Black Americans With Sickle Cell Disease (SCD) Demonstrate Accelerated Epigenetic Pace of Aging Compared to Black Americans Without SCD

Abstract

Sickle cell disease (SCD), an inherited hemoglobinopathy affecting 1 in 365 Black Americans, is a complex chronic disease characterized by red blood cell sickling, vaso-occlusion and subsequent ischemia, hemolytic anemia, organ damage, and ultimately, premature death (1). Despite this significant pathophysiology, clinical outcomes are quite variable, and survival of patients with SCD continues to improve, with current estimates extending past the fifth decade of life (2–4). However, even SCD patients below 50 years demonstrate pronounced acceleration of clinical markers of aging, with physical performance measures similar to individuals without SCD who are 20–30 years older (5,6). One reason for this could be the extensive physical, psychosocial, and psychological stress that patients endure over the course of their lives as a result of SCD (7). Recurrent vaso-occlusive pain crises and chronic anemia cause significant ischemic injury, inflammation, and fatigue. In addition to somatic stress, 24%–30% of SCD patients report having depression, a percentage that is much higher than in the general and Black populations (8), linked to the stress of financial hardship related to frequent hospitalizations and the cost of pain medication, unemployment, low educational attainment, and a lack of social support. Because SCD is a lifelong disease, these stressors are ever present and significantly impact both the body and mind of the patient over the entire life course. Given the profound lifetime stress and accelerated clinical presentation of patients with SCD, we hypothesize that SCD patients are experiencing accelerated biological aging.

Indeed, inflammation and lifetime stress have been linked to accelerated epigenetic aging (9,10), a measure of biological aging that can be assessed with DNA methylation (DNAm). Changes to DNAm over time not only occur as the result of environmental exposures and cellular damage but are an expected component of normal biological aging (10). By assessing DNAm at specific cytosine-phosphate-guanine (CpG) loci across the epigenome, one’s epigenetic age, which is an estimate of biological age, can be calculated (11). Many epigenetic clocks have been developed to calculate epigenetic age based on a subset of CpG loci, in an effort to quantify biological age (12). The “first-generation” clocks, Horvath (11) and Hannum (13), simply estimate chronological age but correlate poorly with mortality risk. The “second-generation” clocks, PhenoAge (14) and GrimAge (15), were specifically developed to associate with mortality, including clinical parameters and biomarkers with known mortality risk associations aimed at refining biological aging prediction. However, these clocks were created using cross-sectional data on individuals of widely varying chronological age or were trained on mortality, rather than directly on measures of biological aging. Unlike prior epigenetic clocks, the Dunedin Pace of Aging Calculated from the Epigenome (DunedinPACE) (16) measure was developed using data from a longitudinal birth cohort, with individuals of the same age assessed four times from age 26 to 45, thereby greatly reducing cohort effects, survival bias, and disease process bias. Importantly, the DunedinPACE epigenetic measure of biological aging was trained on the Pace of Aging (17), which was derived from longitudinal assessment of 19 biomarkers spanning multiple organ systems, and could provide a more refined metric in the context of a lifelong chronic disease like SCD. Utilizing several epigenetic clocks, we recently described estimates of epigenetic aging in a smaller subset of SCD patients, comparing those estimates with chronological age among SCD, with inconsistent results (18). However, without a non-SCD comparison group, it is not possible to determine if SCD patients exhibit accelerated epigenetic aging compared with those without the disease. Moreover, increasing evidence suggests the presence of racial disparities in epigenetic age acceleration, with Black individuals being epigenetically older than White individuals (14,19–21). These disparities are likely influenced by factors such as experiences of racism and other psychosocial stressors, economic and educational disadvantages, and unequal access to health care (22). Thus, given that SCD predominantly affects people who are Black in the United States, it is important to compare the pace of aging to other Black individuals without SCD in order to understand the aspects of aging that could be attributed to the disease process rather than racial disparities.

