key: cord-273253-rgqvdzna
authors: Skowronski, D. M.; Sekirov, I.; Sabaiduc, S.; Zou, M.; Morshed, M.; Lawrence, D.; Smolina, K.; Ahmed, M. A.; Galanis, E.; Fraser, M.; Singal, M.; Naus, M.; Patrick, D. M.; Kaweski, S.; Mill, C.; Reyes, R. C.; Kelly, M. T.; Levett, P. N.; Petric, M.; Henry, B.; Krajden, M.
title: Low SARS-CoV-2 sero-prevalence based on anonymized residual sero-survey before and after first wave measures in British Columbia, Canada, March-May 2020
date: 2020-07-15
journal: nan
DOI: 10.1101/2020.07.13.20153148
sha: 
doc_id: 273253
cord_uid: rgqvdzna

Background: The province of British Columbia (BC) has been recognized for successful SARS-CoV-2 control, with surveillance data showing amongst the lowest case and death rates in Canada. We estimate sero-prevalence for two periods flanking the start (March) and end (May) of first-wave mitigation measures in BC. Methods: Serial cross-sectional sampling was conducted using anonymized residual sera obtained from an outpatient laboratory network, including children and adults in the Greater Vancouver Area (population ~3 million) where community attack rates were expected to be highest. Screening used two chemiluminescent immuno-assays for spike (S1) and nucleocapsid antibodies. Samples sero-positive on either screening assay were assessed by a third assay targeting the S1 receptor binding domain plus a neutralization assay. Age-standardized sero-prevalence estimates were based on dual-assay positivity. The May sero-prevalence estimate was extrapolated to the source population to assess surveillance under-ascertainment, quantified as the ratio of estimated infections versus reported cases. Results: Serum collection dates spanned March 5-13 and May 15-27, 2020. In March, two of 869 specimens were dual-assay positive, with age-standardized sero-prevalence of 0.28% (95%CI=0.03-0.95). Neither specimen had detectable neutralizing antibodies. In May, four of 885 specimens were dual-assay positive, with age-standardized sero-prevalence of 0.55% (95%CI=0.15-1.37%). All four specimens had detectable neutralizing antibodies. We estimate ~8 times more infections than reported cases. Conclusions: Less than 1% of British Columbians had been infected with SARS-CoV-2 when first-wave mitigation measures were relaxed in May 2020. Our findings indicate successful suppression of community transmission in BC, but also substantial residual susceptibility. Further sero-survey snapshots are planned as the pandemic unfolds.

British Columbia (BC) is the third-most populous and westernmost province of Canada, with more direct flights from mainland China to the Vancouver International Airport than any other airport in North America or Europe. In that context, BC health officials have long prepared for the potential introduction of emerging respiratory pathogens, with prior success in the early detection and containment of imported human cases of severe acute respiratory syndrome coronavirus (SARS-CoV-1) in 2003 and avian influenza H7N9 in 2015 [1, 2] . Thereafter, BC also reported the first importation of COVID-19 from outside China (Iran) on February 20; the first super-spreading event on March 6; and the first long-term care facility outbreak (and associated death) on March 9 [3] .

Despite these firsts, BC has been recognized for its successful control of COVID-19 during the winter-spring 2020. Early success was attributed in part to timely alerting; sustained media messaging through a single health official; rapid development and deployment of diagnostic testing; and restriction of staff movement between long-term care facilities. As elsewhere in Canada, on March 12 the province recommended against non-essential travel that Skowronski Version: July 13, 2020 5 by age group across successive waves of the influenza A(H1N1)pdm09 pandemic [12] [13] [14] [15] , and has contributed findings to meta-analyses of the World Health Organization (WHO) [15] .

In February 2020, BCCDC adapted this protocol for SARS-CoV-2 assessment. We present baseline findings for two cross-sectional serum collections (or snapshots) flanking the start (March) and end (May) of first wave activity and population-level control measures. The goal of these serial snapshots was to establish baseline and early pandemic sero-prevalence for future attack rate comparison; to estimate cumulative incidence, residual susceptibility and the extent to which community transmission was suppressed; and to assess surveillance underascertainment across the winter-spring 2020 period in BC.

As shown in Figure 1 , the first snapshot captured specimens collected March 5-13 and the second snapshot captured specimens collected May 15-27, 2020 . For each snapshot, at least 500μL of anonymized residual sera were obtained from patients attending one of ~80 diagnostic service centres of the only outpatient laboratory network in the Lower Mainland, BC (LifeLabs).

