Antibody response to SARS-CoV-2 for more than one year − kinetics and persistence of detection are predominantly determined by avidity progression and test design

Background Antibody detection of SARS-CoV-2 requires an understanding of its variation, course, and duration. Methods Antibody response to SARS-CoV-2 was evaluated over 5–430 days on 828 samples across COVID-19 severity levels, for total antibody (TAb), IgG, IgA, IgM, neutralizing antibody (NAb), antibody avidity, and for receptor-binding-domain (RBD), spike (S), or nucleoprotein (N). Specificity was determined on 676 pre-pandemic samples. Results Sensitivity at 30–60 days post symptom onset (pso) for TAb-S/RBD, TAb-N, IgG-S, IgG-N, IgA-S, IgM-RBD, and NAb was 96.6%, 99.5%, 89.7%, 94.3%, 80.9%, 76.9% and 92.8%, respectively. Follow-up 430 days pso revealed: TAb-S/RBD increased slightly (100.0%); TAb-N decreased slightly (97.1%); IgG-S and IgA-S decreased moderately (81.4%, 65.7%); NAb remained positive (94.3%), slightly decreasing in activity after 300 days; there was correlation with IgG-S (Rs = 0.88) and IgA-S (Rs = 0.71); IgG-N decreased significantly from day 120 (15.7%); IgM-RBD dropped after 30–60 days (22.9%). High antibody avidity developed against S/RBD steadily with time in 94.3% of patients after 430 days. This correlated with persistent antibody detection depending on antibody-binding efficiency of the test design. Severe COVID-19 correlated with earlier and higher antibody response, mild COVID-19 was heterogeneous with a wide range of antibody reactivities. Specificity of the tests was ≥99%, except for IgA (96%). Conclusion Sensitivity of anti-SARS-CoV-2 assays was determined by test design, target antigen, antibody avidity, and COVID-19 severity. Sustained antibody detection was mainly determined by avidity progression for RBD and S. Testing by TAb and for S/RBD provided the highest sensitivity and longest detection duration of 14 months so far.


Introduction
SARS-CoV-2 infection was declared a pandemic by WHO on March 11th, 2020 [1]. SARS-CoV-2 is a novel human coronavirus that can cause severe respiratory illness. Primary diagnosis is performed in the first 1-2 weeks after the onset of symptoms by direct detection of SARS-CoV-2 by nucleic acid amplification test (NAT) or antigen testing from respiratory secretions of nasal or throat swabs [2,3]. Specific antibodies to SARS-CoV-2 develop relatively rapidly, with most patients becoming seropositive within 15-21 days [2] of infection, while viral load decreases and patients eventually become virus-negative. Antibody testing can therefore aid diagnosis in the acute phase adjunct to PCR or antigen testing [4], and identify previous SARS-CoV-2 infection.
Antibody detection may thus be useful to assess antibody response after infection or vaccination, for serosurveillance studies, and to distinguish vaccine-induced seropositivity from natural infection [4,5]. However, interpretation of SARS-CoV-2 antibody responses remains challenging because of considerable heterogeneity among individuals [6], and because the results of SARS-CoV-2 antibody assays may vary widely [7]. In addition, it is still unclear how binding antibodies relate to neutralizing antibodies [6,8,9]. Besides, there remains the crucial question how long an antibody response persists.
The aim of this study was to characterize the course and duration of antibody responses over more than one year in symptomatic patients covering the range of different COVID-19 symptom severity levels and by using 12 anti-SARS-CoV-2 tests encompassing various test designs, detected antibody types, target antigens, and by evaluating test sensitivity, antibody titers, neutralizing activity and antibody avidity.
Samples were collected post symptom onset (pso) for a period of up to 430 days. All individuals tested positive for SARS-CoV-2-NAT and were symptomatic for COVID-19. Patients reported being unvaccinated or samples were collected prior to vaccine availability (12/21/2020). In total 828 samples from 390 patients from three sites were analyzed: The infectious diseases' outpatient department of the University of Frankfurt provided 752 serum samples from 365 patients. Patient samples had been prospectively collected since May 11th, 2020, from day 5 − 430 days (median 147 days) pso, and retrospectively categorized at sample time by reviewing individuals' worst prior disease state, using symptom severity scores of 0-8 according to the contemporary WHO ordinal scale classification [10]. The study was approved by the Frankfurt University Ethics Committee (Votes-No. 11/17 & 20-748), and patients gave written informed consent to be admitted.
Forty-five samples were from Erlangen and Nürnberg (Germany) from two family clusters with total 14 COVID-19 patients. Serum collection was eight days pso for one patient and 42, 171, 311, 407 days pso in the other patients. Thirty-one samples of 11 patients were from the medical service of the Paul-Ehrlich-Institute collected 5 − 392 days pso. Patients of these latter two populations had mild COVID-19 symptoms (scores 1-2 [10]). Ethics committee approval was obtained from the Hessische Landesärztekammer (60314 Frankfurt am Main, Germany, ballot no. 2020-1664-evBO), including written informed consent from patients participating in the study.
Testing was carried out according to manufacturers' instructions at the IVD Testing Laboratory at the Paul-Ehrlich-Institute. Sensitivity of anti-SARS-CoV-2 detection was calculated by standard formula, relative to the reported positive first diagnosis of SARS-CoV-2 infection by NAT. Antibody quantification was performed with the Liaison SARS-CoV-2 S1/S2 IgG test (Diasorin SpA, 13040 Saluggia, Italy), calibrated in arbitrary units (AU/mL) and a measuring range of 15-400 AU/ml. The surrogate virus neutralization test (sVNT), (Genscript USA Inc., NJ 08854) detecting ACE-2 blocking Ab was used to refer to neutralizing antibodies (NAb) based on its correlation to virus neutralization test [11]. An inhibition rate of ≥20% was the cutoff for positive neutralization. All samples were subjected to avidity testing by recomLine SARS-CoV-2 avidity reagent (Mikrogen GmbH, 82061 Neuried, Germany) and quantified using BLOTrix Reader BTR48 and recomScan 3.4 test strip analysis software (Mikrogen GmbH). High avidity was defined as antibody binding capacity of ≥60%. Correlation between antibody quantification, avidity, neutralization and COVID-19 symptom severity were calculated using Spearman rank analysis [12]. Relative antibody binding efficiency (affinity) between different tests was determined by endpoint titration: samples were serially diluted (4-fold) up to the intercept with the assay's cutoff. The titer was calculated by linear interpolation, with a higher titer representing a higher relative affinity of the respective test. Specificity of the tests was determined by standard formula and 95% confidence intervals (Clopper-Pearson for binomial distributed data), using pre-pandemic samples 576 plasma blood donation retention samples of 2014 obtained from the Red Cross Blood Donor Service NSTOB (Springe, Germany) and 100 serum samples from US blood donation units purchased from Trina Bioreactives (Nänikon, Switzerland) in 2011.

