key: cord-0315691-ed9javzj
authors: Cromer, D.; Steain, M.; Reynaldi, A.; Schlub, T. E.; Wheatley, A. K.; Juno, J. A.; Kent, S. J.; Triccas, J. A.; Khoury, D. S.; Davenport, M. P.
title: SARS-CoV-2 variants: levels of neutralisation required for protective immunity
date: 2021-08-13
journal: nan
DOI: 10.1101/2021.08.11.21261876
sha: 577d14a4eac1f82e88d35d376df5ba4aec88fed5
doc_id: 315691
cord_uid: ed9javzj

A number of SARS-CoV-2 variants of concern (VOC) have been identified that partially escape serum neutralisation activity elicited by current vaccines. Recent studies have also shown that vaccines demonstrate reduced protection against symptomatic infection with SARS-CoV-2 variants. Here we integrate published data on in vitro neutralisation and clinical protection to understand and predict vaccine efficacy against existing SARS-CoV-2 variants. We find that neutralising activity against the ancestral SARS-CoV-2 is highly predictive of neutralisation of the VOC, with all vaccines showing a similar drop in neutralisation to the variants. Neutralisation levels remain strongly correlated with protection from infection with SARS-CoV-2 VOC (r=0.81, p=0.0005). We apply an existing model relating in vitro neutralisation to protection (parameterised on data from ancestral virus infection) and find this remains predictive of vaccine efficacy against VOC once drops in neutralisation to the VOC are taken into account. Modelling of predicted vaccine efficacy against variants over time suggests that protection against symptomatic infection may drop below 50% within the first year after vaccination for some current vaccines. Boosting of previously infected individuals with existing vaccines (which target ancestral virus) has been shown to significantly increase neutralising antibodies. Our modelling suggests that booster vaccination should enable high levels of immunity that prevent severe infection outcomes with the current SARS-CoV-2 VOC, at least in the medium term.

1 SARS-CoV-2 variants: levels of neutralisation 5 required for protective immunity Deborah Cromer 1* , Megan Steain 2,3 , Arnold Reynaldi 1 , Timothy E Schlub 1,4 , Adam K Wheatley 5 , Jennifer A Juno 5 , Stephen J Kent 5, 6 , James A Triccas 3,7# , David S Khoury 1#* , Miles P Davenport 1# . 10

The global spread of SARS-CoV-2 has resulted in significant morbidity, mortality, and social disruption. Several vaccines have been deployed that protect against symptomatic SARS-CoV-2 infection (reviewed in 1 ). Vaccines in current use incorporate the ancestral (Wuhanlike) virus or viral spike protein as an immunogen. Both vaccination and prior infection have been shown to provide a degree of protection against symptomatic and severe infection with 60 essentially homologous virus [2] [3] [4] . Recently, several SARS-CoV-2 variants of concern (VOC) have emerged that display increased transmissibility and / or reduced in vitro neutralisation by sera from subjects infected with the ancestral strain or immunised with current vaccines [5] [6] [7] .

Initial reports from clinical trials or from breakthrough community infections suggest that current vaccines may be less protective against symptomatic infection with some SARS-65

CoV-2 variants 8- 12 . In addition, studies also demonstrate that waning antibody levels correlate with reduced protection over time 13, 14 . Thus, a major question is the extent to which existing vaccines are likely to protect against variants of concern and how existing vaccines might be used to combat the threat of variants. 70 Current vaccines have been shown to elicit different levels of neutralising antibody in vaccinated subjects, ranging from ~0.2-fold to ~4-fold of the levels seen in early convalescence [15] [16] [17] . Studies analysing vaccine-induced neutralising antibody responses have reported varying levels of reduction in neutralisation titre against variants of concern (VOC).

