Introduction

Since January 2020, the pandemic of coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has been a global public health emergency. Several types of vaccine against SARS-CoV-2 have been developed, of which mRNA vaccines are the most widely used because they can efficiently prevent severe illness, hospitalization, and death [1, 2]. The mRNA vaccines are primarily administered in a two-dose schedule, but humoral immunity gradually declines, resulting in reduced protection against COVID-19 [3, 4]. Therefore, a third booster vaccination has been administered in many countries [5, 6].

Patients with chronic kidney disease (CKD), especially those with end-stage kidney disease (ESKD) undergoing maintenance hemodialysis (HD), are at high risk of COVID-19 severity and death [7, 8]. Although patients undergoing HD are often excluded from vaccine trials for safety reasons, the results of many cohort studies have shown that SARS-CoV-2 mRNA vaccination reduces the risk of COVID-19 infection, severe illness, and death in patients receiving dialysis [9, 10]. Therefore, SARS-CoV-2 mRNA vaccination is now recommended for patients undergoing HD. However, patients with ESKD generally have a weak immune response to vaccines. Patients with ESKD have a reduced ability to produce antibodies to vaccine antigen owing to combined abnormalities of B cells, T cells, and antigen-presenting cells [11, 12]. For example, post-vaccination antibody positivity has been reported to be 44% versus 96% for hepatitis B virus and 40% versus 65% for seasonal influenza in patients with ESKD versus healthy controls, respectively [13, 14].

Similarly, immune responses to COVID-19 mRNA vaccines are weak in patients with ESKD [15]. Neutralizing antibody titers and SARS-CoV-2 spike protein antibody titers are reported to be lower in HD patients compared with those in healthy controls 1 to 2 months after SARS-CoV-2 mRNA vaccination [16,17,18], and importantly, both antibody titers have strong correlation [19,20,21]. Although there are few reports of long-term observation of post-vaccination antibody titers in patients with ESKD, studies in Israel and Germany have reported that dialysis patients had lower antibody titer levels 6 months after vaccination compared with healthy individuals [22, 23]. However, to our knowledge, there are no reports of antibody titers being examined up to 6 months after SARS-CoV-2 mRNA vaccination in patients receiving HD in Asia, including in Japan.

This was a multi-institutional retrospective cohort study. We aimed to examine post-vaccination antibody titers over time in patients with ESKD undergoing maintenance HD and healthy controls in Japan. SARS-CoV-2 spike protein antibody titers were examined 1 and 6 months after the second dose of the BNT162b2 (Pfizer-BioNTech) mRNA vaccine in 412 patients undergoing HD and 156 health care workers. Additionally, we compared the attenuation speed of antibody titers after vaccination between HD patients and healthy controls.

Materials and methods

Study design and participants

We conducted a multi-institutional retrospective study at five clinics (Kamioooka Jinsei Clinic, Yokohama, Japan; Yokohama Jinsei Hospital, Yokohama; Bunko Jin Clinic, Yokohama; Kanazawa Clinic, Yokohama; and Oppama Jinsei Clinic, Yokosuka, Japan). These clinics offered free testing opportunities to measure SARS-CoV-2 spike protein antibody titers for patients undergoing HD and health care workers at the clinics. The testing was not conducted as a research project, but as a health checkup service provided by these clinics for those who wanted to know their antibody titers, with a total of two or three opportunities before and/or after the vaccinations. These clinics also declared that the testing for antibody titers was only for those who requested it, and patients and staff would never be penalized for not taking the test. The informed consent was clinically obtained by their physicians before each test. Not all participants attended all of the testing opportunities because it was not mandatory. The measurement of antibody titers was outsourced to an external laboratory and the blood samples were immediately discarded by the laboratory after measurement. The results were notified to the participants themselves and recorded in their medical records for patients undergoing HD and in health care information for medical staff. After we designed this study, we retrospectively collected the data from these medical records. This process was described in our research protocol and approved by the Ethics Committees of Yokohama City University Hospital (IRB approval number: F 220,200,037). We enrolled 840 participants, divided into a group who underwent HD three times a week (i.e., the hemodialysis group) and a group of healthy health care workers (i.e., the control group). All participants were aged over 20 years, had received the first vaccination between April and July 2021, and had received the second vaccination between May and August 2021, according to the Japanese government COVID-19 vaccination strategy. All participants had been vaccinated according to the vaccination schedule specified in the BNT162b2 (Comirnaty®, Pfizer-BioNTech) vaccine summary of product characteristics.

