DETERMINANTS OF RESTING CEREBRAL BLOOD FLOW IN SICKLE CELL DISEASE

Abstract

Stroke is common in children with sickle cell disease and results from an imbalance in oxygen supply and demand. Cerebral blood flow (CBF) is increased in patients with sickle cell disease to compensate for their anemia, but adequacy of their oxygen delivery has not been systematically demonstrated. This study examined the physiological determinants of CBF in 37 patients with sickle cell disease, 38 ethnicity matched control subjects and 16 patients with anemia of non-sickle origin. Cerebral blood flow was measured using phase contrast MRI of the carotid and vertebral arteries. CBF increased inversely to oxygen content (r² = 0.69, p < 0.0001). Brain oxygen delivery, the product of CBF and oxygen content, was normal in all groups. Brain composition, specifically the relative amounts of grey and white matter, was the next strongest CBF predictor, presumably by influencing cerebral metabolic rate. Grey matter/white matter ratio and CBF declined monotonically until the age of 25 in all subjects, consistent with known maturational changes in brain composition. Further CBF reductions were observed with age in subjects older than 35 years of age, likely reflecting microvascular aging. On multivariate regression, CBF was independent of disease state, hemoglobin S, hemoglobin F, reticulocyte count and cell free hemoglobin, suggesting that it is regulated similarly in patients and control subjects. In conclusion, sickle cell disease patients had sufficient oxygen delivery at rest, but accomplish this only by marked increases in their resting CBF, potentially limiting their ability to further augment flow in response to stress.

INTRODUCTION

Stroke is one of the most devastating complications in sickle cell disease. Prior to routine transcranial Doppler screening, the risk of major stroke was 11% by the age of 18, with a peak age of incidence in the early school age years¹. Aggressive transcranial Doppler (TCD) screening and chronic transfusion protocols have lowered overt stroke rates ten fold^2,3, but silent strokes remain problematic^4,5. A recent review by DeBaun and Kirkham suggests that silent stroke prevalence rises 1.4% per year with no plateau, even in cohorts managed with modern practices⁶.

Chronic anemia is one of the strongest risk factors for stroke and all cause mortality in SCD⁶. Several studies have denoted elevations in cerebral blood flow (CBF) in SCD, presumably to preserve resting oxygen delivery^7–10. However, the adequacy of that compensatory hyperemia has never been assessed or compared to patients with anemias from other etiologies^8,11. In fact, there is a paucity of data exploring CBF in African Americans, in general, despite their twofold risk of cerebrovascular accidents¹².

To address these shortcomings, we compared resting CBF and oxygen delivery in 36 SCD patients, 37 ethnically matched controls, and 16 patients with anemia other than SCD (Thalassemia major, congenital dyserythropoetic anemia, and spherocytosis). We also measured CBF response to 100% oxygen administration.

METHODS

All studies were performed according to Good Clinical Practice and the Declaration of Helsinki. Informed consent was obtained from all patients under a protocol approved by the Committee on Clinical Investigation at the Children’s Hospital Los Angeles.

Three patient cohorts were studied. The first cohort consisted of sickle cell disease subjects 12 years old or older with SS, SB₀, and SC genotypes who were free from known strokes. Patients were excluded if they had a hospitalization within the month prior to the study visit. The second cohort consisted of African and Hispanic Americans older than 12 years of age who had no prior history of neurologic insult or serious, chronic illness requiring daily medications. Most were first or second-degree relatives of the patients studied. The third cohort consisted of patients with chronic anemia, excluding sickle cell anemia, who were otherwise healthy and age-matched to the first cohort. The etiology of anemia consisted of thalassemia major (beta thalassemia or E-beta thalassemia, N=13), congenital dyserythropoietic anemia (N=1), autoimmune hemolytic anemia (N=1), and spherocytosis (N=2). Exclusion criteria were prior neurologic insult or major medical problems outside of their chronic anemia.

At the beginning of the study visit, patients underwent measurement of vital signs and phlebotomy. Complete blood count, reticulocytes, quantitative hemoglobin electrophoresis, lactate dehydrogenase and cell free hemoglobin levels were analyzed in the clinical laboratory.

