This single-blind, multicenter, randomized, controlled trial involved healthcare professionals from four academic hospitals in the Netherlands (see protocol, available in full text of this article at NEJM.org).9 The trial followed the principles of the Declaration of Helsinki and was approved by the Medical Ethics Review Committee of the Erasmus Medical Center and the local assessment committees of the participating centers. All participants gave written informed consent prior to registration. Qiagen supplied QuantiFERON SARS-CoV-2 assay kits (starter kits and extended kits for research use only) but had no role in the design, data collection or analysis of the experiment. The authors vouch for the accuracy and completeness of the data and for the fidelity of the experiment to the protocol.
Participants and randomization
Healthcare professionals were eligible to participate if they were between 18 and 65 years of age and did not have serious coexisting factors or conditions (eg receiving cancer treatment, use of immunosuppressive agents, dependence on dialysis or receiving a fixed organ). or bone marrow transplantation) or a history of SARS-CoV-2 infection (either laboratory confirmed or reported by the participant).9 A list of the inclusion and exclusion criteria can be found in the protocol. The representativeness of the experimental population is described in Table S1 in the Supplementary Appendix, available at NEJM.org.
Participants had been vaccinated with Ad26.COV2.S 3 months prior to enrollment and were randomly assigned a 1: 1: 1: 1 ratio to not receive a booster or to receive an Ad26.COV2.S booster, an mRNA-1273 booster or a BNT162b2 booster. The pre-specified prime boost interval was 84 days (interquartile range, -7 to 21). Randomization was stratified by trial site after written informed consent from participants. In addition, half of the participants in each group were randomly selected for analyzes of the S-specific T cell response.
At the first trial visit, participants received a booster by injection into the deltoid muscle. The volume and appearance of the vaccines assigned were hidden from the participants to maintain blindness. Vaccine doses were administered according to the Summary of Product Characteristics for Ad26.COV2.S (≥8.92 × 10)10 viral particles), mRNA-1273 (100 μg) and BNT162b2 (30 μg).
Participants randomly assigned to the non-booster group were informed of their task at the first trial visit and did not receive a placebo injection due to ethical concerns. Blood samples were collected at the first and second test visits (at 0 and 28 days). Booster assignments were blinded 8 days after the boosters were administered, after participants completed a questionnaire on reactogenicity.
Safety assessments included monitoring of reactions reported by participants after Ad26.COV2.S priming dose and after boosters. Perceived severity was assessed using a modified 4-point Food and Drug Administration toxicity scale (where 0 indicates no symptoms, 1 mild symptoms that do not interfere with daily activities, 2 moderate symptoms that interfere with daily activities, and 3 severe symptoms, preventing daily activities).20
In addition, participants reported whether the side effects were present every day from the day of injection until 7 days after injection. Adverse reactions were reported using an electronic questionnaire completed by participants 8 days after receiving a booster. Adverse events that occurred after the previously administered priming dose were reported by uptake (approximately 3 months after the priming injection) and were subject to potential recall bias. Other serious adverse reactions and evoked local or systemic reactions were reported by participants in a questionnaire, by email or by telephone. Safety monitoring (biochemical blood tests and a haematological assessment) was performed on days 0 and 28.
The analysis of humoral and cellular immune responses is described in the Supplementary Methods section of the Supplementary Appendix. Briefly, to confirm that participants had not been exposed to SARS-CoV-2, SARS-CoV-2 nucleocapsid (N) -specific antibodies were measured in all baseline samples and in samples obtained from a sample of participants in nonbooster group that had unexpected responses on day 28. On days 0 and 28 after booster vaccination, S-specific binding antibodies were measured using a quantitative anti-spike IgG assay (Liaison SARS-CoV-2 TrimericS IgG assay, DiaSorin) .21.22 Neutralizing antibodies to infectious SARS-CoV-2 D614G (Global Initiative on Sharing All Influenza Data Sequence, hCov-19 / Netherlands / ZH-EMC-2498) were assessed by a plaque reduction neutralization test (PRNT) in Vero E6 cells. S-specific T cell responses were assessed with an interferon-γ release assay (QuantiFERON, Qiagen) on days 0 and 28 after booster vaccination, as previously described.23
The sample size was determined based on available data.9,15,17 We calculated that 108 participants per. group (432 in total) would give the experiment 80% power at a unilateral 2.5% significance level to detect a log-transformed difference of 0.2 in antibody levels among the groups with 25% SARS-CoV -2 seropositivity at baseline and an expected loss of 25% for follow-up.
The baseline characteristics of each group, including immune responses, are described. Continuous variables at baseline are presented as medians and interquartile ranges. Median differences across the four groups were compared with the use of the Kruskal-Wallis test. Categorical variables are presented as numbers and percentages, and differences between the groups were compared with the use of Fisher's exact test.
The primary endpoint was the log-transformed level of S-specific IgG-binding antibodies 28 days after booster vaccination. We used Mann – Whitney U tests to assess the differences in log-transformed titer values for the following three comparisons: Ad26.COV2.S booster without booster, Ad26.COV2.S booster with BNT162b2 booster and Ad26.COV2.S booster with mRNA- 1273 booster. In a post hoc analysis, we also compared the BNT162b2 booster with an mRNA-1273 booster. Effect sizes (beta coefficients) and 98.3% confidence intervals were estimated using quantile regression, where we varied the reference category to estimate each contrast.
The pre-specified secondary endpoints were levels of neutralizing antibodies, S-specific T cell responses, and reactogenicity. In addition, we analyzed the following variables in a post hoc way. We classified participants as having a response or no response based on a predetermined cutoff value (according to the manufacturer's instructions or an external validation cohort for each analysis), and we compared cross-group responses using Fisher's exact test. To correct for baseline values, we also assessed differences in the median factor change in log10-transformed values for S-specific IgG-binding antibody levels, neutralizing antibody levels and S-specific T cell responses before the booster compared to after the booster. Spearman's correlation coefficient and linear regression were calculated to examine the relationship between binding antibody levels and neutralizing antibody levels and between binding antibody levels and S-specific T cell responses in samples obtained before and after booster vaccination. Linear regressions accompany the beta coefficients and 95% confidence intervals. These analyzes do not check for multiple comparisons, and the conclusions may not be reproducible.
To assess the comparability of the experimental groups with adjustment for baseline titer values, we performed a quantile regression on the log-transformed S-specific IgG-binding antibody levels 28 days after booster vaccination with group, recruitment center and log-transformed baseline titer value as covariates. For the secondary endpoints, we analyzed the database on pairwise deletion without imputation.
Statistical analyzes were performed using GraphPad Prism software, version 9.1.2, and RStudio software, version 4.0.5. We pre-specified that a P value of less than 0.017 was considered to indicate statistical significance (using Bonferroni correction at the 0.05 level for the three comparisons for the pre-specified primary endpoint).