A cross-sectional, case-control study of intracranial arterial wall thickness and complete blood count measures in sickle cell disease

Summary

In individuals with sickle cell disease (SCD), tissue oxygen delivery is sensitively dependent on abilities to increase cerebral blood flow and volume through relaxation of smooth muscle lining intracranial arteries. We hypothesized that anemia extent and/or circulating markers of inflammation lead to concentric macrovascular arterial wall thickening, visible on intracranial vessel wall imaging (VWI) MRI. Adult and pediatric SCD (n=69; age=19.9±8.6 years) and age- and sex-matched control (n=38; age =22.2±8.9 years) participants underwent 3 Tesla VWI; two raters measured basilar and bilateral supraclinoid internal carotid artery (ICA) wall thickness independently. Mean wall thickness was compared with demographic, cerebrovascular, and hematological variables (significance: two-sided p<0.05). Mean vessel wall thickness was elevated (p<0.001) in SCD (1.07±0.19 mm) compared to control (0.97±0.07 mm) participants after controlling for age and sex. Vessel wall thickness was higher in participants on chronic transfusion therapy (p=0.013). No significant relationship between vessel wall thickness and (i) flow velocity, (ii) hematocrit, (iii) white blood cell count, or (iv) platelet count was observed, however trends (p<0.10) for wall thickness increasing with decreasing hematocrit and increasing white blood cell count were noted. Findings are discussed in the context of how anemia and circulating inflammatory markers may impact arterial wall morphology.

Introduction

Sickle cell disease (SCD) is a genetic disorder that leads to sickling and premature erythrocyte hemolysis, causing decreased oxygen carrying capacity of arterial blood (1, 2). Individuals with SCD often do not exhibit conventional radiographic risk factors for stroke. For instance, only 16% of individuals with SCD meet typical radiological criteria for cervical or intracranial vasculopathy on magnetic resonance angiography (MRA) (3), yet approximately 64% of individuals demonstrate evidence of vessel tortuosity (3) and approximately 50% of adults with SCD will have an overt stroke or silent cerebral infarct by age 30 years (4, 5). As such, cerebral ischemia in individuals with SCD is likely a result of alternative vascular or metabolic changes at the tissue level.

More specifically, hemodynamic compensation is most frequently provided through increases in cerebral blood flow (ml blood / 100g tissue / minute) and volume (ml blood / 100g tissue) (6–10) via autoregulatory changes in microvascular cerebrovascular reactivity. Cerebrovascular reactivity and associated reserve capacity can be interrogated directly by measuring blood flow and volume during administration of pharmacological (i.e., acetazolamide) or respiratory (i.e., hypercapnia) stimuli (6–8), and these protocols have been successfully applied in individuals with SCD to elucidate reduced reserve capacity. Furthermore, cerebrovascular reactivity has been demonstrated in patients with non-atherosclerotic intracranial stenosis to reduce with increasing intracranial arterial wall thickening (11), yet whether arterial wall thickening is present in individuals with SCD has not been well studied.

Multiple mechanisms could underlie arterial wall thickening in SCD, yet these are largely unexplored. For instance, high flow velocity and potential intravascular turbulence may lead to elevated pulsatile pressure, in addition to arterial wall shear stress or oxidative stress (12–16). Additionally, arterial aging is often thought of in terms of atherogenic contributors, yet this process depends sensitively on multiple cellular and molecular factors, often inflammatory in nature (17). In support of this premise, coronary artery ectasia has been found to be present in pediatric and young adult individuals with SCD, and such ectasia was directly associated with white blood cell count (18), an indirect marker of inflammation. Ongoing hemolysis with resultant endothelial activation could potentially contribute to arterial wall changes, which could have clinical impact by altering cerebrovascular reactivity or could provide a marker of hemodynamic stress.

As such, it is logical that effects from anemia manifest as morphological remodeling of the arterial vessel wall; in vivo measurements of intracranial vessel wall thickness are now becoming possible and more frequently implemented using vessel wall imaging (VWI) magnetic resonance imaging (MRI) (19).

The purpose of this study is to measure intracranial arterial wall thickness using 3-Tesla VWI MRI in adult and pediatric participants with SCD and compare values to those obtained from healthy age-matched controls, as well as to (i) flow velocities in the same vessels and (ii) hematological variables obtained from complete blood cell counts. The overarching hypothesis is that intracranial arterial vessel wall thickness is elevated in participants with SCD, and the degree of elevation scales with anemia extent.

