key: cord-0873928-eulg3yyn
authors: Grandjean, L.; Saso, A.; Torres Ortiz, A.; Lam, T.; Hatcher, J.; Thistlethwaite, R.; Harris, M.; Best, T.; Johnson, M.; Wagstaffe, H.; Ralph, E.; Mai, A.; Colijn, C.; Breuer, J.; Buckland, M.; Gilmour, K.; Goldblatt, D.; Team, The Co-Stars Study
title: Long-Term Persistence of Spike Antibody and Predictive Modeling of Antibody Dynamics Following Infection with SARS-CoV-2
date: 2020-11-23
journal: nan
DOI: 10.1101/2020.11.20.20235697
sha: 0d43b4397653925d8c8d8ad5972c231dfa8dd919
doc_id: 873928
cord_uid: eulg3yyn

Background: Antibodies to Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) have been shown to neutralize the virus in-vitro. Similarly, animal challenge models suggest that neutralizing antibodies isolated from SARS-CoV-2 infected individuals prevent against disease upon re-exposure to the virus. Understanding the nature and duration of the antibody response following SARS-CoV-2 infection is therefore critically important. Methods: Between April and October 2020 we undertook a prospective cohort study of 3555 healthcare workers in order to elucidate the duration and dynamics of antibody responses following infection with SARS-CoV-2. After a formal performance evaluation against 169 PCR confirmed cases and negative controls, the Meso-Scale Discovery assay was used to quantify in parallel, antibody titers to the SARS-CoV-2 nucleoprotein (N), spike (S) protein and the receptor-binding-domain (RBD) of the S-protein. All seropositive participants were followed up monthly for a maximum of 7 months; those participants that were symptomatic, with known dates of symptom-onset, seropositive by the MSD assay and who provided 2 or more monthly samples were included in the analysis. Survival analysis was used to determine the proportion of sero-reversion (switching from positive to negative) from the raw data. In order to predict long-term antibody dynamics, two hierarchical longitudinal Gamma models were implemented to provide predictions for the lower bound (continuous antibody decay to zero, 'Gamma-decay') and upper bound (decay-to-plateau due to long lived plasma cells, 'Gamma-plateau') long-term antibody titers. Results: A total of 1163 samples were provided from 349 of 3555 recruited participants who were symptomatic, seropositive by the MSD assay, and were followed up with 2 or more monthly samples. At 200 days post symptom onset, 99% of participants had detectable S-antibody whereas only 75% of participants had detectable N-antibody. Even under our most pessimistic assumption of persistent negative exponential decay, the S-antibody was predicted to remain detectable in 95% of participants until 465 days [95% CI 370-575] after symptom onset. Under the Gamma-plateau model, the entire posterior distribution of S-antibody titers at plateau remained above the threshold for detection indefinitely. Surrogate neutralization assays demonstrated a strong positive correlation between antibody titers to the S-protein and blocking of the ACE-2 receptor in-vitro [R2=0.72, p<0.001]. By contrast, the N-antibody waned rapidly with a half-life of 60 days [95% CI 52-68]. Discussion: This study has demonstrated persistence of the spike antibody in 99% of participants at 200 days following SARS-CoV-2 symptoms and rapid decay of the nucleoprotein antibody. Diagnostic tests or studies that rely on the N-antibody as a measure of seroprevalence must be interpreted with caution. Our lowest bound prediction for duration of the spike antibody was 465 days and our upper bound predicted spike antibody to remain indefinitely in line with the long-term seropositivity reported for SARS-CoV infection. The long-term persistence of the S-antibody, together with the strong positive correlation between the S-antibody and viral surrogate neutralization in-vitro, has important implications for the duration of functional immunity following SARS-CoV-2 infection.

