Several vaccines have been developed since the beginning of the coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [1]. However, studies evaluating the humoral responses against SARS-CoV-2 have concluded that vaccine efficacy substantially declines over months [2^-4]. The emergence of SARS-CoV-2 variants, such as the omicron variant, further adds to the concern about waning antibody levels despite the administration of multiple vaccine doses [5]. Recent animal studies have revealed a host-protective role of T cells against SARS-CoV-2, particularly when the humoral immune response is insufficient [6, 7]. SARS-CoV-2-specific T cell responses in vaccinated and convalescent humans have been shown to be conserved across variants of concern, even when the humoral responses were insufficient [8, 9]. Jung, et al. [10] found that interferon gamma (IFN-γ), tumor necrosis factor, and interleukin 2 production by T cells is important for protection against SARS-CoV-2. IFN-γ is secreted by CD4⁺ and CD8⁺ T cells following activation. The IFN-γ release assay (IGRA) is well established for measuring T cell-dependent IFN-γ production during tuberculosis. Given that monitoring cellular responses may be as important as assessing antibody titers in vaccinated populations, we evaluated T cell-dependent IFN-γ production in response to the ChAdOx1-nCoV-19 and BNT162b2 vaccines—two major vaccines administered to the Korean population.

The present study was performed at Konkuk University Medical Center (KUMC), Seoul, Korea, from March to December 2021. We used data from 91 healthy study participants who provided written informed consent. Among them, 36 participants received two doses of the ChAdOx1-nCoV-19 vaccine and 30 received two doses of the BNT162b2 vaccine according to their respective vaccination schedules. For the ChAdOx1-nCoV-19 group, T cell responses were measured three weeks and three and five months after the administration of the second dose. In the BNT162b2 group, T cell responses were measured three and six months after the second dose. T cell responses were also measured two weeks after a third dose of BNT162b2 in the ChAdOx1-nCoV-19 group (heterologous booster) and three weeks after a third dose of BNT162b2 in the BNT162b2 group (homologous booster). Twenty-five unvaccinated participants with SARS-CoV-2 anti-nucleocapsid IgG-negative results (SARS-CoV-2 IgG, Abbott Laboratories, Sligo, Ireland), indicating no past infections, were assigned as controls.

For each group, T cell responses against SARS-CoV-2 were measured using Covi-FERON ELISA (SD Biosensor, Suwon, Korea). Briefly, 1 mL of whole blood was collected into each of the three assay tubes: Nil, original spike protein (SP) antigen, and mitogen. The Nil tube was used as a negative control to adjust for the background noise, with an upper limit of 0.80 international units (IU)/mL. The original SP antigen tube was coated with a specific SP antigen derived from SARS-CoV-2 20I/501Y.V1 variant (lineage B.1.1.7), with an upper limit of 10.00 IU/mL. The mitogen tube was used as a positive control. All tubes were gently mixed, incubated at 37°C for 16 hours, and centrifuged at 2,300×g for 15 minutes to extract plasma. A human IFN-γ ELISA was used to determine the IFN-γ levels in the plasma samples. According to the manufacturer’s instructions, the final IFN-γ response was defined as positive when the value of the IFN-γ level in the original SP tube minus that in the Nil tube was ≥0.25 IU/mL. A result with a Nil tube value higher than the upper limit was interpreted as indeterminate. In addition to the IFN-γ response, the anti-receptor-binding domain (RBD) IgG titer (SARS-CoV-2 IgG II Quant, Abbott Laboratories; manufacturer’s cut-off, 50 arbitrary units [AU]/mL) was measured in each group at the aforementioned time points. The upper limit for SARS-CoV-2 IgG II Quant was 40,000 AU/mL, and results above this limit were retested according to the dilution protocol. The Mann–Whitney U test was used to compare the median age among the three groups. The Wilcoxon test was used to compare the median IFN-γ response and anti-RBD IgG titer between the two vaccinated groups. The independent samples t-test was used to compare the IFN-γ response between the vaccinated groups. The Pearson correlation coefficient (r) between the IFN-γ level and anti-RBD IgG titer for each group was interpreted according to the standards established by Mukaka [11]: 0.90–1.00 as very high, 0.70–0.90 as high, 0.50–0.70 as moderate, 0.30–0.50 as low, and 0.00–0.30 as negligible correlation. MedCalc Statistical Software (version 20.112; MedCalc Software, Ostend, Belgium) was used for all statistical analyses. This study was approved by the Institutional Review Board of KUMC (file No. 2021-06-022-006).

The median age of all participants was 37 years (range, 19–58 years), and 73% of the participants were female. The median age of the participants in the control group (33 years, 19–45 years) was lower than that of the participants in the ChAdOx1-nCoV-19 (40 years, 22–58 years) (P=0.06) and BNT162b2 groups (40.5 years, 24–58 years) (P=0.06). There was no significant difference in median age between the vaccinated groups (P=0.65). All individuals in the control group showed a negative IFN-γ response and anti-RBD IgG titer according to the manufacturer’s cut-off. The median IFN-γ response and anti-RBD IgG titer in the control group were significantly lower than those in the vaccinated groups at any time point (P<0.001). Initially, the IFN-γ response was significantly higher in the vaccinated groups than in the control group. The IFN-γ response gradually decreased over time and increased steeply after the administration of a booster dose (Table 1, Fig. 1A, B). The BNT162b2 vaccine induced a higher IFN-γ response than the ChAdOx1-nCoV-19 vaccine at three months after administration (P=0.007). The post-booster results did not differ significantly between the vaccinated groups (P=0.967). The anti-RBD IgG titer showed a similar trend (Table 1, Fig. 1C, D), except that it decreased significantly between three and five months after the second dose in the ChAdOx1-nCoV-19 group (P<0.001) unlike the IFN-γ response, which did not change significantly over this period (P=0.699). The correlation between the IFN-γ response and anti-RBD IgG titer was low overall (r=0.46, P<0.001) and was negligible in each group at each time point (r range, –0.02–0.19).