However, the lack of a standardised assay to measure in vitro neutralisation means that the 75 absolute serum dilution titres sufficient to neutralise either ancestral or variant viruses differ considerably between laboratories 18 . There are also now a number of studies reporting vaccine efficacy against variants, which also indicate a variable reduction in efficacy 10, 19 . Our previous work has shown a correlation between neutralising antibody levels and protection from SARS-CoV-2 infection and has derived a model for predicting vaccine efficacy from 80 mean neutralisation titres 14 . However, this model was developed based on neutralisation and protection from the ancestral virus, and whether this remains predictive for efficacy against SARS-CoV-2 VOC has not been determined. A predictive model of vaccine protection against SARS-CoV-2 variants would allow us to forecast the ongoing utility of current vaccines in the face of emerging variants, and to model the utility of boosters and other 85 strategies to extend the duration of vaccine protection.

In this study we aim to determine the utility of existing vaccines (targeting ancestral virus) for protecting against VOC. We first compare the observed in vitro neutralisation titres from convalescent subjects and vaccinees immunised with 4 different vaccines against all current 90 4 SARS-CoV-2 VOCs: namely alpha (B.1.1.7), beta (B.1.351), gamma (P.1) and delta (B.1.617.2). We find that the reported reductions in serum neutralisation activity against SARS-CoV-2 VOC in vitro is independent of whether immunity was established by prior infection or vaccination, and was consistent between vaccine platforms. Combining this in vitro data with clinical studies of vaccine efficacy, we show that in vitro neutralisation is 95 significantly correlated with protection from infection with SARS-CoV-2 variant virus in vaccinated individuals. This relationship is consistent with the existing model relating in vitro neutralisation titre to observed protection, suggesting this model can be used to predict vaccine efficacy against SARS-CoV-2 variants. Using this model, we study the effects of immune boosting of previously infected subjects. This work suggests that losses of vaccine 100 effectiveness in the face of evolving variant viruses and waning immunity may be partially offset by timely booster immunisations.

All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint

Results:

It has previously been shown that neutralising antibodies against ancestral virus correlate with vaccine efficacy 14, [20] [21] [22] . In order to determine if neutralisation responses continue to correlate with protection for VOC, a robust estimate of the loss of neutralising activity 110 against VOC for each vaccine is needed. We examined the loss of neutralisation against VOC by combining data from 16 published studies which directly compared neutralisation titre against ancestral (Wuhan-like / D614G) strains and the VOC (see supplementary table 1) 5, 7, [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] . Different laboratories used distinct in vitro assays to measure neutralisation of SARS-CoV-2 (supplementary table 1) 37 , and reported considerably different means and fold-115 changes in neutralisation to variants (Fig 1a) . However, within a given study the observed drop in neutralisation titre for a given variant was very similar for both convalescent and vaccinee serum, suggesting a strong study-specific (or assay-specific) effect (supplementary Figure S1 ). To explore the contribution of assay-specific and vaccine-specific effects, we used a regression model (with censoring) to aggregate data from all studies and included 120 potential for assay-specific, variant-specific and vaccine-specific effects (supplementary analysis and methods). From this regression we found that neutralisation against ancestral virus was strongly associated with neutralisation against a particular variant, and that after adjusting for variant and laboratory effects, whether immunity was acquired through infection or vaccination (and which vaccine was used) was not a significant factor associated 125 with the loss of neutralisation (p=0. 26 

Several studies have now shown reduced efficacy of vaccination against infection with SARS-CoV-2 variants 10, 11, 19, [38] [39] [40] [41] (supplementary table 2 ). These studies incorporated a variety of study designs, including both randomised controlled trials (RCT) 10, 19, 40, 41 and All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint 6 observational case-control studies 38, 39 . Vaccine efficacy against the variants was assessed either by separately analysing the number of infections with variant virus 38 or on the 140 assumption (based on epidemiological data) that the vast majority of infections were with variant virus 19, 40 . In addition to these differences in study design, other factors such as the definitions of mild, moderate, and severe infection also differed by study (Supplementary Table 2 ). We used the correlation between ancestral and variant neutralisation titres described above (Figure 1b) to estimate the neutralisation level of each vaccine-variant combination, 145 and related this to the protection observed for that vaccine-variant combination ( Figure S4 ).

