Humoral and cellular immune response after COVID-19 vaccination in patients with sickle cell disease on hydroxyurea

Key Points

• •Patients with SCD who received a 2-dose COVID-19 mRNA vaccine showed normal spike-specific antibody and memory B-cell concentrations.
• •Spike-specific T-cell frequencies were normal, but their capacity to produce type 1 cytokines was reduced in these patients.

Visual Abstract

Abstract

Patients with sickle cell disease (SCD) are at increased risk of COVID-19–related mortality compared with healthy individuals, even after vaccination. To what extent impaired vaccine-induced immunity contributes to this risk is unknown. We prospectively investigated vaccine immunogenicity in 31 patients with SCD who received COVID-19 mRNA vaccines. All patients used hydroxyurea. We quantified humoral and cellular immune responses 4 weeks after the second and third vaccinations and compared the results with those of age-, sex-, and vaccine-matched healthy individuals. Irrespective of higher naïve and lower memory B-cell subsets, serum neutralizing spike glycoprotein 1 immunoglobulin G antibody concentrations and frequencies of spike-specific memory B cells were similar to those in healthy individuals at each time point. Frequencies of CD4⁺ and CD8⁺ T cells in patients with SCD were also comparable with controls; however, type 1 cytokine production by spike-specific T cells was reduced. Reduced cytokine production and lower antibody production correlated with higher serum fetal hemoglobin levels, suggesting an association with hydroxyurea use. Although a third vaccination improved neutralizing antibody and memory B-cell responses, T helper 1 cytokine production tended to remain lower in patients than in controls. Our data point toward delayed or reduced vaccine-induced immunity, in line with previous reports, which may contribute to the increased risk of COVID-19–related mortality reported in these patients. This trial was registered at www.ccmo.nl as NL-OMON51241.

Introduction

Respiratory infections in patients with sickle cell disease (SCD) are often associated with a complicated course of disease, typically resulting in high infection-related mortality.¹ This was also observed with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection.2, 3, 4, 5, 6 Although vaccination protected healthy individuals from COVID-19–related death, the higher risk of COVID-19–related mortality in patients with SCD was shown to persist regardless of vaccination.^7,8

Viral infections in patients with SCD are often complicated by vaso-occlusive crisis and may lead to acute chest syndrome. Comorbidities associated with SCD, such as dilated cardiomyopathy, nephropathy, asplenia, and pulmonary hypertension, further increase the risk of severe COVID-19 and COVID-19–related mortality.9, 10, 11

A less explored factor that could contribute to severe COVID-19 in patients with SCD may be impaired innate or adaptive immunity associated with SCD, resulting in reduced immunogenicity of COVID-19 vaccines in these patients. Impaired immunity may be caused by functional asplenia or by ischemia–reperfusion and continuous hemolysis, which lead to oxidative stress and chronic inflammation.^12,13 The use of hydroxyurea enhances fetal hemoglobin production, suppresses hemolysis, prevents vaso-occlusive crises, and has been associated with reduced chronic inflammation.14, 15, 16 However, it has also been suggested that hydroxyurea may compromise immunity through effects on lymphocytes and other immune cells.^12,13 Delayed and reduced vaccine immunogenicity to yellow fever and measles vaccines has been reported in patients with SCD on hydroxyurea.^17,18 Although COVID-19 mRNA vaccination induced antibody concentrations similar to those in healthy individuals,19, 20, 21, 22 much less is known about cellular immune responses to COVID-19 vaccines in patients with SCD. To investigate the hypothesis that vaccination-induced humoral or cellular immunity may be impaired in patients with SCD, thereby contributing to the increased risk of severe COVID-19, we performed a prospective, observational study to investigate in detail humoral and cellular immunity after 2- and 3-dose COVID-19 mRNA vaccination in patients with SCD on hydroxyurea and in age-matched healthy individuals.

Methods

Study participants

This study is part of a larger, prospective, multicenter, observational cohort study on COVID-19 vaccine immunogenicity in patients with hematologic conditions (COBRA KAI study²⁰). Thirty-one patients with SCD on hydroxyurea were included. Participants received 2 mRNA-1273 vaccinations (Moderna) and 1 mRNA-BNT162b2 booster vaccination (Pfizer) between March 2021 and September 2022, according to the guidelines of the Dutch National Institute for Public Health and the Environment. Study protocols were approved by the institutional review board of the Amsterdam University Medical Center, Amsterdam, The Netherlands. Serum and peripheral blood mononuclear cells (PBMC) were collected before and 28 days after each vaccination and additionally at 6 and 16 months after the first vaccination (Figure 1A). Control cohorts consisted of healthy individuals, matched for age, sex, vaccine type, and number of vaccine doses, extracted from 2 sources: (1) the VITAL project (vaccine and infectious diseases in the aging population), a public-private consortium investigating infectious diseases and the effects of vaccinations, coordinated by the Dutch National Institute for Public Health and the Environment²³; and (2) laboratory personnel from Leiden University Medical Center and Erasmus University Medical Center^24,25 (Figure 1A). They are collectively referred to as healthy controls. All participants provided informed consent before study onset or before sample collection. Participants who tested positive for COVID-19 before vaccination were excluded from analyses.

