Clinical predictors of SARS-CoV-2 vaccine immunogenicity in kidney transplant recipients at a rural center

Abstract

BACKGROUND

Kidney transplant recipients (KTR) have impaired immune responses to vaccination and remain at increased risk for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection despite mRNA coronavirus disease 2019 (COVID-19) vaccination. Data from rural and medically underserved transplant populations remain limited.

AIM

To evaluate serologic response to a two dose mRNA SARS-CoV-2 vaccination series and identify clinical factors associated with antibody response in KTR.

METHODS

This single center retrospective observational study included adult KTR who completed a two dose mRNA COVID-19 vaccination series and had post vaccination anti spike IgG testing performed using stored serum samples. Multivariable logistic regression was used to identify predictors of seroconversion. Among responders, linear regression was used to evaluate factors associated with quantitative antibody titers.

RESULTS

A total of 108 KTR were included, of whom 63 (58.3%) achieved seroconversion. Higher estimated glomerular filtration rate showed independent association with seroconversion, while higher mycophenolic acid dose showed inverse association with response. Black race and basiliximab induction demonstrated association with seroconversion, though estimates were imprecise. Age, sex, body mass index, vaccine type, tacrolimus levels, and comorbidities were not independently associated with response. Among responders, no clinical or demographic variables were significantly associated with antibody titers.

CONCLUSION

After two dose mRNA vaccination, fewer than two thirds of KTR developed detectable anti spike antibodies. Serologic response was associated with allograft function and antimetabolite exposure, while antibody magnitude was not explained by routine clinical factors in this cohort. These findings provide real world data from a rural transplant population and support consideration of augmented vaccination strategies in KTR.

Key Words: Kidney transplantation; SARS-CoV-2 vaccination; mRNA vaccine; Seroconversion; Immunosuppression; Allograft function

Core Tip: In a rural kidney transplant population, fewer than two thirds of recipients developed detectable antibodies after a standard two dose mRNA severe acute respiratory syndrome coronavirus 2 vaccination series. Serologic response was associated with allograft function and antimetabolite exposure, while routine demographic and clinical factors did not predict antibody magnitude among responders. Because immune assessment was limited to a binding antibody assay and clinical outcomes were not evaluated, these findings describe serologic response rather than clinical protection and should be interpreted accordingly.

INTRODUCTION

Kidney transplant recipients (KTR) require lifelong immunosuppressive therapy[1], which substantially impairs immune responses to vaccination. Early studies during the coronavirus disease 2019 (COVID-19) pandemic demonstrated that solid organ transplant recipients, including KTR, have markedly lower seroconversion rates after standard mRNA severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccination compared with immunocompetent individuals[2-4]. As a result, transplant recipients remain at increased risk for SARS-CoV-2 infection despite completion of primary vaccination series.

Prior studies have implicated older age, diabetes mellitus, and the intensity and composition of immunosuppressive regimens are linked to poor antibody responses[5]. In particular, antimetabolite therapy such as mycophenolic acid (MPA), high dose corticosteroids, and costimulation blockade have been associated with impaired serologic responses[5,6]. Other factors, including time from transplantation and vaccine type, may also influence vaccine response, although findings across studies have been variable[7,8].

While impaired vaccine immunogenicity in KTR is well documented, clinical data from rural and resource constrained transplant settings remain limited. In addition, many such settings serve populations that are underrepresented in prior studies, including a high proportion of Black patients. In this retrospective observational study, we evaluated the serologic response to a two dose mRNA SARS-CoV-2 vaccination series among KTR at a rural tertiary transplant center. The study assessed the proportion of patients achieving seroconversion and examined clinical factors associated with antibody response, providing real world epidemiologic data from a rural transplant population.

MATERIALS AND METHODS

Study design and ethics

This was a single-center, retrospective observational study evaluating SARS-CoV-2 vaccine responses among adult KTR. The study was approved by the Institutional Review Board at East Carolina University, with a waiver of informed consent for the use of de-identified clinical data and stored biospecimens. The study was conducted in accordance with the principles of the Declaration of Helsinki. The authors have read the STROBE Statement checklist of items, and the manuscript was prepared and revised according to the STROBE Statement checklist.