MRI examinations were performed on a Philips Achieva 3 Tesla magnet using an 8-element head coil. CBF was measured using single slice phase contrast velocity measurements with the following parameters: repetition time 12.3 ms, echo time 7.5 ms, field of view 260×260 mm, slice thickness 5 mm, signal averages 10, acquisition matrix 204 × 201, reconstruction matrix 448 × 448, bandwidth 244 Hz/pixel, and velocity encoding gradient of 200 cm/s. Details of localization and analysis have been previously published, as well as flow data from a subset of the control and anemia control cohorts¹³. Whole brain 3D T1 and T2 weighted images were collected for brain segmentation and volume purposes¹³. MR angiography was also performed to evaluate for large vessel stenoses. T1, T2 and MR angiography images were read for white matter disease and stenosis by a licensed neuroradiologist, blinded to disease status. At the end of the imaging study, CBF was reassessed after patients had been breathing 100% oxygen through a non-rebreathing mask for ten minutes.

Oxygen content was estimated from hemoglobin according to the well known relationship¹⁴:

$$\mathit{Oxygen}\mathit{content}=\mathit{1.34}\times \mathit{Hemoglobin}\times O_{\mathit{2}}\mathit{Saturation}+\mathit{0.003}\times p{O}_{\mathit{2}}$$ (1)

where oxygen saturation was measured by pulse oximeter and pO₂ is the partial pressure of oxygen, which was assumed to be 100 torr on room air. We did not estimate oxygen content on 100% oxygen because arterial blood gases were not performed. Oxygen delivery to the brain can then be calculated as the product of cerebral blood flow and oxygen content, or conversely CBF can be written as follows:

$$\mathit{CBF}=\mathit{Oxygen}\mathit{Delivery}/\mathit{Oxygen}\mathit{Content}$$ (2)

Intuitively, one would then expect CBF to vary inversely with oxygen content. To control for the nonlinearity in equation [2], we log transformed all variables in our regression analyses for CBF and oxygen delivery. Log transformation improved the normality of all the predictive variables. Following log transformation, equation (2) becomes

$$\mathit{Log}(\mathit{CBF})=\mathit{log}(\mathit{Oxygen}\mathit{Delivery})-\mathit{log}(\mathit{Oxygen}\mathit{Content}),$$ (3)

Stepwise linear regression was performed using JMP Pro 11 (SAS, Cary, NC).

RESULTS

Patient demographics are summarized in Table 1. The anemia control group (ACTL), when the two spherocytosis patients were excluded, had hemoglobin levels (10.3 ± 0.9 g/dl) comparable to the SCD patients. Differences in CBC parameters and reticulocytes among the groups were predictable. Cell-free hemoglobin was equally elevated in the SCD and ACTL patients, but LDH was only elevated in the SCD patients. Diastolic blood pressures were low in both SCD and ACTL patients, consistent with decreased systemic vascular resistance found in anemic patients. SCD patients were mildly desaturated compared to the other subjects, but only 3 patients had oxygen saturations below 93%. Body habitus was comparable in all three groups. The SCD patients were predominantly SS patients with 50% receiving chronic transfusion. Because many of the control subjects were family members, the population was roughly balanced between AS and AA. ACTL patients were predominantly AA hemoglobin; 3 E-beta thalassemia patients had AE electrophoresis but their native erythropoiesis was effectively suppressed.

Table 1.