Figure 1.

Timeline and selection of study participants. (A) Timeline of vaccination schedule, blood collection, and measurement of vaccination response. (B) Selection of study participants. A minimum of 5 × 10⁶ cells per PBMC vial was used as the only criterion for random selection of patients for analyses of spike-specific B and T cells.

SARS-CoV-2–specific antibody concentrations and neutralizing capacity

Serum immunoglobulin G (IgG) antibody concentrations against the spike glycoprotein 1 (S1), nucleocapsid (N), and receptor-binding domain (RBD) antigens of SARS-CoV-2 were quantified and calibrated against the international reference antibody standard for COVID-19 as recommended by the World Health Organization (WHO/BS/2020.2402), in binding antibody units (BAU) per milliliter. Seroconversion was defined as an S1 IgG antibody concentration ≥10 BAU/mL.^20,24 Antibody neutralization capacity was tested using lentiviral-based pseudoviruses expressing the spike protein of wild-type SARS-CoV-2 (D614G) and Omicron BA.1, as described previously.^26,27

SARS-CoV-2 spike- and RBD-specific B cells

To identify SARS-CoV-2 spike- and RBD-specific B cells after both vaccinations, a minimum of 5 × 10⁶ PBMC per time point were stained with a viability dye and antibodies directed to CD3, CD14, CD16, CD19, CD20, CD27, CD38, IgD, IgM, and IgG (supplemental Table 1), combined with fluorescently labeled SARS-CoV-2 spike proteins (BV421 and BB515) and RBD protein (AF647), as described previously (supplemental Tables 2A and 3).^28,29 Living B cells were identified as viability stain⁻CD19⁺CD3⁻CD14⁻CD16⁻ cells, of which differentiation was determined by IgD and CD27, clustering them into naïve (IgD⁺CD27⁻), nonswitched memory (IgD⁺CD27⁺), classical (switched) memory (IgD⁻CD27⁺), and atypical (IgD⁻CD27⁻) B cells. We used 2 fluorescently labeled SARS-CoV-2 spike proteins (BV421 and BB515) to identify SARS-CoV-2 spike-specific B cells and to exclude nonspecific binding. A pre–COVID-19 pandemic donor sample was used as negative control, and a convalescent or COVID-19–vaccinated donor sample was used as positive control. Flow cytometry analyses were performed on the LSRFortessa (BD Biosciences), and data were analyzed using FlowJo software (version 10; TreeStar) and GraphPad Prism 9.0.1.

SARS-CoV-2 spike-specific T cells

T-cell assays were performed as described earlier.^30,31 In brief, up to 2 × 10⁶ PBMC per patient were incubated with an overlapping pool of 15-mer spike peptides (SB Peptide). As SARS-CoV-2 T-cell epitopes have been largely conserved across variants of concern,32, 33, 34 this pool can be considered representative during the whole study. Dimethyl sulfoxide was used as a negative control, and a pool consisting of cytomegalovirus, Epstein-Barr virus, influenza, and extra (CEFX) peptides was used as a positive control (supplemental Table 2A-B). Cells were incubated for 16 hours in the presence of brefeldin A, stained with a viability dye, fixed, permeabilized, and incubated with antibodies directed against CD3, CD4, CD8, CD154, CD137, CD69, interferon gamma (IFN-γ), tumor necrosis factor α (TNF-α), interleukin-2 (IL-2), IL-4, IL-17, PD-1, FOXP3, and CXCR5 (supplemental Tables 1 and 3). Flow cytometry analysis was performed on a 3-laser Aurora (Cytek Biosciences) and analyzed using GraphPad Prism 9.0.1. Spike-specific T-cell frequencies were based on CD137⁺/CD154⁺ cell percentages for CD4⁺ T cells and on CD137⁺CD69⁺ cell percentages for CD8⁺ T cells, both corrected for dimethyl sulfoxide background. Only samples with >5000 events in the CD4⁺ and CD8⁺ gates were analyzed. Positive response threshold for spike-specific T cells was set at 0.05% for CD4⁺ T cells and 0.025% for CD8⁺ T cells. Threshold frequencies were determined in an independent cohort of healthy individuals.³¹ Analysis of cytokine-producing spike-specific T cells (IFN-γ, TNF-α, IL-2, IL-4, and IL-17) and circulating spike-specific T follicular helper (cTfh) cells (CD4⁺CXCR5⁺PD1⁺) was only performed if the response positivity threshold was met, and >25 events were measured in the CD154⁺/CD137⁺ and CD137⁺CD69⁺ gates for CD4⁺ and CD8⁺ T cells, respectively. All time points were measured simultaneously per patient to minimize technical variance within each patient. High-dimensional reduction analysis was conducted using OMIQ (www.omiq.ai), in which an optimized uniform manifold approximation and projection was generated based on CD4, CD8, CD154, CD137, CD69, IFN-γ, TNF-α, and IL-2 expression.

