SARS-CoV-2 Convalescent Sera Binding and Neutralizing Antibody Concentrations Compared with COVID-19 Vaccine Efficacy Estimates Against Symptomatic Infection

word count: 172 12 Text word count: 3,282 13 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. 2 Footnotes 14 1Potential conflicts of interest have been disclosed within the ICMJE Disclosure Forms. 15 2This work was supported by the United States Centers for Disease Control and Prevention. 16 3This work has not been presented at any national or international meeting. 17 4Correspondence: Adi V. Gundlapalli, MD, PhD, Center for Surveillance, Epidemiology, and 18 Laboratory Services, United States Centers for Disease Control and Prevention, Atlanta, Georgia, 19 30333; [email protected] 20 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 3

21 Previous vaccine efficacy (VE) studies have estimated neutralizing and binding antibody 22 concentrations that correlate with protection from symptomatic infection; how these estimates compare 23 to those generated in response to SARS-CoV-2 infection is unclear. Here, we assessed quantitative 24 neutralizing and binding antibody concentrations using standardized SARS-CoV-2 assays on 3,067 25 serum specimens collected during July 27, 2020-August 27, 2020 from COVID-19 unvaccinated persons 26 with detectable anti-SARS-CoV-2 antibodies using qualitative antibody assays. Quantitative neutralizing 27 and binding antibody concentrations were strongly positively correlated (r=0.76, p<0.0001) and were 28 noted to be several fold lower in the unvaccinated study population as compared to published data on 29 concentrations noted 28 days post-vaccination. In this convenience sample, ~88% of neutralizing and 30 ~63-86% of binding antibody concentrations met or exceeded concentrations associated with 70% 31 COVID-19 VE against symptomatic infection from published VE studies; ~30% of neutralizing and 1-32 14% of binding antibody concentrations met or exceeded concentrations associated with 90% COVID-33 19 VE. These data support observations of infection-induced immunity and current recommendations 34 for vaccination post infection to maximize protection against symptomatic COVID-19. 35 36

SARS-CoV-2, COVID-19, quantitative anti-SARS-CoV-2 IgG, neutralizing antibodies, 37 correlation, antibody, immune, protection, correlate of protection 38 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 4

39 As of October 18, 2021, 66.7% of the US population 12 years of age and older has been fully 40 vaccinated for COVID-19 [1]. A recent Gallup survey found that 18% of Americans would not agree to 41 be vaccinated if a US Food and Drug Administration (FDA)-approved COVID-19 vaccine were 42 available to them immediately at no cost [2]. One of the primary reasons cited for vaccine hesitancy was 43 a history of SARS-CoV-2 infection and resultant antibodies. 44 Cumulative evidence indicates that SARS-CoV-2 antibodies are protective against SARS-CoV-2 re-45 infection [3]. A series of non-human primate challenge studies demonstrated the central role of SARS-46 CoV-2 neutralizing antibodies in protection from re-infection [4-6]. A randomized clinical trial 47 involving the subcutaneous administration of REGEN-COV, a combination of two SARS-CoV-2 48 neutralizing monoclonal antibodies, or placebo within 96 hours of SARS-CoV-2 exposure demonstrated 49 that REGEN-COV prevented symptomatic COVID-19 and asymptomatic SARS-CoV-2 infection [7]. A 50 longitudinal study of >12,000 health care workers showed that SARS-CoV-2 infection-induced 51 protective immunity lasts for at least 6 months [8]. 52 The establishment of the First World Health Organization (WHO) International Standard for Anti-53 SARS-CoV-2 Immunoglobulin [9] for quantitative assessment of neutralizing and binding antibody 54 concentrations has made it possible for COVID-19 vaccine efficacy (VE) studies [10-13] to describe and 55 propose standardized immune correlates of protection against symptomatic infection (or risk of 56 symptomatic infection) in fully-vaccinated persons (i.e., 2 weeks after their second dose in a 2-dose 57 series, such as Pfizer-BioNTech (BNT162b2), Moderna (mRNA-1273), or AstraZeneca (ChAdOx1) 58 vaccines OR 2 weeks after a single-dose vaccine, such as Johnson & Johnson’s Janssen [JNJ-78436735] 59 vaccine) across vaccine trials that have used different antibody assays. Comparing antibody 60 concentrations of unvaccinated persons with anti-SARS-CoV-2 antibodies to COVID-19 vaccinated 61 cohorts and to estimated antibody concentrations associated with COVID-19 vaccine effectiveness 62 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 5 would accomplish three goals: 1) improve understanding of the population distribution of antibody 63 concentrations in response to infection and vaccination, 2) help establish relationship of quantitative 64 antibody levels to those associated with protection against illness from published studies, and 3) inform 65 better targeted messaging for universal COVID-19 vaccination. 