Setting

The study was performed at the East Carolina University Health Transplant Center, the only transplant center serving the largely rural population of eastern North Carolina. The center functions as a major regional tertiary care facility for a medically underserved population.

Study population

The study population included adult KTR in eastern North Carolina who had completed a primary two-dose mRNA COVID-19 vaccination series using either BNT162b2 (Pfizer-BioNTech) or mRNA-1273 (Moderna) during the study period.

Sampling and eligibility criteria

Eligible participants were identified through institutional transplant databases. Inclusion criteria were age 18 years or older, completion of a two-dose mRNA COVID-19 vaccination series, and availability of a stored post-vaccination serum sample for SARS-CoV-2 spike antibody testing. Patients were excluded if post-vaccination antibody testing was unavailable or if vaccination occurred outside the institutional vaccination program. A consecutive sampling strategy was used to maximize capture of eligible patients.

Data collection

Clinical data were extracted from the electronic medical record. Variables collected included age, sex, race and ethnicity, body mass index (BMI), comorbid conditions including diabetes mellitus and hypertension, transplant characteristics including donor type, transplant vintage, kidney-only vs simultaneous kidney-pancreas transplantation, and prior transplant history. Immunosuppression data included induction therapy with thymoglobulin, basiliximab, or alemtuzumab, maintenance immunosuppression with tacrolimus, MPA, and corticosteroids, and vaccine type received.

Antibody testing and outcome definitions

Antibody levels were measured using the LABScreen COVID Plus assay (One Lambda, CA, United States), a microbead-based flow cytometry assay detecting IgG antibodies against the full SARS-CoV-2 spike protein. Seropositivity was defined according to manufacturer recommendations as a mean fluorescence intensity of 7500 or greater. Patients with antibody levels below this threshold were classified as non-responders.

Statistical analysis

Patients were categorized as responders or non-responders based on antibody status. Categorical variables were compared using χ² or Fisher exact tests, and continuous variables were compared using Student t tests or Mann-Whitney U tests, as appropriate. Descriptive statistics were reported as means with standard deviations, medians with interquartile ranges, or proportions. A two-tailed P value less than 0.05 was considered statistically significant. Multivariable logistic regression was used to identify independent predictors of seroconversion, adjusting for clinically relevant covariates including age, race, estimated glomerular filtration rate (eGFR), immunosuppressive regimen, and induction therapy. Covariates were selected a priori based on biological plausibility and prior literature demonstrating associations with vaccine immunogenicity in KTR. Results were reported as odds ratios (OR) with 95% confidence intervals (95%CI).

Among responders, linear regression analysis was performed to evaluate factors associated with antibody titer levels. Predictor variables included demographic factors, transplant characteristics, immunosuppressive therapy, and vaccine type. Beta coefficients with 95%CI were reported. All analyses were conducted using STATA version 14 (StataCorp). This study used standard statistical methods commonly applied in observational clinical research, including descriptive statistics and regression analyses. A formal external biostatistical review was not performed.

RESULTS

Patient characteristics and serologic response

A total of 108 KTR were included in the analysis. Overall, 63 patients (58.3%) achieved seroconversion after a two-dose mRNA vaccination series, while 45 patients (41.7%) were classified as non-responders. Baseline characteristics stratified by serologic response are detailed in Table 1.

Table 1 Comparison of demographic and clinical characteristics between responders and non-responders, n (%).

Variable	Non-responders (n = 45)	Responders (n = 63)	P value
Mean age (years)	60.8	58.9	0.59
Mean BMI	31.5	30.6	0.51
Mean eGFR (mL/minute/1.73 m2)	44.1	54.7	0.01
Male	23 (51.1)	30 (47.6)	1
Caucasian	11 (24.4)	14 (22.2)	0.82
Black	33 (73.3)	48 (76.2)	0.82
Tacrolimus	41 (91.1)	62 (98.4)	0.2
MPA	44 (97.8)	57 (90.5)	0.24
Steroid	43 (95.6)	63 (100.0)	0.17
Basiliximab	5 (11.1)	16 (25.4)	0.09
Thymoglobulin	34 (75.6)	43 (68.3)	0.4
Diabetes mellitus	12 (26.7)	24 (38.1)	0.3
Hypertension	16 (35.6)	23 (36.5)	1
Moderna vaccine	23 (51.1)	36 (57.1)	0.7
Pfizer vaccine	21 (46.7)	26 (41.3)	0.56
Kidney only transplant	42 (93.3)	56 (88.9)	0.52
Days from transplant to 1st vaccine	2200.3	2156.6	0.73

MPA: Mycophenolic acid; BMI: Body mass index; eGFR: Estimated glomerular filtration rate.

