Seroprevalence of SARS-CoV-2 Infection in Cincinnati Ohio USA from August to December 2020

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the agent responsible for Coronavirus disease 2019 (COVID19) was first identified in Wuhan, China in late 2019 (1,2), which thereafter spread across the globe resulting in the current pandemic. In the United States of America (USA) more than 525,000 deaths have been attributed to this virus as of March 2021, with approximately 29 million total documented cases of infection, of which over 980,000 were in the State of Ohio (3). Current efforts to curtail and possibly end the pandemic are largely dependent on the vaccine program underway across the USA and the world, with the goal of achieving herd immunity and dramatic reductions in viral spreading (4,5). Another important consideration in achieving herd immunity and the eventual end to the current pandemic is quantifying background rates of previous infection within the population (4). In general, previously infected individuals are resistant to new infection for a period, and/or they appear to have a reduced severity of disease if re-infected (6). Thus, characterizing levels of previous infection within communities across the USA will inform progression toward herd immunity with optimized vaccination strategies.

SARS-CoV-2 contains a large surface facing glycoprotein called Spike (S, ∼190 kDa) that facilitates binding of the virus to the receptor angiotensin converting enzyme 2 (ACE2) on host cells, which after proteolytic cleavage of S mediates viral cellular involution and infection (7,8). The receptor-binding domain (RBD, 30 kDa) is part of the S protein that directly interacts with ACE2, and antibodies against the RBD region can mediate neutralization and protection from viral infection (7,8). The S protein is also a primary component of immunogenicity for the host response against the virus that can produce some degree of lasting immunity (7). Thus, it is not surprising that a primary strategy for vaccine development against SARS-CoV-2 involves ectopic expression of the S protein to generate host specific neutralizing antibodies, particularly those directed against the RBD region (7,8).

Rates of seropositivity for SARS-CoV-2 in voluntary blood donors in the Greater Cincinnati Metropolitan Area (GCMA) of Ohio USA were examined from August 13th - December 8th of 2020, just prior to the beginning of the nationwide vaccination program. A modified serological enzyme-linked immunosorbent assays (ELISA) assay developed by Krammer and colleagues (9,10) was implemented to examine 9,550 individuals of age rage 16-91 years old for SARS-CoV-2 S protein IgG antibodies to quantify rates of prior infection in the GCMA. All positives were assayed separately for concomitant IgG sero-reactivity against the RBD portion of the S protein. A cross-section of the positive ELISA data was further examined using the Luminex immunodetection platform for potential cross-reactivity with the 4 endemic human cold-causing coronaviruses (HCoV- 229E, -NL63, -OC43, and -HKU1) (11,12). Since some individuals donated blood more than once over this 5-month period, rates of decline in S protein antibody levels were documented, suggesting some potential reduction in immunity. Finally, rates of S protein seropositivity in the GCMA due to ethnicity, gender, and geographic subregions were examined.

Results

To document rates of previous SARS-CoV-2 infection in the GCMA leading up to the onset of the national vaccination program, blood samples from randomly presenting volunteer donors through the Hoxworth Blood Center were collected and analyzed for antibodies against the S protein by ELISA, and positives were also analyzed for RBD antibodies in a separate ELISA. Exactly 10514 samples were collected and processed from August 13th through December 8th of 2020, which represented 9550 unique donors and we determined that 802 were positive for S protein antibodies, for an overall prevalence of 8.40% (Table 1). The accuracy of this reported rate of past infection is dependent on how the laboratory ELISA was implemented and verified, which is detailed in the Methods section. The ELISA itself was based on an established protocol and set of biological reagents described by Krammer and colleagues (9,10), which was granted Emergency Use Authorization by the US Food and Drug Administration. However, a more robust protein production system based in expi-CHO cells was used to generate S and RBD proteins, which yielded greater amounts than observed using expi293 cells described by Krammer and colleagues (9,10) (Supplemental Figure 1). An improved purification protocol with quality control analytics was also instituted (Supplemental Figure 1) to ensure the purity of the preparations and that the S protein was in the trimeric configuration (Supplemental Figure 1). However, the ELISA-based reactivity of S protein antigen generated in expiCHO cells or expi293 cells was essentially identical (data not shown). The ELISA positivity cut-off was set as 3-standard deviations above background, calculated with serum samples from 2019 before the onset of the pandemic. The actual experimental value from the ELISA was 0.4039 for S protein IgG antibody and 0.4826 for RBD as optical density (OD) units, so that readings higher than these values were interpreted as positive (Table 2 and Figure 1). Importantly, this was a blinded study in which SARS-CoV-2 testing was performed post-hoc on de-identified specimens and did not modify or influence donor recruitment, eligibility or deferral for this donation or subsequent donations.