Since appearing as a cluster of pneumonia cases in December 2019 in Wuhan, China, Coronavirus disease has rapidly spread worldwide 1 . As of October 26 th , there have been 43,187,134 cases, resulting in over 1.1 million deaths and a global health crisis, with significant social, economic and public health implications 2 . COVID-19 is caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), an enveloped RNA β-coronavirus 3 . Specific immunoglobulin (IgG) antibody responses to the SARS-CoV-2 trimeric spike (S) protein, nucleoprotein (N) protein and the receptor-binding domain (RBD) develop between 6-15 days following disease-onset 4 . The S-protein, which contains the RBD, binds to host cells via the angiotensin-converting-enzyme-2 (ACE-2) receptor, and membrane fusion occurs before viral entry 5, 6 . The N-protein plays an important role in transcription enhancement and viral assembly 7 .

SARS-CoV-2-specific antibodies, particularly to the S-and RBD-antigens, have been shown to correlate with T-cell responses and viral neutralization in vitro as well as to protect against disease in animals, following passive transfer of convalescent serum or selected monoclonal antibodies [8] [9] [10] [11] [12] . It is unclear, however, whether re-infection can occur in humans who mount a humoral response following primary SARS-CoV-2 infection and achieve viral clearance. Recent case reports have emerged describing new respiratory samples positive for SARS-CoV-2 RNA after confirmed negativity, although these are few compared to the worldwide scale of infection, with several potential explanations proposed 13, 14 . Furthermore, no recurrence of disease was reported in rhesus macaques or Syrian hamsters that were re-challenged in the presence of detectable endogenous antibodies (although protection against infection varied between studies). 9, 15, 16 . These findings highlight the importance of characterising humoral dynamics following SARS-CoV-2 infection. SARS-CoV IgG and neutralizing antibodies have been shown to commonly persist up to 2-3 years post-infection, particularly in hospitalized patients, 17, 18 with recent reports demonstrating seropositivity as late as 12-17 years after infection 19, 20 . Following severe disease caused by Middle Eastern Respiratory Syndrome (MERS), antibodies have been detected up to 34-months post-infection 21, 22 . Existing longitudinal studies of SARS-CoV-2 are limited by inadequate modeling of antibody dynamics, short duration, low sampling density and frequency of longer-term follow-up [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] . Fitting Locally Estimated Scatterplot Smoothing (LOESS) or equivalent lines of best fit to the data [23] [24] [25] 33 also fails to provide a mathematical framework for evaluating long-term antibody responses.

In order to evaluate antibody kinetics and longevity following SARS-CoV-2 infection, we undertook the prospective Covid-19 Staff Testing of Antibody Responses Study (Co-STARS). Seropositive and symptomatic participants were followed up monthly with repeated antibody titer quantification. Detailed demographic, clinical and socioeconomic data were collected and mathematical models developed to characterize longitudinal humoral kinetics from initial antibody boosting to subsequent decay. To predict long-term antibody dynamics, we fitted two different models based on the gamma distribution: one which assumed persistent antibody decay 34 , and an alternate that allowed for an eventual plateau, 35, 36 to account for sustained antibody production by long-lived plasma cells (LLPCs).

After providing informed consent, a total of 3555 individuals -all healthcare workers at Great Ormond Street Hospital -were enrolled in the study. Of this group, 349 were both symptomatic, seropositive by the MSD assay and provided 2 or more monthly samples for the primary outcome analysis of antibody dynamics. These 

Multivariate analysis demonstrated that fever, rigors, ageusia, anosmia, a previous medical condition, high BMI and Black Asian Minority Ethnic (BAME) backgrounds were all associated with higher peak spike protein antibody titers (Table 1) . No variables were identified to be independently associated with the rate of antibody decay. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint

Serial monthly serological measurements from 349 participants who provided 2 or more samples following the onset of symptoms demonstrated a rapid rate of decay of the N-antibody relative to the S and RBD antibody ( Figure 1 ). The spike antibody assay detected a total of 342/349 (98%) participants who were seropositive to any one of the S, RBD or N-antibodies. In comparison the RBD and N-assays detected 332/349 (95%) and 333/349 (95%) respectively. The sensitivity of the RBD and N-assays further declined with time relative to the S-antibody assay. At 200 days following the onset of symptoms, only 75% of 349 participants tested positive for N-antibody whereas 99% remained positive for S-antibody ( Figure 2 ).

