Highlights
- •COco is a prospective cohort study conducted in the Netherlands, 2020-2021.
- •Population of non-medical contact professions with regular re-exposure to SARS-CoV-2.
- •COco evaluates the dynamics of antibody levels following exposure to SARS-CoV-2.
- •Very low seroreversion rates were found, suggesting long-term natural immunity.
- •IgG antibody levels initially decreased, but remained detectable up to 12 months.
Abstract
Objective
To evaluate dynamics of antibody levels following exposure to SARS-CoV-2 during 12 months in Dutch non-vaccinated hairdressers and hospitality staff.
Methods
In this prospective cohort study, blood samples were collected every three months for one year, and analyzed using a qualitative total antibody ELISA and a quantitative IgG antibody ELISA. Participants filled out questionnaires, providing information on demographics, health and work. Differences in antibody levels were evaluated using Mann-Whitney U and Wilcoxon Signed Rank tests. Beta coefficients (B) and 95% confidence intervals (95%CI) were calculated using linear regression.
Results
Ninety-five of 497 participants (19.1%) had ≥1 seropositive measurement before their last visit using the qualitative ELISA. Only 2.1% (2/95) seroreverted during follow-up. Of the 95 participants, 82 (86.3%) tested IgG seropositive in the quantitative ELISA too. IgG antibody levels significantly decreased in the first months (p<0.01), but remained detectable up to 12 months in all participants. Higher age (B, 10-years increment: 24.6, 95%CI: 5.7-43.5) and higher BMI (B, 5kg/m² increment: 40.0, 95%CI: 2.9-77.2) were significantly associated with a higher peak of antibody levels.
Conclusions
In this cohort, SARS-CoV-2 antibodies persisted for up to one year after initial seropositivity, suggesting long-term natural immunity.
Keywords
INTRODUCTION
Since the start of the Coronavirus Disease 2019 (COVID-19) pandemic in early 2020, over 564 million confirmed cases and over 6 million deaths have been reported globally (July 22nd, 2022)(WHO, 2022a). In the Netherlands (total population: 17.6 million), 8.3 million inhabitants have officially been diagnosed with COVID-19 (WHO, 2022b). This number is an underestimation of the actual number of cases, since not all people were tested for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections. Hence, a significant percentage of the population has developed natural immunity against SARS-CoV-2 at some point during the pandemic (Chvatal-Medina et al., 2021).
Research on the natural immune response following exposure to SARS-CoV-2 could aid to better understanding the duration of protective immunity. Such information is crucial in assisting public health decisions, for example for estimating the effects of restricting social activities, but also in weighing the benefit of vaccinating more people worldwide versus providing extra doses to those having developed natural immunity.
In most studies examining natural immunity after COVID-19, the study population consisted of patients who had a severe infection usually requiring hospitalization, or healthcare workers (Chvatal-Medina et al., 2021, Post et al., 2020). However, such populations may not adequately represent the general population, while it is especially important for policymakers to have knowledge on the duration of protective immunity in the overall population.
A small number of studies investigated the immune response after natural SARS-CoV-2 infection in a generic population (Anand et al., 2021, Dan et al., 2021, Deisenhammer et al., 2021, Demonbreun et al., 2021, He et al., 2021, Kučinskaitė-Kodzė et al., 2021, Luo et al., 2021, Petersen et al., 2021). These studies report contrasting data regarding the duration of detectable antibody levels. Various factors may have influenced these seemingly contrasting findings, such as differences in assay used or study population. Many studies have a cross sectional design, while the dynamics of antibody development after infection can vary widely among individuals (Luo et al., 2021). Hence, longitudinal studies are needed to improve our understanding of the duration of natural immunity against SARS-CoV-2 in the general population.
Previously, we introduced COco, a Dutch cohort study evaluating antibodies against SARS-CoV-2 in 497 hairdressers and hospitality staff (Mioch et al., 2021). Antibodies were measured up to 12 months in non-vaccinated individuals. Here, we evaluated the dynamics of antibody levels in this healthy population of individuals who frequently have contact with other people at work, thereby being potentially exposed to the coronavirus, while not being trained to take measures to prevent infection, similar to the general population. We studied both the presence and quantity of antibodies over time, and tested whether baseline variables were associated with antibody peak levels and dynamics.
METHODS
Study design and population
COco is a prospective cohort study which evaluated SARS-CoV-2 antibodies in non-medical contact-based professions in the province of North-Brabant in the Netherlands. Its design, recruitment and population has been described previously (Mioch et al., 2021). Hairdressers and hospitality staff (n=497) were recruited in June/July 2020; individuals were followed up to one year during four visits or until vaccination. No participants were hospitalized for COVID-19. In the current study, we selected participants who tested seropositive before their last visit to analyze antibody titers over time.
Participation was voluntary after providing written informed consent.
Data collection and analyses
Blood samples were collected from each participant at four timepoints during approximately one year. During venipuncture, 3.5ml blood was drawn and serum was analyzed by the Microvida Laboratory for Medical Microbiology using the qualitative Wantai SARS-CoV-2 total antibody ELISA (Wantai Biological Pharmacy Enterprise Co., Ltd., Beijing, China) per the manufacturer's instructions (Mioch et al., 2021). The manufacturer defined an absorbance to cut-off ratio (A/C.O.) ≥1.1 as seropositive, A/C.O. 0.9-1.0 borderline seronegative, and A/C.O. <0.9 seronegative. Additionally, we divided seropositivity with an A/C.O. ≥10.0, 2.0-9.9, and 1.1-1.0 into strongly seropositive, seropositive, and weakly seropositive, respectively. As such, these test results were evaluated qualitatively and semi quantitively, similar to earlier studies (Nilsson et al., 2021).
