key: cord-0784324-kl0e1esr
authors: Prete, C. A.; Buss, L. F.; Abrahim, C. M. M.; Salomon, T.; Crispim, M. A. E.; Oikawa, M. K.; Buccheri, R.; Grebe, E.; da Costa, A. G.; Fraiji, N. A.; Carvalho, M. d. P. S. S.; Alexander, N.; Faria, N. R.; Dye, C.; Nascimento, V. H.; Busch, M. P.; Sabino, E. C.
title: Reinfection by the SARS-CoV-2 P.1 variant in blood donors in Manaus, Brazil
date: 2021-05-12
journal: nan
DOI: 10.1101/2021.05.10.21256644
sha: a5175a17a424cd82132e645b5d67859275f64ccd
doc_id: 784324
cord_uid: kl0e1esr

The city of Manaus, north Brazil, was stricken by a severe epidemic of SARS-Cov-2 in March 2020, reaching a seroprevalence of 76% by October 2020. Nevertheless, in late November an abrupt increase in hospitalizations and deaths hit Manaus, causing higher number of deaths compared to the first epidemic wave. It has been hypothesized that virus lineages circulating in the second wave, namely the P.1 variant of concern first detected in early December in Manaus, could be better at evading immunity generated in response to previous infection with other lineages. In order to estimate the reinfection rate during the resurgence of SARS-CoV-2 in Manaus, we tested serial samples from 238 unvaccinated repeat blood donors using a SARS-CoV-2 anti-N IgG chemiluminescence microparticle assay. Blood donors were divided into six groups that reflected the inferred sequence of infection and reinfection with non-P.1 and P.1 variants. We assumed that reinfections induce a recrudescence (or boosting) of plasma anti-N IgG antibody levels, yielding a V-shaped time series of antibody reactivity levels. We infer that 16.9% (95% CI [9.48%, 28.5%]) of all presumed P.1 infections that were observed in 2021 were reinfections. If we also include cases of probable or possible reinfections (defined by considering the time period when the antibody levels are expected to grow after recovery and the range of half-lives for antibody waning after seroconversion), these percentages increase respectively to 25.8% (95% CI [16.7%, 37.4%]), and 31.0% (95% CI [21.4%, 42.5%]). Our data suggest that reinfection due to P.1 is common and more frequent than what has been detected by traditional epidemiologic, molecular and genomic surveillance of clinical cases.

Approximately 76% of Manaus' inhabitants had been infected with SARS-CoV-2 eight months after the first reported case in March 2020 1 . Nevertheless, a second epidemic wave occurred in the city, coinciding with the emergence of a new SARS-CoV-2 Variant of Concern (VOC) in November 2020 denoted P.1, corresponding to 87% of all infections in January 2021 2 .

Mutations that are associated with immune escape and could increase the risk of reinfection have been postulated to explain the resurgence COVID-19 in Manaus 2,3 . To address this question, we retrieved and tested serial samples from 3,655 repeat donors from Manaus. From these, we selected all unvaccinated donors with three or more donations, which included at least one during the first epidemic wave (before July 1 st 2020) and at least one in January-March 2021, and excluded two donors that had their first positive donation in November or December 2020, when it is not possible to determine if the infection was caused by P.1, leading to 238 donors included in this study.

The samples were tested using a SARS-CoV-2 anti-N IgG chemiluminescence microparticle assay (CIMA, Abbott Park, IL, USA). The reactivity on this assay consistently wanes during convalescence 1 and we hypothesized that reinfection would induce a recrudescence (or "boosting") of plasma anti-N IgG antibody levels, yielding a V-shaped time series of antibody reactivity levels. We partitioned the 238 repeat donors into six groups that reflect the inferred sequence of infection and reinfection with non-P.1 and P.1 variants (Figure 1 ). See Supplemental Appendix for a detailed description of the classification criteria.