Mean age (58.9 years vs 60.8 years; P = 0.59) and mean BMI (30.6 kg/m² vs 31.5 kg/m²; P = 0.51) were similar between responders and non-responders. Responders had a higher mean eGFR compared with non-responders (54.7 mL/minute/1.73 m² vs 44.1 mL/minute/1.73 m², P = 0.01). Sex distribution (47.6% female vs 51.1% male; P = 1.00) were comparable, with most patients identifying as Black in both groups. Maintenance immunosuppression use, including tacrolimus (98.4% vs 91.1%; P = 0.20), MPA (90.5% vs 97.8%; P = 0.24), and corticosteroids (100.0% vs 95.6%; P = 0.17), was similar across groups. Diabetes mellitus, hypertension, vaccine type, transplant type, and time from transplant to the first vaccine dose were also similar between groups. Induction therapy with basiliximab was used more often among responders than non-responders (25.4% vs 11.1%, P = 0.09), while thymoglobulin use did not differ significantly.

Factors associated with seroconversion

Multivariable logistic regression results are presented in Table 2. Higher eGFR rate was associated with increased odds of seroconversion OR = 1.05, 95%CI: 1.02-1.09, P = 0.002. Black race was associated with serologic response compared with other racial groups (OR = 43.08, 95%CI: 1.01-1839.96, P = 0.049). Basiliximab induction was also associated with seroconversion (OR = 31.3, 95%CI: 1.45-677.30, P = 0.028). Higher MPA dose was inversely associated with seroconversion (OR = 0.99, 95%CI: 0.99-1.00, P = 0.01). Thymoglobulin induction (P = 0.066) and time from transplant to vaccination (P = 0.082) did not meet statistical significance. Other covariates, including age, sex, BMI, tacrolimus level, diabetes mellitus, hypertension, vaccine type, transplant type, donor type, prior transplant, and rituximab exposure, were not associated with seroconversion in the model.

Table 2 Multivariate logistic regression analysis of factors associated with antibody response.

Variable	Odds ratio	Standard error	95%CI	P value
Age (per year increase)	0.95	0.028	0.90-1.01	0.101
Gender (male vs female)	0.75	0.44	0.24-2.39	0.631
Race
Caucasian vs other	23.88	47.14	0.50-1143.83	0.108
Black vs other	43.08	82.52	1.01-1839.96	0.049
BMI	0.99	0.049	0.89-1.09	0.767
eGFR	1.05	0.018	1.02-1.09	0.002
Tacrolimus level	0.93	0.14	0.70-1.24	0.637
MPA dose (mg/day)	0.99	0.002	0.99-1.00	0.01
Basiliximab induction (yes vs no)	31.3	49.1	1.45-677.30	0.028
Thymoglobulin Induction (yes vs no)	16.44	25	0.83-323.82	0.066
Days from transplant to vaccine	1	0.0003	1.00-1.00	0.082
Days from 2nd vaccine to sample testing	1.01	0.0035	1.00-1.01	0.055
Diabetes mellitus	1.39	0.8	0.47-4.13	0.564
Hypertension	1.71	0.94	0.61-4.79	0.307
Vaccine type
Moderna vs other	0.78	0.46	0.28-2.18	0.639
Pfizer vs other	0.57	0.34	0.20-1.66	0.307
SPK or kidney alone	1.03	0.62	0.31-3.42	0.956
History of rituximab use	0.99	0.8	0.20-4.93	0.993
Type of donor (living vs deceased)	1.62	0.92	0.55-4.78	0.38
Previous transplant (yes vs no)	0.82	0.58	0.24-2.75	0.747
Primary ESRD etiology	1.41	0.36	0.90-2.19	0.135
Constant	0.01	0.02	0.00-60.94	0.263

Reference categories for regression models were female sex, other race, absence of induction therapy, absence of calcineurin inhibitor, absence of antimetabolite therapy, absence of steroid therapy, receipt of non-mRNA vaccine, deceased donor transplantation, and absence of the listed comorbidities, unless otherwise specified. BMI: Body mass index; eGFR: Estimated glomerular filtration rate; MPA: Mycophenolic acid; SPK: Simultaneous pancreas kidney; ESRD: End stage renal disease; 95%CI: 95% confidence interval.

