key: cord-315448-bosazmlm
authors: Crawford, Katharine H D; Dingens, Adam S; Eguia, Rachel; Wolf, Caitlin R; Wilcox, Naomi; Logue, Jennifer K; Shuey, Kiel; Casto, Amanda M; Fiala, Brooke; Wrenn, Samuel; Pettie, Deleah; King, Neil P; Greninger, Alexander L; Chu, Helen Y; Bloom, Jesse D
title: Dynamics of neutralizing antibody titers in the months after SARS-CoV-2 infection
date: 2020-09-30
journal: J Infect Dis
DOI: 10.1093/infdis/jiaa618
sha: 
doc_id: 315448
cord_uid: bosazmlm

Most individuals infected with SARS-CoV-2 develop neutralizing antibodies that target the viral spike protein. Here we quantify how levels of these antibodies change in the months following SARS-CoV-2 infection by examining longitudinal samples collected between ~30 and 152 days post symptom onset from a prospective cohort of 32 recovered individuals with asymptomatic, mild, or moderate-severe disease. Neutralizing antibody titers declined an average of about four-fold from one to four months post symptom onset. This decline in neutralizing antibody titers was accompanied by a decline in total antibodies capable of binding the viral spike or its receptor-binding domain. Importantly, our data are consistent with the expected early immune response to viral infection, where an initial peak in antibody levels is followed by a decline to a lower plateau. Additional studies of long-lived B-cells and antibody titers over longer time frames are necessary to determine the durability of immunity to SARS-CoV-2.

onset, sera from most infected individuals can bind to the viral spike and neutralize infection in vitro [5, 7, 9] . The reciprocal dilution of sera required to inhibit viral infection by 50% (NT 50 ) is typically in the range of 100 to 200 at 3-4 weeks post symptom onset [10] , although neutralizing titers range from undetectable to >10,000 [2, 5, 9] .

There are currently limited data on the dynamics of neutralizing antibodies in the months following recovery from SARS-CoV-2. For most acute viral infections, neutralizing antibodies rapidly rise after infection due to a burst of short-lived antibody-secreting cells, and then decline from this peak before reaching a stable plateau that can be maintained for years to decades by long-lived plasma and memory B cells [11, 12] . These dynamics have been observed for many viruses, including influenza [13] , RSV [14] , MERS-CoV [15] , SARS-CoV-1 [16, 17] , and the seasonal human coronavirus 229E [18] .

Several recent studies have tracked antibody levels in individuals who have recovered from infection with SARS-CoV-2 for the first few months post symptom onset [5, 7, 8, [19] [20] [21] [22] . Most of these studies have reported that over the first three months, antibodies targeting spike decline several fold from a peak reached a few weeks post symptom onset [5, 7, 19] , suggesting that the early dynamics of the antibody response to SARS-CoV-2 are similar to those for other acute viral infections.

A c c e p t e d M a n u s c r i p t 5 Here we build on these studies by measuring both the neutralizing and binding antibody levels in serial plasma samples from 32 SARS-CoV-2-infected individuals across a range of disease severity with follow-up as long as 152 days post symptom onset. On average, neutralizing titers decreased ~4-fold from ~30 to >90 days post symptom onset. This decline in neutralizing titers was paralleled by a decrease in levels of antibodies that bind spike and its receptor-binding domain (RBD).

Nonetheless, most recovered individuals still had substantial neutralizing titers at three to four months post symptom onset.

Plasma samples were collected as part of a prospective longitudinal cohort study of individuals with SARS-CoV-2 infection. Individuals 18 years or older with laboratory confirmed SARS-CoV-2 infection were eligible for inclusion. Individuals who were HIV+ were excluded from this study due to concerns that antiretroviral treatment may affect our pseudotyped lentivirus neutralization assay. Individuals were recruited from three groups: inpatients, outpatients, and asymptomatics. Inpatients were hospitalized at Harborview Medical Center, University of Washington Medical Center, or Northwest Hospital in Seattle, WA and were enrolled while hospitalized. Outpatients were identified through a laboratory alert system, email and flyer advertising, and through identification of positive COVID-19 cases reported by the Seattle Flu Study [23] . Asymptomatic individuals in this study were recruited through outpatient testing and identified when they answered "None" on their symptom questionnaire. They were confirmed to be symptom-free for the first 30 days after diagnosis.

