Longitudinal dynamics of humoral immune responses following serial COVID-19 vaccinations and breakthrough infections in dialysis patients

ABSTRACT

Patients with end-stage kidney disease (ESKD) undergoing dialysis have impaired vaccine responses. Many countries recommend extended primary series and regular COVID-19 boosters for this group, but data on long-term humoral responses after repeated doses or breakthrough infection are limited. In this prospective cohort study of 498 dialysis patients, most received homologous ChAdOx1 nCoV-19 as a primary series, mRNA-1273 for the third and fourth doses, and bivalent Moderna vaccines for later doses. Neutralizing antibodies (via surrogate virus neutralization test) and anti – receptor-binding domain (RBD) antibodies were measured up to 15 months post-vaccination or infection. The primary endpoint was seroprotection (neutralization inhibition ≥30%); the secondary was anti-RBD seroprotection (≥100 U/mL). Seroprotection increased from 16% (neutralizing) and 2% (anti-RBD) after the first dose to ~100% after the third and subsequent doses. Neutralizing inhibition rose from 5% (dose 1) to ~90% (doses 4–6). Anti-RBD titers declined 66%–85% by 6 months and >90% by 12–15 months. Younger age, receipt of a fourth dose, and vaccine platform were significant predictors of neutralizing titers. Breakthrough infection led to higher and more sustained anti-RBD titers, particularly in patients with hybrid immunity. Patients with stronger immunity had fewer symptoms. In conclusion, serial COVID-19 vaccinations elicited robust humoral responses in dialysis patients, with younger age, receipt of a booster dose, and vaccine platform, emerging as significant predictors. Although antibody titers declined over time, they were better maintained in those with hybrid immunity. These findings support the implementation of personalized booster strategies to optimize protection in immunocompromised populations.

Plain Language Summary

Patients receiving dialysis are at higher risk of severe illness and death from COVID-19. While COVID-19 vaccines have been effective in the general population, their performance in dialysis patients remains less well understood, especially after multiple booster doses and natural infections. This study followed a large group of dialysis patients for over 15 months to evaluate how their antibody levels changed after up to six COVID-19 vaccine doses and/or breakthrough infections. We found that younger patients developed stronger antibody responses than older patients. Receiving a booster vaccine dose further boosted antibody levels, and vaccine type also mattered: mRNA-1273 produced stronger responses, while MVC-COV1901 produced weaker responses. Those who had both vaccination and infection – known as “hybrid immunity” – had stronger and longer-lasting responses. Interestingly, individuals with asymptomatic infections showed better antibody responses than those with symptoms, suggesting that a more effective immune system might help prevent noticeable illness. Although antibody levels gradually declined over time, functional immunity remained detectable for many months. These findings support personalized vaccine strategies in immunocompromised populations.

Introduction

The coronavirus disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2), continues to pose a global health threat despite advances in clinical care and widespread population-level immunity.^1–3 Even in the post-pandemic era, new cases persist, maintaining public concern.^4,5 Patients with end-stage kidney disease (ESKD) undergoing dialysis remain at elevated risk for severe outcomes due to immune dysfunction and high comorbidity burden.^6,7 Compounding this vulnerability is a diminished vaccine-induced immune response, necessitating modified vaccination strategies.^8–10 As a result, health authorities in Taiwan and worldwide have implemented extended primary series and multiple booster doses of COVID-19 vaccines for this population.^11,12 However, data on the long-term durability and dynamics of humoral responses – especially across six vaccine doses or following breakthrough infections – are limited.

Although vaccination has reduced COVID-19–related morbidity and mortality, dialysis patients were excluded from most initial vaccine trials, leaving gaps in knowledge about their immunogenicity.¹³ While short-term serologic responses after early vaccinations have been reported in ESKD populations, few studies have longitudinally assessed antibody waning, seroprotection durability, or seroconversion across multiple doses or post-infection.¹⁴ Additionally, comparisons across vaccine platforms, breakthrough infection outcomes, and patient-level predictors of stronger responses (e.g., age) remain underexplored.¹⁵

Understanding humoral immunity dynamics is essential to optimizing vaccination in immunocompromised populations. Prior studies have shown that repeated mRNA vaccination enhances neutralizing responses in transplant and dialysis patients, though titers wane rapidly.^16–18 Breakthrough infection may confer longer-lasting protection – a phenomenon termed “hybrid immunity” – associated with more sustained responses in both healthy and high-risk populations.^19,20

This prospective cohort study aimed to address these gaps by evaluating longitudinal changes in neutralizing and anti – receptor-binding domain (RBD) antibody responses following up to six doses of COVID-19 vaccines or breakthrough SARS-CoV-2 infection in a large cohort of dialysis patients. We also examined predictors of robust humoral responses and compared outcomes across vaccine types, infection status, and patient subgroups.

Methods

Study design and participants

This prospective observational cohort study was conducted from July 2021 to July 2023 at a dialysis unit of a tertiary hospital in northern Taiwan. It was approved by the Institutional Review Board of Far Eastern Memorial Hospital (FEMH-110101-E) and adhered to the Declaration of Helsinki. Eligible participants were adults ( >20 years) receiving maintenance hemodialysis (HD) or peritoneal dialysis (PD), and willing to receive COVID-19 vaccination. Health care workers (HCWs) over 20 years old with normal renal function served as the control group. Written informed consent was obtained from all participants. The study followed STROBE guidelines.²¹

Vaccination schedule

Most dialysis patients received their first vaccine dose in June 2021, with HCWs vaccinated earlier in May. The primary series comprised two doses spaced three months apart in dialysis patients and two months apart in HCWs. In response to the Omicron variant, a third dose was approved for high-risk populations in December 2021. Dialysis patients received this dose one month after the second, as part of an extended primary series, while HCWs received it five months after their second dose. Both groups received their third dose between December 2021 and March 2022. Fourth to sixth doses were administered ≥3 months apart.

