Samples obtained at each timepoint were tested for serum levels of antibodies against SARS-CoV-2 spike (S1) protein receptor binding domain (RBD) (S1-AB) and antibodies against the SARS-CoV-2 nucleocapsid antigen (N) (N-AB). S1-AB served as marker of infection as well as vaccination, whereas N-AB served as marker of infection only .
N-AB are not expected to increase upon mRNA-vaccination against spike (S1) protein receptor binding domain (RBD) in the absence of infection. Consequently, N-AB-positive participants were assumed to having undergone asymptomatic infection with SARS-CoV-2 virus before vaccination or during the post-vaccination period monitored in the study. 1.8% (n = 2) of the participants (n = 113, details: Suppl.Table 1) were positive for N-AB and S1-AB in the sample obtained before the first vaccination, indicating that they had already undergone inapparent or unregistered infection(s) with SARS-CoV-2 virus. Another participant was tested positive for a SARS-CoV-2 infection by PCR test. According to the vaccination regimen of the institution, all participants were vaccinated irrespective of their serological state. However, in our study, N-AB- and PCR-positive participants were kept separate in statistical analyses.
S1-AB levels before and at various timepoints after vaccination: 99.1% (n = 109) of the N-negative participants (n = 110) and 100% (n = 116, details: Suppl.Table 1) were tested positive after first and second vaccination, respectively. All N-negative participants monitored six months after the second vaccination (n = 95, details: Suppl.Table 1) were still S1-AB-positive. These values are in good agreement with vaccination responses observed elsewhere . Mean-values of S1-AB were 169 U/ml (0.4–1,004 U/ml) after first vaccination, increased to 5704 U/ml (213 − 17,764 U/ml) after second vaccination and dropped again to 1,019 U/ml (69–5,220 U/ml) six months after the second vaccination. A synoptic representation of S1-AB values obtained at the various timepoints of observation is given in Fig. 1A.
Alterations of antibody levels over time where highly significant (p < 0.001), giving rise to highly inhomogeneous timecourses of sero-responses (Fig. 1B). Inter-individual divergence started with immediate vaccination responses: Certain participants showed a huge increase of antibody-levels from an above-median level after the first vaccination to an even higher level above median after the second vaccination (Index Pat. A), whereas other participants responded with sub-median rises in S1-AB to the first vaccination and exhibited no significant further increase following the second vaccination (Index Pat. B) Fig. 1B. Timecourses of S1-AB levels during six months after the second vaccination were even more heterogeneous, encompassing a drop to as low as 2.6% (250 U/ml of 9,724 U/ml, Index Pat. C) as well as maintenance of as much as 67.5% (5.220 of 7.725 U/ml, Index Pat. D) of the initial S1-AB-level reached after the second vaccination. Most notably, S1-AB levels immediately after second vaccination exhibited only a very moderate correlation (r = 0.54, p < 0.001) with corresponding residual S1-AB levels observed six months later. The rather poor linear regression of that data (r² =0.16, p < 0.001) suggests that immediate humoral vaccination response and long-termed maintenance of humoral immunity are not stringently linked in quantitative terms. (Fig. 1C).
Samples of participants having undergone SARS-CoV-2 infection before vaccination (n = 3, details: Suppl.Table 1) were identified by increased serum levels of N-AB and/or positive virus-PCR. These samples exhibited many-fold higher levels of S1-AB. After first vaccination, the mean value of S1-AB in post-infection samples was 47,738 ± 3,002U/ml as opposed to 169 ± 16.6 U/ml in non-infected samples. After second vaccination, the mean value of S1-AB in post-infection samples was 43,001 ± 1,532 U/ml as opposed to 5,704 ± 322.9 U/ml in non-infected samples. These differences were highly significant (p < 0.001). In the long run, the augmenting effect of SARS-CoV-2 infection on vaccination response started to diminish. At six months after the second vaccination the mean value of S1-AB in post-infection samples was 3,070 ± 417 U/ml as opposed to 1.019 ± 88.5 U/ml in non-infected samples. This difference was still significant (p = 0.001) but quantitatively less pronounced than at the timepoints directly after vaccination (see: Fig. 1A and Suppl. Figure 1).