Adjustment for time from second vaccine dose to antibody testing did not materially alter associations with seroconversion. The median testing interval was 139.5 days (interquartile range, 72 days to 199.5 days).

Factors associated with antibody titers among responders

Among the 63 responders, linear regression analysis evaluating predictors of quantitative antibody titers (Table 3) identified no statistically significant associations. The model explained a limited proportion of variability in antibody titers (R² = 0.16).

Table 3 Linear regression analysis of factors associated with log-transformed antibody titer among responders.

Variable	Coefficient	Standard error	95%CI	P value
Age (per year increase)	-22.6	47.5	-118.2 to 73.1	0.64
Gender (male vs female)	-1153.4	1138.2	-3444.4 to 1137.6	0.32
Race	-148.9	1488.7	-3145.5 to 2847.6	0.92
Days from transplant to first vaccine	-0.2	0.5	-1.2 to 0.7	0.62
Days from 2nd vaccine to sample testing	2.1	7.8	-13.6 to 17.9	0.79
Tacrolimus level	-42.5	303.5	-653.5 to 568.5	0.89
MPA dose (mg/day)	1.9	2.9	-3.9 to 7.8	0.51
Diabetes mellitus	-1957.9	1701	-5381.8 to 1466.0	0.26
Hypertension	-98.3	1568.2	-3254.8 to 3058.2	0.95
Vaccine type	-232.3	1011.9	-2269.2 to 1804.6	0.82
SPK or kidney alone	-1152.1	2005	-5188.1 to 2883.8	0.57
History of rituximab use	950.3	1837.7	-2748.7 to 4649.4	0.61
Type of donor (living vs deceased)	-547.1	1798.1	-4166.5 to 3072.2	0.76
Previous transplant (yes vs no)	-2029.9	1537	-5123.7 to 1063.8	0.19
Primary ESRD etiology	-202.5	332.3	-871.3 to 466.3	0.55
Constant	23867.3	5273.3	13252.8 to 34481.8	0

Reference categories for regression models were female sex, other race, absence of induction therapy, absence of calcineurin inhibitor, absence of antimetabolite therapy, absence of steroid therapy, receipt of non-mRNA vaccine, deceased donor transplantation, and absence of the listed comorbidities, unless otherwise specified. MPA: Mycophenolic acid; SPK: Simultaneous pancreas kidney; ESRD: End stage renal disease; 95%CI: 95% confidence interval.

DISCUSSION

In this single center cohort of KTR from eastern North Carolina, 58.3% developed detectable anti SARS-CoV-2 spike antibodies after completion of a two-dose mRNA COVID 19 vaccination series. This finding is consistent with prior reports demonstrating substantially reduced humoral immunogenicity in transplant recipients, with published seroconversion rates generally ranging from approximately 30% to 50% and pooled estimates around 40% across large observational cohorts and meta-analyses[2,3,9]. Although the seroconversion rate observed in this study was numerically higher than that reported in some series[10], it remains markedly lower than the near universal antibody response observed in immunocompetent populations. Differences in cohort composition, immunosuppressive burden, timing of antibody assessment, and assay characteristics may contribute to variability across studies. Taken together, these data reinforce the prevailing evidence that a standard two dose mRNA vaccination strategy frequently fails to elicit adequate humoral immunity in KTR.

In multivariable analysis, Black race was associated with higher odds of seroconversion following two-dose mRNA vaccination groups (OR = 43.08, 95%CI: 1.01-1839.96, P = 0.049). Racial differences in vaccine-induced antibody responses among KTR have not been consistently demonstrated in prior studies, which have largely emphasized clinical determinants such as immunosuppressive intensity and graft function[5]. The magnitude of the observed association in this cohort was accompanied by wide confidence intervals, indicating statistical instability, likely secondary to sparse data bias or small non-Black comparator groups within the model. In addition, baseline pre-vaccination serostatus was not available, and differential prior exposure to SARS-CoV-2 may have contributed to observed post-vaccination seropositivity. In a rural, medically underserved region, higher community exposure and asymptomatic infection could result in unrecognized hybrid immunity, potentially influencing antibody detection after vaccination[11]. Taken together, this association should be interpreted as hypothesis generating and warrants further evaluation in larger, well-characterized transplant cohorts with baseline serologic assessment.

