phase 3 randomized, active-comparator, observer-blind, non-inferiority study registered with the Japan Registry of Clinical Trials with the identifier jRCT2031230424. The study was conducted in accordance with the Declaration of Helsinki, Good Clinical Practice guidelines, and all national and regional ordinance, and was approved by the relevant ethical committees at each study site (). The study is reported as per CONSORT guidelines (CONSORT Checklist). S1 Table S1

The first participant was enrolled on 9 January 2024, the initial observation period lasted for 4 weeks post-vaccination, and the follow-up period was planned to continue up to 26 weeks post-vaccination. In the current manuscript, we report the results up to 4 weeks post-vaccination.

The study included adults and children aged ≥12 years enrolled from 15 sites across Japan (). Written informed consent was obtained from each participant or their legal representative prior to initiation of study procedures. Eligible participants were stratified into 4 subpopulations: (A) those with a history of both SARS-CoV-2 infection and COVID-19 vaccination, (B) those with a history of SARS-CoV-2 infection but without a history of COVID-19 vaccination, (C) those without a history of SARS-CoV-2 infection but with a history of COVID-19 vaccination, or (D) those without any history of SARS-CoV-2 infection or COVID-19 vaccination prior to the date of informed consent. Prior infection was defined as a positive test result for SARS-CoV-2 (either a reverse transcription-polymerase chain reaction test, other nucleic acid detection test, or SARS-CoV-2 antigen test), or a diagnosis of COVID-19. For participants in subpopulations A and B, the positive test result for SARS-CoV-2 was required to have been documented >3 months prior to enrolment, as persons who had tested positive for SARS-CoV-2 infection or were diagnosed with COVID-19 within 3 months before the date of informed consent, or had symptoms suggestive of SARS-CoV-2 infection at the time of informed consent, or who had a positive SARS-COV-2 test at the time of eligibility evaluation were deemed ineligible for the study. Participants in subpopulation D were categorized as non-immune based on an absence of vaccination history, plus an absence of self-reported infection history, plus a negative N-antibody test prior to administration of the study vaccine. S1 Table

Key exclusion criteria were the presence of serious cardiovascular, renal, hepatic, blood, neuropsychiatric or developmental disorder, thrombocytopenia, or coagulopathy; history of vaccine-related seizures or epilepsy, or history of anaphylaxis or severe allergy due to administration of food, drugs, cosmetics or vaccines; current or prior history of myocarditis or pericarditis; or a previous diagnosis of immunodeficiency or a close relative with congenital immunodeficiency.

The study design is shown in Fig 1A. Participants were divided into subpopulations A–D according to their prior infection/vaccination status. Enrolled participants were then randomly assigned in a 1:1 ratio via stratified block randomization to receive either DS-5670d (60 µg of mRNA) or monovalent BNT162b2 (containing an antigen from omicron XBB.1.5, 30 µg of mRNA). There were four stratification factors: (i) study site; (ii) age (≥12 and <18 years, ≥ 18 and <65 years, ≥ 65 years); (iii) history of previous SARS-CoV-2 infection (present or absent); (iv) history of previous SARS-CoV-2 vaccination (present or absent). Random assignment was the responsibility of an independent statistician, and vaccines were dispensed and administered unblinded site personnel. Investigators were blinded to treatment, and uploaded participant data via an interactive web response system. Participants and their legal representatives, the study monitor and members of the study advisory board, sponsor and collaborators, and personnel performing antibody titer determination were also blinded to vaccine assignment.

Participants received either DS-5670d (60 µg of mRNA) or monovalent BNT162b2; vaccines were dispensed and administered intramuscularly into the deltoid region of the upper arm on day 1.

The study objective was to verify the non-inferiority of DS-5670d to BNT162b2 in terms of immunogenicity, and to evaluate safety.

The primary efficacy endpoint was geometric mean titer (GMT) of blood neutralizing activity against SARS-CoV-2 (omicron variant XBB.1.5) and seroresponse rate at day 29 (4 weeks) after study vaccine administration in participants with any history of SARS-CoV-2 infection and/or vaccination (the combined ABC subpopulation). Seroresponse was defined as a ≥ 4-fold increase in the blood neutralizing activity against SARS-CoV-2 after study vaccine administration, compared with the value before administration. The key secondary efficacy endpoint was blood neutralizing activity GMT and seroresponse rate at day 29 after study vaccination in all study participants regardless of infection or vaccination history (the combined ABCD subpopulation). The incidence of COVID-19 (confirmed by the investigator based on a positive SARS-CoV-2 infection test plus any COVID-19 symptoms) up to one month (day 29) and six months after study vaccination was also evaluated. Exploratory efficacy endpoints (GMT of blood neutralizing activity and seroresponse rate at 8, 12, 16, 20, and 26 weeks after study vaccine administration) are not reported here.

