Abstract
The SARS-CoV-2 variant of concern Omicron has increased infectivity and immune escape compared with previous variants. Despite a vast majority (∼90%) of the population of Santa Fe city (Argentina) had been vaccinated and/or had been infected by SARS-CoV-2 prior to the arrival of Omicron, the epidemic wave caused by this variant was by far the largest one experienced in the city. Nevertheless, the severity was much lower than that of previous waves. A serosurvey conducted within a month prior to the arrival of Omicron allowed to assess the humoural defences preceding the wave and to evaluate their consequences on infection risk and severity. Santa Fe is a city of 430000 inhabitants, the survey was conducted on 1452 citizens, 514 of which were followed-up until March 2022. A high proportion of the sampled individuals had immunological memory against COVID-19 at the arrival of Omicron (almost 90%), many of whom had high antibody levels. The anti-spike IgG titres were strongly associated with the number of vaccine shots and the vaccine platform received, and also depended markedly on prior COVID-19 diagnosis and the days elapsed since last antigen exposure (vaccine shot or natural infection). In turn, various analytical approaches consistently showed that preceding antibody titres were strongly correlated with COVID-19 incidence and severity of symptoms during the Omicron-dominant wave. Also, receiving a vaccine shot during the wave reduced the COVID-19 risk drastically (15-fold). Here we present real world data showing that COVID-19 incidence and severity during the Omicron-dominant wave was lowest in individuals with high antibody levels, which highlight the importance of maintaining high defences through vaccination in the presence of immune-escaping variants.
Introduction
As of May 2022, the COVID-19 pandemic continues to occur despite the acquired defenses developed in a large proportion of people due to having been infected and/or vaccinated against SARS-CoV-2. Several viral variants have evolved, prevailing the ones that achieved enhanced transmissibility and immune escape compared to prior variants (Tian et al. 2022). Until November 2021, some strains had become prominent and have caused new outbreaks worldwide. These were considered variants of concern, and were named Alpha, Beta, Gamma, and Delta. A new variant, B.1.1.529, was first detected in samples collected on 11th November 2021 in Botswana and on 14th November 2021 in South Africa. On 26th November, the WHO defined it as the fifth variant of concern, naming it Omicron (Tian et al. 2022). So far, Omicron is the variant with the largest number of mutations, many of which provide increased infectivity and immune escape compared with previous variants (Hu et al. 2022; Duong et al. 2022).
The dynamics of COVID-19 have been heterogeneous since the beginning of the pandemic (Beldomenico 2020). While countries like United Kingdom and Germany have gone through several epidemic waves, others like Thailand and Vietnam had their first wave only after over a year had passed since SARS-CoV-2 began to circulate in those countries. In Argentina, by early December 2021 there had been two waves, the first one by mid-2020, related to the arrival and spread of the virus, and the second one in 2021 associated with the seasonality of respiratory viruses. Omicron was confirmed in Argentina on 5th December 2021, and a few days later it triggered the largest epidemic wave that occurred in the country so far, with the number of daily cases at the peak being several times higher than in the two previous waves.
In Santa Fe city, the infection dynamics reflected what was observed elsewhere in the rest of the country (Figure 1). Santa Fe is a city of around 430.000 inhabitants. By mid-December 2021, 12.9% of the citizens had been diagnosed with COVID-19, 90.6% had received a first dose of an anti-SARS-CoV-2 vaccine, 79.2% a second dose, and 10.3% a third one (data provided by the Municipality of Santa Fe city). The Omicron-dominant wave in Santa Fe city began around 18th December 2022, and the number of daily cases started to decline by mid-January 2022, returning to levels as low as before the wave by the end of February (Figure 1).