For the additional quantitative analyses, we included all samples from all participants who had a seropositive (A/C.O. ≥1.1) or borderline seronegative (A/C.O. 0.9-1.0) sample in the qualitative ELISA at a measurement before their final visit. These samples allowed evaluation of antibody dynamics after an initial seropositive measurement. In addition, we randomly selected several participants who were seropositive at their final visit only to evaluate agreement between the two assays. The samples were analyzed by the National Institute for Public Health and the Environment (RIVM) using the Anti-SARS-CoV-2 QuantiVac ELISA (IgG) test (EUROIMMUN Medizinische Labordiagnostika AG, Lübeck, Germany). This is a quantitative ELISA determining the concentration of IgG antibodies against the S1 antigen (including Receptor-Binding Domain (RBD)) of SARS-CoV-2. It was performed on participants’ serum using the EUROIMMUN Analyzer I platform per the manufacturer's instructions. This ELISA has an estimated sensitivity and specificity of 90.3% and 99.8%, respectively (EUROIMMUN, 2020). Antibody levels ≥11 relative units per milliliter (RU/ml) were considered seropositive, ≥8-<11RU/ml borderline, and <8RU/ml seronegative. For in-depth analyses of IgG antibody dynamics, we selected all participants with ≥1 seropositive result before their final visit. Additionally, we considered inclusion of participants with ≥1 borderline test result before their final visit. Three participants had a borderline test result but no seropositive test result before the last visit. One participant was included for in-depth analyses, as the participant had a borderline test result followed by a seropositive test result during the last two visits. The other two participants with borderline test results were excluded from the analyses: these participants only had borderline and seronegative test results using the quantitative ELISA, while all four test results using the more sensitive qualitative ELISA were seropositive. Therefore, we assumed that the quantitative measurements were incorrect and therefore, we excluded these from the analyses.
To signify the development of IgG antibody levels, we calculated differences between participants’ highest and lowest antibody levels. When antibody levels had decreased by 50% or increased by 100%, this was labeled as decreasing or increasing antibody levels, respectively. Doubled antibody concentrations after an initial ≥50% decrease were considered fluctuating antibody levels. Other changes were labeled as stable antibody levels. To calculate rates (percentage in antibody level changes per three months), we subtracted the highest from the baseline antibody concentration and divided it by the time between those measurements if participants had increasing antibody levels; for those with decreasing antibody levels, the antibody level peak and antibody concentration at the last visit were used. Participants with fluctuating antibody levels were censored for rate calculations when antibodies increased.
In addition, three web-based questionnaires were collected. The baseline questionnaire, collecting information such as demographics, has been published previously (Mioch et al., 2021). After enrolment, participants answered weekly a follow-up questionnaire (supplementary questionnaire 1), which collected information on health and work in the past week. Once every three months, the weekly questionnaire was expanded with questions regarding SARS-CoV-2 transmission risk outside work (supplementary questionnaire 2). We analyzed data from the baseline questionnaire; the variables we used are described in more detail in the supplementary data. Furthermore, we analyzed several variables from the follow-up questionnaires: symptoms possibly related to COVID-19, positive PCR test result (yes/no), not shaking hands as a measure to prevent SARS-CoV-2 transmission (yes/no), being part of contact tracing (yes/no) and having had contact with someone who tested positive for SARS-CoV-2 (yes/no). Severity of COVID-19 related symptoms was divided into two categories, based on whether participants did or did not report fever.
Statistical analyses
Descriptive statistics and frequencies were used to analyze baseline characteristics. Baseline characteristics were compared using chi-square tests or Fisher's exact test (for cell frequency n≥5 or n<5, respectively) for dichotomous categorical variables, and Mann-Whitney U tests for numerical variables. The ordinal categorical variable education (low, middle and high) was analyzed using a Mann-Whitney U test. Seroreversion curves were constructed using the Kaplan-Meier method (Kaplan and Meier, 1958). Differences in antibody levels were evaluated using Mann-Whitney U tests for independent samples (e.g., comparing participants with and without fever) and Wilcoxon Signed Rank tests for dependent samples (differences in antibody levels within individuals at different visits). Non-parametric tests were used due to the non-normal distribution of antibody levels, as determined using Shapiro-Wilk tests. Uni- and bivariable linear regression models were used to calculate beta coefficients (B) and their respective 95% confidence intervals (95%CI) for variables associated with the IgG antibody level peak. Univariable logistic regression model was used to calculate odds ratios and their respective 95% confidence intervals for variables predicting a decrease in antibody levels over time. P<0.05 was considered statistically significant. Trend lines were drawn based on the mean antibody levels of all participants at 0, 3.25, 7.25 and 10.75 months. Analyses were conducted using SPSS Statistics 24.0. The survival curve and boxplot were plotted using R4.1.2; scatterplots including trend lines and 95% confidence intervals were constructed using Microsoft Excel.
RESULTS
In total, 497 individuals were included in the COco-study. In June/July 2020, 11.3% (56/497) of participants tested positive for SARS-CoV-2 antibodies using the qualitative total antibody ELISA. This percentage increased to 13.7% (62/454), 25.1% (103/410) and 32.0% (110/344) in subsequent measurements in September/October 2020, January 2021 and February-June 2021, respectively. In total, 95 participants (19.1%) had at least one seropositive measurement before their last follow-up visit. Six of 95 participants (6.3%) had a weak seropositive test result at their first seropositive measurement, while 69.4% (65/95) had a strong seropositive test result (Figure 1). Of the 65 strongly seropositive participants, all but one (98.5%) remained strongly seropositive at all follow-up visits. In the study population, only two participants seroreverted (Supplementary Figure 1); both initially had weak seropositivity (Figure 1). Antibody levels of the other four participants with initial weak seropositive test results increased at subsequent visits. All participants who had an A/C.O≥2.0 at their first seropositive measurement, kept having seropositive measurements A/C.O≥2.0 at follow-up visits (Figure 1). Hence, the qualitative total antibody ELISA suggested limited seroreversion.
