In this dual-center observational cohort study, we screened 192 patients on dialysis between March 2020 and April 2021 at the Division of Nephrology of the University Hospital of Heidelberg and at the associated Kidney Center Heidelberg for inclusion (Figure 1). Eligible participants were either hemodialysis patients after PCR-confirmed COVID-19 (Group 1), uninfected hemodialysis patients after two-time mRNA BNT162b2 (BioNTech) vaccination (Group 2), or previously COVID-19 infected individuals who received a single-dose of BNT162b2 ≥ 6 months after infection (Group 3) (Figure 1). Group 3 consists of seven patients from Group 1 and 6 patients without available sera who are not part of Group 1. We exclusively included patients on hemodialysis.

The anti-S1 IgG index, SARS-CoV-2 specific neutralizing antibodies, and a bead-based differentiation of SARS-CoV-2 target antibodies were measured in 43 BNT162b2 vaccinated dialysis patients without prior SARS-CoV-2 infection 3 weeks after the first and 3 weeks after the second vaccination. In 25 unvaccinated dialysis patients after COVID-19, humoral responses were measured 6 weeks after the onset of infection. The onset of infection was defined as the day with the first symptoms and the diagnosis was made by positive PCR. Vaccination was performed twice in all 43 patients of Group 1 at week 0 and 3. To exclude patients with prior SARS-CoV-2 infection or infection during follow-up, we measured antibodies to the nucleocapsid protein before enrollment and after the first and the second vaccination. SARS-CoV-2 infected dialysis patients who died within the first 6 weeks after disease onset were excluded. In addition, 13 dialysis patients received a single-dose of BNT162b2 vaccine at least 6 months after onset of COVID-19 disease and humoral responses were subsequently measured 3 weeks after single-dose vaccination. To exclude a decisive effect of age and dialysis vintage on humoral responses, we performed a subgroup analysis with age-matched and dialysis vintage-matched COVID-19 infected (N = 20), fully vaccinated (N = 30), and previously infected single-dose vaccinated (N = 13) dialysis patients, respectively (Supplementary Tables 1–3). To detect breakthrough infections after vaccination or previous infection, we performed rapid SARS-CoV-2 antigen testing before each individual dialysis session (three times weekly) and PCR testing once a week.

The study (ethics number: S-061/2021) was approved by the ethics committee of the University of Heidelberg on 02/15/2021 and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all study participants.

The IgG response against the S1 protein was measured by the SARS-CoV-2 IgG Assay (Siemens, Eschborn, Germany) and the total antibody response against the nucleocapsid protein by the Elecsys Anti-SARS-CoV-2 assay (Roche, Mannheim, Germany) according to the manufacturer's instructions. A semi-quantitative index of <1 was classified as negative, and a value of ≥1 or higher as positive. This cutoff for detection gives a specificity of 99.9% with a sensitivity of 96.4% for the Siemens spike assay.

The binding-inhibition potency of serum samples was detected by a plate-based SARS-CoV-2 surrogate virus neutralizing assay (Medac, Wedel, Germany) (13, 14). This assay mimics the virus-host interaction by direct protein-protein interaction using purified receptor-binding domain protein from the viral spike protein and the host cell receptor angiotensin converting enzyme 2 (ACE2). The antibodies in serum samples were incubated with SARS-CoV-2 receptor-binding domain horse-radish peroxidase and added to ACE2 coated wells. The reactions were developed using 3,3′,5,5′-tetramethylbenzidine as substrate. Optical density at 450 nm was measured in each well and the percent (%) inhibition was calculated as presented in the Supplementary Material. A cut-off for viral neutralization of ≥30% inhibition of receptor-binding domain:ACE-2 binding was applied according to the manufacturer's instructions, resulting in a specificity of 100% with a sensitivity of 98%. The SARS-CoV-2 anti-S1 IgG index (Siemens, Eschborn, Germany) correlated with the neutralizing antibody activity measured by the surrogate virus neutralization assay in our cohort (Supplementary Figure 1).

We determined IgG antibodies against different SARS-CoV-2 target antigens by a multiplex bead-based assay (One Lambda Inc., West Hill, CA) for the Luminex platform (LabScreen COVID Plus), including besides the SARS-CoV-2 nucleocapsid protein, four distinct fragments of the SARS-CoV-2 spike protein, namely the full spike protein, the S1 protein, the receptor-binding domain of the spike protein, and the S2 protein (15). Additionally, we measured S1 fragments from six other coronaviruses, namely HCoV-229E, HCoV-HKU1, HCoV-NL63, HCoV-OC43, MERS-CoV, and SARS-CoV-1. Antibody detection was performed according to the manufacturer's instructions and the mean fluorescence intensity (MFI) was analyzed on a Luminex 200 device (Luminex Corporation, Noord-Brabant, The Netherlands). The MFI cutoff values for each of the 11 proteins are shown in Supplementary Table 4.

