Possible influence of UK SARS-CoV-2 variant(s) on efficacy of ChAdOx1 nCoV-19 vaccine
Of note, comparison of vaccine efficacy data between the two different cutoff dates for the participants in LD/SD cohort (the only that remained constant in numbers and therefore comparable in longitudinal analyses), the values may suggest a slight decrease of protection (3 ). Indeed, the relative risk of infection in LD/SD group at the first data cutoff date (November 4, 2020) was 0.10 (95% CI 0.03–0.33) and 0.41 (95% CI 0.17–0.99) for symptomatic and asymptomatic disease, respectively (2 ), while they were subsequently estimated to be respectively 0.19 (95% CI 0.10–0.38) and 0.51 (95% CI 0.28–0.93) at the second data cutoff (December 7, 2020) (3 ), possibly suggesting a slight decrease of vaccine efficacy during the last time period of about a month. Indeed, at the first data cutoff date (Nov 4, 2020), within the LD/SD group, the symptomatic infected individuals were 3 in 1367 participants (0.2%) in vaccinated group and 30 in 1374 participants (2.2%) in control group. Instead, at the second data cutoff date (Dec 7, 2020), within the same LD/SD group, the symptomatic infected individuals were 10 in 1396 participants (0.7%) in vaccinated group and 51 in 1402 participants (3.6%) in control group. That means that during the time between the two data cutoff dates (basically the month of November) there were 7 symptomatic infections in 1393 participants (1396 minus 3 already infected) (0.50%) in vaccinated group and 21 in 1372 participants (1402 minus 30 already infected) (1.53%) in control group, which correspond to 0.33 (95% CI 0.14–0.77) of relative risk of symptomatic infection that is about 3 time higher than that of the first time period. Notably, the LD/SD cohort was enrolled between May 31 and June 10, 2020 (2 ) and most of them had a boost dose about 3 months later (median 84 days, interquartile range 77—91) (3 ), that means that booster doses occurred between the end of August and the beginning of September (2 ). Therefore, at the two data cutoff dates (Nov 4 and Dec 7, 2020), LD/SD cohort was respectively analysed about two and three months after the booster dose. During the longitudinal study, the frequency of infected individuals in control group was 2.2% in the first time period (data cutoff date Nov 4) and 1.53% in the second time period (between the two data cutoff dates). Since in UK the frequency of infected individuals increased during the month of November, it confirms that the second time period was shorter than the first; nevertheless, the frequencies of spontaneous infection were somehow similar (and comparable) between the two groups. In addition, during the time between the two data cutoff dates, a similar trend of the relative risk of infection was observable for the asymptomatic (transmissible) infection, for which the vaccination in LD/SD cohort reached a relative risk of 0.64 ([95% CI 0.28–1.48], it was 0.41 [95% CI 0.17–0.99] on Nov 4, 2020), suggesting that the booster dose provided a protection for two months after which the immune protection seems to start to wane. Instead, no waning of vaccine efficacy was detected between 22 and 90 days after single standard dose. Indeed, the relative risk of symptomatic infection remained stable (median 0.24 [95% CI 0.14–0.41]) until 90 days (3 months) after vaccination, despite a 34% reduction (GMR 0.66 [95% CI 0.59–0.74]) of anti-SARS-CoV-2 spike IgG responses after 90 days from the peak at day 28 (median 5496 AU/ml for participants aged 56–69 years and 9807 AU/ml for participants aged 18–55 years) (3, 5 ). Intriguingly, the relative risk of symptomatic infection after a single standard dose within 90 days (0.24 [95% CI 0.14–0.41]) was similar to that between the two data cutoff dates (about 60 and 90 days after booster dose) in LD/SD group (0.33 [95% CI 0.14–0.77]). In this regard, it is expected that the prime-boost regimen can induce both higher levels of neutralising antibodies and longer time protection (for several months) than a single dose. Indeed, 28 days after the second dose (about one month before the first data cutoff date), the anti-SARS-CoV-2 spike IgG responses in LD/SD group were extremely high (median 39670 AU/ml, [IQR 21068–66338] 9-11 week interval and 49584 AU/ml, [IQR 31122–81163] ≥12 week interval for participants aged 18–55 years) compared with those induced after 28 days by single (standard or low) doses (single low dose, median 6439 AU/ml [IQR 4338-10640] for participants aged 18–55 years) (3,5 ), suggesting that the decrease of anti-SARS-CoV-2 immune protection may not be due to the decline of anti-SARS-CoV-2 antibodies. Since the protection induced by a single dose lasted for at least 3 months (despite a significant reduction of anti-SARS-CoV-2 spike IgG responses) and at 28 days after the second dose (one month earlier than the first data cutoff, November 4, 2020) participants in LD/SD cohort displayed an expression of anti-SARS-CoV-2 spike IgG about five fold higher than single (standard or low) dose (3,5 ), it is unlikely that the decreased vaccine efficacy at the second data cutoff date (Dec 7, 2020) may depend on concentrations of anti-SARS-CoV-2 antibodies. Rather, the relatively low protective efficacy recorded during the month of November after the booster dose of ChAdOx1 nCoV-19 vaccine (relative risk of symptomatic infection 0.33 [95% CI 0.14–0.77]) compared to single dose analyses (relative risk of symptomatic infection 0.24 [95% CI 0.14–0.41]) assessed between 22 and 90 days (for which most data were collected between June and October) suggests that other factors might be at play late after the booster dose. Of note, during the same period of time (basically the month of November), the vaccine efficacy of single standard dose (between 91 and 120 days after administration) similarly wane, reaching a relative risk of symptomatic infection of 0.68 (95% CI 0.19–2.42) (3 ), thus suggesting a common factor that led to a general reduction of ChAdOx1 nCoV-19 vaccine efficacy in that specific temporal period. In this regard, the impairment of vaccine-induced immune protection could be due to reduction not only in concentration, but also in specificity of anti-SARS-CoV-2 neutralizing antibodies. Therefore, it is possible that loss of antibody recognition might be involved in the reduction of immune protection “in vivo”. In this regard, evaluation of anti-SARS-CoV-2 spike IgG responses has been assessed using the ancestral spike protein, it is therefore possible that the emergence SARS-CoV-2 variants with spike protein mutations may be at the origin of the discrepancy between anti-SARS-CoV-2 spike IgG responses and vaccine efficacy late during the vaccine trials. In this regard, the B.1.1.7 (UK) variant, which carries several mutations including spike protein, started circulating in England in late September and became the dominant lineage in December (8 ). In the UK, the proportion of the B.1.1.7 variant has increased from 0.1% in early October to 49.7% in late November among sequences available at 19 December 2020 (8 ), suggesting a cause-effect relationship between B.1.1.7 expansion and a possible decrease of efficacy of ChAdOx1 nCoV-19 vaccine against symptomatic infection during the two data cutoff dates of the reports (Nov 4, 2020 and Dec 7, 2020).