The study cohort exhibited considerable heterogeneity in vaccine combinations and inter-dose intervals due to evolving government policy, vaccine availability, and individual medical factors. Most participants began with ChAdOx1 nCoV-19 and transitioned to mRNA-based boosters (typically mRNA-1273). Interval variation reflected personalized clinical timing and public health updates. This real-world complexity enhances the generalizability of our findings to diverse clinical settings.

Humoral immune response assessment

We measured neutralizing and anti-RBD binding antibodies one month after the first dose, and at one and three months after each subsequent dose (2^nd–6^th). Anti-RBD antibodies were also assessed at three-month intervals between doses to monitor waning.

Neutralizing antibody levels targeting the wild-type receptor-binding domain (RBD) of the SARS-CoV-2 spike protein were assessed using the Surrogate Virus Neutralization Test (SVNT) assay (GenScript, USA), previously validated by Tan et al.²² A threshold of < 30% was considered negative for neutralizing antibody response, in line with the manufacturer’s instructions, which report a specificity of 99.93% and sensitivity of 95–100%. Quantitative results (in IU/mL) were provided only for samples with ≥30% inhibition.

Binding antibody responses to the wild-type RBD of the SARS-CoV-2 spike protein were evaluated using the Elecsys Anti – SARS-CoV-2 S assay (Roche Diagnostics), with titers ≥0.8 U/mL considered positive. A higher cutoff of ≥100 U/mL was used to indicate anti-RBD seroprotection, based on published associations with neutralizing activity in transplant and dialysis patients.²³

Breakthrough SARS-CoV-2 infection assessment

To distinguish vaccine-induced from infection-induced responses, anti-nucleocapsid (N) antibodies were assessed once at study enrollment using the Elecsys Anti – SARS-CoV-2 N assay (Roche), as N antibodies are elicited by natural infection but not by spike-only vaccines. This assay has a reported analytical specificity of 100% and a clinical specificity of 99.98%. Anti-N testing was not repeated at subsequent time points, because a positive result cannot reliably differentiate acute, prior, or repeated infections in dialysis patients. Instead, participants were surveyed monthly for infection history, and medical records were reviewed. Symptomatic breakthrough infections were defined by positive PCR and reported symptoms. Asymptomatic cases were defined as a ≥4-fold increase in anti-RBD titers between two time points without recent vaccination or reported symptoms; anti-N positivity and neutralization titers were not used in this definition. For analyses restricted to infection-naïve dialysis patients, individuals who developed breakthrough infection, regardless of symptoms, were excluded from the point of infection onwards and subsequently evaluated within the breakthrough infection cohorts.

Data collection

Demographics, comorbidities, medication use, and routine labs were collected from electronic records and structured questionnaires. For dialysis patients, data included dialysis vintage, Kt/V, serum albumin, hemoglobin, ferritin, intact parathyroid hormone (PTH), and normalized protein catabolic rate (nPCR).

Outcomes

The primary endpoint was neutralizing antibody seroprotection (≥30% inhibition). The secondary endpoint was anti-RBD seropositivity (≥100 U/mL).²³ We also analyzed geometric mean (GM) neutralization inhibition and anti-RBD titers over time. Additional endpoints included seroconversion factors (geometric mean fold increase), seroconversion rates (≥4-fold rise), and high-response rates (≥16-fold increase). Subgroup analyses compared responses by vaccine type and dose. We also assessed predictors of neutralizing titers and waning kinetics post-vaccination and post-infection. Comparisons were made between symptomatic vs. asymptomatic infections and dialysis vs. control groups. Age-matched analyses addressed potential confounding.

Statistical analysis

Continuous variables were presented as means ± SD or medians with interquartile ranges; categorical variables as counts and percentages. Antibody titers and seroconversion factors were reported as GMs with 95% CIs.²⁴ Longitudinal predictors of neutralizing antibody titers were evaluated using linear mixed-effects models that incorporated repeated measures across the 2^nd to 4^th vaccine doses and time-varying covariates. Neutralizing antibody titers were expressed as percent changes, and results are presented as estimates with 95% confidence intervals. Variables included in the models were age, sex, body mass index, dialysis adequacy (Kt/V), diabetes status, dialysis vintage, vaccine dose, vaccine type, serum albumin, hemoglobin, ferritin, intact parathyroid hormone, and normalized protein catabolic rate. Both basic models (adjusted for time from vaccination to sampling) and multivariate models (adjusted for all covariates simultaneously) were fitted, and overall P values were provided for categorical variables with three levels. Comparisons used Wilcoxon-Mann-Whitney, Chi-square, Fisher’s exact, or Kruskal-Wallis tests with Dunn’s post hoc analysis, as appropriate. A two-sided P-value <.05 was considered statistically significant. All statistical analyses were performed using SAS software (version 9.4; SAS Institute), and data visualizations were created using Microsoft Excel and R (version 4.0.5; R Foundation for Statistical Computing, Vienna, Austria).