Of all the 59 presumed P.1 infections that were observed in 2021, we infer that 10 (16.9%, 95% CI [9.48%, 28.5%]) were reinfections. If probable reinfections are also included, these percentages increase to 25 The main limitation of this study is that donors were not sampled frequently enough to robustly detect cases of reinfection, leading to the possible existence of undetected cases of reinfection. We attempted to resolve this issue by classifying the degree of evidence and identifying probable and possible reinfections. Further, repeat negative donors may not represent truly unexposed individuals, since not all PCR+ individuals produce antibodies to nucleocapsid proteins and because sparse sampling may have resulted in missing the positive interval. Finally, blood donors are biased towards asymptomatic and mild infections; therefore, our rates of reinfection cannot be extrapolated to persons who had more severe primary infection.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 12, 2021.

Despite these caveats, our data suggest that reinfection due to P.1 is common and more frequent than has been detected by traditional epidemiologic, molecular and genomic surveillance of clinical cases 4, 5 . This is because few infected persons are tested by PCR in Brazil, thus PCRpositive patients have a small chance of having another positive PCR test, even if the reinfection rate is substantial. These results reinforce concerns over the risk of reinfection particularly as variants continue to evolve and demonstrate that repeat blood donor serosurveillance is valuable for documenting rates and correlates of reinfection. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Reference 8

. CC-BY-NC-ND 4.0 International license It is made available under a 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 May 12, 2021. ; https://doi.org/10.1101/2021.05.10.21256644 doi: medRxiv preprint

We divided blood donors into 6 non-overlapping groups. Supplemental Figure 1 shows a flowchart describing the procedure used to classify donors, and Supplemental Figure 2 shows an illustration of a case of reinfection that falls into different groups depending on the sequence of dates of sample collection. From all repeat blood donors, we selected only donors with three or more donations because it is not possible to infer reinfection based on two time points. We also required donors to have one positive donation between March 1 st , 2020 and June 30 th , 2020, and one positive donation after January 1 st , 2021. The objective of this requirement is to avoid selecting donors that had their first sample collected many months after the date of infection, since they may have already seroreverted when their first sample was collected.

To define the groups, we assume that all positive cases in 2021 are due to P.1 because of the high prevalence of P.1 in early January 1 . We excluded 31 vaccinated donors and two donors that had their first positive result in November and December 2020, when it was not possible to determine if the infection was caused by P.1 due to its low prevalence at that time, leading to 238 selected donors. With this exclusion, no infections observed in 2020 are due to P.1 because P.1 had an insignificant prevalence before November 2020 1 . The definition of the groups also depends on two predefined parameters Δ min and max used to define the expected behavior of nonreinfected individuals. There parameters are estimated based on donations that occurred before the incidence of P.1 became significant (i.e., donations up to and including October 2020).

The definition of probable and possible reinfections depends on the maximum half-life of NC antibodies on the CIMA assay ( max ) and the minimum interval between donations (Δ min ). Donors that tested positive in 2021 and whose S/C curve decayed with half-lives higher than all half-lives measured before November 2020 were treated as possible reinfections because they may have had unobserved (due to sparse sampling) V-shaped curves. The objective of defining Δ min is to avoid misclassifying donors as reinfected when samples were collected during the seroconversion periodthat is, we consider that Δ min is much greater than the period of seroconversion. Before estimating these parameters, we added to all donors an artificial negative donation with CIMA result 0.01 S/C in February 28, 2020, before the beginning of the epidemic in Manaus. This is because SARS-CoV-2 had not yet been introduced to the population, which was presumably completely immunologically naïve at that time.

