Between November 21, 2020, and March 22, 2021, a total of 5973 participants underwent screening; in the full analysis set, 3460 were randomly assigned to receive AZD7442 and 1737 were randomly assigned to receive placebo. The last participant received an injection on March 29, 2021. In both groups, the median follow-up time from receipt of AZD7442 or placebo to the primary analysis was 83 days, and the median 6-month follow-up was 196 days. The primary analysis was conducted after 30% of the participants had elected to become aware of their randomized assignment (e.g., in order to consider Covid-19 vaccination).
The primary safety analysis included 3461 participants who had received AZD7442 and 1736 participants who had received placebo; 1 participant was assigned to receive placebo but incorrectly received AZD7442. The primary efficacy analysis included 3441 participants who had received AZD7442 and 1731 participants who had received placebo.
All the participants underwent SARS-CoV-2 RT-PCR testing at baseline; 25 of 5197 participants (0.5%; 19 in the AZD7442 group and 6 in the placebo group) tested positive after receiving AZD7442 or placebo and were excluded from the primary efficacy analysis. In the AZD7442 group, 1413 participants had become aware of their randomized assignment; 1406 participants (99.5%) had elected to become aware of their randomized assignment because they wanted to consider receiving a Covid-19 vaccine. In the AZD7442 group, 1161 received a Covid-19 vaccine (Fig. S1). In the placebo group, 749 participants had become aware of their randomized assignment; 742 participants (99.1%) had elected to become aware of their randomized assignment because they wanted to consider receiving a Covid-19 vaccine. In the placebo group, 853 received a Covid-19 vaccine. The percentages of participants with data that were censored because of loss to follow-up or discontinuation, unblinding, or vaccination were balanced between the AZD7442 and placebo groups (Tables S2 and S3).
The demographic and clinical characteristics at baseline were similar in the two groups (Table 1) and were consistent with those of the broader population of persons with SARS-CoV-2 infection (Table S4). The mean age was 53.5 years, 43.4% of the participants were 60 years of age or older, 46.1% were female, 14.5% identified as Hispanic or Latinx, 73.0% were White, and 17.3% were Black. At baseline, a large proportion of the participants were considered by the investigators to have an increased risk of an inadequate response to Covid-19 vaccination (73.3%) or exposure to SARS-CoV-2 (52.5%), and 77.5% had coexisting conditions that placed them at high risk for progression to severe Covid-19 disease.
At the data-cutoff date for the primary analysis, at least one adverse event was reported in 1221 of 3461 participants (35.3%) in the AZD7442 group and 593 of 1736 participants (34.2%) in the placebo group (Table 2). Most adverse events were mild or moderate in intensity. The most common adverse event of special interest was an injection-site reaction, which occurred in 2.4% of the participants in the AZD7442 group and in 2.1% of those in the placebo group. The incidence of serious adverse events was similar in the two groups (Table S5).
Eight deaths occurred; two deaths (both in the placebo group) were assessed by the adjudication committee and were confirmed by testing as being related to Covid-19 (Table 2). The other deaths were the result of illicit-drug overdose (in 4 participants [0.1%], 2 each in the AZD7442 and placebo groups), myocardial infarction (in 1 participant [<0.1%] in the AZD7442 group), and renal failure (in 1 participant [<0.1%] in the AZD7442 group). All the deaths were assessed by the site investigators as being unrelated to AZD7442 or placebo.
The safety analysis at the median 6-month data-cutoff date revealed no additional adverse events of special interest (Table S6) or unexpected longer-term safety signals (Table S7). Nine deaths occurred in the AZD7442 group, and seven deaths occurred in the placebo group; none were considered by the investigator to be related to AZD7442 or placebo. No additional Covid-19–related deaths occurred.
At the data-cutoff date for the primary analysis, symptomatic SARS-CoV-2 RT-PCR–positive illness had occurred in 8 of 3441 participants (0.2%) in the AZD7442 group and in 17 of 1731 participants (1.0%) in the placebo group. The primary efficacy analysis met the statistical criterion for trial success; that is, it showed a significantly lower incidence of symptomatic SARS-CoV-2 RT-PCR–positive illness in the AZD7442 group than in the placebo group (relative risk reduction, 76.7%; 95% confidence interval [CI], 46.0 to 90.0; P<0.001) (Table 3). The unadjusted relative risk reduction was the same as the relative risk reduction after adjustment for age. The between-group differences in the key supportive analyses and the key secondary end point (Table S8) were statistically significant within the testing strategy.
