Vaccine breakthrough infections with SARS-CoV-2 Omicron induced a higher level of protection compared to triple vaccination and contributed to herd immunity on a population level. To address the underlying immunological mechanisms, we studied the evolution of SARS-CoV-2-specific antibody and Tcell responses during vaccination and upon breakthrough infection in Bavarian residents between February 2021 and December 2022. Further, we investigated the temporal distance between completed vaccination and break-through infection, as well as any occurring re-infection. Each vaccination significantly increased peak neutralization titers against Wuhan, Delta, and Omicron BA.5 with simultaneous increases in circulating spike-specific Tcell frequencies. After vaccination, Omicron BA.5 neutralization titers were most significantly associated with a reduced hazard rate for SARS-CoV-2 infection, also when accounting for spikespecific Tcell responses. Yet, 97% of triple vaccinees became SARS-CoV-2 infected, often within a few months after their third vaccination. Breakthrough infections further boosted neutralization magnitude and breadth, broadened virusspecific Tcell responses to non-vaccine-encoded antigens and protected with an efficiency of 88% from further infections by December 2022. This effect was then assessed by utilizing mathematical modelling, which accounted for time-dependent infection risk in Bavaria, as well as the antibody and Tcell concentration at any time point after breakthrough infection. Our findings suggest that cross-variant protective hybrid immunity induced by vaccination and breakthrough infection was an important contributor to the reduced virus transmission observed in Bavaria in late 2022 and thereafter.
Research Article
Evolution of protective SARS-CoV-2-specific B- and T-cell responses upon vaccination and Omicron breakthrough infection
https://doi.org/10.21203/rs.3.rs-3411238/v1
This work is licensed under a CC BY 4.0 License
Journal Publication
published 30 Apr, 2024
Version 1
posted
You are reading this latest preprint version
The emergence and dominance of the SARS-CoV-2 Omicron variant, in late 2021, resulted in high infection rates across populations with high vaccination coverage (1, 2). In Germany, roughly 63% of the population was vaccinated by summer 2021 and then boosted in autumn/winter 2021 (3). Likewise, a large proportion of the Bavarian population and in particular health care workers started SARS-CoV-2 vaccinations in the first half of 2021 and were boosted with a third RNA vaccination between October and December 2021, just before the emergence of Omicron in Bavaria (1). Despite the high vaccination rates, SARS-CoV-2 Omicron rapidly replaced Delta by efficiently spreading among vaccinated individuals during multiple waves until December 2022, albeit having a lower risk of severe COVID-19 (4, 5). Since then, the reported case numbers substantially declined and remained on a comparatively low level as of beginning of 2023.
Parameters defining protective immunity to Omicron are likely comparable to those established in the pre-Omicron era; neutralizing antibody titers positively correlated with protection from SARS-CoV-2 infection (6, 7, 8, 9). Antibody transfer using spike (S)-specific IgG were sufficient to mediate protection from SARS-CoV-2 infection in a preclinical infection model (8). We have recently reported that nucleocapsid (N)-specific Tcell responses associated with control of primary SARS-CoV-2 infection in the upper airways before antibody seroconversion (10). Moreover, memory T cells mediated protection from infection in several pre-clinical models; vaccine-induced airwaytissueresident T cells targeting the N protein protected mice upon SARS-CoV-1 challenge (11). Similarly, intranasal vaccination with N and membrane protein (M, both are structural proteins of the virion) encoding vectors induced protective Tcell responses in a non-human primate model (12). Indeed, accumulating evidence suggests that SARS-CoV-2-specific Tcell responses contribute to control of infection (13, 14, 15). Furthermore, intramuscular vaccinations followed by BTI result in a state of “hybrid immunity”, characterized by high S-specific antibody responses, broadening of virus-specific Tcell responses to non-vaccine-encoded antigens and formation of virus-specific Tcell memory in the airway epithelium (16,17, 18). Yet, while BTI induces a high degree of protective immunity, the adaptation of antibody and Tcell responses after BTI and their contribution to reduce infection spread within the population and development of further waves is incompletely understood. We therefore investigated the dynamics of SARS-CoV-2-variant-specific neutralizing antibody and T-cell responses against vaccineencoded and non-encoded viral antigens during vaccination, as well as before and after Omicron BTI spanning the transition from the pandemic to endemic phase (February 2021 to December 2022). Mathematical modelling was then used to assess the protective effect of antibody and T-cell responses upon hypothetical re-exposure to neutralizationsensitive and neutralizationresistant SARS-CoV-2 variants, respectively.
Study population
To study vaccine-induced SARS-CoV-2-specific antibody and Tcell responses, study participants (n = 50) were recruited from Munich and surrounding regions in Bavaria, Germany. Figure 1 and Table 1 summarize the characteristics of the study population including vaccination types, SARS-CoV-2 infection events, the observation time for each individual and also shows the concurrent SARS-CoV-2 incidence in Bavaria over the study period (February 2021 to December 2022). The median age was 31 years (range 21–57 years) and 68% (34/50) were female. Out of 50 individuals, 18 were recruited before their first vaccination, 23 before administration of the second vaccination and an additional 9 participants were recruited after the second vaccination dose. 34 subjects (indicated in the underlying data) were followed longitudinally over an extended period from the first quarter of 2021 until December 2022. 16 participants were followed only for a limited time or lost to follow up before receiving a third vaccination. mRNA vaccines accounted for the majority of first immunizations, followed by adenoviral vector vaccines. Second and third booster vaccines were almost exclusively mRNA vaccines. Of these 34 vaccinated subjects, 33 were subsequently SARS-CoV-2 infected between November 2021 and December 2022 with a median time of 157.5 days after the last vaccination (range: 10–406 [days]). 31 of 33 individuals received three vaccinations and then were infected, while two participants were infected after the second vaccination. Infecting SARSCoV-2 variants were determined by imputation. One individual was determined to be infected with Delta, eight subjects with Omicron BA.1, 17 subjects with Omicron BA.2 and 14 were imputed to be infected with BA.5 or BA.5 sub-lineages. All infections induced N-specific antibody seroconversion (Supplementary Fig. 1, source data provided), and none were associated with hospitalization.