Results

Participants characteristics

A total of 498 dialysis patients (438 hemodialysis, 60 peritoneal dialysis) were enrolled, with a mean age of 63 ± 12 years; 60% were female, and 49% had diabetes. The average body mass index (BMI) was 24 ± 4 kg/m² (Table 1). The study flow diagram is illustrated in Supplemental Figure 1.

Table 1.

Characteristics of study participants.

Characteristics	n = 498
Hemodialysis (n,%)	438 (88)
Peritoneal dialysis (n,%)	60 (12)
Age (year)	63 ± 12
Female (n,%)	201 (60)
Body height (cm)	162 ± 9
Dry weight (kg)	63 ± 13
Body mass index (kg/m2)	24 ± 4
Serum creatinine (mg/dL)	10.9 ± 2.5
Vintage (year)	6 ± 6
Coronary artery disease (n,%)	122 (26)
Congestive heart failure (n,%)	32 (7)
History of cerebrovascular accident (n,%)	39 (8)
Diabetes mellitus (n,%)	230 (49)
Hypertension (n,%)	292 (62)
Gout (n,%)	65 (14)
Liver cirrhosis (n,%)	6 (1)
On immunosuppressant (n,%)	9 (2)
On steroids (n,%)	10 (2)
On chemotherapy (n,%)	4 (1)
First vaccine type (n,%)	498 (100)
ChAdOx1 nCov19 (Oxford-AstraZeneca)	492 (99)
mRNA-1273 (Moderna)	6 (1)
Second vaccine type (n,%)	477 (96)
ChAdOx1 nCov19 (Oxford-AstraZeneca)	461 (97)
mRNA-1273 (Moderna)	13 (3)
MVC-COV1901 (Medigen)	2 (1)
BNT162b2 (Pfizer-BioNtech)	1 (1)
Primary vaccine series (n,%)	477 (96)
Homologous (ChAdOx1 nCov19/ChAdOx1 nCov19)	461 (97)
Homologous (mRNA-1273/mRNA-1273)	6 (1)
Heterologous (ChAdOx1 nCov19/mRNA-1273)	7 (1)
Heterologous (ChAdOx1 nCov19/Medigen)	2 (1)
Heterologous (ChAdOx1 nCov19/BNT)	1 (1)
Third vaccine type (n,%)	455 (91)
mRNA-1273 (Moderna) (full dose, 100 μg)*	429 (94)
BNT162b2 (Pfizer-BioNtech)	13 (3)
MVC-COV1901 (Medigen)	13 (3)
Type of combination in 1–3 doses	455 (91)
AZ/AZ/Moderna*	419 (92)
AZ/AZ/Medigen	11 (2)
AZ/AZ/BNT	11 (2)
Moderna/Moderna/Moderna*	5 (1)
AZ/Moderna/Moderna*	4 (1)
AZ/Moderna/BNT	2 (1)
AZ/Medigen/Medigen	2 (1)
AZ/BNT/Moderna*	1 (1)
Fourth vaccine type (n,%)	399 (80)
mRNA-1273 (Moderna)	374 (94)
BNT162b2 (Pfizer-BioNtech)	6 (2)
MVC-COV1901 (Medigen)	6 (2)
NVX-CoV2373 (Novavax)	3 (1)
Moderna bivalent BA.1	5 (1)
Moderna bivalent BA.4/5	5 (1)
Fifth vaccine type (n,%)	170 (34)
Moderna bivalent BA.1	102 (60)
Moderna bivalent BA.4/5	67 (39)
BNT162b2 (Pfizer-BioNtech)	1 (1)
Sixth vaccine type (n,%)	4 (1)
Moderna bivalent BA.4/5	4 (100)
Interval between first vs second dose (days)	98 (92, 100)
Interval between first vs third dose (days)	198 (196, 201)
Interval between first vs fourth dose (days)	308 (306, 308)
Interval between first vs fifth dose (days)	506 (486, 570)
Interval between first vs sixth dose (days)	611 (574, 650)
Antibody measurement at 1 month after the first dose (1D1M) (days)	21 (21, 23)
Antibody measurement at 1 month after the second dose (2D1M) (days)	42 (42, 48)
Antibody measurement at 3 months after the second dose (2D3M) (days)	78 (77, 84)
Antibody measurement at 6 months after the second dose (2D6M) (days)	161 (154, 163)
Antibody measurement at 1 month after the third dose (3D1M) (days)	42 (42, 43)
Antibody measurement at 3 months after the third dose (3D3M) (days)	98 (97, 98)
Antibody measurement at 6 months after the third dose (3D6M) (days)	155 (147, 161)
Antibody measurement at 9 months after the third dose (3D9M) (days)	271 (261, 284)
Antibody measurement at 12 months after the third dose (3D12M) (days)	365 (356, 371)
Antibody measurement at 15 months after the third dose (3D15M) (days)	446 (441, 469)
Antibody measurement at 1 month after the fourth dose (4D1M) (days)	16 (16, 16)
Antibody measurement at 3 months after the fourth dose (4D3M) (days)	79 (79, 79)
Antibody measurement at 6 months after the fourth dose (4D6M) (days)	170 (170, 170)
Antibody measurement at 9 months after the fourth dose (4D9M) (days)	261 (261, 261)
Antibody measurement at 12 months after the fourth dose (4D12M) (days)	359 (359, 359)
Antibody measurement at 15 months after the fourth dose (4D15M) (days)	443 (443, 443)
Antibody measurement at 1 month after the fifth dose (5D1M) (days)	30 (24, 36)
Antibody measurement at 3 months after the fifth dose (5D3M) (days)	86 (79, 93)
Antibody measurement at 6 months after the fifth dose (5D6M) (days)	170 (153, 181)
Antibody measurement at 9 months after the fifth dose (5D9M) (days)	261 (254, 268)
Antibody measurement at 1 month after the sixth dose (6D1M) (days)	29 (22, 36)
Antibody measurement at 3 months after the sixth dose (6D3M) (days)	92 (78, 99)