66 Numerous SARS-CoV-2 serological assays have been developed over the course of the COVID-19 67 pandemic to measure virus-specific antibody responses [14]. Persons who recover from SARS-CoV-2 68 infection or receive a COVID-19 vaccine typically develop virus-specific neutralizing antibodies, with 69 most of these antibodies directed against the immunodominant receptor binding domain (RBD) of the 70 spike (S) protein [15, 16]. Efficacy trials of the ChAdOx1 [11] and mRNA-1273 [12] vaccines showed 71 that higher anti-SARS-CoV-2 S IgG, anti-SARS-CoV-2 RBD IgG, and SARS-CoV-2 neutralizing 72 antibody concentrations were correlated with a reduced risk of symptomatic infection. Both trials 73 determined antibody concentrations associated with varying levels of VE against symptomatic COVID-74 19 [11, 12]. The ChAdOx1 vaccine trial estimated that 70% and 90% VE against symptomatic COVID-75 19 was associated with 50% neutralizing antibody titer (NT50) concentrations of 3.7 and 64.1 76 international units per mL (IU/mL), respectively, and 70% and 90% VE was associated with anti-RBD 77 IgG antibodies concentrations of 165.0 and 2360.0 binding antibody units per mL (BAU/mL), 78 respectively [11]. The mRNA-1273 vaccine trial estimated that 70% and 90% VE against symptomatic 79 COVID-19 was associated with NT50 concentrations of 4.0 and 83.0 IU/mL, respectively, and 70% and 80 90% VE was associated with anti-RBD IgG antibodies concentrations of 8.0 and 775.0 BAU/mL, 81 respectively [12]. However, it remains unclear how these antibody concentrations compare to those 82 generated in response to SARS-CoV-2 infection in COVID-19 unvaccinated persons. To answer this 83 question, we tested 3,067 pre-vaccination sera collected as part of a nationwide commercial laboratory 84 seroprevalence study led by the US Centers for Disease Control and Prevention (CDC) [17] that had 85 previously tested positive for anti-SARS-CoV-2 antibodies using standardized anti-SARS-CoV-2 RBD 86 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 6 IgG and SARS-CoV-2 pseudovirus neutralization antibody assays, and then compared these antibody 87 concentrations to concentrations measured in COVID-19 vaccinated cohorts (28 days post-vaccination) 88 and to those associated with 70% and 90% COVID-19 VE against symptomatic infection for two of the 89 vaccines for which such data are published to date. 90

91 Specimen Source and Study Design 92 Sera were collected by two US-based commercial laboratories for routine or acute clinical care (e.g., 93 cholesterol screening or a sick visit) as part of a nationwide seroprevalence study; detailed methods have 94 previously been described [17]. Briefly, blood specimens collected for COVID-19-related reasons were 95 excluded. Sera were tested [17] for anti-SARS-CoV-2 antibodies using one of the following three 96 qualitative assays issued emergency use authorization by the FDA that were in use in the clinical 97 laboratories: 1) Architect™ SARS -CoV-2 IgG Assay (nucleocapsid (N) protein; Abbott, Chicago, IL), 98 2) VITROS® Anti-SARS-CoV-2 IgG Assay (S protein; Ortho-Clinical Diagnostics, Raritan, NJ), and 3) 99 Elecsys® Anti-SARS-CoV-2 Assay (N protein; Roche, Indianapolis, IN) (Supplementary figure 3). 100 Among 84,683 total serum samples collected during July 27, 2020-August 27, 2020 identified as anti-101 SARS-CoV-2 antibody positive and had linked age and sex information (as part of the larger 102 serosurveillance study) [17], 3067 serum specimens were selected by convenience sampling for reflex 103 testing with anti-SARS-CoV-2 quantitative IgG and neutralizing antibody assays. Data on race, 104 ethnicity, and occurrence or date of SARS-CoV-2 infection or symptoms were not available; data on 105 prior SARS-CoV-2 quantitative reverse transcription polymerase chain reaction qRT-PCR test results 106 were not available for most persons. 107 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 7 Ethics 108 This activity was reviewed by CDC and was conducted consistent with applicable federal law and 109 CDC policy (45 C.F.R. part 46, 21 C.F.R. part 56; 42 U.S.C. Sect. 241(d); 5 U.S.C. Sect. 552a; 44 110 U.S.C. Sect. 3501 et seq.). Informed consent was waived, as all data were deidentified. 