To that end, we used DNAm data to compare DunedinPACE epigenetic pace of aging scores in 131 Black Americans with SCD from the Outcome Modifying Genes in SCD (OMG-SCD) (3) and the Duke SCD Implementation Consortium (Duke SCDIC) (23) cohorts to 1391 non-Hispanic Black (NHB) American veterans without SCD from the U.S. Department of Veterans Affairs (VA) Mid-Atlantic Mental Illness Research, Education, and Clinical Center (MIRECC) cohort (24). By selecting the MIRECC cohort as ancestry-matched non-SCD controls, we were able to compare the DunedinPACE epigenetic measure of aging in SCD patients to that of NHB individuals with and without a different chronic disease, namely posttraumatic stress disorder (PTSD), a condition previously associated with accelerated epigenetic aging (25).

Method

Cohorts

SCD subjects (total n = 134) were ascertained from the OMG-SCD cohort (3) (n = 45) and the Duke SCDIC cohort (23) (n = 89). The OMG-SCD study included adult subjects (≥18 years at the time of enrollment) diagnosed with SCD who were enrolled at 5 sickle cell centers in the Southeastern United States between 2002 and 2015 for the purpose of identifying genetic and genomic factors associated with clinical variability in SCD. Participants provided informed consent and study protocols were approved by local institutional review boards (IRB) or ethics committees. A subset of OMG-SCD individuals who survived to at least 50 years of age were selected for the present analysis, utilizing DNA samples collected at the time of enrollment (n = 45). The SCDIC is a multisite research program intended to develop a longitudinal registry of SCD individuals. It currently maintains a cohort of 2 423 adults (age ≥ 18 years) in the United States. Of these adults, 89 individuals collected through the Duke University Medical Center site and who had DNA methylation data available from a previous study (18) were utilized in the current analysis. SCD patients were classified as having either severe or mild disease based on their genotype at the beta-globin locus as follows: HbSS and HbSβ⁰ genotypes were classified as severe, while HbSC and HbSβ⁺ genotypes were classified as less severe.

Control individuals were derived from a multisite study of U.S. Afghanistan and Iraq-era veterans. The VA Mid-Atlantic MIRECC (24) began study data collection in 2005 as a regional cohort data repository to facilitate future mental health research focused on the millions of the troops returning from post-9/11 deployment. Study protocols were approved by the local IRB of each participating VA hospital, and written informed consent was obtained from all participants prior to study enrollment. About half the MIRECC participants self-identify as NHB. The current study included any NHB participants who had DNAm data available (n = 1392). Notably, none of these veterans were diagnosed with SCD. We further confirmed the clinical diagnosis by assessing each individual’s genotype at HBB rs334, the causal SCD variant, using imputed GWAS data described elsewhere (26). One individual was homozygous alternate for the sickle mutation and was therefore removed, leaving 1391 individuals for statistical analysis.

Identity-by-Descent Across SCD Cohorts

To ensure sample independence, identity-by-descent (IBD) estimates were generated using whole genome sequencing data obtained from the NHLBI TOPMed program for the OMG-SCD and Duke SCDIC subjects. Variants passing standard QC filters (27) were subsequently filtered to include only biallelic single nucleotide polymorphisms (SNPs) with minor allele frequency >1% that were not missing for >15% of samples. Resulting SNPs were further filtered to remove linkage disequilibrium using a window size of 100, shifting by 25 SNPs at a time, and setting the variance inflation factor to 1.1 (--indep 100 25 1.1), prior to IBD estimation (--genome) in PLINK (28). Three sets of first-degree relatives (IBD ~0.5) were identified across the 2 SCD cohorts; one subject was removed at random from each pair to eliminate sample relatedness. No duplicate individuals were identified. Therefore, 131 SCD subjects were deemed appropriate for analysis.

Generation of Methylation Data

The generation of methylation data in the Duke SCDIC (18) and MIRECC (29) subjects has been previously described. For the OMG-SCD cohort, extracted DNA samples for 45 subjects were used to generate DNAm from the Infinium MethylationEPIC Beadchip (Illumina, San Diego, CA) by the Duke Molecular Genomics Core at the Duke Molecular Physiology Institute. A total of 250 ng of DNA was plated into 96-well plates and bisulfite treated using the EZ DNA Methylation-Direct Kit (Zymo). Genome-wide DNA methylation was measured on the beadchip from these bisulfite-converted DNA samples. The data were preprocessed using Illumina GenomeStudio software. Sample and probe quality control (QC) was performed using minfi (30) and ChAMP (31) R packages. The relative levels of methylation (β) were calculated as the ratio of methylated probe signal to total locus signal intensity. Probe QC and data normalization were performed within each batch using the R package wateRmelon (32). Probes not detected (detection p > .0001) in >10% of samples, as well as probes hybridizing to multiple locations in the genome (cross-reactive) were removed (33–35). Correction of sample data by chip and by chip position was also performed. Raw β values were normalized using the dasen approach (32) and adjustments for both batch and chip were accomplished using ComBat (36) in the R package sva. M-values were calculated from the normalized and adjusted beta values for statistical analysis. Ultimately, no OMG-SCD samples were excluded on the basis of the QC process. Finally, blood cell-type proportions were estimated for all samples using the R package EpiDISH (37).

