key: cord-1006762-mbiny5cs
authors: Khoury, D. S.; Cromer, D.; Reynaldi, A.; Schlub, T. E.; Wheatley, A. K.; Juno, J. A.; Subbarao, K.; Kent, S. J.; Triccas, J. A.; Davenport, M. P.
title: What level of neutralising antibody protects from COVID-19? .
date: 2021-03-11
journal: nan
DOI: 10.1101/2021.03.09.21252641
sha: 35f862ea1339cdcba1f4951509659ca096d74f86
doc_id: 1006762
cord_uid: mbiny5cs

Both previous infection and vaccination have been shown to provide potent protection from COVID-19. However, there are concerns that waning immunity and viral variation may lead to a loss of protection over time. Predictive models of immune protection are urgently needed to identify immune correlates of protection to assist in the future deployment of vaccines. To address this, we modelled the relationship between in vitro neutralisation levels and observed protection from SARS-CoV-2 infection using data from seven current vaccines as well as convalescent cohorts. Here we show that neutralisation level is highly predictive of immune protection. The 50% protective neutralisation level was estimated to be approximately 20% of the average convalescent level (95% CI = 14-28%). The estimated neutralisation level required for 50% protection from severe infection was significantly lower (3% of the mean convalescent level (CI = 0.7-13%, p = 0.0004). Given the relationship between in vitro neutralization titer and protection, we then used this to investigate how waning immunity and antigenic variation might affect vaccine efficacy. We found that the decay of neutralising titre in vaccinated subjects over the first 3-4 months after vaccination was at least as rapid as the decay observed in convalescent subjects. Modelling the decay of neutralisation titre over the first 250 days after immunisation predicts a significant loss in protection from SARS-CoV-2 infection will occur, although protection from severe disease should be largely retained. Neutralisation titres against some SARS-CoV-2 variants of concern are reduced compared to the vaccine strain and our model predicts the relationship between neutralisation and efficacy against viral variants. Our analyses provide an evidence-based prediction of SARS-CoV-2 immune protection that will assist in developing vaccine strategies to control the future trajectory of the pandemic.

3

SARS-CoV-2 has spread globally over the last year, infecting an immunologically naïve population and causing significant morbidity and mortality. Immunity to SARS-CoV-2 induced either through natural infection or vaccination has been shown to afford a degree of protection and/or reduce the risk of clinically significant outcomes. For example, seropositive 60 recovered subjects have been estimated to have 89% protection from reinfection 1 , and vaccine efficacies from 50 to 95% have been reported 2 . However, the duration of protective immunity is unclear, with primary immune responses inevitably waning [3] [4] [5] , and ongoing transmission of increasingly concerning viral variants that escape immune control 6 .

A critical question at present is to identify the immune correlate(s) of protection from SARS-CoV-2 infection and therefore predict how changes in immunity will be reflected in clinical outcomes. A defined correlate of protection will permit both confidence in opening up economies and facilitate rapid improvements in vaccines and immunotherapies. In influenza infection, for example, a hemagglutination inhibition (HAI) titre of 1:40 is thought to provide 70 50% protection from influenza infection 7 (although estimates range from 1:17 -1:110 8, 9 ) . This level was established over many years using data from a standardised HAI assay 10 applied to serological samples from human challenge and cohort studies. At present, however, there are few standardised assays for assessing SARS-CoV-2 immunity, little data comparing immune levels in susceptible versus resistant individuals, and no human challenge 75 model 11 .

The data currently available for SARS-CoV-2 infection includes immunogenicity data from phase 1 / 2 studies of vaccines and data on protection from preliminary reports from phase 3 studies and in seropositive convalescent individuals (see Supplementary Tables 1 and 2) . 80

While antiviral T and B cell memory certainly contribute some degree of protection, strong evidence of a protective role for neutralising serum antibodies exists. For example, passive transfer of neutralising antibodies can prevent severe SARS-CoV-2 infection in multiple animal models 12, 13 and Regeneron has recently reported similar data in humans 14 . We therefore focus our studies on in vitro virus neutralisation titres reported in studies of 85 vaccinated and convalescent cohorts. Unfortunately, the phase 1 / 2 studies all use different assays for measuring neutralisation. Normalisation of responses against a convalescent serum standard has been suggested to provide greater comparability between the results from 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 phase 3 studies the timeframes of study and case-definitions of   infection also vary amongst studies (see Supplementary Table 2) . Recognising these limitations, we aimed to investigate the relationship between vaccine immunogenicity and protection.