B.1.1.7 variant contains 8 spike mutations in addition to D614G, including one mutation (N501Y) in receptor binding domain (RBD), two deletions (69-70del and 144del) in the N-terminal domain (NTD) of the spike, and one mutation (P681H) near the furin cleavage site (9-11 ). However, the B.1.1.7 variant seems susceptible to neutralising antibodies elicited by ancestral spike vaccines (9 ), rather it has an enhanced binding to ACE-2, a higher reproduction and an increased transmission that gives it a competitive advantage in humans (9,10 ). Nevertheless, neutralization by serum samples from recipients of vaccines with ancestral spike was moderately reduced and a subset of monoclonal antibodies to the RBD of spike is less effective against the B.1.1.7 variant (9 ), raising the possibility of a moderate increased risk of infection and virus transmission after vaccination with ancestral spike sequences. In line with this possibility, another report found that B.1.1.7 is refractory to neutralization by most monoclonal antibodies to the NTD of the spike and relatively resistant to a few monoclonal to ancestral RBD, which could cause escape from neutralizing antibody control in vivo, thus threatening the protective efficacy of current vaccines (11 ). In this regard, a recent sequencing of the B.1.1.7 variant revealed the presence of the E484K mutation (first identified in South Africa) (12 ) and several studies showed reduced neutralising activity of monoclonal antibodies from convalescent or vaccinated individuals against virus mutants containing the E484K mutation (13-16 ). Moreover, the presence of the N439K mutation, which has emerged independently in multiple variant lineages, has been shown to increase both spike binding affinity for human ACE2 and resistance to several anti-SARS-CoV-2 neutralizing antibodies, which give SARS-CoV-2 variants carrying N439K a selective advantage (17 ). Altogether these observations suggest that neutralising antibodies elicited by ancestral spike vaccines induce cross-protection from B.1.1.7 variant; however, they may not be able to fully protect from the UK variant, and in particular, from its transmission. In this regard, a post-hoc analysis of the efficacy of ChAdOx1 nCoV-19 vaccine against B.1.1.7 variant have shown that clinical efficacy against symptomatic infection was 70.4% (95% CI 43.6–84.5) (while it was 81.5% [95% CI 67.9–89.4] for non-B.1.1.7 lineages) and it was 28.9% (95% CI -77.1–71.4) against asymptomatic infection (18 ); values that are very similar to those calculated in the present report for the LD/SD group between November 4 and December 7, which were respectively 67.2% (95% CI 23.0–86.0) and 35.9% (95% CI -47.6–72.2). In line with these observations, neutralisation activity of vaccine-induced antibodies in a live-virus neutralisation assay has been shown to be about nine times lower against the B.1.1.7 variant than against the ancestral lineage (GMR 8.9 [95% CI 7.2–11.0]) (18 ). Notably, participants were recruited between May 31 and Nov 13, 2020, therefore before expansion and acquired new mutations (evolution) of B.1.1.7 variant. It is therefore likely that the relative increase risk of symptomatic infection in LD/SD cohort between the two data cutoff dates of the reports (Nov 4, 2020 and Dec 7, 2020) may be related to the emerging UK variant(s), which was dominant in that period during the UK pandemic (8 ). Therefore, although actual vaccines are able to protects from severe forms of COVID-19 and its transmission, the present work suggests that ChAdOx1 nCoV-19 vaccine may not be able to block UK variant transmission as for ancestral virus. Whether this possibility may occur only with ChAdOx1 nCoV-19 vaccine or also with other vaccines based on the ancestral spike sequences is still unclear because data for longitudinal studies as those performed for the ChAdOx1 nCoV-19 vaccine are not available. Nevertheless, a recent preprint report (data from 1 December 2020 to 3 April 2021, a period in which B.1.1.7 was dominant) showed no difference of protection between a single dose of ChAdOx1 nCoV-19 and BNT162b2 vaccine (after either one or two vaccine doses) in a representative sample across the UK (19 ).