| n | 50 |
|---|---|
| Sex % Male Female | |
| 32% (16/50) | |
| 68% (34/50) | |
| Median Age (Range) | 30 (21–57) |
| Vaccination 1 (B/M/A/J)1 | 32/1/11/6 |
| Vaccination 2 (B/M/A/J)1 | 40/6/1/0 |
| Vaccination 3 (B/M/A/J)1 | 31/9/0/0 |
| Vaccination 4 (B/M/A/J)1 | 2/0/0/0 |
| Median time between vaccination and infection (days, range) | 157.5 (10–406) |
| Infection strain (Imputed)* (D/BA. 1/2/5/BQ. 1)2 | 1/8/17/13/1 |
| 1 B = BioNTech, M = Moderna, A = AstraZeneca, J = Janssen | |
| 2 D = Delta strain, BA.1/2/5/BQ.1 = Omicron BA.1, BA.2, BA.5, BQ.1 strain, | |
| *Imputed using the infection date and the Bay-VOC database. The strain with the highest likelihood is presented here. | |
97% of triplevaccinated study participants, who were followed up until December 2022, experienced a BTI within 14 months of their last vaccination, while only four of these became infected a second time during the third and fourth quarter of 2022. In summary, three vaccinations were insufficient to provide protection from symptomatic Omicron infection. Omicron breakthrough infection protected with a high efficiency of 88% (30 out of 34) from a second infection by December 2022 (Fig. 1). When limiting the observation time to 230 days after three vaccinations or BTI, a difference can be observed where the majority vaccinated were infected thereafter. However, after this BTI, only one subject was reinfected a second time within 230 days of the first BTI (Fig. 2).
SARS-CoV-2 neutralizing antibody and Tcell response dynamics during vaccination
We assessed the dynamics of SARS-CoV-2 neutralizing activity against three clinically relevant variants of concern (VOCs) by determining the neutralizing antibody titers (IC50) against Wuhan, Delta, and Omicron BA.5 (Fig. 3A-C, source data provided). The median, interquartile ranges, and number of tested subjects are provided for all study visits grouped in time bins (Supplementary Table 1). Neutralizing antibody responses against all three variants followed a similar dynamic pattern, albeit at different intensity. IC50 values for Wuhan neutralization were consistently higher as compared to Delta or Omicron BA.5, whereas Delta and Omicron BA.5 neutralization was comparable after vaccination (Supplementary Table 1). Each vaccination induced a peak response that was at least 11- to 21-fold higher compared to the last measured time point before vaccination (Supplementary Table 1). Beyond 150 days after the second vaccination titers had waned to pre 2nd vaccination levels, while the third vaccination boosted median IC50 titers 10-15-fold (depending on the tested variant) and remained significantly higher beyond 150 days as compared to the levels after the second vaccination for all tested variants (p < 0.001, n = 10, excluding participants with BTI before 3rd vaccination). We next wanted to assess neutralization breadth for different variants and combined singlevariant data using a magnitude-breadth analysis approach(17, 18), in which areas under the curve (AUC) were determined for every single serum sample, based on the neutralization data against all three variants. A slight, but statistically significant increase in neutralization magnitude-breadth was observed after each vaccination, dwindling quickly within 90 days (Fig. 3D).
Next, we assessed the dynamics of T-cell responses against the vaccineencoded spike (S), and the non-vaccineencoded nucleocapsid (N) and membrane (M) antigens (source data provided). Representative gating for CD4 + or CD8 + IFNɣ+ T cells is shown in Supplementary Fig. 2. 31%, 72% and 94% of vaccinees had detectable S-specific Tcell responses after the first, second and third vaccination respectively and median frequencies of circulating S-specific CD4 + T cells increased from a median of 0.002% of CD4 T cells at baseline to 0.006% at 8–35 days after the 1st vaccination, and to 0.017% and 0.032% at 8–35 days after the 2nd and 3rd vaccination, respectively and slightly declined thereafter (Fig. 4A). Frequencies of circulating S-specific CD8 + T cells followed a similar dynamic pattern (Fig. 4B), albeit with overall fewer IFNɣ+ cells.
BTI induces persisting high levels of cross-neutralizing antibody responses and Tcell reactivity to structural proteins of the virion
Next, we assessed how BTI affected neutralization of the three tested SARS-CoV-2 variants and of spike, nucleocapsid and membrane-specific Tcell responses. BTI induced a 1.5- to 3.3-fold increase in IC50 values (depending on the variant) compared to peak levels at 8 to 35 days after the third vaccination. Neutralization remained high beyond 150 days (Fig. 3A-C; Supplementary Table 1); even beyond 150 days after BTI median IC50 neutralization values were comparable to peak levels directly after the third vaccination for all three tested variants. Likewise, the BTI increased the magnitude-breadth by a substantial degree (2-fold, p < 0.0001, Fig. 3D), resulting in a sustained response for at least 150 days, exceeding what has been observed during the same time frame, after the third vaccination (2-fold, p = 0.052). The longitudinal IC50 values against the three variants tested (Wuhan, Delta and Omicron BA.5) for the 34 subjects followed up is shown in Supplementary Fig. 3.
Omicron BTI also boosted S-specific CD4 + Tcell frequencies up to a median of 0.041% after BTI (at 8–35 days, Fig. 4, Supplementary Table 2), which persisted beyond 150 days at 11-fold higher median compared to prevaccination levels. Simultaneously, S-specific CD8 + T cells increased after BTI (Fig. 4B), confirming our previous results from the pre-Omicron era (17). Moreover, peak frequencies of vaccine-induced S-specific CD8 + T cells after the 3rd vaccination are strongly correlated with post-BTI frequencies; at 8–35 days: r = 0.8, p < 0.0001; at 36–90 days: r = 0.5, p < 0.04; and > 150 days: r = 0.7, p = 0.01 (Supplementary Fig. 4). Similar associations were also found for S-specific CD4 + Tcell responses (8–35 days: r = 0.4, p = 0.03 and > 150 days: r = 0.6, p = 0.02, Supplementary Fig. 4). BTI elevated CD4 + Tcell responses targeting the non-vaccineencoded nucleocapsid from background frequencies of 0.002% (median) at baseline to 0.025% at 8–35 days post BTI (Fig. 5A). In 17 of 18 participants, N-specific CD4 + Tcell responses remained above median baseline frequency beyond 150 days after BTI (Fig. 5A, Supplementary Table 2). Similar patterns were observed for N-specific CD8 + T cells and membrane-specific Tcell responses (Fig. 5B-D). These results show that Omicron breakthrough infection was associated with persistent high levels of cross-neutralizing antibody responses to spike, as well as broad and persistent Tcell reactivity to the non-vaccineencoded virion antigens N and M.