Despite the variability in study design, we find that predicted serological neutralisation activity against each variant elicited by vaccines is significantly correlated with protection from COVID-19 (r=0.81, p=0.0005, Spearman, Supplementary Figure S4 ).

To test whether our previously developed predictive model could also be used to predict efficacy against VOCs, we took the same model (parameterised from ancestral virus) and 

Predicting vaccine efficacy against severe SARS-CoV-2 infection is significantly more challenging, due to the low numbers of severe infections captured in most efficacy studies 14 .

All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint However, even with this limitation, vaccination has been shown to provide significantly better protection against severe disease than against symptomatic SARS-CoV-2 infection 42 .

Thus we next considered our model's prediction of efficacy against severe outcomes with 175 VOC (Figure 2b ). The 95% confidence intervals are substantially broader for severe than for symptomatic SARS-CoV-2 infection, indicating the greater uncertainty in the data on protection from severe disease. Further, the point estimates of efficacy from the clinical studies contain considerable uncertainty (Supplementary Table 4 ). Even so, all but one of the efficacy studies falls within or above the predicted efficacy confidence limits. It should be 180 noted that our model assumes neutralisation alone drives protection against severe disease, but it is likely that other cellular responses play a critical role in modulating disease severity, and thus the model may underestimate efficacy against severe COVID-19.

The models above and emerging data on breakthrough infections suggest that booster vaccinations may eventually be required to both augment waning immunity and boost responses to variants. A major question is whether boosting with existing vaccines (that all currently incorporate only the ancestral spike) will be effective in providing protection Table S4) 43-48 and additionally, improved the level of cross-reactivity to variants [44] [45] [46] [47] . Although the level of increase varied between studies, vaccination of convalescent individuals led to boosting of 200 neutralisation levels to approximately 12-fold (range, 6.1-fold to 29-fold) higher than that seen in early convalescence (and higher than that seen in any current vaccination regimen) (Figure 3 ).

We use our existing model to predict the impact of boosting, with existing vaccines targeting 205 ancestral virus, on protection against symptomatic (figure 3a) and severe infection with All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint SARS-CoV-2 VOC (figure 3b). This prediction assumes that the relationship between neutralisation and protection continues to hold after boosting and that the drop in titre against the variants is also similar after boosting 47 . As noted above, boosting may actually increase cross reactivity against VOC, and thus the protection shown here is likely a minimum bound 210 on the true protection conferred from boosting. Regardless, this suggests that boosting of previously infected individuals using existing vaccines should lead to both higher neutralising titres and higher protection from symptomatic and severe COVID-19 (figure 3), consistent with a recent study of vaccine efficacy after vaccination of either naïve or previously infected individuals 49 . 215

There is currently more limited data available on the effects of boosting in vaccinated individuals (blue vertical lines, Figure 3 ). Wu et al 50 showed that a third dose of mRNA-1273 delivered 6 months after the initial vaccination boosted neutralisation levels by around 23fold compared to the pre-boost levels, or around 2.5-fold higher than vaccination of naïve 220 individuals 48 . Pan et al 51 have studied the effects of a third dose of CoronaVac delivered either 1 month or 6 months after the second dose. They found that a third vaccination at 6 months boosted responses approximately 3 to 5-fold higher than seen after the initial two dose regime. Interestingly, boosting at 1 month after the initial two dose regime led to only a 1.3-2.1-fold increase, suggesting that delayed boosting may be required. The level of 225 boosting of vaccinated individuals is indicated as vertical blue lines in Figure 3 .