Table 1 . IFN-γ responses and anti-RBD IgG titers after BNT162b2 and ChAdOx1-nCoV-19 vaccinations

Variable	Control	ChAdOx1-nCoV-19-vaccinated				BNT162b2-vaccinated
		3 weeks after 2nd dose	3 months after 2nd dose	5 months after 2nd dose	2 weeks after booster	3 months after 2nd dose	6 months after 2nd dose	3 weeks after booster
N	25	36	32	31	25	30	29	26
IFN-γ response, IU/mL	0 (0–0)	0.31 (0.12–0.69)	0.14 (0.04–0.24)	0.14 (0.06–0.41)	1.92 (0.73–5.01)	0.54 (0.36–1.01)	0.31 (0.16–0.86)	1.93 (1.13–4.00)
Anti-RBD IgG, AU/mL	1 (0–4)	1,097 (592–1,581)	500 (295–760)	278 (160–454)	17,824 (10,763–22,525)	3,236 (2,382–4,578)	1,080 (758–1,515)	20,277 (14,790–31,932)

Data are shown as median (interquartile range).

Abbreviations: AU, arbitrary units; IFN-γ, interferon gamma; IU, international units; N, number; RBD, receptor-binding domain.

Figure 1. IGRA and antibody assay results at the indicated time points after vaccination, with responder portions and percentages defined according to the manufacturers’ cut-offs. (A) IFN-γ response in the ChAdOx1-nCoV-19-vaccinated (yellow dots) and control groups (blue dots). (B) IFN-γ response in the BNT162b2-vaccinated (red dots) and control groups. (C) Anti-RBD IgG titer in the ChAdOx1-nCoV-19-vaccinated and control groups. (D) Anti-RBD IgG titer in the BNT162b2-vaccinated and control groups. The horizontal dotted lines indicate the cut-off for each assay.
Abbreviations: AU, arbitrary units; IFN-γ, interferon gamma; IGRA, IFN-γ release assay; IU, international units; RBD, receptor-binding domain.

This study was the first to concurrently assess cellular and humoral responses against SARS-CoV-2 in a Korean population following homologous and heterologous vaccination regimens. Our study results were consistent with those of previous similar studies [12, 13] in that the cellular and humoral responses increased after vaccination but then gradually decreased over time and increased again after the administration of a booster dose. In the context of the high vaccination rate in Korea, the IFN-γ values obtained from 25 pre-vaccinated subjects may provide valuable data for future studies. Interestingly, in the ChAdOx1-nCoV-19 group, the median IFN-γ response did not change significantly (P=0.699) (Fig. 1A), although the median anti-RBD IgG titer decreased significantly between three and five months after administration (P<0.001) (Fig. 1C). This is partially in line with the data of Shaw, et al. [14], who found that the decrease in the humoral response was greater than that in the cellular response from one to six months after homologous ChAdOx1-nCoV-19 vaccination (mean [95% confidence interval] fold change; 0.23 [0.21–0.26] vs. 0.62 [0.49–0.79]). Furthermore, three weeks after booster administration, we identified a participant in the BNT162b2 group who had an anti-RBD IgG titer (267.7 AU/mL) below the median value (20,277 AU/mL), but an IFN-γ response (1.16 IU/mL) within the interquartile range (1.13–4.00 IU/mL). Importantly, the participant was on treatment with methotrexate, an immune-modifying drug. Therefore, we speculate that humoral and cellular immune responses play complementary roles. Qui, et al. [15] also demonstrated that patients undergoing immune-modifying therapy show a reduced humoral response but a robust T cell response to vaccination. Our study indicates that the two immune responses have a low correlation, at least numerically. Yao, et al. [16] also found a lack of significant correlation between IFN-γ levels and antibody responses in convalescent individuals (r=0.70, P=0.593).

Interestingly, Le Bert, et al. [17] reported on the presence of long-lasting SARS-CoV-2-specific cross-reactive memory T cells in patients recovered from SARS 17 years after its outbreak in 2003. Given the high mutation rate of SARS-CoV-2, it is plausible that the effectiveness of the antibodies induced by the current vaccines diminishes over time [18]. However, reluctance to vaccination is undesirable because cellular immunity still plays an important role in preventing severe infection [19]. It would be interesting to measure IFN-γ responses after booster administration, when antibody titers are expected to be low. Notably, the positive rates of the IGRA were low in the vaccinated groups; however, these were determined according to the manufacturer’s cut-off (0.25 IU/mL). Given the low IFN-γ levels in the control group, further verification of a qualitative cut-off is necessary.

In summary, ChAdOx1-nCoV-19 and BNT162b2 vaccines elicited a substantial cellular response after the second dose, which increased after the administration of a BNT162b2 booster dose. T cell responses must be monitored when assessing the immunogenicity of COVID-19 vaccines. Our results serve as baseline data for future research and development of COVID-19 vaccines for the Korean population.

The SARS-CoV-2 IgG kits were provided by Abbott. Abbott had no role in the study design or data analysis.

All authors accept responsibility for the entire contents of this manuscript. Lee TH analyzed the data and wrote the draft; Moon H-W designed the study and finalized the draft; Nam M and Seo JD participated in data collection; and Kim H, Kim H-R, Hur M, and Yun Y-M participated in data analysis and reviewed the manuscript.

The authors declare no conflicts of interest.

This research was supported by a MD-PhD/Medical Scientist Training Program grant through Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Korea.