Methods

Participants

Adult and pediatric participants with SCD and age- and sex-matched healthy individuals provided informed, written consent in accordance with the local institutional review board and were enrolled for this prospective study.

Participants with SCD were recruited from a local sickle cell clinic continuously between 2014 and 2019. All had either hemoglobin-SS (HbSS) or hemoglobin-Sβ0-thalassemia (HbSβ0) phenotype and no 3T MRI contraindications (dental braces, ferromagnetic or paramagnetic implantable devices that were 3T MRI conditional or prohibitive at the required specific absorption ratio, intracranial clips, or a magnetic foreign body in the eyes). No potential participants failed enrollment due to any of these contraindications, with the exception of one potential participant with dental braces. Prior stroke and treatment (i.e., regular blood transfusion, hydroxyurea, or no treatment) were not exclusion criteria, however treatments were recorded and participants on vs. off regular blood transfusion therapies were considered separately.

Healthy controls were required to be normotensive for age and height with normal low-density lipoprotein levels (defined as < 130 mg/dL). Exclusion criteria for healthy controls: major neurological or psychological diagnosis (e.g., Parkinsonism, Alzheimer’s disease, schizophrenia, bipolar disorder), prior stroke or transient ischemic attack, myocardial infarction, diabetes, chronic smoker, body mass index > 40 kg/m², and greater than 50% stenosis or occlusion of any major cervical artery or first segment of the anterior, middle, or posterior cerebral arteries.

Hematologic measures

Hematocrit, hemoglobin-S percentage (HbS%), white blood cell (WBC) count, red blood cell (RBC) count, platelet count, and red cell distribution width (RDW) were measured in all participants with SCD. In transfusion participants, measurements were made on the day of the scan. In non-transfusion participants, measurements were made within one week of the scan. When unavailable in the electronic medical record, the hemoglobin phenotype was assessed using high performance liquid chromatography.

Magnetic resonance imaging and angiography

All participants received non-contrast 3T brain MRI and MRA with standard anatomical imaging including VWI (3D turbo-spin-echo [TSE] factor = 56; TR/TE = 1500/33 ms; refocusing angle sweep = 40°–120°; in-plane spatial resolution = 0.5 × 0.5 mm) (20, 21), T₂-weighted (3D TSE factor = 15; TR/TE = 4000/83 ms; refocusing angle sweep = 40°–120°; spatial resolution = 0.5 × 0.5 × 2.4 mm), fluid-attenuated inversion recovery (FLAIR; turbo inversion recovery; spatial resolution = 0.6 × 0.6 × 3.0 mm; TR/TI/TE = 11,000/2800/120 ms), diffusion weighted imaging (DWI; single-shot echo planar imaging; spatial resolution = 1.8 × 1.8 × 4 mm; b-values of 0 and 1000 s/mm²), and intracranial time-of-flight MR angiography (3D gradient echo; spatial resolution = 0.4 × 0.4 × 1.4 mm³; TR/TE = 23/3.5 ms). To evaluate whether vessel wall thickening may be related to flow velocities, we performed phase contrast angiography (PCA) of major anterior and posterior vessels using a previously-reported protocol (22) with in-plane spatial resolution=0.5×0.5 mm², TR/TE=20/7 ms, v_enc=40 cm/s. Data were acquired separately in left and right ICAs and left and right vertebral arteries, with four separate slices planned, each orthogonal to the vessel of interest to improve quantitative accuracy. A v_enc=40 cm/s was applied to allow for determination of slow flow near the perimeter of the vessel, and phase unwrapping was applied to quantify any aliased flow velocities in the center of the vessel.

Analysis

MRI and MRA findings were independently evaluated by two board-certified neuroradiologists. Disagreement was resolved by consensus. Number of participants with intracranial vasculopathy and cerebral infarcts was calculated based on brain MRI and MRA findings. Vasculopathy was defined as > 50% stenosis of the first segment of anterior, middle, or posterior cerebral artery and/or intracranial segment of ICA or basilar artery. Cerebral infarcts were diagnosed as hyperintense on FLAIR and hypointense on T₁-weighted sequence approaching CSF signal.