• Download figure
• Open in new tab

Figure 1. Histogram frequency of positive S protein ELISA OD values from 0.4039 to 3.000.

The 802 Spike positive samples were grouped in bins of 0.1 OD units with the initial bin starting at 0.4039, the minimum positive Spike OD value. The maximum Spike OD of all samples was 2.998.

All 802 of 9550 donors with antibodies against the S protein were also evaluated for antibodies against the RBD in separate ELISAs, which showed that 446 were positive for both, a rate of 55.61% (Table 3). However, because the RBD region is ∼85% smaller than the entire S protein, there are fewer total antibodies with specificity to RBD compared with S protein, explaining the reduced sensitivity of the RBD ELISA and likely why only ∼56% concordance was observed (Table 3). Indeed, a sub-analysis of the 802 positive donors for S protein based on ELISA intensity was performed, and the highest 1/3rd OD values (2.0-3.0) showed 97.3% positivity for RBD, while the lowest 1/3rd of OD values (0.4039-1.0) only showed 21.3% positivity for RBD (Table 3). Thus, those donors with the lowest levels of total S protein antibodies were less likely to have detectable RBD antibodies simply based on sensitivity issues.

The Luminex immunodetection platform was also employed to examine potential cross-reactivity with the endemic 4 human cold-causing coronaviruses that might influence S protein or RBD reactivity in the ELISAs, as well as to confirm the validity and sensitivity of the S protein ELISA (Table 4). We selected a group of 11 donor samples that were highly positive for S protein and RBD by ELISA for comparative analysis in the Luminex platform, which as expected showed a rate of 100% positivity for S, RBD and Nucleocapsid (N) protein from SARS-CoV2, but not S protein from MERS-CoV or SARS-CoV-1, which further validates the ELISA protocol (Table 4). Eleven highly characterized negative controls from the S protein ELISA were also examined and these same samples were 100% negative for S protein, RBD and N protein in the Luminex assay (Table 4). To examine the sensitivity of the S protein ELISA we also analyzed 57 donor samples that were RBD negative and relatively low in overall S protein reactivity, and in this group 67% confirmed in the Luminex assay for S protein, and 23% were positive for RBD, but only one sample was positive for N protein (Table 4). These results with the Luminex assay support the validity of the ELISA but also suggest that the ELISA is more sensitive than the Luminex platform in detecting S protein.

We also used the Luminex platform to assess correlation or cross-reactivity between donor antibody status for the S protein and positivity against the 4 human cold causing coronaviruses (HCoV-229E, -NL63, -OC43, and -HKU1). While many donors show antibodies against the 4 hCoVs in all three of our S protein ELISA groups (11 negatives, 11 high positives, and the 57 low positives), there was no correlation between antibodies against the 4-human cold-causing coronaviruses and antibody positivity or lack of positivity for SARS-CoV-2 S or RBD protein (Table 4). Thus, past infection status with the 4-human cold-causing coronaviruses did not impact S protein or RBD positivity.