Vaccine-induced peak neutralizing antibody IC50 titers are associated with risk reduction of Omicron variant infection
In order to address whether vaccine-induced antibody and/or Sspecific Tcell responses at least temporarily protected from SARS-CoV-2 Omicron variant infection; we estimated hazard ratios using Cox regression. As covariates in our model, we separated study volunteers into two groups based on levels of antibody IC50 values or S-specific CD4 + or CD8 + Tcell frequencies at peak response (8 to 35 days after third vaccination). The cutoff values for each covariate were optimized to separate between "protected" and "non-protected" individuals.
Omicron BA.5 neutralization IC50 values above 400 were more significantly associated with a reduced hazard for Omicron infection (HR [95% Cl] = 0.27 [0.1–0.7], p = 0.007, Fig. 6A), compared to Delta (IC50 values above 506, HR [95% Cl] = 0.3 [0.11–0.81], p = 0.019, Fig. 6B) and Wuhan neutralization (IC50 values above 1992, HR [95% Cl] = 0.4 [0.16-1], p = 0.05, Fig. 6C). Of note, for many participants and time points the IC50 values for the Wuhan variant were often above the dynamic range of the neutralization assay, which likely affected this hazard ratio calculation. Circulating S-specific IFNɣ+ CD4 + or CD8 + T cell frequencies were not significantly associated with a change in the infection risk (Supplementary Fig. 5A and B). A multivariate Cox-regression model considering Omicron BA.5 neutralizing antibody levels, S-specific CD4 + and S-specific CD8 + Tcell frequencies, showed a significant risk reduction only for Omicron BA.5 neutralizing antibody levels (HR [95% CI] = 0.3 [0.11–0.8]) (Fig. 6D). Together, these results demonstrate that a vaccineinduced crossneutralizing antibody response was associated with a significant reduction in hazard of Omicron infection when accounting for S-specific CD4 + and S-specific CD8 + Tcell responses.
Effect of neutralizing antibodies on the infection risk against Omicron BA.5
We utilized a mathematical model to determine the role of neutralizing antibodies in protection from Omicron infection during the observed time period and protection from future re-infection by an unknown Omicron strain. A schematic representation of the model can be found in Fig. 7A. This model describes the infection risk of individuals based on (i) their time-dependent antibody IC50 level and (ii) the time-dependent number of infectious individuals in the general population. More precisely, the infection risk is expressed by a hazard rate h(t) which is influenced by the IC50 neutralizing antibody titers of the individual (Antibodies(t)) at any given time and the seven-day SARS-CoV-2 infection incidence (Incidence(t)) per 100,000 inhabitants in Bavaria (Fig. 7A). In this model higher Antibodies(t) reduce infection risk, whereas a higher Incidence(t) increases it. The model contains two parameters: the base risk of infection for the Omicron lineage, β0, and the estimated effect of antibody levels on infection risk, β1. Infection risk parameter were chosen based on the literature (19, 20, 21, 22).
The effect of antibody levels on infection risk, β1, was inferred from the timetoinfection data from individuals as well as the matching antibody levels. Using these parameters (β0 and β1), the infection risks for individuals can be studied and visualized. In Fig. 7B example trajectories of the seven-day incidence (left panel) and IC50 neutralizing antibody titers (right panel) are shown for three representative individuals prior to an infection. Although only a minor variation in the risk of infection was observed between individuals 1 and 3, individual 2 clearly diverges and appears to be protected during the peak incidence in Bavaria between February and May. The higher neutralizing antibody titers observed within individual 2 as compared to individual 1 and 3 potentially contributed to protection (Fig. 7B, right panel).
As the base risk of infection for the Omicron lineage, β0, is not precisely known, we compared different literature estimates of the base risk of infection and our estimated effect β1 of antibody levels on infection risk (Fig. 7C, right panel). Considering BA.5 neutralization data from all 31 BTIs, the effect did not depend strongly on the base infection risk factor β0, leading to similar infection risk reduction for any given antibody IC50 value (Supplementary Fig. 6). The different estimates for β1 result in a reduction of infection risk that quickly rises with an increase in the neutralizing antibody titers. Antibody levels at the upper detection limit (IC50 value of 2560) reduce the risk of infection by almost 100%.
Finally, after determining the effect of the antibodies on the infection risk within our cohort for triple-vaccinated individuals, we utilized the BA.5 IC50 neutralizing antibody titers after BTI (assuming these correlate with neutralization of other Omicron variants (23)) to simulate the risk of a second infection with a SARS-CoV-2 Omicron variant (Fig. 7D). In our simulation, the increased neutralizing antibody levels after BTI strongly reduce the infection risk within 180 days compared to three vaccinations. The model predicts several infections within these 180 days, which were however not observed in our study volunteers.
Modeling the effect of pre-existing nucleocapsid-specific T cells on upper airway viral load upon hypothetical re-infection with a neutralization resistant SARS-CoV-2 variant
To estimate the effect of circulating SARS-CoV-2-specific T cells on the upper airway viral load (UA-VL) during a hypothetical second infection with a completely neutralization resistant virus variant, we employed a linear mixed-effects model previously described by Eser et al. (10). A schematic comparison of estimated viral load levels within 7 days after secondary infection between a patient with or without pre-existing memory SARS-CoV-2 N-specific CD4 + T cells is shown in Fig. 8A. Using measurements from this study we have modeled the frequency of pre-existing Nspecific T after a hypothetical second infection at 180 days after the initial BTI (Fig. 8B). Assuming that no expansion of pre-existing N-specific T cells occurred upon second infection (i.e., the fold expansion is equal to one), a reduction of UA-VL between 7.3% and 23.0% is predicted (Fig. 8C). As highly variable expansion of pre-existing influenza M and N-specific T cells was observed during human Influenza challenge experiments with an average of tenfold at day seven we have investigated the impact of it on the results of the model (24). According to our model, an estimated 10-fold expansion of Nspecific CD4 + T cells, could reduce acute infection UA-VL by 77.5% (53.0%-92.7%) on average compared to previously non-exposed individuals (Fig. 8C). In summary, pre-existing N-specific CD4 + IFNɣ+ T cells induced by a first BTI are likely to associate with a multi-fold reduction of acute phase viral load upon a re-infection compared to first infection.
