Seropositivity in blood donors and pregnant women during the first year of SARS‐CoV‐2 transmission in Stockholm, Sweden

Abstract Background In Sweden, social restrictions to contain SARS‐CoV‐2 have primarily relied upon voluntary adherence to a set of recommendations. Strict lockdowns have not been enforced, potentially affecting viral dissemination. To understand the levels of past SARS‐CoV‐2 infection in the Stockholm population before the start of mass vaccinations, healthy blood donors and pregnant women (n = 5,100) were sampled at random between 14 March 2020 and 28 February 2021. Methods In this cross‐sectional prospective study, otherwise‐healthy blood donors (n = 2,600) and pregnant women (n = 2,500) were sampled for consecutive weeks (at four intervals) throughout the study period. Sera from all participants and a cohort of historical (negative) controls (n = 595) were screened for IgG responses against stabilized trimers of the SARS‐CoV‐2 spike (S) glycoprotein and the smaller receptor‐binding domain (RBD). As a complement to standard analytical approaches, a probabilistic (cut‐off independent) Bayesian framework that assigns likelihood of past infection was used to analyse data over time. Setting Healthy participant samples were randomly selected from their respective pools through Karolinska University Hospital. The study was carried out in accordance with Swedish Ethical Review Authority: registration number 2020–01807. Participants No participants were symptomatic at sampling, and blood donors were all over the age of 18. No additional metadata were available from the participants. Results Blood donors and pregnant women showed a similar seroprevalence. After a steep rise at the start of the pandemic, the seroprevalence trajectory increased steadily in approach to the winter second wave of infections, approaching 15% of all individuals surveyed by 13 December 2020. By the end of February 2021, 19% of the population tested seropositive. Notably, 96% of seropositive healthy donors screened (n = 56) developed neutralizing antibody responses at titres comparable to or higher than those observed in clinical trials of SARS‐CoV‐2 spike mRNA vaccination, supporting that mild infection engenders a competent B‐cell response. Conclusions These data indicate that in the first year since the start of community transmission, seropositivity levels in metropolitan in Stockholm had reached approximately one in five persons, providing important baseline seroprevalence information prior to the start of vaccination.

Background. In Sweden, social restrictions to contain SARS-CoV-2 have primarily relied upon voluntary adherence to a set of recommendations. Strict lockdowns have not been enforced, potentially affecting viral dissemination. To understand the levels of past SARS-CoV-2 infection in the Stockholm population before the start of mass vaccinations, healthy blood donors and pregnant women (n = 5,100) were sampled at random between 14 March 2020 and 28 February 2021.
Methods. In this cross-sectional prospective study, otherwise-healthy blood donors (n = 2,600) and pregnant women (n = 2,500) were sampled for consecutive weeks (at four intervals) throughout the study period. Sera from all participants and a cohort of historical (negative) controls (n = 595) were screened for IgG responses against stabilized trimers of the SARS-CoV-2 spike (S) glycoprotein and the smaller receptor-binding domain (RBD). As a complement to standard analytical approaches, a probabilistic (cut-off independent) Bayesian framework that assigns likelihood of past infection was used to analyse data over time.
Setting. Healthy participant samples were randomly selected from their respective pools through Karolinska University Hospital. The study was carried out in accordance with Swedish Ethical Review Authority: registration number 2020-01807.
Participants. No participants were symptomatic at sampling, and blood donors were all over the age of 18. No additional metadata were available from the participants.
Results. Blood donors and pregnant women showed a similar seroprevalence. After a steep rise at the start of the pandemic, the seroprevalence trajectory increased steadily in approach to the winter second wave of infections, approaching 15% of all individuals surveyed by 13 December 2020. By the end of February 2021, 19% of the population tested seropositive. Notably, 96% of seropositive healthy donors screened (n = 56) developed neutralizing antibody responses at titres comparable to or higher than those observed in clinical trials of SARS-CoV-2 spike mRNA vaccination, supporting that mild infection engenders a competent B-cell response.