Statistical Analysis

DunedinPACE scores were calculated for participants in the current study as described in the original DunedinPACE publication (16), using data from 20 000 CpG probes. This produced an aging score for each individual whereby a value of one represents one year of biological aging per chronological year (ie, expected aging). Therefore, scores above one represent an accelerated pace of biological aging and scores below one represent a slower pace of biological aging compared with chronological age.

All statistical analyses were performed in R. Difference of means of DunedinPACE between SCD patients and NHB veterans were tested using linear models (glm). Chronological age was included as a covariate in all statistical models and least squares means with 95% confidence intervals (CI) are presented. Sensitivity analyses were performed in several ways: excluding veterans from the MIRECC cohort who had a diagnosis of PTSD, excluding veterans from the MIRECC cohort without PTSD, restricting SCD subjects to only those with severe SCD genotypes, performing sex-stratified analyses, and adjusting for estimated cell-type proportions. Unadjusted DunedinPACE means and standard deviations are reported in Supplementary Table 1 for all subject groups examined.

Results

Participant characteristics are shown in Table 1. SCD subjects were significantly younger by chronological age (mean age_SCD = 34.5 years vs mean age_MIRECC = 39.4 years; p = 1.48 × 10⁻⁷) and had a higher proportion of females (62.6% vs 33.3%, p = 4.28 × 10⁻¹¹) compared with NHB veterans (Table 1). All OMG-SCD subjects were classified with severe SCD, while 74.42% of Duke SCDIC subjects had severe SCD. The frequencies of SCD-related complications among the SCD patients in this analysis generally agree with frequencies in the larger OMG-SCD and SCDIC cohorts. Characteristics for the 2 SCD cohorts separately can be found in Supplementary Table 2. Consistent with estimates in the general U.S. population (38), 9.41% of the NHB veterans were heterozygous for rs334, indicating they have sickle cell trait (SCT).

We observed significantly faster DunedinPACE aging scores in individuals with SCD compared with NHB veterans (Figure 1; Table 2; p = 4.49 × 10⁻⁸); the DunedinPACE estimate for those with SCD was 1.165 (95% CI: 1.146–1.185) compared with 1.108 (95% CI: 1.102–1.114) for NHB veterans. Therefore, those with SCD were epigenetically aging 0.684 months more per chronological year than NHB veterans without SCD. This was observed, despite the SCD subjects being significantly younger by chronologic age. Given established associations between DunedinPACE (16) and chronological aging, this makes the observed epigenetic discrepancy even more apparent. Despite differences between the two SCD cohorts, both exhibited a significantly accelerated pace of aging compared with NHB veterans, when considered separately (Supplementary Table 3).

Figure 1.

Epigenetic pace of aging in sickle cell disease (SCD) as measured by DunedinPACE. Blue bars and lines (shifted right) depict SCD patients; red bars and lines (shifted left) depict non-Hispanic Black (NHB) veterans.

Open in new tabDownload slide

Because veterans with PTSD display accelerated epigenetic aging (25,39,40), we removed MIRECC subjects with PTSD to obtain a comparison group that is more aligned with the general population and again compared them to those with SCD. NHB veterans without PTSD (N = 882) had a DunedinPACE estimate of 1.0996 (95% CI: 1.092–1.107), resulting in a larger difference of means compared with SCD subjects, who had a DunedinPACE estimate of 1.163 (95% CI: 1.144–1.182, p = 2.18 × 10⁻⁹; Figure 2; Table 2). Compared with NHB veterans without PTSD, subjects with SCD were epigenetically aging 0.763 months more per chronological year.

Figure 2.