Statistical analysis

Differences between time points were calculated using paired sample t tests for normally distributed data and Wilcoxon signed-rank tests for nonnormally distributed data. Differences between patients and healthy individuals were analyzed using Mann-Whitney U tests and independent samples t tests after log10 transformation, respectively. Given the exploratory nature of this study and the predicted small sample size, no correction for multiple comparisons was done. Pearson correlation was calculated between serum S1 IgG antibody concentrations and in vitro virus neutralization after log10 transformation of both. Spearman correlations were calculated pairwise between clinical parameters and humoral or cellular immune response parameters. Two-sided P values <.05 were considered statistically significant. Analyses were performed using GraphPad Prism 9.0.1 and using R version 4.3.2 with RStudio version 1.3.1093 (Posit Software).

Results

Patient characteristics

A total of 31 patients with SCD on hydroxyurea and 29 healthy individuals were included in the study (Figure 1B; Table 1). The median age of patients was 35 years (interquartile range, 27-47), and 45% of patients were female. Ethnicity was diverse, with the largest number of patients with SCD from African descent (32.3%). Most patients had homozygous SCD (81%). All patients received hydroxyurea therapy, with an average daily dose of 1000 mg (range, 500-1500), and 1 patient used both hydroxyurea and voxelotor. Immunological parameters, including circulating B-, T-, and natural killer cell numbers, were within normal ranges at baseline (Table 2). Three patients and 1 laboratory employee were excluded from analysis because of SARS-CoV-2 infection before the second vaccination. Infection was confirmed by N IgG ≥14.3 BAU/mL on the day of the second vaccination. One patient had a local reaction to the vaccination, which was followed shortly thereafter by a vaso-occlusive crisis; the patient therefore refrained from a third vaccination. A total of 25 patients completed the 2-dose vaccination schedule and follow-up sampling, and 12 patients completed follow-up after the third vaccination (Figure 1B). Details of participants who were excluded or lost to follow-up are depicted in supplemental Table 4.

Table 1.

Baseline characteristics

	Patients	Control cohorts for analyses on
	SCD	Humoral immunity	Cellular immunity
		Healthy individuals	Laboratory personnel
No. of patients	31	19	10
Age, median (IQR), y	35 (27-47)	41 (36-46)	45 (31-46)
Sex (female), n (%)	14 (45)	8 (42)	7 (54)
Ethnicity, n (%)
Mixed ethnicity∗	4 (12.9)	0 (0.0)	0 (0.0)
Middle Eastern	1 (3.2)	0 (0.0)	0 (0.0)
African	10 (32.3)	0 (0.0)	0 (0.0)
Latin American	3 (9.7)	0 (0.0)	0 (0.0)
Mixed, Surinamese	8 (25.8)	0 (0.0)	0 (0.0)
White	0 (0.0)	0 (0.0)	5 (50)
Caribbean	4 (12.9)	0 (0.0)	0 (0.0)
Unknown	0 (0.0)	19 (100)	5 (50)
Hemoglobinopathy, n (%)
HbSS	25 (80.6)	n.a.	n.a.
HbSβ thalassemia	4 (12.9)	n.a.	n.a.
HbSC	2 (6.5)	n.a.	n.a.
SCD-related hospitalization, n (%)
>1 hospital admission in the last year	6 (19.4)	n.a.	n.a.
History of acute chest syndrome	14 (45.2)	n.a.	n.a.

Table shows baseline characteristics of patients with SCD, healthy individuals selected for humoral immunity (VITAL study²³), and laboratory personnel selected for cellular immunity (EMC cohort^24,25) experiments.

EMC, Erasmus University Medical Center; HbSC, compound heterozygous SCD; HbSS, homozygous SCD; IQR, interquartile range; n.a., not applicable; NK, natural killer.

Table 2.

Baseline laboratory measurements

	Patients with SCD	Normal range
HbF, mean (SD), %	11.4 (7.4)	<1.0
Total bilirubin, median (IQR), μmol/mL	27.0 (18.5-41.0)	3.0-19.0
LDH, median (IQR), U/L	417 (325-534)	<251
CRP, median (IQR), mg/L	7.0 (3.7-14.2)	0.0-5.0
Immune cells, median (IQR), ×109/L
Reticulocytes	159 (116-252)	35-105
Leukocytes	6.9 (4.8-8.7)	4.0-10.5
Neutrophils	3.5 (2.4-4.6)	1.8-7. 2
Lymphocytes	2.0 (1.8-2.9)	1.5-4.0
B cells	0.42 (0.25-0.58)	0.10-0.5
CD4+ T cells	0.94 (0.60-1.190)	0.30-1.40
CD8+ T cells	0.42 (0.27-0.59)	0.20-0.90
NK cells	0.37 (0.24-0.66)	0.09-0.60

CRP, C-reactive protein; LDH, lactate dehydrogenase; SD, standard deviation.