Introduction
Densely populated areassuch as the Stockholm region of 2.37 million people (975,000 within city limits)have facilitated the spread of SARS-CoV-2. Evidence suggests that transmission can be curtailed by imposing restrictions on leisure and business activities, as well as by mask usage and contact tracing [1][2][3][4]. In contrast to most comparable countries, Sweden has favoured a strategy in which individuals are encouraged to adhere to a set of basic public health recommendations, while society has remained largely open. Only recentlyfaced with a winter second wavedid the government opt for earlier closing hours for some businesses. The country has reported a significantly higher burden on the public healthcare system to date than its Scandinavian neighbours (Fig. 1a), and there are many reasons for why this might be the case.
Serological enumeration of past infections is critical for estimating viral spread and understanding characteristics of the adaptive immune response to an emerging pathogen [5]. Antibody testing is also important for optimal planning of vaccination campaigns (especially when doses are limited) and could help demonstrate 'immunity' on official documentation. However, as illustrated by the current pandemic, not all antibody tests are of equal sensitivity and specificity (S&S), and closer scrutiny of assays is needed to improve public health measures and our understanding of COVID-19 [6].
To monitor seropositivity in the Stockholm region, we developed highly sensitive and specific SARS-CoV-2 antibody tests based on the native-like spike glycoprotein [7] (alongside a diagnostic clinical laboratory responsible for monitoring seropositivity during the pandemic [8]), as well as novel cut-off-independent statistical methods, and applied them to random, healthy adults in the region throughout 2020 and early 2021. Such approaches, which improve upon strictly thresholding a continuous variable, are critical for accurate seropositivity estimates at individual and population levels. This is especially important in the case of SARS-CoV-2, as asymptomatic/mild infections generate antibody responses of varying titres that can be difficult to classify [8][9][10][11], and the knowledge that individual titres generally wane from peak levels over time [12][13][14][15].
We chose to survey anti-S IgG responses as these are the best indicators of past SARS-CoV-2 infection at the population level (i.e. present in >91% of PCR-positive cases) [16,17]. Spike harbours the ACE2 receptor-binding domain (RBD) and is the major target of the neutralizing antibody response, central to vaccine efforts. For example, antinucleocapsid (N) antibodies were not detectable in a subset of S-seropositive individuals [8,[18][19][20]. For both S, N and RBD, the levels of antibodies in circulation were shown to increase with worsening disease severity [9,11,21], as also reported for SARS-CoV and MERS [22,23].
Blood donors and pregnant women, studied here, are important clinical groups, and pregnant women require further study in relation to infectious disease [24,25]. Moreover, both groups represent good and accessible proxies for adult population health, being generally working age, mobile members of societywithout being enriched for individuals at especially high-risk of SARS-CoV-2 infection, such as healthcare workers or public transportation employees, where seroprevalence may be higher [26]. Similarly, seroprevalence may be higher in children and lower in the elderlywith complementary studies much needed.
All healthy individuals in this study (n = 5,100) were over the age of 18 and symptom-free at sample collection. For ethical reasons, their ages and genders are not revealed, nor is it known whether any participant previously tested positive for the virus. Thus, while we here provide an unbiased assessment of past SARS-CoV-2 infection in two important groups in Stockholm in response to natural infection and before the start of mass adult vaccinations, future studies in other cohorts are needed to generate more integrated data to guide public health strategies.   No metadata, such as age or sex information, were available for the samples in the study. Blood donors must be over 50 kg and over the age of 18, while no upper age limit is established in Stockholm. Pregnant women were sampled as part of routine screening for infectious diseases during the first trimester of pregnancy. Whether a study participant had previously tested positive for SARS-CoV-2 PCR is unknown. Blood donors are required to be healthy for a minimum of two weeks before a donation and all participants reported symptom-free at sample collection. We cannot exclude that a very small number of blood donor samples from 2021 were from individuals who had received a COVID-19 vaccine. Healthcare workers in the region were receiving first doses at the time of writing, along with elderly persons; however, mass vaccinations of adults (including pregnant women) had not yet begun [27,28].

SARS-CoV-2 antigen generation
The plasmid for expression of the SARS-CoV-2 prefusion-stabilized spike ectodomain with a C- Anti-SARS-CoV-2 S and RBD IgG were detectable at up to 1:20,000 serum dilution using this assay [8], and all study samples were here run at 1:100 dilution.