Figure 1Flow chart depicting the percentage of participants with a seropositive test resultᵃ, per visit, as evaluated using the qualitative ELISA assay.
Of the 95 participants with a seropositive measurement before their last follow-up visit in the qualitative ELISA, 82 participants were SARS-CoV-2 IgG seropositive before their last follow-up visit in the QuantiVac IgG ELISA. Baseline characteristics of the 82 participants are summarized in Table 1. The majority (72.0%) were women, had a middle level education (56.1%) and worked in the hospitality industry (61.0%). The median age was 39 years (range 18-68 years). Most participants were born in the Netherlands (97.6%), had no chronic disease (77.8%), did not smoke (82.9%) and used alcohol (90.2%, of whom 56.8% on average ≤1 alcohol unit per day). Median self-reported BMI was 25kg/m² (range 16-38kg/m²).
Table 1Baseline characteristics of participants with at least one seropositive SARS-CoV-2 IgG test result before the last follow-up visit.
| Total group (N=82) | Decreasing antibody levels (N=38) | Non-Decreasing antibody levels (N=44) | P-value | ||
|---|---|---|---|---|---|
| Sex | Male | 23 (28.0%) | 14 (36.8%) | 9 (20.5%) | 0.100 |
| Female | 59 (72.0%) | 24 (63.2%) | 35 (79.5%) | ||
| Age, in years | Median (Min - Max) | 39 (18 - 68) | 42 (18 – 62) | 37 (18 – 68) | 0.477 |
| 25 (30.5%) | 11 (28.9%) | 14 (31.8%) | ||
| 33 (40.2%) | 13 (34.2%) | 20 (45.5%) | ||
| 24 (29.3%) | 14 (36.8%) | 10 (22.7%) | ||
| Born in the Netherlands | No | 2 (2.4%) | 1 (2.6%) | 1 (2.3%) | 1.000ᵇ |
| Household size | Median (Min – Max) | 3 (1 – 15) | 3 (1 – 5) | 3 (1 – 15) | 0.784 |
| 1 person (participant lives alone) | 6 (7.3%) | 1 (2.6%) | 5 (11.4%) | ||
| 2 persons | 27 (32.9%) | 16 (42.1%) | 11 (25.0%) | ||
| 3 persons | 22 (26.8%) | 8 (21.1%) | 14 (31.8%) | ||
| 4 persons | 19 (23.2%) | 11 (28.9%) | 8 (18.2%) | ||
| 5 persons | 5 (6.1%) | 2 (5.3%) | 3 (6.8%) | ||
| >5 persons | 3 (3.6%) | 0 (0.0%) | 3 (6.8%) | ||
| Education level | Low | 16 (19.5%) | 6 (15.8%) | 10 (22.7%) | 0.716 |
| Middle | 46 (56.1%) | 23 (60.5%) | 23 (52.3%) | ||
| High | 20 (24.4%) | 9 (23.7%) | 11 (25.0%) | ||
| Work setting | Hairdresser | 32 (39.0%) | 13 (34.2%) | 19 (43.2%) | 0.406 |
| Hospitality industry | 50 (61.0%) | 25 (65.8%) | 25 (56.8%) | ||
| Workplace location | Breda | 40 (48.8%) | 22 (57.9%) | 18 (40.9%) | 0.585ᶜ |
| Roosendaal | 18 (22.0%) | 6 (15.8%) | 12 (27.3%) | ||
| Etten-Leur | 5 (6.1%) | 1 (2.6%) | 4 (9.1%) | ||
| Oosterhout | 4 (4.9%) | 2 (5.3%) | 2 (4.5%) | ||
| Other municipalities | 15 (18.3%) | 7 (18.4%) | 8 (18.3%) | ||
| Working hours | Median (Min – Max) | 30 (8 – 70) | 35 (8 – 60) | 26 (8 – 70) | 0.166 |
| 8 – 20 hours | 24 (29.3%) | 11 (28.9%) | 13 (29.5%) | ||
| 21 – 40 hours | 33 (40.2%) | 12 (31.6%) | 21 (47.7%) | ||
| ≥ 40 hours | 25 (30.5%) | 15 (39.5%) | 10 (22.7%) | ||
| Chronic disease | Yesᵃ | 19 (23.2%) | 12 (31.6%) | 7 (15.9%) | 0.094 |
| BMI, in kg/m2 | Median (Min – Max) | 25 (16 – 38) | 24 (17 – 33) | 25 (16 – 38) | 0.320 |
| Normal weight | 46 (56.1%) | 22 (57.9%) | 24 (54.5%) | ||
| Overweight | 29 (35.4%) | 14 (36.8%) | 15 (34.1%) | ||
| Obesity | 7 (8.5%) | 2 (5.3%) | 5 (11.4%) | ||
| Current smoker | Yes | 14 (17.1%) | 7 (18.4) | 7 (15.9%) | 0.763 |
| Current smokers: number of cigarettes per day | Median (Min – Max) | 7 (1 – 20) | 10 (1 – 20) | 2 (1 – 10) | 0.128 |
| < 10 | 8 (57.1%) | 2 (28.6%) | 6 (85.7%) | ||
| 10 - 20 | 6 (42.9%) | 5 (71.4%) | 1 (14.3%) | ||
| > 20 | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | ||
| Current alcohol use | Yes | 74 (90.2%) | 34 (89.5%) | 40 (90.9%) | 0.827 |
| Current alcohol users: alcohol units per week | Median (Min – Max) | 6 (0.5 – 41) | 8 (0.5 – 41) | 4 (0.5 – 34) | 0.068 |
| 0.5 - 7 | 42 (56.8%) | 16 (47.1%) | 26 (65.0%) | ||
| 8 – 14 | 12 (16.2%) | 9 (26.5%) | 3 (7.5%) | ||
| 15 - 21 | 8 (10.8%) | 4 (11.8%) | 4 (10.0%) | ||
| > 21 | 12 (16.2%) | 5 (14.7%) | 7 (17.5%) |
Baseline characteristics of participants with decreasing and non-decreasing antibody IgG antibody levels were compared using Wilcoxon Rank Sum tests (numerical variables) or chi-squared tests (categorical variables).