A c c e p t e d M a n u s c r i p t 6 We initially enrolled 34 individuals following RT-qPCR-confirmed SARS-CoV-2 infection. Two individuals (participant IDs (PIDs) 19C and 196C) were seronegative at all timepoints in the neutralization assay and all RBD and spike ELISAs (Supplementary Figure 1) . We then tested these individuals in the Abbott Architect anti-nucleoprotein assay where they were also seronegative, with index values of 0.01 (for both samples from PID 196C) or 0.02 (for both samples from PID 19C), which are far below the threshold for seropositivity of 1.40 [24] . Because these individuals had only a single positive RT-qPCR test (and PID 196C tested negative in 6 subsequent RT-qPCR tests conducted within 15 days of their initial test), and because the Abbott Architect assay has been validated to have very high (95.1-100%) sensitivity by day 17 post symptom onset [24] , we assessed these two individuals were likely not truly infected but rather false positives in a single RT-qPCR viral test. Therefore, they were excluded from all further analyses, resulting in a final cohort of 32 individuals.

Participants or their legally authorized representatives completed electronic informed consent.

Sociodemographic and clinical data were collected from electronic chart review and from participants via a data collection questionnaire (Project REDCap [25] ) at the time of enrollment. The questionnaire collected data on the nature and duration of symptoms, medical comorbidities, and care-seeking behavior (Supplementary Table 1 ). Based on these data, individuals were classified by disease severity as Asymptomatic, Symptomatic Non-Hospitalized, and Symptomatic Hospitalized.

Individuals who were recruited as inpatients were enrolled during their hospital admission and had samples collected during their hospitalization. After hospital discharge, these participants subsequently returned to an outpatient clinical research site approximately 30 days after symptom A c c e p t e d M a n u s c r i p t 7 onset for follow-up. In person follow-up only occurred if participants were asymptomatic as per CDC guidelines. Outpatients and asymptomatic individuals completed their enrollment, data collection questionnaire, and first blood draw at an outpatient visit approximately 30 days after symptom onset (or positive test for asymptomatic individuals). All participants subsequently were asked to return at day 60 and then at day 90 or 120 for follow-up.

The majority of samples collected from participants were from outpatient visits after recovery.

However, the first sample from PID 13, the first three samples from PID 23, and the first six samples from PID 25 were collected during their hospitalizations. Table 1 . For analyses of fold-change, we required individuals to have a sample collected at the 30-day timepoint. The numbers of individuals included in the foldchange analyses are indicated in Table 1. A c c e p t e d M a n u s c r i p t 8 This study was approved by the University of Washington Human Subjects Institutional Review Board.

Whole blood was collected in acid citrate dextrose tubes then spun down, aliquoted, and frozen at -20ºC within 6 hours of collection. Prior to use in this study, plasma samples were heat inactivated at 56ºC for 60 min and stored at 4ºC. Some samples from the early timepoints were stored at -80ºC after heat inactivation and underwent no more than two freeze/thaw cycles. Plasma samples were spun at 2000xg for 15 min at 4ºC immediately prior to use to pellet platelets.

Protein expression and purification SARS-CoV-2 RBD and spike (S-2P trimer [26] ) proteins were produced in mammalian cells as previously described [26] [27] [28] . Proteins were purified from clarified supernatants as described in [28] .

SDS-PAGE was used to assess purity prior to flash freezing and storage at -80°C.

Testing of serum samples with the Abbott Architect SARS-CoV-2 IgG assay was performed according to the manufacturer's instructions for use and as described in [24] . Index values associated with the IgG enzyme-linked immunosorbent assays (ELISAs) to spike and RBD were conducted as described previously [27] , and were based on a published protocol that recently received emergency use authorization from New York State and the FDA [29, 30] . Plasma samples were diluted with five serial AUC was calculated as the area under the titration curve after putting the serial dilutions on a logscale.