We excluded seven participants with documented SARS-CoV-2 infection before or during the study. We also excluded 265 participants 1) who had been treated for cancer within the past 1 year; 2) who had been taking any steroids or immuno-suppressants within the past 1 year; 3) who had been under peritoneal dialysis or renal transplantation; 4) who had been diagnosed with immunodeficiency syndrome; 5) whose records included missing values; or 6) whose anti-S1immunoglobulin (Ig) G antibody titers ≤ 50 AU/mL at Month 1 (n = 8; all in the HD group) [24]. Finally, we selected 412 HD patients and 156 controls.

Measurement

Anti-S1 IgG antibody

Titers of anti-SARS-CoV-2 spike protein S1 IgG antibody were serologically measured at 1, 2, 3, and 6 months after the second dose of vaccine (i.e., Month 1, Month 2, Month 3, and Month 6) in the hemodialysis group. Titers for Month 1, Month 3, and Month 6 were available in the control group. Measurement was not conducted at all time points, but titers at Month 1 and Month 6 were mandatory for inclusion in the current study. We used the Abbott Architect SARS-CoV-2 IgG II Quant chemiluminescent microparticle immunoassay to detect IgG antibodies to the receptor-binding domain of the S1 subunit of the SARS-CoV-2 spike protein, per the manufacturer’s instructions. The reportable measurement range of the assay is 6.8–80,000 AU/mL. As for within-laboratory precision, the coefficient of variation is reported to be between 4.2% and 5.1% [24]. Levels under the detection limit for the antibody were considered equivalent to 6.79 AU/mL in this study because the detection limit was 6.8 AU/mL.

Others

In the HD group, blood samples were collected on the first dialysis day of the week, 3 days after the previous dialysis day. The dry weight was used as body weight in the hemodialysis group, which was determined by each physician, depending on the clinical findings. Body mass index (BMI) was calculated by dividing body weight (kg) by height (m) squared. For details of the other collected data, see Text S1) [25]. The clinical history and dialysis prescription were obtained from the medical records. Information about use of vitamin D, serum albumin levels, hemoglobin levels, and Kt/V is also collected, since these items are reportedly associated with SARS-CoV-2 spike protein antibody titer after the vaccination in patients undergoing HD [22, 23, 26, 27]. Data for the control group were unavailable in this study, except for anti-S1 IgG antibody titers, age, sex, height, body weight, and history of SARS-CoV-2 infection.

Statistical analysis

First, we compared the antibody titers at Month 6 (Ab-titer-6) between the HD group and the control group using the Mann–Whitney U test. Because of the skewed distribution of the titers, we performed analysis of covariance (ANCOVA) to detect the difference in log-transformed Ab-titer-6 with multi-variable adjustment. Model 1 included age, sex, and BMI. Model 2 additionally included these covariates and log-transformed antibody titers at Month 1 (log Ab-titer-1).

Second, we assumed that the changes in antibody titers follow the exponential equation, which is known as the “one-compartment model,” [26] as follows:

$$A(t) = \alpha e^{ \wedge } ( - \beta t)$$
(1)

where A (t) equals the serum titer level at an arbitrary time t, α is a constant that equals the titer at t = 0, and β is a constant known as the elimination rate constant. By natural logarithm transformation of Eq. 1:

$$\log A(t) = \log \alpha - \beta t$$
(2)

According to Eq. 2, we can calculate β using a linear regression model with log-transformed antibody titers as a slope. Therefore, we used a mixed-effects linear regression model with random intercepts and random slopes on the log-transformed antibody titers and calculated the slope (i.e., –β) for each individual case. The onset of time (t = 0) was defined as Month 1 in the model. We compared this slope (–β) between the HD group and the control group using ANCOVA with multi-variable adjustment. Model 1 was unadjusted. Model 2 was adjusted by age, sex, and BMI. Plasma half-life (t = T1/2) was calculated using β [28]:

$$T_{\frac{1}{2}} = \log 2 \cdot \beta^{ \wedge } ( - 1)$$
(3)

We also explored the association of clinical or laboratory characteristics related to HD treatment (e.g., Kt/V, use of vitamin D) with Ab-titer-1 or Ab-titer-6 in the HD group (see Text S1).

We compared the antibody titers at Month 1 (Ab-titer-1) between the HD group and control group using ANCOVA. As a sensitivity analysis, we used a non-linear mixed-effects regression model to directly estimate the constants in Eq. 1 without using log-transformation (see Text S1).