All values are n (%) unless specified otherwise.
a Patients were considered to have a chronic disease if they reported a chronic illness and/or used medication chronically.
b Due to small participant groups (n<5), a Fisher's exact test was used.
c We compared those working in Breda/Roosendaal to those working in other cities or villages.
BMI, body-mass index; min, minimum; max, maximum.
The two participants who seroreverted in the qualitative ELISA, were not IgG seropositive (or borderline) in the quantitative assay. We further examined these two participants. Both reported COVID-19-related symptoms prior to the initial qualitative seropositive test result, including fever in one participant. However, both participants did not have PCR confirmed COVID-19, as they had symptoms before recruitment when PCR testing was limited available in the Netherlands. Both participants were women, aged 17-30 years and 30-50 years. They did not have a chronic disease, did not smoke, reported minimal or no alcohol use and had a Body Mass Index (BMI) below 25kg/m2. As such, we were unable to identify potential determinants for their weak immune response, but also cannot confirm that they had truly been SARS-CoV-2 infected.
We evaluated longitudinal changes in individuals’ IgG antibody levels (Figure 2). Participants were divided into two groups based on changes in antibody levels over time: participants with (n=38, 46.3%) and without decreasing (n=44, 53.7%) antibody levels. Baseline characteristics of these two groups did not differ significantly (Table 1). Participants without decreasing antibody levels were further divided into those with stable (n=36, 43.9%), increasing (n=5, 6.1%) and fluctuating (n=3, 3.7%) antibody levels (Figure 2A), and did not have significantly different baseline characteristics (Supplementary Table 1). We hypothesized that those with fluctuating antibodies had a rise in antibody levels after an initial decline due to re-exposure to SARS-CoV-2. Indeed, one participant with fluctuating antibody levels reported that she tested SARS-CoV-2 positive before antibody levels increased again, but the other two participants did not report any re-exposure to SARS-CoV-2.
Figure 2Longitudinal changes in IgG antibody levels of COco-study participants. Participants were included when having at least one follow-up measurement after their first seropositive test result. Participants (n=82) were grouped based on whether they had decreasing or non-decreasing (stable, increasing or fluctuating) antibody levels (A). Longitudinal changes were plotted of the whole group (B) and separating participants by decreasing antibody levels (n=38) (C) and non-decreasing antibody levels (n=44) (D). Participants (n=82) were grouped based on the percentage decrease or increase in antibody concentration per three months (E).
Show full caption
Antibody levels of the whole group per measurement were plotted. The grey line indicates the trendline; red lines indicate 95% confidence intervals (F). Finally, antibody levels were plotted for the group of participants reporting a positive PCR test (n=29). The grey line indicates the trendline; red lines indicate the 95% confidence intervals. The striped, orange line indicates the median of the first measurement and the striped, blue line indicates the median of the second measurement after the self-reported positive PCR test (G).
Since follow-up time differed between participants, we calculated the rate of change of antibody levels per three months (Figure 2E). Only a minority (11.0%) of participants had a decrease in antibody levels of more than 50% per three months; the majority (69.5%) had a decrease in antibody levels between 0 and 50% per three months.
In addition, we analyzed group changes in antibody levels across the four measurements (Figure 2G). Antibody levels decreased significantly both between the first and second (n=82, median 42RU/ml, interquartile range (IQR) 25-98RU/ml versus median 29RU/ml, IQR 16-75RU/ml, p=0.002) and second and third measurements (n=48, median 27RU/ml, IQR 14-80RU/ml versus median 20RU/ml, IQR 10-52RU/ml, p<0.001), but not between the third and last measurements (n=37, median 19RU/ml, IQR 10-51RU/ml versus median 21RU/ml, IQR 10-60RU/ml, p=0.541). Of the 37 participants who had three follow-up visits after initial seropositivity, all had detectable antibodies at the fourth visit, but IgG levels in six participants (16.2%) were below the borderline of seropositivity (range 6-7RU/ml). When conducting subgroup analyses evaluating antibody level changes in those with confirmed PCR test results (n=29), similar results were observed: antibody levels decreased significantly between the first and second measurement (median 42RU/ml, IQR 26-187RU/ml versus median 25RU/ml, IQR 17-76RU/ml), p=0.007), but IgG antibodies remained detectable in all participants for the study duration (Figure 2G).
Next, we aimed to identify variables associated with higher antibody levels and/or antibody level dynamics.
First, we evaluated whether antibody levels at the first seropositive measurement were associated with longitudinal antibody responses (Figure 3). Participants with fluctuating antibody levels had the highest concentration of IgG antibodies at baseline (n=3, median 76RU/ml, min-max 14-191RU/ml), followed by participants with decreasing (n=38, median 56RU/ml, min-max 13-312RU/ml), stable (n=36, median 41RU/ml, min-max 12-660RU/ml) and increasing antibody levels (n=5, median 15RU/ml, min-max 9-52RU/ml). Participants with increasing antibody levels had significantly lower baseline antibody levels than those with decreasing (p=0.006) or stable (p=0.011) antibody levels; no other significant differences were observed.
Figure 3Antibody concentration of the whole group at the first seropositive measurement, divided into decreasing (n=38), stable (n=36), increasing (n=5) and fluctuating (n=3) antibody levels plotted in a boxplot combined with a scatterplot.