Patient Demographics

	SCD (N=37)	CTL (N=38)	ACTL (N=16)	p
Age (Years)	22.2 ± 8.9	27.0 ± 10.6	23.4 ± 6.7	0.07
Sex	19F, 18M	28F, 10M	10F, 6M
Ethnicity	33 African American 3 Hispanic 1 Middle Eastern	33 African American 4 Hispanic	9 East Asian 1 Italian 3 Subcontinent 4 Hispanic
Hemoglobin (g/dl)	9.7 ± 1.6	13.5 ± 1.3	11.2 ± 2.6	<0.0001
Hematocrit (%)	27.9 ± 4.4	39.8 ± 3.6	32.5 ± 6.6	<0.0001
Mean corpuscular volume ()	85.7 ±10.9	85.2 ± 5.2	83.0 ± 4.6	0.51
Mean corpuscular hemoglobin	29.7 ± 4.3	28.9 ± 2.0	28.4 ± 1.7	0.35
MCHC	34.6 ± 1.1	33.9 ± 1.0	34.3 ± 1.3	0.25
White blood cell count (x 103)	9.9 ± 4.3	5.7 ± 1.7	6.4 ± 2.2	<0.0001
Platelets	286 ± 111	245 ± 59	241 ± 100	0.17
Platelet volume	10.1 ± 0.8	10.6 ± 0.9	10.7 ± 0.9	0.023
Red cell distribution width (%)	19.3 ± 3.9	13.3 ± 1.1	16.5 ± 3.9	<0.0001
Reticulocytes (%)	10.3 ± 6.1	1.4 ± 0.6	1.7 ± 2.2	<0.0001
Lactate dehydrogenase	988 ± 531	525 ± 85	502 ± 319	<0.0001
Free hemoglobin	20.0 ± 21.1	5.4 ± 3.6	22.3 ± 24.6	0.0004
Hemoglobin electrophoresis	29 SS, 14 transfused 4 SC, 0 transfused 4 SB0, 1 transfused		13 AA 3 AE	<0.0001
%A hemoglobin	36.4 ± 33.2	77.8 ± 19.6	93.9 ± 5.5	<0.0001
%S hemoglobin	53.1 ± 30.3	18.5 ± 19.4	0.0 ± 0.0	<0.0001
Systolic blood pressure (mmHg)	112 ± 12	118 ± 12	112 ± 11	0.05
Diastolic blood pressure (mmHg)	61 ± 8	69 ± 10	62 ± 8	0.0017
Heart rate (min−1)	80 ± 12	75 ± 18	79 ± 12	0.26
Respiratory rate (min−1)	18.5 2.5	18.5 2.1	21.7 ± 11.1	0.07
O2 saturation (%)	97.4 ± 2.4 15 < 98% 3 < 93%	99.2 ± 1.1 2 < 98% 0 < 93%	99.1 ± 1.0 1 < 98% 0 < 93%	<0.0001
Height (cm)	164.4 ± 10.8	165.9 ± 7.1	164.7 ± 9.7	0.77
Weight (kg)	65.2 ± 24.6	71.1 ± 19.4	63.6 ± 12.5	0.35
Body mass index (kg/m2)	24.1 ± 8.7	25.8 ± 6.9	23.3 ± 2.8	0.41

MR angiography was abnormal in only one subject, a 37 year old man with sickle cell disease who had mild narrowing of his right internal carotid artery and severe, bilateral proximal anterior cerebral artery stenosis without overt stroke. Small, white matter hyperintensities (silent strokes) greater than expected for age were observed in a total of 16 subjects, 4/38 controls, 1/14 anemia controls, and 11/37 patients with SCD (p<0.05 by ChiSquared analysis).

We first validated the stability of MRI phase contrast measurements in 13 SCD patients by interleaving six CBF assessments between other anatomic and functional imaging series, at roughly six minute intervals. Patient head movement precluded the use of all measurements, thus the number of usable measurements per subject was 5.1 ± 0.5(range 4 to 6, 85% acceptance rate) over a window of 35.5 ± 12.0 (range 22.1 – 61.7) minutes. Coefficient of variation was 4.6 ± 2.5%, comparable to the results we obtained in control subjects¹³.

By analysis of variance, CBF was increased in both SCD (94.0 ± 21.2 ml/100g/min) and ACTL subjects(80.1 ± 16.8 ml/100g/min) compared with control subjects (59.1 ± 9.5 ml/100g/min), p<0.0001. Figure 1 demonstrates that CBF varies inversely with oxygen content, as predicted. There is seamless overlap of all three patient groups. Best fit after log-log transformation and linear fitting to all patients is shown in linear space by the heavy black line (r² = 0.69, p<0.0001). To provide perspective from another imaging modality and different patient group, the light grey lines represent 95% confidence intervals derived using Xenon scintigraphy in older adults without sickle cell disease¹¹; Xenon scintigraphy is a radioactive tracer technique that provides highly accurate estimates of total brain flow¹⁵.

Figure 1. Plot of cerebral blood flow (CBF) versus blood oxygen content for control subjects (solid dots), non-sickle anemia control subjects (x symbols), and sickle cell disease(SCD, open circles).

The solid line is the best linear fit to the log transformed data. The light grey lines are the 95% confidence intervals calculated from a historical cohort of older, nonhemoglobinopathy patients who had a diverse range of hematocrits in whom CBF was measured by Xenon scintigraphy¹¹.

The age of our patients spanned a critical window of brain development. Figure 2A demonstrates the ratio of grey matter to white matter volume as a function of patient age. By using a ratio we demonstrate the maturational effects and avoid interpatient variability in absolute brain volume. There is a sharp decline of relative grey matter volume over adolescence and early adulthood, independent of patient group; this reflects normal synaptic maturation. Using the relationship found in Figure 1, Figure 2B demonstrates the measured CBF as a percentage of CBF predicted by oxygen content alone. The dotted line at 100% indicates when CBF is “matched” to oxygen content to produce a constant oxygen delivery to the brain. The brain is relatively “hyperperfused” until the age of 25, followed by a plateau, and then CBF declines relative to oxygen content after the age of 35 or 40. Children at the age of 11 have 40% greater CBF than patients aged 21 years or older. There is striking concordance between the two representations, suggesting that brain composition, represented by gray white matter ratio, is likely a major determinant of cerebral blood flow.