111 PhenoSense® CoV Neutralizing Antibody Assay 112 Neutralizing antibodies against the SARS-CoV-2 S protein were measured using the PhenoSense 113 CoV Neutralizing Antibody Assay® [18]. To measure SARS-CoV-2 neutralizing antibodies, human 114 immunodeficiency virus-1 pseudovirions expressing the SARS-CoV-2 S protein were prepared by co-115 transfecting HEK293 producer cells with an HIV-1 genomic vector and a SARS-CoV-2 envelope 116 protein expression vector. Serial dilutions of sera were incubated with recombinant pseudovirions and 117 neutralizing activity was assessed by measuring the inhibition of luciferase activity in HEK293 target 118 cells co-expressing the angiotensin-converting enzyme 2 and transmembrane serine protease 2 receptors. 119 Fifty percent neutralizing antibody titers (NT50) were expressed as the reciprocal of the serum dilution 120 conferring 50% inhibition of pseudovirus infection. Calibration of the PhenoSense CoV Neutralizing 121 Antibody Assay® with the First WHO International Standard for Anti-SARS-CoV-2 Immunoglobulin 122 (20/136, National Institute for Biological Standards and Controls, UK) [9] allowed us to generate a 123 calibration factor of 0.0653 (S protein containing G614) and convert NT50 values from titers to 124 international units per mL (IU/mL) by multiplying by the calibration factor . 125 Cov2Quant IgG® Assay 126 Anti-SARS-CoV-2 RBD IgG in sera were quantified using the electrochemiluminescent (ECL) 127 Cov2Quant IgG® Assay [19]. Briefly, sera were incubated with biotin-conjugated, recombinant SARS-128 CoV-2 RBD antigen bound to a streptavidin, carbon-coated microtiter well plate. After washing the 129 plate, the wells were incubated with ruthenium-conjugated anti-IgG and a final wash was performed to 130 remove unbound material. Voltage was applied to the metallic electrode on the base of the plate and a 131 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 8 photomultiplier was used to measure the ECL signal generated from each well. The signal in relative 132 light units (RLU) is directly proportional to the anti-RBD IgG concentration in sera. Each assay batch 133 included a dilution series of affinity-purified human IgG standard that was used to calculate the anti-134 SARS-CoV-2 RBD IgG concentration of each serum. Calibration of the Cov2Quant IgG® Assay with 135 the WHO International Standard allowed us to generate a calibration factor of 25 and convert anti-136 SARS-CoV-2 RBD IgG concentrations from µg/mL to binding antibody units per mL (BAU/mL) by 137 multiplying by the calibration factor. 138 Comparison of Study Antibody Data to Antibody Concentrations Associated with COVID-19 VE 139 Of the COVID-19 vaccines in use around the world currently, serological correlates of protection 140 thresholds associated with VE have been estimated for ChAdOx1 [11] and mRNA-1273 [12] using the 141 First WHO International Standard for Anti-SARS-CoV-2 Immunoglobulin. As the quantitative antibody 142 assays used in our study were calibrated to the same standard, we were able to compare antibody 143 concentrations in our study to those estimated to be associated with 70% and 90% COVID-19 VE to 144 ChAdOx1 and mRNA-1273. 145 Statistical Analyses 146 D ata management tasks, and statistical analyses were performed using SAS version 9.4 (SAS 147 Institute Inc., Cary, NC) and GraphPad Prism 9.0.0 (GraphPad Software, San Diego, CA). T-tests on 148 log10-transformed data were used to determine if there were statistically significant differences between 149 geometric mean anti-SARS-CoV-2 RBD IgG and NT50 concentrations by sex. ANOVAs on log10-150 transformed data were used to determine if there were statistically significant differences between 151 geometric mean anti-SARS-CoV-2 RBD IgG and NT50 concentration according to previous qualitative 152 antibody test and age category. Post-hoc Tukey’s tests were performed to identify previous qualitative 153 antibody tests and age categories with significantly different anti-SARS-CoV-2 RBD IgG and NT50 154 concentrations; p-values were adjusted to account for multiple comparisons. Pearson’s correlation was 155 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 9 used to assess the overall relationship between log10anti-RBD IgG and log10NT50 concentrations, as well 156 as the relationships between the two variables across sex and age category. Two-sided p-values <0.05 157 were considered statistically significant. 158