We investigated the evolution of peripheral SARS-CoV-2-specific antibody and Tcell responses in a longitudinal Bavarian cohort during vaccination and upon subsequent breakthrough infection from 01 February 2021 to 31st of December 2022. The study data were then utilized to assess the risk for re-infection as a function of neutralizing antibody titers and – in case of re-infection with a hypothetical, completely neutralization resistant new variant – to estimate the effect of pre-existing SARSCoV-2specific memory Tcell responses on the upper airway viral load. Despite triple vaccination, 97% of the study participants experienced breakthrough infections within one year of the last vaccination. Re-infection only occurred in 4 individuals consistent with high protective efficiency against re-infection after Omicron BTI. Peak Omicron BA.5 neutralization IC50 titers after the third vaccination were the strongest correlate of transient protection from Omicron infection, while S-specific CD4 + and CD8 + Tcell frequencies did not correlate with a reduced infection risk. Surprisingly, 42% of vaccinees with peak BA.5-IC50 values above 400 (and likely higher IC50s for BA.1/2) were infected within 120 days after the third vaccination in early 2022 (25). During the early phase 3 vaccine efficiency trials, even peak “Wuhan” IC50 values above 100 were associated with a protective efficiency of above 90% (3, 5). Lower protective IC50 in our study may be explained by the fact that phase 3 vaccine trial participants, at that time encountered a virus strain more homologous to the vaccine. In addition, extraordinarily high numbers of virus encounters during the Omicron BA.1/2 wave likely contributed to these BTIs in the presence of high neutralizing antibody titers. Of note, our mathematical model suggested that Omicron BA.5 IC50 titers above 400 reduce the infection risk by more than 90% against Omicron BA.5 infection, which is in the similar range as described for the phase 3 vaccine trials of the pre-Omicronera (26, 27, 28, 29, 30). One possible explanation for the difference in our results compared to the clinical trials could therefore be the high incidence and associated high number of virus encounters in Bavaria in early 2022, which was multi-fold higher as compared to July 2020 to March 2021, when these trials were conducted. Protection from SARS-CoV-2 Omicron infection in Bavaria in early 2022 therefore likely required higher titers and breadth of neutralizing antibodies as compared to the preOmicron era. Some methodological variability such as different amounts of input virus in the neutralization assay may also contribute to these differences in protective titers.
Differences in incident infection dynamics between high and low titer groups became more apparent beyond 4 months after the 3rd vaccination, which, for most participants, coincided with the onset of spring-summer season and a seasonassociated reduction of infection incidence, until emergence of the more virulent BA.4/5 and its sub-lineages (BQ.1, BF.1 and BQ.1.1) starting from May/June 2022 (data obtained from BayVOC) (31).
Omicron BTIs further increased neutralization against the three tested variants significantly. While Wuhan and Delta-specific neutralization was also enhanced, the strongest effect was observed for Omicron BA.5 neutralization, consistent with previous findings. Furthermore, our magnitude-breadth analysis show significantly increased breadth of neutralization after breakthrough infection likely conveying improved protection from re-infection despite the evolution of novel more neutralization resistant variants. Our mathematical model further supported the concept that high neutralizing antibody titers after Omicron BTI contribute to protection from subsequent re-infection. Interestingly, the protection in our cohort was higher than predicted by our model considering only the antibody levels reached after a BTI, consistent with a protective role of cell-mediated immunity against re-infection.
Even long after breakthrough infection, we observed broad memory Tcell recognition against the three tested virus antigens S, M, and N, similar to what was observed in non-BTI Wuhan infection(15, 32) and these - according to our model - possibly contributed to the high level of protection observed in our cohort after BTI. Indeed, pre-existing T cells, in particular those targeting the N protein, have been associated with protection from SARS-CoV-2 infection(13, 14, 33). Moreover, N-specific T-cell responses are also associated with control of SARS-CoV-2 in the upper airways and reduced systemic inflammation (10). The effect of pre-existing T cells on virus control and early Tcell expansion was studied after controlled Influenza challenge of human volunteers (24). Key results from the Eser and Wilkinson studies (10, 24) were therefore utilized to estimate the effect of preexisting SARS-CoV-2-specific memory T cells on upper airway viral load upon a hypothetical re-infection with a neutralization resistant SARS-CoV-2 variant. Assuming that the T-cell expansion during SARS-CoV-2 behaves similar to Influenza and the effect of T cells on virus clearance is unchanged between the first and second infection, individuals with circulating SARS-CoV-2-specific T cells, in particular those targeting N, likely control the neutralization resistant virus efficiently. This process may depend on the level of pre-existing virus-specific memory T cells and should reduce onward transmission and may even abort infection early (13). Both primary and BT infection also result in the formation of airway tissueresident memory T cells, which differs from intramuscular vaccination (16, 18). BTI can therefore be considered as airway immunization. It should be noted that, vaccination induced formation of Tcell memory in the airway tissues is a key determinant for infection outcome in small rodent and non-human primate studies (11, 12, 34, 35). One recent clinical study showed that after BTI, SARS-CoV-2-specific tissueresident memory T cells persist in the upper airways at similar levels up to 3 months, but by month 6 have declined (16). This decay of mucosal adaptive immunity and the recruitment dynamics of virus-specific tissueresident T cells into the airways upon reexposure likely influence the outcome upon virus reexposure and need to be investigated in future studies.
The presented study had several limitations. We did not test neutralization with the autologous infecting virus variant, as most participants were infected during the Omicron BA.1/2 wave. Moreover, Tcell responses were only tested against three immunodominant antigens in blood, and not tested within the airway tissues. Furthermore, the mathematical models used for data analysis provided only a simplified description of real-world processes. The model for the assessment of the infection risk makes, for instance, use of information on the timevarying risk of infection (based on incidence) with a high-time resolution, but constructs estimate for each individuals’ antibody trajectory from observations with low time resolution, which introduces some uncertainty. However, by simulating infections of several individuals from our model, we were able to confirm that the model accurately describes the data presented herein. Further, an accurate account of the exact number of infected people in Bavaria is difficult to obtain, due to reporting issues and only the official numbers reported by the RKI were used in this study. Nevertheless, the infection process in this study could be reliably recapitulated using our model. Thus, we believe the model provides valuable insights into the risk reduction through neutralizing antibodies against SARS-CoV-2.