Age-adjusted DunedinPACE mean scores for 5 subsets of the data set: All non-Hispanic Black (NHB) veterans, NHB veterans without posttraumatic stress disorder (PTSD), NHB veterans with PTSD, all sickle cell disease (SCD) patients, and SCD patients with HbSS or HbSβ⁰ genotypes. SCD patients have significantly higher DunedinPACE scores compared with all NHB veterans (p = 4.49 × 10⁻⁸), NHB veterans without PTSD (p = 2.18 × 10⁻⁹), and NHB veterans with PTSD (p = 5.02 × 10⁻⁵). SCD patients with HbSS or HbSβ⁰ genotypes have significantly higher DunedinPACE scores compared with all NHB veterans (4.2 × 10⁻⁹), NHB veterans without PTSD (p = 1.61 × 10⁻¹⁰), and NHB veterans with PTSD (5.94 × 10⁻⁶).

Open in new tabDownload slide

Next, we restricted the analysis to only SCD subjects with more severe hemoglobin genotypes (N = 109) and observed an even larger increase in DunedinPACE aging scores compared with NHB veterans (1.175 [95% CI: 1.154–1.196] vs 1.109 [1.103–1.114], p = 4.2 × 10⁻⁹; Figure 2; Table 2). Thus, SCD subjects with severe disease genotypes were epigenetically aging 0.794 months more per chronological year than NHB veterans. Finally, we compared SCD subjects with severe hemoglobin genotypes to NHB veterans without PTSD and detected the largest difference in epigenetic pace of aging estimates (1.173 [95% CI: 1.152–1.194] vs 1.1 [1.093–1.107], p = 1.61 × 10⁻¹⁰; Figure 2; Table 2), indicating that subjects with severe SCD genotypes were epigenetically aging 0.874 months more per chronological year than NHB veterans without PTSD. These data support our hypothesis that SCD patients experience a faster rate of biological aging as measured by global epigenetic variation compared with other Black individuals without SCD.

To examine DunedinPACE scores in two chronic disease states, we also compared SCD patients to MIRECC subjects with a PTSD diagnosis. SCD patients (DunedinPACE = 1.165; 95% CI: 1.146–1.185) still displayed a significantly faster pace of aging compared with NHB veterans with PTSD (DunedinPACE = 1.120, 95% CI: 1.110–1.130, p = 5.02 × 10⁻⁵; Figure 2; Table 2), equivalent to 0.546 months more per chronological year. When restricting to SCD patients with severe hemoglobin genotypes, we observed an even larger increase in epigenetic pace of aging compared with NHB veterans with PTSD (1.175 [95% CI: 1.154–1.196] vs 1.121 [95% CI: 1.111–1.131], p = 5.94 × 10⁻⁶; Figure 2; Table 2), equivalent to 0.647 months more per chronological year.

Sex was not associated with DunedinPACE scores in the SCD patients (p = .332), however was associated in the MIRECC cohort (p = 4.51 × 10⁻¹¹), such that NHB female veterans were epigenetically aging 0.528 months more per chronological year than NHB male veterans, as previously reported (41). Although this cohort*sex interaction did not reach statistical significance, it did show a trending association (p = .073; Figure 3). We were unable to match the cases and controls based on sex because the frequencies were nearly opposite in the 2 cohorts: the SCD cohorts were comprised of 62.6% females, while the MIRECC cohort was comprised of 33.3% females. Therefore, we performed sex-stratified analyses to control for the sex*cohort trend we observed. The effect of DunedinPACE was significantly different in the males (p = 1.21 × 10⁻⁵; Table 2), such that the male SCD patients (DunedinPACE = 1.165, 95% CI: 1.134–1.196) were epigenetically aging 0.852 months more per chronological year than the NHB male veterans (DunedinPACE = 1.094, 95% CI: 1.087–1.101). Likewise, SCD females displayed a higher DunedinPACE score compared with NHB female veterans (1.163 [95% CI: 1.139–1.187] vs 1.138 [95% CI: 1.128–1.148]), but this difference did not reach statistical significance (p = .0569; Table 2).

Figure 3.

Age-adjusted DunedinPACE mean scores depicting a trending cohort*sex interaction (p = .073). NHB = non-Hispanic Black; SCD = sickle cell disease.