159 Of the overall convenience sample of 3,067 serum specimens collected during July 27, 2020-August 160 27, 2020 with detectable anti-SARS-CoV-2 antibodies on a qualitative assay, 1,309 were from males 161 (42.7%) and 1,758 (57.3%) were from females, 779 (25.4%) were from persons aged <18 years, 982 162 (32.0%) were from persons aged 18–49 years, 798 (26.0%) were from persons aged 50–64 years, and 163 508 (16.6%) were from persons aged ≥65 years. 164 Most of the serum specimens (n=2568, 83.7%) tested positive for both quantitative SARS-CoV-2 165 neutralizing antibodies and quantitative anti-SARS-CoV-2 RBD IgG; 271 sera (8.8%) tested negative 166 for both neutralizing antibodies and anti-RBD IgG, 83 sera (2.7%) tested negative for neutralizing 167 antibodies only, and 145 sera (4.7%) tested negative for anti-RBD IgG only. Geometric mean NT50 168 concentrations differed significantly according to the qualitative antibody test type that was used to 169 screen sera for study inclusion (p<0.0001; Supplementary Figure 4A), while geometric mean anti-RBD 170 IgG concentrations did not differ significantly according to the qualitative antibody test type that was 171 used to screen sera for study inclusion (p=0.02998; Supplementary Figure 4B). 172 The overall geometric mean NT50 concentration was 28.8 IU/mL (95% confidence interval (CI): 173 26.8, 31.0) and did not differ significantly according to sex (male: 30.9 IU/mL, 95% CI: 27.4, 34.8; 174 female: 27.3 IU/mL, 95% CI: 25.0, 30.0; p=0.1088) but differed significantly according to age category 175 (<18 years: 31.3 IU/mL, 95% CI: 28.4, 34.6; 18–49 years: 23.1 IU/mL, 95% CI: 20.3, 26.2; 50–64 176 years: 38.0 IU/mL, 95% CI: 32.5, 44.4; ≥65 years: 25.0 IU/mL, 95% CI: 20, 31.3; p<0.0001; Figure 177 1A). Persons aged <18 years had significantly higher NT50 concentrations compared to persons aged 18–178 49 years (p=0.0108), and persons aged 50–64 years had significantly higher NT50 concentrations 179 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 10 compared to persons aged 18–49 years (p<0.0001) and persons aged ≥65 years (p=0.0020) (Figure 1A). 180 The overall geometric mean anti-RBD IgG concentration was 162.5 BAU/mL (95% CI: 152.9, 172.7) 181 and did not differ significantly according to sex (male: 171.3 BAU/mL, 95% CI: 155.6, 188.5; female: 182 156.3 BAU/mL, 95% CI: 144.4, 169.1; p=0.1447) but did differ significantly according to age category 183 (<18 years: 218.2 BAU/mL, 95% CI: 199.2, 239.1; 18–49 years: 122.2 BAU/mL, 95% CI: 109.8, 136.1; 184 50–64 years: 189.0 BAU/mL, 95% CI: 167.2, 213.7; ≥65 years: 141.4 BAU/mL, 95% CI: 117.8, 169.8; 185 p<0.0001; Figure 1B). Like the age-associated patterns of neutralizing antibody concentrations, persons 186 aged <18 years had significantly higher anti-RBD IgG concentrations compared to persons aged 18–49 187 years (p<0.0001) and persons aged ≥65 years (p<0.0001), and persons aged 50–64 years had 188 significantly higher anti-RBD IgG concentrations compared to persons aged 18–49 years (p<0.0001) 189 and persons aged ≥65 years (p=0.0147) (Figure 1B). 190 We found an overall significant, strong positive correlation between NT50 concentrations and anti-191 RBD IgG concentrations (r=0.76, p<0.0001; Figure 1A). The strength of the relationship remained 192 consistent across sex (male: r=0.76, p<0.0001; female: r=0.76, p<0.0001; Figure 1B) and increased with 193 increasing age category (<18 years: r=0.67, p<0.0001; 18–49 years: r=0.69, p<0.0001; 50–64 years: 194 r=0.79, p<0.0001; ≥65 years: r=0.88, p<0.0001; Figure 2). 195 NT50 concentrations in the overall convenience sample of 3,067 serum specimens collected during 196 July 27, 2020-August 27, 2020 with detectable anti-SARS-CoV-2 antibodies on a qualitative assay were 197 2.7-fold lower and 8.6-fold lower than concentrations reported from serum specimens collected 28 days 198 post COVID-19 vaccination from ChAdOx1 [11] and mRNA-1273 [12] VE study participants classified 199 as not having COVID-19 during the follow-up period, respectively (Figures 3A and B). Likewise, anti-200 RBD IgG concentrations in the overall convenience sample were 1.5-fold lower and 24.2-fold lower 201 than concentrations reported from serum specimens collected 28 days post COVID-19 vaccination from 202 ChAdOx1 and mRNA-1273 VE study participants classified as not having COVID-19 during the 203 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 11 follow-up period, respectively (Figure 3C and D). Overall, 88.3% of serum specimens in this study met 204 or exceeded the concentration of SARS-CoV-2 neutralizing antibodies associated with 70% ChAdOx1 205 and mRNA-1273 VE, while 32.9% and 27.1% of sera met or exceeded the concentration of SARS-CoV-206 2 neutralizing antibodies associated with 90% ChAdOx1 and mRNA-1273 VE, respectively (Figure 207 1A). For binding antibodies, we found that 63.1% and 86.4% of serum specimens from this study met or 208 exceeded the concentration of anti-RBD IgG antibodies associated with 70% ChAdOx1 and mRNA-209 1273 VE, respectively, and 1.3% and 13.6% of sera met or exceeded the concentration of anti-RBD IgG 210 antibodies associated with 90% ChAdOx1 and mRNA-1273 VE, respectively (Figure 1B). A greater 211 percentage of males and persons aged ≥50 years met or exceeded the concentrations of SARS-CoV-2 212 neutralizing antibodies and anti-RBD Ig associated with 70% ChAdOx1 and mRNA-1273 VE than 213 females and persons aged <50 years, respectively. 214