Normal neutralizing antibody concentrations in patients with SCD

Serum S1 IgG antibody concentrations increased significantly after each mRNA-1273 vaccination (Figure 2A). After 2 vaccinations, S1 IgG concentrations in patients with SCD were similar to those in healthy individuals (P = .41). S1 IgG antibody concentrations had significantly declined in patients with SCD at 6 and 9 months after the second vaccination (Figure 2A). A total of 12 patients with SCD and all healthy individuals received a BNT162b2 booster vaccination at 9 and 8 months after the second mRNA-1273 vaccination, respectively. This led to a significant increase in S1 IgG antibody concentrations in all participants. Patients with SCD reached significantly higher S1 IgG antibody concentrations than 3-dose–vaccinated age-matched healthy individuals (P = .001; Figure 2A). We assessed the neutralization capacity of S1 IgG antibodies after the second and third vaccinations in a random selection of 17 patients. S1 IgG antibody concentrations correlated significantly with antibody neutralization capacity for wild-type SARS-CoV-2 (r = 0.71; P = .003) and the Omicron BA.1 variant of concern (r = 0.59; P = .002) after 2 vaccinations, and this correlation persisted after the third vaccination (wild-type, r = 0.85 [P = .001]; Omicron BA.1, r = 0.69 [P = .002]; Figure 2B; data not shown). The neutralizing capacity per antibody, a measure of antibody maturation, increased after the third vaccination for the Omicron BA.1 variant and was comparable with the neutralizing capacity per antibody for Omicron BA.4/5 (Figure 2C).

Figure 2.

Normal neutralizing S1 IgG antibody concentrations in patients with SCD. (A) S1 IgG concentration at indicated time points in patients with SCD (purple) and age-matched healthy controls (HC; gray). The dotted line indicates the threshold for seroconversion (10 BAU/mL). Medians and interquartile ranges are indicated. (B) Pearson correlation between S1 IgG concentration and S1 IgG neutralization, defined as the serum dilution that inhibits 50% of SARS-CoV-2 Omicron BA.1 infectivity (ID50), shown for second (purple dots) and third vaccinations (purple triangles) in patients with SCD. (C) S1 IgG neutralization capacity after second (purple dots) and third vaccinations (purple triangles) for wild-type SARS-CoV-2, Omicron BA.1, and Omicron BA.4/5 in patients with SCD. Asterisk indicates significance level, and an α significance level of 0.05 is used; pre refers to the time before first vaccination; first, 28 days after first vaccination and at the time of second vaccination; second, 28 days after second vaccination; third, 28 days after third vaccination. BAU, binding antibody units; HC, healthy controls; IC50, inhibitory concentration 50; ID50, Infectious Dose 50; S1; spike 1; SCD, sickle cell disease.

Normal spike-specific memory B-cell responses in patients with SCD

Seven patients from the SCD cohort were selected for B-cell analyses based on availability of frozen PBMC vials collected after the second and third vaccinations. See Figure 3A for a representative example of the gating strategy. Of the total B-cell compartment, significantly higher total percentages of naïve B cells (median, 86.3% vs 65.5%; P = .0001) and lower percentages of unswitched memory (2.9% vs 6.7%; P = .02), classical memory (6.4% vs 17.8%; P = .001), and atypical memory B cells (3.6% vs 12.1%; P = .002) were observed in patients with SCD compared with controls (Figure 3B). We identified SARS-CoV-2–specific memory B cells using fluorescently labeled SARS-CoV-2 spike and RBD proteins. Two vaccinations induced comparable percentages of spike- and RBD-specific memory B cells in patients with SCD compared with age-matched healthy controls, with a median of 0.46% vs 0.21% (P = .06) for spike-specific B cells and a median of 0.16% vs 0.10% (P = .42) for RBD-specific B cells, respectively (Figure 3C; supplemental Figure 1). Because the memory B-cell population in patients with SCD was significantly smaller than healthy controls, the percentage of spike-specific B cells was corrected for the number of live B cells, and no differences were observed between patients with SCD and healthy controls (data not shown). Within the spike-specific memory B-cell pool, comparable percentages of unswitched (median, 6.67% vs 15.0%; P > .05), classical (median, 64.4% vs 53.6%; P > .05), and atypical memory B cells (median, 28.9% vs 23.8%; P > .05) were observed between patients with SCD and healthy controls (Figure 3D; data not shown). Furthermore, percentages of spike-specific IgM⁺ and IgG⁺ memory B cells were not different between the 2 groups (Figure 3E). A third vaccination led to an increase in the frequency of spike-specific memory B cells in patients with SCD (P = .03). The percentage of spike-specific memory B cells did not correlate with S1 IgG antibody concentrations at any time points (data not shown).