In conclusion, this study revealed that Omicron BTI occurred in 97% of a triple vaccinated study population and induced a state of hybrid immunity characterized by higher cross-reactive, more persistent neutralizing antibody responses and broader SARS-CoV-2-specific Tcell responses. These characteristics aligned with a high level of protection from re-infection, likely reflecting events on a population scale in Bavaria. It should be noted that Omicron waves occurred until October 2022 in the Bavarian population but waned thereafter despite the onset of the winter season. These data therefore suggest that BTIinduced hybrid was an important contributor to the reduced virus transmission observed in the early winter of 2022.
Study subjects and design
A total of 50 participants – residing in Bavaria – were recruited into this study between February 2021 and February 2022 under the umbrella of the longitudinal KoCo19 Study (36). 41 of these were recruited before or shortly after receiving their first SARS-CoV-2 vaccination during the vaccination campaign in the first half of 2021, while additional 9 volunteers were recruited at later time points. Subjects were then followed longitudinally for up to 18 months until December 2022. Whole blood and plasma samples were collected during multiple visits before and after first, second and third vaccinations and after SARS-CoV-2 breakthrough infection. Written informed consent was obtained from all participants in accordance with the principles of the Declaration of Helsinki. This study is part of the ORCHESTRA project (Connecting European Cohorts to increase common and Effective SARS-CoV-2 Response) and was approved by the Ethics Committee of the Faculty of Medicine at LMU Munich (20–371).
Assessment of SARS-CoV-2 strains
SARS-CoV-2 breakthrough infections within this cohort occurred from November 2021 to December 2022 and were confirmed by SARS-CoV-2 N-specific seroconversion. The infecting SARS-CoV-2 variant was imputed using the Bavarian Variants of Concern (Bay-VOC) database at weekly resolution and using > 50% likelihood of a specific variant as a cutoff (https://www.bay-voc.lmu.de/surveillance.xhtml). This database tracks the sequence of the SARS-CoV-2 variants of concern using data provided by hospitals and institutes in the region of Bavaria, Germany throughout the pandemic (starting from November 2020).
Pseudovirus neutralization assay
The capacity of sera to neutralize different SARS-CoV-2 variants was determined using a lentiviral pseudotype assay, as described previously (32, 37). In brief, an inoculum of 105 rlu/well of lentiviral particles expressing luciferase and pseudotyped with the according variant-specific SARS-CoV-2 spike protein was neutralized with a 2-fold serum dilution series (starting from 1 in 20) for 1 h. Luciferase activity was determined 48 h post infection of HEK293T-ACE2-cells using BrightGlo (Promega Corp, Madison, WI, USA). The 50% inhibitory concentration (IC50) of the sera was calculated with Prism 8 GraphPad (San Diego, CA, USA) after normalizing to non-infected and infected cells. Magnitude-breadth analysis was performed as described by others (38). In brief for each serum and time point, a “survival curve” was calculated, with the “event” of interest being neutralization of one variant and the “time-to-event” representing the titer required to get 50% neutralization. Despite data censoring, areas under the curve were calculated to compare different sera or groups, as with three variants, the AUC provided a better estimate than the 50% maximum Magnitude-Breadth.
Binding antibody titers
Serological assays to test for SARS-CoV-2-specific binding antibodies were performed as previously published (36, 39). EDTA plasma was used to quantify binding antibodies specific for S and N protein using Roche Elecsys anti-nucleocapsid (Ro-N-Ig) (Roche, Mannheim, Germany). All assays were performed according to manufacturer’s instructions. A value above 0.8 counts on the Ro-N-Ig was considered as a positive response towards the SARS-CoV-2 nucleocapsid.
Flow cytometry
PBMCs were isolated within 6 h of blood collection via density gradient centrifugation (Cytiva Sweden AB) and stimulated immediately with peptide pools representing the nucleocapsid, spike and membrane proteins of SARS-CoV-2 (1 µg/ml/peptide, Miltenyi Biotec) for 16 h at 37oC in the presence of anti-CD28 (clone L293, 1 µg/ml, BD Biosciences), anti-CD49d (clone L25, 1 µg/ml, BD Biosciences), and brefeldin A (5 µg/ml, Sigma-Aldrich). Negative control wells lacked stimulants, and positive control wells contained staphylococcal enterotoxin B (SEB, 0.6 µg/ml, Sigma-Aldrich). Cells were then stained with anti-CD4–ECD (clone SFCI12T4D11, Beckman Coulter), anti-CD8–APC-AF750 (clone B9.11, Beckman Coulter), anti-CD57–APC (clone HNK-1, BioLegend), anti-PD1–PE-Cy5.5 (clone NAT105, BioLegend), and anti-CXCR5–PE-Cy7 (clone J252D4, BioLegend). Labelled cells were fixed and permeabilized using a FoxP3 / Transcription Factor Staining Buffer Set (eBioscience) and further stained intracellularly with anti-CD3–APC-AF700 (clone UCHT1, Beckman Coulter), anti-IFN-γ–FITC (clone 4S.B3, BioLegend), anti-IL2–PE (clone MQ1-17H12, BioLegend), anti-TNF-α–BV510 (clone mAb11, BioLegend), anti-CTLA-4–BV421 (clone BNI3, BioLegend), anti-Ki-67–BV605 (clone Ki-67, BioLegend), and anti-CD40L–BV785 (clone 24–31, BioLegend). Samples were acquired using a CytoFLEX Flow Cytometer (Beckman Coulter). Data analysis was performed using FlowJo software version 10 (FlowJo LLC). Background subtraction was performed by subtracting IFNɣ+ T cell frequencies in the negative control from those in the antigen stimulated sample.
Statistical modelling
Basic statistical analyses were performed using non-parametric tests in Prism version 8 (GraphPad). For Kaplan-Meier curves the patients that have not completed threefold vaccination and those who were observed for less than 230 days after third vaccination or after BTI, respectively, were excluded from the data used for the curve. The line depicting post third vaccination behaviour is based on 32 individuals, post BTI curve – on 18 subjects. The percentage was calculated with respect to the number of patients meeting the criteria; the R package survival was used for the analysis (40).