215 Similar to the ChAdOx1 [11] and mRNA-1273 [12] VE trial serologic testing results reported in the 216 literature, we demonstrate a statistically significant, strong positive correlation between standardized 217 anti-SARS-CoV-2 RBD IgG and SARS-CoV-2 NT50 concentrations, across all sex and age categories, 218 using sera from unvaccinated persons with previous SARS-CoV-2 infection. Based on these broad 219 correlations, our results suggest that high-throughput, quantitative anti-SARS-CoV-2 RBD IgG assays 220 can be used as surrogates for SARS-CoV-2 pseudovirus neutralization assays that are considered 221 mechanistic correlates of protection [12]. Performing high-throughput, commercially available 222 quantitative IgG assays is logistically feasible and practical on a large scale compared to performing 223 time- and resource-intensive neutralization assays. Thus, there is a continued need for FDA-approved 224 quantitative anti-SARS-CoV-2 S and RBD IgG assays in the management of the pandemic to facilitate 225 assessment of antibody concentrations associated with protection from infection and/or disease [14]. 226 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 12 While our finding that most serum specimens in this study met or exceeded antibody concentrations 227 associated with 70% COVID VE estimates support the real-world observations of infection-induced 228 immunity from re-infection, it is important to note that less than 33% of sera from unvaccinated persons 229 with previous SARS-CoV-2 infection met or exceeded antibody concentrations associated with 90% 230 ChAdOx1 and mRNA-1273 vaccine efficacy. This suggests that not all persons with a history of SARS-231 CoV-2 infection generate antibody responses of sufficient magnitude, as measured by a single, randomly 232 timed blood test, to protect them against symptomatic re-infection. In addition to the circulation of the 233 hyper-transmissible Delta variant of SARS-CoV-2, this finding may also partially explain the rise in 234 COVID-19 cases that occurred in the United States in August 2021 despite a national serosurvey in May 235 2021 showing that 83.3% of the population had SARS-CoV-2 infection- and/or vaccine-induced 236 antibodies [20]. Meanwhile, at least in the short-term (28 days), COVID-19 vaccination can provide 2.7 237 to 8.6-fold higher neutralizing antibody concentrations and 1.5 to 24.2-fold higher binding antibody 238 concentrations. Thus, our results support current guidance that all eligible persons should consider 239 COVID-19 vaccination to maximize their protection against symptomatic COVID-19 regardless of their 240 SARS-CoV-2 infection history. 241 Our sampling approach of selecting specimens identified as positive using at least one qualitative 242 anti-SARS-CoV-2 antibody assay would have selected for persons who had more severe disease and 243 thus higher antibody concentrations, and/or in whom antibodies had not waned. This selection bias may 244 have led us to overestimate the percentage of persons with antibody concentrations meeting or 245 exceeding those concentrations associated with 70% and 90% COVID-19 VE and would have missed 246 persons infected with SARS-CoV-2 who did not mount a measurable post-infection immune response or 247 had antibodies below detectable levels. 248 Our study has several limitations. First, due to the absence of information on whether persons had 249 symptoms or if testing had been done or, if so, the date of COVID-19 symptom onset and/or a positive 250 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 13 SARS-CoV-2 qRT-PCR or antigen test, we were unable to calculate the number of days that had elapsed 251 between infection and collection of serum for antibody testing and thus the measurements might not 252 reflect peak antibody concentrations. The ChAdOx1 and mRNA-1273 VE trials measured serum 253 antibody concentrations 28 days after the second vaccine dose, presumably at the peak of the measurable 254 humoral immune response to the vaccine. Therefore, if the sera used in this study were collected >28 255 days after SARS-CoV-2 infection, it is possible that the percentage of persons with antibody 256 concentrations meeting or exceeding those concentrations associated with 70% and 90% COVID-19 VE 257 were underestimated. However, as a result of the sharp rise in COVID-19 cases in the United States in 258 mid-June 2020 (Supplementary Figure 2) [21], >50% of cases reported in the country prior to July 27, 259 2020 (date first specimens used in this study were collected) occurred after June 15, 2020. This indicates 260 that the majority of sera in this study were likely to have been collected within 73 days of SARS-CoV-2 261 infection, a time before IgG antibody concentrations would be expected to have significantly waned 262 from peak post-infection levels [22]. 