As a next step, we compared S1-AB serum levels with corresponding virus neutralizing activity of the sera. For that purpose, all samples were probed for their potency to inhibit the binding of biotin-labelled ACE2-receptor to immobilised recombinant SARS-CoV-2-S1/-RBD (NeutraLISA, EUROIMMUN), which is considered a practical diagnostic surrogate for the neutralization of cytopathic effects of the full viable virus as determined in cell culture. After first vaccination and six months after second vaccination levels of S1-AB correlated strongly with corresponding virus-neutralization capacity of the sera (r2 = 0.774 to 0.845). Immediately after second vaccination, a similar analysis appeared not meaningful since the upper measuring limit of the NeutraLISA at 100% was already attained by sub-median levels of S1-AB. Thus, the limited dynamic range rendered the NeutraLISA uninformative in the situation of recent re-immunisation. Similar results were obtained by cPass (not shown). The two surrogate assays exhibited excellent linear correlations across all time points (r² = 0.774 to 0.932, p < 0.001) (Suppl. Figure 1). It should be noted that the cPass assay appeared slightly more sensitive in the low range (after first vaccination) but yielded similar values (around 98%) after second vaccination.
In summary, the two surrogate assays for virus neutralization capacity fail to provide meaningful additional information regarding immediate vaccination responses. However, they may be useful in long-termed monitoring of humoral vaccination responses. To follow up on the latter notion, NeutraLISA data obtained at six months after second vaccination was scrutinized for relevance. Based on comparisons with WHO-standards and full virus endpoint dilution neutralization test (full virus NT), inhibition values of ≥ 35% obtained by the NeutraLISA in post-infection sera are proposed to indicate effective virus neutralization potency . However, according to own unpublished observations the neutralization potency of antibodies induced by S1-spike protein-directed vaccination may be overestimated by these surrogate assays as compared to the results obtained by full virus NT, currently considered the reference assay. To follow up on this notion, samples collected six months after second vaccination were re-tested with a full virus NT. For 95 samples interpretable results were obtained. Within these samples, the surrogate assays showed strong correlations with full virus NT (r² = 0.79, p < 0,001 for NeutraLISA, r²=0.77, p. < 0,001 for cPass) (Fig. 2A, B), which confirms previous studies . However, in the low range, positive-negative discrimination by the surrogate assays did not sufficiently match the results of full virus NT. Most notably, the surrogate assays yielded a significant number of false-positive results (5/89 in both tests) (Fig. 2A, B, inserts), suggesting that they may not be a safe companion diagnostic for long-termed monitoring of vaccination with mRNA-based vaccines such as Spikevax (Moderna).
Consequently, we addressed the question, which other diagnostic tool or staged strategy could possibly improve the safety of serologic monitoring of long-termed vaccination responses. First, we investigated whether a full virus NT titre ≥ 10 at six months after vaccination could possibly be predicted from the quantitative levels of S1-AB measured either directly or six months after second vaccination. S1-AB levels measured directly after second vaccination were poorly correlated with full virus NT obtained six months later (r² = 0.54, p < 0.001), which is expected given the equally poor correlation with quantitative S1-AB determined six month later (Fig. 1C). However, S1-AB levels measured six months after second vaccination exhibited a reasonably strong correlation with neutralizing capacity determined by full virus NT at the same time (r² = 0.79,
p < 0.001) (Fig. 3A), allowing to define a cut-off at 1000 U/ml discriminating a major portion (35/ 89 of the full virus NT-positive samples from all full virus NT-negative samples (Fig. 3A, insert). Incidentally, the fraction above that cut-off encompassed all samples having undergone infection in addition to double vaccination (Fig. 3B, black circles).
The remaining 63/89 samples below the cut-off (i.e. exhibiting S1-AB levels < 1000 U/ml six months after second vaccination) (Fig. 3B, symbols below dashed line) encompassed all six full virus NT – negative samples (Fig. 3B, closed red circles) but also 57 full virus NT – positive samples (Fig. 3B, open circles below dashed line). In search of a practical diagnostic tool allowing to discriminate within this group NT – negative and – positive samples, we reassessed the corresponding results of the surrogate assays for virus-neutralization. Upon re-adjusting the cut-off level of NeutraLISA and cPass to 64 and 72%, respectively, it was possible to thereby discriminate 5/6 true full virus NT – negative samples within the samples having S1-AB levels < 1000 U/ml (Fig. 4).
In summary, the staged diagnostic strategy applied six months after second vaccination detected five out of six full virus-NT negative samples, i.e., it had a sensitivity for presumably insufficient virus-neutralization capacity of 83.3%. As little as 14/89 (using NeutraLISA) or 6/89 (using cPass) were thereby falsely classified as virus-NT negative, i.e. corresponding specificity values were 84.2 and 96.2% for NeutraLISA and cPass, respectively.