Data are presented as mean ± SD, median (IQR) or counts (percentages), as appropriate. *Participants received a full dose (100 μg) of mRNA-1273 (Moderna) vaccine for the third dose and a half dose (50 μg) for other doses.

Abbreviation. AZ, ChAdOx1 nCov-19 (Oxford-AstraZeneca); BNT, BNT162b2 (Pfizer-BioNtech); IQR, interquartile range; SD, standard deviation.

Vaccine administration

Nearly all patients (97%) received two homologous doses of ChAdOx1 nCoV-19. For the additional (third) dose, 94% of patients received a full 100 μg dose of mRNA-1273—double the standard 50 μg dosage for the general population. Subsequent boosters (fourth to sixth doses) were predominantly mRNA-based or bivalent formulations, with 60% and 39% receiving BA.1 or BA.4/5 bivalent vaccines, respectively, for the fifth dose.

Timing and sampling

Median intervals between the first and second doses, and third through sixth doses, were 98, 198, 308, 506, and 611 days, respectively. Antibody levels were assessed ~1 month post-vaccination, with additional follow-up at 3, 6, 9, 12, and 15 months after the third and fourth doses.

Antibody responses after six doses

Figures 1–3 and Supplemental Table 1 present the humoral antibody responses at multiple time points following up to six doses of COVID-19 vaccines among infection-naïve dialysis patients.

Figure 1.

Seroprotection rates based on (A) percent neutralization inhibition, and (B) anti – receptor-binding domain (RBD) antibody titers after six COVID-19 vaccine doses among dialysis patients.

Seroprotection was defined as percent neutralization ≥30% (A) and anti-RBD antibody titers ≥100 U/mL (B). Panel A presents 11 time points; Panel B presents 22 time points following up to six COVID-19 vaccine doses among dialysis patients.

Abbreviations. D, dose; M, month; RBD, receptor binding domain.

Figure 2.

Humoral antibody response based on (A) percent neutralization inhibition, (B) neutralizing antibody titers (U/mL), and (C) anti-RBD antibody titers (U/mL) after six COVID-19 vaccine doses among dialysis patients.

Each dot represents an individual participant. Error bars indicate geometric mean (A: percent; B & C: titer) with 95% confidence intervals. Numeric GM (A) and GMT (B, C) are shown at each time point. Data include 11 time points for panels A and B, and 22 time points for panel C.

Abbreviations. CI, confidence intervals; D, dose; GM, geometric mean; GMT, geometric mean titer; M, month; RBD, receptor binding domain.

Figure 3.

Seroconversion factor of (A) neutralizing antibodies, and (B) anti-RBD antibodies after six COVID-19 vaccine doses in dialysis patients.

Abbreviations. D, dose; GMT, geometric mean titer; RBD, receptor binding domain.

At one month after the first dose (1D1M), seroprotection rates were low (neutralization: 16% in Figure 1A, anti-RBD: 2% in Figure 1B) but rose to 78% and 86% after the second dose and ~ 100% after the third dose. Responses remained high after doses 4–6, except for MVC-COV1901 recipients, who had lower seroprotection (Supplemental Figure 2A).

Geometric mean neutralization inhibition rose from 5% (1D1M) to 91%–93% (4D1M–6D1M) and remained stable at 3 months (Figure 2A; Supplemental Table 1). Neutralizing antibody titers increased with each dose: 5 U/mL (1D1M) to 5,464 U/mL (6D3M), with no decline at 3 months post-dose (Figure 2B; Supplemental Table 1).

The GMTs of anti-RBD antibody titers also increased progressively: 1.6 at 1D1M 29,521 at 5D1M, and 23,691 at 6D1M (Figure 2C). For doses two through six, anti-RBD titers peaked at one month and gradually declined thereafter. The MVC-COV1901 vaccine again showed a weaker response in dialysis patients, with lower values across all antibody metrics: 43% for neutralization inhibition (vs. 81–96% for other vaccines), 157 U/mL for neutralizing antibody titers (vs. 771–2329 U/mL), and 2126 U/mL for anti-RBD titers (vs. 11130–22,165 U/mL). Apart from this, vaccine-specific trends mirrored the overall response patterns across all three humoral markers (Supplemental Figure 3A–C; Supplemental Table 2).

Seroconversion factors and rates were highest after the first two doses but declined with successive boosters (Figures 3, Supplemental Table 3; Figure 4). These patterns were consistent across vaccine types (Supplemental Table 4).

Figure 4.

Waning of anti-RBD antibody titers at various time points after six vaccine doses in dialysis patients.

Antibody waning was calculated by subtracting titers at 1 month post-vaccination from values at later time points (e.g., 3 months minus 1 month post-vaccination).

Abbreviations. D, dose; M, month; RBD, receptor binding domain.