Neutralization assays were conducted using pseudotyped lentiviral particles as described in [31] , with a few modifications. First, we used a spike with a cytoplasmic tail truncation that removes the A c c e p t e d M a n u s c r i p t 10 last 21 amino acids (spike-∆21). The map for this plasmid, HDM-SARS2-Spike-delta21, is in Supplementary File 1 and the plasmid is available from Addgene (Plasmid #155130). We used a spike with a C-terminal deletion because, since publishing our original protocol [31] , other groups have reported that deleting spike's cytoplasmic tail improves titers of spike-pseudotyped viruses [32] [33] [34] [35] . Indeed, we found that the C-terminal deletion increased the titers of our pseudotyped lentiviral particles without affecting neutralization sensitivity (Supplementary Figure 2) . At 50-52 hours post-infection, luciferase activity was measured using the Bright-Glo Luciferase Assay System (Promega, E2610) as described in [31] , except luciferase activity was measured directly in the assay plates. Two "no plasma" wells were included in each row of the neutralization plate and fraction infectivity was calculated by dividing the luciferase readings from the wells with plasma by the average of the "no plasma" wells in the same row. After calculating fraction infectivities, we used the neutcurve Python package (https://jbloomlab.github.io/neutcurve/) to calculate the A c c e p t e d M a n u s c r i p t 11 plasma dilution that inhibited infection by 50% (IC50) by fitting a Hill curve with the bottom fixed at 0 and the top fixed at 1. NT 50 s for each plasma sample were calculated as the reciprocal of the IC50.

Individuals whose plasma was not sufficiently neutralizing to interpolate an IC50 using the Hill curve fit were assigned an NT 50 of 20 (the limit of our dilution series) for plotting in Figures 1A, 1C, and 2B and for fold-change analyses in Figure 1B . Results from SARS-CoV-2 spike-pseudotyped lentivirus neutralization assays have been shown to correlate well with full virus SARS-CoV-2 neutralization assays [36, 37] . Nonetheless, in an effort to help standardize comparisons between neutralization assays, we also ran our assay with a standard serum sample from NIBSC (Research Reagent for Anti-SARS-CoV-2 Ab, NIBSC code: 20/130). This sample had an NT 50 of ~3050 (Supplementary Figure 3) .

Raw data for each sample, including IC50, NT 50 , AUC and relevant demographic data (age, sex, 

We used spike-pseudotyped lentiviral particles [31] to measure neutralizing antibody titers in the longitudinal plasma samples from all 32 infected individuals ( Figure 1A) . All individuals had detectable neutralizing antibody titers (NT 50 >20) at their first convalescent plasma sample, which was generally collected roughly one month post symptom onset. These data are consistent with prior studies showing that most SARS-CoV-2 infected individuals develop neutralizing antibodies [5, 7, 9] . Qualitative inspection of Figure 1A shows that these titers modestly decreased for most A c c e p t e d M a n u s c r i p t 13 individuals over the next few months, although the dynamics were highly heterogeneous across individuals.

To quantify the dynamics of neutralizing antibody titers over time, we calculated the fold change at ~60 and >90 days post symptom onset relative to the ~30 day timepoint, excluding any individuals who lacked a 30-day sample. Taken across all individuals, neutralizing titers significantly declined from 30 to 60 days, and again from 60 to 90 days (see legend of Figure 1B for details). At >90 days, the median neutralizing titer was reduced 3.8-fold relative to the 30-day value ( Figure 1B) . However, most individuals (27/32) still had detectable neutralizing titers at the last timepoint.

We compared the dynamics of neutralizing antibody titers between individuals with different disease severity (Figure 1C) . Individuals with more severe disease tended to have higher neutralizing antibody titers during early convalescence, consistent with prior studies [5, 38, 39] . Specifically, at both ~30 and ~60 days post symptom onset, individuals who required hospitalization had significantly higher neutralizing antibody titers than individuals who did not ( Figure 1C) . From ~30 to >90 days post symptom onset, the NT 50 for symptomatic hospitalized individuals decreased ~18-fold, which is significantly more than the ~3-fold decrease in the NT 50 for non-hospitalized individuals (p=0.03, Wilcoxon rank-sum test) (Supplementary Figure 4) . By >90 days post symptom onset, neutralization titers were not significantly different between disease severity groups (Figure 1C) . At all timepoints, asymptomatic individuals had neutralization titers similar to those of symptomatic non-hospitalized individuals. For all plasma samples, we also used ELISAs to measure IgA, IgM, and IgG binding to the RBD of spike, and IgG binding to the full spike ectodomain [29] . Figure 2A shows each individual's IgA, IgM, and IgG binding antibody titers as quantified by area under the curve (AUC) of the ELISA readings (see Methods for detailed description). Like neutralizing antibody titers, these antibody binding titers tended to decrease over time, although there was substantial variation among individuals. All the ELISA-measured antibody-binding titers are clearly correlated with neutralizing antibody titers ( Figure 2B) .