During the last two months of 2021, we conducted a survey collecting relevant information on COVID-19 and measuring anti-spike IgG antibodies in randomly selected households of Santa Fe city and from citizens that volunteered to participate in the study. This provided the opportunity of having an assessment of the acquired humoural defences of the population of Santa Fe city immediately prior to the arrival of the Omicron-dominant wave. In March 2022, after the wave was over, the study participants were asked to complete a second questionnaire indicating if they were diagnosed with COVID-19 after 15th December 2021, if they got additional vaccine shots, and other relevant information. The data collected allowed us to pursue three goals:
describing the acquired humoural immunity of the population immediately prior the arrival of Omicron
assessing how such immune status was acquired (i.e. previous infection and/or different vaccination schemes), and
evaluating if those humoural defences predicted the risk and severity of infection during the wave.
Materials and methods
Source of the data
A random sample of 1000 households of Santa Fe city was provided by the Instituto Provincial de Estadísticas y Censos. Those households were visited from the end of October to mid-December 2021, and the occupants were invited to participate in a study that involved answering a first questionnaire and providing a blood sample to measure IgG antibodies against SARS-CoV-2. In addition, volunteers were invited to participate by announces in the local media. We collected data from 414 people from randomly selected households and 1041 volunteers.
The first questionnaire included queries on sex, age, having been diagnosed with COVID-19 (with dates), COVID-19 severity and duration, vaccine shots received (with type and dates), close contacts with COVID-19 cases, co-morbidities, among other information.
Those that were sampled after 15th November 2021, were asked to complete a second questionnaire in March 2022. This allowed us to follow the participants from whom there was an antibody measurement within a month prior to the Omicron-dominant wave. The second questionnaire inquired information for the period from 18th December of 2021 through 28th February 2022, including close contact with cases during that period, COVID-19 diagnosis, vaccine shots, and disease severity and duration. We obtained responses to the second questionnaire from 514 participants.
All procedures were carried out under the approval of the Ethics and Biosafety Committee of the Scientific and Technological Centre of Santa Fe of the Argentine Council for Research and Technology (CCT Santa Fe CONICET). All participants signed an informed consent.
Quantification of IgG
Levels of anti-SARS-CoV-2 spike protein IgG were quantified by COVID AR IgG immunoassay developed by Instituto Leloir in Argentina (Ojeda et al. 2021), following the manufacturer’s instructions. This IgG immunoassay kit consists of a solid phase ELISA that utilizes the trimer of native protein S and a receptor binding domain as antigens, obtained by recombinant DNA techniques produced in human cells.
Briefly, 40 µl of fingertip capillary blood samples were diluted 1:6 in the diluent provided in the SEROKIT developed by Instituto Leloir, and kept refrigerated. At the laboratory, 200 µl of each diluted blood sample was transferred to 96-well plates and incubated at 37°C for one hour. IgG specific for spike protein was captured on the plate, and subsequently the wells were thoroughly washed 6 times to remove unbound material. Anti-Human IgG, HRP-linked antibody was then used to recognize the bound IgG. A mix of HRP substrate and TMB (1:1) was added to develop color. The magnitude of optical density at 450 nm is proportional to the quantity of IgG specific for spike protein. To estimate antibody levels, sample optical densities were converted to concentrations expressed in UI/ml by using a lineal model built with the optical densities (response variables) obtained in each plate from two sets of known dilutions of the positive control at 50, 100, 200 and 400 UI/ml. These dilutions were the independent variable, included as a polynomial term (with lineal and quadratic terms), to address possible non-linearity of the dilution-OD relationship. The R2 of that model was checked to confirm that the value was >0.85, otherwise all samples were analysed again in a new plate.
Statistical analysis
All statistical analyses were done using the Software R version 4.2.0 (The R Foundation for Statistical Computing). The analyses were conducted in three steps, to pursue three complementary goals, as follows.
The first step aimed to characterise the acquired humoural defences in Santa Fe at the arrival of the Omicron-dominant wave. This part consisted of descriptive statistics of the IgG levels overall and by age group.