Correlates of antibody response

In longitudinal mixed-effects models incorporating repeated neutralizing antibody measures across the 2^nd to 4^th vaccinations and time-varying covariates, younger age ( <65 years) was a consistent predictor of higher neutralizing antibody titers (estimate: +43%, 95% CI 10–86%, P < .01). Compared with the second dose, the third dose showed a directionally positive but nonsignificant association (estimate: +610%, 95% CI −32 to 7,295, P = .10), whereas the fourth dose was significantly associated with higher titers (estimate: +1,267%, 95% CI 30 to 14,225, P = .03). By vaccine platform (mRNA-1273 as reference), MVC-COV1901 was independently associated with lower neutralizing titers (estimate: −83%, 95% CI −93 to −54, P < .01), while NVX-CoV2373, ChAdOx1 nCoV-19 and BNT162b2 did not differ significantly from mRNA-1273. Other covariates – including sex, body mass index, dialysis adequacy, diabetes status, vintage, serum albumin, hemoglobin, ferritin, intact parathyroid hormone, and normalized protein catabolic rate – were not independently associated with longitudinal neutralizing titers (all P > .05; Table 2).

Table 2.

Factors associated with percent change in neutralizing antibody titers following serial COVID-19 vaccination in dialysis patients (linear mixed-effects models).

	Basic model	Between-group difference	Overall	Multivariate model	Between-group difference	Overall
Variables	Estimate (95% CI)	P value	P value	Estimate (95% CI)	P value	P value
Age (year)
≥65	0 (reference)			0 (reference)
<65	51 (16, 96)	<.01		43 (10, 86)	<.01
Sex
Female	0 (reference)			0 (reference)
Male	21 (−7, 59)	.16		15 (−13, 54)	.33
BMI (kg/m2)
<22	0 (reference)		.69	0 (reference)		.69
22 to 25	9 (−20, 50)	.59		9 (−20, 48)	.58
≥25	15 (−17, 61)	.4		16 (−18, 64)	.39
Kt/V
Low	0 (reference)			0 (reference)
High	−12 (−32, 13)	.31		−8 (−28, 18)	.52
Diabetes mellitus
Yes	0 (reference)			0 (reference)
No	35 (4, 76)	.03		23 (−7, 62)	.14
Vintage (year)
≥10	0 (reference)		.29	0 (reference)		.63
3 to 10	−14 (−39, 20)	.37		−4 (−31, 34)	.81
<3	−25 (−48, 7)	.12		−15 (−41, 23)	.39
Vaccine dose
2nd dose	0 (reference)		<.01	0 (reference)		<.01
3rd dose	538 (418, 687)	<.01		610 (−32, 7295)	.1
4th dose	1122 (657, 1874)	<.01		1267 (30, 14225)	.03
Vaccine type
mRNA-1273	0 (reference)		<.01	0 (reference)		.01
NVX-CoV2373	187 (−46, 1429)	.22		23 (−79, 609)	.82
MVC-COV1901	−83 (−94, −55)	<.01		−83 (−93, −54)	<.01
BNT162b2	11 (−61, 214)	.85		44 (−55, 363)	.54
ChAdOx1 nCov19	−84 (−87, −81)	<.01		12 (−89, 1072)	.93
Serum albumin (g/dL)
<3.8	0 (reference)		.14	0 (reference)		.22
3.8 to 4.1	31 (−6, 83)	.11		29 (−3, 73)	.08
≥4.1	2 (−27, 43)	.9		18 (−14, 62)	.31
Hemoglobin (g/dL)
<10.6	0 (reference)		.38	0 (reference)		.12
10.6 to 12.1	11 (−18, 49)	.51		0.6 (−22, 29)	.96
≥12.1	−10 (−35, 25)	.53		−21 (−41, 5)	.11
Ferritin (ng/ml)
<379	0 (reference)			0 (reference)
≥379	7 (−17, 36)	.61		−9 (−27, 13)	.39
Intact PTH (pg/ml)
<78	0 (reference)		.39	0 (reference)		.2
78 to 433	−11 (−36, 22)	.46		−16 (−37, 11)	.22
≥433	9 (−24, 57)	.65		5 (−24, 46)	.78
nPCR
<0.96	0 (reference)		<.01	0 (reference)		.92
0.96 to 1.34	42 (6, 91)	.02		4 (−19, 34)	.73
≥1.34	94 (35, 178)	<.01		6 (−23, 48)	.71

Neutralizing antibody titers were analyzed using linear mixed-effects models incorporating repeated measures across the 2^nd to 4^th vaccine doses and time-varying covariates. “Basic models” included the time interval from vaccination to antibody measurement and the covariates themselves. “Multivariate models” included all covariates simultaneously, in addition to adjustment for time interval. Percent change estimates are shown with 95% confidence intervals; positive values indicate increased titers and negative values indicate decreased titers. For covariates with three levels, an overall P value for comparison across groups is provided. Kt/V was dichotomized at the median: the high Kt/V group was defined as weekly Kt/V >1.8 for peritoneal dialysis patients and Kt/V (Daugirdas) >1.5 for hemodialysis patients; the low Kt/V group included values at or below these thresholds.

Abbreviations: BMI, body mass index; CI, confidence interval; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone.

Waning immunity

As shown in Figure 4, anti-RBD titers declined 33%–59% at 3 months, 66%–85% at 6 months and >90% by 12–15 months. In contrast, neutralizing titers increased at 3 months post-vaccination, indicating asynchronous kinetics (Supplemental Figure 5).