Vessel wall thickness measurements were performed using OsiriX (Pixmeo, Bernex, Switzerland) independently by one board-certified neuroradiologist and one 3^rd year radiology resident. All VWI images were first reviewed to determine the quality of the study and whether vessel walls were discernable from adjacent soft tissues and CSF for measurements. Images were then reformatted in a plane orthogonal to the vessel course (Fig. 1) at the following locations: supraclinoid ICA (left and right separately) and mid basilar artery between the superior cerebellar artery and anterior inferior cerebellar artery. These segments were the focus of the study due to the higher established reproducibility of vessel wall thickness measurements in these vessels in non-atherosclerotic vascular disease patients at the afforded spatial resolution of 3-T VWI MRI, compared with more distal measurements in the middle cerebral arteries (20, 21). Luminal diameters and outer vessel wall diameters were measured perpendicular to the vessel course using the OsiriX multiplanar reformatting tool. Vessel wall thickness was calculated as one-half the difference of the outer vessel wall and luminal diameters.

Figure 1.

Representative basilar vessel wall measurements from sickle cell disease (SCD) participants with low (patient 1) and high (patient 2) vessel wall thicknesses. For each participant, 3D data were reconstructed along the course of the vessel of interest by two radiologists and measurements were made in the transverse plane

Flow velocities were calculated from PCA data in each of the four vessels using an identical approach as outlined previously (22); magnitude and phase images were converted to temporally-averaged (over the cardiac cycle) velocity images (23). Flow velocities were calculated at approximately the same location as the ICA VWI arterial wall thickness measurements. However, posterior circulation velocities were only performed in the vertebral arteries approximately 10 mm proximal to the confluence of the vertebral and basilar arteries. While expected to be highly correlated, vertebral and basilar artery wall velocities may differ, and therefore for statistical testing we compared only the ICA wall thickness and ICA velocities, which were obtained in the same vessels and locations.

Statistical methods

Standard descriptive statistics including mean, median, and standard deviations were calculated for continuous variables, as well as frequency for categorical variables. Investigations for outliers were performed for hematologic measures using criterion of more than 2.5 standard deviations from the mean.

To test the hypothesis that vessel wall thickness was higher in participants with SCD versus healthy controls, we performed multiple regression using the mean vessel wall thickness (from the ICA and basilar arteries) as the dependent variable and (i) age, (ii) sex, and group status (control or SCD) as independent variables. This comparison included all control (n=38) and SCD (n=69) participants. It is also possible that the ICA and basilar arteries may have a different dependence on the above covariates, and therefore as a secondary analysis we performed two separate regressions with either (i) mean ICA thickness or (ii) basilar artery thickness as the dependent variable and the same independent variables as above. We required a two-sided p<0.05 for significance. For the three comparisons, this yielded a critical Bonferroni-corrected p-value of p<0.0167. Corresponding model significance (F-statistic) and variable coefficients were recorded.

To understand how any hypothesized increase in vessel wall thickness in SCD is associated with flow velocity, anemia extent, or circulating markers of inflammation (WBC or platelet counts), we excluded participants on regular blood transfusion (which will increase variability in most of the hematological markers). This resulted in 50 SCD participants remaining. In these participants, we performed separate calculations of Spearman’s rank correlation coefficient by comparing mean vessel wall thickness with (i) flow velocity, (ii) hematocrit, (iii) WBC count, and (iv) platelet count. Data are presented as scatter plots with associated Spearman’s ρ and two-sided p-values. These separate bivariate analyses were performed only to provide additional information between variables as a secondary analysis; multiple regression was not performed due to concerns with over-fitting in the smaller sample size of 50 participants.

For completeness, we also present summary statistics for controls, SCD participants on regular transfusion, and SCD participants not on regular transfusions. A Wilcoxon signed-rank test was applied to evaluate differences in wall thickness between groups with two-sided p<0.05 required for significance.