Interestingly, of the 802 S protein ELISA positive donors, 108 donated blood or blood products at least 2 times from August to December 2020, and hence the study monitored maintenance or loss of antibody reactivity over time (Table 5). Analysis of these repeat donors showed 38 that had significant antibodies against S protein by ELISA on their first donation but that fell just below the assay threshold on the second donation, with an average time of 53 days (Table 5). In contrast, 24 donors maintained significant antibody reactivity and remained positive between the 2 donations, with an average time span of 69 days. However, of the 38 that lost positivity by the 2nd donation the initial composite ELISA OD value was 0.5960, while the group of 24 donors that maintained ELISA positivity between the 1st and 2nd donation had a much higher OD value of 1.47, which compares to a total average of 1.23 OD units across all 802 S protein positive donors (Table 5). Thus, the group of 38 repeat donors that lost their antibody positive status on the 2nd donation likely reflects the low starting point of antibody reactivity in conjunction with the known gradual loss of antibodies to SARS-CoV-2 over time (13), which now was below the sensitivity of the assay.

Sample collection from August 13th - December 8th of 2020 was then split in half, which showed a significant increase in rates of positivity from the first to the second periods (Table 6). Specifically, rates of S protein ELISA positivity were 7.56% across the GCMA from August 13th through October 10th (58 days), compared with 9.24% from October 10th through December 8th, 2020 (58 days). Rates of S protein reactivity were also analyzed by month, which showed a temporal increase, culminating in a value of 12.86% in the portion of December that was evaluated (Table 6 and Figure 2). Thus, the GCMA emerged into the national vaccination phase of the pandemic with a background level of ≥13% of individuals likely with some degree of protection from SARS-CoV-2 re-infection.

• Download figure
• Open in new tab

Figure 2. Percent positive S protein ELISA seroprevalence by month from August 13, 2020 to December 8, 2020.

Month by month breakdown of the percent positive rate for 9550 unique donors. In August, 76 of 924 donors were positive (8.225%); September, 215 of 2821 donors were positive (7.621%); October, 189 of 2842 donors were positive (6.650%); November, 224 of 2201 donors were positive (10.177%); and December, 98 of 762 donors were positive (12.861%). p-value was determined from the z score for each month compared to the August positive rate for significance. Only December was significant as compared to August (z=-3.114, p=0.0018 Two-tailed) while September, October and November were not significantly different from each other. The number of samples collected per month is shown in each bar.

Within the donor population dataset, we also evaluated age ranges for rates of prior infection. The 9550 donors spanned in age from 16 to 91 years, and we analyzed rates of ELISA S protein positivity based roughly on decade increments. Interestingly, the youngest group of donors from 16-30 years of age had the highest rates of antibodies against S protein compared with individuals in their 30s and 40s, while adults over 60 years of age had significantly lower rates (Table 7). This significant reduction in rates in donors over 60 may reflect greater adherence to social protective guidelines or due to reduced antibody levels in older age groups, such that they more quickly lose responsiveness in the S protein-based ELISA. Finally, gender association with rates of S protein antibodies was 8.52% in males versus 8.28% in females, which was not statistically different (Table 7).

Finally, we also analyzed geographic subregions within the GCMA for rates of S protein antibodies. Unfortunately, sampling was not large enough to examine rates based on individual zip codes, although statistical evaluation of the GCMA as larger subregions was possible, such as West versus East side, as split by Interstate 75 (I-75), and as rates in Ohio versus Kentucky, as split by the Ohio River (Table 8 and Figure 3). The data show a rate of 9.63% on the West side of Cincinnati versus 8.13% on the East side, while the Ohio portion of the GCMA was 8.79% versus 7.03% in the adjoining Kentucky region (Table 8 and Figure 3). Thus, the West side of Cincinnati had the highest rates of past SARS-CoV-2 infectivity, while the adjoining Kentucky region of the GCMA had the lowest.

• Download figure
• Open in new tab

Figure 3. Seroprevalence of protein antibodies in the Greater Cincinnati Metropolitan Area.