The assumption for normality and homoscedasticity was visually inspected using histograms and normal quantile–quantile plots. Possible violations of the assumptions in the regressions were examined by visually inspecting scatter plots of residuals and predicted values. All analyses were performed with R version 4.0.4 (The R Project for Statistical Computing, Vienna, Austria) using the default strings of the nlme 3.1–152, lme4 1.1–26, and car 3.0–10 packages. Significance was defined as p < 0.05 using two-sided tests.

Results

The final analytic sample for the primary analyses included 412 patients in the HD group (mean ± standard deviation [SD] of age was 69.2 ± 12.2 years and 28.4% were female patients), and 156 health care workers in the control group (mean ± SD of age was 45.0 ± 12.4 years and 76.3% were female individuals). Clinical history, laboratory data, and prescription of dialysis in the HD group are shown in Table 1. In total, 221 (53.6%) HD patients had a history of hypertension and 192 (46.6%) had a history of diabetes; 87.4% had been treated with the online hemodiafiltration method.

Table 1 Characteristics of participants at baseline
Full size table

The SARS-CoV-2 spike protein antibody titers and positivity rate (i.e., ≥ 50 AU/mL) at each follow-up point are shown in Table 2 and Table S1, respectively. The HD group showed significantly lower levels of Ab-titer-6: median (95% confidence interval [CI]) 353.4 (178.4, 656.3) AU/mL versus 812.0 (498.3, 1342.7) AU/mL, (p < 0.001). 24 (5.8%) participants in the HD group showed seronegative levels at Month 6 whereas all control participants remained seropositive until Month 6 (Table S1). Log-transformed Ab-titer-6 levels were 5.8 ± 1.1 AU/mL and 6.8 ± 0.9 AU/mL, respectively. After multi-variable adjustment, the adjusted mean difference in log-transformed Ab-titer-6 between the HD group and the control group was –0.28 (–0.48, –0.08) (p < 0.001). Using back-transformation of this difference, we calculated the adjusted ratio of Ab-titer-6 between the two groups (i.e., the HD group [AU/mL] per the control group [AU/mL]), which was estimated as 0.76 (0.62, 0.93). At Month 1, the HD group showed significantly lower levels of Ab-titer-1: median (95% CI) 2617.1 (1296.7, 5240.8) AU/mL versus 7285.4 (4403.9, 11,000.0) AU/mL. The adjusted mean difference in Ab-titer-1 was –0.45 (–0.72, –0.17), (p < 0.01). Fig. 1 shows the exponential changes in titers and linear changes in the log-transformed levels.

Table 2 Changes in titers of SARS–CoV–2 spike protein antibody from the 2nd vaccination
Full size table
Fig. 1
figure 1

Changes in the titers of anti-SARS-CoV-2 spike protein antibody. A Exponential changes in the absolute values of anti-SARS-CoV-2 spike protein antibodies are shown. Dots and error bars show the back-transformed means ± standard deviation of antibody titers in the control group (circular dots) and in the HD group (triangular dots) from the natural logarithmic form. B Linear changes in the natural log-transformed titers of SARS-CoV-2 spike protein antibodies are shown. Dots and error bars show the means ± standard deviation of log-transformed antibody titers in the control group (circular dots) and the HD group (triangular dots). M1–M6 = Month 1 – Month 6, which means 1–6 months after the second vaccination, respectively. HD hemodialysis

Full size image

Using a linear mixed-effects model toward the log-transformed titers in the overall group, we calculated each individual's slope (i.e., –β) and intercept (i.e., logα). Table 3 shows the estimates between the HD group and the control group. The estimated –β values in the HD group and control group were –4.7 ± 1.1 and –4.7 ± 1.4 (year−1), respectively. In both the unadjusted and adjusted models, no significant difference in the slope (–β) was found between the two groups. Each plasma half-life (t = T1/2) estimated in the adjusted model was 1.70 (1.49, 1.99) months in the HD group and 1.78 (1.59, 2.02) months in the control group, respectively. We also constructed a non-linear mixed-effects regression model as a sensitivity analysis. Results were similar in terms of comparable –β values between the HD group and control group, with a minimal random effect: α = 4309.1 (AU/mL), –β = – 5.1 (year−1), T1/2 = 1.64 (months) in the HD group and α = 9388.6 (AU/mL), –β = –5.1 (year−1), T1/2 = 1.64 (months) in the control group. The estimated exponential curves of the titers in the two groups, which were well fitted for the actual observed titers, are shown in Figure S1.