In order to understand the effect of antibodies on reduction of infection risk, Cox regression with proportional hazards assumption was performed with the R survival package (Fig. 6, Supplementary Fig. 5). 95% confidence intervals for hazard ratios were obtained with the R package gtsummary (41). Here, a hazard ratio below 1 indicates a reduction in hazard by presence of antibodies, CD4 + or CD8 + T cells.
Infection model
To estimate the effect of neutralizing antibodies on the risk of getting infected after three vaccinations against SARS-CoV-2 an infection model was created. For this model, we assumed two states “susceptible” (S) and “infected” (I). We derived the unconditional probability density of becoming infected within time t according to the framework used in standard survival analysis by applying a hazard rate that is tailored to an infection process. Hence, our approach differs from the semi-parametric Cox regression. The time-dependent hazard rate for the infection is assumed to follow the equation:
Here, Incidence(t) is the seven-day incidence per 100,000 inhabitants in Munich at time t (source RKI COVID Datenhub, https://npgeo-corona-npgeo-de.hub.arcgis.com/). The base risk of infection β0 was fixed according to estimates of contact rates from SEIR models (19, 20, 21). Additionally, the hazard rate is influenced by the IC50 neutralizing antibody titers Antibodies(t) against the SARSCoV-2 Omicron variant BA.5. The reductive effect of the antibodies on the infection risk is captured by the linear scaling factor β1 in the term 1 – (tanh (β1 × log (Antibodies(t))). Large values of β1 reduce the risk of infection, whereas values close to 0 would indicate no effect. Furthermore, the term tanh (β1 × log (Antibodies(t))) can be used to describe the percent reduction in infection risk conveyed by the neutralizing antibodies. Note that tanh(x) becomes negative if x falls below zero. However, this never occurred in this study. Additionally, we tested our model by replacing tanh with the logistic function and obtained similar results.
We estimated β1 from infection times after three vaccinations of the study volunteers. The neutralizing antibody levels of each individual were estimated beforehand using a linear mixed effects model. Here, we assumed an exponential decay of the antibodies after third vaccination. For estimation of the parameters of the mixed effects model, the R package lme4 was used (42). The infection model was implemented in julia and the parameters were estimated using the julia interface to the NLopt optimizer library (https://github.com/JuliaOpt/NLopt.jl) (43). After having estimated the effect of the antibodies, we compared model simulations after third vaccination to the observed data by drawing infection times for each study volunteer from the previously derived infection time distribution. Given these simulations, we calculated the 2.5% and 97.5% simulation quantile at each observed infection time for visualization. With good agreement of model and data, the model was used to predict secondary infections with Omicron BA.5 after BTI for several individuals. Therefore, we used measurements of neutralizing antibodies after BTI, reported COVID-19 cases in Munich and estimates for β0 (≈ 11.55) and β1 (≈ 0.36), to draw the time of infection within 180 days after BTI for the 30 individuals of this study for 1000 times each from the infection time distribution.
Reduction in viral load by pre-existing nucleocapsid specific T cells
In order to estimate the influence of pre-existing N-specific CD4 + IFNɣ+ T cells, we first estimated the average number of pre-existing T-cells at 180 days after BTI using an exponential decay model. For the estimation, we employed the R package nlme (44). The model was calibrated on the flow cytometric measurements of CD4 + IFNɣ+ T-cells after BTI assuming a peak response at 14 days after infection (10). Using this model, the average T-cell concentration could be calculated at 180 days after BTI. At that time a secondary infection was assumed, which would lead to the expansion of T-cells. We tested different expansion factors for seven days after infection. Given the expanded T-cell population at day seven after infection, we used a linear mixed effects model developed by Eser et al. (10) to estimate the reduction in upper airway viral load between two virtual individuals: one with pre-existing N-specific CD4 + IFNɣ+ T-cells and one without. For the individual without pre-existing T-cells, the Tcell concentration between day four and seven after symptom onset measured by Eser et al. was used. This measurement was taken to be a baseline T-cell level and added to the expanded T cell concentration of the individual with pre-existing T cells. Then, the model from Eser et al. was evaluated for both virtual individuals and the ratio of viral loads was calculated.
The authors declare no competing interests.
Acknowledgment:
We would like to thank all study participants. This study would also not have been possible without the passionate contribution of the staff of the Division of Infectious Diseases and Tropical Medicine at the University Hospital, LMU Munich, Helmholtz Zentrum München.
Funding:
We are grateful for the financial support of the Bavarian State Ministry of Science and the Arts (KoCo19)
Deutsche Forschungsgemeinschaft, Germany (reference number: GE 2128/3-1 and HO 2228/12-1)
Bavarian State Ministry of Science and the Arts, Germany (Project “Langfristige Nachverfolgung der zellulären und humoralen Immunantwort nach SARS-CoV-2-Infektion und Impfung” (490 2021) and Project “Eindämmung, Behandlung und Erforschung der Erkrankung mit dem neuartigen Coronavirus COVID-19” (FOR-COVID))
Ministry for Education and Research, Germany (proj. nr.: 491 01KI20271)
The ORCHESTRA project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 101016167. The views expressed in this paper are the sole responsibility of the author and the Commission is not responsible for any use that may be made of the information it contains.
Author Contributions:
RW, SE, AS, JHe and GC contributed to the experimental work and analyzed neutralizing antibody responses.
MA, TE and KH contributed to the experimental work and analyzed T cell responses.
RR and AW analyzed binding antibody responses.
CP, MHu and JH developed the mathematical models.
MA, SE, JH, CP, NC, CG and OB performed the statistical analysis and prepared figures.
The human cohort study was planned by CG, IK, AW, LO, and MH (KoCo19-sub-study).
CG, JH and MA conceived the study.
MA, CJ, and IK contributed to the clinical work. All authors contributed to study data discussion, writing and revision of the manuscript.
Code availability
All code is available on Zenodo.