To compare neutralisation titres across studies we determined the mean and standard deviation (on log-scale) of the neutralisation titre in each study. These were normalised to the mean convalescent titre using the same assay in the same study (noting that the definition of convalescence was also not standardised across studies and a variable number of convalescent samples are studied). We then compared this normalised neutralisation level in 100 each study against the corresponding protective efficacy reported from the phase 3 clinical trials. Despite the known inconsistencies between studies, comparison of normalised neutralisation levels and vaccine efficacy demonstrates a remarkably strong non-linear relationship between average neutralization level and reported protection across different vaccines (Spearman r=0.905; p=0.0046, Figure 1A ). 105

To further dissect the relationship between immunogenicity and protection in SARS-CoV-2 we considered the parallels with previous approaches to estimating a '50% protective titre' in influenza infection. These historic studies in influenza involved comparison of HAI titres in infected versus uninfected subjects (in either natural infection or human challenge studies), 110 and used logistic or receiver-operating characteristics (ROC) approaches to identify an HAI titre that provided protection [7] [8] [9] 16, 17 ) . We adapted these approaches to analyse the existing data on 'average neutralisation level' in different studies and the observed level of protection from infection (details of statistical methods are provided in the Supplementary material).

We first fitted a logistic model to estimate the '50% protective neutralisation level' (across all studies) that best predicted the protective effect observed in each study (consistent with the use of a logistic function to model protection in influenza serological studies 16, 17 ). We estimate from this model a 50% protective neutralisation level of 19.9% (95% CI = 14.1% -28.1%) of the mean convalescent level ( Figures 1A and 1B) , and that this model provided a 120 good explanation of the relationship between average neutralisation level and protection across the studies. Since the model is dependent on the mean and distribution of 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 5 neutralisation levels, we also estimated these using different approaches, leading to similar estimates ( Fig. S2 and Supplementary Methods).

To relax the assumption that neutralisation levels are normally distributed in the above model, we also estimated the protective level using a distribution-free approach applied to the raw data for individual neutralisation levels reported in the studies. This method finds a protective neutralisation threshold that maximises the chances of classifying individuals as protected or not protected based on their neutralisation level being above or below the 130 threshold and on the observed protective efficacy in Phase 3 trials (i.e.: it does not rely on neutralisation levels being normally distributed). We refer to this as the 'protective neutralisation classification model'. Using this classification approach the estimated protective threshold was 28.6% (CI= 19.2% -29.2%) of the average convalescent level. As expected, the estimated protective level using the classification method was slightly higher 135 than the 50% protective level estimated using the logistic method, as the classification method essentially estimates a level of 100% protection instead of 50% protection. This analysis suggests that the 50% neutralisation level for SARS-CoV-2 is approximately 20% of the average convalescent titre. This analysis shows that the relationship between the 140 average protective levels and the observed protective efficacy across different vaccines can be predicted based on consideration of the level and distribution of neutralisation titres. To test the potential utility of this in predicting the protective efficacy of an unknown vaccine, we repeated our analysis with a 'leave-one-out' approach. That is, we fitted all possible groups of seven vaccine or convalescent studies and used this to predict the efficacy of the 145 8th. Figure 1C shows the results of using the logistic model of protection to predict the efficacy of each vaccine from the results of the other seven. In addition, after fitting the model to the data for eight vaccine / convalescent studies, the phase 3 results of another vaccine were released in a press release on 3 March 2021. Using the observed neutralisation level (a mean of 79.2% of the convalescent titre in that study (See Supplementary Table 1 ) 150 the predicted efficacy of the new vaccine was 79.4%, (95% Predictive Interval: 76.0%-82.8%), which is in very close agreement with the reported efficacy of 80.6% 18 , and suggests good predictive value of the model ( Figure 1A ).

As discussed above, a major caveat of our estimate of the relative protective level of 155 antibodies in SARS-CoV-2 infection is that it includes aggregation of data collected from 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 6 diverse neutralisation assays and clinical trial designs. Clearly, a more standardised approach to assays and trials design would allow refinement of these analyses 11 , although this has not occurred so far. In addition, the association of neutralisation with protection across these studies does not prove that neutralising antibodies are mechanistic in mediating protection. It 160 is possible that neutralisation is correlated with other immune responses, leading to an apparent association (as has been suggested for the use of HAI titre in influenza 19-21 ).

Another important refinement of this approach would be to have standardised measures of other serological and cellular responses to infection to identify if any of these provide a better predictive value than neutralisation. However, despite these limitations it is tempting to 165 consider the implications of this protective titre for immunity to SARS-CoV-2 infection. Figure S3 ). Although this comparison relies on limited data, it suggests that decay of vaccine-induced neutralisation is similar to that observed after natural infection.