Table 3 Comparison of slope for SARS-CoV-2 spike protein antibody from one month after the 2nd vaccination
Full size table

In exploratory analyses, older age, lower hemoglobin level, and lower albumin level were related to a lower level of log Ab-titer-1 in both the unadjusted and adjusted models (Table S2). Use of vitamin D was significantly associated with a higher level of log Ab-titer-1 (estimates: 0.23 [0.02, 0.44)], p < 0.05). Concerning the association with Ab-titer-6, lower Ab-titer-1 was significantly associated with a lower log Ab-titer-6 in both the unadjusted and adjusted models (estimates per 1 SD: 0.90 [0.83, 0.97], p < 0.001). Similar to the association with log Ab-titer-1, older age and lower albumin level were associated with lower levels of log Ab-titer-6 in the adjusted model using age, sex, and BMI. However, after adjustment for the Ab-titer-1, these did not reach statistical significance (Table S3).

Discussion

To the best of our knowledge, this is the first report in Japan to compare the SARS-CoV-2 spike protein antibody titers between patients undergoing HD and healthy individuals over a long period, up to 6 months after the second dose of SARS-CoV-2 mRNA vaccine. Additionally, in the present study, we examined the attenuation speed (slope) of SARS-CoV-2 spike protein antibody titer after the second vaccination in HD patients and healthy controls. The results showed that SARS-CoV-2 spike protein antibody titers in HD patients were significantly lower than those in healthy controls at 6 months as well as at 1 month (peak time) after the second vaccination. Intriguingly, the attenuation speed of SARS-CoV-2 spike protein antibody titers after vaccination was not obviously different between HD patients and healthy controls. Furthermore, multivariate analysis revealed that low peak antibody titers (1 month after the second vaccination) contributed to the low antibody titers at 6 months in HD patients. Collectively, these results suggest that SARS-CoV-2 spike protein antibody titers may be maintained over a longer period if peak levels of antibody titer after vaccination can be raised sufficiently in HD patients. An increase in antibody titers with a third booster vaccine may be the key to protection against COVID-19 among patients undergoing HD.

In individuals with no history of SARS-CoV-2 infection, anti-SARS-CoV-2 spike protein antibody titers reach their peak 2 weeks to 1 month after the second dose of the BNT162b2 vaccine [29,30,31]. In HD patients, the increase in antibody titer after vaccination is delayed by approximately 1 week compared with that in healthy controls but reaches its peak well past 1 month after the second vaccination [15, 31]. HD patients are less likely to have an increase in SARS-CoV-2 spike protein antibody titer after vaccination compared with healthy individuals. Peak levels of SARS-CoV-2 spike protein antibody titer in HD patients after vaccination are reportedly around 1/9 to 1/3 those in healthy individuals [17, 18, 26, 32]. In the present study among Japanese patients receiving HD, antibody titers 1 month after the second vaccination were significantly lower than those in healthy controls and were approximately one-third those of healthy individuals.

In patients undergoing HD, the results of previous studies have demonstrated that use of immunosuppressive and anti-cancer agents, age, low serum albumin, low lymphocyte count, and low Kt/V are associated with lower peak levels of SARS-CoV-2 spike protein antibody titer after vaccination [16, 17, 26, 32,33,34]. Additionally, calcitriol treatment is reportedly associated with SARS-CoV-2 spike protein IgG concentration 1 month after the second dose of the BNT162b2 vaccine in HD patients [27]. As a molecular mechanism by which vitamin D enhances the immune response to vaccines, dendritic cells activated by stimulation via toll-like receptors after vaccination are further activated by vitamin D and migrate to lymphoid tissues to stimulate antigen-specific T and B cells, resulting in sufficient antibody production [35, 36]. In the present study, SARS-CoV-2 spike protein antibody titers 1 month after the second vaccination were positively correlated with use of vitamin D, serum albumin and hemoglobin levels and negatively correlated with age. However, Kt/V was not associated with peak antibody titer 1 month after vaccination in the present study, which may be owing to differences in patient backgrounds, including therapeutic condition and health insurance systems for HD between Israel, Western countries, and Japan.