Safety endpoints included the occurrence of solicited adverse events (AEs), unsolicited AEs, serious AEs (SAEs), and clinically relevant changes in laboratory values. AEs were coded using the Medical Dictionary for Regulatory Activities version 26.1. Health status was recorded in an electronic diary by participants or their representative. Solicited AEs (injection site events [redness, swelling, induration, pain, warmth, and pruritus] and systemic events [fever, malaise, headache, rash, and myalgia]) were collected up to day 8 post-administration, and unsolicited AEs and SAEs up to day 29. Laboratory tests were performed pre-vaccination and at day 29 post-vaccination.

To confirm non-inferiority of DS-5670d versus BNT162b2 at day 29, the lower limit of the two-sided 95% confidence interval (CI) of the GMT ratio was required to be above the non-inferiority margin of 0.67, and the lower limit of the two-sided 95% CI for the difference in the seroresponse rate was required to exceed −10% in favor of DS-5670d. Based on this, plus taking into account anticipated dropout rates (approximately 10% due to withdrawal or lack of data for analysis), the number of participants in the combined ABC subpopulations required to power the immunogenicity non-inferiority comparison was 690 (of whom 345 would receive DS-5670d and 345 would receive BNT162b2). With a sample size of 690, a study vaccine assignment ratio of 1:1, and a common standard deviation (SD) of neutralizing activity (common logarithmic value) of 0.533 (determined from prior studies in participants aged ≥18 years; Daiichi Sankyo Co., Ltd, data on file), the statistical power to detect the non-inferiority for GMT would be 98.2% at 1-sided significance level of 0.025, and the power to detect non-inferiority for the seroresponse rate would be 87.2% at 1-sided significance level of 0.025. Thus, the power for non-inferiority which simultaneously satisfied both primary endpoints was 85.6% (i.e., 98.2% × 87.2%) assuming that non-inferiority for GMT and seroresponse rate were independent.

Among the 690 planned participants for the combined ABC subpopulations, 100 participants (50 per vaccine group) were required to be vaccine-naïve (i.e., included in subpopulation B). No sample size requirement or limitation was set for subpopulation D as the number of eligible participants for this group (no prior infection, no prior vaccination) was considered likely to be small; however, as many eligible participants as possible were enrolled into this subpopulation. With a total combined ABCD sample size of 790 participants, and assuming that the GMT ratio (DS-5670d group/ BNT162b2 omicron XBB.1.5 group) for blood neutralizing activity against SARS-CoV-2 at 4 weeks after study vaccine administration was 1.0, and the seroresponse rate was 80.0%, the statistical power to detect non-inferiority for GMT would be 99.1% and for the seroresponse rate would be 91.2%, both at a 1-sided significance level of 0.025. The power for non-inferiority which simultaneously satisfied these key secondary endpoints was 90.4% (i.e., 99.1% × 91.2%) assuming that the two non-inferiority outcomes were independent.

A sequential procedure was followed for the immunogenicity analyses. First, non-inferiority was tested for the combined ABC subpopulations; only if non-inferiority was confirmed was testing continued for the combined ABCD subpopulation. Immunogenicity analyses were conducted based on the treatment group to which each participant was assigned; the primary analysis population was the immunogenicity-evaluable per protocol set (PPS), which included participants who received a dose of study drug, had a pre-administration and at least one post-administration immunogenicity measurement, and had no protocol violations that could affect the immunogenicity evaluations. Safety analyses were conducted based on the study drug that was actually administered. Unsolicited AEs were evaluated using the safety analysis set, which included all participants who received at least one dose of study drug; solicited AEs were evaluated in the solicited safety analysis set, which included all participants in the safety analysis set for whom information on the occurrence of at least one solicited AE was available.

Baseline participant characteristics were recorded descriptively as number (%), mean (SD), or median (range). Adjusted GMT and two-sided 95% CI were calculated using an analysis of covariance (ANCOVA) model, with blood neutralizing activity as the dependent variable, study vaccine administration group as an independent variable, and baseline value and subpopulation as covariates. The adjusted GMT ratio was also calculated based on the ANCOVA model. We evaluated the assumptions of the ANCOVA analysis in terms of multicollinearity of covariates, homogeneity of variance, and homogeneity of regression; none of the tests indicated a significant impact on the comparisons between DS-5670d and BNT162b2 omicron XBB.1.5 (). For the adjusted seroresponse rate, the between-group difference was calculated using the Mantel-Haenszel method, with subpopulation included as a stratum. The COVID-19 incidence rate was calculated, using the PPS, as the number of cases per 1,000 person-years; curves were generated using Kaplan-Meier methodology and 95% CIs were derived using the Clopper-Pearson exact method. Solicited AEs were recorded in an electronic diary by each participant or their legal representative; the presence or absence of solicited AEs and their severity were recorded up to 8 days post-vaccine administration, and any other (unsolicited) AEs were recorded up to day 29. The number (%) of participants with events were calculated according to the respective analysis set and tabulated. All statistical calculations were performed using SAS software version 9.4 or later (SAS Institute, Inc., Cary, NC, USA). S1 Text

A total of 779 participants were enrolled into the study, of whom 394 were assigned to receive DS-5670d and 385 to receive BNT162b2 omicron XBB.1.5. The disposition and analysis populations are presented in Fig 1B. One participant in each group did not receive study vaccination, so the safety analysis set included 393 participants in the DS-5670d group and 384 in the BNT162b2 group. Overall, the safety analysis set included 286 participants in subpopulation A, 133 in subpopulation B, 330 in subpopulation C, and 28 in subpopulation D.