ORCHESTRA/KoCo19 working group: Mohamed Ibraheem Mohamed Ahmed, Emad Alamoudi, Jared Anderson, Maximilian Baumann, Marc Becker, Marieke Behlen, Jessica Beyerl, Rebecca Böhnlein, Isabel Brand, Anna Brauer, Vera Britz, Jan Bruger, Friedrich Caroli, Lorenzo Contento, Max Diefenbach, Jana Diekmannshemke, Paulina Diepers, Anna Do, Gerhard Dobler, Jürgen Durner, Ute Eberle, Judith Eckstein, Tabea Eser, Volker Fingerle, Jonathan Frese, Felix Forster, Turid Frahnow, Günter Fröschl, Christiane Fuchs, Mercè Garí, Otto Geisenberger, Leonard Gilberg, Kristina Gillig, Philipp Girl, Arlett Heiber, Christian Hinske, Janna Hoefflin, Tim Hofberger, Michael Höfinger, Larissa Hofmann, Sacha Horn, Kristina Huber, Christian Janke, Ursula Kappl, Charlotte Kiani, Arne Kroidl, Michael Laxy, Ronan Le Gleut, Reiner Leidl, Felix Lindner, Silke Martin, Rebecca Mayrhofer, Anna-Maria Mekota, Hannah Müller, Katharina Müller, Dafni Metaxa, Leonie Pattard, Michel Pletschette, Michael Pritsch, Stephan Prückner, Konstantin Pusl, Peter Pütz, Ernst-Markus Quenzel, Katja Radon, Elba Raimúndez, Camila Rothe, Nicole Schäfer, Yannik Schälte, Paul Schandelmaier, Lara Schneider, Sophie Schultz, Mirjam Schunk, Lars Schwettmann, Heidi Seibold, Peter Sothmann, Paul Stapor, Fabian Theis, Verena Thiel, Sophie Thiesbrummel, Niklas Thur, Julia Waibel, Claudia Wallrauch, Franz Weinauer, Simon Winter, Julia Wolff, Pia Wullinger, Houda Yaqine, Sabine Zange, Eleftheria Zeggini, Thomas Zimmermann, Anna Zielke.
- Flechsler J, Eberle U, Dangel A, Hepner S, Wimmer C, Lutmayr J, et al. Molecular SARS-CoV-2 surveillance in Bavaria shows no Omicron transmission before the end of November 2021. Infection. 2022;50(3):761–6.
- Taylor L. Covid-19: Omicron drives weekly record high in global infections. BMJ. 2022;376:o66.
- Maier BF, Wiedermann M, Burdinski A, Klamser PP, Jenny MA, Betsch C, et al. Germany's fourth COVID-19 wave was mainly driven by the unvaccinated. Commun Med (Lond). 2022;2:116.
- Varianten bei COVID-19 Patienten in Bayern 2022 [Available from: https://www.bay-voc.lmu.de/surveillance.xhtml.
- Petrone D, Mateo-Urdiales A, Sacco C, Riccardo F, Bella A, Ambrosio L, et al. Reduction of the risk of severe COVID-19 due to Omicron compared to Delta variant in Italy (November 2021 - February 2022). Int J Infect Dis. 2023;129:135–41.
- Gilbert PB, Montefiori DC, McDermott AB, Fong Y, Benkeser D, Deng W, et al. Immune correlates analysis of the mRNA-1273 COVID-19 vaccine efficacy clinical trial. Science. 2022;375(6576):43–50.
- Feng S, Phillips DJ, White T, Sayal H, Aley PK, Bibi S, et al. Correlates of protection against symptomatic and asymptomatic SARS-CoV-2 infection. Nat Med. 2021;27(11):2032–40.
- Corbett KS, Nason MC, Flach B, Gagne M, O'Connell S, Johnston TS, et al. Immune correlates of protection by mRNA-1273 vaccine against SARS-CoV-2 in nonhuman primates. Science. 2021;373(6561):eabj0299.
- Mercado NB, Zahn R, Wegmann F, Loos C, Chandrashekar A, Yu J, et al. Single-shot Ad26 vaccine protects against SARS-CoV-2 in rhesus macaques. Nature. 2020;586(7830):583–8.
- Eser TM, Baranov O, Huth M, Ahmed MIM, Deak F, Held K, et al. Nucleocapsid-specific T cell responses associate with control of SARS-CoV-2 in the upper airways before seroconversion. Nat Commun. 2023;14(1):2952.
- Zhao J, Zhao J, Mangalam AK, Channappanavar R, Fett C, Meyerholz DK, et al. Airway Memory CD4(+) T Cells Mediate Protective Immunity against Emerging Respiratory Coronaviruses. Immunity. 2016;44(6):1379–91.
- Ishii H, Nomura T, Yamamoto H, Nishizawa M, Thu Hau TT, Harada S, et al. Neutralizing-antibody-independent SARS-CoV-2 control correlated with intranasal-vaccine-induced CD8(+) T cell responses. Cell Rep Med. 2022;3(2):100520.
- Swadling L, Diniz MO, Schmidt NM, Amin OE, Chandran A, Shaw E, et al. Pre-existing polymerase-specific T cells expand in abortive seronegative SARS-CoV-2. Nature. 2022;601(7891):110–7.
- Kundu R, Narean JS, Wang L, Fenn J, Pillay T, Fernandez ND, et al. Cross-reactive memory T cells associate with protection against SARS-CoV-2 infection in COVID-19 contacts. Nat Commun. 2022;13(1):80.
- Wyllie D, Jones HE, Mulchandani R, Trickey A, Taylor-Phillips S, Brooks T, et al. SARS-CoV-2 responsive T cell numbers and anti-Spike IgG levels are both associated with protection from COVID-19: A prospective cohort study in keyworkers. medRxiv. 2021:2020.11.02.20222778.
- Lim JME, Tan AT, Le Bert N, Hang SK, Low JGH, Bertoletti A. SARS-CoV-2 breakthrough infection in vaccinees induces virus-specific nasal-resident CD8 + and CD4 + T cells of broad specificity. J Exp Med. 2022;219(10).
- Ahmed MIM, Diepers P, Janke C, Plank M, Eser TM, Rubio-Acero R, et al. Enhanced Spike-specific, but attenuated Nucleocapsid-specific T cell responses upon SARS-CoV-2 breakthrough versus non-breakthrough infections. Front Immunol. 2022;13:1026473.
- Poon MML, Rybkina K, Kato Y, Kubota M, Matsumoto R, Bloom NI, et al. SARS-CoV-2 infection generates tissue-localized immunological memory in humans. Sci Immunol. 2021;6(65):eabl9105.