Open in new tabDownload slide

Because DNAm was measured in whole blood, we estimated leukocyte proportions in all samples for inclusion as technical covariates in our linear models. However, estimated proportions of CD8T cells, B cells, and monocytes were significantly different by case status, and consequently, confounded with the analysis groups of interest, namely SCD compared with non-SCD (Supplementary Table 4). Including cell-type estimates in the primary model comparing SCD subjects to NHB veterans resulted in a slightly attenuated, but still significant effect of DunedinPACE (p = 7.2 × 10⁻⁷): SCD patients (DunedinPACE = 1.162, 95% CI: 1.142–1.182) were epigenetically aging 0.64 months more per chronological year than NHB veterans without SCD (DunedinPACE = 1.108, 95% CI: 1.103–1.114).

Discussion

In this study, we provide preliminary evidence for an accelerated epigenetic pace of aging in subjects with SCD compared with NHB subjects without SCD. To our knowledge, this is the first report of accelerated epigenetic aging in SCD patients compared with ancestry-matched controls. These effects were observed despite the SCD subjects being significantly younger in terms of chronological age at the time of sample collection. The differences in DunedinPACE aging score estimates between SCD patients and individuals without SCD ranged from 0.057 to 0.073, depending on the subset of subjects used, corresponding to an increased pace of aging of 6.8 to 8.7 months over the course of each decade of life. This increased pace of aging is comparable to that observed in breast cancer survivors compared with controls (42), providing context for the accelerated rate of aging we report in SCD patients. Although this acceleration in the epigenetic pace of aging may help explain why SCD patients appear more similar clinically to individuals without SCD who are 20–30 years older (5,6), it is likely only one piece of the puzzle, and more work is needed to uncover the still unidentified factors influencing the rapid aging seen in SCD patients.

Importantly, the average DunedinPACE scores for both the SCD patients and NHB veterans reported here were higher than the benchmark aging score (1.0) for the original DunedinPACE measure (16), suggesting that all subjects in the current analysis were aging faster than the reference population. By removing veterans with PTSD, we attempted to make our comparison group more aligned with the general population, but recognize that veterans without a diagnosis of PTSD still evidence accelerated epigenetic aging as a result of stressors and experiences related to military service, such as traumatic brain injury and exposure to toxic chemicals (25,43,44). Conversely, when comparing the DunedinPACE score of SCD patients to those with another chronic disease, PTSD, we continued to observe pronounced epigenetic age acceleration. Future studies should compare the rate of epigenetic aging in SCD patients to that of healthy nonveteran Black Americans without SCD or other chronic diseases. We anticipate that SCD patients will demonstrate an even more accelerated pace of epigenetic aging in that comparison.

As part of a larger study, we have recently shown that the NHB veterans included in this analysis were aging more rapidly than veterans of European descent, equivalent to 0.8 months of additional aging per year, and that acceleration was not attributable to genetic ancestry, but more likely, to experiences of discrimination, structural disadvantages, or other social determinants (41). Similarly, a recent study by Shen et al showed that African American race was associated with higher DunedinPACE scores in a population-based study of socioeconomically diverse adults in Baltimore, Maryland (21). This is particularly important when interpreting the accelerated aging we observed among SCD patients in this study, who are predominantly Black. Efforts to slow aging and improve health for SCD patients should include not only improved clinical intervention, but also efforts to combat racial disparities.

Due to the nature of the 2 populations examined, the proportion of males and females in the SCD and veteran cohorts was nearly opposite. Although the proportion of women in the military is well below that of men (45), OMG-SCD patients were only selected for the current study if they had survived to at least 50 years of age, who we and others have shown are more likely to be female (5,46), thus possibly resulting in a selection bias for those with the lowest disease burden. Despite this, the rate of epigenetic aging was not statistically different by sex among SCD patients. The rate of epigenetic aging in NHB female veterans was, however, much closer to that of the SCD patients than the NHB male veterans, perhaps as a consequence of increased rates of childhood and sexual trauma (47), and added stress incurred by NHB females during military service. Although this strong sex association in the veteran sample is driving the sex*cohort relationship we observed, larger sample sizes are necessary to more fully understand the relationship between estimates of biological aging and sex in SCD patients.