Individuals with severe disease had higher antibody binding titers at early timepoints. Specifically, individuals who were hospitalized as part of their care had higher IgG, IgA, and IgM binding responses than asymptomatic or symptomatic non-hospitalized individuals at ~30 days post symptom onset (Figure 2C) . By ~60 days post symptom onset, anti-RBD IgM levels were no longer significantly different between severity groups, and by >90 days post symptom onset, binding responses did not differ between severity groups for any antibody subtype. This trend is consistent with data in Figure 1C showing that neutralizing antibody responses were higher for individuals with more severe disease early during convalescence, but reached similar levels across all diseaseseverity groups by >90 days post symptom onset. Among all patients, regardless of disease severity, IgA and IgM levels decreased more than IgG levels from ~30 to >90 days post symptom onset, consistent with other studies [7, 8, 19, 22] .

A c c e p t e d M a n u s c r i p t 15 We have measured the dynamics of neutralizing antibody titers over the first three to four months following infection with SARS-CoV-2 in a well-characterized prospective longitudinal cohort of individuals across a range of disease severities. The titers of neutralizing antibodies declined modestly, with the titers at three to four months post symptom onset generally about four-fold lower than those at one month. This decline in neutralizing antibodies was paralleled by a decline in antibodies binding to the viral spike and its RBD. This decline is generally similar in magnitude to that reported in several other recent studies of antibody dynamics in the months immediately following SARS-CoV-2 infection [5, 7, 19] .

Individuals with more severe disease tended to have higher peak antibody responses at one to two months post symptom onset, consistent with many other studies reporting higher early titers in severely ill SARS-CoV-2 infected individuals [5, 6, 38, 39] . However, by three to four months post symptom onset, neutralizing antibody titers among individuals with severe disease were no longer significantly higher than those of individuals with mild symptoms or even asymptomatic infections.

Therefore, it seems possible that the large peak in antibody production in severely ill individuals wanes more dramatically than in milder cases, consistent with severe disease often leading to an exaggerated burst of short-lived antibody-secreting cells [40, 41] .

Importantly, most individuals in our study still had substantial neutralizing antibody titers at three to four months post symptom onset. While some recent studies have interpreted a modest drop in titers in the first few months after infection as alarming, it is entirely consistent with antibody responses to other respiratory viruses. Acute infection is always associated with an initial peak in antibody titers due to a burst of short-lived antibody-secreting cells [42] . For many other infections, titers decline from this initial peak but then reach a stable plateau that is maintained for years or A c c e p t e d M a n u s c r i p t 16 even decades by long-lived plasma cells and memory B cells that can be recalled during subsequent infections [12] [13] [14] [15] [16] 18, 43, 44] .

The modest declines in antibody titers that we observe over time do have implications for efforts to collect convalescent patient plasma for use in treatment of sick individuals [45] . FDA guidelines suggest minimum cutoffs for the antibody activity in such convalescent plasma (e.g., NT 50 > 160; [39] ). Our results suggest that plasma from convalescent donors collected in the first few months post symptom onset will be more likely to meet these cutoffs; others have made a similar observation [46] . Additionally, our results indicate that if an individual is donating convalescent plasma over time, each plasma sample should be tested for antibody titers to ensure that they remain above the FDA cutoff.

The limitations of this study include the small number of samples, particularly in the asymptomatic and symptomatic hospitalized groups, and recruitment of participants from a single study site, which potentially limits the generalizability of these results. Furthermore, since symptom-onset date relies on individual recollections, it is difficult to precisely match the timing of blood draws across all participants. Additionally, we only had follow-up to about four months post symptom onset and only measured plasma antibody responses. Further studies over longer time frames and with direct interrogation of plasma and memory B cells will be necessary to determine longer term durability of immunity to SARS-CoV-2, as well as its relationship to protection against re-infection [47] .