Probabilistic seroprevalence estimations
Prior to analysis, each sample OD was logtransformed and standardized by dividing by the mean OD of 'no sample control' wells on that plate or other plates run on the same dayto reduce batch variation. This resulted in more similar distributions for 2019 blood donor samples with 2020 blood donors and pregnant volunteers. Our Bayesian approach is presented in detail in Christian and Murrell [30]. Briefly, we used a logistic regression over anti-RBD and anti-S training data (n = 595 historical blood donor controls and n = 138 SARS-CoV-2 PCR+ individuals across the clinical spectrumall of whom developed anti-S IgG and 97% did so against the RBD) to model the relationship between the ELISA measurements and the probability that a sample is antibody-positive. We adjusted for the training data class proportions and used these adjusted probabilities to inform the seroprevalence estimates for each time point. Given that the population seroprevalence cannot increase dramatically from one week to the next, we constructed a prior over seroprevalence trajectories using a transformed Gaussian Process and combined this with the individual class-balance adjusted infection probabilities for each donor to infer the posterior distribution over seroprevalence trajectories. We compared our Bayesian approach to the output of more established probabilistic algorithms (specifically, an ensemble learner from support vector machines (SVM) and linear discriminant analysis (LDA) -SVM-LDA) that we have previously developed for ELISA measurements.
The sensitivity and specificity (S&S) of our IgG ELISA assays were as follows: • To compute confidence intervals for S&S, we dichotomized predictions of seropositivity at prob > 0.5 or <= 0.5 and computed average sensitivity, specificity and 95% confidence intervals for each fold in the cross-validation via Wilson's method before averaging over all folds. The Bayesian model does not make assignments of seropositivity to individual samples, but rather integrates over the uncertainty in relationship between infection and OD450, as well as the uncertainty due to sampling, to provide population-level seroprevalence estimates. Thus, S&S cannot be calculated for this approach.

Results
Blood donor (BD) and pregnant women (PW) serum samples (n = 100 of each per sampling week) were selected at random from their respective pools and the IgG response against SARS-CoV-2 S glycoprotein trimers and the smaller RBD subunit was measured in all sera using established ELISA assays extensively validated using sera from confirmed infections [8] (Fig. 1b,c); anti-S and RBD responses are highly correlated, with lower titres generally observed for the smaller RBD.
Test samples were run alongside historical (SARS-CoV-2-negative) control sera (n = 595 blood donors from spring 2019) over the course of the study. Seropositivity for S and RBD according to conventional 3 or 6 standard deviations (SD) (from the mean of negative control sera) assay cut-offs is presented in Table 1 and Fig. 2a. Notably, seropositivity was not significantly different between blood donors and pregnant women over the study period (Fig. 2b,c).
However, the many measurements between the 3 and 6 SD cut-offs for both or a single antigen (Fig. 1b,c) pose a problem when assigning case to low values, uncertainty that can significantly skew seroprevalence estimates and is undesirable at the individual level. Therefore, to provide an accurate seropositivity estimate for our population and to model seropositivity changes over time, we developed a cut-off-independent, probabilistic Bayesian framework that models the log odds that a sample is antibody-positive based on anti-S and anti-RBD IgG responses in training data, in this case SARS-CoV-2 PCR+ COVID-19 patients (n = 136) across the clinical spectrum (including asymptomatic/mild cases) [8].
Using this more quantitative approach that better considers the wide range of responses present in the population and shares information between  sampling weeks, we found seropositivity to increase sharply at the start of the pandemic (Fig. 2d). By the time, the COVID-19 death rate in the country was at low levels during August, following the first wave, the seroprevalence trajectory increased at a steady, but slower rate in approach to the winter second wave (Fig. 2d)in agreement with continued viral spread in the Stockholm population during a summer recess in cases and mortality, and consistent with persistent individual antibody responses over a 9-month period [31].  (Table 1 and Table S1). The seroprevalence level again increased more rapidly between mid-December and the end of February 2021, consistent with the winter peak in mortality and infections, and reached 19.2% (95% Bayesian CI [15.1-24.4]) of the population by last sampling. Thus, approximately one in five healthy adults in Stockholm showed evidence of past natural SARS-CoV-2 infection before the start of mass adult vaccinations in the country. An equal-weighted probabilistic SVM-LDA learner that we had previously optimized for ELISA measurements showed highly consistent results (Table S2).  Fig. 2e,f), with titres comparable to those engendered by recently approved COVID-19 mRNA vaccines that were shown to be protective in clinical phase 3 trials [32,33]. These observations support that asymptomatic/mild infection generates an antibody response that provides a degree of protection upon re-exposure, although interindividual heterogeneity, environmental factors and different SARS-CoV-2 variants will play a role in individual outcomes.