If the relationship between neutralisation level and protection that we observe crosssectionally between different vaccines is maintained over time, we can use our model to predict how the observed waning of neutralisation titres might affect vaccine effectiveness.

Important caveats to such an extrapolation are that (i) it assumes that neutralisation is a major mechanism of protection (or that the mechanism of protection remains correlated with 185 neutralisation), although B cell memory and T cell responses may be more durable [3] [4] [5] 23 (and indeed B cell responses have been shown to increase following infection 3 ), (ii) it applies the decay of neutralisation observed in convalescence to the vaccine data, and (iii) it assumes that the decay in titre is the same regardless of the initial starting titre (whereas others have suggested faster decay for higher initial levels 24 ). These limitations notwithstanding, we used 190 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 7 the estimated half-life of neutralisation titre of 90 days derived from a study of convalescent individuals 5 and modelled the decay of neutralisation and protection over the first 250 days after vaccination (Figure 2A ). Our model predicts that waning neutralisation titre will have non-linear effects on protection from SARS-CoV-2 infection, depending on initial vaccine efficacy. For example, a vaccine starting with an initial efficacy of 95% would be expected to 195 maintain 58% efficacy by 250 days. However, a response starting with an initial efficacy of 70% would be predicted to drop to 18% efficacy after 250 days. This analysis can also be used to estimate how long it would take a response of a given initial efficacy to drop to 50% (or 70%) efficacy, which may be useful in predicting the time until boosting is required to maintain a minimal level of efficacy ( Figure 2B ). Clearly data generated from standardised 200 assays are needed to track the long-term decay of post-vaccination immune responses and their relationship to clinical protection. However, this model provides a framework that can be adapted to predict outcomes as further immune and protection data becomes available.

Indeed, if a disconnect between the decay of neutralisation titre and protection is observed, this may be a direct pointer to the role of non-neutralising responses in protection. 205

In addition to the effect of declining neutralisation titre over time, reduced neutralisation titres and reduced vaccine efficacy to different viral variants have also been observed 6,25-28 .

For example, it has been reported that the neutralisation titre against the B.1.351 variant in vaccinated individuals is between 7.6-fold and 9-fold lower compared to the early Victoria 210 variant 29 . Our model predicts that a lower neutralisation titre against a variant of concern will have a larger effect on vaccines for which protective efficacy against the wild-type virus was lower ( Figure 2C ). For example, a five-fold lower neutralisation titre is predicted to reduce efficacy from 95% to 67% in a high efficacy vaccine, but from 70% to 25% for a vaccine with lower initial efficacy. 215

The analysis above investigates vaccine (and convalescent) protection against detectable SARS-CoV-2 infection (using the definitions provided in the different phase 3 and convalescent studies, see Supplementary Table 2 ). However, it is thought that the immune response may provide greater protection from severe infection than from mild infection. To 220 investigate this, we also analysed data on the observed level of protection from severe infection where this was available (Supplementary Table 3 , noting that the definition of severe infection was not consistent across studies). As there have been under 100 severe infections reported across all the phase 3 trials combined, the confidence intervals on the 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 8 level of protection from severe infection are broad. The neutralisation level for 50% 225 protection from severe infection was 3.1% of the average convalescent level (CI= 0.75% -12.7%), which was significantly lower than the 20% level required for protection from any infection (p = 0.0004, likelihood ratio test, Supplementary Table 4 ). An important caveat to this analysis is the implicit assumption that neutralisation titre itself confers protection from severe infection, while cellular responses may also be important 30-32 . 230

The estimated neutralisation level for protection from severe infection is approximately 6 times lower than the level required to protect from any infection. Thus, a higher level of protection against severe infection is expected for any given level of vaccine efficacy against mild (any) SARS-CoV-2 infection. Assuming that this relationship remains constant over 235

time, it appears likely that immunity to severe infection may be much more durable than overall immunity to any infection. Long-term studies of antibody responses to other vaccines suggest that these responses generally stabilise with half-lives >10 years 33, 34 . Therefore we perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in 36, 37 . Similarly, after influenza virus vaccination, protective efficacy is thought to decline by around 7% per month 38 . Our modelling and predictions are based on a number of assumptions on the mechanisms and rate of loss of immunity. Important priorities for the field are the development of standardised assays to measure neutralisation and other immune responses, as well as standardised clinical trial protocols. These data will allow further testing 265 and validation of other potential immune correlates of protection. However, our study develops a modelling framework for integrating available, if imperfect, data from vaccination and convalescent studies to provide a tool for predicting the uncertain future of SARS-CoV-2 immunity.