263 Second, 13.6% of persons had anti- RBD IgG concentrations below the cutoff value of the assay 264 (37.5 BAU/mL). However, the anti-RBD IgG concentration associated with 70% mRNA-1273 VE (8.0 265 BAU/mL) is below the cutoff value of the Cov2Quant IgG® Assay used in this study. Therefore, it is 266 possible that some persons below the assay cutoff value might have had anti-RBD IgG concentrations 267 greater than or equal to 8.0 BAU/mL, leading us to underestimate the percentage of persons with 268 antibody concentrations meeting or exceeding those concentrations associated with 70% mRNA-1274 269 COVID-19 VE. 270 Third, extrapolating these population-level results to individual-level clinical decision making with 271 regard to immune protection based on a single antibody test poses a challenge. The sera tested here were 272 untimed with regard to onset of infection and we have no knowledge of the immune status of the 273 individuals in terms of co-morbidities and medications. These challenges are also applicable to post-274 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 14 vaccination sera in terms of certain populations responding less well to vaccines in terms of neutralizing 275 and binding antibodies as compared to healthy volunteers in vaccine trials and limited availability of 276 standardized quantitative assays [3]. 277 Fourth, our findings , as well as the timing of the ChAdOx1 CoV-19 and mRNA-1273 VE trials, 278 predate the circulation of the Delta variant of SARS-CoV-2 that exhibits a decreased sensitivity to post-279 vaccination antibody neutralization compared to the Alpha variant of the virus [23]. Therefore, our data 280 is not able to address potential protection of persons infected SARS-CoV-2 in mid-2020 against 281 infection with the Delta variant of the virus. 282 Finally, the efficiency and redundancy of the immune system, especially for the prevention of severe 283 disease, in terms of humoral [12], cellular immunity, memory and anamnestic response on second 284 exposure to SARS-CoV-2 likely contribute to protection beyond an estimate indicated solely by single 285 antibody concentrations [6, 24]. 286 In conclusion, we demonstrate that in this sample, most non- vaccinated persons with qualitative 287 antibody evidence of prior infection had quantitative antibody concentrations against SARS-CoV-2 that 288 met or exceeded antibody levels associated with 70% VE. However, only a small proportion had 289 antibody responses that met or exceeded levels associated with 90% VE. Our findings suggest that 290 persons with a prior COVID-19 would benefit from vaccination to maximize protective antibody 291 concentrations against symptomatic COVID-19. 292 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 15 Notes 293 Acknowledgments. The authors thank Marla Williams, Morgan Herle, Wade Tanico, and Paul Theobald 294 from Labcorp for help received in planning and execution of sample logistics and data acquisition. A 295 special thanks goes to the reference laboratory operators that performed Cov2Quant IgG® and 296 PhenoSense® CoV Neutralizing Antibody assays. We also thank members of the Multistate Assessment 297 of SARS-CoV-2 Seroprevalence (MASS) steering group, including Lyle Peterson, Margaret Honein, 298 Adam MacNeil, Eduardo Azziz-Baumgartner, Francisco Averhoff, Sridhar Basavaraju, Tina Benoit, 299 Carla Black, Kevin Berney, Ryan Weigand, Michele Owen, Ruchi Pancholy, and Lisa Grohskopf. 300 Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily 301 represent the official position of the Centers for Disease Control and Prevention. 302 Financial support. This work was supported by the United States Centers for Disease Control and 303 Prevention, Atlanta, Georgia. 304 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 16