Sensitivity and titers of SARS-CoV-2 antibody tests over time
(v) IgG-N showed high initial sensitivity of 93.4% (mean of two IgG-N tests) up to 120 days pso and then dropped rapidly to 14.5% by 430 days pso. (vi) TAb-N sensitivity was 99-100% until day 120 pso and decreased only slightly to 96.8% by day 430 pso. TAb-N titers also decreased after 120 days of pso slowly and steadily, with a similar trend as IgG-N. (vii) IgM-RBD dropped rapidly after 30-60 days to low sensitivity of 16.1% at 430 days pso but showing a wider range between individual patients including persistent IgM. (viii) Antibody avidity increased with time, primarily for RBD and S. In severe COVID-19, high avidity (>60%) developed rapidly within 30-60 days pso, whereas in mild COVID-19 avidity progressed slowly reaching high avidity (>60%) on average after 120 days pso. Most patients (93.5%) had high avidity for RBD at 430 days pso. There was significant correlation (p < 0.001) of avidity with IgG-S and NAb (Rs 0.55, 0.56), adjusted for time lag until maturation to high avidity (Fig. 4). Increasing avidity suggests enhancement of antibody affinity in the tests, as verified in an antibody titration experiment (Fig. 5). Twelve patients of varying COVID-19 severity (ten score 1-3, two score 4), each with four follow-up samples (mean 54, 142, 238, and 324 days pso) and mean avidity for RBD of 26.7, 53.5, 66.7, and 71.1%, respectively, were plotted against the endpoint titers of each sample at the cutoff of the respective assay. TAb-RBD titers (176, 310, 680, and 858) increased ( Fig. 5A) with the increasing avidity to RBD/S (Fig. 5B), while titers for IgG-S and IgA-S were overall lower (IgG-S: 19, 10, 11, 9; IgA-S: 6, 6, 5, 5) and slightly decreasing (Fig. 5C). In comparison, TAb-N and IgG-N titers decreased 5-to 6-fold relative to TAb-RBD and IgG-S tests, consistent with decreasing avidity to N. Fig. 5A also shows that despite equal rates of decline in N-based assays versus S, absolute TAb-N titers were nevertheless 5-120-fold higher than IgG-N. NAb showed a similar titer profile as IgG-S and IgA (Fig. 5C), but with baseline 2-3 and 5-fold higher titers than IgG-S and IgA-S, respectively.

Test specificity
Specificity ( Table 2) of the tests was 99.0-100% for IgG, TAb and NAb tests, 98.8% for IgM and 96% for IgA-S, independent of the target antigen, and no detectable cross-reactivity with human endemic coronaviruses by immunoblot (data not shown).

