1. Validation of an in-house ELISA
We have estimated the qualitative (seropositive or seronegative) performances of our in-house ELISA among several COVID-19 RT-PCR positive samples. ROC analysis (Figure 1) gave us very satisfactory performances, the overall sensitivity was 96% CI95% [91.5%-98.5%] and the sensitivity after D14 is 95.9% CI95% [89.8%-98.9%] (93 true positive out of 97 patients sampled after D14). For specificity calculation, we have used 116 pre-pandemic sera collected in 2017 as true negative sera. Hence, the specificity of our in-house ELISA was 97.5% CI95% [92.8%-99.5%]. The accuracy of the test was very high 0.968 CI95% [0.939-0.985] and the positive and negative predictive values were 66.7% CI95% [39.6%-85.9%] and 99.8% CI95% [99.4%-99.9%], respectively. The comparison of our qualitative results to those of the Vidas® anti-RBD and Elecsys® anti-N showed better performances than the first (AUC 0,841 CI95%[0,727-0,883], p-value <0.05) but similar performances with the second (0,885 CI95%[0,808-0,938], p-value >0.05) (Figure 2). Therefore, we can consider our developed in-house ELISA is at least as good as the commercial techniques.
Then, we have compared the quantitative results of our in-house ELISA to those of the commercial cPass® GenScript® FDA-EUA. Correlation between anti-S titers and percentage of inhibition by the surrogate sero-netralization ELISA showed a strong correlation (r=0.5 CI95% [0.236-0.696], p-value<0.05) (Figure 3).
According to the cPass® a cut-off of 30% is recommended by the manufacturer as an indicator of the presence of virus neutralizing antibodies. In extrapolation to our in-house test, this corresponds to serums with titers of 59 BAU/ml.
2. Evaluation of the humoral immune responses induced by different COVID-19 vaccines.
The in-house validated ELISA for COVID-19 serology seems to be efficient in assessing the immune responses induced by the different vaccines used in the National Vaccination Program of Tunisia. We have analyzed 115 recipient of a COVID-19 vaccine and analyzed their immune responses by the quantitative in-house ELISA. Since most of the vaccine platforms are based only on the SARS-CoV-2 Spike and since a history of COVID-19 in vaccine recipients cannot be always confirmed, we have analyzed all the serum samples for anti-N antibodies. Hence, we have discriminated vaccine recipients of either BNT162b2, mRNA-1273, JNJ-78436735, Vaxzevria, or Sputnik V with Elecsys® anti-N. When it is positive, we consider the vaccine recipient previously SARS-CoV-2 infected and when it is negative as never infected by the virus. For vaccine recipients of inactivated viruses (Coronavax and Sinopharm), we have shown that the vaccine induced anti-N antibodies are almost undetectable. Therefore, in our analysis when we have high anti-N Elecsys® titres of Coronavax and Sinopharm, they are considered previously infected by the SARS-CoV-2.
All the results are compiled in Table 1, and they are grouped by vaccine brand and sub grouped according to a previous SARS-CoV-2 infection and the number of received shots. In addition, the titers were classified in different intervals and correspond to the percentages of each category of vaccine recipients in each interval. Based on Dimeglio et al. report (12), we have chosen the following intervals of antibody titers. Lower than 10 BAU/ml as seronegative; Between 10 and 50 BAU/ml as a poor immune response; Between 50 and 200 BAU/ml as borderline protective immune response; Between 200 and 400 BAU/ml as satisfactory immune response; Between 400 and 2,500 BAU/ml as high immune response; Between 2,500 and 10,000 BAU/ml as very high immune response; And higher than 10,000 as hyperimmune response.
In convalescent COVID-19 Patients sampled before the beginning of the vaccination campaign, only 1% mounted very or high immune responses in each group. They represent 11% with satisfactory immune responses. The remaining have either induced borderline protective immune responses (42.6%), or a poor response (35%) and even seronegative (10%). Mean antibody titer of all the convalescent patients was 118.23 BAU/ml.