- Volz E, Mishra S, Chand M, Barrett JC, Johnson R, Geidelberg L, et al. Assessing transmissibility of SARS-CoV-2 lineage B.1.1.7 in England. Nature. 2021;593(7858):266–9.
- Ito K, Piantham C, Nishiura H. Relative instantaneous reproduction number of Omicron SARS-CoV-2 variant with respect to the Delta variant in Denmark. J Med Virol. 2022;94(5):2265–8.
- Dandekar R, Barbastathis G. Neural Network aided quarantine control model estimation of global Covid-19 spread. 2020.
- SPI-M-O: Consensus Statement on COVID-19: Scientific Advisory Group for Emergencies; 2021 [Available from: https://www.gov.uk/government/publications/spi-m-o-consensus-statement-on-covid-19-3-june-2021.
- Tran TT, Vaage EB, Mehta A, Chopra A, Tietze L, Kolderup A, et al. Titers of antibodies against ancestral SARS-CoV-2 correlate with levels of neutralizing antibodies to multiple variants. NPJ Vaccines. 2022;7(1):174.
- Wilkinson TM, Li CK, Chui CS, Huang AK, Perkins M, Liebner JC, et al. Preexisting influenza-specific CD4 + T cells correlate with disease protection against influenza challenge in humans. Nat Med. 2012;18(2):274–80.
- Cheng SS, Mok CK, Li JK, Ng SS, Lam BH, Jeevan T, et al. Plaque-neutralizing antibody to BA.2.12.1, BA.4 and BA.5 in individuals with three doses of BioNTech or CoronaVac vaccines, natural infection and breakthrough infection. J Clin Virol. 2022;156:105273.
- Polack FP, Thomas SJ, Kitchin N, Absalon J, Gurtman A, Lockhart S, et al. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N Engl J Med. 2020;383(27):2603–15.
- Baden LR, El Sahly HM, Essink B, Kotloff K, Frey S, Novak R, et al. Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine. N Engl J Med. 2021;384(5):403–16.
- Falsey AR, Sobieszczyk ME, Hirsch I, Sproule S, Robb ML, Corey L, et al. Phase 3 Safety and Efficacy of AZD1222 (ChAdOx1 nCoV-19) Covid-19 Vaccine. N Engl J Med. 2021;385(25):2348–60.
- Sadoff J, Gray G, Vandebosch A, Cardenas V, Shukarev G, Grinsztejn B, et al. Safety and Efficacy of Single-Dose Ad26.COV2.S Vaccine against Covid-19. N Engl J Med. 2021;384(23):2187–201.
- Logunov DY, Dolzhikova IV, Shcheblyakov DV, Tukhvatulin AI, Zubkova OV, Dzharullaeva AS, et al. Safety and efficacy of an rAd26 and rAd5 vector-based heterologous prime-boost COVID-19 vaccine: an interim analysis of a randomised controlled phase 3 trial in Russia. Lancet. 2021;397(10275):671–81.
- Kimura I, Yamasoba D, Tamura T, Nao N, Suzuki T, Oda Y, et al. Virological characteristics of the SARS-CoV-2 Omicron BA.2 subvariants, including BA.4 and BA.5. Cell. 2022;185(21):3992–4007 e16.
- Cantoni D, Mayora-Neto M, Nadesalingam A, Wells DA, Carnell GW, Ohlendorf L, et al. Neutralisation Hierarchy of SARS-CoV-2 Variants of Concern Using Standardised, Quantitative Neutralisation Assays Reveals a Correlation With Disease Severity; Towards Deciphering Protective Antibody Thresholds. Front Immunol. 2022;13:773982.
- Jay C, Ratcliff J, Turtle L, Goulder P, Klenerman P. Exposed seronegative: Cellular immune responses to SARS-CoV-2 in the absence of seroconversion. Front Immunol. 2023;14:1092910.
- Teijaro JR, Verhoeven D, Page CA, Turner D, Farber DL. Memory CD4 T cells direct protective responses to influenza virus in the lungs through helper-independent mechanisms. J Virol. 2010;84(18):9217–26.
- Manfredi F, Chiozzini C, Ferrantelli F, Leone P, Pugliese K, Spada M, et al. Induction of SARS-CoV-2 N-specific CD8 < sup>+</sup > T cell immunity in lungs by engineered extracellular vesicles associates with strongly impaired viral replication. bioRxiv. 2023:2023.01.19.524762.
- Radon K, Bakuli A, Putz P, Le Gleut R, Guggenbuehl Noller JM, Olbrich L, et al. From first to second wave: follow-up of the prospective COVID-19 cohort (KoCo19) in Munich (Germany). BMC Infect Dis. 2021;21(1):925.
- Sampson AT, Heeney J, Cantoni D, Ferrari M, Sans MS, George C, et al. Coronavirus Pseudotypes for All Circulating Human Coronaviruses for Quantification of Cross-Neutralizing Antibody Responses. Viruses. 2021;13(8).
- Huang Y, Gilbert PB, Montefiori DC, Self SG. Simultaneous Evaluation of the Magnitude and Breadth of a Left and Right Censored Multivariate Response, with Application to HIV Vaccine Development. Stat Biopharm Res. 2009;1(1):81–91.
- Pritsch M, Radon K, Bakuli A, Le Gleut R, Olbrich L, Guggenbuehl Noller JM, et al. Prevalence and Risk Factors of Infection in the Representative COVID-19 Cohort Munich. Int J Environ Res Public Health. 2021;18(7).
- Therneau T, Lumley T. Package Survival: A Package for Survival Analysis in R. R Package. version 3.5-3. 2015.
- Daniel D. Sjoberg KW, Michael Curry, Jessica A. Lavery and Joseph Larmarange. Reproducible Summary Tables with the gtsummary Package. The R Journal. 2021;13:1:570–80.
- Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software. 2015;67(1):1–48.
- Bezanson J, Edelman A, Karpinski S, Shah VB. Julia: A Fresh Approach to Numerical Computing. SIAM Review. 2017;59(1):65–98.
- Pinheiro J, Bates D, DebRoy S, Sarkar D, Team RC. nlme: Linear and Nonlinear Mixed Effects Models. R Package Version 3.1–137. 2018.
- Supplementary.pdf
Supplementary Tables and Figures