Monte-Carlo Markov traces converged well for both the gamma distribution and decay-to-plateau curves demonstrating a stable model fit to the data (Supplementary Figure 1) . The maximum R-hat for any parameter was 1.0035, while the minimum effective sample size (ESS) was 842.8 (Supplementary data, Table S1 ).

Comparison of goodness of fit between models showed that for all antigens the decay-to-plateau model provided a better fit to the data than the gamma-decay model, although this difference was not statistically 

Measured weekly average titer data for each of the S-, RBD-, and N-antibodies demonstrated that peak antibody response to infection was itself a plateau/slowly increasing line. Antibody titers rapidly increased during the first 3 weeks with prolonged high titers reached and maintained between week 4 and week 10 after the onset of symptoms. The peak antibody response for the S-antibody, RBD, and N-antibody from both raw weekly average serial titer and modeled data occurred at 40 [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] days respectively. This was supported by the both the gamma-decay and gamma-plateau models which provided a similar close fit to this early stage of the humoral response ( Figure 4 a, b, c).

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint

The modeled half-life under the gamma-decay model and the gamma-plateau model were also very similar and both models showed a rapid decay of the N-relative to the RBD-and S-antibody. The half-life for the N-, RBD-and S-antibody was 60 days [95% CI 52- Under the gamma-plateau model, the S-antibody was characterized by a slow decay, with an eventual stabilized plateau at 1825 days [95% CI 250-3700] and none of the posterior probability distribution of the titers at the eventual plateau crossed the threshold for a negative test, whereas 75% of the posterior probability distribution for the N-antibody crossed the threshold for a negative test by 610 days.

There was a sigmoidal relationship between raw antibody titers and percentage binding/ACE-2 receptor 

This prospective cohort study of antibody responses following SARS-CoV-2 infection has demonstrated that 99% of 349 healthcare workers symptomatic with SARS-CoV-2 remained seropositive for the spike protein antibody 200 days after symptoms. Our study is the first to provide a mathematical modeling framework capable of predicting the long-term dynamics of the 3 key SARS-CoV-2 antibodies following natural infection.

Even under our most pessimistic assumptions of continuous exponential decay, 95% of individuals were predicted to remain seropositive to S-antibody at 465 days [95% CI 370-575 days] while our more optimistic upper bound gamma-decay model predicted a permanent long-lasting plateau of detectable S-antibody.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint These data contradict conclusions from studies that have reported rapid waning of antibodies after a few months 29, 31, 32 . Furthermore, our findings are in line with the duration of humoral responses observed following [17] [18] [19] [20] . Importantly, the long-lasting S-and RBD-antibodies also correlated well with a surrogate SARS-CoV-2 neutralization assay of ACE-2-receptor-blocking, strongly suggesting that long-term measurable S-antibody levels are functionally important. When the humoral correlates of protection against reinfection are known, our model of longitudinal S-antibody dynamics will therefore enable predictions to be made about the duration of long-lasting protective immunity following infection with SARS-CoV-2.

In contrast to the S-antibody, the N-antibody was observed to serorevert in 56/349 participants over the course of the study alone and had a modeled half-life of 60 days. This has important implications for diagnostic testing, epidemiological modeling and public health decision making that often rely on the N-antibody to estimate SARS-CoV-2 seroprevalence. This finding may also explain some unexpectedly low population level prevalence estimates in high burden countries 37 and confound the finding that children with multi-system inflammation have a higher S:N ratio compared to adults 30 . Antibodies peaked at 30-40 days, this is significantly longer than other reports that are likely to have missed the prolonged peak/plateau due to inadequate sampling density 38 . It is notable that the delay between the SARS-CoV-2 epidemic curve and the mini-epidemic of paediatric multi-system inflammatory syndrome coincides temporally with peak antibodies 39 .