270 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 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 March 11, 2021. ;

The reported mean neutralisation level (x-axis) from phase 1 / 2 trials and protective efficacy from phase 3 trials (y-axis) for seven vaccines, as well as the protection observed in a seropositive convalescent cohort are shown (details of data sources in Supplementary Tables 1 and 2 ). Mean / point estimates are indicated by dots (95% 295 confidence intervals are indicated as whiskers). Red solid line indicates the best fit of the logistic model and shaded red indicates the 95% predictive interval of the model. The mean neutralisation level and protective efficacy of the Covaxin vaccine is indicated as a green circle (data from this study was only available after modelling was complete and did not contribute to fitting). 300 B) Schematic illustration of the logistic approach to identifying the protective neutralisation level. The data for each study includes the mean and distribution of the in vitro neutralisation titre against SARS-CoV-2 in vaccinated or convalescent subjects (as a proportion of the titre in convalescent subjects)(green/red bell curve), accompanied by a level of protective efficacy for the same regimen. The efficacy is 305 illustrated by the proportion of bell curve 'protected' (green) versus 'susceptible (red) for individual studies (shaded areas in between reflect the changing risk). The modelling fits the optimal 50% protective neutralisation level (blue dashed line, shaded areas indicate 95% confidence limits) that best estimates the correct levels of protection observed across the different studies. 310 C) Predictions of the 'leave one out' analysis. Modelling was repeated multiple times using all potential sets of seven vaccination / convalescent studies to predict the efficacy of the eighth study. Dots indicate point estimates and whiskers indicate 95% confidence / predictive intervals. 315

A) Predicting the effects of declining neutralisation titre. Assuming the observed 320 relationship between neutralisation level and protection is consistent over time, we estimate the decline in efficacy for vaccines starting with different levels of early protection. The model assumes a half-life of neutralisation titre of 90 days over the first 250 days (as observed in a convalescent cohort 5 ). Coloured lines indicate the predicted trajectory for groups starting with different initial efficacy. 325 B) Modelling the time for efficacy to drop to 70% (red line) or 50% (blue line) for scenarios with different initial efficacy (indicated on the y-axis). For example, for a group starting with an initial protective efficacy of 90% the model predicts that 70% efficacy will be reached after 131 days, and 50% efficacy after 221 days. C) Estimating the impact of viral antigenic variation on vaccine efficacy. In vitro studies 330 have shown that neutralisation titres against some SARS-CoV-2 variants are reduced compared to titres against wild-type virus. If the relationship between neutralisation and protection remains constant, we can predict the difference in protective efficacy against wild-type and variant viruses from the difference in neutralisation level. For cohorts with a given initial protective efficacy measured against wild-type (vaccine 335 strain) virus (x-axis), we model the impact of a two-fold (red), 5-fold (green) or 10fold (blue) reduction in neutralisation titre to a variant virus. The y-axis indicates the 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 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 March 11, 2021. ; https://doi.org/10.1101/2021.03.09.21252641 doi: medRxiv preprint 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 Supplementary Tables 1 and 2 ). Mean / point estimates are indicated by dots (95% confidence intervals are indicated as whiskers). Red solid line indicates the best fit of the logistic model and shaded red indicates the 95% predictive interval of the model. The mean neutralisation level and protective efficacy of the Covaxin vaccine is indicated as a green circle (data from this study was only available after modelling was complete and did not contribute to fitting). (B) Schematic illustration of the logistic approach to identifying the protective neutralisation level. The data for each study includes the mean and distribution of the in vitro neutralisation titre against SARS-CoV-2 in vaccinated or convalescent subjects (as a proportion of the titre in convalescent subjects)(green/red bell curve), accompanied by a level of protective efficacy for the same regimen. The efficacy is illustrated by the proportion of bell curve 'protected' (green) versus 'susceptible (red) for individual studies (shaded areas in between reflect the changing risk). The modelling fits the optimal 50% protective neutralisation level (blue dashed line, shaded areas indicate 95% confidence limits) that best estimates the correct levels of protection observed across the different studies. (C) Predictions of the 'leave one out' analysis. Modelling was repeated multiple times using all potential sets of seven vaccination / convalescent studies to predict the efficacy of the eighth study.