Map image courtesy of Google Maps (https://www.google.com/maps). Greater Cincinnati Metropolitan Area (GCMA) as defined by the United States Census Bureau (https://data.census.gov/cedsci/map?q=All%20counties%20in%20Ohio&g=310M500US17140&tid=ACSDP5Y2019.DP02&layer=VT_2019_310_M5_PY_D1&cid=DP02_0093PE&palette=Teal&break=5&classification=Natural%20Breaks&mode=customize) and reported in the 2010 AGE, RACIAL, GENDER AND MARITAL STRUCTURE OF GREATER CINCINNATI. The Cincinnati Metropolitan Statistical Area (MSA) includes: Butler, Brown, Clermont, Hamilton, and Warren Counties in Ohio; Boone, Bracken, Campbell, Gallatin, Grant, Kenton, and Pendleton Counties in Kentucky; and Dearborn, Franklin, and Ohio Counties in Indiana. The indicated data show percent of seropositivity for Spike antibodies shown by the indicated regions.

Discussion

To our knowledge the current study is the first to report rates of SARS-CoV-2 seroprevalence in the GCMA immediately preceding the national vaccination program. The total cumulative rate as of December 2020 was ∼13%, which should help achieve herd immunity in the GMCA more rapidly in conjunction with the national vaccination program (14,15). With respect to immunity after viral infection, 150,000 previously infected individuals in Ohio and Florida were tracked from March 2020 -August 2020 and shown to be largely resistant to subsequent re-infection, like protection achieved with vaccination (6). In the current study here, the highest rates of past infection were observed in individuals under 30 years of age, and more generally on the West side of the GCMA compared to the East side and adjoining regions of Kentucky. Data trends failed to reveal a difference in background levels of past infectivity based on ethnicity in the GCMA as analyzed in Caucasians, African Americans, Hispanics, and Asians, although the total sampling pool of the later 3 ethnic groups was too low to achieve statistical certainty.

The detection by seroprevalence of past infection of SARS-CoV-2 is dependent on the quantitative measures of the S protein-based IgG-dependent ELISA. Thus, exact details surrounding implementation and associated quality control measures are critical in achieving accuracy of past purported infectivity rates within the population. A modified ELISA protocol from Mount Sinai Icahn School of Medicine in New York City (9,10) was implemented, which was given an Emergency Use Authorization (EUA) by the US Food and Drug Administration (FDA) in April of 2020 (16). Additional quality control measures included the use of 60 serum samples obtained prior to the onset of the pandemic as true negatives in generating a background value for the S protein ELISA. The entire ELISA process was also automated with robotic liquid handling and plate reading systems. Hence, the values reported here are accurate in representing the rates of past infectivity in the donor samples evaluated within the GCMA. However, it is likely that the ELISA testing platform used here will miss individuals in whom the levels of antibody have dropped below the level of detection, as previously reported (17). Thus, not all individuals infected in the early months of the pandemic will maintain a positive reading in this ELISA and as such, the composite rate calculated from the S protein ELISA likely underestimates the true rates of past infectivity in the GCMA, which is likely several percentage points higher (see below).

Individuals who present to donate blood or other blood products are not a true cross-sectional representation of a metropolitan area. Indeed, such individuals are pre-screened for communicable diseases or behaviors that are high risk for attaining such diseases. The ethnicity of the GCMA blood donor volunteers, especially during the current pandemic, is 90% or greater Caucasian and hence, under-represented in other ethnicities such as African Americans and Hispanics (18). Moreover, blood donors also tend to be healthy and have lower prevalence of acquired chronic health conditions. However, given that this study was a post-hoc analysis of de-identified specimens, no selection bias based on donor suspicions of past infection were involved. Indeed, as discussed below, the rates of seropositivity observed here are in line with rates calculated by PCR analysis across the entire state of Ohio since the inception of the pandemic (3).

