Durability of protection and immunogenicity of AZD1222 (ChAdOx1 nCoV-19) COVID-19 vaccine over 6 months

Abstract

BackgroundWe report updated safety, efficacy, and immunogenicity of AZD1222 (ChAdOx1 nCoV-19) from an ongoing phase 3 trial.MethodsAdults at increased risk of SARS-CoV-2 infection were randomized (2:1), stratified by age, to receive 2 doses of AZD1222 or placebo. The primary efficacy end point was confirmed SARS-CoV-2 reverse-transcriptase PCR-positive (RT-PCR-positive) symptomatic COVID-19 at 15 or more days after a second dose in baseline SARS-CoV-2-seronegative participants. The 21,634 and 10,816 participants were randomized to AZD1222 and placebo, respectively.FindingsData cutoff for this analysis was July 30, 2021; median follow-up from second dose was 78 and 71 days for the double-blind period (censoring at unblinding or nonstudy COVID-19 vaccination) and 201 and 82 days for the period to nonstudy COVID-19 vaccination (regardless of unblinding) in the AZD1222 and placebo groups, respectively. For the primary efficacy end point in the double-blind period (141 and 184 events; incidence rates: 39.2 and 118.8 per 1,000 person years), vaccine efficacy was 67.0% (P < 0.001). In the period to nonstudy COVID-19 vaccination, incidence of events remained consistently low and stable through 6 months in the AZD1222 group; for the primary efficacy end point (328 and 219 events; incidence rates: 36.4, 108.4) and severe/critical disease (5 and 13 events; incidence rates: 0.6, 6.4), respective vaccine efficacy estimates were 65.1% and 92.1%. AZD1222 elicited humoral immune responses over time, with waning at day 180. No emergent safety issues were seen.ConclusionAZD1222 is safe and well tolerated, demonstrating durable protection and immunogenicity with median follow-up (AZD1222 group) of 6 months.Trial registrationClinicalTrials.gov NCT04516746.FundingAstraZeneca; US government.

Keywords: Adaptive immunity; COVID-19.

Figure 1. Participant disposition during trial.

Overall number randomized and number randomized to AZD1222 are 1 lower than in the primary analysis (8) due to identification of double-counting of 1 participant. In the placebo arm, 1 participant was not included in the primary analysis due to record deactivation, but has been reinstated at this analysis. FVAS population numbers differ from those in the primary analysis because, with additional follow up, additional participants achieved the milestone of 15 days after the second dose and became eligible for these populations, but also some participants were excluded from these populations based on newly obtained information regarding prior infections, baseline serology and receipt of nonstudy COVID-19 vaccinations.

Figure 2. Forest plot of estimated efficacy of AZD1222 versus placebo in the double-blind period.

Plot shows estimated AZD1222 efficacy 15 or more days after the second dose for the primary and secondary efficacy end points in the FVAS population for the double-blind period, with censoring for unblinding or nonstudy COVID-19 vaccination (AZD1222, n = 17,617; placebo, n = 8528). Total follow-up was 3.60 and 1.55 × 1000 person years in the AZD1222 and placebo groups. The dotted vertical line represents the nominally statistically significant criterion of a lower CI greater than 30% applicable to the primary end point and is shown for reference. VE was calculated as (1 minus relative risk) × 100, with relative risk estimated using Poisson’s regression model with robust variance adjusted for follow-up time and with trial group and age group (18–64 versus ≥65 years) as covariates. *Per 1000 person years. †The FVAS includes all participants who were SARS-CoV-2 seronegative at baseline; this population (n = 18,450 in AZD1222 group; n = 8960 in placebo group) includes participants regardless of prior SARS-CoV-2 infection. ‡P < 0.001; ^§P = 0.03. IR, incidence rate.

Figure 3. Estimated efficacy of AZD1222 versus placebo for the prevention of COVID-19 and SARS-CoV-2 infection during the period to nonstudy COVID-19 vaccination.