In further analyses (Table 2), we did not identify any measured baseline characteristic that significantly predicted decreasing antibody levels. Those with a chronic disease seemed to have more frequently decreasing antibody levels (OR 2.4, 95%CI 0.9-7.0) while women seemed to have decreasing antibody levels less frequently (OR 0.4, 95%CI 0.2-1.2), but these differences were not statistically significant (p=0.099 and p=0.103, respectively).
Table 2Associations between participant characteristics and longitudinal changes in SARS-CoV-2 IgG antibody levels
| Dependent variable: Decreasing antibody level (yes/no) | |
|---|---|
| Independent variable | OR (95%CI) |
| Fever (ref: no fever) | 1.3 (0.5 – 3.1) |
| Female sex (ref: male sex) | 0.4 (0.2 – 1.2) |
| Age (10-years increment) | 1.1 (0.8 – 1.6) |
| Chronic disease (ref: no chronic disease) | 2.4 (0.9 – 7.0) |
| BMI (5 kg/m² increment) | 0.7 (0.4 – 1.3) |
| Current smoker (ref: not currently smoking) | 1.2 (0.4 – 3.8) |
| - Smoking (5 cigarettes per day increment) | 2.7 (0.9 – 8.3) |
| Current alcohol user (ref: currently no alcohol use) | 0.9 (0.2 – 3.7) |
| - Alcohol quantity (7 alcohol units per week increment) | 1.2 (0.8 – 1.7) |
BMI, Body-mass index; CI, confidence interval; OR, odds ratio; Ref, reference.
Odd ratios and their 95% confidence intervals were calculated using a univariable logistic regression model.
Since previous studies suggested that those with severe disease developed higher antibody levels (Chvatal-Medina et al., 2021, Lynch et al., 2020), we evaluated whether disease severity predicted antibody levels and antibody level dynamics in our cohort (Figure 4). Since participants were not hospitalized for COVID-19, we grouped participants based on whether they reported fever. While participants with fever had higher antibody levels at every visit, these differences were only significant at the last measurement (n=14, mean 83RU/ml, 95%CI 20-146RU/ml versus n=24, mean 26RU/ml, 95%CI 16-36RU/ml, p=0.050). However, this difference was particularly driven by one participant with fever. She reported a SARS-CoV-2 reinfection between the third and fourth measurement and therefore, antibody levels increased from 22RU/ml to 412RU/ml. After excluding her from the analyses, those with fever still had higher mean antibody levels at the last visit (n=13, mean 58RU/ml, 95%CI 25-91RU/ml versus n=24, mean 26RU/ml, 95%CI 16-36RU/ml), but this difference was not significant (p=0.095). Both participants with and without fever had a tendency towards decreasing antibody levels over time (Figure 4).
Figure 4Longitudinal changes in the mean IgG antibody levels of the whole group. Participants were grouped based on whether they had fever (fever: n=52; no fever: n=30). Vertical lines indicate 95% confidence intervals.
Analyses of other baseline characteristics indicated that a higher age (B, 10-years increment: 24.6, 95%CI 5.7-43.5) and BMI (B, 5kg/m² increment: 40.0, 95%CI: 2.9-77.2) were significantly associated with antibody peak levels (Table 3). These associations remained significant after adjusting for disease severity (B: 25.0, 95%CI: 6.0-43.9, and B: 42.2, 95%CI: 4.9-79.5, respectively).
Table 3Associations between participant characteristics and peak of SARS-CoV-2 IgG antibody levels
| Dependent variable: Peak of antibody level | Univariable | Bivariable |
|---|---|---|
| Variables | B (95%CI) | B (95%CI) after adjusting for fever |
| Fever (ref: no fever) | 24.2 (-32.6 – 80.9) | - |
| Female sex (ref: male sex) | -44.2 (-104.5 – 16.1) | -43.4 (-103.9 – 17.1) |
| Age (10-years increment) | 24.6 (5.7 – 43.5)* | 25.0 (6.0 – 43.9)* |
| Chronic disease (ref: no chronic disease) | 47.8 (-16.4 – 112.0) | 46.3 (-18.2 – 110.8) |
| BMI (5 kg/m² increment) | 40.0 (2.9 – 77.2)* | 42.2 (4.9 – 79.5)* |
| Current smoker (ref: not currently smoking) | -41.7 (-114.0 – 30.7) | -41.4 (-113.9 – 31.1) |
| - Smoking (5 cigarettes per day increment) | -8.2 (-40.8 – 24.3) | -10.0 (-40.9 – 20.9) |
| Current alcohol user (ref: currently no alcohol use) | 26.7 (-65.7 – 119.0) | 26.9 (-65.6 – 119.4) |
| - Alcohol quantity (7 alcohol units per week increment) | -15.0 (-36.4 – 6.5) | -15.0 (-36.4 – 6.5) |
CI, confidence interval; B, beta coefficient; Ref, reference; *, p<0.05.
In the second column, beta coefficients and their 95% confidence intervals were calculated using a univariable linear regression model. The last column reports beta coefficients for the variable listed in the first column, after adjusting for severity of COVID-19 related symptoms (fever/no fever) using a bivariable linear regression model.
DISCUSSION
In this longitudinal cohort study on SARS-CoV-2 antibody levels in Dutch hairdressers and hospitality staff, we found little evidence for waning natural immunity. IgG antibody levels decreased somewhat in about half of participants during the follow-up of up to 12 months but remained detectable. In the qualitative Wantai SARS-CoV-2 total antibody assay, two participants seroreverted, having a seropositive test result followed by a seronegative test result. The seropositivity could not be confirmed by the quantitative IgG assay. Evaluating the A/C.O. ratio, using the Wantai assay semi quantitatively (Nilsson et al., 2021), these two participants only had weak seropositivity with an A/C.O. ratio just above the cut-off for seropositivity as determined by the manufacturer. This suggested a very weak immune response or false positive test result. All other participants continued to have detectable immune responses.