Figure 3.

Frequencies of SARS-CoV-2 spike-specific memory B cells in patients with SCD. (A) Representative example of flow cytometry gating strategy in a patient after second vaccination to detect SARS-CoV-2 spike- and RBD-specific B cells. All events were gated on lymphocytes, single cells, and CD19⁺CD3⁻CD14⁻CD16⁻ living B cells. SARS-CoV-2 spike double-positive and RBD-positive B cells were gated from memory B-cell subsets (combining unswitched memory [IgD⁺CD27⁺], classical memory [IgD⁻CD27⁺], and atypical memory [IgD⁻CD27⁻]). (B) Naïve, unswitched memory, classical memory, and atypical memory B cells as the percentage of the overall B-cell pool in patients with SCD and HC, collected after the second COVID-19 mRNA vaccination. (C) Spike-specific memory B cells as the percentage of memory B cells in patients with SCD after the second and third vaccinations compared with 2-dose–vaccinated HC. (D) Classical memory B cells as the percentage of the spike-specific memory B-cell pool in patients with SCD and HC. (E) IgM and IgG isotype of spike-specific memory B cells in patients with SCD and HC. Asterisk indicates significance level for the comparison between groups and time points. An α significance level of 0.05 is used. Second refers to 4 weeks after the second vaccination; third, 4 weeks after the third vaccination. DUMP, dump channel containing CD3, CD14, CD16; FSC-A, forward scatter area; Q1, naive B cells; Q2, unswitched memory B cells; Q3, switched memory B cells; Q4, atypical memory B cells.

Normal spike-specific T-cell frequencies in patients with SCD

A total of 15 patients with SCD, representative of the cohort (supplemental Table 5), were selected for T-cell analyses. Before vaccination, subsets of naïve and memory CD4⁺ T cells were similar to healthy controls, whereas higher percentage of naïve and lower percentages of memory and effector-memory CD8⁺ T cells were observed in patients with SCD than in healthy controls (supplemental Figure 2). Spike-specific CD4⁺ T cells were detected in all patients and healthy controls after the first vaccination (Figure 4A-B). The frequency increased after the second vaccination and remained stable during the 6-month follow-up. A third vaccination did not further increase the percentages of spike-specific CD4⁺ T cells. Percentages of spike-specific CD4⁺ T cells were not significantly different from healthy controls at any point in time (Figure 4B). After 2 vaccinations, spike-specific CD8⁺ T cells were detected in 11 of 15 patients with SCD (73.3%), which was similar to those in the controls (6/8 [75.0%]; P = .62; Figure 4C). Control CEFX-specific CD4⁺ and CD8⁺ T-cell frequencies were constant over time and also comparable with those in healthy controls (Figure 4D-E).

Figure 4.

Frequencies of spike-specific CD4⁺ and CD8⁺ T cells after COVID-19 mRNA vaccination in patients with SCD and HC. (A) Representative example of flow cytometry gating strategy for activation induced marker expression (AIM) CD4⁺ and CD8⁺ T cells. All events were gated on lymphocytes, single cells, viable cells, CD3⁺, and subsequently either CD4⁺ or CD8⁺. (B) Spike-specific CD4⁺ T cells of total CD4⁺ T-cell pool over time in patients with SCD, compared with HC. (C) Spike-specific CD8⁺ T cells of total CD8⁺ T-cell pool over time in patients with SCD, compared with HC. (D) CEFX-specific CD4⁺ T cells over time in patients with SCD, compared with HC. (E) CEFX-specific CD8⁺ T cells over time in patients with SCD, compared to HC. The dotted line indicates the threshold for a positive T-cell response. Comparisons between patients and HC are indicated with an asterisk. The asterisk indicates significance level, and an α significance level of 0.05 is used. DMSO, dimethyl sulfoxide.

Higher percentage of spike-specific cTfh cells after second vaccination in patients with SCD

The percentage of cTfh cells of CD4⁺ T cells was similar between patients with SCD and healthy controls at all time points (Figure 5A). Two vaccinations induced higher percentages of spike-specific cTfh cells in patients with SCD than in controls (Figure 5B). In the ensuing 6 months, the percentage of spike-specific cTfh cells declined significantly in patients with SCD, reaching values comparable with controls. Because Tfh cells help B cells to develop into memory B cells and antibody-producing plasma cells, we investigated the correlation between spike-specific cTfh cells and humoral immune parameters. Frequencies of spike-specific cTfh cells did not correlate with S1 IgG antibody concentration or with the percentage of spike-specific memory B cells at any time point after vaccination (data not shown).