305 1. US Centers for Disease Control and Prevention. COVID-19 Vaccinations in the United States. 306 Available at: https://covid.cdc.gov/covid-data-tracker/#vaccinations_vacc-total-admin-rate-total. 307 Accessed October 14, 2021. 308 2. Jones JM. About One in Five Americans Remain Vaccine-Resistant. Available at: 309 https://news.gallup.com/poll/353081/one-five-americans-remain-vaccine-resistant.aspx. Accessed 310 September 12, 2021. 311 3. US Centers for Disease Control and Prevention. Science Brief: SARS-CoV-2 Infection-induced and 312 Vaccine-induced Immunity. Available at: https://www.cdc.gov/coronavirus/2019-313 ncov/science/science-briefs/vaccine-induced-immunity.html. Accessed November 5, 2021 2021. 314 4. Chandrashekar A, Liu J, Martinot AJ, et al. SARS -CoV-2 infection protects against rechallenge in 315 rhesus macaques. Science 2020; 369:812-7. 316 5. McMahan K, Yu J, Mercado NB, et al. Correlates of protection against SARS- CoV-2 in rhesus 317 macaques. Nature 2021; 590:630-4. 318 6. Corbett KS, Nason MC, Flach B, et al. Immune correlates of protection by mRNA-1273 vaccine 319 against SARS-CoV-2 in nonhuman primates. Science 2021; 373:eabj0299. 320 7. O'Brien MP, Forleo- Neto E, Musser BJ, et al. Subcutaneous REGEN-COV Antibody Combination 321 to Prevent Covid-19. N Engl J Med 2021; 385:1184-95. 322 8. Lumley SF, O'Donnell D, Stoesser NE, et al. Antibody Status and Incidence of SARS- CoV-2 323 Infection in Health Care Workers. N Engl J Med 2021; 384:533-40. 324 9. Mattiuzzo G, Bentley EM, Hassall M, et al. Establishment of the WHO International Standard and 325

Panel for anti-SARS-CoV-2 antibody (WHO/BS/2020/2403): World Health Organization, 326 2020. 327 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 17 10. Earle KA, Ambrosino DM, Fiore-Gartland A, et al. Evidence for antibody as a protective correlate 328 for COVID-19 vaccines. Vaccine 2021; 39:4423-8. 329 11. Feng S, Phillips DJ, White T, et al. Correlates of protection against symptomatic and asymptomatic 330 SARS-CoV-2 infection. medRxiv 2021. 331 12. Gilbert PB, Montefiori DC, McDermott A, et al. Immune Correlates Analysis of the mRNA-1273 332 COVID-19 Vaccine Efficacy Trial. medRxiv 2021. 333 13. Khoury DS, Cromer D, Reynaldi A, et al. Neutralizing antibody levels are highly predictive of 334 immune protection from symptomatic SARS-CoV-2 infection. Nat Med 2021; 27:1205-11. 335 14. Gundlapalli AV, Salerno RM, Brooks JT, et al. SARS-CoV-2 Serologic Assay Needs for the Next 336 Phase of the US COVID-19 Pandemic Response. Open Forum Infect Dis 2021; 8:ofaa555. 337 15. Ju B, Zhang Q, Ge J, et al. Human neutralizing antibodies elicited by SARS-CoV-2 infection. Nature 338 2020; 584:115-9. 339 16. Robbiani DF, Gaebler C, Muecksch F, et al. Convergent antibody responses to SARS-CoV-2 in 340 convalescent individuals. Nature 2020; 584:437-42. 341 17. Bajema KL, Wiegand RE, Cuffe K, et al. Estimated SARS-CoV-2 Seroprevalence in the US as of 342 September 2020. JAMA Intern Med 2021; 181:450-60. 343 18. Markmann AJ, Giallourou N, Bhowmik DR, et al. Sex disparities and neutralizing antibody 344 durability to SARS-CoV-2 infection in convalescent individuals. medRxiv 2021. 345 19. Kappelman MD, Weaver KN, Boccieri M, et al. Humoral Immune Response to Messenger RNA 346 COVID-19 Vaccines Among Patients With Inflammatory Bowel Disease. Gastroenterology 2021. 347 20. Jones JM, Stone M, Sulaeman H, et al. Estimated US Infection- and Vaccine-Induced SARS-CoV-2 348 Seroprevalence Based on Blood Donations, July 2020-May 2021. JAMA 2021. 349 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 18 21. US Centers for Disease Control and Prevention. Trends in Number of COVID-19 Cases and Deaths 350 in the US Reported to CDC, by State/Territory. Available at: https://covid.cdc.gov/covid-data-351 tracker/#trends_dailycases. Accessed October 14, 2021. 352 22. Egbert ER, Xiao S, Colantuoni E, et al. Durability of Spike Immunoglobin G Antibodies to SARS-353 CoV-2 Among Health Care Workers With Prior Infection. JAMA Netw Open 2021; 4:e2123256. 354 23. Planas D, Veyer D, Baidaliuk A, et al. Reduced sensitivity of SARS-CoV-2 variant Delta to 355 antibody neutralization. Nature 2021; 596:276-80. 356 24. Sette A, Crotty S. Adaptive immunity to SARS-CoV-2 and COVID-19. Cell 2021; 184:861-80. 357 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 19 Figure 1. SARS-CoV-2 50% neutralizing antibody titer (NT50) concentrations and anti-SARS-CoV-2 358 receptor binding domain (RBD) IgG concentrations for the overall convenience sample of 3,067 serum 359 specimens collected during July 27,2020-August 27, 2020 with detectable anti-SARS-CoV-2 antibodies 360 on a qualitative assay, and according to sex and age category. A) SARS-CoV-2 NT50 concentrations in 361 international units per mL (IU/mL) and B) Anti-SARS-CoV-2 RBD IgG concentrations in binding 362 antibody units per mL (BAU/mL). Horizontal bars represent geometric means, vertical error bars 363 represent 95% confidence intervals, and dashed horizontal lines represent assay cut-off values for 364 seropositivity. P-values from t-tests (sex) and post-hoc Tukey tests (age class) are shown for each sex 365 and age class comparison. Bolded p-values denote statistical significance (p<0.05). Dotted horizontal 366 lines represent antibody concentrations associated with 70% and 90% ChAdOx1 [11] and mRNA-1273 367 [12] vaccine efficacy (VE). The percentage of sera with antibody concentrations that meet or exceed the 368 concentrations represented by each of the horizontal dotted lines are shown below the charts. 