Abovementioned results match with other published studies on natural immunity after COVID-19. Previous studies in healthcare workers or severely ill COVID-19 patients generally reported that antibodies remain detectable for a long time after natural infection (Feng et al., 2021, Gallais et al., 2021, Gong et al., 2022, Huynen et al., 2022, Liu et al., 2021, Pilmis et al., 2022, Wang et al., 2021, Yang et al., 2022). However, these are subpopulations which may not reflect the general population, as healthcare workers are trained to work with protective equipment, and severely ill COVID-19 patients required hospitalization while most patients in the general population do not (Wiersinga et al., 2020). Our results demonstrate that antibodies also remain detectable in a group comparable to a general population.
Studies examining natural immunity in the general population are more limited and have contrasting results regarding the duration of detectable antibodies after COVID-19. While most studies are in line with our findings (Anand et al., 2021, Dan et al., 2021), a study in non-hospitalized individuals in Chicago reported that only half of the initially seropositive participants (N=87) had detectable IgG SARS-CoV-2 antibodies after 3–4 months (Demonbreun et al., 2021). A potential explanation for these seemingly contrasting data is that sample collection and assay differed. Demonbreun et al. (2021) used a self-sampled dried-blood spot assay, which may be less accurate, as the estimated sensitivity ranges between 81.5% and 89.4% (Oved et al., 2020). Alternatively, our participants may have continued to have detectable antibodies due to frequent re-exposure at work, resulting in natural boosting. Regardless, our subpopulation is relatively comparable to the current general population: as most countries have reduced or eliminated restrictive measures, people are regularly exposed to the virus, similar to our participants who were exposed due to social interactions at work, but at times, particularly when circulation of the virus increases, people take measures to prevent exposure, similar to our study population too. As such, our study contributes to an expanding body of literature suggesting long-term natural immunity after a SARS-CoV-2 infection in the general population. Such real-world evidence will aid policymakers and others in making decisions regarding social restrictions and vaccination, both for SARS-CoV-2 (for example, in populations or countries with low vaccine-induced immunity) and potential introductions of new coronaviruses.
We aimed to identify variables associated with antibody peak levels and dynamics. Lower baseline antibody levels were associated with increasing antibody levels over time, but likely reflect a short time between infection and measurement. Higher age and BMI were associated with higher antibody peak levels. This is in contrast with studies that found a negative association between antibody levels and BMI after COVID-19 vaccination (Kara et al., 2022, Watanabe et al., 2022), but is in agreement with a previous study examining the immune response after natural COVID-19 infection in Dutch blood donors, which used data unrelated to COco (Vinkenoog et al., 2022). This positive association between antibody peak levels after a natural infection and age or BMI could be related to the increased risk for severe COVID-19 disease in these populations, eventually resulting in stronger antibody responses (Chvatal-Medina et al., 2021, Gallo Marin et al., 2021, Lynch et al., 2020, Zhou et al., 2021). Chronic disease seemed to be associated with decreasing antibody levels in our cohort as well, potentially due to a (relative) immune deficiency, but this difference was not significant. Although not significant either, chronic disease seemed to be associated with higher antibody peak levels (B: 47.8, 95%CI: -16.4-110.8). Potentially, comorbidities are associated with increased disease severity, which in turn correlates with higher antibody levels (Gallo Marin et al., 2021).
Previous studies in hospitalized patients similarly suggested that severe illness was associated with higher SARS-CoV-2 antibody responses (Chvatal-Medina et al., 2021). Another longitudinal study, in which the population primarily consisted of recovered, non-hospitalized patients, also found an association between fever and higher IgG concentrations (Vinkenoog et al., 2022). Although participants who reported fever generally had higher antibody levels in our cohort, these differences were not significant, potentially due to the small sample size or recall bias.
A strength of our study is that our participants had diverse levels of education and socioeconomic status, in contrary to most studies which overrepresent the highly educated (Enzenbach et al., 2019). Another strength is the longitudinal measurements within individuals. While more frequent blood sample collection would have been preferred, we consider our data more reliable than cross-sectional data, as we and others have shown that antibody peak levels vary widely between individuals (Vinkenoog et al., 2022).
This study has several limitations. We evaluated SARS-CoV-2 total antibody and IgG levels, but did not include neutralizing antibodies. However, the IgG ELISA measured anti-RBD IgG, and since RBD is the main target for neutralizing antibodies, we assumed similar results (Ju et al., 2020). Second, participants had frequently not been tested by PCR due to limited availability, and PCR test results were self-reported. As a result, our study may include false-positive test results and we may have missed individuals without detectable antibody levels after infection. However, the specificity of both used antibody tests is high (≥99%) (EUROIMMUN, 2020, Havers et al., 2020), and cross-reactivity with other respiratory viruses, including non-SARS-CoV-2 coronaviruses, was non-existent (Harritshøj et al., 2021). Third, our population size limited the statistical power to identify variables associated with decreasing antibody levels. Pooled analyses may provide more details. Finally, we had not objectively confirmed the time of COVID-19 infection in all participants, while the time between infection and first measurement will affect antibody levels. However, dynamics remained similar when analyzing the subgroup of participants whose date of positive PCR test was available.
In conclusion, after an initial immune response due to a natural SARS-CoV-2 infection, we observed a reduction in antibody levels in approximately half of the studied seropositive hairdressers and hospitality staff in the Netherlands. However, most participants continued to have detectable antibody levels for up to one year. Therefore, our real-world data results suggest long-term immune protection after natural infection. Future studies should investigate whether this detectable natural immune response also results in less (severe) reinfections after one year, especially considering that this virus is continuously evolving into new variants.
