Main

Vaccination protects against fatal courses of SARS-CoV-2 infection, also in older adults1,2. Induction of neutralizing serum antibodies was observed after two intramuscular applications of the BNT162b2 mRNA COVID-19 vaccine in people >80 years of age3. However, recent outbreaks among older vaccinees4 and antibody responses inferior to those observed in younger vaccinees3 have prompted discussion on the necessity of a third vaccination.

Here we compared vaccine-induced humoral and cellular immune responses to SARS-CoV-2 in 51 individuals aged >80 years (older adults) and in 46 control individuals (young) (45 individuals aged 20–44 years (Table 1 and Supplementary Table 1) plus one 53-year-old woman, not included in the mean age calculation in Table 1, but included in all figures, accordingly. All participants were randomly recruited, COVID-19-naïve and not acutely ill; they were vaccinated twice, at day 0 and day 21, with BNT162b2 in a vaccination centre (older adults) or a doctor’s practice (young) in Marburg, Germany, March–May 2021. Analysis of spike-specific IgG, neutralization capacity against SARS-CoV-2 and SARS-CoV-2-reactive CD4 T cells (positive for the markers CD40L and IFNγ) in peripheral blood revealed strong induction of humoral and cellular immunity in response to vaccination (Fig. 1a–c). Notably, the neutralization capacity after two BNT162b2 doses increased even against the delta variant (B.1.617.2), although reactivity towards the wild type (B.1) was higher (Fig. 1b). This finding confirms previous data on antibodies3. As for spike-specific CD4 T cells, our data vary from this report3, as young and older groups demonstrated a further increase (10-fold average) between first and second dose (Fig. 1c, day 21 versus day 35). This was previously not noted3, potentially due to the different method used for their quantification (FluoroSpot), as compared to the herein applied multi-parameter flow cytometry (Extended Data Fig. 1).

Table 1 Donor characteristics
Fig. 1: Humoral and cellular SARS-CoV-2 immunity in >80- and 20–53-year-old study participants vaccinated with the BNT162b2 vaccine.
figure 1

ac, SARS-CoV-2-spike-specific serum IgG antibody titres (a), serum titres of 100% virus neutralization (VNT 100) for SARS-CoV-2 wild type (B.1) or its delta variant (B.617.2) (b) and percentages of SARS-CoV-2-spike-specific CD4 T cells (c) were analyzed in 51 donors aged >80 years (blue symbols) and 46 donors aged 20–53 years (yellow symbols) before (day 0; a,c), 21 d after the first (a,c) and 14 d after the second BNT162b2 (Pfizer-Biontech) vaccination (day 35; ac). Each symbol represents one donor. Horizontal lines indicate medians, dotted lines indicate the cut-off for antibody positivity at 35.2 BAU ml−1 (a) and 8 (reciprocal titre) for VNT (b). A cut-off value for determining reactive T cells could be considered at 0.01% as shown previously6 (c). P values determined by two-tailed Mann–Whitney test.

Source data

Fig. 2: Humoral and cellular SARS-CoV-2 immunity in >80-year-old initial low-/non-responders is rescued after a third dose of the BNT162b2 mRNA vaccine.
figure 2

ae, Combined presentation per person of the SARS-CoV-2-spike-specific serum IgG antibody and percentages of SARS-CoV-2-spike-specific CD4 T cells at day 35 in young (a) and aged donors (b). Five individuals (red triangles) mounted only low or no specific antibody and T-cell responses (b), and were vaccinated a third time in week 16 (W16) after the first dose (ce). SARS-CoV-2-spike-specific serum IgG antibody titre (c), percentages of SARS-CoV-2-spike-specific CD4 T cells (d) and serum titre of 100% virus neutralization (VNT 100) for SARS-CoV-2 wild type (B.1) or its delta variant (B.1.617.2) (e) were measured in week 18 (W18). (f) T-cell response to SEB of the young and older participants measured at the indicated days or weeks after the first dose. Each dot, square and triangle represents one donor. Blue dots: aged responders; red triangles: initial low-/non-responders to the SARS-CoV-2 spike glycoprotein; yellow squares: young donors. Horizontal lines indicate medians, dotted lines indicate the cut-off for antibody positivity at 35.2 BAU ml−1 for SARS-CoV-2-spike-specific IgG (a,c) and 8 (reciprocal titre) for VNT (e). P values determined by one-tailed Wilcoxon matched-pairs signed-rank test (ce).