The second step’s goal was to assess vaccine performance in terms of IgG production. For this, we evaluated antibody titres as a function of number of vaccine doses, and comparing the most frequent vaccination schemes. The antibody levels were the response variables, which were transformed by calculating the square root to approach normality. Two sets of lineal models were run, one containing the polynomial term ‘number of doses + (number of doses)2’ as variable of interest (to take into account possible non-linearity of the titre-dose relationship), and the second set including only the 4 vaccination schemes most frequently observed, to compare antibody levels among them. In both models, the independent variables ‘COVID-19’ and ‘days from last exposure’ (vaccine shot or detected infection; whatever happened last), were included to control for potential confounding phenomenon. The vaccination schemes used for the second model were: two Astra-Zeneca vaccines (viral vector vaccine; N= 411), two Sinopharm (inactivated vaccine; N= 334), two Sputnik V vaccines (viral vector vaccine; N= 260), the combination of Spunik V and Moderna (viral vector + mRNA vaccines; N= 155), and two Pfizer/BioNTech (mRNA vaccine; N= 25).
The third step used information from the second questionnaire to conduct a longitudinal analysis that enabled assessment of how the vaccines and antibody levels influenced the incidence of COVID-19 during the Omicron-dominant wave in Santa Fe. In addition, we looked at associations between antibody levels and COVID-19 symptoms severity and duration among those that were infected during the Omicron-dominant wave. For this third step, the period in which participants were followed to assess new detected infections by SARS-CoV-2 was from December 18th to February 28th (72 days).
In order to establish an association between vaccination status and the incidence of COVID-19 during the Omicron-dominant wave, we built two Generalized Lineal Models (GLMs) with a binomial response (COVID-19 positive or not). The first model used number of vaccine doses as the independent variable of interest (levels= 0 - 4), and the second model was constructed with a subset that included the most frequent vaccination schemes, to compare among them. Based on the results obtained in the second step of our analyses (see below), the schemes evaluated in this second model were: vector × 2 (2 Astra-Zeneca or Sputnik V; N= 220), inactivated × 2 (2 Sinopharm; N= 118), vector + mRNA (Sputnik V + Moderna; N= 55), and vector × 3 (3 doses of either Astra-Zeneca or Sputnik V; N= 32). In view of the results obtained in the previous step and with the GLMs above, we constructed a third model that explored the effect of the days elapsed from the last vaccine shot to 1st February 2022 (when the wave was fading out quickly), taking into account only individuals that were vaccinated up to 20th January 2022 (N= 385).
To assess associations between antibody levels and the incidence of COVID-19 during the Omicron-dominant wave, we used two different approaches with subsets of data that excluded participants who received a vaccine shot between seven days prior of blood sample collection and 25th December 2021 (N= 484), so that the antibodies measured reflected the levels present at the onset of the Omicron-dominant wave. The first approach consisted of a GLM with a binary response that evaluated associations between antibody levels, while adjusting by a number of relevant factors, detailed below. The variable of interest was also included in a separate model as a dichotomous factor, setting those with antibody levels >400 UI/ml as 1, and the rest as 0. The second approach used Cox Proportional Hazard survival analysis to reinforce the findings of the GLM, in this case only using information of the days in which participants were expected to have antibody levels similar to what was measured. That is, a participant was followed until it was diagnosed with COVID-19 or until she/he received a new vaccine shot.
Finally, in the subset of samples that was diagnosed with COVID-19 during the wave (N= 174) the associations between antibody levels and COVID-19 severity and duration were assessed with ordinal regression models, where the responses were 3-level ordinal variables, as follows. Disease severity was measured by asking in the second questionnaire whether they had no or very mild symptoms (e.g. light sore throat, nasal congestion; level 1), mild symptoms (e.g. one or two days of fever and/or light malaise, not requiring bed rest; level 2), or moderate symptoms (e.g. bed rest was required; level 3). The participants were also asked if hospitalization was required, as a 4th level, but none chose this option. As for the duration of COVID-19 symptoms (excluding loss of smell), the three levels were: one day or less (level 1), two to five days (level 2), and more than 5 days (level 3).