CONTRIBUTIONS
Dymphie Mioch: Conceptualization, Methodology, Investigation, Formal analysis, Writing – original draft. Leonard Vanbrabant: Methodology, Formal analysis, Writing - review & editing. Johan Reimerink: Methodology, Formal analysis, Writing - review & editing. Sandra Kuiper: Conceptualization, Methodology, Formal analysis, Writing - review & editing, Funding acquisition. Esther Lodder: Conceptualization, Methodology, Formal analysis, Writing - review & editing. Wouter den Bijllaardt: Conceptualization, Methodology, Investigation, Formal analysis, Writing - review & editing. Jan Kluytmans: Conceptualization, Methodology, Investigation, Formal analysis, Writing - review & editing. Michel D. Wissing: Conceptualization, Methodology, Investigation, Formal analysis, Writing – original draft, Writing - review & editing, Supervision, Funding acquisition, Visualization.
ACKNOWLEDGEMENTS
The authors wish to thank all hairdressers and employees in the hospitality industry who voluntarily participated in this study. We would further like to thank the Dutch National Institute for Public Health and the Environment (RIVM) for providing the ELISA kits used to measure seroprevalence in our study population and for conducting the quantitative ELISA. We would like to thank the employees of Microvida for conducting the ELISA tests, and Anita Blerck, Desiree Bosker and Karin de Vries at the regional public health service of West-Brabant (GGD) for contacting participants to ensure participation in questionnaires and schedule appointments. We would also like to thank Royal Horeca the Netherlands (KHN) and the municipality of Breda for their help with the recruitment of participants. Finally, we would like to thank the management team of the regional public health service of West-Brabant, in particular Mark van Beers, and Fion Brouwer at the RIVM and Ariëne Rietveld, public health physician at the RIVM and regional public health service of Hart voor Brabant, for their support for and input in this study.
Funding source: The COco-study receives financial support from the regional public health service (GGD) of West-Brabant and the Dutch National Institute for Public Health and the Environment (RIVM). These are government-funded, non-commercial organizations, and they did not influence any aspect of the content of the study.
ETHICAL APPROVAL
The COco study has been approved by Medical research Ethics Committees United (MEC-U) at Nieuwegein, the Netherlands (project number A20.247/R20.041). It follows laws and guidelines on research with human subjects, including international standards such as the Declaration of Helsinki.
REFERENCES
Anand SP, Prévost J, Nayrac M, Beaudoin-Bussières G, Benlarbi M, Gasser R, et al. Longitudinal analysis of humoral immunity against SARS-CoV-2 Spike in convalescent individuals up to 8 months post-symptom onset. Cell Rep Med 2021;2(6):100290.
Chvatal-Medina M, Mendez-Cortina Y, Patiño PJ, Velilla PA, Rugeles MT. Antibody Responses in COVID-19: A Review. Frontiers in Immunology 2021;12(1208).
Dan JM, Mateus J, Kato Y, Hastie KM, Yu ED, Faliti CE, et al. Immunological memory to SARS-CoV-2 assessed for up to 8 months after infection. Science 2021.
Deisenhammer F, Borena W, Bauer A, Kimpel J, Rudzki D, Schanda K, et al. 6-month SARS-CoV-2 antibody persistency in a Tyrolian COVID-19 cohort. Wien Klin Wochenschr 2021;133(7-8):351-8.
Demonbreun AR, McDade TW, Pesce L, Vaught LA, Reiser NL, Bogdanovic E, et al. Patterns and persistence of SARS-CoV-2 IgG antibodies in Chicago to monitor COVID-19 exposure. JCI Insight 2021;6(9).
Enzenbach C, Wicklein B, Wirkner K, Loeffler M. Evaluating selection bias in a population-based cohort study with low baseline participation: the LIFE-Adult-Study. BMC Medical Research Methodology 2019;19(1):135.
EUROIMMUN. Product data sheet QuantiVac ELISA IgG. 2020.
Feng C, Shi J, Fan Q, Wang Y, Huang H, Chen F, et al. Protective humoral and cellular immune responses to SARS-CoV-2 persist up to 1 year after recovery. Nature Communications 2021;12(1):4984.
Gallais F, Gantner P, Bruel T, Velay A, Planas D, Wendling MJ, et al. Evolution of antibody responses up to 13 months after SARS-CoV-2 infection and risk of reinfection. EBioMedicine 2021;71:103561.
Gallo Marin B, Aghagoli G, Lavine K, Yang L, Siff EJ, Chiang SS, et al. Predictors of COVID-19 severity: A literature review. Reviews in Medical Virology 2021;31(1):e2146.
Gong X, Cui P, Wu H, Pan H, Teng Z, Yuan F, et al. Long-term immune responses in patients with confirmed novel coronavirus disease-2019: a 9-month prospective cohort study in Shanghai, China. BMC Infect Dis 2022;22(1):240.
Harritshøj LH, Gybel-Brask M, Afzal S, Kamstrup PR, Jørgensen CS, Thomsen MK, et al. Comparison of 16 Serological SARS-CoV-2 Immunoassays in 16 Clinical Laboratories. J Clin Microbiol 2021;59(5).
Havers FP, Reed C, Lim T, Montgomery JM, Klena JD, Hall AJ, et al. Seroprevalence of Antibodies to SARS-CoV-2 in 10 Sites in the United States, March 23-May 12, 2020. JAMA Internal Medicine 2020;180(12):1576-86.
He Z, Ren L, Yang J, Guo L, Feng L, Ma C, et al. Seroprevalence and humoral immune durability of anti-SARS-CoV-2 antibodies in Wuhan, China: a longitudinal, population-level, cross-sectional study. Lancet 2021;397(10279):1075-84.
Huynen P, Grégoire C, Gofflot S, Seidel L, Maes N, Vranken L, et al. Long-term longitudinal evaluation of the prevalence of SARS-CoV-2 antibodies in healthcare and university workers. Sci Rep 2022;12(1):5156.
