Measurement of anti SARS-CoV-2 RBD IgG in saliva: validation of a highly sensitive assay and effects of the sampling collection method and correction by protein

Assay description

The assay developed is a two-step sandwich type immunoassay based on the amplified luminescent proximity homogeneous technology (AlphaLISA^®), and uses acceptor beads conjugated to a commercially available recombinant SARS-CoV-2 spike glycoprotein receptor binding domain antigen (Rekom Biotech, Spain) (RBD-AB), streptavidin-coated AlphaScreen donor beads (St-Dbeads), and a byotinilated mouse antibody specific for human IgG that showed no cross reactivity with human IgA, IgD, IgE or IgE (MABX1901-1KC, Merck, Germany) (Figure 1). All AlphaLISA^® reagents were purchased from Perkin Elmer (MA, USA) and conjugation of the beads was performed following manufacturer protocols.

Figure 1:

Schematic illustration of the AlphaLISA^® for anti-SARS-CoV-2 RBD IgG detection.

RBD-AB, acceptor bead conjugated to receptor binding domain; DB, donor bead; B-Anti hIgG, anti-human IgG biotinylated antibody.

Briefly, samples and standards (5 μL, pre-diluted 1:4,000 (serum) and 1:4 (saliva) in assay buffer) along with 20 μL of a mix containing 25 μg/mL of RBD-AB and 6 nM biotinylated mouse anti-human IgG in assay buffer 1× (AlphaLISA Assay Buffer 10×; PerkinElmer) were added to 96-well plates. The plates were covered with a lid and incubated at room temperature for 1 h. Then, in a second step, 25 μL of 40 μg/mL Streptavidin Donor beads (PerkinElmer) were added to each well and incubated for 30 min in a final reaction volume of 50 μL per well. After that the AlphaLISA^® signal (excitation at 680 nm, emission at 615 nm) was measured on a EnSpireAlpha plate Reader (PerkinElmer (MA, USA)).

A standard curve was developed by diluting an inactivated positive serum with alpha buffer up to different dilutions and results were given in arbitrary units (AU) per mL as previously reported [15, 16].

Analytical validation

Within run precision was calculated by measuring patient pools (from serum and saliva) with high and low IgG values 5 times in a single analytical run. The pools were made mixing samples from vaccinated and unvaccinated individuals to create samples with high and low concentration respectively. Between run precision was calculated by measuring the same pools in five different days. For the linearity study two samples were serially diluted (serum: 1:4,000; 1:8,000; 1:16,000; 1:32,000; 1:64,000 dilutions; saliva: 1:4; 1:8; 1:16; 1:32 dilutions) with the buffer used for the assay. The limit of detection was calculated as the lowest concentration of IgG which could be distinguished from a zero sample, and was taken as the mean + 3 standard deviations (SD) of 12 replicates of the blank (buffer used for the assay).

Sample collection and processing methods

To evaluate the effects of different sampling conditions, and correction by protein content, paired saliva and serum samples of 35 vaccinated participants (23 with BNT162b2, Comirnaty from Pfizer/BioNTech, 12 with ChAdOx1-S, Vaxzevria from Astra Zeneca) at three timepoints post-vaccination were collected: 48–72 h (T1), 3 weeks (T2) and 4 months post second dose (T3). All participant had received two doses of the vaccine separated 21 days in case of Comirnaty and 12 weeks in case of Vaxzevria and were not previously affected by COVID-19. They were 24 females and 11 males; average age (SD)=44 (11) years. Participants were requested to refrain from eating, drinking, and performing basic oral hygiene for 1 h before sample collection. All participants were informed of the purpose and experimental procedures of the study and signed a written informed consent form prior to their participation. The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of the University of Murcia (protocol code 2952/2020 and date of approval 22/02/2021) for studies involving humans.

To study the possible effect of the collection method on the results obtained, saliva samples were obtained under supervision by two different collection methods: passive drool and stimulated saliva by the use of a Salivette^® device (Sarstedt, Germany). In passive drool method, saliva was collected for 2 min by passive flow directly into a 15 mL conical and then centrifuged (4,000× g, 10 min, 4 °C). In stimulated samples participants were instructed to chew the cotton swab of the Salivette^® device during 1 min and then the swab was transferred into the Salivette^® tube and centrifuged at 4,500 rpm for 10 min. No samples showed blood contamination as determined by visual inspection. Paired blood samples were obtained immediately after saliva collection using standard procedures and transferred to tubes with a coagulation activator.

Serum was separated from blood by centrifuging at 3,000 rpm for 5 min. Inactivation of potential infectious viruses in serum and saliva was performed by incubation with NP-40 to a final concentration of 0.5% for 30 min and were stored at −80 °C until further analysis. We confirmed in pilot studies (data not shown) that NP-40 and storage at −80 °C up to 6 months had no effect on anti-RBD IgG titers.

Effect of total protein content of saliva

Results from anti-RBD IgG in each saliva sample were divided by its total protein content to evaluate the possible influence of protein content of each sample on the results. For this purpose a spectrophotometric assay (protein in urine and CSF, Spinreact, Spain) previously validated for human saliva [17] was used to measure the total protein content (mg/mL) of saliva. The method was adapted to an automatic analyzer (Olympus UA600, Olympus Diagnostica GmbH, Ennis, Ireland) following the manufacturer’s guidelines and showed less than 15% inter and intra-assay imprecision and was linear under serial dilution.