For all models used in step 3, potential confounding phenomena was controlled for by including in the models relevant independent variables, as follows. Age (in years, and assessed separately as a single term or polynomial) was included in all models. Also in all models except for the Cox Proportional Hazard one, receiving a new vaccine during the wave period was included as a two-level independent variable, as those that got a booster shot within the follow up period had changes in both the antibody levels and the vaccination scheme. In the few cases in which the vaccine shot was received late in the wave (after 15th February 2022), the observation was discarded. The number of known close contacts with COVID-19 cases was used as a proxy of exposure, and used for the GLMs assessing associations with COVID-19 incidence. Close contact was defined as being within 3 m distance or indoors for over 15 minutes with someone who was diagnosed with COVID-19, and the contact happened within the period that went from two days prior the onset of symptoms and seven days after the onset of symptoms. The contacts were set at four levels, 0, 1, 2, 3, and 4 or more close contacts with cases. Prior COVID-19 was included in models that assessed associations between vaccination status and COVID-19 incidence during the wave (in these analyses, ‘suspect’ cases were removed). Finally, the presence of co-morbidities (high blood pressure, diabetes, obesity, heart disease, chronic pulmonary disease) was included in the models assessing the influence of antibody levels on COVID-19 severity and duration. All these variables used for adjustment purposes were dropped from the models if they were not important for the model’s goodness of fit, as indicated by AIC comparisons.
Results
Description of the sample
We obtained answers to the first questionnaire and blood samples from 1455 people. Of those, 57.3% were female and 43.7% were male. The mean age was 41 years old, being the minimum 5 months old and the maximum 95 years old.
Almost three quarters (74.7%) of the participants had not been diagnosed with COVID-19 at the time of answering the first questionnaire, but 2.4% of those suspected having been infected. One quarter (24.8%) was diagnosed with COVID-19 once, and 0.4% twice.
Regarding the vaccination regime, 6.9% of the participants were not vaccinated at the time of sampling, 5% had one dose, 83.6% had two doses, 4.5% had three doses and 0.07% had four doses. The vaccination scheme most frequently applied in the sample was two Astra-Zeneca vaccines, followed by two Sinopharm, two Sputnik V vaccines, and the combination of Spunik V and Moderna. At the time of the sampling, Pfizer vaccines were being used for youngsters aged 13 to 18 years old, having 25 participants of our study two doses, and 19 one dose.
Characterization of the acquired humoural defences prior to the Omicron-dominant wave
Anti-spike IgG were detected in 88.7% (1289/1453) of the samples. Among those that received at least one dose of an anti-COVID-19 vaccine, 7.4% (100/1354) did not have detectable IgG. Among the non-vaccinated (N=99), 63.6% (63/99) did not have detectable antibodies. Of the unvaccinated that had antibodies, 71% had not been diagnosed with COVID-19 nor suspected having been infected.
Figure 2 shows the distribution of the antibody levels measured in our sample. More than one third of the participants (35.4%) had antibody levels considered to be high (>400 UI/ml). The overall mean antibody level was 290 UI/ml, but it varied by age group (Table 1). Among the age group considered to be of high risk (>60 years old), the vast majority was vaccinated (98.1%), but 17% was vulnerable because they had no detectable IgG (6%) or had low antibody levels (11% with <40 UI/ml). However, most aged 60 and above had high antibody levels (65% with >400 UI/ml). The high level of antibodies observed in those aged 13-20 is due to the good performance of the vaccine received by that age group and the shorter time elapsed from the last shot.
Antibody levels according to vaccine doses and schemes
Those participants that were not vaccinated had a mean IgG titer of 62.5 UI/ml, while the mean level was 287.9 UI/ml, 293.5 UI/ml, and 567.8 UI/ml for those that receive one, two or three doses, respectively. A third dose increased the antibody levels significantly, and there was strong positive correlation with prior COVID-19 and strong negative correlation with days that elapsed from last exposure (vaccine or infection) (Figure 3, Table 2). The latter negative association reflects the quick waning of IgG levels.