Figure 5.

cTfh cells in patients with SCD compared with HC. (A) Percentage of circulating Tfh cells CXCR5⁺PD1⁺ (cTfh) within the total population of CD4⁺ T cells in patients (blue) and HC (gray) over time. The gating strategy for cTfh cells is indicated in Figure 4A. (B) Percentage of cTfh cells within the population of spike-specific CD4⁺ T cells in patients (blue) and HC (gray) over time. The asterisk indicates significance level, and an alpha significance level of 0.05 is used.

Lower cytokine-producing spike-specific T cells in patients with SCD

The predominant cytokines produced by spike-specific CD4⁺ T cells were T helper 1 (Th1) cytokines IFN-γ, TNF-α, and IL-2 (Figure 6A-C), whereas the production of Th2 cytokine IL-4 and Th17 cytokine IL-17 was low in both groups (supplemental Figure 3). After 2 vaccinations, the percentage of spike-specific CD4⁺ T cells producing IFN-γ, TNF-α, and/or IL-2 was significantly lower in patients with SCD than in healthy controls (47.6% vs 66.0%, respectively; P = .01; Figure 6A). Lower cytokine production after the second vaccination was also observed in spike-specific CD8⁺ T cells in patients with SCD, but this difference was not statistically significant (Figure 6B). The lower percentage of cytokine-producing spike-specific CD4⁺ T cells in patients with SCD was also observed after analysis in an unbiased manner (uniform manifold approximation and projection; Figure 6C). Due to the low number of events, this analysis was not performed for spike-specific CD8⁺ T cells. In particular, spike-specific CD4⁺ T cells producing IFN-γ and TNF-α (median IFN-γ, 21% vs 40% [P = .005]; median TNF-α, 24% vs 34% [P = .02]; Figure 6D) and spike-specific CD8⁺ T cells producing IFN-γ (median IFN-γ, 8% vs 20%; P = .11) were reduced in frequency (supplemental Figure 4). In patients with SCD, more spike-specific CD4⁺ T cells without cytokine production and fewer polyfunctional spike-specific CD4⁺ T cells were detected than controls (Figure 6E). After the third vaccination, cytokine production by spike-specific CD4⁺ T cells increased in some patients (IFN-γ, n = 3/6 patients; TNF-α, n = 1/6; IL-2, n = 3/6), demonstrating heterogeneity in this group, which was less pronounced in healthy controls. After the third vaccination, differences in frequencies of cytokine-producing CD4⁺ T cells between patients and controls tended to remain lower, although no longer statistically significant (Figure 6C-E). The percentage of spike-specific CD4⁺ T cells producing cytokines correlated with S1 IgG antibody concentrations in patients with SCD after the 2-dose vaccination regimen (r = 0.77; P = .04; Figure 6F) but not after the third vaccination (r = −0.57; P = .20; data not shown). No differences between patients with SCD and healthy controls were observed in cytokine production by CEFX-specific CD4⁺ or CD8⁺ T cells (supplemental Figure 5).

Figure 6.

Impaired cytokine production by spike-specific T cells. (A) Percentage of spike-specific CD4⁺ T cells producing IFN-γ, TNF-α, and/or IL-2 in patients (blue) and HC. (B) Percentage of spike-specific CD8⁺ T cells producing IFN-γ, TNF-α, and/or IL-2 in patients (blue) and HC. (C) UMAP analysis of raw flow cytometry data of CD154, CD137, CD69, IFN-γ, TNF-α, and IL-2 expression of spike-specific CD4⁺ T cells. Data were obtained 28 days after the second and third vaccinations in patients and HC. (D) Percentages of cytokine-producing spike-specific CD4⁺ T cells over time for patients (blue) and HC stratified by Th1 cytokines (IFN-γ and TNF-α) and IL-2. (E) Proportion of spike-specific CD4⁺ T cells that produce 0, 1, 2, or 3 cytokines (IFN-γ, TNF-α, and IL-2) in patients (blue) and HC (gray) after the second and third vaccinations. (F) Spearman correlation between S1 IgG antibody concentration and the percentage of spike-specific CD4⁺ T cells producing cytokines in patients with SCD after the second vaccination. The asterisk indicates significance level, and an α significance level of 0.05 is used. FITC-A, Fluorescein Isothiocyanate-A; PE-A, Phycoerythrin-A; APC-A, Allophycocyanin-A; UMAP, uniform manifold approximation and projection.