Source data

Several important differences were noted between young and old vaccinees’ immune responses. First, the overall antibody and CD4 T-cell response was lower in older vaccinees at a high level of significance (Fig. 1a–c, day 35), as also shown by the stimulation index (Extended Data Fig. 2a). Second, while responses were comparable across young donors, substantial heterogeneity was observed in older donors, both in antibody and CD4 T-cell responses, whereby several older adults showed scarce or even no reaction. Third, some older adults had high frequencies of responding CD4 T cells before vaccination (Fig. 1c, day 0), probably reflecting cross-reactive activities gained during previous encounters with other coronaviruses, as demonstrated before5,6.

By combining results for antibodies and CD4 T cells for each vaccinee (Fig. 2a,b), we identified five older adults retaining very low levels of specific serum IgG together with almost an absence of spike-reactive CD4 T cells (Fig. 2b, red triangles). Remarkably, these older adults were potentially not protected by the previous two doses of the vaccine. Among them, donor #31 received methotrexate for rheumatoid arthritis, while no history of immune modulating medication or disease was evident in the remaining four (Supplementary Table 1). Additionally, no evident difference between responders and low-/non-responders was identified by age and the Charlson comorbidity index (CCI; Extended Data Fig. 3a,b and Supplementary Table 1). Furthermore, there was no significant difference (P = 0.15 by Fisher’s exact test) in the distribution of male and female participants between groups (Table 1).

No similar non-responder was found in the young cohort (Fig. 2a). Notably, our local authorities (Regional Council of Giessen, Hesse, Germany) informed us about breakthrough infections in several retirement homes. These infections occurred between 1 and 3 months after the second vaccination with BNT162b2 and 5 out of 45 infected residents >80 years succumbed to infection. Thus, the lethality of these breakthrough infections is remarkably similar to the frequency of low-/non-responders in our older-adults study cohort. No COVID-19 infections were recorded in our cohorts until August 2021.

Aiming to enhance SARS-CoV-2 immunity, all 5 low-/non-responders received a third dose of the BNT162b2 mRNA vaccine during week 16 after the first dose. At that day, blood analyses demonstrated absence of specific immunity to SARS-CoV-2 (Fig. 2c–e). The third vaccination was well tolerated. Most importantly, 2 weeks later, 4 out of 5 vaccinees, including donor #31, demonstrated robust spike-specific T-cell and antibody responses comparable with those detectable in responders after two-dose vaccination (Fig. 2c–e and Extended Data Figs. 1 and 2b). In donor #54, a healthy man without obvious morbidities, SARS-CoV-2-specific immunity also increased, although only to low levels. He was meanwhile vaccinated a fourth time, again with BNT162b2, which unfortunately still yielded an insufficient response (46.97 BAU ml−1).

Our data show that older adults initially hardly responding to two-dose vaccination can mount a virus-specific adaptive immune response after a third BNT162b2 dose. While the reason for primary unresponsiveness in our older-adults cohort remains unclear, BNT162b2 unresponsiveness in older adults is not fixed, and can be overcome by repeated vaccination. To confirm overall intact adaptive immune competence in low-/non-responders, we tested for antibody and CD4 T-cell reactivity towards control pathogens unrelated to SARS-CoV-2. Measles virus (MV)- or varicella-zoster virus (VZV)-specific IgG did not differ between low-/non-responsive donors and all other aged donors at baseline (day 0, Extended Data Fig. 4). Additionally, the T-cell responsiveness towards staphylococcal enterotoxin B (SEB) was comparable to that seen in the other older vaccinees throughout the observation period (Fig. 2f and Extended Data Fig. 1). A similar response was observed in the young cohort (exemplified by their data on day 0), or in older adults (>80 years) who have recovered from COVID-19 infection several months before (Extended Data Fig. 1). These results demonstrate that the kinetics of T-cell activation in our assay conditions are similar throughout participants and evaluated timepoints. Furthermore, even if our cohort is small, we demonstrate that initial unresponsiveness to vaccination is not indicative of an overall lack of immune competence; consequently, older adults who previously did not respond or responded poorly to vaccination, will probably benefit from repeated vaccination with BNT162b2. Accordingly, in very recent studies of patients after allogeneic hematopoietic stem-cell transplantation or on haemodialysis, only a subgroup of vaccinees reacted by an increase in antibody levels after a third vaccination, while T-cell reactivity was not analysed7,8. A third vaccination was recently shown to increase protection against COVID-19 in people >60 years9. This effect probably combines overcoming of the herein described primary non-responsiveness, and a booster effect in primary responders whose antibody titre may have gradually declined.

