Characteristics of the Participants
The data-cutoff date for the primary vaccine efficacy analysis was August 20, 2021. Of the 25,170 participants who were recruited, 24,141 underwent randomization in the intention-to-treat population. These participants had no virologic evidence of Covid-19 before receiving the trial injection. The safety population included 24,076 participants who had received one or more trial injections. The per-protocol population included 20,090 participants who had received two trial injections as scheduled and had no major protocol deviations. Participants may have discontinued their involvement in the trial after qualification as part of the per-protocol population (shown in the bottom set of boxes). An additional 10 participants withdrew from the study (4 in the vaccine group and 6 in the placebo group), and the timing of their discontinuation (by day 21, by day 42, or after day 42) could not be ascertained with confidence.
Details regarding the enrollment and randomization of the participants are provided in Figure 1. A total of 24,141 participants underwent randomization and were included in the intention-to-treat population. Demographic information and details after randomization were not available for 2 participants, who were not included in either efficacy or safety analyses.
The demographic and clinical characteristics of the participants in the intention-to-treat and per-protocol populations are provided in Tables S2 and S3. The median age of the participants was 29 years (range, 18 to 86), and 14.8% of the participants were seropositive at baseline. The intention-to-treat population included 21,651 healthy adults younger than 65 years of age (89.7%), 127 healthy adults 65 years of age or older (0.5%), and 2361 adults with coexisting illnesses (9.8%) (Table S4). The participants included 12,293 men (50.9%) and 11,846 women (49.1%). Participants were predominantly White (88.8%), and 82.0% reported having Hispanic or Latinx ethnicity. The high representation of Hispanic or Latinx participants reflected contributions from multiple clinical sites in Central and South America.
Despite an effort to enroll older adults and those with coexisting illnesses, these populations were underrepresented because of the timing of the trial during a surge of the B.1.617.2 (delta) and P.1 (gamma) variants in the trial areas, which increased the risk of participating in a clinical trial in those populations. The timing of the trial also led many participants to exercise their protocol-sanctioned option to withdraw or undergo unblinding to access another vaccine. This factor resulted in a progressive imbalance between the vaccine and placebo groups that was managed with the use of person-year denominators to calculate efficacy outcomes. The relevance and representativeness of the trial populations are discussed in Table S5. The participants who discontinued were more likely to be male or to identify as White, but there were no major differences between the groups (Table S6).
As of August 20, 2021, a total of 176 cases of Covid-19 that were predicted to contribute to the primary vaccine efficacy end point had been identified. Adjudication subsequently confirmed 165 cases. Data for 10 participants (9 in the vaccine group) were removed according to the protocol owing to unblinding before symptom onset; data for 1 participant did not meet the criteria for inclusion in the primary analysis of vaccine efficacy. An additional 7 participants in the placebo group who were subsequently adjudicated as having met the primary analysis criteria were not included in the analysis because of incomplete information by the data-cutoff date. Thus, the primary vaccine efficacy analysis included 165 cases in the intention-to-treat analysis and 157 cases in the per-protocol analysis.
In the intention-to-treat population, the median duration of time until data censoring for the efficacy analysis was 1.5 months (interquartile range [IQR], 0.8 to 2.0) in the vaccine group and 1.4 months (IQR, 0.8 to 1.9) in the placebo group. Among the 165 adjudicated cases, 40 occurred in the vaccine group and 125 in the placebo group, for an incidence rate of 0.025 (95% confidence interval [CI], 0.018 to 0.033) and 0.080 (95% CI, 0.068 to 0.096) per person-year, respectively.
The overall vaccine efficacy was 69.5% (95% CI, 56.7 to 78.8) in the intention-to-treat population, regardless of serostatus at baseline. In the per-protocol population, 39 cases occurred in 10,554 participants in the vaccine group and in 118 of 9536 participants in the placebo group, for a vaccine efficacy of 71.0% (95% CI, 58.7 to 80.0) (Table S7).