The persistence of detectable S-and/or RBD-antibody compared to the rapid decay of the N-antibody has also been observed in convalescent sera obtained from SARS survivors, seventeen years after infection 19 , although the exact underlying mechanisms warrant further investigation. Differences in the epitope structure 40 , immunogenicity and presentation to B-cells may distinctly impact the production, maturation and longevity of the plasma cells that secrete these antibodies [41] [42] [43] [44] . Distinct T-helper cell interactions at the germinal centre may further determine B-cell and humoral dynamics, as previously observed in the context of the response to different HIV proteins 45 . Independently or in parallel, transcriptional programs and epigenetic imprinting may also selectively influence the kinetics and survival of both N-versus S-antibody-producing long-lived plasma cells 43, 44, 46 . Finally, cross-reactive memory, rather than naive B cells, may play a role in responses targeting nucleoprotein, given that it is more conserved across CoVs than RBD 47 IgG and neutralizing antibodies within the first 3 months following infection, particularly in mildly symptomatic cases. In July, our earlier publicly available pre-print of serial antibody and modeled data demonstrated longer lasting spike antibody and rapidly decaying N-antibody 48 . In comparison, others have reported that the S-antibody and/or RBD-antibody correlate with neutralizing responses and decay slowly, persisting during the study period, up to at 90-150 days post-infection [23] [24] [25] 49 . These studies, however, are limited by their shorter sampling time frame, lower sampling density and lack of appropriate modeling to predict antibody trajectory.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint Implementing LOESS lines of best fit to the data [23] [24] [25] 29 or comparing the variance of average antibody titers at different time intervals 31, 32, 49 does not permit evaluation of long-term antibody trajectory.

Our study is strengthened by the density, frequency, and duration of longitudinal sampling collection. The parallel evaluation of absolute antibody titers by the chemiluminescent MSD assay to three major SARS-CoV-2 proteins also enabled us to demonstrate the decay of the N-antibody relative to the S-and RBD-antibodies. Importantly, this is the first study to provide a mathematical framework for long-term SARS-CoV-2 antibody responses, modeling both the peak and decay following infection and enabling realistic best-case and worst-case predictions of future antibody titers. Our work provides a detailed, shareable and reproducible model, with parameters that are useful for epidemiological purposes. Additionally, some of our parameters are fitted at both the population and individual level which is informative when inspecting risk factors and variability in the population. A third possible trajectory may be that the humoral response stabilizes but then continues to decline ('plateau then decay'), albeit at a slower rate, as previously demonstrated in the context of vaccine-induced HPV responses and Hepatitis A infection 35, 36 . Further serological measurements of seropositive recruits will take place in 6 months' time to confirm which of these three models is superior in the longer term.

None of the seropositive healthcare workers identified in this study required hospitalization. This is important, given that the overwhelming majority of COVID-19 cases are not hospitalized. Our study population is, therefore, representative of most community SARS-CoV-2 infections 50 . Severe disease has however been associated with higher antibody titers and a longer duration of antibody response following both SARS and is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint

One potential limitation of this study is the fact that only 38% of participants had an available confirmatory positive PCR result. In order to mitigate this concern, prior to the study taking place, a formal performance evaluation of the MSD assay was undertaken among 169 confirmed PCR positive SARS-CoV-2 participants that demonstrated a 97.9% sensitivity and 97.4% specificity at 21 days post infection 57 . This makes the proportion of false positive serological tests likely to be small and therefore have little impact on our findings. Viral neutralization assays remain the gold-standard in vitro correlate of protection; as such, the lack of formal 'authentic' neutralization tests is another limitation of the study. However, ACE-2 receptor competition assays, such as the MSD competitive binding assay, have been shown to correlate well with formal viral neutralization assays, enabling their use as a suitable surrogate functional test 58 .

Whilst the severity of infection among our study participants is likely to be representative of community infection, our findings may be biased to healthcare workers. Recent studies have hypothesized that previous exposure to seasonal CoVs -to which healthcare workers may be disproportionately exposed -may confer some protection against SARS-CoV-2 12,19,53-56,58 and may need to be accounted for when modeling transmission or longevity dynamics 59 . Our estimates of the time-to-negativity are also dependent on the negative thresholds and lower limits of detection of the assay, respectively. However, our model fits, estimates of the rate of decay and the raw serial antibody titer trajectory (Figure 1 ) are not dependent on the threshold for a negative test.