Specimens were collected as shown geographically distributed across this region in Figure 3 [18] . Consistent with the WHO Unity protocol [19] , sera were obtained from 100 patients within All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.13.20153148 doi: medRxiv preprint Serological testing Sera were screened by two chemiluminescent immuno-assays (CLIA). Specimens seropositive on either screening assay were subjected to a third CLIA as well as neutralization assay.

Screening assays, both currently approved by Health Canada [20] , included:

1. Detection of total antibody (IgA, IgG and IgM) to recombinant spike (S1) protein using the Vitros XT 7600 analyzer (Ortho-Clinical Diagnostics, Rochester, New York) [21] . The patient sample signal was divided by the calibrator signal, with resultant signal to cut-off (S/C) ratios of <1.00 and ≥1.00 considered negative or positive, respectively. Public Health England reports overall sensitivity of 85% (95% confidence interval (CI)=76. 5 at ≥14 days and 100% (95.1-100) at ≥17 days after symptom onset [24] . All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.13.20153148 doi: medRxiv preprint Further testing of sero-positive specimens included:

3. Detection of total (IgG, IgM) antibodies to the S1 receptor-binding domain (S1-RBD) by the ADVIA Centaur XPT system (Siemens Healthineers, Erlangen, Germany) [21] . Samples with S/C ratios of <1.0 and ≥1.0 were considered negative or positive, respectively. As of July 1, 2020, this assay had not received Health Canada approval but has FDA Emergency Use Authorization [20, 21] . For a similar Siemens assay based upon the Atellica-IM analyzer, Public Health England reports overall sensitivity of 86.0% (95%CI=77.6-92.1), 89.4% (95%CI=80. .0) at ≥14 days and 92.4% (95%CI=84.2-97.2) at ≥21 days after symptom onset, with specificity of 100% (95%CI=99.1-100) [22] . 4 . Standard virus neutralization assay as described by Zakhartchouk et al. [25] . Each serum specimen was heat inactivated at 56°C for 30min and duplicate serial 2-fold dilutions from 1:8 to 1:4096 were each incubated with 100 TCID50 of SARS-CoV-2 for 2h, then added to monolayers of Vero-E6 cells. Seed virus was obtained from the National Microbiology Laboratory designated as SARS-CoV-2/2020/0044. Monolayers were examined after 72h for characteristic cytopathogenic effect (CPE). The inverse of the highest serum dilution to inhibit CPE was deemed the antibody titre. Geometric mean titres (GMT) of duplicate tests are presented.

The number and proportion of specimens testing positive by each screening assay are presented by age group and snapshot with 95%CIs derived by exact method. We explored the proportion sero-positive on either screening assay and further taking into account sensitivity of 85% independently applied to each screening assay while assuming perfect specificity All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. [4] , summarized as ~2000 reported cases for this analysis.

There were 1759 sera included in both snapshots; 241 specimens were set aside because of low sample volume of which 163 (68%) were children 0-4 years and 39 (16%) were children 5-9 years. For the March snapshot, 870 and 869 sera were screened for antibodies to S1 and nucleocapsid, respectively; for the May snapshot 889 and 885, respectively, were tested ( Table   1) . At each snapshot, median age (45 and 45 years, respectively) and the proportion female (51% All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. no March specimens were positive on both screening assays and only when additionally assessed on the third CLIA, did 2 specimens show dual-assay positivity, to S1 and S1-RBD. Both were identified among adults 40-59 years of age (2/199; 1.01%; 95%CI=0.12-3.58). None of the seropositive specimens in March had detectable neutralizing antibodies ( Table 2) .

Two of 869 sera were dual-assay positive at the March snapshot giving a crude seroprevalence of 0.23% (95%CI=0.03-0.83) and age-standardized sero-prevalence of 0.28% (95%CI=0.03-0.95).