Results

69 participants with SCD (age = 19.9 ± 8.6 years, range = 6–39 years; sex = 55.1% female) and 38 age-matched healthy controls (age = 22.2 ± 8.9 years, range = 8–39 years; sex = 50% female) were enrolled. Cohorts were matched for age (p=0.12) and sex (p=0.56). No control participants had evidence of intracranial vasculopathy or infarcts, whereas in the SCD cohort, 13.0% had evidence of intracranial vasculopathy, 49.3% silent cerebral infarcts strokes, and 15.9% prior overt stroke (Table I). 19 SCD participants were on regular transfusion therapy, whereas 50 were not receiving regular transfusions. One participant not receiving transfusions declined hydroxyurea therapy, however all others (49/50; 98%) were on hydroxyurea therapy.

Table I.

Characteristics of the control and sickle cell disease participants. Additional variables for sickle cell disease participants, separated by treatment group, are summarized in Table II.

	Controls	Sickle cell disease
Count	38	69
Age (median), years	22.2±8.9 (21.5)	19.9±8.6 (19.0)
Sex, percent female	50	55
ICA thickness (median), mm	1.00±0.08 (1.00)	1.07±0.19 (1.01)
Basilar thickness (median), mm	0.90±0.10 (0.89)	1.07±0.17 (1.08)
Intracranial thickness (median), mm	0.97±0.07 (0.97)	1.07±0.14 (1.05)
Vasculopathy, percent present	0	13
Silent cerebral infarct, percent present	0	49.3
Overt stroke, percent present	0	15.9
Hemoglobin-S, percent	0	65.8±19.7 (71.1)

The mean vessel wall thickness (across both ICAs and basilar artery) was 1.07 ± 0.19 mm in SCD participants, compared to 0.97 ± 0.07 mm in control participants. Higher arterial wall thickness was present for SCD vs. control participants in both the basilar and ICAs, however the discrepancy was larger for basilar artery (SCD vs. control: 1.07 ± 0.17 mm vs. 0.90 ± 0.10 mm) than ICA (SCD vs. control: 1.07 ± 0.19 mm vs. 1.00 ± 0.08 mm). These differences were evaluated statistically using multiple regression, separately modeling age and sex. On multiple regression, considering the mean vessel wall thickness as the dependent variable and (i) age, (ii) sex, and (iii) group (SCD vs. control) as independent variables, the regression model was significant with F-statistic=6.01 (F-significance = 0.0008): no significant relationship between wall thickness and age (p=0.66) or sex (p=0.78) was found, but a significant relationship was found for mean vessel wall thickness being larger in SCD vs. control participants (p<0.001). For the secondary analysis considering ICA and basilar arteries separately, similarly no significant relationship was found with age or sex; the uncorrected p-values for group were significant for the basilar (p<0.001) and ICA (p=0.048) wall thicknesses. The critical two-sided p-value for Bonferroni correction of the three regressions (considering mean wall thickness, basilar wall thickness, or mean ICA wall thickness), was be 0.017, and therefore the mean overall wall thickness and mean basilar wall thickness values only were significant after multiple comparison correction. These findings were from the cumulative 107 measurements (38 control and 69 SCD) and are summarized in Table II. Representative case examples are shown in Fig. 1.

Table II.

Measurements in sickle cell disease participants on regular transfusion (left) vs. not on regular transfusion (right). All but one of the 50 participants not on regular transfusion was on hydroxyurea. Continuous measures are reported as mean ± standard deviation, with median in parentheses.

	Sickle cell disease (regular transfusion)	Sickle cell disease (non-transfusion)
Count	19	50
Age (median), years	18.4±7.7 (18.3)	20.5±9.0 (19.8)
Sex, percent female	58	54
Intracranial vessel wall thickness (median), mm	1.14±0.16 (1.12)	1.05±0.12 (1.02)
Vasculopathy, percent present	31.6	6
Silent cerebral infarct, percent present	68.4	42.0
Overt stroke, percent present	42.1	6.0
Hemoglobin-S (median), percent	44.7±14.6 (44.8)	73.8±15.0 (76.6)
White blood cell count (median), 103/μL	12.5±3.7 (13.5)	9.4±3.9 (8.7)
Platelet count (median), 103/μL	415.9±197.0 (405)	350.1±166.7 (319.5)
Arterial velocity (median), cm/s	33.8±5.6 (35.4)	34.9±5.9 (35.2)
Red blood cell count (median), 106/μL	3.1±0.5 (3.1)	2.6±0.6 (2.5)
Hematocrit (median), percent	26.5±3.4 (27)	24.9±4.1 (24.8)
Red cell distribution width (median), percent	18.2±2.7 (18.1)	19.5±3.2 (19.1)