Previous studies have shown a decrease in neutralising antibody titres over the first 8 months of infection, with a half-life of around 3-4 months 52 , and recent work has also shown a decrease in vaccine protection over this period 13 . This waning immunity, coupled with the 230 drop in neutralisation titres to the VOC, has the potential to reduce protection over the first year after vaccination or infection (Figure 4) 13, 14 . However, vaccination of previously infected individuals or boosting of previously vaccinated individuals has the potential to raise neutralising antibody levels above those seen after primary vaccination. Assuming that the decay of neutralisation titres after boosting is consistent with decay after primary infection 52-235 54 , vaccination of convalescent individuals is predicted to provide 69% protection from symptomatic infection (lower bound CI: 47%) and 94% protection from severe infection (lower bound CI: 74%) even against the most escaped VOC (beta) 6-months after boosting (figure 4) 49 . Although there are limited data on the effects of boosting of vaccinated individuals, preliminary data suggests that boosting with mRNA vaccines may be capable of 240 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint 9 achieving similar levels of neutralising responses (Figure 3) 50 . Further work is required to understand the optimal vaccination and boosting schedules that might achieve high levels of immunity to SARS-CoV-2 infection against all current variants for at least a 6 month interval (figure 4) 49 . Together this data and modelling suggest that vaccination of previously infected individuals, even with existing vaccines targeting ancestral virus, will provide robust 245 protection against the current VOC, considerably prolonging the duration of efficacy of existing vaccines against these variants.

All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint

The deployment of highly effective vaccines against SARS-CoV-2 is contributing to major reductions in disease in many countries. However, the spread of novel viral variants with increased transmissibility and potential for reduced susceptibility to neutralisation by vaccine-induced serological responses has raised concerns about the durability of vaccine protection. Thus, understanding how the emergence of SARS-CoV-2 variants affects the 255 ability of vaccines to neutralise and protect against infection is an important priority. Here we integrate data from in-vitro neutralisation assays and efficacy studies incorporating a number of vaccines in widespread use. A major question is to understand whether immunity from existing vaccines shows a similar level of cross-reactivity to SARS-CoV-2 variants, or whether some vaccine platforms elicit more broadly cross-reactive neutralising antibody 260 responses. After accounting for significant variation between the different assays used, we find that vaccine-elicited antibody responses display similar levels of cross reactivity between vaccine platforms, and show little difference to that seen in convalescent subjects. This is Table 2 ). We previously derived a model to predict vaccine efficacy from in vitro neutralisation titre against (ancestral) SARS-CoV-2 14 . In the present work we test the utility of this model in predicting vaccine efficacy 275 against SARS-CoV-2 variants. We find that in vitro neutralisation remains strongly correlated with protection against SARS-CoV-2 variants and the observed protection is consistent with the predictions of the model (Fig 2) . A feature of this model that is likely to be useful for future vaccine development is that it provides not just a predicted efficacy, but also a lower confidence bound on the predicted efficacy against each variant for vaccines of a 280

given potency (measured as neutralisation titre to ancestral virus). It should be noted that all reported vaccine efficacies were above this lower bound ( Figure 2 ). However, while the model predictions are consistent with observed vaccine efficacy against symptomatic SARS-All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint CoV2 infection (Figure 2a) , we have relatively little data on vaccine efficacy against severe SARS-CoV-2 infection and mortality (Figure 2b) . Thus, while vaccine protection against 285 severe infection is an important question explored in our study, it is also the result for which we have the lowest confidence based on existing data. Interestingly, the results also hint that the design of study may play some role in the observed efficacy level. For example, the four randomised control trials are placed relatively symmetrically around the predictive line (one above and three below the mean prediction line) (Figure 2a) . However, all 10 observational 290 test-negative case control (TNCC) studies are placed above the line. It is known that testnegative study designs have a number of potential confounders 55,56 , and our analyses suggest future studies should investigate whether TNCC trials have a tendency to over-estimate vaccine efficacy.

The observed correlation between neutralisation and protective efficacy does not prove that neutralisation is the sole mechanism of protection, particularly since many other immune responses (such as cellular responses) are often highly correlated with neutralisation 57 . In addition, although the observed efficacies fall within the 95% predictive interval of the model, it is notable that neutralisation seems better correlated with protection at high 300 neutralisation levels than at lower levels ( Figure 2 ). The uncertainty in the model at low neutralisation is somewhat matched by uncertainty in the efficacy estimates themselves, arising from the generally smaller size of the RCT of vaccines against VOC (see confidence intervals in Supplementary Figure S4 ). In addition, it is also possible that neutralising responses are a good predictor of efficacy at high neutralisation levels, but other mechanisms 305 such as cellular immunity may contribute a more substantial role at lower neutralisation levels 58, 59 . This suggests that future studies investigating alternative immune correlates of protection should be focused on outcomes in populations immunized with more modestly protective vaccines that induce lower neutralising antibody levels.

