Antibody Levels in Healthcare Workers During the COVID-19 Gamma Variant Outbreak in Manaus, Brazil

We analyzed samples from a prospective, observational cohort study of HCW in 2 hospitals in Manaus, Amazonas State, Brazil. Enrollment and follow-up of HCW occurred from 31 March to 31 May 2021. HCWs who consented to participation had a nasal swab collected weekly regardless of symptoms for 4 consecutive weeks. Nasal swabs were tested for SARS-CoV-2 by real-time reverse transcription polymerase chain reaction (rRT-PCR). Trained study staff collected data on demographics, medical history, previous history of COVID-19, symptoms, healthcare and community exposures, and COVID-19 vaccination history using participants' COVID-19 vaccination cards or the state immunization registry. A blood specimen was obtained at enrollment (visit 1) and at the end of follow-up period (visit 4) unless the HCW was found to be infected. If infected, a blood specimen was collected at 14 and 28 days after symptom onset or date of positive specimen, if asymptomatic. HCWs with 2 documented doses of COVID-19 vaccination who tested rRT-PCR positive for SARS-CoV-2 at least 14 days after the second dose were classified as postvaccination infections. Previous infection was based on self-report of having a positive diagnosis either by laboratory test or from a physician.

Nasal swabs were placed in viral transport medium, stored at 2–4°C, and transported in coolers within 48 hours after collection to Laboratório Central do Amazonas for rRT-PCR for SARS-CoV-2 using AllplexTM 2019-nCoV Assay [18]. rRT-PCR results were expressed as the cycle threshold (Ct) for the gene encoding the nucleocapsid (NC) protein as previously described [19]. All samples with sufficiently low Ct values (Ct <30) were defined as a case and underwent whole-genome sequencing using COVIDSeq library preparation kit (Illumina) at Fiocruz-Amazonas, as previously described [20].

Whole blood (4 mL) was collected in ethylenediaminetetraacetic acid tubes and the plasma from these samples was tested using previously validated in-house enzyme-linked immunosorbent assay (ELISA) for SARS-CoV-2 NC immunoglobulin G (IgG) antibody titers [6], and commercially available ELISA for SARS-CoV-2 S1 IgG titers, EUROIMMUN Anti–SARS-CoV2 S1 Curve ELISA [21]. For the in-house ELISA, an anti-SARS-CoV-2 NC IgG antibody reactivity index (RI) was expressed as the ratio between optical density of the patient sample and the negative control. All samples with RI ≥1.5 were considered positive. For the commercial anti–SARS-CoV-2 S1 IgG ELISA, we used the manufacturer's cutoff of >35.2 binding antibody units per milliliter (BAU/mL) to define reactivity. After assessing serologic titer distribution in our cohort, we further classified reactive samples as low titer if the BAU/mL was from 35.3–260.03 and high titer if the BAU/mL was >260.03. This classification is consistent with data from ChAdOx1 nCoV-19 clinical trial that found a titer of ≥264 BAU/mL to be correlated with protection against SARS-CoV-2 infection [22]. Finally, we performed the SARS-CoV-2 NeutraLISA (Euroimmun) assay, a surrogate test used to determine the neutralizing capacity of anti–SARS-CoV-2 antibodies by detecting inhibition of protein interaction between the angiotensin-converting enzyme 2 receptor and the viral receptor binding domain (RBD) in the S1 domain of the spike protein. The results were expressed as a percent neutralization [23].

Vaccinated HCWs with or without postvaccination infection were tested for neutralizing antibodies against the Gamma and Delta variants of SARS-CoV-2. PRNT90 was performed as previously described [24] with modifications. Briefly, the plasma samples were heat inactivated (56°C for 1 hour) and 2-fold serial dilutions were performed in maintenance medium 199 (M199). Next, diluted serum was incubated with the SARS-CoV-2 variants for 1 hour at 37°C and 5% CO₂. For the serum-virus complexes, we used an equal amount of virus suspension containing 100 plaque forming units. Then, antibody-virus mixture was added to 24-well plates containing a confluent monolayer of Vero CCL-81 cells (ATCC, Manassas, VA, USA) and incubated for 1 hour at 37°C and 5% CO₂. Subsequently, 1.5 mL of the overlay medium (medium 199 containing 1.5% carboxymethylcellulose and 5% fetal bovine serum) was added to each well and incubated for 4 days at 37°C and 5% CO₂. The plates were fixed using 10% formaldehyde for 4 hours and then stained with 1% Crystal Violet for 2 hours. Plaque counting was performed under an optical microscope (Zeiss, Oberkochen, Germany). After counting the plaques, the neutralizing antibody titers were determined based on the PRNT90, considering the sera that reduced the number of plaques by 90% compared to the average of the virus control wells to be positive.