When comparing the four vaccination schemes most frequently applied while adjusting by prior COVID-19 infection and days elapsed from last exposure, we observed very significant differences in antibody levels (Table 3; Figure 4). The scheme with inactivated vaccines showed significantly lower antibody levels than all other schemes, both schemes of viral vectors (Astra-Zeneca and Sputnik V) performed similarly, and the schemes combining vector and mRNA (Sputnik V + Moderna) and two mRNA (Pfizer) showed the highest levels, not statistically different between them (Table 3).
Vaccination status and COVID-19 incidence during the Omicron-dominant wave
The number of vaccine doses received prior to the arrival of Omicron did not have an effect on the COVID-19 infection risk during the wave (Table 4). A deeper analysis taking into account age, prior COVID-19, receiving a vaccine dose during the wave, and close contact with cases, showed that the number of doses was not associated with infection risk, although contact with cases and vaccination during the wave were strong predictors of COVID-19 incidence (Table 5). When comparing the vaccination schemes most commonly used, three doses of viral vector vaccines significantly reduced the probability of COVID-19 during the wave compared to all other vaccination schemes (Table 6). In addition, it was found that the time elapsed from the last vaccine shot was positively correlated with probability of COVID-19 during the wave. For every 10 days that passed from the last vaccine dose, the odds of being diagnosed with COVID-19 during the wave increased by 3% (Table 7).
Antibody levels and COVID-19 incidence during the Omicron-dominant wave
There was a strong negative correlation between antibody levels preceding the Omicron-dominant wave and COVID-19 incidence, both in the analysis with GLM (Figure 5, Table 8 and Table 9) and the Cox Proportional Hazard model (Table 10). Participants with antibody levels >400 UI/ml at the onset of the wave had 67% less chances of being diagnosed with COVID-19 during the wave (Table 9). At a given day, for every 100 UI/ml IgG increase in antibody levels, the risk of infection decreases 9%, adjusting by recent vaccine shot and contact with cases (Table 10). In addition, receiving a vaccine shot after the onset of the wave and the number of close contacts with cases were strong predictors of COVID-19 risk in all models.
Antibody levels and COVID-19 severity and duration
The ordinal regression model showed that antibody levels were strongly associated with the severity of the symptoms (Table 11, Figure 6). For every 100 UI/ml increase in the IgG level, the odds of being more likely to have higher disease severity (mild or moderate symptoms versus none or very mild symptoms) decreases 34.8%, holding constant new vaccine shot, age and presence of co-morbidities. The model looking at the association between antibody levels and duration of the symptoms showed a negative trend, but not statistically significant, although borderline (p=0.05; Table 12, Figure 7).
Discussion
The arrival of the variant Omicron was associated with the largest wave of COVID-19 cases in Santa Fe city and elsewhere in Argentina (Figure 1), despite immediately prior to the wave a vast majority of the citizens (>90%) had been vaccinated and/or had been infected by SARS-CoV-2. Here we characterised the acquired humoural defences in Santa Fe prior to the arrival of Omicron, estimating that a high proportion of the population had immunological memory against COVID-19 (i.e. almost 90% of our sample had detectable antibodies), many of whom had high antibody levels. Although the number of cases of the Omicron-dominant wave was much higher than in the previous two waves, the mortality due to COVID-19 was considerably lower. Before the Omicron-dominant wave there had been 856 deaths over 55969 cases (case-fatality= 1.5%), and during the wave there were 54 deaths over 36166 cases (case-fatality= 0.15%) (official records of Municipality of Santa Fe). This 10-fold lower impact could be attributable to the high level of defences here described, which is also supported by the finding of decreased disease severity as antibody levels increased. However, because there is evidence that suggests that Omicron may be less pathogenic than previous variants (Ulloa et al. 2022; Wolter et al. 2022) it is difficult to infer whether how much of the reduced severity is due to immunological experience and how much attributable to virus evolution making new variants less pathogenic (Bhattacharyya et al. 2022).