The combined effects of waning immunity and reduced recognition of the SARS-CoV-2 VOC suggest that vaccine boosters may be needed to maintain high levels of protection from symptomatic SARS-CoV-2 infection. This raises a number of important questions about the potential benefits of boosting and whether variant-specific immunogens will be required.

Our analysis suggests maximising neutralising antibody responses to the ancestral virus, 315 through booster vaccination of previously infected individuals (with ancestral immunogens), should be an effective strategy to broadly increase neutralisation titres against SARS-CoV-2 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint variants. However, the limited studies of boosting in vaccinated subjects raise a number of important questions. Firstly, the optimal timing of boosting is unclear. Most boosting studies in convalescent or previously vaccinated individuals occurred around 6 months after infection 320 or vaccination. One study comparing early versus late boosting of vaccinees would appear to suggest a benefit in delaying to six months 51 . This is consistent with an observed rise in memory B cells within the first months after infection 53 . Beyond the first months, van Gils and colleagues found a similar increase in binding antibody levels in subjects infected 1 to 15 months before vaccination 60 . It is also presently unclear whether all vaccines will boost 325 immunity to a similar extent, or whether homologous or heterologous boosting might be perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in 

The authors declare no competing interests. 385

All authors contributed to the data collection, design of the study, writing of the manuscript and revision of the manuscript. DSK, DC, AR, TES and MPD contributed to the modelling and statistical analysis of the data. 390

All data and code will be made freely available on GitHub upon publication.

This work would not be possible without the many scientists who generously provided the published data analysed in this study, either through making the data directly available through the original publication or through providing it upon request. The authors thank these scientists for their contribution and the individual sources of data are indicated in the references and supplementary tables. 400 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint 

(1)

(2) All rights reserved. No reuse allowed without permission.

perpetuity.

preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in

The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint available studies (number of studies indicated).

All rights reserved. No reuse allowed without permission.

perpetuity.

preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in

The copyright holder for this this version posted August 13, 2021. All rights reserved. No reuse allowed without permission.

perpetuity.

preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in

The copyright holder for this this version posted August 13, 2021. Supplementary Table S4) , and assumes decay after boosting is the same as after initial infection or primary vaccination. 480 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint Figure S1 : In vitro neutralisation of SARS-CoV-2 variants. (A) Change in neutralisation titre of all vaccinees normalised against the change in neutralisation titre seen in convalescent individuals in the same study. For convalescent subjects, the mean for each study is one (since titres are normalised to convalescent). For 635 different vaccination groups, the difference between the drop in titre in convalescent individuals in the same study is shown. Horizontal bars indicate the weighted mean for that vaccine / variant combination. Vaccination groups show changes in neutralisation titre that closely match that of convalescent subjects in the same study. (B) For each laboratory the mean change in neutralisation titre observed in convalescent subjects and different vaccine 640 groups is shown. Although estimates of change in neutralisation vary between laboratories, within a given laboratory the change in neutralisation titre is congruent between convalescent and different vaccine groups. (C) Normalisation against BNT162b2 vaccinee sera; Panel A normalises vaccine responses against convalescent sera (which are not consistently defined across different studies). To check that our conclusions are robust to the reference serum 645 used, we also analysed the subset of studies in which sera from individuals vaccinated with BNT162b2 was available and normalised against the change in neutralisation titre seen in BNT162b2-vaccinated subjects. As can be seen, the dominant effect of laboratory is still evident when normalised against BNT162b2 sera. All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint Figure S2 : Neutralisation of ancestral virus predicts neutralisation of variants. The mean neutralisation titre against the ancestral virus (x-axis) and the mean neutralisation titre against the VOC (y-axis) is shown for individual studies. The predicted line for a 1:1 relationship is indicated (dashed blue line). The observed mean drop in neutralisation titre 660 across all vaccines and convalescent subjects is indicated by an arrow, and the predicted levels of variant neutralisation are indicated by a dashed red line (shading indicates 95% CI) are shown. The results for individual studies are variable because of differences between assays and Figure 2 reports the mean across all studies for each vaccine / variant combination. 665 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint Figure S3 : Mean drop in neutralisation titre against SARS-CoV-2 variants. The mean fold-drop in neutralisation titre reported for different SARS-CoV-2 variants is shown (with 95% CI). The number of subjects and studies contributing to this is also indicated. 675 