Five indicators of immune response were analyzed: IgG levels against Spike protein, IgG levels against Nucleocapsid protein, NeutraLISA percent neutralization, PRNT90 levels against SARS-CoV-2 Gamma-variant, and PRNT90 levels against SARS-CoV-2 Delta-variant. IgG levels were available and analyzed for the full cohort (N = 771 participants), while NeutraLISA and PRNT90 levels were available for a subset (N = 190).

Immune responses were plotted by time since last vaccination dose, and nonparametric locally estimated scatterplot smoothing were estimated to describe longitudinal trends in immune response decay. This analysis was performed for all participants and stratified by age category.

Pairwise correlation plots between the 5 immune response variables were created and pairwise Pearson correlation coefficients were calculated.

HCWs were grouped into 4 vaccine status categories: full, partial, unvaccinated, and unknown. Full vaccination was defined as having received the full course of doses approved for that vaccine (ie, 2 doses for CoronaVac, 2 doses for ChAdOx1) at least 14 days before sample collection. Partial vaccination was having received at least 1 but not the 2 doses of vaccine at least 14 days before sample collection.

The HCWs who were SARS-CoV-2 positive by PCR during follow-up were further analyzed. Trends in their IgG and PRNT90 levels were plotted for each HCW individually. A Gaussian model was used to assess predictors of immune response based on baseline risk factors.

Baseline IgG levels against Spike protein, Neutralisa percent neutralization, and PRNT90 levels against SARS-CoV-2 Gamma variant and Delta variant were compared between participants who remained PCR– throughout the entire study versus the 16 that had a positive PCR test during the study. Statistical significance for this comparison was measured with the Kruskal–Wallis test. All analyses were performed in R Version 4.1.2 (R Foundation, Vienna, Austria).

This project was reviewed and approved by the institutional review board from Universidade de São Paulo (reference #4.611.867) and by the Brazilian National Ethics Committee (reference #4.858.909). This activity was reviewed by the Centers for Disease Control and Prevention (CDC), classified as nonresearch, and conducted in compliance with applicable federal laws and CDC policies. Written informed consent was obtained from all participants before enrollment.

Baseline S1 IgG levels at enrollment and end of follow-up are described in Figure 1B. During follow-up, 53.6% (392/731) of the participants remained below 260 BAU/mL anti-SARS-CoV-2 S1 IgG antibody titers (Figure 1B). Anti–SARS-CoV-2 S1 IgG antibody titers declined with time since the most recent vaccine dose for IgG levels against the Spike and Nucleocapsid proteins (Figure 2, Supplementary Figure 1), for NeutraLISA neutralization percentage, and for PRNT90 titers against Gamma and Delta variants (Figure 3). The steepest declines for S1 IgG and neutralizing antibody responses occurred in the first 50 days following their most recent vaccine dose, with a plateau between 50 and 100 days postdose (Figure 3). Responses were generally higher in individuals with reported previous COVID-19 (Figure 2B, Supplementary Table 1). Amongst individuals with previous COVID-19, an infection in the previous 90 days resulted in the highest levels of immune response (Figure 2C). Trends in the decay of the immune response did not vary by age except for the age group >60 years (Supplementary Figure 2).

The 5 measures of immune response we evaluated had statistically significant positive correlations between each other (). The highest correlation was between PRNT90 titers to Gamma and Delta variants, followed by PRNT90 titers to Gamma variant and Spike IgG levels. PRNT90 titers to Delta were largely not detectable for low to moderate IgG responses, only appearing for individuals with high IgG or NeutraLISA responses (bottom row of). Supplementary Figure 3 Supplementary Figure 3

Median PRNT90 titers against Delta were 4 times lower (interquartile range [IQR]: 2) compared to median PRNT90 titers against Gamma in the same sample (). Of the samples with non-zero PRNT90 titers against Gamma, 37.9% (131/346) had undetectable PRNT90 activity against Delta. Conversely, 100.0% (215/215) of samples with observed, non-zero PRNT90 titers against Delta also had detectable PRNT titers against Gamma. Supplementary Figure 4

A total of 16 participants had positive SARS-CoV-2 rRT-PCR+ results at 1 point during recruitment or follow-up, with 12 of these being fully vaccinated, 1 partially vaccinated, and 3 unvaccinated (). When considering IgG antibody response to spike protein, only 3 showed increase following infection. When considering PRNT titers against the Gamma variant, 9 of the 16 showed increases following infection (). Supplementary Figure 5 Supplementary Figure 5

The only immune response level at baseline associated with testing rRT-PCR + during the follow-up was PRNT baseline titers (Figure 4). Baseline titers were lower in those who tested rRT-PCR + (median Log10 titers [IQR]: Gamma, 1.5 [3]; Delta, 0 [0.25]) during follow-up versus those who were rRT-PCR– throughout the study period (median Log10 titers [IQR]: Gamma, 3 [2]; Delta, 1 [2]) (Figures 4C and 4D).