Despite these limitations, our study shows that titers of neutralizing and binding antibodies targeting SARS-CoV-2 spike remain detectable in most individuals out to >90 days post symptom onset. While M a n u s c r i p t 21 M a n u s c r i p t A c c e p t e d M a n u s c r i p t M a n u s c r i p t 27 

Neutralizing Antibodies Responses to SARS-CoV-2 in COVID-19

Neutralizing antibody responses to SARS-CoV-2 in a COVID-19 recovered patient cohort and their implications

Antibody dynamics to SARS-CoV-2 in Asymptomatic and Mild COVID-19 patients. medRxiv

Antibody responses to SARS-CoV-2 in patients with COVID-19

Longitudinal evaluation and decline of antibody responses in SARS-CoV-2 infection

Antibody responses to SARS-CoV-2 in patients of novel coronavirus disease 2019

Neutralizing and binding antibody kinetics of COVID-19 patients during hospital and convalescent phases

Evidence for sustained mucosal and systemic antibody responses to SARS-CoV-2 antigens in COVID-19 patients

Rapid Generation of Neutralizing Antibody Responses in COVID-19 Patients

Convergent antibody responses to SARS-CoV-2 in convalescent individuals

The Multifaceted B Cell Response to Influenza Virus

Duration of humoral immunity to common viral and vaccine antigens

Antibody dynamics of 2009 influenza A (H1N1) virus in infected patients and vaccinated people in China

Impaired Antibody-mediated Protection and Defective IgA B-Cell Memory in Experimental Infection of Adults with Respiratory Syncytial Virus

MERS-CoV Antibody Responses 1 Year after Symptom Onset, South Korea

Longitudinally Profiling Neutralizing Antibody Response to SARS Coronavirus with Pseudotypes

Neutralizing antibody response and SARS severity

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

Decline of humoral responses against SARS-CoV-2

Clinical and immunological assessment of asymptomatic SARS-CoV-2 infections

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

SARS-CoV-2 infection induces sustained humoral immune responses in convalescent patients following symptomatic COVID-19

Early Detection of Covid-19 through a Citywide Pandemic Surveillance Platform

Performance characteristics of the Abbott architect SARS-CoV-2 IgG assay and seroprevalence in

Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Serological identification of SARS-CoV-2 infections among children visiting a hospital during the initial Seattle outbreak

Elicitation of potent neutralizing antibody responses by designed protein nanoparticle vaccines for SARS-CoV-2. bioRxiv

A serological assay to detect SARS-CoV-2 seroconversion in humans

SARS-CoV-2 Seroconversion in Humans: A Detailed Protocol for a Serological Assay, Antigen Production, and Test Setup

Protocol and Reagents for Pseudotyping Lentiviral Particles with SARS-CoV-2 Spike Protein for Neutralization Assays

A replication-competent vesicular stomatitis virus for studies of SARS-CoV-2 spike-mediated cell entry and its inhibition

Neutralizing antibody and soluble ACE2 inhibition of a replication-competent VSV-SARS-CoV-2 and a clinical isolate of SARS-CoV-2. Cell Host Microbe

Rapid isolation of potent SARS-CoV-2 neutralizing antibodies and protection in a small animal model

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

Human neutralizing antibodies elicited by SARS-CoV-2 infection

A novel receptor-binding domain (RBD)-based mRNA vaccine against SARS-CoV-2

SARS-CoV-2 Neutralizing Antibody Responses Are More Robust in Patients with Severe Disease

Relationship between Anti-Spike Protein Antibody Titers and SARS-CoV-2 In Vitro Virus Neutralization in Convalescent Plasma

Immunologic perturbations in severe COVID-19/SARS-CoV-2 infection. bioRxiv [Internet

IgA dominates the early neutralizing antibody response to SARS-CoV-2. medRxiv

Somatically Hypermutated Plasmodium-Specific IgM+ Memory B Cells Are Rapid, Plastic, Early Responders upon Malaria Rechallenge

Original antigenic sin priming of influenza virus hemagglutinin stalk antibodies

Others. Investigational COVID-19 Convalescent Plasma-Emergency INDs. Accessed on April

Longitudinal analysis of the humoral response to SARS-CoV-2 spike RBD in convalescent plasma donors

Neutralizing antibodies correlate with protection from SARS-CoV-2 in humans during a fishery vessel outbreak with high attack rate

We thank Marion Pepper for helpful input, Drs. David Koelle and Anna Wald for sharing reagents, and Andrea Loes and Meei-Li Huang for experimental assistance. We thank Ariana Magedson, DylanMcDonald and Nicholas Franko for assistance with enrollment. We additionally thank all our research participants in the HAARVI study for their generosity in participation. 

A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t