In summary, this prospective cohort study has shown that the SARS-CoV-2 S-antibody, which correlated well with functional receptor blocking in-vitro , remained detectable in 99% of individuals up to 200 days post infection. In comparison, the N-antibody waned rapidly with a half-life of 60 days and 54/349 participants seroreverting over the course of the study. This study therefore has immediate consequences for diagnostic testing and public health decision making that often depend on the N-antibody as a reliable measure of past infection. Our most pessimistic continuous decay model, predicted that 95% of individuals would continue to have detectable spike antibody at 465 days while our gamma-plateau predicted that spike antibody would plateau at detectable levels indefinitely. The long-term presence of functional S (and RBD-antibody) has important implications for the duration of protective immunity following natural infection. It remains to be seen whether the SARS-CoV-2 vaccine candidates will replicate the long-lasting spike antibody duration observed and modeled here following natural infection.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint

Co-STARS is a 1 year single-centre, two-arm, prospective longitudinal cohort study of healthcare workers at a central London paediatric hospital Great Ormond Street Hospital for Children (GOSH). The study was approved to start by the United Kingdom NHS Health Research Authority on 29th April 2020 and registered on ClinicalTrials.gov (NCT04380896). Informed consent was obtained from all participants. The Study Protocol and Supplementary Materials submitted with this paper include detailed methods, power calculations and the data analysis approach.

All hospital staff members ≥18 years of age were eligible for the study, provided they did not display symptoms consistent with SARS-CoV-2 infection at recruitment. Those significantly immunosuppressed or those who had previously received blood products (including immunoglobulins or convalescent sera) since September 2019 were excluded from the study.

After providing informed consent, participants undertook a detailed, standardised online questionnaire at study entry. This included socio-demographic factors, details of previous exposure to and symptomatic episodes consistent with COVID-19, any subsequent complications, previous SARS-CoV-2 diagnostic test results, past medical and contact history, and a comprehensive assessment of risk factors for exposure, susceptibility to infection and severe disease. Blood samples were also taken at baseline and each follow-up visit for determination of SARS-CoV-2 serology.

Serum antibodies titers were measured by the Meso Scale Discovery (MSD) Chemiluminescent binding assay that simultaneously detects and quantifies anti-SARS-CoV-2 IgG specific for trimeric S-protein, RBD and N-protein. Assay qualification and performance were evaluated as described in our accompanying methods is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint

An MSD® 96-well Custom Competition Assay designed to measure the inhibition of ACE-2 receptor binding to S or RBD by serum-derived antibody (MSD, Maryland) was run on 94 serial samples from 46 participants (two participants had 3 serial samples) in order to establish in vitro correlates of functional immunity.

All seropositive participants were followed up monthly (ongoing) for repeat antibody testing. Seronegative participants will be followed up 6-monthly. At each follow-up visit, participants completed a shortened version of the baseline questionnaire, focussing on any recurrent COVID-19 exposure and/or symptoms.

The primary outcome of the study was to establish humoral dynamics following SARS-CoV-2 infection. Nested sub-studies studies will explore the secondary outcome measures including the incidence of SARS-CoV-2 re-infection, the dynamics of the cellular response, IgA dynamics and the clinical and demographic factors that are associated with SARS-CoV-2 infection.

Power calculations were based on a negative exponential model of antibody decay from the peak using the pwr.f2.test function in R. We assumed a study power of 80% and explored a variety of hypothesized effect sizes (decreases in antibody titers over 1 year) and co-variates on the required study size with an alpha of 0.05 (Supplementary Materials, Study Protocol).

To calculate the proportion of individuals that serorevert over the course of the study, we performed a survival analysis to account for censoring using the survival package in R. An "event" was defined as a persistent negative test after the first positive test, while positive tests were counted as "censored" events.

The dynamics of antibody response following infection with SARS-CoV-2 were estimated by fitting two mixed effects models based on a gamma curve on every sample that had more than three antibody observations. Two gamma models were chosen ("gamma-plateau" and "gamma-decay") to enable modeling of an optimistic upper bound estimate (eventual stabilization of decay to a plateau) to a pessimistic lower bound estimate is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint secrete antibody with a short half-life, followed by a subsequent robust long-lived plasma cell (LLPC) response that maintains circulating, high-specificity, high-avidity antibody long-term 61 .