HbF percentage correlated with vaccine immunogenicity outcomes

To determine whether reduced cytokine production by spike-specific CD4⁺ T cells in patients with SCD was related to hydroxyurea use, we investigated the association between immunological parameters and the mean corpuscular volume (MCV) and the percentage of fetal hemoglobin (HbF) before vaccination. HbF percentage and MCV are generally accepted correlates of hydroxyurea adherence and response.^33,34 HbF percentage at baseline, before vaccination, was negatively correlated with S1 IgG antibody concentrations after 2 vaccinations (r = −0.52; P = .02) and with the percentages of Th1 cytokine–producing spike-specific CD4⁺ T cells (r = −0.50; P = .008; Figure 7). To verify whether this correlation was confounded by individual differences in HbF percentage before the start of hydroxyurea, we recalculated correlation coefficients based on the difference in HbF percentage before the start of hydroxyurea and HbF percentage at the start of this study. Both associations remained significant (r = −0.80 [P = .005]; r = −0.91 [P = .001]; supplemental Figure 6). None of the other immunological outcomes showed a significant correlation with baseline HbF levels (data not shown). After the third vaccination, associations were no longer significant. MCV did not correlate with immunological parameters (data not shown).

Figure 7.

Correlation between HbF percentage and vaccination-induced immune responses. (A) Correlation (Spearman rank test) between HbF percentage at study start and S1 IgG concentrations after the second vaccination in patients with SCD. (B) Correlation (Spearman rank test) between HbF percentage at study start and the percentage of cytokine-positive spike-specific CD4⁺ T cells after the second vaccination in patients with SCD.

Antibody maturation improved by SARS-CoV-2 infection in vaccinated patients with SCD

Of 31 patients with SCD included in this study, 6 patients became infected with SARS-CoV-2 before or during the study. Infection was confirmed by N IgG ≥14.3 BAU/mL at any time point during the study. These patients were not included in the vaccine immunogenicity analyses (for baseline characteristics, supplemental Table 4). None of these patients experienced severe COVID-19, as defined by hospitalization or COVID-19–related death. S1 IgG antibody concentrations after vaccination in these patients were comparable with those in patients with SCD who were only vaccinated. However, the neutralizing capacity per antibody against SARS-CoV-2 was higher in patients who had been infected than in patients with SCD who were vaccinated but never infected (supplemental Figure 7).

Discussion

In this prospective cohort study of patients with SCD on hydroxyurea, we confirm that COVID-19 mRNA vaccination resulted in frequencies of spike-specific B and T cells, as well as robust neutralizing antibody responses, similar to those in healthy individuals.^19,21,22,35 However, the capacity of vaccine-induced spike-specific T cells to produce Th1 cytokines (IFN-γ and TNF-α) was significantly reduced compared with those in healthy controls. A third vaccination improved cytokine production of spike-specific CD4⁺ T cells in some patients, but most patients with SCD did not reach cytokine-producing spike-specific CD4⁺ T-cell levels comparable with those in vaccinated healthy individuals.

Previous studies on influenza, yellow fever, and measles vaccination have demonstrated that patients with SCD on hydroxyurea exhibited a reduced or delayed vaccine-induced antibody response.^17,18,36 It is therefore remarkable that COVID-19 mRNA vaccines elicited a robust neutralizing antibody response in these patients. After COVID-19 booster vaccination, an even higher antibody concentration was observed than in healthy controls, which cannot be explained by age, vaccine type, or the number of vaccine doses, because these parameters were matched between patients and controls. The time interval between the second and third vaccinations differed between healthy individuals (8 months) and patients (9 months) but not enough to account for the observed differences. The higher concentrations of circulating Tfh cells observed in patients with SCD could offer an explanation, because Tfh cells help B cells to differentiate into memory and antibody-producing cells. It may be hypothesized that higher cTfh cell concentrations at the time of second vaccination led to better B-cell maturation, with higher S1 IgG concentrations after repeated antigen exposure. Nevertheless, a correlation between cTfh cell frequencies and antibody concentrations could not be demonstrated, potentially due to the overall homogeneous high antibody responses after vaccination in patients with SCD. A limitation of our study is that patients and controls were not matched for ethnicity. Ethnicity has been reported as a factor in humoral immunity^37,38; however, the observation that COVID-19 vaccination induced higher antibody concentrations in patients with SCD than in healthy individuals has also been reported in an ethnicity-matched cohort.¹⁹

We observed a skewing of B-cell subsets toward a naïve phenotype in patients with SCD compared with those controls, in line with previous reports.^39,40 In these studies, a defective differentiation from naïve to memory B cells has been suggested,⁴⁰ which may contribute to dysregulated humoral immune responses, as observed after influenza, yellow fever, and measles vaccinations.^17,18,36 In our study, we found robust neutralizing antibody responses and did not observe an impaired differentiation because the percentages of spike-specific memory B-cell subsets were comparable with those in controls. This may reflect a higher immunogenic capacity of mRNA vaccines.²⁵