Three participants were excluded due to major protocol deviations (two in the DS-5670d group had taken a concomitant medication prohibited by the protocol, and one in the BNT162b2 group had been mistakenly enrolled in the study despite meeting exclusion criteria). After additional exclusions due to a lack of immunogenicity data, the immunogenicity-evaluable PPS (combined subpopulations ABCD) included 378 participants in the DS-5670d group and 375 in the BNT162b2 group. For the primary efficacy endpoint, the 28 participants in subpopulation D were excluded; thus, the immunogenicity-evaluable ABC subpopulation comprised 362 in the DS-5670d group and 363 in the BNT162b2 group.

The demographics and baseline characteristics of participants are described in Table 1. The median age of participants was 46.0 years (range 12–90 years), with the majority aged between 18–65 years (644/777, 82.9%). Slightly more than half of participants were male (425/777, 54.7%), and had a history of SARS-CoV-2 infection (419/777, 53.9%). Many participants had received prior vaccination (616/777, 79.3%), with monovalent (original strain) vaccines in 313 (40.3%) of participants and bivalent (original strain/omicron BA.4–5) vaccines in 275 (35.4%). Characteristics were similar across the study vaccine groups, and there were no notable differences between the safety analysis set and the immunogenicity-evaluable PPS.

The randomization and stratification schedule implemented for the study aimed to overcome potential bias from factors such as age, location, and prior immune history. However, subpopulation B (those with a history of SARS-CoV-2 infection but without a history of COVID-19 vaccination) contained a slightly greater proportion of children, while subpopulation C (those without a history of SARS-CoV-2 infection but with a history of COVID-19 vaccination) contained a slightly greater proportion of elderly patients. This was not unexpected, and likely reflects real-world COVID-19 precautions and vaccine roll-out in Japan, where elderly people took measures to avoid infection and became eligible for vaccination prior to young adults and children who were more likely to have become infected at home or school. Other characteristics across the subpopulations were similar, except where specified in terms of immune status (). S2 Table

In the primary analysis (combined ABC subpopulations), the adjusted GMT ratio of DS-5670d to BNT162b2 was 1.218 (95% CI, 1.059, 1.401). Seroresponse rates were 87.3% in the DS-5670d group and 82.9% in the BNT162b2 group; the adjusted difference was 4.5% (95% CI, –0.70, 9.71). Both of these results exceeded the prespecified non-inferiority margins (Table 2 and S1 Fig), and the study met the primary endpoint. Anonymized individual data underlying the neutralizing titer calculations are provided by sex in S3 Table and by age in S4 Table.

In the secondary analysis (combined ABCD subpopulations), the adjusted GMT ratio was 1.185 (95% CI, 1.025, 1.371), and the adjusted difference in seroresponse rate was 3.7% (95% CI, –1.39, 8.87). Again, both results exceeded the prespecified non-inferiority margins (Table 2).

An analysis of blood neutralizing activity and seroresponse according to key subgroups (age group, sex, and subpopulation) is shown in Fig 2. The GMT ratio favored DS-5670d in patients with a history of vaccination, and also in those with infection history but without a history of vaccination. Violin plots illustrating immunogenicity according to study arm are provided as supplementary information, with the overall population shown in S2 Fig, by age category in S3 Fig, and by subpopulation in S4 Fig.

The incidence rate of COVID-19 to day 29 after study vaccination is presented in Fig 3. Both study vaccines appeared to be effective in preventing investigator-confirmed COVID-19 during this period, with two events reported in the DS-5670d group and three in the BNT162b2 group. The cumulative incidence at 6 months was < 5% in both groups, and no clear between-group difference in the incidence rate over the 6-month period following vaccine administration was observed (S5 Fig).

AEs occurring during the study are summarized in Table 3. The majority of participants experienced at least one AE, with most AEs judged by the investigator to be vaccine-related. There were very few serious AEs across the study vaccine groups; none of the serious AEs were judged to be study vaccine-related.

The majority of solicited AEs were mild or moderate in intensity (Fig 4). The most frequently reported solicited injection site AE across both study vaccine groups was mild pain. Unsolicited treatment-emergent AEs (all causality and study vaccine-related) are described in S5 Table. Overall, there were no major differences in the incidence or severity of solicited or unsolicited AEs between the study vaccine groups.