Although we sought to use the largest possible sample of SCD patients with epigenetic data available, we recognize that the 2 SCD cohorts utilized are quite different with respect to chronological age, sex, SCD severity, and some SCD-related complications (Supplementary Table 2). For instance, the frequencies of hydroxyurea usage and splenectomy among the SCDIC patients were higher, likely due to changing clinical practices, while the frequency of stroke was lower in OMG-SCD patients, likely reflecting the effect of stroke on overall survival. Despite these cohort differences, we observed a significantly accelerated pace of epigenetic aging in each SCD cohort separately compared with NHB veterans without SCD (Supplementary Table 3), suggesting that the association we identified is not due to sampling bias, but rather to the pathophysiology of SCD. This is important because even though the OMG-SCD patients were selected for longevity and represent a group of patients with less critical end-organ damage, they still evidence an accelerated pace of aging, indicating that this effect remains even in the “healthiest” SCD patients.

Epigenetic studies in whole blood can be biased if cell-type proportions are not addressed, as patterns of DNA methylation have been shown to be different in different types of leukocytes (48). However, SCD patients are known to have altered leukocyte proportions as a consequence of their disease, particularly chronic inflammation due to hemolysis. Other SCD-specific factors known to affect leukocyte count include hemoglobin genotype, reduced splenic function or splenectomy, and hydroxyurea usage. Therefore, it was unsurprising that we detected lower CD8T counts, and higher B cell and monocyte counts in the SCD patients compared with those without SCD (Supplementary Table 4). Consequently, leukocyte proportions were confounded with our analysis groups of interest, namely SCD versus non-SCD. For these reasons, we chose not to include the leukocyte estimates in our main analysis. Still, we performed a sensitivity analysis including leukocyte proportions as covariates to ensure that the effects we observed with DunedinPACE were not solely due to differences in cell-type estimates. Indeed, when including cell-type estimates as covariates, we still observed a significantly accelerated pace of aging in SCD patients compared with NHB veterans. Because of this finding, and the work by Belsky and colleagues (16) showing that DunedinPACE effect sizes are robust to covariate adjustment of estimated cell counts, we are confident that the associations we report are not an artifact of cell-type heterogeneity.

Of note, within the MIRECC cohort, NHB veterans with SCT did not show sex- and age-adjusted accelerated DunedinPACE aging scores compared with NHB veterans with the wild-type beta-globin genotype (p = .9789). Although SCT is relatively benign, several studies have correlated SCT with increased risk for adverse health outcomes (49) and proteomic analysis among subjects with SCT has identified proteins previously associated with kidney function or injury, hemolysis, and inflammation (50). Replication in other cohorts is necessary to confirm the lack of accelerated biological aging we observed in the MIRECC cohort.

Several considerations should be noted when interpreting the results of this analysis. First, DunedinPACE was developed using individuals of primarily European descent. Here, we have applied it to individuals of primarily African descent. More research utilizing longitudinal datasets is necessary to investigate how well DunedinPACE translates to individuals of different ancestral backgrounds. Also, some of the SCD patients utilized here were selected for longevity and therefore have low disease burden. Similarly, the use of a military control group likely resulted in participants with more stress than the general population. The effect of these 2 biases would have been to diminish any differences rather than augment them. Thus, we are encouraged by the strong associations we report here, but also recognize that much larger sample sizes in unselected SCD cohorts compared with the general population of Black Americans is necessary to confirm, and perhaps strengthen, the findings of this study. Finally, we were unable to determine if the accelerated epigenetic pace of aging we observed is associated with increased mortality, as nearly all SCD subjects were still living at the time of analysis. To more fully understand this association, future work should examine the relationship between epigenetic measures of biological aging and specific types of end-organ damage, overall SCD severity, and survival in a larger SCD cohort. Future studies should also examine the relationship between physical measures of aging, such as gait speed and frailty, with epigenetic measures of biological aging.

In summary, SCD patients are experiencing a remarkable accelerated rate of epigenetic aging, compared with NHB veterans without SCD, in concordance with the significant multi-organ damage and lifetime stress that is inherent to SCD and with the premature reduction in physical performance observed clinically among these patients. This increased pace of aging is occurring regardless of the patient’s chronological age, sex, or disease severity. Given the widely recognized clinical and prognostic variability of SCD, the assessment of epigenetic aging may prove useful to identify which SCD patients would most benefit from clinical interventions to reduce mortality. In the long run, we anticipate these measures would likely benefit future studies investigating interventions designed to improve health among SCD patients, as well as provide an imperative first step toward improved patient care.