In this cohort, age, sex, and BMI were not independently associated with serologic response following two dose mRNA vaccines. Vaccine type (BNT162b2 vs mRNA-1273) also did not significantly influence serostatus, and no associations were observed with common comorbid conditions, including diabetes or hypertension. Although prior studies have reported lower vaccine responsiveness with increasing age or modest differences between mRNA vaccine platforms, these effects were not evident in the present analysis[5,12]. In this relatively modest sample size, it is plausible that the influence of these demographic and clinical factors was secondary to the more dominant effects of the maintenance and induction immunosuppressive regimens.

Prior studies in KTR have consistently demonstrated reduced vaccine induced humoral responses in the setting of more intensive immunosuppressive regimens, particularly those involving T cell depleting agents or costimulation blockade[9,13,14]. Immunosuppressive regimen characteristics showed the most consistent associations with serologic response in this cohort. In the multivariable model, basiliximab induction was associated with higher odds of seroconversion. However, this finding should be interpreted cautiously given the imprecision of the estimate, wide confidence intervals and the likelihood of confounding by indication and selection bias. Induction choice is nonrandom and often reflects baseline immunologic risk, era effects, and center specific practice patterns, which may also correlate with maintenance immunosuppression intensity and overall clinical stability. Consistent with this concern, basiliximab use differed only modestly between responders and non-responders in unadjusted comparisons. Taken together, the observed association is best viewed as hypothesis generating and may reflect characteristics of patients selected for basiliximab rather than a direct effect of induction strategy on vaccine responsiveness.

Higher MPA dose (modeled per mg/day) was inversely associated with seroconversion in the multivariable model, although the per unit effect size was small and interpretation depends on the dose scale used. The impact of antimetabolite therapy on dampening vaccine immunogenicity is well documented, as these agents fundamentally impair B-cell proliferation and humoral recruitment[10]. Although the odds ratio per unit increase is numerically near unity, the significance of the P value reinforces the finding that the overall burden of antimetabolite exposure remains a primary barrier to achieving seropositivity in transplant recipients. The marginal nature of this association, with a confidence interval touching 1, may be attributed to the high prevalence of MPA use across the entire cohort (93.5%), which limits the power to detect clear dose-dependent differences. These data complement ongoing research into temporary antimetabolite modulation as a potential strategy to enhance vaccine responses, though such approaches require careful individualization based on allograft stability and rejection risk.

Higher allograft function, assessed by eGFR, was independently associated with seroconversion following two dose mRNA vaccination (OR = 1.05, P = 0.002) in the study. Preserved graft function likely reflects greater clinical and immunologic stability, whereas reduced eGFR is associated with higher comorbidity burden, uremia related immune dysfunction, and diminished vaccine induced antibody production. Although higher eGFR does not indicate normal immune competence, KTRs with better allograft function appear more likely to mount humoral responses than those with advanced graft dysfunction. This finding is consistent with prior studies[14-16] and suggests that eGFR is a clinically useful marker for identifying patients with a comparatively higher probability of seropositivity. These associations in this study should be interpreted cautiously given wide confidence intervals and the potential for model instability in a modest-sized cohort. These results reinforce the need for individualized vaccination strategies in KTRs.

Among recipients who achieved seroconversion, linear regression analysis did not identify significant clinical or demographic predictors of quantitative anti spike IgG titers. The low model R² (about 16%) indicates that routinely measured clinical variables, including immunosuppressive drug exposure and metabolic factors, explain only a small proportion of the variability in antibody magnitude. These findings suggest that while standard clinical parameters may help distinguish responders from non-responders, they are poor predictors of the depth of the humoral response once seropositivity is achieved. This observation is consistent with prior reports showing that many seropositive KTRs generate relatively low and heterogeneous antibody levels compared with immunocompetent populations[17,18]. The substantial unexplained variance in our model likely reflects the influence of intrinsic immune factors not captured in routine clinical data, such as B cell memory formation, T cell help, or preexisting immune priming. The absence of associations between antibody magnitude and immunosuppressive exposure, including tacrolimus levels or MPA dose, further supports the concept that once an initial immune threshold is crossed, quantitative antibody production may be governed by individual or stochastic biological processes rather than readily measurable treatment variables. These data emphasize that a binary “seropositive” result may mask a wide spectrum of quantitative responses, reinforcing the need for more sophisticated immunologic monitoring, such as neutralizing capacity or cellular assays, to accurately assess protective immunity in this vulnerable population.