Antibodies provide protection either through direct obstruction of infection or through their ability to leverage the immune system to eliminate pathogens. In vaccine clinical trials, the neutralising antibody titre produced is highly correlated with protective effect and the durability of the protection (He et al. 2021). In this study it was found that the level of antibodies of the participants were strongly associated with the number of vaccine shots and the vaccine platform received, and also depended markedly on prior COVID-19 diagnosis and the days elapsed since last exposure (vaccine shot or infection). Those that received a third vaccine dose had much higher antibody levels than participants that got 2 or less shots at the time of sampling, which was caused by the booster effect already documented (Kanokudon et al. 2022), and the shorter time since the last shot. When adjusting by prior COVID-19 and days from last shot, vaccine schemes showed different performance in terms of immunogenicity. Two inactivated vaccines (Sinopharm) conferred the lowest antibody levels, and schemes that used mRNA platforms (Sputnik + Moderna or Pfizer × 2) the highest titres, whereas both vector vaccines (Astra Zeneca × 2 or Sputnik V × 2) performed between the other two schemes. This is in agreement with what was reported previously (Kanokudon et al. 2022; Kudlay & Svistunov 2022).
Despite the strong association between vaccination status and antibody levels, an association of the former with COVID-19 incidence during the wave was only apparent when comparing 3 doses of a vector vaccine versus 2 doses of different platforms (inactivated × 2, vector × 2, vector+mRNA; Table 6). This might result from the large variability in immunogenicity of the vaccines used and the strong correlation observed with days elapsed from the last vaccine shot (as those with 3 shots had been vaccinated more recently). It was documented that anti-spike IgG wane quickly (Bayart et al. 2021; Levin et al. 2021), and here we confirmed this in a real-world study (Table 2) and showed consequences of waning defences on infection risk (Table 7).
Prior studies have shown increased antibody evasion and greater breakthrough infection risk of Omicron, compared with previous variants (Mannar et al. 2022; Hu et al. 2022). However, although reduced, the binding of IgG antibodies to the Omicron Spike antigen is maintained, and recent data suggests that extraneutralising antibodies contribute to disease control (Bartsch et al. 2022). This partial immune escape implicates that higher defence levels would be required to reduce the risk and severity of COVID-19 caused by the Omicron variant. Here we present evidence of this with real world data. Anti-spike IgG levels and variables that cause antibodies to rise (i.e. prior COVID-19 and a recent boost shot) or are associated with higher antibody levels (i.e. days from last antigen exposure) were strong drivers of COVID-19 risk and severity. Our results strongly suggest that to reduce the impact of highly transmissible and immune-escaping variants like Omicron, there is need of keeping the defences high. Therefore, booster vaccine shots in anticipating or during the emergence of an epidemic wave are highly recommended.
In conclusion, the arrival of the Omicron variant caused the largest COVID-19 epidemic experienced in Santa Fe city since the beginning of the pandemic, but the case-fatality observed was 10-fold lower than that of previous waves. The increased number of cases may be caused by the immune escape and high transmissibility of Omicron while the high immune defences existing in the population at the time of its arrival most likely contributed the low impact observed. Disease risk and severity was lowest in individuals with high antibody levels, which highlight the importance of maintaining high defences through vaccination in the presence of immune-escaping variants.
Data Availability
All data produced in the present study are available upon reasonable request to the authors
Acknowledgements
This work was funded by the Agencia Santafesina de Ciencia, Tecnología e Innovación (Grant # DEMES-2020-0008). The Instituto Provincial de Estadísticas y Censos of Santa Fe province provided the random sample of households. Laboratorio Lemos, Instituto Leloir, Dr. Andrea Gamarnik and Dr. Marcelo Yanovsky donated part of the assays used in this work. Special thanks to all participants for accepting taking part in this study.