Gamma-decay and gamma-plateau models were fitted to the entire antibody response curve from day 0 (symptom onset) to the peak and then subsequent decay of each SARS-CoV-2 antibody. The confidence limits around the curves were derived by repeated sampling from the posterior distribution of the different model parameters, including individual effects.

The gamma-decay model assumed a uninterrupted continuous decay, given by the formula:

Eq1:

where f(t) is the log antibody titer at time t after symptom onset. The gamma function is described in terms of the shape (a) and scale (b), parameterized to reduce confounding of the parameters. β 1 is the initial titer value at baseline; β2 determines the level of antibody rise; and u is an individual effect.

To account for the contribution of long lived plasma cells, a third term was added to the model, allowing for a long-term plateau expressed as: Eq2:

Where β 3 represents the long-term plateau, and k is the rate at which the latter term rises.

The relationship between ACE-2 receptor blocking and antibody titers was modeled with a 4 parameter generalized logistic curve, where the percentage binding at titer level t is given by:

Eq3:

with parameters a,b, and c that represent, respectively, the upper receptor blocking asymptote, the growth rate, and the titer at which maximum growth occurs. The parameter d is an asymmetry factor that affects the point of inflection on the y axis.

14 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org /10.1101 /10. /2020 All the models were fitted using RSTAN in R (R: A language and environment for statistical computing.

Foundation for Statistical Computing, Vienna, Austria). For each model, we ran 4 independent chains for 15,000 iterations for the gamma distributions, and 10,000 iterations for the sigmoid model. Model comparison was performed using Pareto-smoothed importance sampling leave-one-out cross-validation (PSIS-LOO) as implemented in the loo R package .

This study was approved by the UK Health Research Authority (www.hra.nhs.uk) and registered with www.clinical-trials.gov (NCT04380896). Written informed consent was obtained from all participants before recruitment to the study.

We would like to dedicate this article to the staff members who died of COVID-19 at Great Ormond Street Hospital during the first wave of the pandemic. We would also like to thank all the staff at Great Ormond Street 

The authors have declared that no competing interests exist.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint 

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 23, 2020. 

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

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

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.20.20235697 doi: medRxiv preprint

Clinical Characteristics of Coronavirus Disease 2019 in China

An interactive web-based dashboard to track COVID-19 in real time. The Lancet Infectious Diseases

A Novel Coronavirus from Patients with Pneumonia in China

Immune responses and immunity to SARS-CoV-2

European Centre for Disease Prevention and Control

Characterization of spike glycoprotein of SARS-CoV-2 on virus entry and its immune cross-reactivity with SARS-CoV | Nature Communications

Nucleocapsid Protein Recruitment to Replication-Transcription Complexes Plays a Crucial Role in Coronaviral Life Cycle

Human neutralizing antibodies elicited by SARS-CoV-2 infection

Syrian hamsters as a small animal model for SARS-CoV-2 infection and countermeasure development

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

Rapid Generation of Neutralizing Antibody Responses in COVID-19 Patients

Case series and systematic review of literature

Genomic evidence for reinfection with SARS-CoV-2: a case study. The Lancet Infectious Diseases

Primary exposure to SARS-CoV-2 protects against reinfection in rhesus macaques

SARS-CoV-2 infection protects against rechallenge in rhesus macaques

Disappearance of antibodies to SARS-associated coronavirus after recovery

Longitudinal profile of antibodies against SARS-coronavirus in SARS patients and their clinical significance

Serological differentiation between COVID-19 and SARS infections

Long-Term Persistence of IgG Antibodies in SARS-CoV Infected Healthcare Workers. medRxiv [Internet] 2020

Persistence of Antibodies against Middle East Respiratory Syndrome Coronavirus. Emerg Infect Dis

Antibody Response and Disease Severity in Healthcare Worker MERS Survivors

Persistence of serum and saliva antibody responses to SARS-CoV-2 spike antigens in COVID-19 patients

