Several studies on the durability of humoral response in subjects recovered from SARS-CoV-2 infection showed that both binding and neutralizing antibody levels decrease only modestly at month 8 after the infection (1, 2). This evidence initially suggested that vaccinated persons and previously infected would experience a low number of breakthrough infections. However, the durability of immunity has been called into question by the mounting evidence of reinfections after natural recovery (3). Moreover, a progressive decline in humoral immune response has been shown after vaccination (4). In our experience, in a cohort of healthcare workers, this decline was shown to start from d90 after the first shot (5). These results were in agreement with larger cohort studies (4) and suggest that after vaccination or infection, several mechanisms of immunity exist both at the antibody level and at the level of cellular immunity.

Moderna and Pfizer vaccines using a mutated sequence of the receptor-binding domain (RDB) that contains two consecutive prolines, lysine 986, and valine 987 (6) have been associated with high protection rates (7). Accumulating evidence demonstrates that the two doses of the BNT162b vaccine elicit either high IgG or neutralizing antibody responses (8, 9). Neutralizing antibodies were shown to correlate with protection and may be used to assess effective vaccine-induced humoral response (10) However, there is scarce applicability of neutralizing assays in the routine practice as neutralizing tests are complex, time-consuming, and not always comparable across assays (11). In addition, a time-dependent neutralizing activity regression relationship with IgG levels has been demonstrated (4).

It has recently been shown that fully vaccinated people remain at the risk for SARS-CoV-2 infections and Pfizer's CEO announced in October 2021, the need for a booster within 12 months of the first dose (12–14). In a recent study from Israel, involving participants 60 years old, 5 months after two doses of BioNTech/Pfizer vaccine, rates of infection and severe illness were lower among those who received a booster injection as compared to participants who did not (15).

Evidence suggests that humoral response alone may not offer sufficient protection against either infection or disease, and SARS-CoV-2-specific cellular immunity may be more stable and longer-lasting than humoral immunity (1). It has been, therefore, hypothesized, based on experimental models, that CD4+ and CD8+ T-cells and production of IFN-γ play an important role in vaccination immune response (16).

We analyzed – by age, gender, and previous SARS-CoV-2 infection history –the binding and neutralizing antibody response induced by the BioNTech/Pfizer vaccine 180 days after the second vaccine shot in our cohort of almost 400 healthcare workers longitudinally followed up to 180 days after the second dose of BioNTech/Pfizer. The subjects' early humoral response had been previously reported to decline 90 days after the first vaccine dose (5). Spike-specific T-cell-mediated reactivity using an IFN-γ release assay, with the aim to gather information about cellular immune response, was also evaluated.

Our analysis was based on the medical data from the multicenter longitudinal study (Covidiagnostix, funded by the Italian Ministry of Health) to investigate the antibody response in Healthcare workers vaccinated with BioNTech/Pfizer starting from February 11, 2020, and ending on April 11, 2021. All the subjects received two vaccine injections 21 days apart. The planned testing time for binding antibodies was day 0 (d0) (before the first dose), day 7 (d7), day 21 (d21), day 31 (d31) after the first shot, and day 90 (d90) 60 days after the second shot, day 180 (d180) days after the second shot corresponding to 210 days after the first shot, respectively.

We excluded the participants who do not have the complete set of blood sample collection. Blood samples were collected into clot activator BD vacutainer tubes (Becton Dickinson, Franklin Lakes, NJ, USA). The margin of sampling window for each time-point was of 2 days.

Our study investigated the IgG and neutralizing response in naive and experienced HW previously shown to be able to mount a strong IgG response at d31 (5). At d180 after the second BioNTech/Pfizer vaccine shot, among naive, all HW had binding antibody levels higher than the assay threshold, although only 8.1% had results higher than the highest assay threshold. At variance, 1/3 of subjects had neutralizing antibodies titers below LLoQ, while titers ≥ 320 generally associated with protection, were observed in very few cases (1.2%). Converting neutralizing antibody titers in International Unit (IU/ml) by running in the same neutralization assay, the first SARS-CoV-2 WHO International Standard (NIBSC 20/136)¹, we observed that only 32.78% of our patients had 50% protective neutralizing antibody. Our results appear in keeping with those reported in two studies from Israel, where the majority of the population was vaccinated using the BioNTech/Pfizer or Moderna vaccine. The first study on over 1,000,000 persons (596,618 vaccinated and 596,618 non-vaccinated) demonstrated high efficacy of vaccines, not only in disease prevention but also in infection transmission up to 42 days after the first vaccination (7). A second more recent study with longer follow up from the same Country, showed that 39 (2.6%) out of 1,497 fully vaccinated HW became infected during 14 weeks after their second dose of the BNT162b2 (BioNTech/Pfizer) vaccine; all the infected had lower neutralizing antibody levels than their uninfected colleagues during the peri-infection period (23). In our study, only 6 subjects (2.2%) experienced a breakthrough infection. All of them were older and had median neutralizing antibody levels lower than the median of the uninfected population. Although we are aware that our sample size is limited, our results appear in line with those reported in Israel.

