Evidence of prior SARS-COV-2 infection by binding antibodies and neutralisation
Our study population in Lagos, Nigeria who received at least one dose of the AZ1222 vaccine comprised 140 participants with a median age of 40 (interquartile range: 33, 48), 73 (52%) of whom were males. In order to analyse the proportion of participants in this urban population previously exposed to SARS-COV-2, we tested all baseline samples (n=140) for anti-N IgG using a flow cytometry based assay18 and found 62/140 participants were positive prior to administration of first vaccine dose, demonstrating 44% SARS-CoV-2 anti-N IgG seroprevalence at baseline prior to vaccination. Using the anti-RBD algorithm, 21/78 (27%) anti-N IgG negative subjects were additionally positive for anti-RBD IgG prior to vaccination – yielding a total seroprevalence of 83/140 (59%).
Our study population from Kumasi, Ghana, enrolled prior to vaccination comprised 527 participants with a median age of 33 (interquartile range: 25, 47), 295 (56%) of whom were males. We tested all baseline samples (n=527) for anti-N IgG by flow cytometry 18 and found 147 participants were positive, demonstrating 28% SARS-CoV-2 anti-N IgG seroprevalence at baseline prior to vaccination. Using RBD positivity, we found additional 77/382 (20%) anti-N negative participants were positive for anti-RBD IgG prior to vaccination indicating a previously waned anti-N IgG and exposure proportion of 224/527 (43%).
To explore this phenomenon further we used neutralisation assays as described previously17. Baseline neutralising GMT (geometric mean titre) of ID50s in the Nigerian study population when stratified by anti-N status was 431 vs 47 in IgG anti-N positive and negative participants respectively, suggestive of the presence of neutralizing antibodies against SARS-CoV-2 in subjects negative for SARS-CoV2 anti-N Ab prior to vaccination. Of the 24 individuals anti-N Ab negative at baseline, 12/24 had ID50 above the cut-off of 20. In these individuals, binding antibodies to S were also detectable, and neutralisation correlated with IgG anti-S and IgG anti-RBD levels (r=0.71 and r=0.73) respectively indicating prior infection in at least half of those who were N Ab negative at baseline (Supplementary Figure 1a). In the Ghanaian population, the baseline GMT of ID50s against WT PV in the study population when stratified by anti-N status was 106 vs 45 in IgG anti-N positive and negative participants respectively, again suggestive of the presence of neutralizing antibodies against SARS-CoV-2 in subjects negative for SARS-CoV2 anti-N Ab prior to vaccination. Of the 32 individuals anti-N Ab negative at baseline, 15/32 had ID50 above the cut-off of 20. In these individuals, binding antibodies to S were also detectable, and neutralisation moderately correlated with IgG anti-S and IgG anti-RBD levels (r=0.57 and r=0.55) respectively also indicating prior infection in at least half of those who were N Ab negative at baseline (Supplementary Figure 1b). Both the anti-Spike binding antibody and neutralisation analyses were consistent and suggest that N antibody underestimates the true prevalence of prior infection by approximately 50%.
Waning of N antibody and reinfections over time
12/49 (24.5%) became anti-N IgG negative at 1 month post second-dose and no further participants lost anti-N IgG positivity between 1- and 3-months post second-dose. Of note, one subject who was anti-N IgG positive at baseline, became anti-N IgG negative at 1 month and then became anti-N IgG positive 3 month post second-dose with a 7-fold increase in anti-N IgG titres between 1 month and 3 months post second-dose - strongly suggestive of re-infection. In Ghanaian participants we similarly observed loss of IgG anti-N in 7 of 45 (15.6%) participants with follow up serum samples at 1 month post vaccination.
Longitudinal neutralising and binding antibody responses following vaccination
Of the 140 Nigerian participants recruited (Figure 1), 49 had plasma samples available at baseline prior to vaccination and at two follow-up timepoints post vaccination for neutralization assays (Table 1). Median age was 39 (31, 46) and 47% were male. Half of the participants, 25/49 (51%) were IgG anti-N positive at baseline, and the geometric mean titre (GMT) of neutralizing antibodies associated with 50% neutralization (ID50) against WT PV across the entire study population was 145 ± 4.5(GMT±s.d) (Table 1; Figure 2a) with significantly lower titres observed against the Delta and Omicron variants with GMT titres 75 ± 3.6(GMT±s.d) and 55 ± 3.0(GMT±s.d) (p=0.0001 and p<0.0001) respectively. Amongst Ghanaian participants, of the 527 participants recruited (Figure 1), 45 had plasma samples available at baseline prior to vaccination and at two-month post vaccination for neutralization assays (Table 1). Almost one-third of participants were IgG anti-N positive at baseline, and the geometric mean titre (GMT) of neutralizing antibodies associated with 50% neutralization (ID50) against WT PV across the entire study population was 57 ± 3.0 (GMT±s.d) (Table 1; Figure 2a) with significantly lower titres observed against the Delta and Omicron variants with GMT titres 37±2.4 (GMT±s.d) and 29±1.8 (GMT±s.d) (p<0.001 and p<0.001) respectively.