Persistence and decay of human antibody responses to the receptor binding domain of SARS-CoV-2 spike protein in COVID-19 patients

Orthogonal SARS-CoV-2 Serological Assays Enable Surveillance of Low-Prevalence Communities and Reveal Durable Humoral Immunity

S1074761320304453

Rapid Decay of Anti-SARS-CoV-2 Antibodies in Persons with Mild Covid-19

28. Loss of Anti-SARS-CoV-2 Antibodies in Mild Covid-19

Longitudinal evaluation and decline of antibody responses in SARS-CoV-2 infection. medRxiv [Internet] 2020

SARS-CoV-2 infection induces robust, neutralizing antibody responses that are stable for at least three months. medRxiv [Internet] 2020

Dynamics of neutralizing antibody titers in the months after SARS-CoV-2 infection

Durability of neutralizing antibodies and T-cell response post SARS-CoV-2 infection

SARS-CoV-2 infection induces sustained humoral immune responses in convalescent patients following symptomatic COVID-19. medRxiv [Internet] 2020

Individual and Population Trajectories of Influenza Antibody Titers Over Multiple Seasons in a Tropical Country

Modeling the long-term antibody response of a human papillomavirus (HPV) virus-like particle (VLP) type 16 prophylactic vaccine. Vaccine [Internet]

Living on Three Time Scales: The Dynamics of Plasma Cell and Antibody Populations Illustrated for Hepatitis A Virus

Prevalence of SARS-CoV-2 in Spain (ENE-COVID): a nationwide, population-based seroepidemiological study. The Lancet

S0140673620314835

Quantifying antibody kinetics and RNA detection during early-phase SARS-CoV-2 infection by time since symptom onset. eLife

Multisystem Inflammatory Syndrome in Children in the United States

Role of Multivalency and Antigenic Threshold in Generating Protective Antibody Responses. Front Immunol

Antibodies and B Cell Memory in Viral Immunity

Mechanisms that determine plasma cell lifespan and the duration of humoral immunity

The Maintenance of Memory Plasma Cells. Front Immunol

Factors Affecting Early Antibody Secreting Cell Maturation Into Long-Lived Plasma Cells. Front Immunol

HIV-1 Envelope Induces Memory B Cell Responses That Correlate with Plasma Antibody Levels after Envelope gp120 Protein Vaccination or HIV-1 Infection

Long-lived plasma cells: a mechanism for maintaining persistent antibody production

Structural Genomics of SARS-CoV-2 Indicates Evolutionary Conserved Functional Regions of Viral Proteins. Viruses [Internet] 2020

Humoral Response Dynamics Following Infection with SARS-CoV-2. medRxiv [Internet] 2020

Robust neutralizing antibodies to SARS-CoV-2 infection persist for months

Estimates of the severity of coronavirus disease 2019: a model-based analysis. The Lancet Infectious diseases 2020;0(0). -Google Search

2C+et+al.+Estimates+of+the+severity+of+coronavirus+disease+2019%3A+a+model-based+analysis.+T he+Lancet+Infectious+diseases+2020%3B0(0).&aqs=chrome..69i57

Neutralizing Antibody Response and SARS Severity

A systematic review of antibody mediated immunity to coronaviruses: kinetics, correlates of protection, and association with severity

SARS-CoV-2-specific T cell immunity in cases of COVID-19 and SARS, and uninfected controls

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

Broad and strong memory CD4 + and CD8 + T cells induced by SARS-CoV-2 in UK convalescent individuals following COVID-19

Targets of T Cell Responses to SARS-CoV-2 Coronavirus in Humans with COVID-19 Disease and Unexposed Individuals

Evaluation of a novel multiplexed assay for determining IgG levels and functional activity to SARS-CoV-2

A human monoclonal antibody blocking SARS-CoV-2 infection

Projecting the transmission dynamics of SARS-CoV-2 through the postpandemic period

Validation of an extraction-free RT-PCR protocol for detection of SARS-CoV2 RNA. medRxiv [Internet] 2020

B Cell Responses: Cell Interaction Dynamics and Decisions

Supplementary