The already known significant decline in BNT162b2 vaccine protection more than 120 days after the second dose, in our study, conducted in the region of Puglia with a low community incidence rate (positivity index on December 16, 2021, was 2.4%)², was associated with the rate of breakthrough infections comparable to those reported by Bergwek (23) and were significantly lower than the rates reported among unvaccinated subjects³.

In keeping with the decreased severity of the disease in vaccinated individuals who acquire SARS-CoV-2 infection, all our patients with breakthrough infections were mild. A persistently positive swab result was observed in almost all (mean positivity duration 4.5 ± 2.3 weeks). Whether a possible further decrease in vaccine effectiveness against hospitalization after a longer interval from vaccination occurs was impossible to evaluate in our population given the mandatory administration of a third vaccine dose to the HW in Italy that started from November 22, 2021, based on the evidence that booster dose may mitigate the risk of transmission, disease, and deaths in all the age groups (24)⁴.

Reliable detection of the T-cell-mediated immune response was explored in our study by IFN-γ production. Most of the subjects showed robust IFN-γ production after S-protein stimulation of peripheral blood cells. Results below the threshold of the assay were observed in only 12 (4.6%) naive, suggesting that lack of T-cell reactivity is a rare event even after a long interval from the second vaccine shot. This evidence was also confirmed by the cytoflorimetric analysis (manuscript in preparation). Moreover, as shown by the linear regression model, higher T-cell reactivity was observed in patients with higher neutralizing antibody levels. These results are in agreement with those reported by Schiffner et al. (25, 26). Consequently, the combination of these two assays seems to provide predictive information on protective immune reactions. Nevertheless, we need to keep in mind that neutralizing titers may be impractical to assess routinely, whereas IFN-γ evaluation as an expression of lymphocyte activity may be easier to use than other more complex CD4+ and CD8+ cellular response assessment methods.

Whether the decay of serum antibody levels is a good indicator for the timing of booster administration remains to be determined. Identifying immune correlates of protection (or lack thereof) from SARS-CoV-2 is critical in predicting how the expected antibody decay will affect clinical outcomes, if and when a booster dose will be needed, and whether vaccinated persons are protected (23, 26). Surely antibody decay represents one of the initial predisposing factors to breakthrough infections. However, while cellular and humoral immunity to SARS-CoV-2 is critical to control primary infection and correlates with severity of disease, the degree of vaccine protection from breakthrough infections may be an expression of the initial immune response rather than of the decay of antibody levels, since memory cells are expected to respond to future exposures. Moreover, while correlates of protection have been developed for other infections such as influenza (27) by challenge experiments in humans (28), no study has defined correlate of protection until a recent one that focused on correlates of protection against symptomatic COVID-19 (29, 30). This study highlights that there is no single threshold value for different assays (31). In our small group of subjects who experienced a breakthrough infection, we had the opportunity to both identify a common source of infection in an unvaccinated index case and to show low median neutralizing antibody titers and higher median age.

The use of the same mRNA vaccine with a similar schedule and similar interval between vaccination and post-vaccination antibody assessment strengthen this study. Moreover, evaluating one of the longest delays between the second vaccine dose and both IgG and neutralizing antibody assessment has the advantage of using the IFN-γ spike-specific-induced T-cell immune response assay that allows simultaneous cellular responses evaluation. Finally, we had the opportunity to trace the incident breakthrough infection and to investigate its possible predictors. Limitations of our study are the relatively small sample size, the homogeneous demographic characteristics of our patients, young and healthy in the majority of cases. A further disadvantage is the relatively low prevalence of SARS-CoV-2 infection in our region as compared to others in Italy. This may prevent the exportability of our findings to the general population with different ages and co-morbidities.

In conclusion, our study shows that although the low humoral response is relatively long-lasting, high IgG levels are extremely rare in naive subjects. Only a third of subjects maintained neutralizing responses. In terms of T-cell, IFN-γ production after specific stimulation, a very limited number of subjects resulted unable to produce this cytokine over a period of 180 days after the second shot. IFN-γ testing could be used as surrogate testing for cellular immune responses. The results attained in our small group of subjects with breakthrough infection suggest that simultaneous neutralizing antibody titers below 20, binding antibody levels below 200 BAU/ml, and IFN-γ <1000 in subjects older than 58 years may be considered an alarming condition.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://zenodo.org/record/5728042#.Ycy6BWnSJPw.

The studies involving human participants were reviewed and approved by EC IST Giovanni Paolo II IRCCS, BARI at Fondazione Casa Sollievo della Sofferenza San Giovanni Rotondo. The patients/participants provided their written informed consent to participate in this study.

AMang: conceptualization, data curation, formal analysis, investigation, methodology, project administration, supervision, validation, visualization, writing – original draft, and writing – review & editing. VP: data collection and writing – original draft. GC: formal analysis and data collection. VG: investigation and writing – original draft. AMane: formal analysis, investigation, writing – original draft, and writing – review & editing. PC: formal analysis and investigation. EM: writing – review & editing. FG: visualization. AMai: validation. SA: formal analysis and investigation. All authors contributed to the article and approved the submitted version.

This study was partially funded by Ministry of Health of Italy, Bando Ricerca COVID-19; Project Number: COVID-2020-12371619; project title: COVIDIAGNOSTIX—Health Technology Assessment in COVID serological diagnostics.

EM, PC, and AMane were employed by VisMederi Srl. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.