All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted August 13, 2021. ; https://doi.org/10.1101/2021.08.11.21261876 doi: medRxiv preprint Figure S4 : Correlation between in vitro neutralisation and observed protection. The relationship between neutralisation and protection derived from data on ancestral virus is shown (mean as solid line, shading is 95% CI). The predicted in vitro neutralisation titres against variants (based on the titres reported against ancestral virus in phase I/II studies, and 685 adjusted for the mean drop in neutralisation titre to variants reported in supplementary Figure  S3 ) are shown for each vaccine, along with the observed efficacy against VOC (see supplementary table 2). Note that whereas in Figure 2 the model curve is adjusted by the mean drop in neutralisation to VOC, here the mean neutralisation titres for each vaccine / variant combination are adjusted for this drop. 690 

Accelerated COVID-19 vaccine development: milestones, lessons, and prospects

Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine

Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine

Antibody Status and Incidence of SARS-CoV-2 Infection in Health Care Workers

Antibody resistance of SARS-CoV-2 variants B.1.351 and B.1.1.7. 490

Reduced sensitivity of SARS-CoV-2 variant Delta to antibody neutralization

Reduced neutralization of SARS-CoV-2 B.1.617 by vaccine and convalescent serum

Effects of SARS-CoV-2 variants on vaccine efficacy and response strategies

Evidence for increased breakthrough rates of SARS-CoV-2 variants of concern in BNT162b2-mRNA-vaccinated individuals

Efficacy of the ChAdOx1 nCoV-19 Covid-19 Vaccine against the 500 B.1.351 Variant

SARS-CoV-2 Delta VOC in Scotland: demographics, risk of hospital admission, and vaccine effectiveness

Vaccine Breakthrough Infections with SARS-CoV-2 505 Variants

Six Month Safety and Efficacy of the BNT162b2 mRNA COVID-19 Vaccine. medRxiv

Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection

Safety, tolerability, and immunogenicity of an inactivated SARS-CoV-2 vaccine in healthy adults aged 18-59 years: a randomised, double-blind, placebo-controlled, phase 1/2 clinical trial

An mRNA Vaccine against SARS-CoV-2 -Preliminary Report. 515

Phase 1-2 Trial of a SARS-CoV-2 Recombinant Spike Protein Nanoparticle Vaccine

Measuring immunity to SARS-CoV-2 infection: comparing assays and animal models

Efficacy of the NVX-CoV2373 Covid-19 Vaccine Against the B.1.1.7 Variant. medRxiv

Evidence for antibody as a protective correlate for COVID-19 vaccines

Covid-19 Breakthrough Infections in Vaccinated Health Care 525 Workers

Correlates of Protection against symptomatic and asymptomatic SARS-CoV-2 infection. medRxiv

Evidence of escape of SARS-CoV-2 variant B.1.351 from natural and vaccine induced sera

Sensitivity of infectious SARS-CoV-2 B.1.1.7 and B.1.351 variants to neutralizing antibodies

Resistance of SARS-CoV-2 variants to neutralization by monoclonal and serum-derived polyclonal antibodies

Antibody evasion by the P.1 strain of SARS-CoV-2

Reduced antibody cross-reactivity following infection with B.1.1.7 than with parental SARS-CoV-2 strains. bioRxiv