In the past year countless seroprevalence studies have been published or uploaded to preprint servers from across the USA, although very few have thus far extended to December 2020. A few of these past studies are particularly relevant and interesting to consider in relation to the current analysis. One such study examined 252,882 blood donors over 24 centers across the USA from the months of June and July 2020 (19). Vassallo et al., utilized the Ortho VITROS Anti-SARS-CoV-2 total immunoglobulin assay for the S1 region of the S protein for IgG, IgA and IgM (19), which was different from the full-length S protein ELISA detecting IgG serum levels that was produced in house and employed here. Vassallo et al., reported a rate of 1.83% in June and 2.26% in July across their entire USA sample population. Within these data were results from Chicago, which showed a rate of 2.76% in June and 3.34% in July (19). By comparison, another seroprevalence study from the Chicago area that analyzed 1545 solicited volunteers, showed a seroprevalence of 19.8% from June 24 through September 6, 2020 (20). This later study utilized the same S protein ELISA protocol from Krammer and colleagues (9,10) that we also used to investigate the GCMA, although their volunteer sample collection was based on advertising for serology evaluation (20). Hence, select community-based recruitment variables and the technical aspects of the immunodetection platform could underlie the widely disparate results discussed here. However, based on PCR measured SARS-CoV-2 molecular detection from the beginning of the pandemic until March of 2021, there were approximately 980,000 Ohioans infected with the virus. This equates to a rate of 8.23% of the state’s population, which is generally consistent the data showing a rate of ∼12% seropositivity as of December of 2020 in the GCMA. More interestingly, extrapolation of these data 3 additional months to the time this report was uploaded in March 2021, the rate of past infectivity in the GCMA is likely ∼16%, and as stated above this approximation almost certainly slightly under-estimates the true rate due to the known gradual decline in antibody levels and the associated drop below the threshold of the S protein-based ELISA (13,17). Given all these factors and the fact that blood donors likely represent a slightly healthier cross-section of the GCMA, an overall estimated rate of past infection of roughly 20% seems reasonable for the present day in mid-March 2021.

Another study examined 177,919 seemingly random adult blood samples from across 50 States that spanned from July 27 through September 24, which reported a range of values from just under 1% to over 20%. This study used 3 different automated clinical laboratory immunodetection platforms for either S or N protein. Within their data set samples collected from August 24 - September 24, 2020 overlap with part of our collection time, and in Ohio they report values of 2.8 to 5.0% (21). This value is also consistent with another more limited seroprevalence study with 727 samples from across the entire state of Ohio in July 2020, which incorporated mathematical modeling to predict a rate of 7% past viral infectivity (22), a value that is close to our actual data from the adjacent month of August in the GCMA.

While there are a multitude of seroprevalence studies in the literature from across the USA, only a handful appeared directly relevant to the interpretation of this study. These past studies generally support our conclusions and suggest that the ELISA implemented here was rigorous and properly calibrated, and that the donor subgroup used generated a reasonable approximation of the GCMA and likely the entire state of Ohio. The data in this study establish a rate of ∼13% past SARS-CoV-2 infectivity within the GMCA by the end of 2020, and extrapolation to the present day (March of 2021) approximates a rate of 16% past infectivity, and likely even as high as 20% if depreciation in blood antibody levels over time is considered. This knowledge can impact the deployment of the vaccination program to most rapidly achieve herd immunity (15). For example, previously infected individuals might only need 1 vaccine dosage for full protection compared with a naïve individual who requires 2 vaccinations (with Moderna and Pfizer vaccines). Indeed, previously infected and recovered individuals produce a strong immunologic reaction after a single dosage of the Pfizer vaccine that is comparable to the standard 2 dose routine in naïve individuals (23); and using this information and associated strategy would augment the relative supply of the vaccine in attempting to achieve herd immunity more rapidly.

Methods

Competing interest

None

Data Sharing

Requests datasets or protocols used in this study should be directed to the corresponding author and will be made available to investigators upon reasonable request.

Sources of Funding

This study was supported the Howard Hughes Medical Institute (to J.D.M). J.D.M was also supported by an internal grant from Cincinnati Children Research Foundation to conduct these studies on COVID-19. Contributions to this work by P.S. were financially assisted by The John Hauck Foundation, Fifth Third Bank.

Ethics and Reporting

Blood samples were collected from volunteer donors presenting to the Hoxworth Blood Center following United States FDA regulations and American Association of Blood Banks (AABB) guidelines with a signed standard donor consent form. Specimens were de-identified and as such, the University of Cincinnati Institutional Review Board (FWA #: 000003152) ruled that these blood donor samples and their analysis as constituting non-human research for the proposed study of SARS-CoV-2 serological responsiveness.