Demographical data
Among the participants, 170 (73.6%) were females (mean age = 43 years; range, 23-66 years) and 61 (26.4%) were males (mean age = 43 years; range, 23-64 years). One-hundred and fifty eight subjects (68.3%) were COVID-19 naïve and were categorized as seronegative at baseline while 73 (31.6%) were seropositive. All demographical data of the population are presented in Supplementary Table 1.
Anti-NCP antibodies
Among the cohort, anti-NCP antibodies remained stable in seropositive participants up to 6 months compared to pre-vaccinal titers (P > 0.05) (Supplementary Figure 1). At the individual level, 3 participants (1.3%) had a significant increase in their anti-NCP antibody levels. The first subject was seronegative before vaccination and had a positive RT-PCR 93 days since the first vaccine dose. The B.1.1.7 variant was identified by sequencing. The subject had a close contact with an infected patient and was asymptomatic. The second subject was seropositive at baseline with a documented positive RT-PCR in April 2020 and reported a positive RT-PCR 71 days after the first dose. The subject developped minor symptoms. No sequencing was available. The third subject was seronegative before vaccination and reports a positive RT-PCR carried out in the context of persisting flu-like symptoms, only 2 days after the first dose.
IgG and total antibodies
In seronegative individuals, the maximal antibody response was reached at day 28 with a mean total antibodies titer of 2,204 U/mL (95%CI: 1,833 – 2,575 U/mL) and a mean IgG titer of 18,785 AU/mL (95%CI: 16,020 – 21,549 AU/mL). A continuous decrease was observed up to day 180 with an observed mean total antibodies titer of 998 U/mL (95%CI: 848 –1,148) and an observed mean IgG titer of 1,949 AU/mL (95%CI: 1,565 – 2,332) which represent a decrease of 54.7% and 89.6%, respectively (Table 1, Figure 1). In seropositive individuals, the maximal antibody reponse was reached at day 28 and day 42 for total and IgG antibodies, respectively. The mean total antibodies titer was 16,935 U/mL (95%CI: 15,112 – 18,759) and the mean IgG titer was 30,678 AU/mL (95%CI: 26,600 – 34,755). A continuous decline was also observed between day 28 or day 42 and day 180 with a total antibodies titer of 4,270 U/mL (95%CI: 3,324 – 5,215) which represent a decrease of 74.8% and 79.4%, respectively (Table 1, Figure 1). All participants still had detectable anti-S antibodies 6 months after the first vaccine dose (i.e. total antibodies titer ≥ 0.8 U/mL and IgG titer ≥ 50 AU/mL).
Considering each time point separately, anti-S titers of seropositive individuals were always statistically higher compared to seronegative individuals (P < 0.0001), except for IgG at the 6-month timepoint (Table 1). The difference of titers between seronegative and seropositive individuals was higher when measuring total antibodies compared to IgG, but the difference tends to decrease over time (Table 1, Figure 1).
Using the kinetics model, the time to maximum concentration (Tmax) in seronegative subjects for total antibodies and IgG was comparable with 36.3 days (95% CI: 30.2 – 42.5) versus 34.5 days (95% CI: 31.7 – 37.2). The estimated half-life (T1/2) for total antibodies was 114 days (95% CI: 87 – 167) and was significantly longer than the 21 days (95% CI: 13 – 65) obtained for IgG. In seropositive subjects, the Tmax for total antibodies and IgG were also comparable (23.3 days (95% CI: 18.7 – 28.0) versus 25.0 days (95% CI: 20.4 – 29.9), and was shorter compared to seronegatives. The estimated half-life for total antibodies was 68 days (95% CI: 54 – 90) and slightly longer compared to IgG (i.e. 53 days, 95% CI: 40 – 79) (Figure 2).
According to the model, a mean time of 229 days (95% CI: 134 – 277) in seronegatives and 529 days (95% CI: 283 – 623) in seropositives would be needed to cross the threshold of 50 AU/mL for the IgG assay. For the total antibody assay, a mean time of 830 days (95% CI: 508 – 1,000) in seronegatives and 718 days (95% CI: 425 – 826) in seropositives would be needed to cross the threshold of 15 U/mL which was defined by the manufacturer as a cut-off for detection of inhibitory effects.21 Using the threshold of 133 U/mL,22 the mean time needed would be 470 days (95% CI: 341 – 585) and 507 days (95% CI: 359 – 591) in seronegative and seropositive subjects, respectively.
Among the 1,443 samples analyzed on both assays, IgG and total anti-S antibodies showed an almost perfect agreement (i.e. Cohen’s kappa = 0.97) with a Spearman’s correlation coefficient of 0.892 (95% CI: 0.881 – 0.902 ; P < 0.0001) (Supplemental Figure 2).
Neutralizing antibodies
In the 42 seronegative subjects included in this subgroup analysis, NAbs increased from a mean dilution factor of 11.9 (95% CI: 9.96 – 13.8) at day 0 to 1,955 (95%CI: 1,287 – 2,622) at day 28, which represents an increase of 99.4% (P < 0.0001) with all subjects having detectable NAbs at day 28. At day 90 and day 180, the mean dilution factors decreased to 127.6 (95% CI: 84.3 – 170.9) and 26.1 (95%CI: 20.1 – 32.1), which represents, respectively, a significant decrease of 93.5% and 98.7% compared to day 28. At these timepoints, the positivity rates dropped at 95.2% and 45.0% (Figure 3), respectively. In the 18 seropositive subjects, 72.2% of the subjects had detectable NAbs at baseline. At day 28, the mean dilution factor increased from 43.8 to 2,091 which represents an increase of 97.9% (P < 0.001). At day 90 and day 180, the mean dilution factors were 163.1 (95%CI: 83.5 – 242.6) and 30.5 (95% CI: 18.2 – 42.7) which represents a significant decrease of 92.2% and 98.5%, respectively. All subjects had detectable levels of NAbs at day 28 and day 90 but the positivity rate decreases to 44.4% at day 180. Considering each timepoint separately, NAbs of seropositive individuals were not statistically different compared to seronegative individuals (P > 0.9998). The kinetic model found that the estimated half-life of NAbs in this subgroup was 16 days (95% CI: 9 to 59 days) and that the time to reach the negativity cut-off was 135 days (95% CI: 55 – 179 days). A significant correlation with total antibodies was found (r = 0.63, P < 0.0001) (Supplemental Figure 3) but the strength of agreement was moderate with the manufacturer’s cut-off (Cohen’s kappa = 0.60). The use of alternative cut-off (15 and 133 U/mL) did not increase the agreement (0.58 and 0.54, respectively). For the IgG assay, we observed a better correlation (r = 0.78, P < 0.0001), with still a moderate strength of agreement (Cohen’s kappa = 0.66) (Supplemental Figure 3). No alternative cut-off could enhance the observed agreement.