Breakthrough infections

Among the dialysis cohort, 38% (n = 187) experienced breakthrough infections confirmed by PCR or antigen testing. Anti-RBD titers peaked at 1 month post-infection in most subgroups and declined thereafter, except for patients receiving their fifth dose before infection, who peaked at 3 months (Figure 5).

Figure 5.

Anti-RBD antibody responses following breakthrough SARS-CoV-2 infection among dialysis patients who received: (A) 4 vaccine doses before infection, (B) 4^th dose after prior infection, (C) 5 doses before infection, and (D) 5^th dose after prior infection.

Each dot indicates a participant’s anti-RBD titer (U/mL). Error bars represent GMT with 95% CI. Numeric GMTs are displayed at each time point.

Abbreviations. CI, confidence intervals; GMT, geometric mean titer; M, month; RBD, receptor binding domain.

Comparison of waning: vaccination vs. breakthrough infection

In patients with breakthrough SARS-CoV-2 infections, antibody waning was markedly attenuated compared to vaccine-only recipients. As shown in Supplemental Figure 6, patients with hybrid immunity – either from infection followed by vaccination or vice versa – maintained significantly higher anti-RBD titers over time. For example, among 4-dose recipients, anti-RBD titers declined by 67% at 6 months post-infection followed by vaccination versus 84% in the vaccine-only group (Supplemental Figure 6A). Similarly, 5-dose recipients with breakthrough infection retained over 80% of their peak titers at 9 months, compared to a 90% drop in vaccine-only counterparts (Supplemental Figure 6B). These findings underscore the superior durability of hybrid immunity in dialysis patients, with important implications for booster strategies.

Asymptomatic vs. symptomatic infections

Among 115 asymptomatic patients identified serologically, post-infection anti-RBD titers and seroconversion factors were significantly higher than in symptomatic cases (GMR for titers: 1.63; 95% CI, 1.27–2.09; P < .01), suggesting stronger immune control in subclinical infections (Supplemental Table 6; Figure 7).

Dialysis vs. controls

Dialysis patients showed reduced responses vs. age-matched controls after early vaccine doses (e.g., 6% vs. 49% neutralization at 1D1M; P < .01). However, at 2D3M and 4D1M, dialysis patients had higher titers, potentially due to longer inter-dose intervals and mRNA-1273 use. Nevertheless, dialysis patients experienced lower peak titers and higher viral loads during breakthrough infection (Supplemental Figures 8–10; Tables 7–9).

Discussion

This prospective cohort study offers one of the most detailed longitudinal assessments to date of humoral immune responses following serial COVID-19 vaccinations and breakthrough infections in dialysis patients. While early responses to initial vaccine doses were limited, additional mRNA-based boosters – particularly the third and fourth – elicited robust seroprotection and marked increases in antibody titers. Over time, anti-RBD titers declined significantly, though more gradually in those with hybrid immunity. Importantly, younger patients consistently mounted stronger neutralizing responses, and vaccine platform played a role: MVC-COV1901 was associated with lower neutralizing titers compared with mRNA-1273, whereas NVX-CoV2373, ChAdOx1 and BNT162b2 did not differ significantly. These findings emphasize the combined importance of age, booster dosing, and vaccine platform in shaping long-term immunity in dialysis patients. We also observed a dissociation in waning kinetics between anti-RBD and neutralizing antibody responses: neutralizing titers increased at three months post-vaccination despite concurrent declines in anti-RBD levels. This asynchronous pattern may reflect ongoing affinity maturation, and it underscores the limitations of relying solely on anti-RBD binding antibody levels to evaluate immune protection. Furthermore, asymptomatic breakthrough infections were associated with stronger post-infection antibody responses than symptomatic cases, suggesting that individuals with greater humoral competence may more effectively control infection. The identification of 115 asymptomatic cases – confirmed by serologic surges – suggests underestimation of infection burden in this population. Comparisons with age-matched healthy controls revealed diminished antibody responses in dialysis patients after the first and second vaccine doses, though this gap narrowed following subsequent boosters, particularly with longer vaccination intervals and consistent use of mRNA-based vaccines. Nonetheless, dialysis patients continued to exhibit lower peak antibody titers and higher viral loads at the time of breakthrough infection, underscoring persistent immunologic vulnerability in this high-risk population.

The significant immunologic gain occurred following the third dose, aligning with prior studies in immunosuppressed populations.^18,23 This robust increase is also evident in our cohort from both anti-RBD and neutralizing antibody titers (Figures 1 and 2). In the longitudinal mixed-effects model, the fourth dose – but not the third – was independently associated with higher neutralizing titers, suggesting that while the third dose produced a major quantitative rise, the incremental effect became statistically distinguishable only after the fourth exposure. Taiwan’s policy of administering full-dose (100 μg) mRNA-1273 to dialysis patients likely contributed to the strong responses observed with later boosters. Our extended follow-up – through six doses and over 15 months – offered valuable insight into long-term antibody dynamics. Boosters continued to enhance titers, especially with mRNA or bivalent vaccines, but progressive decline was still observed (66–85% at 6 months, >90% by 12–15 months), highlighting the need for ongoing serologic monitoring and timely re-dosing in this vulnerable group.