Statistical analysis

Arithmetic means, medians, 25th–75th percentile and CVs were calculated using standard descriptive statistical procedures and software (Excel 2013, Microsoft Corp., WA, USA,). For post-vaccination samples the D′Agostino & Pearson omnibus normality test was performed to evaluate the normality of distribution giving a nonparametric distribution. Therefore, data were log transformed to get normality and a one-way repeated measures ANOVA followed by uncorrected Fisher LSD test was performed to compared values obtained at different times after vaccine. A p-value<0.05 is considered to be statistically significant in all cases. Correlation coefficients between serum and salivary concentrations and between salivary collection methods ware determined using Spearman correlation analysis. The strength of the correlations tested was assessed by the Rule of Thumb [18], according to which an R value between 0.90 and 1 was considered to have very high correlation, 0.70 to 0.90 high correlation, 0.50 to 0.70 moderate correlation, 0.30 to 0.50 low correlation and less than 0.30 little if any correlation. To determine positivity, a cut-off value was generated by measuring 15 prepandemic serum and saliva samples. The cut-off was taken as the mean + 3 SD of this population as previously reported [10].

Statistical analyses were performed using GraphPad Prism 8 software (GraphPad Software Inc., CA, USA). A stand-alone power program (G-Power) was employed to compute achieved power (1-β error probability) with a significance level of α=5% in the IgG postvaccination study. The statistical power was higher than 0.99 for both serum and saliva samples.

Analytical validation of the AlphaLISA assay

Intra-assay CVs for IgG assay were in a range between 6 and 7% for serum and 6 and 13% for saliva. Inter-assay CVs were between 12 and 19% for serum and 13–15% for saliva (Tables 1 and 2). Dilution of two human serum and saliva samples with high IgG concentration resulted in linear regression equations, with determination coefficients close to 1 (Figure 2). The analytical limit of detection of the assay developed was 0.53 AU/mL.

Figure 2:

Linearity under dilution of two human serum (A) and saliva (B) samples serially diluted with the AlphaLISA^®-buffer.

R2, coefficient of determination.

Figure 3:

Linear correlation between IgG in serum and saliva obtained by two different collection methods.

r, Spearman correlation coefficient.

Effect of collection method

Significantly higher values of IgG anti RBD were observed at T2 in saliva obtained by flow and Salivette^® (median value: 104.3 AU/mL; 71.32 AU/mL; p<0.0001 for both flow and Salivette respectively) and also in serum (median value: 141,520 AU/mL; p<0.0001) in comparison with T1 (median values: 3.08; 0.84 and 18,960 AU/mL for flow, Salivette^® and serum respectively) and T3 (median values: 5.98; 2,08; 18,236 AU/mL for flow, Salivette^® and serum respectively) (Figures 3 and 4). In addition, significantly higher IgG values were found in T3 compared with T1 only in case of saliva obtained by both methods, but not in serum. A significant high correlation was observed between saliva obtained by the two different collection methods and serum (r=0.90; p<0.001 for passive flow and serum; and r=0.86; p<0.001 for Salivette^® and serum) (Figure 5).

Figure 4:

Saliva anti-SARS-CoV-2 RBD IgG.

The bars show median and whisker shows 75th percentile. T1, 48–72 h post second dose; T2, 3 weeks post second dose; T3, 4 months post second dose. F, passive drool; S, Salivette^®. Asterisks indicated significant differences between groups: *p<0.05; ****p<0.0001.

Figure 5:

Serum anti-SARS-CoV-2 RBD IgG.

The bars show median and whisker shows 75th percentile. T1, 48–72 h post second dose; T2, 3 weeks post second dose; T3, 4 months post second dose. Asterisks indicated significant differences between groups: ****p<0.0001.

When the results obtained in saliva by the two methods of collection were compared, significantly lower values of IgG anti RBD (p<0.0001) were observed in samples obtained with the use of Salivette^® device (median: 4.55 AU/mL) compared with those obtained by passive flow (8.7 AU/mL). In addition, a higher number of values under detection limit of the method were found when Salivette device was used (29 by Salivette vs. 10 by passive flow, specially at T1). A significant high correlation (r=0.86 p<0.0001) was found between samples obtained by both collection methods.

With regards to detection of individuals with IgG values higher that the cut-off, in saliva, a higher number of positives was obtained by passive flow compared with Salivette^® although the number of positives in saliva was lower than in serum, especially at T1 (Table 3).

Effect of total protein content of saliva

When data were corrected according to their total protein content, significantly higher values of IgG anti RBD per mg of protein were observed in T2 in saliva (p<0.0001) for both flow and Salivette in comparison with T1 and T3 (Figure 6).

Figure 6:

Saliva anti-SARS-CoV-2 RBD IgG corrected by total protein. The bars show median and whisker shows 75th percentile. T1, 48–72 h post second dose; T2, 3 weeks post second dose; T3, 4 months post second dose. F, Passive drool; S, Salivette^®. Asterisks indicated significant differences between groups: ***p<0.001; ****p<0.0001.

The IgG correlation between serum and saliva corrected by protein content was moderate (r=0.70, p<0.0001) when saliva was obtained by flow, showing a higher correlation with serum when saliva was used without correction (r=0.90; p<0.001). However, when Salivette^® was used (r=0.86, p<0.001) the correction by protein gave a similar correlation with serum than when no correction was used.

The correlation between values with and without protein content correction in samples obtained by passive flow and Salivette^® was high (r=0.85 and 0.88 respectively, p<0.0001).