Shown are the results of the efficacy analysis of the CoVLP+AS03 vaccine in preventing Covid-19 according to subgroup of participants (Panel A) and according to variant (Panel B) in the intention-to-treat population; 22 samples had not yet been sequenced. In both panels, the incidence rate was calculated as the number of cases per person-year. Variant-specific values probably overestimate the true vaccine efficacy, since cases that were positive on polymerase-chain-reaction (PCR) assay but had no available sequencing data were asymmetrically distributed (14 in the vaccine group and 7 in the placebo group). The implications of these missing data for the variant-specific efficacy estimates are discussed in the Supplementary Appendix. Race or ethnic group was reported by the participants. NA denotes not applicable.
In the intention-to-treat population, the vaccine efficacy was 68.9% (95% CI, 55.0 to 78.9) in healthy adults younger than 65 years of age and 78.7% (95% CI, 30.2 to 95.1) in those with coexisting illnesses (Figure 2A). In the per-protocol population, the vaccine efficacy was 70.9% (95% CI, 57.7 to 80.4) and 76.8% (95% CI, 21.5 to 94.8), respectively. In both the intention-to-treat and per-protocol populations, among healthy adults who were 65 years of age or older, Covid-19 was detected in only 1 participant in each group, so the vaccine efficacy could not be determined. In the overall intention-to-treat population, 3 severe cases that led to 2 hospitalizations occurred in the placebo group.
Shown is the cumulative incidence of adjudicated Covid-19 in the intention-to-treat population (Panel A), in the per-protocol population (Panel B), and according to the presence of the delta or gamma variant in the intention-to-treat population (Panel C), starting 7 days after the second vaccination. Covid-19 cases were confirmed on the basis of PCR-positive nasopharyngeal swabs and were independently adjudicated by a subcommittee of the data and safety monitoring committee.
Overall vaccine efficacy in preventing moderate-to-severe disease (post hoc analysis) was 78.8% (95% CI, 55.8 to 90.8) in the intention-to-treat population. Among the participants who were seronegative at baseline, the vaccine efficacy against moderate-to-severe disease was 86.0% (95% CI, 66.2 to 95.1). There were no Covid-19–related deaths. The cumulative incidence curves for all cases and for cases caused by the delta and gamma variants are presented in Figure 3.
During the trial period, circulating variants differed in the participating countries, according to genomic sequences that were shared through GISAID, a global science initiative that provides open access to genomic data (Fig. S1).24 The delta and gamma variants dominated in Argentina and Brazil, and the B.1.1.7 (alpha) and delta variants dominated in North America and the United Kingdom. Participants with positive results for Covid-19 were largely from Argentina (59 cases), Brazil (53 cases), and the United States (47 cases).
Of the 165 cases in the intention-to-treat population that were included in the primary vaccine efficacy analysis, sequencing data were available for 122 participants (73.9%); samples that were obtained from an additional 21 participants with confirmed cases (12.7%) could not be sequenced. All sequenced strains were variants of the original strain: delta in 56 participants (45.9%), gamma in 53 (43.4%), alpha in 6 (4.9%), B.1.621 (mu) in 4 (3.3%), and C.37 (lambda) in 3 (2.5%).
When the viral load in the upper respiratory tract was sufficient to permit sequencing, the overall variant-specific efficacy estimates were 87.8% (95% CI, 73.0 to 95.3) for the gamma variant, 74.0% (95% CI, 51.7 to 86.8) for the delta variant, and 100% for the alpha, lambda, and mu variants (Figure 2B). However, these values probably overestimate the true variant-specific efficacy, since cases that were PCR-positive but had negative results on sequencing were asymmetrically distributed (14 in the vaccine group and 7 in the placebo group). The implications of the missing data for the variant-specific efficacy estimates are discussed in the Supplementary Appendix.
Panel A shows viral loads for patients for whom data were available in the two groups, arranged in violin plots. Within each plot, the red dashed line indicates the median, and the black dashed line indicates the lower limit of detection (LLOD). Panel B shows mean viral loads, presented as log10 virus copies per milliliter, according to subgroup. All analyses were performed in the intention-to-treat population. NE denotes not estimable.