For recipients of Pfizer/Biontech, when only one shot is administered and when their anti-N Elecsys® test were negative, the induced immune responses were either satisfactory or borderline in 16.7% and 25%, respectively. The remaining were either seronegative (16.7%) or mounted a poor immune response (41.7%). Overall, the mean antibody titer was 126 BAU/ml, roughly the same as in convalescent patients. When the second shot was administered, there is still 36.4% of them with borderline protective responses. The remaining of them have reached at least satisfactory immune responses with 45.5% as high responder. The mean antibody titer was 532 BAU/ml, largely higher than convalescent patients (p-value 0.0004). In the Tunisian COVID-19 vaccination program, whenever a person is previously confirmed infected, they only receive one shot of vaccination. In this category of vaccine recipients, we have also confirmed an eventual previous infection by an anti-N Elecsys® positive test. Hence, in this category of Pfizer/Biontech vaccine recipients, hypeimmunized (16.1%) and inducer of very high (61.3%) or high (13%) immune responses were detected. Only 6.5% and 3% of them induced rather satisfactory or poor immune responses, respectively. The mean antibody titer was 7,962 BAU/ml, very largely higher than convalescent patients (p-value 0,000000) and to two doses of Pfizer/Biontech vaccine COVID-19 negative recipients (p-value 0.043). When people received two shots of Pfizer/Biontech vaccine, normally they should have not been previously infected by the SARS-CoV-2. However, since sometimes there are lacks of transparency from vaccine recipients, or due to the occurrence of a mild form of the disease not easy to diagnose, they can be administered two shots of Pfizer/Biontech vaccine. Nevertheless, with the anti-N Elecsys® positive test, we have been able to identify them. Although, only six of them are here represented, half of them have mounted rather borderline protective immune responses. The remaining have mounted either high, or very high or hyperimmune responses (one recipient in each category). Their mean antibody titer was 3,064 BAU/ml.
Recipients of Moderna mRNA COVID-19 vaccine are only represented by 5 individuals, all of them previously infected by the SARS-CoV-2, as confirmed by a positive anti-N test and their mean antibody titer was 1,707 BAU/ml. They have received only a single shot of vaccine and they have mounted either high (four recipients) or very high (one-recipient) immune responses.
For individuals that have received inactivated virus vaccines we have plotted the results of both brands (Sinovac and Sinopharm). When receiving only one shot of the vaccine, most of them (66.7%) mounted borderline immune responses. The remaining either mounted a satisfactory immune response or were seronegative, at rates of 16.7% in each category. Their mean antibody titer was 132 BAU/ml very close to that of convalescent patients. When two doses of inactivated virus vaccines were administered, there is still 16.7% of them seronegative. The remaining either mounted poor or borderline immune responses (41.7%, each group). Their mean antibody titer was 49 BAU/ml, non-statistically different to that of convalescent patients (p-value 0.439) and to one shot recipient (p-value 0.06). In presumably infected persons, based on a positive anti-N serology, inactivated virus vaccine recipients induced almost the same level of immune responses, whether they received one, two or three shots. Hence, more than 40% of them induced high immune responses and around 40% satisfactory antibody titers. Their pooled mean antibody titer was 458 BAU/ml, higher to convalescent patients (p-value 0.0004) and lower to Pfizer/Biontech COVID-19 positive recipients after one shot (p-value 0.027) or two shots (p-value 0.05).
For vaccines based on recombinant adenoviruses, only few recipients could be recruited in this investigation. After vaccination with one or two shots of AstraZeneca vaccine, SARS-CoV-2 infected individuals mounted at least high level of antibody titers. With Johnson and Johnson vaccine only three recipients were recruited, two of them did not seroconvert and the third only mounted a borderline protective response. Finally, with Sputnik V vaccine, non-infected recipients induced a rather borderline response in at least 60% of them. Out of two infected individuals and vaccinated with Sputnik V, there is one with a very high level of induced immune response. The second recipient induced a satisfactory immune response.