At the data-cutoff date for the primary analysis, SARS-CoV-2 RT-PCR–positive severe or critical illness (defined in Table S9) had occurred in none of the 3441 participants in the AZD7442 group and in 1 of the 1731 participants (0.1%) in the placebo group. In the median 6-month follow-up data analyses, an additional four cases of severe or critical Covid-19 were reported, for a total of five cases, all of which occurred in the placebo group. Six participants (0.2%) in the AZD7442 group and none of the participants in the placebo group had emergency department visits for symptoms that were consistent with Covid-19; the 6 participants were not hospitalized, and 3 of them subsequently tested positive for Covid-19. Results of the post hoc analysis of Covid-19–related hospitalizations are provided in the Supplementary Results section in the Supplementary Appendix.
As compared with the primary analysis of the primary end point, the median 6-month follow-up data analyses showed an even lower incidence of symptomatic Covid-19 (Table 3) in the AZD7442 group than in the placebo group, with a relative risk reduction of 82.8% (95% CI, 65.8 to 91.4). This increase in the efficacy estimate since the time of the primary analysis was driven by a greater percentage of events in the placebo group (1.2%, in 12 of 960 participants) than in the AZD7442 group (0.1%, in 3 of 2003 participants) during months 3 through 6, as compared with months 0 through 3 (Table S10).
Estimates are based on Poisson regression with robust variance with the use of a full model or reduced model. An estimated relative risk reduction greater than 0 favored AZD7442. Panel A shows the relative risk reduction according to the baseline demographic and clinical characteristics of the participants. The relative risk reduction with AZD7442 could not be estimated for participants of American Indian or Alaskan Native heritage or those with immunosuppressive disease or sickle cell disease, because there were no instances of SARS-CoV-2 RT-PCR–positive symptomatic illness in participants in those subgroups. Panel B shows the relative risk reduction according to the participants’ coexisting conditions. The body-mass index (BMI) is the weight in kilograms divided by the square of the height in meters. Immunosuppressive treatment is medication that suppresses the immune response, and immunosuppressive disease is a medical condition that could suppress the immune response, regardless of treatment. CI denotes confidence interval, COPD chronic obstructive pulmonary disease, Covid-19 coronavirus disease 2019, NE not estimable, RT-PCR reverse-transcriptase–polymerase chain reaction, and SARS-CoV-2 severe acute respiratory syndrome coronavirus 2.
Shown are Kaplan–Meier curves from a time-to-event analysis of the proportion of participants in the full preexposure analysis set who had a first SARS-CoV-2 RT-PCR–positive symptomatic illness, with a median follow-up of 6 months. The hazard ratio and corresponding 95% confidence interval were obtained from a Cox proportional-hazards model with the group as a covariate and with the stratification factor of age at informed consent (≥60 years or <60 years). Tick marks indicate censored data. The inset shows the same data on an enlarged y axis.
The efficacy of AZD7442 was consistent across subgroups of participants with data that could be evaluated; all point estimates of the relative risk reduction in the incidence of symptomatic illness with AZD7442 as compared with placebo were greater than 44% (Figure 1). Among participants who were at increased risk for either an inadequate response to Covid-19 vaccination or exposure to SARS-CoV-2, the relative risk reductions (80.7% and 82.6%, respectively) were similar to the relative risk reduction in the overall population in the primary efficacy analysis (76.7%). The time until symptomatic illness was longer with AZD7442 than with placebo (hazard ratio, 0.17; 95% CI, 0.08 to 0.33) (Figure 2).
Serum levels of AZD7442 remained elevated for 6 months after administration (Fig. S2A). The geometric mean (±SD) serum level of AZD7442 was 18.9±2.1 μg per milliliter at day 8 and 24.0±1.8 μg per milliliter at day 29, which translated to a SARS-CoV-2 geometric mean neutralizing antibody titer of 493.1 (95% CI, 469.3 to 518.1) at day 8 and 677.3 (95% CI, 647.1 to 709.0) at day 29 in an 80% plaque-reduction neutralization test that used wild-type virus (Fig. S2B). These titers were 16 times and 22 times as high, respectively, as those from samples of convalescent plasma obtained from patients with Covid-19.22
Viral genotypic data collected at illness visits were available for 7 of 11 symptomatic participants in the AZD7442 group and 13 of 31 symptomatic participants in the placebo group. Eleven of these participants were infected with a SARS-CoV-2 variant of concern,27 including 1 participant with B.1.351 (beta) in the AZD7442 group and 10 participants in the placebo group (5 participants with B.1.1.7_1 [an alpha subvariant] and 5 participants with B.1.617.2 [delta]) (Table S11).