Ju B, Zhang Q, Ge J, Wang R, Sun J, Ge X, et al. Human neutralizing antibodies elicited by SARS-CoV-2 infection. Nature 2020;584(7819):115-9.
Kaplan EL, Meier P. Nonparametric Estimation from Incomplete Observations. Journal of the American Statistical Association 1958;53(282):457-81.
Kara Z, Akçin R, Demir AN, Dinc HO, Kocazeybek B, Yumuk VD. Risk factors influencing seroconversion after inactive SARS-CoV-2 vaccination in people living with obesity. Obes Facts 2022.
Kučinskaitė-Kodzė I, Simanavičius M, Šimaitis A, Žvirblienė A. Persistence of SARS-CoV-2-Specific Antibodies for 13 Months after Infection. Viruses 2021;13(11).
Liu C, Yu X, Gao C, Zhang L, Zhai H, Hu Y, et al. Characterization of antibody responses to SARS-CoV-2 in convalescent COVID-19 patients. J Med Virol 2021;93(4):2227-33.
Luo H, Camilleri D, Garitaonandia I, Djumanov D, Chen T, Lorch U, et al. Kinetics of anti-SARS-CoV-2 IgG antibody levels and potential influential factors in subjects with COVID-19: A 11-month follow-up study. Diagn Microbiol Infect Dis 2021;101(4):115537.
Lynch KL, Whitman JD, Lacanienta NP, Beckerdite EW, Kastner SA, Shy BR, et al. Magnitude and kinetics of anti-SARS-CoV-2 antibody responses and their relationship to disease severity. Clin Infect Dis 2020.
Mioch D, Kuiper S, van den Bijllaardt W, van Jaarsveld CHM, Kluytmans J, Lodder E, et al. SARS-CoV-2 antibodies in employees working in non-medical contact-intensive professions in the Netherlands: Baseline data from the prospective COco-study. Preventive Medicine Reports 2021;24:101594.
Nilsson AC, Holm DK, Justesen US, Gorm-Jensen T, Andersen NS, Øvrehus A, et al. Comparison of six commercially available SARS-CoV-2 antibody assays-Choice of assay depends on intended use. Int J Infect Dis 2021;103:381-8.
Oved K, Olmer L, Shemer-Avni Y, Wolf T, Supino-Rosin L, Prajgrod G, et al. Multi-center nationwide comparison of seven serology assays reveals a SARS-CoV-2 non-responding seronegative subpopulation. EClinicalMedicine 2020;29:100651.
Petersen MS, Hansen CB, Kristiansen MF, Fjallsbak JP, Larsen S, Hansen JL, et al. SARS-CoV-2 Natural Antibody Response Persists for at Least 12 Months in a Nationwide Study From the Faroe Islands. Open Forum Infect Dis 2021;8(8):ofab378.
Pilmis B, Elkaibi I, Péan de Ponfilly G, Daikha H, Bouzid A, Guihot A, et al. Evolution of anti-SARS-CoV-2 immune response in a cohort of French healthcare workers followed for 7 months. Infect Dis Now 2022;52(2):68-74.
Post N, Eddy D, Huntley C, van Schalkwyk MCI, Shrotri M, Leeman D, et al. Antibody response to SARS-CoV-2 infection in humans: A systematic review. PLoS One 2020;15(12):e0244126.
Vinkenoog M, Steenhuis M, Brinke AT, van Hasselt JGC, Janssen MP, van Leeuwen M, et al. Associations Between Symptoms, Donor Characteristics and IgG Antibody Response in 2082 COVID-19 Convalescent Plasma Donors. Front Immunol 2022;13:821721.
Wang H, Yuan Y, Xiao M, Chen L, Zhao Y, Haiwei Z, et al. Dynamics of the SARS-CoV-2 antibody response up to 10 months after infection. Cell Mol Immunol 2021;18(7):1832-4.
Watanabe M, Balena A, Tuccinardi D, Tozzi R, Risi R, Masi D, et al. Central obesity, smoking habit, and hypertension are associated with lower antibody titres in response to COVID-19 mRNA vaccine. Diabetes/Metabolism Research and Reviews 2022;38(1):e3465.
WHO. WHO Coronavirus (COVID-19) Dashboard. 2022a.
WHO. WHO Coronavirus (COVID-19) Dashboard: the Netherlands. 2022b.
Wiersinga WJ, Rhodes A, Cheng AC, Peacock SJ, Prescott HC. Pathophysiology, Transmission, Diagnosis, and Treatment of Coronavirus Disease 2019 (COVID-19): A Review. Jama 2020;324(8):782-93.
Yang Y, Yang M, Peng Y, Liang Y, Wei J, Xing L, et al. Longitudinal analysis of antibody dynamics in COVID-19 convalescents reveals neutralizing responses up to 16 months after infection. Nat Microbiol 2022;7(3):423-33.
Zhou Y, Chi J, Lv W, Wang Y. Obesity and diabetes as high-risk factors for severe coronavirus disease 2019 (Covid-19). Diabetes Metab Res Rev 2021;37(2):e3377.
Appendix. Supplementary materials
Article Info
Publication History
Accepted:
November 21,
2022
Received in revised form:
October 28,
2022
Received:
August 2,
2022
Publication stage
In Press Journal Pre-ProofFootnotes
Declaration of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Identification
Copyright
© 2022 The Author(s). Published by Elsevier Ltd on behalf of International Society for Infectious Diseases.
User License
Creative Commons Attribution (CC BY 4.0) | How you can reuse 
Elsevier's open access license policy
Creative Commons Attribution (CC BY 4.0)
Permitted
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article
- Reuse portions or extracts from the article in other works
- Sell or re-use for commercial purposes
Elsevier's open access license policy
00616-6&id=gr1.jpg){kind=link}
00616-6&id=gr2.jpg){kind=link}
00616-6&id=gr3.jpg){kind=link}
00616-6&id=gr4.jpg){kind=link}