Overall, we show lower immune responses against SARS-CoV-2 in aged versus young vaccinees, a finding which is also reflected in the antibody neutralization capacity against the SARS-CoV-2 delta variant. Nevertheless, 90% of individuals aged >80 years established adaptive SARS-CoV-2-specific immunity after receiving two doses of the BNT162b2 mRNA vaccine. However, low-/non-responders can be identified. Therefore, our data are suggestive of the importance of routine screening for spike-specific immunity in this population at risk, to assess the extent of immunity after two doses of BNT162b2. Screening should be unbiased and not limited to conditions of immunodeficiency or targeted immunosuppression. Should such tests reveal lack of specific immunity, re-vaccination should be considered.

Methods

Our research complies with all relevant ethical regulations. The study of patients with COVID-19 and vaccinations against COVID-19 was approved by the ethics committee of the medical faculty of the Philipps University Marburg (study number 40/21-12032021) and participants gave written informed consent according to the Consensus-based Clinical Case Reporting Guideline (CARE) guidelines and in compliance with the Declaration of Helsinki principles.

Study participants

Blood samples were obtained from older adults aged >80 years by venipuncture before and at additional timepoints indicated in Fig. 1 after primary and secondary vaccination by injection of Tozinameran (BNT162b2 vaccine, Comirnaty) in the deltoid muscle, at a vaccination centre in Marburg, Germany (Extended Data Table 1). On the basis of the lack of appropriate response to routine vaccination, the vaccination centre decided to vaccinate 5 individuals a third time with BNT162b2 vaccine 16 weeks after day 0, and blood samples were again obtained immediately before and 2 weeks after the third vaccination.

Analyses were performed between March and May 2021 and in August 2021 for third vaccination candidates. In May 2021 the study also obtained samples from unvaccinated elderlies >80 years of age living in a retirement home who have recovered from COVID-19 after an outbreak with SARS-CoV-2 variant B.1.221 in January 2021. All donors provided informed consent to participate in the study. Charlson comorbidity index10 was calculated according to https://www.mdcalc.com/charlson-comorbidity-index-cci#evidence.

Sample processing and clinical lab

Blood serum was isolated from Serum Separator Clot Activator tubes (Greiner Bio-One, Germany) according to the manufacturer’s instructions and stored at −80 °C until analysis.

Peripheral blood mononuclear cells (PBMCs) were isolated from fresh heparinized whole blood by density gradient centrifugation over Pancoll human (Pan Biotech) after dilution with an equal volume of PBS at room temperature. PBMCs were washed twice (500 × g, 10 min, 4 °C) in cold PBS supplemented with 0.2% BSA, counted manually, and resuspended in RPMI 1640 media (Gibco, Life Technologies) supplemented with penicillin, streptomycin and 10% human AB serum (all from Sigma) at 5 × 106 cells per ml.

Assessment of antigen-specific T cells

Antigen-reactive T-cell responses were analysed using a protocol based on previous work11. A detailed protocol is given in the Supplementary Information. Briefly, 5 × 106 PBMCs were stimulated with either SARS-CoV-2 spike protein peptide mix (wild type, Miltenyi Biotec), SEB (0.7 µg ml−1, kindly provided by Prof. Bernhard Fleischer, Bernhard Nocht Institute of Tropical Medicine, Hamburg, Germany), or with an equal volume of water as a control, in the presence of anti-CD28 (5 µg ml−1) and monensin (1 µg ml−1) for 12 h. Brefeldin A (1 µg ml−1) was added 2 h after the start of the stimulation. The stimulation was stopped by adding 2 nM EDTA. Dead-cell labelling was performed by resuspending the cell pellet in 500 µl PBS supplemented with 1:1,000 amine reactive Zombie Aqua Fixable Viability dye (Biolegend), and PBMCs were fixed for 20 min using 2% formaldehyde solution (Thermo Scientific). Thereafter, cells were stained with a cocktail of antibodies as detailed in the supplementary section. Acquisition was performed on a MACSQuant 16 flow cytometer (Miltenyi Biotec).