Next, we analyzed only the SCD participants not on regular transfusion therapy (n=50) to gain additional information on any possible contributors to the increased vessel wall thickness while reducing confounds from treatment effects. Fig. 2 shows the results of the relationship between mean vessel wall thickness and (A) flow velocity, (B) platelet count, (C) hematocrit, and (D) WBC count. Three participants met outlier criteria for platelet counts and one for WBC count and were excluded from analysis; these participants are reviewed specifically in the Discussion. No comparison yielded a significant relationship on separate bivariate analysis, however while there was no evidence of any trend for a relationship with flow velocity or platelet count, there was a trend (p<0.10) for vessel wall thickness increasing with increasing WBC count and also decreasing hematocrit.

Figure 2.

Separate bivariate analyses of the relationship between mean vessel wall thickness and (A) flow velocity, (B) platelet count, (C) hematocrit, and (D) white blood cell count. No comparison met criteria for statistical significance, however trends (p<0.10) were observed when considering the association between mean vessel wall thickness and white blood cell count and hematocrit.

Table II summarizes descriptive statistics for measurements between transfused and non-transfused participants. The mean vessel wall thickness was statistically higher in the transfused vs. non-transfused cohort (p=0.013). As transfusions are frequently clinically indicated in patients with a history of overt stroke, vasculopathy, or progressive silent cerebral infarcts, the transfused group had higher incidences of each of these factors as expected. The modest sample size of transfused participants (n=19) prevented these effects from being evaluated rigorously, however implications are considered in the Discussion. Group level vessel wall values are summarized in Fig. 3 and multiple regression findings are summarized in Table III.

Figure 3.

Mean vessel wall thickness measurements for healthy control and all SCD participants are shown to left. On the right, SCD participants on regular blood transfusion are shown separately.

Table III.

Results of multiple regression analyses using mean vessel wall thickness across all intracranial vessel wall segments (dependent variable) and (above) all control vs. all SCD participants, (lower) control vs. only SCD participants not receiving regular transfusion. For case-control analyses, a positive coefficient indicates a positive relationship with mean vessel wall thickness.

	Coefficient	Standard Error	t-stat	p-value	Lower 95%	Upper 95%
Healthy controls vs. all sickle cell disease
Group (SCD=1; control=0)	0.0993	0.0242	4.1108	0.0001	0.0514	0.1472
Age (years)	−0.0006	0.0013	−0.4365	0.6634	−0.0032	0.0021
Sex (0=F;1=M)	−0.0063	0.0230	−0.2742	0.7845	−0.0520	0.0394
Healthy controls vs. sickle cell disease (non-transfusion)
Group (SCD=1; control=0)	0.0735	0.0220	3.3393	0.0013	0.0297	0.1173
Age (years)	0.0003	0.0013	0.2377	0.8127	−0.0023	0.0029
Sex (0=F;1=M)	−0.0005	0.0218	−0.0231	0.9816	−0.0439	0.0429

Discussion

Intracranial vessel wall imaging was applied at the standard clinical field strength of 3 Tesla in adults and children with SCD. After controlling for age and sex, significant mean vessel wall thickening of the combined ICAs and basilar arteries was observed in participants with SCD compared to healthy controls. The vessel wall thickness was also highest in the sub-group of participants on regular blood transfusion, who had higher incidence of vasculopathy, silent cerebral infarcts, and overt stroke. Preliminary leads regarding possible relationships between increased vessel wall thickness and (i) anemia extent and (ii) white blood cell count were noted, however these trends did not reach criteria for statistical significance.

These findings can first be considered in light of the growing field of intracranial VWI. VWI can not only identify structural abnormalities such as stenosis, aneurysm and dissection, but also characterize carotid artery plaque composition and identify vulnerable plaque characteristics (24–27). There have been some studies evaluating the use of VWI to measure intracranial vessel wall thickness in vivo in healthy individuals, which demonstrated good reproducibility of VWI to measure supraclinoid ICA and basilar artery wall thicknesses with intraclass correlation coefficients 0.83–0.94 and no significant differences in basilar wall thicknesses due to age or sex in early adulthood (20, 21, 28). These prior studies are consistent with the current study in which no significant dependence of arterial wall thickness on age or sex was observed.