Measurement of viral load over time after diagnosis was a prespecified outcome of the trial. However, the between-group difference in sequencing success (nearly twice as high in the placebo group as in the vaccine group) prompted analysis of viral load only at the time of Covid-19 diagnosis. Such analysis revealed a higher viral load in the placebo group by a factor of more than 100 (3.46 log10 copies per milliliter in the vaccine group vs. 5.65 log10 copies per milliliter in the placebo group) (Figure 4). The median viral load in cases that were PCR-positive but sequence-negative was at the limit of detection (2.08 log10 [or 120] copies per milliliter) as compared with more than 500,000 copies per milliliter in cases that were PCR- and sequence-positive. Viral loads in the breakthrough cases with the delta variant in the vaccine group were lower by a factor of 42 than those in the placebo group (3.65 log10 vs. 5.27 log10 copies); the corresponding values in the breakthrough cases with the gamma variant were lower by a factor of 269 (3.78 log10 vs. 6.21 log10 copies). A similar trend of lower viral loads in the vaccine group was observed in breakthrough cases that were classified as mild (by a factor of 138) or moderate (by a factor of 426).
Participants were monitored for local and systemic solicited adverse events through 7 days after administration of the trial injection. Participants who reported having no adverse events or who had missing data make up the remainder of the 100% calculation (not shown). For each category, adverse events are classified as follows: grade 1, mild; grade 2, moderate; and grade 3, severe. In addition, six grade 4 events (potentially life-threatening) were reported in 3 participants after the second injection: in 2 participants in the vaccine group (1 participant with chills, general discomfort, headache, and muscles aches and 1 participant with fever) and in 1 participant in the placebo group (headache) (proportional representation not visible in the graph). If a participant had different grades of the same adverse event, the highest grade was reported. If any of the solicited adverse events persisted beyond day 7 after administration of the trial injection, it was recorded as an unsolicited adverse event. Fever was defined as an oral temperature of at least 38.0°C.
The safety population included 24,076 participants (12,036 in the vaccine group and 12,040 in the placebo group). Solicited adverse events up to 7 days after each dose were analyzed in 7819 participants (Figure 5). Both local and systemic solicited adverse events were predominantly mild-to-moderate and transient (duration, 1 to 3 days). More local and systemic adverse events were reported in the vaccine group than in the placebo group. Overall, local solicited adverse events were reported after the first and second doses combined in 3819 participants (92.3%) in the vaccine group and in 1677 (45.5%) in the placebo group. Systemic solicited adverse events were reported in 3612 participants (87.3%) and 2394 (65.0%), respectively.
In the two groups, local reactogenicity was primarily injection-site pain (Table S9). The most common systemic adverse events were headache, myalgia, fatigue, and a feeling of general discomfort (Table S10), with more participants in the vaccine group reporting these events. Grade 2 and 3 local and systemic adverse events occurred more frequently after the second dose. Grade 3 (severe) local adverse events were reported in 2.1% of the participants in the vaccine group and in less than 0.1% of those in the placebo group after the second dose; grade 3 systemic reactions were reported in 3.1% and 0.5%, respectively. No grade 4 (life-threatening) local adverse events were reported, but 3 participants reported grade 4 systemic adverse events after the second dose: 2 in the vaccine group (chills, headache, muscles aches, and a feeling of general discomfort in 1 participant and fever in the other) and 1 in the placebo group (headache).
The occurrence and intensity of unsolicited adverse events in each group are provided in Tables S11 and S12, along with data regarding serious and medically attended adverse events, events leading to withdrawal from the trial, events of special interest, and deaths. The incidence of unsolicited adverse events after the first or second dose was slightly higher in the vaccine group than in the placebo group (22.7% vs. 20.4% up to 21 days after the second dose; 4.2% vs. 4.0% from day 43 to day 201). Unsolicited adverse events with a frequency of at least 1% after the first or second dose according to the MedDRA preferred terms are listed in Table S13. The incidence of serious adverse events was similar in the two groups up to 21 days after the first or second doses (24 participants in the vaccine group [0.2%] vs. 16 in the placebo group [0.1%]) and between days 43 and 201 (19 [0.2%] vs. 22 [0.2%], respectively). One participant in the placebo group reported two serious adverse events (aortic thrombosis and peripheral artery thrombosis) that were considered by the site investigator to be related to the trial injection. No deaths were considered to be related to the vaccine.