Adverse events assessment included medical monitoring of local and systemic adverse events for 10 days after the first and the second vaccine dose at each in-house dialysis treatment. Adverse events were assessed using a 12-item questionnaire inquiring previously mentioned side-effects and the use of pain medication after vaccine reception. The questionnaire is presented in the Supplementary Material. Patients were further screened for infection with COVID-19 three times a week by rapid antigen testing and once a week by PCR testing.

Data are expressed as median and interquartile range (IQR). Results of different groups were compared by applying the Kruskal-Wallis test with Dunn's post-test. Statistical analysis of categorical data was performed using the chi-square (χ²) test. Statistical significance was assumed at a P < 0.05. The statistical analysis was performed using GraphPad Prism version 9.0.0 (GraphPad Software, San Diego CA, USA).

Twenty-five patients had a PCR-confirmed COVID-19 disease, 43 patients received the standard two-dose regimen of BNT162b2 mRNA vaccine, and 13 patients received a single-dose BNT162b2 vaccination at least 6 months after the onset of COVID-19 disease (Figure 1). Baseline characteristics of the different groups are shown in Table 1. We included only patients on hemodialysis in our study. COVID-19 infected patients were significantly younger compared to fully vaccinated or previously infected dialysis patients vaccinated with a single-dose (median age 74 vs. 82 vs. 79 years; P = 0.002). An age- and dialysis vintage-adjusted subgroup analysis was performed and baseline characteristics as well as COVID-19 related clinical characteristics are shown in Supplementary Tables 1, 2, respectively. Antibody responses were determined 6 weeks after the onset of COVID-19 disease, 3 weeks after the first and 3 weeks after the second vaccine dose, or 3 weeks after single-dose vaccination in patients with onset of COVID-19 disease in median (IQR) 6.3 (6.1–6.7) months before.

The clinical courses of individuals infected with SARS-CoV-2 are shown in Table 2. According to previously used grading systems for COVID-19 disease severity (16, 17), three patients had critical (12%), three severe (12%), fourteen moderate (56%), and five mild (20%) COVID-19 disease courses (Table 2; Supplementary Table 4). No individual was asymptomatic. Dialysis patients with critical COVID-19 disease courses were more often male and suffered more frequently from diabetes (for both P < 0.05; Table 2). Four patients died within 6 weeks after onset of COVID-19 disease and were subsequently excluded (Figure 1).

We did not detect symptomatic or asymptomatic breakthrough infections in any of the three different groups during follow-up.

Dialysis patients after SARS-CoV-2 infection tended to have a higher anti-S1 IgG level as compared to dialysis patients after two times BNT162b2 vaccination (median index 15.8 vs. 6.3; IQR 5.6–37.1 vs. 0.8–14.2; P = 0.060; Figure 2A). Whereas, 25/25 (100%) of SARS-CoV-2 infected dialysis patients had antibody levels above the index cutoff for detection, only 31/43 (72%) of twice vaccinated patients exceeded this threshold.

We also determined the SARS-CoV-2 neutralizing capacity in both groups. Sera of infected patients showed a significantly higher inhibition than sera of twice vaccinated dialysis patients (median of percent inhibition 88.0 vs. 50.7; IQR 71.5–95.5 vs. 26.4–81.0; P = 0.018; Figure 2A). An age- and dialysis vintage-adjusted subgroup analysis revealed similar results for neutralizing antibody capacity (P = 0.047; Figure 2B; Supplementary Table 3).

Antibodies against the full spike protein, the S1 spike protein, and the receptor-binding domain protein were comparable between both groups with a median (IQR) MFI of 20,862 (16,750–23,315), 8,908 (5,124–13,010), and 12,301 (7,128–17,355) in infected dialysis patients and 21,797 (11,773–23,769), 7,808 (2,901–12,556), and 12,933 (3,929–18,058) in fully vaccinated dialysis patients, respectively (Figure 3A). Reactivity of S1 proteins from six other coronaviruses were also comparable between both groups (Supplementary Figure 2). However, infected dialysis patients had higher MFI-values for the spike S2 and the nucleocapsid protein with a median (IQR) of 8,553 (6,808–14,041) and 13,076 (10,452–16,591) as compared to fully vaccinated patients with 1,515 (381–4,346) and 0 (0–139) MFI-values (P = 0.007 and P < 0.001; Figure 3A). Age- and dialysis vintage-adjusted subgroup analysis confirmed these findings (Figure 3B; Supplementary Table 3). For each individual, the cumulative MFI-value for all SARS-CoV-2 proteins tested and the proportional MFI-value as a percentage of each component protein are shown in Figure 4.