Our results are consistent with other dialysis cohorts showing significant benefits from repeated boosting.^25,26 In a 66-patient hemodialysis study, a fifth BA.4/5 booster induced robust neutralization against Omicron subvariants, with responses surpassing those after the fourth dose²⁶ Similarly, a 55-patient hemodialysis study reported that the bivalent fifth dose increased anti-spike IgG by over sevenfold in infection-naïve individuals and broadened neutralization breadth, though responses depended strongly on pre-booster antibody levels.²⁵ Extending beyond dialysis populations, a large Japanese cohort (n = 1,763; 7,376 serial measurements) demonstrated that each successive booster, including the fifth, raised antibody baselines and slowed waning, with prior infection further sustaining durability.²⁷ Compared with these cohorts, our study contributes (i) a larger dialysis sample at dose 5 with standardized serial follow-up out to nine months, (ii) concurrent assessment of anti-RBD　binding and neutralizing responses to capture asynchronous kinetics, and (iii) stratification by hybrid-immunity phenotypes (pre- and post-booster infections). Together, these convergent findings across dialysis and general populations reinforce the clinical rationale for a fifth booster in vulnerable hemodialysis patients and clarify that benefits extend beyond short-term antibody peaks to improved functional capacity and durability.

Vaccine-induced immune responses vary by platform, with notable differences observed in both magnitude and durability.²⁸ In our cohort, all vaccine types elicited measurable humoral responses; however, mRNA-based and bivalent vaccines consistently outperformed protein subunit vaccines such as MVC-COV1901 in terms of seroconversion rates and antibody titers. In line with previous studies, these findings support the preferential use of mRNA and bivalent platforms in dialysis patients, where immunogenicity is a critical consideration.²⁹ Nonetheless, a recent retrospective study by Chen et al. reported no significant difference in the risk of clinical composite outcomes between hemodialysis patients receiving mRNA-based versus protein-based booster vaccines.³⁰ This discrepancy underscores the need for future studies that integrate both immunologic and clinical endpoints to more accurately assess the effectiveness of MVC-COV1901 in this population.

Previous studies have also demonstrated that adenoviral vector vaccines elicit weaker immune responses compared to mRNA vaccines in dialysis patients.³¹ Our findings further corroborate this, showing that dialysis patients who received an adenoviral vector vaccine – based primary series exhibited significantly lower seroprotection rates, antibody titers, and neutralizing activity compared to age-matched healthy controls (Supplemental Figures 8–10; Tables 7–9). Importantly, this immunologic disparity was substantially mitigated following administration of a full-dose (100 μg) mRNA-1273 (Moderna) vaccine as the third (additional) dose. These results lend support to Taiwan’s national policy of extending the primary vaccine series for dialysis patients by providing an enhanced mRNA booster.

Multivariate analysis confirmed younger age as a predictor of stronger neutralizing responses, consistent with age-related immunosenescence.^8,32 Immunosenescence is characterized by a progressive decline in both innate and adaptive immune functions, including diminished T- and B-cell activity, reduced formation and persistence of memory lymphocytes, impaired phagocytic function, and increased production of pro-inflammatory cytokines.³³ These age-associated alterations contribute to weaker vaccine-induced immunity and faster waning of protective antibody responses in older individuals.

An intriguing observation was the divergence in waning patterns between neutralizing and anti-RBD binding antibodies. Specifically, we found that while anti-RBD levels steadily declined by 38–59%, neutralizing antibody titers paradoxically increased by approximately 5–52% at 3 months post-vaccination (Supplemental Figure 5), suggesting asynchronous kinetics between anti-RBD binding and neutralizing antibody responses. The magnitude of these divergent changes far exceeded expected assay variability, as the coefficient of variation for the Roche Elecsys assay ranges only from 0.9% to 2.9%. In addition, because most participants were recruited from the same dialysis unit and followed prospectively, the likelihood of systematic sampling bias is minimal. This divergence most likely reflects ongoing affinity maturation, and clonal selection of high-affinity memory B cells, which enhance neutralization potency despite declining total antibody concentrations.^34–36 Our findings are consistent with prior studies showing that neutralizing capacity can be preserved or even augmented over time through the evolution of B-cell responses, even as overall binding titers wane.^37–40 Collectively, these findings highlight the importance of evaluating functional antibody activity – not just binding titers – when assessing long-term immunity, particularly in immunocompromised hosts.

Our study also distinguishes the immunological effects of breakthrough infection. Humoral immune response may differ between natural immunity (obtained by infection) and vaccine-elicited immunity.⁴¹ These immunity are even more different from the combined immunity to SARS-CoV-2 induced by both of natural infection and vaccination has been named hybrid immunity.⁴² In line with previous literature, our study showed that patients with previous SARS-CoV-2 infection mount strong immune responses to COVID-19 vaccines.^43–45 In our study, patients with breakthrough infection exhibited significantly higher and more durable anti-RBD titers than those who were only vaccinated, suggesting that hybrid immunity (infection plus vaccination) may offer superior long-term protection. Notably, those infected after vaccination demonstrated less antibody waning than those vaccinated alone, a finding with direct implications for booster strategies in dialysis patients.

Our study is the first to characterize asymptomatic SARS-CoV-2 infections in dialysis patients. Asymptomatic cases were identified through dynamic antibody profiling – specifically, significant rises in anti-RBD titers unaccompanied by vaccination – suggesting subclinical viral exposure. While baseline anti-RBD titers were similar, asymptomatic individuals showed significantly greater post-infection titers and seroconversion factors (Supplemental Table 6; Figure 7). This suggests immune competence may influence clinical manifestations. The high frequency of undetected infections also implies a greater infection burden than recognized clinically, with implications for infection surveillance and patient counseling.