Among specimens included from the May snapshot, 6/889 (0.67%; 95%CI=0. 25-1.46) were S1 sero-positive and 7/885 (0.79%; 95%CI=0.32-1.62) were nucleocapsid positive ( Table   1 ). Nine of 885 (1.02%; 95%CI=0.47-1.92%) specimens were positive on either assay ( Table 2) , with age-standardized proportion of 1.03% (95%CI=0.45-2.00). Accounting for imperfect test All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. Four specimens were positive on both S1 and nucleocapsid screening assays and when additionally tested the same four specimens were also S1-RBD and neutralization assay positive;

whereas, the 5 specimens positive on just one or the other screening assay were neither S1-RBD nor neutralization assay positive ( Table 2 ). The four consistently positive sera included 1/195 children <20-years-old (0.51%; 95%CI=0.01-2.82) and 3/297 adults 30-59-years-old (1.01%; 95%CI=0.21-2.92). The S1, nucleocapsid and S1-RBD signals on these four specimens were each high at ≥127, ≥2.63 and ≥4.84, respectively. The two sera with the highest S1 signals (283 and 325, respectively) and S1-RBD signals (>10.0) also had the highest neutralizing antibody titres (GMTs of 64 and 256, respectively) although one patient with strong S1, nucleocapsid, and S1-RBD signals (162, 7.03, >10.0) had only low-level neutralization titre (GMT of 8).

Four of 885 sera were dual-assay positive at the May snapshot giving a crude seroprevalence of 0.45% (95%CI=0.12-1.15) and age-standardized sero-prevalence of 0.55% (95%CI=0.15-1.37%). Applying this 0.55% age-standardized sero-prevalence point estimate to the Lower Mainland source population (~3 million), we estimate about 16,500 infections, or about 8 times higher than the ~2000 cases reported.

We report the first SARS-CoV-2 sero-prevalence estimates from Canada. Our estimates are based on anonymized residual sera obtained from an outpatient laboratory network in an area of British Columbia where community attack rates were anticipated to be highest. Serial sampling captured two periods flanking the start and end of the first pandemic wave and the All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.13.20153148 doi: medRxiv preprint implementation of population measures to mitigate its impact. Using a convenient and efficient sero-survey protocol previously used by the BCCDC for emerging pathogen risk assessment [12] [13] [14] [15] [16] [17] , we estimate SARS-CoV-2 cumulative incidence (community-wide infection rates) remained below 1% throughout the winter-spring 2020 period in BC.

For efficiency in serological testing we used commercially-available, high-throughput CLIAs and for improved reliability applied an algorithm involving multiple viral targets. This included two screening assays, one targeting the S1 and another the nucleocapsid, with a third CLIA applied to positive specimens that targeted the S1-RBD [20, 21] . All positive specimens were further assessed by gold-standard neutralization assay [25] . We estimated sero-prevalence based on dual-assay positivity and report cumulative incidence of 0.28% by the start of first wave population-level measures in March. Had we interpreted sero-positivity based on either screening assay, the age-standardized proportion would have just reached 1.0% and taking into account test sensitivity as low as 85%, sero-positivity might have reached 1.5%. In fact, just two of 869 March specimens were sero-positive to both S1 and S1-RBD, but none were sero-positive on both S1 and nucleocapsid screening assays and none had detectable neutralizing antibodies. In that regard, the only two dually-positive specimens in March may have been false-positives and community-level attack rates could be even lower than we estimate by that snapshot. Potential cross-reactivity between SARS-CoV-2 and other coronaviruses may explain inconsistent findings in March and will be explored in follow-up investigations [27, 28] . Alternatively, SARS-CoV-2 cases, notably mild or asymptomatic infections, may not have mounted a serological response or sustained neutralizing titres [29] [30] [31] . A similar inability to detect neutralizing antibodies among specimens positive on enzyme immunoassay was also recently reported from Hong Kong [32] . All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.13.20153148 doi: medRxiv preprint By the end of the first pandemic wave and easing of restrictions in May, we estimate sero-prevalence and cumulative incidence remained low at 0.55%. Had we interpreted seropositivity based on either screening assay, the age-standardized proportion would have still been just 1.0%, and accounting for a test sensitivity of 85%, sero-positivity might have reached 1.2%.

Unlike the earlier March snapshot, several May specimens showed S1 plus nucleocapsid positivity, and the same four specimens were also S1-RBD and neutralization assay positive, confirming true infections. Overall, just 20 of 1754 screened sera overall were positive on any assay (1.14%;95%CI=0.70-1.76), a low proportion indicating that if non-specific reactivity or cross-reactivity with other endemic coronaviruses is an issue [27, 28] , it is not a prominent one.

Furthermore, all sera that were positive for S1-RBD were also positive for S1 antibodies; and all sera that were positive for S1 and that had a neutralizing antibody response were also nucleocapsid sero-positive. In combination, these findings highlight that in lieu of labourintensive screening by gold-standard neutralization assay, an efficient algorithm for seroprevalence estimation might rely upon orthogonal (two-tiered) high-throughput CLIA testing [33] . In particular, assays targeting S1 or S1-RBD, plus nucleocapsid, should provide reliable and specific detection of SARS-CoV-2 sero-positivity. Such orthogonal approach, however, requires further validation.