Our findings have important clinical implications for the management of immunization in KTRs. The suboptimal seroconversion rate observed after a standard two dose mRNA vaccine series in our cohort (58.3%) reinforces the clinical limitations of conventional vaccination strategies in KTRs. In response to suboptimal vaccine responses, transplant and public health communities have recommended additional mRNA vaccine doses for immunocompromised patients. Early studies demonstrated that a substantial proportion of initial nonresponders achieved seroconversion following a third vaccine dose[19]. These findings supported adoption of a three dose primary mRNA vaccine series for transplant recipients and other immunosuppressed populations. Despite augmented vaccination, a clinically relevant proportion of KTRs remain seronegative or exhibit suboptimal antibody responses[20,21]. In this context, adjunctive strategies such as modification of immunosuppressive regimens have been explored. Our finding of a significant inverse association between MPA dose and seroconversion aligns with existing clinical trial and observational data examining the impact of antimetabolite exposure on vaccine immunogenicity. While randomized studies evaluating alternative booster platforms have demonstrated mixed results, antimetabolite modulation has been associated with numerically higher response rates in selected settings without a clear increase in short term rejection risk[9,17]. Within the broader framework of augmented vaccination strategies, our findings support ongoing efforts to refine immunization approaches for transplant recipients based on biologically relevant and clinically observable risk factors.

This study has several limitations that warrant consideration. As a single center analysis, the findings reflect local clinical practice and modest sample size limited statistical power to detect smaller effect sizes. The retrospective and observational design introduces the potential for residual confounding. In addition, antibody measurements were obtained from residual clinical samples collected at variable time points following vaccination, limiting the ability to characterize peak antibody responses or immune kinetics. Pre-vaccination SARS-CoV-2 serostatus was unavailable, and unrecognized prior infection may have influenced post-vaccination antibody responses. Seroconversion was defined using a single binding antibody assay measuring anti-spike IgG levels, without assessment of neutralizing antibody activity or cellular immune responses and therefore does not directly equate to clinical protection. The modest sample size limited statistical power to detect smaller effect sizes. Given the modest number of seroconversion events, the multivariable model may be susceptible to overfitting and sparse data bias, and estimates should be interpreted cautiously. High prevalence of MPA based maintenance immunosuppression in the cohort constrained comparative analyses across different immunosuppressive regimens. Long term clinical outcomes, including breakthrough SARS-CoV-2 infection and allograft rejection, were not captured. Although impaired vaccine immunogenicity in KTR is well established, the purpose of this study was not to define novel immunologic mechanisms, but rather to provide a hospital based epidemiologic description of real-world vaccine response patterns in an underrepresented rural transplant population. These findings contribute to ongoing infection prevention discussions by identifying patient level factors that may inform vaccination strategies, immunosuppression management, and post vaccination surveillance in similar healthcare settings.

CONCLUSION

In this single center study of KTR from a rural transplant program, fewer than two thirds of patients developed detectable anti SARS-CoV-2 spike antibodies following a standard two dose mRNA vaccination series. Serologic response was independently associated with allograft function and antimetabolite exposure, while demographic and most clinical variables were not predictive. Among those who seroconverted, the magnitude of the antibody response was not explained by routinely measured clinical factors, underscoring the limitations of binary serologic assessment and standard clinical predictors in defining vaccine induced immunity in this population. Collectively, these findings reinforce the need for augmented vaccination strategies and highlight clinically observable factors that may inform risk stratification and immunization approaches in KTR, particularly in rural and resource constrained settings.

ACKNOWLEDGEMENTS

The authors thank the clinical and laboratory staff of the East Carolina University Health Transplant Center for their support in specimen handling and data retrieval. The authors also acknowledge the transplant recipients whose clinical data contributed to this study.