Limited studies have been performed to evaluate endothelial thickness and function in individuals with SCD. One study evaluated common carotid artery elasticity by using ultrasound to assess the cross-sectional compliance and distensibility of the carotid arteries based on changes in diameters during systole and diastole in a pediatric population (29). The study concluded that children with SCD did not demonstrate significant endothelial dysfunction or change in arterial stiffness. However, comparable impact of vessel wall morphology in adults with SCD has not been investigated, nor the relation of any possible vessel wall changes with flow velocity, anemia extent, or circulating markers of inflammation. As adults have been shown to respond differently to SCD therapies (30), and SCI prevalence increases with age (5), it is logical that disease chronicity may also influence the extent of arterial wall changes, and it may be inappropriate to transpose pediatric findings to adults.

Increased mean intracranial wall thickness (measured over intracranial segments of the ICAs and basilar artery) was observed in participants with SCD compared to controls. Intracranial ICA vessel wall thickness between the SCD and control groups was less different compared to basilar artery segments: the ICA thickness did meet criteria for significance on multiple regression at two-sided p=0.048, but did not meet significance criteria after adjusting for multiple comparisons (critical p-value for three comparisons p=0.017). This less significant finding may be physiological, and close inspection of Table I demonstrates that in SCD participants the vessel wall thickness of all vessel segments was similar (1.07 mm), whereas there was more variation in wall thickness between ICAs (1.00 mm) and basilar arteries (0.90 mm) in controls. Additionally, technical challenges to obtaining accurate measurements in the ICAs may be relevant. For example, more variable CSF signal suppressions in this region has been described (21). However, when a subset (n = 38) of scans were repeated with additional CSF suppression utilizing a delay alternating with nutation for tailored excitation preparation module (DANTE) (31), ICA measurements were not significantly different compared to the scan without the additional CSF suppression.

It has previously been shown that reduced hematocrit is associated with higher cerebral blood flow (32, 33) and also cervical blood flow velocity (6). These findings motivated our hypothesis that due to decreased hematocrit and oxygen carrying capacity, increased blood volume and flow velocities with possibly elevated pulsatile pressure and wall shear stress may lead to morphological vessel wall remodeling ultimately manifesting as largely concentric vessel wall thickening. However, we did not observe any trend or significant relationship between the increased vessel wall thickness in the non-transfused participants and flow velocity measured from phase contrast angiography.

These findings suggest that the vessel wall thickening may be more closely associated with circulating markers of inflammation or anemia extent, and this possibility was separately investigated. In support of this, we recorded standard measures from complete blood cell counts, as well as other hematologic measures including hemoglobin-S percentages and platelet counts. We observed that the vessel wall thickness trended higher (p=0.0975) in individuals who were more anemic (i.e., lower hematocrit) and had some circulating markers of inflammation such as increased WBC count (p=0.0940), however these relationships did not meet criteria for statistical significance. It is well-characterized that elevated WBC count, along with reduced hemoglobin and arterial oxygen saturation, are associated with poor neurological outcomes in individuals with SCD; see (34) and references therein. Elevated white cells may exacerbate SCD symptoms and disease severity by adhering to blood vessel walls (35), with other cells, and cause corresponding steno-occlusion. White cells also stimulate the vascular endothelium to increase expression of adhesion molecule ligands, and are centrally involved in inflammatory reactions (36). Therefore, a direct relationship between vessel wall thickness and WBC count may not be unexpected, however additional participants, likely followed longitudinally (given relatively large variations in WBC count even in otherwise healthy individuals), will be necessary to better understand this relationship.

We observed no trend between platelet count and mean vessel wall thickness. We considered outliers in our dataset as we did observe that several participants with very high platelet counts did bias the correlation analysis, even with the non-parametric test. For example, three outlier participants had very elevated platelets greater than 600 10³/μL. One had a single platelet count of 934 10³/μL but WBC count remained normal and reviewing multiple complete blood cell counts over a two month period revealed only the single elevated platelet count. Another participant had persistently elevated platelets of 600–700 10³/μL over multiple years with normal WBC count; this was either idiopathic or due to splenectomy. A third participant had platelets that were persistently elevated in the range of 500–900 10³/μL. In this case, cause was unclear. Infection was suspected with WBC counts of 9–16 10³/μL, but there was no evidence of fever and no history of splenectomy. Abnormal platelet counts may also indicate more severe sepsis or organ failure, however no participants in this study had evidence of such effects. These findings highlight that single cross-sectional measures of circulating markers may be insufficient. To more fully evaluate any possible relationship between inflammation with vessel wall thickness, longitudinal assessment of WBC and platelet counts, and perhaps other markers of inflammation, would be useful.