Multiple other countries have reported sero-prevalence estimates [24, 32, [34] [35] [36] [37] [38] [39] [40] , but study populations, sampling approach, time frame, as well as antibody assays and algorithms must be taken into account when comparing findings. Survey approaches based on blood donation, invitation or other voluntary participation may over-estimate sero-prevalence due to selfselection bias. Few other studies have capitalized on residual patient specimens to try to mitigate that bias [32, 37, 39] , and in that regard, our findings are best compared to another such study All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. [39] . Note that the latter estimate using residual patient specimens is substantially lower than estimated by invitation to adult grocery store patrons in New York City between April 19-28 (22.7%;95%CI=21.5-24.0) [40] . For the Puget Sound Region, US authors estimated about 11 times as many infections compared to reported cases [39] , a comparable order of magnitude to our estimate of about 8-times as many infections.

There are limitations to this analysis. Delay in generating an antibody response and uncertainty in the duration of antibody persistence, particularly among mild or asymptomatic cases and for neutralizing titres [27, [29] [30] [31] , may lead to under-estimation of sero-prevalence.

Specimens were anonymized without accompanying clinical detail such as symptom history or testing indication. We restricted to outpatient specimens and show both snapshots to be generally representative of the source population for age and sex. However, residual clinical specimens are more likely to come from people with underlying comorbidity and may differ from the rest of the population in their exposure risk, immune response and healthcare seeking. We assume sampling within the Lower Mainland provides an upper range of community sero-prevalence for BC; however, this does not preclude discrete pockets experiencing higher attack rates. Given manifold uncertainties, our ratio of estimated infections versus reported cases is an imprecise approximation, meant only to gauge the likely order of magnitude of surveillance underascertainment. In addition to other considerations, reported surveillance tallies may include cases that were imported or accrued within care facilities or other settings under-represented in our community-based sero-sampling. Finally, with such low sero-prevalence, including just one All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.13.20153148 doi: medRxiv preprint infected child in 375 pediatric sera tested, we were limited in our ability to conduct subset comparisons.

In conclusion, we present a convenient, efficient and reliable approach for serial population-based sero-prevalence monitoring based on anonymized residual sampling (mitigating self-selection bias) and orthogonal high-throughput testing (improving predictive value). We estimate <1% of British Columbians were infected with SARS-CoV-2 by the time first wave restrictions were relaxed in May. These sero-prevalence findings reinforce other surveillance data indicating successful suppression of SARS-CoV-2 transmission throughout the winter-spring 2020 period in BC. This success, however, constitutes a double-edged sword, further highlighting substantial residual susceptibility. Our sero-prevalence protocol is readily amenable to comparison across serial snapshots and these are planned at relevant intervals as the pandemic unfolds.

All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. 

DMS is Principal Investigator on grants from the Michael Smith Foundation for Health Research in support of this work. MK received grants/contracts paid to his institution from Roche, Hologic and Siemens. No other authors have conflicts of interest to disclose. All rights reserved. No reuse allowed without permission.

(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 this preprint this version posted July 15, 2020. . https://doi.org/10.1101/2020.07.13.20153148 doi: medRxiv preprint Table 1 . Results of SARS-CoV-2 sero-survey screening by chemiluminescent assay for antibodies to spike (S1) and nucleocapsid proteins, by age group, March and May 2020 snapshots, Lower Mainland, BC, Canada Table 2 . Specimens sero-positive on either spike S1 or nucleocapsid screening assay, including additional S1 receptor binding domain and neutralization assay findings, March and May 2020

Coordinated response to SARS

Avian Influenza A(H7N9) Virus Infection in 2 Travelers Returning from China to Canada

COVID-19 (Novel Coronavirus)

British Columbia Centre for Disease Control. COVID-19 Surveillance Reports

Annual Estimates of the Resident Population for the United States, Regions, States

Coronavirus disease (COVID-19): Outbreak update

Provisional death counts and excess mortality

Estimation of Excess Deaths Associated With the COVID-19 Pandemic in the United States

Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (SARS-CoV-2)

Prevalence of Asymptomatic SARS-CoV-2 Infection: A Narrative Review

Immuno-epidemiologic correlates of pandemic H1N1 surveillance observations: higher antibody and lower cell-mediated immune responses with advanced age