Our study has several strengths, including prospective design, long-term follow-up (up to 15 months), comprehensive serologic assessments, and a large dialysis cohort. However, some limitations merit consideration. First, the observational design may introduce residual confounding. Second, we focused on humoral immunity; cellular immune responses, which are also critical to COVID-19 protection, were not assessed. Third, small sample sizes for certain vaccine subtypes and the sixth dose may limit generalizability in those subgroups. Fourth, the antibody assays used in this study were originally designed to target the ancestral SARS-CoV-2 strain. While antibody titers measured by these assays consistently reflected immunologic changes following vaccination or breakthrough infection, elevations in antibody levels may not translate directly to clinical protection, particularly against Omicron or later variants. Fifth, the unique vaccine schedule in Taiwan (ChAdOx1 priming and full-dose mRNA-1273 boosting) may limit generalizability to other settings where BNT162b2 or protein-based vaccines were used. Lastly, clinical outcomes such as reinfection rates or hospitalization were not analyzed in relation to antibody titers.

Conclusion

In this longitudinal cohort of dialysis patients, serial COVID-19 vaccinations – particularly with mRNA-based boosters – elicited progressive improvements in humoral immunity, with high seroprotection rates and elevated anti-RBD titers following the third and fourth doses. Breakthrough infections further enhanced immune durability, highlighting the synergistic effect of hybrid immunity. Younger age, receipt of a booster dose, and vaccine platform emerged as independent predictors of neutralizing responses. Notably, the observed dissociation between anti-RBD binding and neutralizing antibody kinetics highlights the need to assess both quantitative and functional immune parameters when evaluating long-term vaccine effectiveness, especially in immunocompromised populations. Asymptomatic breakthrough infections conferred more robust post-infection antibody responses compared to symptomatic infections, suggesting that individuals with greater humoral competence may mount more effective immune control with fewer clinical manifestations. The high frequency of undetected asymptomatic infections also implies that the true burden of infection in dialysis patients may be underestimated, reinforcing the need for ongoing surveillance and immunologic monitoring in this vulnerable population. Compared to age-matched healthy controls, dialysis patients showed attenuated responses after early vaccine doses, but this gap narrowed after subsequent boosters, particularly with optimal vaccine selection and longer inter-dose intervals. Despite these gains, dialysis patients remained immunologically vulnerable, as evidenced by lower peak antibody titers and higher viral loads at the time of infection. Taken together, our findings underscore the importance of individualized vaccination strategies and close serologic monitoring in this high-risk population. Enhancing early protection and maintaining durable immunity through optimized booster timing and vaccine platform selection may help mitigate infection-related risks in patients receiving dialysis.

Biographies

Wan-Chuan Tsai, MD, PhD is a nephrologist and clinical researcher at Far Eastern Memorial Hospital in Taiwan. His academic interests focus on the immunologic responses to vaccination in patients with chronic kidney disease and the optimization of care for dialysis patients. He is committed to improving patient outcomes through translational research and evidence-based clinical practices in nephrology and immunology.

Yu-Sen Peng, MD, PhD, is Vice President of Far Eastern Memorial Hospital in Taiwan and a practicing nephrologist. His research focuses on chronic kidney disease, dialysis, uremia-related complications, and immunologic responses to vaccination. He has led multiple institutional and collaborative studies aimed at improving outcomes and advancing uremia-related care in patients with kidney disease.

Funding Statement

This study was supported by research grants to Dr. Wan-Chuan Tsai from the Far Eastern Memorial Hospital, New Taipei City, Taiwan [FEMH-2022-C-007, FEMH-2023-C-008, FEMH-2024-C-011, and FEMH-2025-C-008]. The funders had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; the preparation, review, and approval of the manuscript; or the decision to submit the manuscript for publication.

Disclosure statement

No potential conflict of interest was reported by the author(s).

CRediT authorship contribution statement

Wan-Chuan Tsai: Conceptualization; Data curation; Formal analysis; Funding acquisition; Methodology; Resources; Software; Validation; Visualization; Writing – original draft; Writing – review & editing.

Fang-Yeh Chu: Conceptualization; Data curation; Methodology; Validation; Writing – review & editing.

Yen-Ling Chiu: Methodology; Resources; Supervision; Writing – review & editing.

Hon-Yen Wu: Data curation; Formal analysis; Resources; Supervision; Writing – review & editing.

Ju-Yeh Yang: Investigation; Resources; Supervision; Writing – review & editing.

Mei-Fen Pai: Data curation; Investigation; Project administration; Resources; Supervision; Validation; Writing – review & editing.

Shih-Ping Hsu: Investigation; Project administration; Resources; Software; Supervision; Validation; Writing – review & editing.

Kuei-Tung Tung: Data curation; Investigation; Resources; Validation; Writing – original draft; Writing – review & editing.

Kai-Hsiang Shu: Investigation; Project administration; Resources; Validation; Writing – review & editing.

Wan-Yu Lin: Formal analysis; Methodology; Software; Validation; Visualization; Writing – review & editing.

Yu-Sen Peng: Conceptualization; Investigation; Methodology; Resources; Supervision; Writing – review & editing.

Data sharing statement

All data are included in the manuscript and/or supporting information.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21645515.2025.2561176