With respect to long-term humoral immunity after vaccination, several studies among health care workers have reported that SARS-CoV-2 spike protein antibody titers and neutralizing antibodies progressively decline up to 6 months after vaccination [3, 4, 37, 38]. Studies of HD patients in Germany showed that SARS-CoV-2 spike protein antibody titers peaked 1 month after the second vaccination in dialysis patients and declined to approximately 50% of peak values 3–4 months later [5, 23, 39]. The results of the present study showed that 6 months after the second dose of the BNT162b2 vaccine, SARS-CoV-2 spike protein antibody titers were markedly decreased (approximately one-seventh) compared with the peak value 1 month after vaccination.

There are few reports directly comparing antibody titers between patients receiving HD and healthy controls over a long period. A recent study in Israel reported that SARS-CoV-2 spike protein antibody titers in dialysis patients 6 months after the second dose of the BNT162b2 vaccine were significantly lower (approximately one-third) than those in healthy controls [40]. Consistent with this previous report, the results of the present study demonstrated that SARS-CoV-2 spike protein antibody titers in Japanese HD patients were significantly lower (approximately one-third) than those in healthy controls 6 months after receiving the second dose of the BNT162b2 vaccine. With respect to factors associated with long-term antibody titers, studies among health care workers have shown that low antibody titers 6 months after vaccination are related to age, immunosuppressive medications, low initial doses of vaccine, and low peak values 1 month after vaccination [3, 38]. In dialysis patients, age, use of immunosuppressive agents, low albumin levels, and low hemoglobin levels are reportedly associated with low antibody titers 6 months after vaccination [22, 23]. In our study, age and low albumin levels were associated with low antibody titers 6 months after receiving the second shot of the BNT162b2 vaccine. However, these factors were no longer associated with antibody titers at 6 months, after adjustment for antibody titers at 1 month. Combined with the present results that attenuation speed did not differ between HD patients and healthy controls, low peak levels after vaccination may contribute to low antibody titers over the long term in HD patients.

The relationship between COVID-19 mRNA vaccine antibody titers and prevention of severe illness is controversial. In patients aged 60 years or older, the rate of severe illness due to SARS-CoV-2 infection reportedly increased 1.8 (95% CI 1.1–2.9) 6 months after the second vaccination compared with 3 months after the vaccination [41]. On the other hand, mRNA vaccines have the effects of conferring cellular immunity and immune memory in addition to humoral immunity. Cellular immunity is known to suppress the severity of SARS-CoV-2 infection independent of anti-SARS-CoV-2 spike protein IgG levels [42]. Several studies have reported that the preventive effects of mRNA vaccine on severe illness sustain even after a marked decrease in SARS-CoV-2 spike protein antibody titers [43, 44]. Therefore, the results of the present study, which showed lower antibody titers in response to mRNA vaccination in patients undergoing HD compared to controls, could not conclude that the effects of mRNA vaccination on prevention of the severity of SARS-CoV-2 infection is inferior in patients undergoing HD compared to controls.

The present study had some limitations. First, participants’ backgrounds differed significantly between the two groups. Especially, the age distribution and the gender are very different between the HD patient group and the control group. To select medical staff as the control group is an important limitation. Second, this study had a retrospective study design. Third, it was difficult to exclude asymptomatic participants infected with SARS-CoV-2. Fourth, only adult patients undergoing HD (not less than 20 years old) were included. Fifth, the only vaccine investigated was BNT162b2 (Pfizer-BioNTech); patients who received other mRNA vaccines were not included. Sixth, we examined SARS-CoV-2 spike protein antibody titers but not neutralizing antibodies or cellular immunity. Therefore, further studies are needed to elucidate the immunoprotective effects of SARS-CoV-2 vaccines against COVID-19 in patients undergoing HD.

In conclusion, the present study findings demonstrated that SARS-CoV-2 spike protein antibody titers in patients undergoing HD in Japan were significantly lower than those in healthy controls 6 months after vaccination. In HD patients, low antibody titers 6 months after vaccination were associated with low peak levels whereas the attenuation speed of antibody tiers did not differ from that of healthy controls. Taken together, these results indicate the importance of higher peak antibody titers in HD patients. Given the fact that low hemoglobin and albumin levels were found to be associated with low peak antibody titers, it is important for HD patients to improve their nutrition and address anemia when receiving vaccination. Furthermore, in dialysis patients, SARS-CoV-2 spike protein antibody titers are reported to increase markedly by approximately 25-fold after the third booster dose of BNT162b2 vaccine compared with the second dose [45]. Therefore, a sufficient increase in antibody titers with a third booster vaccine may be the key to protection against COVID-19 in patients receiving HD.