Let ( i , Δ i ) be the number of donors that have at least one pair of successive positive results before November 2020 separated by an interval Δ ≥ Δ and decaying with half-life ≥ i . The function ( , Δ ) represents the number of possible reinfections observed in 2020. We first estimate Δ min as the smallest Δ > 0 such that there exists a ∈ [0,365] such that ( , Δ ) = 0. After estimating Δ min , max is assigned as the smallest ∈ [0,365] such that ( , Δ min ) = 0. Using this approach, we obtain Δ min = 87 days and T max = 287 days. It is worth noting that . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Given the values of max and Δ min , six groups of donors are defined as:

Donors that never had a positive donation. It is not possible to say that all persistently seronegative donors were not infected, since some infected donors may have had already seroreverted at the date of sample collection, or not seroconverted at all. b. Donors with three consecutive rising positive results, the last being in 2021. Since the minimum interval between successive donations is 60 days for men and 90 days for women in Brazil, donors with three consecutive rising positive results would apparently be seroconverting for more than 120 days > Δ min , a possibility that we rule out due to the definition of Δ min . Donors following this rule have likely had an unobserved S/C decay after the second rising result, but seroconverted again after being reinfected. (E) Probable reinfection by P. 1 Donors with two consecutive rising positive results, the last being in 2021, separated by an interval Δ ≥ Δ min . We hypothesize that donors following this rule have had an unobserved antibody decline after the first positive sample, and then seroconverted again after being reinfected. A minimum interval between donations is required to avoid misclassifying donors sampled during the seroconversion period as probable reinfections.

Donors with two consecutive positive results, the last being in 2021, separated by an interval Δ ≥ Δ min and decaying with half-life ≥ max . Donors following this rule may have had an S/C decay with half-life compatible to the half-lives observed in 2020, but may have had an S/C boosting due to reinfection, yielding an apparent half-life higher than max . The constraint Δ ≥ Δ min is important to avoid overestimating the half-life if the individual is still seroconverting after the first positive sample. Some donors following this rule may not be cases of reinfections, since donors with chronic infection may present a very slow antibody decay rate.

We define the CIMA test to be positive if the measured signal-to-cutoff (S/C ratio) is higher or equal to 0.49. This is the lowest value of range defined by the manufacturer ( Even though the assay has high sensitivity and specificity, it produces results that are subject to measurement error, which results in variation in S/C that does not reflect a biological change, but is simply variation within the limit of precision of the test. If this variation is not small, sequential donations may have a V-shaped curve even if reinfection has not occurred, leading to an overestimation of the reinfection rate. To assess the amount of measurement error, we tested 200 samples in replicate from blood donors that donated in February 2021 in São Paulo.

Supplemental Figure 2 shows the measured S/C for the first and the second test of each sample. The absolute deviation of each pair of measured S/C had a median of 0.00 and a 95% confidence interval of [0.00, 0.09]. If only positive results were considered, the median deviation increases to 0.02% (95% CI [0.00, 0.16]), and the relative deviation obtained by dividing the absolute deviation by the first result has median 1.21% (95% CI [0.00%, 7.3%]) for positive results.

Therefore, the assay employed in this study yields results with a small amount of measurement error. For this reason, a sequence of serial samples is unlikely to be misclassified as a case of reinfection due to measurement noise. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 12, 2021. ; https://doi.org/10.1101/2021.05.10.21256644 doi: medRxiv preprint

Three-quarters attack rate of SARS-CoV-2 in the Brazilian Amazon during a largely unmitigated epidemic

Genomics and epidemiology of the P.1 SARS-CoV-2 lineage in Manaus

Resurgence of COVID-19 in Manaus, Brazil, despite high seroprevalence

Risk of SARS-CoV-2 reinfection after natural infection

Assessment of protection against reinfection with SARS-CoV-2 among 4 million PCR-tested individuals in Denmark in 2020: a population-level observational study

Genomics and epidemiology of the P.1 SARS-CoV-2 lineage in Manaus

Three-quarters attack rate of SARS-CoV-2 in the Brazilian Amazon during a largely unmitigated epidemic

This work was supported by the Itaú Unibanco "Todos pela Saúde" program and by a Medical