Although the dynamics and frequencies of spike-specific T cells were comparable with those observed in healthy individuals, the percentage of Th1 cytokine–producing spike-specific T cells was lower in patients with SCD. Reduced TNF-α levels in patients with SCD receiving hydroxyurea have been reported during vaso-occlusive crises and in steady-state conditions.⁴¹ However, our data do not indicate an intrinsically impaired cytokine production of T cells because cytokine-producing CEFX-specific T-cell responses were comparable with those in healthy individuals. We speculate therefore that lower Th1 cytokine production by spike-specific T cells might be a manifestation of a hampered primary T-cell response in patients with SCD, which was only partially reversed by a third vaccination. Additionally, higher frequencies of spike-specific cTfh cells observed 4 weeks after the second vaccination may point toward a difference in the immune response dynamics of patients with SCD compared with those in controls. Whether these differences in the immune responses result from adaptive immune alterations described in patients with SCD,39, 40, 41 from hydroxyurea therapy, or from other factors, such as (treatment of) iron overload, splenic dysfunction, genetic predispositions, concurrent medication, or concurrent vaso-occlusive crises, requires further investigation.

The presence of spike-specific T cells and neutralizing S1 IgG is associated with a reduced severity of COVID-19.42, 43, 44, 45, 46 Despite comparable S1 IgG concentrations and spike-specific T-cell frequencies between 2-dose–vaccinated patients with SCD and controls, higher COVID-19–related mortality has been reported in patients with SCD.⁷ The reduced cytokine production by spike-specific CD4⁺ T cells may contribute to this risk, because Th1 cytokines produced by antigen-specific CD4⁺ T cells are crucial for immune responses against viral infections. These cytokines help to recruit other immune cells, support the development of antigen-specific B cells into memory B cells, and assist CD8⁺ T cells in killing infected cells.^47,48 We did not observe lower spike-specific antibody or memory B-cell concentrations; however, we did observe a trend toward decreased levels of cytokine-producing CD8⁺ T cells in patients compared with those in controls. Taken together, T-cell–mediated viral clearance may be impaired in patients with SCD on hydroxyurea, which could contribute to the increased risk of severe COVID-19 in this population. However, factors inherent to SCD, such as the occurrence of vaso-occlusive crisis or acute chest syndrome triggered by even mild viral infections, and SCD-associated comorbidities, may also play a significant role herein.⁴⁹

Interestingly, a negative correlation between HbF and S1 IgG and Th1 cytokine production was observed. Although this may raise concerns, it does not outweigh the clear protective effect of hydroxyurea against virus-induced vaso-occlusive crises with potentially fatal pulmonary complications such as acute chest syndrome.

Limitations of the study, in addition to the lack of an ethnicity-matched control group, include the lack of patients with SCD who do not use hydroxyurea, as well as the fact that only a limited number of patients completed follow-up after the third vaccination.

Although reduced or delayed immune responses to other types of vaccines have been observed, our study indicates that COVID-19 mRNA vaccines elicited robust neutralizing antibody responses in patients with SCD. This could be due to the strong immunostimulatory capacity of mRNA vaccines, which, in contrast to some of the other types of vaccines, engages B and T cells and induces persistent cellular and humoral immunity. mRNA vaccination may be more effective than other vaccine types in patients with SCD and may significantly contribute to the protection of these patients with a heightened risk of viral infection–related complications. We encourage future research into adaptive immunity in patients with SCD and the potential effects of hydroxyurea.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

A complete list of the members of the COBRA KAI Study Team appears in “Appendix.”

Appendix

The members of the COBRA KAI Study Team are Johan J.H. van Meerloo, Gaby P. Smit, Dorine Wouters, Ester M. M. van Leeuwen, Hetty J. Bontkes, Neeltje Kootstra, Iris M. J. Kant, Thecla Graas, Belle Toussaint, Sterre de Jong, Shahan Darwesh, Sandjiv S. Mahes, Dora Kamminga, Matthijs Koelewijn, Gino Faber, Guus Beaumont, Marije D. Engel, R. Cheyenne N. Pierie, Suzanne R. Janssen, Gino Faber, Kazimierz Groen, Judith A. Burger, Joey H. Bouhuijs, Paul A. Baars, Edith van Dijkman, Jarom Heijmans, Yara Y. Witte, Rogers A. Nahui Palomino, Said Z. Omar, Sonja Zweegman, Arnon P. Kater, Caya van den Vegt, Ilonka Arends-Halbesma, Emma de Pater, Margriet J. Dijkstra, Josien van Beek, Nynke Y. Rots, Esther Siteur-van Rijnstra, Dennis M. de Rooij, Rogier W. Sanders, Meliawati Poniman, Wouter Olijhoek, Jacqueline van Rijswijk, Tim Beaumont, Lusia Çetinel, Louis Schellekens, Yvonne M. den Hartogh, Jacqueline Cloos, Suzanne S. Weijers, Saïda Tonouh-Aajoud, Selime Avci, Elianne Roelandse-Koop, Willem A. Dik, and Sandra Vogels-Nooijen.