Funding

This work was supported in part by the National Institute of Nursing Research (R21NR020017 to A.A.-K. and M.R.K.); the National Heart, Lung, and Blood Institute (R01HL68959 and R01HL079915 to M.J.T. and U01HL133964 to P.T. and N.S.); the National Center for Advancing Translational Sciences of the National Institutes of Health (UL1TR002553 to A.A.-K. and C.O.); the Clinical Science Research and Development (CSR&D) Service of VA ORD (IK2CX002694 to K.J.B., IK2CX000525 to N.A.K., and lK6BX003777 to J.C.B.); and the Biomedical Laboratory Research and Development (BLR&D) Service (I01BX002577 to J.C.B.). The authors also received support from the VA Mid-Atlantic Mental Illness Research, Education and Clinical Center (MIRECC), the Mental Health and Research Services of the Durham VA Healthcare System, the Durham VA Geriatrics Research, Education, and Clinical Center (GRECC), and the Department of Psychiatry and Behavioral Sciences at the Duke University School of Medicine.

Conflict of Interest

N.S. is a consultant for Global Blood Therapeutics (GBT)/Pfizer, Forma, Agios, Vertex, and bluebird bio, is a speaker for GBT/Pfizer and Alexion, and performs research on GBT/Pfizer. P.T. is a consultant for CSL Behring. J.S. is a consultant for Disc Medicine and Editas Medicine and has research funding from Agios Pharmaceuticals. All other authors declare no competing financial interests.

Acknowledgments

We sincerely thank all of the sickle cell disease patients and veterans who volunteered for this study. The views expressed in this manuscript are solely those of the authors and do not necessarily reflect the position or policy of the funding agencies, the VA, or the United States government. The VA Mid-Atlantic MIRECC Workgroup includes Pallavi Aurora, PhD, Jean C. Beckham, PhD, Patrick S. Calhoun, PhD, Eric Dedert, PhD, Eric B. Elbogen, PhD, Tate F. Halverson, PhD, Robin A. Hurley, MD, Jason D. Kilts, PhD, Angela Kirby, MS, Anna T. Magnante, PsyD, Sarah L. Martindale, Ph.D, Brandy S. Martinez, PhD, Christine E. Marx, MD, MS, Scott D. McDonald, PhD, Scott D. Moore, MD, PhD, Victoria O’Connor, PhD, Rajendra A. Morey, MD, MS, Jennifer C. Naylor, PhD, Jared Rowland, PhD, Robert D. Shura, PsyD, Cindy Swinkels, PhD, & Elizabeth E. Van Voorhees, PhD, and H. Ryan Wagner, PhD.

Author Contributions

M.E.G.: Formal analysis (lead), Data curation (equal), Software (lead), Writing—Original Draft (lead), Writing—Review & Editing (lead). B.L.: Formal analysis (supporting), Writing—Review & Editing (supporting). K.B.: Software (supporting), Writing—Review & Editing (supporting) M.D.: Data curation (equal), Writing—Review & Editing (supporting). D.H.: Data curation (equal), Writing—Review & Editing (supporting). Q.Y.: Supervision (supporting), Writing—Review & Editing (supporting). P.T.: Funding acquisition (equal), Writing—Review & Editing (supporting). N.S.: Funding acquisition (equal), Writing—Review & Editing (supporting). F.L.: Funding acquisition (supporting), Writing—Review & Editing (supporting). C.O.: Funding acquisition (equal), Writing—Review & Editing (supporting) Mid-Atlantic VA MIRECC Workgroup: Writing—Review & Editing. J.S.: Funding acquisition (supporting), Writing—Review & Editing (supporting). H.C.: Writing—Review & Editing (supporting). N.K.: Funding Acquisition (equal), Writing—Review & Editing (supporting). J.B.: Funding Acquisition (equal), Writing—Review & Editing (supporting). M.K.: Funding acquisition (equal), Writing—Review & Editing (supporting). M.T.: Funding acquisition (equal), Writing—Review & Editing (supporting). A.A.-K.: Conceptualization (lead), Funding acquisition (equal), Supervision (equal), Writing—Original draft (supporting), Writing—Review & Editing (supporting).

References