FlowJo version 10 (BD) and OMIQ.ai were used for analysing flow cytometry data. Flow cytometry standard files underwent quality control and, where applicable, anomaly removal by FlowAI12.

Quantification of SARS-CoV-2-specific antibodies

Serum antibodies against the recombinantly expressed S1 subunit of the SARS-CoV-2 spike protein were quantified using the Anti-SARS-CoV-2-QuantiVac-ELISA run on the automated EuroLab Workstation (Euroimmun, Germany), following the manufacturer’s protocol. Sera exceeding the detection range of the assay were pre-diluted 1:10 or 1:100 and measured again. Results obtained in relative units per ml were converted into binding antibody units (BAU) per ml by multiplication with the factor 3.2, according to the manufacturer’s instructions. Results in BAU ml−1 were calibrated against the ‘First WHO International Standard for anti-SARS-CoV-2 immunoglobulin (NIBSC code: 20/136)’. The lower cut-off for this assay is at 35.2 BAU ml−1.

SARS-CoV-2 neutralization tests (VNT 100)

Human sera were heat-inactivated for 30 min at 56 °C and diluted in a two-fold dilution series in 96-well cell culture plates (1:4 to 1:512). One hundred plaque-forming units (PFU) of SARS-CoV-2 were added in the same volume to the serum dilutions. The following SARS-CoV-2 virus isolates were used: German isolate BavPat1/2020; European Virus Archive Global #026 V-03883 (Genbank: MZ558051.1) and the Delta variant, B.1.617.2. The sequence of the viruses was confirmed. Following incubation at 37 °C for 1 h, approximately 20,000 Vero C1008 cells (ATCC, Cat. no. CRL-1586, RRID: CVCL_0574) were added. Plates were then incubated at 37 °C with 5% CO2, and cytopathic effects were evaluated at day 4 or day 6 (Delta variant) post infection. Neutralization was defined as the absence of cytopathic effects in the serum dilutions. The reciprocal neutralization titre was calculated from the highest serum dilution without cytopathic effects as a geometric mean based on three replicates. The lower detection limit of the assay is 8 (reciprocal titre), corresponding to the first dilution of the respective serum. Two positive controls were used as inter-assay neutralization standards and quality control for each test. Neutralization assays were performed in the BSL-4 laboratory of the Institute of Virology at Philipps University Marburg, Germany.

Virus-specific antibodies

IgG antibodies against measles and VZV were quantified using the commercial Siemens Enzygnost IgG ELISA kit that was run on an automated Siemens BEPIII system. Values were quantified using the alpha method, according to the manufacturer’s instructions. The cut-off for VZV-specific IgG was 50 mIU ml−1, and 150 mIU ml−1 for measles-specific IgG.

Statistical analysis and reproducibility

Prism version 9 (GraphPad software) was used to display data, and perform descriptive statistics and significance testing.

To determine differences between the two cohorts (older adults vs young), a two-tailed Mann–Whitney test was applied to evaluate the level of significance. For assessing donor-specific responses over time (5 low-/non-responders), the one-tailed Wilcoxon matched-pairs signed-rank test was used. To test for differences in age, CCI and sex between responders and low-/non-responders, we applied the unpaired two-tailed t-test, the Wilcoxon rank-sum test with continuity correction and the Fisher’s exact test, respectively. No statistical method was used to predetermine sample size.

Assessment of T cells after antigen-specific stimulation was performed in a total of 29 independent batches, with up to 30 donor samples per batch. Data of a SEB-stimulated control sample included in each batch showed consistent results across all batches. Quantification of SARS-CoV-2-specific antibodies was run according to clinical routine standards, including all required calibrators and controls. Data from 5 vaccinees were excluded, on the basis of their decision to decline study participation, death (not related to the study) or previous COVID-19 infection. Only age-group assignment was available to the investigators during experiments and outcome assessment. Moreover, the experiments and analyses were performed in three independent laboratories. The experiments were not randomized regarding age of the vaccinees.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.