Discussion
The variation, time course, and duration of antibody responses to SARS-CoV-2 over 14 months after the onset of the pandemic were examined. Criteria were test sensitivity, antibody titers of the tests over time in relation to the severity of COVID-19 symptoms, target antigens of the tests, types of antibodies detected, association with NAb and antibody avidity. Symptom distribution of COVID-19 patients was representative for the overall distribution with mild symptomatic COVID-19 representing the majority of cases [13,14]. COVID-19 severity is known to correlate with antibody levels and persistence [15][16][17][18]. In this study, the impact of COVID-19 symptom severity was differentiated between two groups, which were not limited to    outpatient/hospitalized treatment but also by need for respiratory therapy, i.e. scores 1-3 versus scores 4-7 (requiring oxygen therapy or ventilation) [19,20]. The picture within the mild COVID-19 group was heterogeneous with a wide range of antibody reactivities partly in overlap with severe COVID-19, as previously reported [18,21], and a corresponding variability between the different tests. Nevertheless, the results demonstrate a distinct test-specific pattern over time, allowing differences between tests to be classified. In summary, sensitivity and duration of antibody detection was dependent on test design and progression to high avidity. The sandwich design of TAb assays, and RBD/S-based assays, resulted in higher sensitivity over time based on increasing avidity for S/RBD. N-based assays were less sensitive than the corresponding S-based assays consistent with decreasing avidity for N. (i) TAb-S/RBD tests thus showed the highest and longest-lasting sensitivity, with stable antibody titers throughout the 14-month observation period and no predictable detection endpoint. (ii) The sensitivities of IgG-S/RBD and IgA-S were stable until 300 days pso followed by moderate decrease of antibody titers and in sensitivity of about 10-16% until 430 days pso. (iii) N-based assays showed an early decline in antibody titers starting from day 120 pso. Although sensitivity (test-positivity) of TAb-N decreased only slightly by day 430 pso, due to its sandwich design, declining antibody titers suggest a further decrease in sensitivity. For IgG-N, the decline was much faster after 120 days pso with short-lived sensitivity becoming insignificant after 240-300 days pso. (iv) IgM-RBD sensitivity was essentially confined to the early phase (30-60 days pso), then declined rapidly, although not consistently, since some individuals showed persistent IgM. Therefore, IgM does not appear to be suitable for accurate timing of infection. (v) NAb, targeting RBD, showed consistently high or slightly increased base sensitivity (>20% inhibition) of 90-94% during 430 days pso, indicating more sustained duration of neutralization than published [18,21]. Neutralization potency correlated strongly over time with quantitative IgG-S titers (Rs  0.88) [6,18,23] and with IgA-S (Rs 0.71) involved in mucosal immunity [22]. (vi) Antibody avidity developed mainly for RBD and S and increased slowly over time, from about 120 days pso to high avidity (>60%), remaining stable or increasing further up to 430 days pso. Most patients attained high avidity, 76% at one year and 94% after 430 days pso. Considering the time lag of avidity maturation, there was positive correlation with IgG-S antibody titers and neutralizing activity (Rs 0.55, 0.56). In contrast to the immunogenic RBD and S, [24], avidity against N was low, with 20-26% of patients showing high avidity.
Increasing avidity reflects affinity maturation of antibodies, mediation of cross-reactivity, and formation of long-lived plasma cells that can secrete antibodies in absence of antigen [9,[24][25][26][27][28][29]. Mature antibodies formed after acute infection may have up to 100-fold higher affinity [30,31]. On the other hand, the test design determines the extent to which a test benefits from higher affinity. This was confirmed by evaluation of the endpoint titers, reflecting the relative affinity and concentration of antibody in the sample. The sandwich test design of TAb-RBD assays exploited high avidity most effectively with 5-fold higher titers at high versus low avidity, and 9-to 93-fold or 31-to 158-fold higher titers than the corresponding IgG-S or IgA-S test in the indirect assay format. In contrast, N-based TAb and IgG endpoint titers decreased 5-to 6-fold over time relative to S-based assays, in line with decreasing avidity to N. Nevertheless, TAb-N due to its sandwich design [32], provided 5-120-fold higher baseline titers than indirect IgG-N (as the ratio TAb-S/RBD to IgG-S), explaining its high test-positivity in clinical samples and making TAb-N suitable for long-term detection of anti-N. The sVNT assay for NAb, detecting TAb using a competitive test design, showed a base sensitivity between TAb sandwich and indirect IgG-S and IgA-S assays, with a time course similar to the latter.
Sensitivity of anti-SARS-CoV-2 tests in this study compared with high specificity of 99% or greater (except IgA) in 676 pre-pandemic blood donation samples, indicating reliable positive predictive value.
A limitation of the study is that no asymptomatic individuals were included. Nevertheless, the results of this study with the mildly symptomatic patients suggest a basically similar picture in asymptomatic individuals.
In conclusion, antibody detection showed a distinct pattern dependent on the antibody type detected, the target antigen, antibody avidity level, the test design, and severity of COVID-19. Duration of detection was mainly driven by avidity progression for S/RBD and its exploitation by the respective test design. TAb assays based on S/RBD showed high sensitivity and persistent antibody detection for more than 14 months with consistent antibody titers and no predictable end of detection so far.

Funding statement
The study was supported by the German Federal Ministry of Health.

Declaration of Competing Interest
The authors declare no conflict of interests.