The severity of COVID-19 disease courses did not correlate with results of any of the assays performed (Supplementary Figure 3).

With a median (IQR) of 274 (151–791), dialysis patients with prior infection developed 15- to 34-fold higher anti-S1 IgG levels than age- and dialysis vintage-adjusted patients after infection or after two times vaccination (for both P < 0.001; Figures 2A,B; Supplementary Table 3).

The inhibition capacity of SARS-CoV-2 antibodies was also significantly higher in previously infected single-dose vaccinated patients compared to patients after infection or patients after two times vaccination, respectively (P = 0.002 and P < 0.001). Median (IQR) percent inhibition was 97.6 (97.2–98.0) in single-dose vaccinated patients with previous infection as compared to 88.0 (71.5–95.5) and 50.7 (26.4–81.0) inhibition in age- and dialysis vintage-adjusted patient after infection or two-times vaccination (Figure 2B; Supplementary Table 3).

Moreover, single-dose vaccinated patients with prior infection showed a strong and broad reactivity against all four different fragments of the SARS-CoV-2 spike protein, whereas reactivity against the nucleocapsid protein was low (Figure 4). All 13 individuals showed MFI-values above the detection cutoff for the full spike, the spike S1, the spike receptor-binding domain, and the spike S2 protein with a median (IQR) MFI of 24,017 (23,699–24,107), 22,174 (19,849–22,800), 22,736 (22,406–23,337), and 22,254 (19,490–23,179), respectively (Figure 3A). Mean fluorescence intensity-values for all four fragments were significantly higher as compared to age- and dialysis vintage-adjusted patients after infection or two-times vaccination (Figure 3B; Supplementary Table 3). Reactivity of S1 proteins against the four community coronaviruses and MERS-CoV was comparable between groups, whereas reactivity against the SARS-CoV-1 S1 protein was significantly higher in previously infected dialysis patients vaccinated with a single-dose (for both P < 0.001; Supplementary Figure 2). Reactivity against the nucleocapsid protein declined in single-dose vaccinated dialysis patients 6 months after SARS-CoV-2 infection (Figure 3A). The median (IQR) MFI of 4,654 (3,251–6,344) was significantly lower as compared to age- and dialysis vintage-adjusted patients 6 weeks after SARS-CoV-2 infection with a median (IQR) MFI of 13,074 (10,452–16,591) (P < 0.05; Figure 3B).

Figure 5 shows all measurements of humoral responses for all subgroups. Individual antibody courses were determined from before second vaccination to after second vaccination. In addition, antibody courses from after COVID-19 in unvaccinated patients to after single-dose vaccination in patients with prior infection were determined in patients with available sera at both time points (7 out of 13 patients). SARS-CoV-2 anti-S1 IgG, neutralizing antibody activity, and reactivity against all four different fragments of the SARS-CoV-2 spike protein significantly increased from before second vaccination to after second vaccination (for all P < 0.001) and from after COVID-19 in unvaccinated patients to after single-dose vaccination in patients with prior infection, respectively (for all P < 0.05; Figures 5A–F). In contrast, antibody levels against the nucleocapsid protein significantly decreased over time from after COVID-19 in unvaccinated patients to after single-dose vaccination in patients with prior infection (P < 0.05; Figure 5G).

Local responses were low after both the first and second BNT162b2 vaccination in dialysis patients without prior SARS-CoV-2 infection (Supplementary Figure 4). Only 2/43 (4.6%) reported mild-to-moderate pain at the injection site within 2 days after the first injection and 4/43 (9.3%) after the second injection. Systemic responses were only reported after the second dose of BNT162b2. Systemic events were fatigue 2/43 (4.6%), headache 2/43 (4.6%) and muscle ache 1/43 (2.3%).

Single-dose vaccination in previously infected dialysis patients was well-tolerated in all patients. Only 3/13 (23.1%) had pain at the injection site after single-dose vaccination and systemic reactions were not reported (Supplementary Figure 4).