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 Assuming the observed relationship between neutralisation level and protection is consistent over time, we estimate the decline in efficacy for vaccines starting with different levels of early protection. The model assumes a half-life of neutralisation titre of 90 days over the first 250 days (as observed in a convalescent cohort5). Coloured lines indicate the predicted trajectory for groups starting with different initial efficacy. (B) Modelling the time for efficacy to drop to 70% (red line) or 50% (blue line) for scenarios with different initial efficacy (indicated on the y-axis). For example, for a group starting with an initial protective efficacy of 90% the model predicts that 70% efficacy will be reached after 131 days, and 50% efficacy after 221 days. (C) Estimating the impact of viral antigenic variation on vaccine efficacy. In vitro studies have shown that neutralisation titres against some SARS-CoV-2 variants are reduced compared to titres against wild-type virus. If the relationship between neutralisation and protection remains constant, we can predict the difference in protective efficacy against wild-type and variant viruses from the difference in neutralisation level. For cohorts with a given initial protective efficacy measured against wild-type (vaccine strain) virus (x-axis), we model the impact of a two-fold (red), 5-fold (green) or 10-fold (blue) reduction in neutralisation titre to a variant virus. The y-axis indicates the predicted new efficacy against the variant strain. Dashed line indicates equal protection against wild-type and variant strains. Details of the data and modelling are provided in the Supplementary Material. 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 projections beyond 250 days rely on an assumption of how the decay in SARS-CoV-2 neutralisation titre will slow over time. In addition, the modelling only projects how decay in neutralisation is predicted to affect protection. Other mechanisms of immune protection may play important roles in providing long-term protection that are not captured in this simulation.

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 March 11, 2021. ; https://doi.org/10.1101/2021.03.09.21252641 doi: medRxiv preprint

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

Looking beyond COVID-19 vaccine phase 3 375 trials

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

Evolution of antibody immunity to SARS-CoV-2

Immunological memory to SARS-CoV-2 assessed for up to 8 months 380 after infection

Increased Resistance of SARS-CoV-2 Variants B.1.351 and B.1.1.7 to Antibody Neutralization

The role of serum haemagglutination-inhibiting antibody in protection against challenge infection with 385 influenza A2 and B viruses

Relationship between haemagglutination-inhibiting antibody titres and clinical protection against influenza: development and application of a bayesian random-effects model

Hemagglutination Inhibition Antibody Titers as a Correlate of 390 Protection for Inactivated Influenza Vaccines in Children

International Laboratory Comparison of Influenza Microneutralization Assays for A(H1N1)pdm09, A(H3N2), and A(H5N1) Influenza Viruses by CONSISE

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

Isolation of potent SARS-CoV-2 neutralizing antibodies and protection from disease in a small animal model

Correlates of protection against SARS-CoV-2 in rhesus 400 macaques

Regeneron Reports Positive Interim Data with REGEN-COV™ Antibody Cocktail used as Passive Vaccine to Prevent COVID-19. (Regeneron Press release

Establishment of the WHO International Standard and Reference Panel for anti-SARS-CoV-2 antibody (WHO/BS/2020/2403)

A framework for assessing immunological correlates of protection in vaccine trials

Comparison of neutralizing and hemagglutination-inhibiting antibody responses for evaluating the seasonal influenza vaccine

Durability of Responses after SARS-CoV-2 mRNA-1273 425 Vaccination

Functional SARS-CoV-2-Specific Immune Memory Persists after Mild COVID-19

Stable neutralizing antibody levels six months after mild and severe COVID-19 episode

SARS-CoV-2 variants show resistance to neutralization by many monoclonal and serum-derived polyclonal antibodies

mRNA vaccine-elicited antibodies to SARS-CoV-2 and circulating variants

Serum Neutralizing Activity Elicited by mRNA-1273 Vaccine -435 Preliminary Report

Neutralizing Activity of BNT162b2-Elicited Serum -Preliminary Report

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

Adaptive immunity to SARS-CoV-2 and COVID-19

Antigen-Specific Adaptive Immunity to SARS-CoV-2 in Acute COVID-19 and Associations with Age and Disease Severity

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

Duration of humoral immunity to common viral and vaccine antigens

Heterogeneity and longevity of antibody memory to viruses and 450 vaccines

Update: FDA Issues Policies to Guide Medical Product Developers Addressing Virus Variants

The time course of the immune response to experimental coronavirus infection of man

Influenza A Reinfection in Sequential Human Challenge: 460 Implications for Protective Immunity and "Universal" Vaccine Development

Intraseason waning of influenza vaccine protection: Evidence from the US Influenza Vaccine Effectiveness Network

14