We did observe that vessel wall thickness was higher in the transfused vs. non-transfused groups. As transfusion was given per clinical indication and was not a research procedure, these groups fundamentally differed as the transfusion cohort had higher clinical indicators of disease severity including overt stroke, vasculopathy, and silent cerebral infarcts (Table II). Unfortunately it is not possible to evaluate from these cross-sectional data when the stroke or infarct occurred, or whether these infarcts are recent and potentially related to progressive morphological vessel wall changes, or are simply older and unrelated. These data suggest, but do not confirm, that vessel wall thickness may be higher in individuals with more advanced indicators of disease. Vascular regulation and oxygen extraction at the microvascular level may also be impaired due to vessel wall damage that currently cannot be evaluated at the spatial resolution of human vessel wall sequences. Prior work in other populations has however shown that vessel wall disease is related to impaired cerebrovascular reactivity magnitude and timing (11), which is logical in SCD as well and may be the topic of future studies. Vascular reactivity is well known to be largely mediated by autoregulation and relaxation of smooth muscle located within the middle tunica media layer of the arterial wall (37–40). Reduced cerebrovascular reactivity potential and associated autoregulatory capacity is of high clinical significance in SCD owing to well-known dependence of brain health on this compensatory property (41–43), yet whether intracranial vessel wall thickening is present in SCD is not well-established.

Limitations of this study include that intracranial vessel wall thickness is near the spatial resolution of the sequence. However, VWI has been shown to be reproducible in measuring vessel wall thickness of supraclinoid ICAs and basilar arteries in healthy participants with intraclass correlation coefficients of 0.83–0.94 (20, 21), and in this work we focus on these larger vessels rather than smaller vessels (e.g., middle, anterior, and posterior cerebral arteries) where the method may be less accurate. Second, due to the limited number of prior studies using VWI for in vivo intracranial vessel wall thickness measurements, there is no gold standard for measurement guidance or comparison. We hope with this study, we will establish precedence for future studies to reference and compare. Third, the 38 healthy controls were not matched in number to the patient cohort, however they were matched for age and sex and were well-characterized in terms of neuroimaging and absence of cerebrovascular risk factors. In addition, Cogswell et al demonstrated that there was no significant change of basilar wall thickness, at the spatial resolution applied here, across the lifespan in early adulthood of healthy participants (p = 0.45) and there was significant difference in basilar wall thickness between males and females (p = 0.42) (20). Fourth, flow velocity measurements were made from phase contrast angiography rather than transcranial Doppler ultrasound (TCD) which is clinically more common in pediatric SCD patients. However, TCD velocity is frequently applied intracranially at the level of the MCAs (where vessel wall measurements are less reliable), and TCDs are not routinely used in adults with SCD owing to no clinical trial demonstrating efficacy for triaging adult SCD patients for transfusion therapies. Similar phase contrast measurements to those used here have been compared with TCD, and findings suggest similar information (44) with variability between methods frequently attributed to TCD user-dependence and insonation angle variations (45). As PCA generally reports lower velocities compared to TCD (44), quantitative values should not be compared directly. Fifth, our hematological measures were made on a single day, whereas many of these measures will vary based on issues unrelated to SCD (i.e., recent infection). Future studies that monitor ongoing changes in these circulating measures of inflammation and hemolysis would be useful, as well as in sequence with direct measures of arteriolar resistance and reserve capacity using respiratory or pharmacological stimuli.

In conclusion, increased combined mean arterial wall thickness of the ICA and basilar arteries was observed in SCD patients compared to healthy controls. This wall thickness was higher in the participants on regular blood transfusions, which are generally clinically indicated in patients with stroke, vasculopathy, and progressive silent cerebral infarcts. Future studies that perform longitudinal measures of anemia extent and inflammation are warranted to better understand the source of these vessel wall changes, as well as their potential clinical relevance.