Prevalence of seroprotection against the pandemic (H1N1) virus after the 2009 pandemic

Pre-and post-pandemic estimates of 2009 pandemic influenza A(H1N1) seroprotection to inform surveillance-based incidence, by age, during the 2013-2014 epidemic in Canada

Estimating age-specific cumulative incidence for the 2009 influenza pandemic: a meta-analysis of A(H1N1)pdm09 serological studies from 19 countries

SARS-CoV-2 Sero-survey

H3N2v and other influenza epidemic risk based on age-specific estimates of sero-protection and contact network interactions

Cross-reactive and vaccine-induced antibody to an emerging swine-origin variant of influenza A virus subtype H3N2 (H3N2v)

Government of British Columbia. Population projections

Population-based age-stratified seroepidemiological investigation protocol for COVID-19 infection

Authorized medical devices for uses related to COVID-19: List of authorized testing devices

United States Food and Drug Administration

COVID-19: laboratory evaluations of serological assays

Performance Characteristics of Four High-Throughput Immunoassays for Detection of IgG Antibodies against SARS-CoV-2

Performance Characteristics of the Abbott Architect SARS-CoV-2 IgG Assay and Seroprevalence in

Augmentation of immune responses to SARS coronavirus by a combination of DNA and whole killed virus vaccines

Efficient interval estimation for age-adjusted cancer rates

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

Severe Acute Respiratory Syndrome Coronavirus 2-Specific Antibody Responses in Coronavirus Disease Patients

No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity

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

SARS-CoV-2 Infections and Serologic Responses from a Sample of

Antibody responses to SARS-CoV-2 at 8 weeks postinfection in asymptomatic patients

Seroprevalence of SARS-CoV-2 in Hong Kong and in residents evacuated from Hubei province, China: a multicohort study

Interim Guidelines for COVID-19 Antibody Testing

Seroprevalence of IgG antibodies against SARS coronavirus 2 in Belgium: a prospective cross-sectional study of residual samples

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

No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity

Seroprevalence of anti-SARS-CoV-2 IgG antibodies in Geneva, Switzerland (SEROCoV-POP): a population-based study

Estimation of seroprevalence of novel coronavirus disease (COVID-19) using preserved serum at an outpatient setting in Kobe, Japan: A crosssectional study

Seroprevalence of immunoglobulin M and G antibodies against SARS-CoV-2 in China

Seroprevalence of Antibodies to SARS-CoV-2 in Six Sites in the United States

Cumulative incidence and diagnosis of SARS-CoV-2 infection in New York

a Of 200 sera collected from children 0-4 and 5-9 years of age, sufficient sera were available for this initial serological analysis for 25/101 and 71/99, respectively. Most included children < 10 years old were therefore 5-9 years of age (i.e. 71/96; 74%). b One specimen was tested for antibodies to S1 but not nucleocapsid because below the minimum sample volume established by our protocol c Of 200 sera collected from children 0-4 and 5-9 years of age, sufficient sera were available for this initial serological analysis for 13/100 and 89/100, respectively. Most children < 10 years old were therefore 5-9 years of age (i.e. 89/102; 87%). d Two specimens were assessed for antibodies to S1 but not nucleocapsid because below the minimum sample volume established by our protocol e Four specimens in total from the May snapshot were assessed for antibodies to S1 but not nucleocapsid because below the minimum sample volume established by our protocol S/C=Signal to cut-off ratio; GMT=geometric mean titre; M=male; F=female Grey shading indicates specimens sero-positive on individual assays or in combination included in primary seroprevalence estimationd (-) sero-negative; (+) sero-positive a The patient sample signal was divided by the calibrator signal, with resulted signal to cut-off (S/C) ratios of <1.00 and ≥1.00 considered non-reactive (-) or reactive (+), respectively. b The patient sample signal was divided by the calibrator signal, with resulted signal to cut-off (S/C) ratios of <1.40 and ≥1.40 considered negative (-) or positive (+), respectively. c The patient sample signal was divided by the calibrator signal, with resulted signal to cut-off (S/C) ratios of <1.0 and ≥1.0 (+) considered non-reactive (-) or reactive (+), respectively. Note: as of July 1, 2020, this assay had not received Health Canada approval and was only conducted on specimens that were + on either of the preceding two assays. d Primary sero-prevalence estimation required positivity on at least two assays (dual assay positivity). e Not tested because below the minimum sample volume established by our protocol All rights reserved. No reuse allowed without permission.