Discussion
To characterize immunological responses to SARS-CoV-2 and safeguard public health, it is critical to monitor the level of population immunity after natural infection, especially in settings with different public health measuresallowing for concurrent and retroactive evaluation of different strategies. As Sweden has taken a unique public health approach to mitigate the effects of the virus, data from the country provide an important contrast to comparable settings.
Serology is amenable to studies of large cohorts and remains the gold standard for determining previous exposure to pathogens. Several studies have highlighted the protective role antibodies play in controlling SARS-CoV-2 infection in humans [34,35] and animal models [36,37], and potent neutralizing and convergent antibodies were rapidly isolated from infected donor samples [38][39][40]. Indeed, mounting evidence suggests that most persons (>91%) previously infected with SARS-CoV-2 develop virus-specific antibodies, including following mild infections [8,16,17,31]. Critical ongoing and future research is required to determine the duration of serological responses and Bcell memory in those infected, as well as following vaccination.
Early research has shown that the majority of mildly infected individuals maintain detectable virus-specific B-cell responses for at least 8-10 months [17,31]. Together with the steadily increasing seropositivity observed in this study and elsewhere, such as New York City [41], these findings help allay early concerns that immunity waned in the short term, although individuals with mild infection can have antibody titres close to the assay boundary. Data suggest that~10% of SARS-CoV-2 seropositive individuals can lose detectable titres 10 months post-PCR test [12,15,17], although this proportion remains to be reported in larger and different cohorts. Therefore, the levels of past infection in this cohort may be slightly higher than we report. Anti-SARS-CoV antibodies have been reported to persist for 3 years following more severe infection [42,43], while detectable responses to seasonal coronaviruses generally wane within 1-2 years and immunity is affected by rhythms in their circulation [44][45][46]. Importantly, as the activated B-cell pool and antibody response contracts following viral clearance (and declines in function with age [47,48]), the identification of low but persistent antibody responses (that may be highly effective) remains a challenge using conventional serological methods.
Notably, we observed virus-neutralizing titres in our cohorts to be comparable to those engendered by the first mRNA vaccines of Pfizer/BioNTech [33] and Moderna [34], supporting that natural infection generates neutralizing immunity to the infectious strain, as do studies of healthcare workers [35] and elderly care home residents and staff [49], in whom re-infection rates were substantially reduced. The potential endemicity and continued evolution of SARS-CoV-2 may alter the nature of protective immunity (e.g. possible re-infection with a different strain), and knowledge of strain-specific antibody responses should be used to optimally coordinate vaccine interventions and inform public health measures.
Epidemiologically, our data indicate that approximately one year since virus emergence in Sweden, approximately one in five adults in the categories studied here (blood donors and pregnant women) in Stockholm had been previously infected by the virus. The situation in Stockholm is like that observed in New York City, USA, where despite high rates of infection and mortality [50], herd immunity was not attained within six months of the outbreak, with seropositivity reaching~20% of the general adult population [41]. Despite an estimated >60% IgG seroprevalence after the first wave [51], Manaus, Brazil, was also not spared a severe second wave of infections [52]. As SARS-CoV-2 variants continue to emerge [53], seroprevalence measurements reported from many countries highlight a continued need to curtail viral transmission through social restrictions and effective vaccination programmes.
Our study supports that Stockholm has suffered a higher number of infections (following the first wave) than most European locations studied [54][55][56][57][58][59], in agreement with high per capita mortality. However, direct comparisons between sites are complicated by differences in the demographics of the study subjects as well as the S&S of the assay used for antibody detectiony [60]. Longitudinal studies over longer timescales and in different cohorts are urgently needed for improved understanding of population immune responses after SARS-CoV-2 infection and COVID-19 vaccination.

Conclusions
These data highlight a relatively high SARS-CoV-2 seroprevalence in Stockholm, Sweden, following one year of community transmission. Population serology informs disease epidemiology and public health approaches and can help target vaccines where they are most efficacious.