369 Figure 2. Correlations between SARS-CoV-2 50% neutralizing antibody titer (NT50) concentrations in 370 international units per mL (IU/mL) and anti-SARS-CoV-2 receptor binding domain (RBD) IgG 371 concentrations in binding antibody units per mL (BAU/mL) for the convenience sample of 3,067 serum 372 specimens collected during July 27,2020-August 27, 2020 with detectable anti-SARS-CoV-2 antibodies 373 on a qualitative assay. A) Overall, B) by sex, and C) by age class. 374 Figure 3. SARS-CoV-2 50% neutralizing antibody titer (NT50) concentrations in international units per 375 mL (IU/mL) and anti-SARS-CoV-2 receptor binding domain (RBD) IgG concentrations in binding 376 antibody units per mL (BAU/mL) for the overall convenience sample of 3,067 serum specimens 377 collected during July 27,2020-August 27, 2020 with detectable anti-SARS-CoV-2 antibodies on a 378 qualitative assay compared to 28 days post COVID-19 vaccination concentrations in ChAdOx1 [11] and 379 mRNA-1273 [12] VE study participants classified as not having COVID-19 during the follow-up period. 380 A) Median and interquartile range SARS-CoV-2 NT50 concentrations for the overall convenience 381 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 20 sample compared to concentrations in VE study participants who received the ChAdOx1 vaccine and 382 were classified as not having COVID-19 during the follow-up period, B) Geometric mean and 95% 383 confidence interval SARS-CoV-2 NT50 concentrations for the overall convenience sample compared to 384 concentrations in VE study participants who received the mRNA-1273 vaccine and were classified as 385 not having COVID-19 during the follow-up period, C) Median and interquartile range anti-SARS-CoV-386 2 RBD IgG concentrations for the overall convenience sample compared to concentrations in VE study 387 participants who received the ChAdOx1 vaccine and were classified as not having COVID-19 during the 388 follow-up period, and D) Geometric mean and 95% CI anti-SARS-CoV-2 RBD IgG concentrations for 389 the overall convenience sample compared to concentrations in VE study participants who received the 390 mRNA-1273 vaccine and were classified as not having COVID-19 during the follow-up period. 391 Horizontal bars represent medians or geometric means, and vertical error bars represent the interquartile 392 range surrounding the median or 95% confidence intervals surrounding the geometric mean. 393 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint Overall Male Female <18 18—49 50—64 ³65 1 10 100 1000 10000 100000NT50 (IU/mL) A 0.8914 0.2213 0.2456 0.1088 0.002<0.00010.0108 88.3% 88.3% 27.1% 32.9% 87.1% 87.0% 31.1% 35.9% 74.5% 74.5% 24.1% 30.6% 95.3% 95.3% 20.4% 26.6% 79.3% 79.3% 21.9% 26.7% 89.1% 89.1% 37.1% 80.5% 80.5% 32.1% 38.2% 70% ChAdOx1 VE 70% mRNA-1273 VE 90% mRNA-1273 VE 90% ChAdOx1 VE43.7% 70% ChAdOx1 and mRNA-1273 VE 90% ChAdOx1 VE 90% mRNA-1273 VE Cut-off for seropositivity B Overall Male Female <18 18—49 50—64 ³65 10 100 1000 10000Anti-RBD IgG (BAU/mL) 70% mRNA-1273 VE 90% ChAdOx1 VE 90% mRNA-1273 VE 70% ChAdOx1 VE Cut-off for seropositivity 0.3978 0.0147 0.3382 0.1447 0.0147<0.0001<0.0001 63.1% 86.4% 13.6% 1.3% 64.9% 86.0% 16.3% 1.6% 58.5% 72.9% 11.4% 1.0% 71.9% 93.7% 10.0% 0.3% 51.6% 74.7% 8.6% 0.8% 67.4% 88.0% 19.0% 63.2% 78.3% 20.3% 3.0% 70% ChAdOx1 VE 70% mRNA-1273 VE 90% mRNA-1273 VE 90% ChAdOx1 VE1.8% for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint 1 10 100 1000 10000 100000 10 100 1000 10000 NT50 (IU/mL) Anti-RBD IgG (BAU/mL) Overall (r=0.76, p<0.0001) A 1 10 100 1000 10000 100000 10 100 1000 10000 NT50 (IU/mL) Anti-RBD IgG (BAU/mL) Male (r=0.76, p<0.0001) Female (r=0.76, p<0.0001) B 1 10 100 1000 10000 100000 10 100 1000 10000 NT50 (IU/mL) Anti-RBD IgG (BAU/mL) <18 (r=0.67, p<0.0001) 18—49 (r=0.69, p<0.0001) 50—64 (r=0.79, p65 (r=0.88, p<0.0001) C for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint Overall ChAdOx1 Vaccine 0.1 1 10 100 1000 10000 100000 NT50 (IU/mL) A Overall mRNA-1273 Vaccine 1 10 100 1000 10000Anti-RBD IgG (BAU/mL) D Overall mRNA-1273 Vaccine 0.1 1 10 100 1000 10000 100000 NT50 (IU/mL) B Overall ChAdOx1 Vaccine 1 10 100 1000 10000Anti-RBD IgG (BAU/mL) C for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted November 26, 2021. ; https://doi.org/10.1101/2021.11.24.21266812doi: medRxiv preprint