(A) Cumulative incidence of SARS-CoV-2 RT-PCR–positive symptomatic illness occurring 15 or more days after the second dose (time 0 = day 15 after the second dose) in the FVAS population for the period to nonstudy COVID-19 vaccination (AZD1222, n = 19,569; placebo, n = 8868), with censoring for nonstudy COVID-19 vaccination, regardless of unblinding. (B) Incidence of SARS-CoV-2 RT-PCR–positive symptomatic illness events and decrease in the at-risk population over time from first dose during the period to nonstudy COVID-19 vaccination. The at-risk population curves show the numbers of participants in the FVAS who have not been censored and were available for analysis at the corresponding time point. Cumulative incidence of (C) severe or critical symptomatic COVID-19 and (D) SARS-CoV-2 infection, as defined by seroconversion rate from negative at baseline to positive for SARS-CoV-2 nucleocapsid antibody at 15 or more days after the second dose, regardless of symptoms, in the FVAS population for the period to nonstudy COVID-19 vaccination (AZD1222, n = 19,569; placebo, n = 8868), with censoring for nonstudy COVID-19 vaccination, regardless of unblinding. For panels A, C, and D, time to first event was from time of second dose administration, calculated as follows: (date of SARS-CoV-2-positive test) – (date of second dose of AZD1222 or placebo + 14 days) + 1. For censored participants, censoring time was from date of second dose of AZD1222 or placebo plus 14 days to the last time observed before data cutoff (July 30, 2021). Cumulative incidences were estimated using the Kaplan-Meier method. Cumulative incidence curves were truncated at the point at which less than 10% of the starting population remained at risk. IR, incidence rate per 1000 person years.

Figure 4. Forest plot of efficacy estimates of AZD1222 versus placebo in the period up to receipt of nonstudy COVID-19 vaccination.

Plot shows estimated AZD1222 efficacy 15 or more days after the second dose for the primary, secondary, and exploratory efficacy end points in the FVAS population for the period to nonstudy COVID-19 vaccination, with censoring for nonstudy COVID-19 vaccination, regardless of unblinding (AZD1222, n = 19,569; placebo, n = 8868). Total follow-up was 9.01 and 2.02 × 1000 person years in the AZD1222 and placebo groups. The dotted vertical line represents the nominally statistically significant criterion of a lower CI greater than 30% applicable to the primary end point and is shown for reference. VE was calculated as (1 minus relative risk) × 100, with relative risk estimated using Poisson’s regression model with robust variance adjusted for follow-up time and with trial group and age group (18–64 versus ≥65 years) as covariates. *Per 1000 person-years. †Results from an IPCW method applied to right censoring and using standardized weights. See Statistics section of Methods for methodology and Supplemental Table 11 for additional analyses and information. ‡The FVAS includes all participants who were SARS-CoV-2 seronegative at baseline; this population (n = 20,479 in AZD1222 group; n = 9312 in placebo group) includes participants regardless of prior SARS-CoV-2 infection.

Figure 5. Neutralizing antibody responses over time in the AZD1222 group in the immunogenicity substudy population.

Box and whisker plots showing SARS-CoV-2 neutralizing antibody quantitation over time, (A) overall, and by (B and C) age and (D–G) race/ethnicity (data on race/ethnicity missing for n = 21 participants). Participants were censored at the earliest date of nonstudy COVID-19 vaccination, positive test for SARS-CoV-2 nucleocapsid antibodies, RT-PCR–positive SARS-CoV-2 symptomatic infection, or last trial contact. The y axes shows 1/dilution; for conversion to the WHO International Standard, see Supplemental Methods. Box indicates interquartile range, whiskers indicate range, horizontal line in box indicates median, and dot in box indicates mean.

Figure 6. Incidence of SARS-CoV-2 variants, numbers at risk, and unblinding/receipt of nonstudy COVID-19 vaccination over the time course of the trial.

(A) Incidence of variants observed in cases of RT-PCR–confirmed SARS-CoV-2 infection in the placebo and AZD1222 arms of the trial, truncated at the point at which less than 10% of the starting population remained at risk, and incidence of confirmed cases in population data from the US, Peru, and Chile during the time of study (data derived from COVID-19 Data Repository by the Center for Systems Science and Engineering [CSSE] at Johns Hopkins University, ref. , available at https://github.com/owid/covid-19-data/tree/71b0337018fe20d469aa9014e3a8003d900a2b5b, commit ID: 71b0337018fe20d469aa9014e3a8003d900a2b5b; and from GISAID, EPI_SET_220825fk, https://doi.org/10.55876/gis8.220825fk), along with timing of participants being on study, with censoring for (B) unblinding or nonstudy COVID-19 vaccination, or (C) nonstudy COVID-19 vaccination only in the FVAS populations for these 2 analyses.