Reduced neutralization of SARS-CoV-2 B.1.1.7 variant by convalescent and vaccine sera

Increased resistance of SARS-CoV-2 variant P.1 to antibody neutralization

SARS-CoV-2 variant B.1.1.7 is susceptible to neutralizing antibodies elicited by ancestral spike vaccines

Neutralization of SARS-CoV-2 Variants B.1.429 and B.1.351

Immune response to SARS-CoV-2 variants of concern in vaccinated individuals

Reduced sensitivity of infectious SARS-CoV-2 variant B.1.617.2 to monoclonal antibodies and sera from convalescent and vaccinated individuals. 550 bioRxiv

Infection and Vaccine-Induced Neutralizing-Antibody Responses to the SARS-CoV-2 B.1.617 Variants

Neutralising antibody activity against SARS-CoV-2 VOCs B.1.617.2 and B.1.351 by BNT162b2 vaccination

AZD1222-induced neutralising antibody activity against SARS-CoV-2 Delta VOC

Measuring immunity to SARS-CoV-2 infection: comparing assays and animal models

Covid-19 Vaccine against the B.1.1.7 and B.1.351 Variants

Effectiveness of Covid-19 Vaccines against the B.1.617.2 (Delta) Variant

Efficacy of NVX-CoV2373 Covid-19 Vaccine against the B.1.351 565 Variant

Safety and Efficacy of Single-Dose Ad26.COV2.S Vaccine against Covid-19

Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an interim analysis of four randomised

SARS-CoV-2 Antibody Response in Persons with Past Natural Infection

Previously infected vaccinees broadly neutralize SARS-CoV-2 variants. medRxiv

mRNA vaccination boosts cross-variant neutralizing antibodies elicited by SARS-CoV-2 infection

Distinct antibody and memory B cell responses in SARS-CoV-2 naive and recovered individuals following mRNA vaccination

Impact of circulating SARS-CoV-2 variants on mRNA vaccine-580 induced immunity in uninfected and previously infected individuals. medRxiv

Serum Neutralizing Activity of mRNA-1273 against SARS-CoV-2 Variants. bioRxiv

Protection afforded by the BNT162b2 and mRNA-1273

COVID-19 vaccines in fully vaccinated cohorts with and without prior infection. medRxiv, 2021

Preliminary Analysis of Safety and Immunogenicity of a SARS-CoV-2 Variant Vaccine Booster. medRxiv

Immunogenicity and safety of a third dose, and immune persistence of 590

CoronaVac vaccine in healthy adults aged 18-59 years: interim results from a doubleblind, randomized, placebo-controlled phase 2 clinical trial. medRxiv

Immunological memory to SARS-CoV-2 assessed for up to 8 months after infection

Evolution of immune responses to SARS-CoV-2 in mildmoderate COVID-19

Longitudinal Analysis Reveals Distinct Antibody and Memory B Cell Responses in SARS-CoV2 Naive and Recovered Individuals Following mRNA Vaccination

Measurement of Vaccine Direct Effects Under the Test-Negative Design

Vaccine Effectiveness Studies in the Field

Humoral and circulating follicular helper T cell responses in recovered patients with COVID-19

Correlates of protection against SARS-CoV-2 in rhesus macaques

Prospects for durable immune control of SARS-CoV-2 and 610 prevention of reinfection

Single-dose SARS-CoV-2 vaccine in a prospective cohort of COVID-19 patients. medRxiv

A booster dose is immunogenic and will be needed for older adults who have completed two doses vaccination with CoronaVac: a randomised, double-blind, 615 placebo-controlled, phase 1/2 clinical trial. medRxiv

Reduced Risk of Reinfection with SARS-CoV-2 After COVID-19 Vaccination -Kentucky

Escape of SARS-CoV-2 501Y.V2 from neutralization by convalescent plasma

COVID boosters for wealthy nations spark outrage

Beta (B.1.351) Delta (B.1.617.2)