Overall, neutralizing Ab titres to WT 1 month after second dose in Nigerian participants were 2579 ± 4.2(GMT±.s.d). As expected, lower levels of neutralization were observed against the Delta [549 ± 3.7(GMT±.s.d); p<0.0001)] and Omicron variants [269 ± 3.4(GMT±.s.d); p<0.0001] at 1 month, representing a fold reduction of 4.7 and 9.6 respectively (Figure 2). The GMT for Delta and Omicron was only around 100, nearly a log lower in comparison to WT (Figure 2). Positive anti-N IgG Ab status at baseline was associated with significantly higher titres of neutralizing antibodies following vaccination across all tested VOC (Figure 2). Importantly, those with anti-N Abs present at baseline did not experience waning of responses between months 1 and 3 post second dose (Supplementary Figure 2).
In Ghana, neutralizing Ab titres to WT PV two-month after second dose were 1049± 6.7(GMT±.s.d). Lower levels of neutralization were observed against the Delta [453 ± 7.4 (GMT±.s.d); p<0.0001)] and Omicron variants [95 ± 5.3(GMT±.s.d); p<0.0001] at 2 month, representing a fold reduction of 2.3 and 11.0 fold respectively (Figure 2).
As observed in the Lagos population, positive anti-N IgG Ab status at baseline was associated with significantly higher titres of neutralizing antibodies following vaccination across tested VOC except the Omicron variant (Figure 2). When we compared neutralisation in N negative participants post second dose, the GMT for WT were: 1423 ± 3.9 (GMT±.s.d) in Nigerian participants and 773 ± 7.4 (GMT±.s.d) in Ghanaian participants respectively.
Waning of neutralising responses
Overall, there was no decline in neutralising antibody titres at 3 months for WT, Delta, or and Omicron compared to 1 month post vaccination in Nigerian participants (Figure 2a). By contrast, when data were stratified by anti-N IgG status at any timepoint, there was a significant decline in neutralization between 1 month and 3 months post second-dose across all variants tested for participants who were N antibody negative throughout (Figures 2 and 3, p=0.04). The GMT in these individuals for Delta and Omicron was approximately 100, nearly a log lower in comparison to WT (Figure 2B). Participants with anti-N Abs present at baseline did not experience waning of responses between months 1 and 3 post second dose (Supplementary Figure 2a and 2c), despite frequent loss of N antibody over time (Supplementary Figure 3). When we examined binding antibodies over time in the group as a whole, we saw very small decreases for Wu-1 and Omicron Spike IgG but not for Wu-1 RBD (Supplementary Figure 4a). When we analysed the data for those N antibody negative, waning of binding antibodies was more evident (Supplementary Figure 4b, Supplementary Figure 5). We were not able to assess waning due to absence of samples beyond one month post second dose in Ghana.
Vaccine breakthrough infection
To evaluate the proportion of participants with vaccine breakthrough infection after two doses of AZD1222 vaccine, we tested anti-N IgG in subjects who were anti-N IgG negative at baseline (n=78) and became positive between 1- and 3-months post second dose and found 7/49 (14%) with de-novo infection, with one additional participant demonstrating both reinfection and breakthrough infection to yield a total breakthrough rate of 8/49 (16%, Figure 4 and Supplementary Figure 5). These individuals also experienced increase in antibodies to S and RBD that mirrored N antibody dynamics (Figure 4). We were also able to measure binding antibodies to Omicron that were around a log lower in titre as compared to Wu-1 binding antibodies as expected (Figure 3).
To investigate whether suboptimal immune response was related to subsequent breakthrough, we compared the neutralizing antibody titres 1 month post second dose between those with (n=8) or without breakthrough infection (n=15). We found no significant difference in neutralization between the groups (p=0.36, Figure 5a left panel). However, and as expected, neutralizing titres were higher at the last time point in individuals who had experienced vaccine breakthrough infection (with no evidence of infection prior to vaccine), indicating a boosting effect of infection in addition to vaccine (Figure 5a right panel). We noted that the increase in titres against Delta PV observed in breakthrough was significantly greater than the increase for WT and Omicron PVs, coincident with the Delta wave of infection in mid 2021.