Figure 2A and 2B. Ratio of grey matter to white matter volume (2A) and relative cerebral blow flow as a function of age (2B) in control (CTL), anemic control (ACTL), and sickle cell disease (SCD) patients.

The relative proportion of grey matter falls sharply during the second decade of life as synapses are pruned during normal maturation; this process continues until around 25 years of age. Brain composition is relatively stable during young adulthood and middle age but further grey matter loss occurs in senescence. The y-axis represents the ratio of observed CBF compared with the CBF predicted by oxygen content along (the solid line from Figure 1). During adolescence and young adulthood, CBF is significantly higher than predicted by oxygen content alone (depicted by the horizontal dashed line at 100%). This increased flow parallels the higher proportion of grey matter observed in Figure 2. As the brain matures to an adult composition, less CBF is needed to feed the metabolically demanding grey matter. CBF also declines in middle aged subjects, disproportionally to changes in brain composition. Other groups have attributed this decline to vascular aging.

To unravel the relative contributions of oxygen content, age and grey matter volume on cerebral blood flow, we performed univariate and multivariate regression, also including demographic factors from Table 1. Final multivariate results are summarized in Table 2, calculated for SCD patients alone, non SCD patients combined (ACTL + CTL) and all subjects. All variables were log-transformed (natural log) prior to inclusion in the model. O₂ content, grey matter volume and age were consistently the strongest predictors of cerebral blood flow for all three groups. LDH was a negative predictor of CBF in CTL and ACTL patients, as we showed previously¹³, but not in SCD patients. MCH and MCV were co-linear (r² = 0.91) and were positively associated with CBF in all groups, although accounting for only a small amount of the variance. Combined r² ranged from 0.68 to 0.78. Importantly, CBF was independent of transfusion status, hemoglobin S%, hemoglobin F%, blood viscosity, and markers of hemolysis. Findings were independent of the presence or absence of white matter hyperintensities. Exclusion of the single patient with an abnormal MRA did not alter the strength or significance of the CBF predictors.

Table 2.

Multivariate Analysis for CBF and Oxygen Delivery

	SCD (N=37)		CTL & ACTL (N=54)		All Subjects (N=91)
CBF	Beta	p	Beta	p	Beta	p
O2 Content	−0.954	0.0007	−.648	<0.0001	−1.00	<0.0001
Grey Matter	0.927	0.0000	0.273	0.0002	0.751	<0.0001
Age	−0.195	0.030	−0.246	0.011	−0.198	0.0005
LDH	~	~	−0.384	0.0054	~	~
MCH (MCV)	0.436	0.014	0.751	0.018	0.431	0.004
Combined r2	0.73		0.68		0.78
Oxygen Delivery	Beta	p	Beta	p	Beta	p
GM/WM ratio	0.5168	0.00032	~	~	0.328	<0.0001
Age	−0.177	0.030	−0.212	0.002	−0.184	0.0004
LDH	~	~	−0.271	0.021	~	~
MCH (MCV)	0.444	0.016	0.574	0.056	0.378	0.001
Combined r2	0.45		0.32		0.33

Oxygen delivery to the brain is simply the product of oxygen content and cerebral blood flow, and was 11.7 ± 2.0 ml O₂/100g/min for SCD, 11.4 ± 1.7 ml O₂/100g/min, for ACTL and 10.5 ± 1.6 ml O₂/100g/min for CTL, p=0.03. The coefficient of variation was quite low (14.9% – 17.1%), indicating tight control over cerebral oxygen delivery. Predictors of oxygen delivery are summarized in Table 2 (bottom half). Not surprisingly, the same predictors of CBF are also important in determining oxygen delivery, with brain composition and age being the strongest overall modulators. Combined r² was smaller, 0.32 – 0.45, reflecting the smaller dynamic range of oxygen delivery compared with cerebral blood flow.

All three populations reacted similarly to the 100% oxygen challenge suggesting a common pattern of response. CBF dropped in most subjects, inversely to the baseline oxygen content (r² = 0.24, p<0.0001) as predicted by equation 2. Patients who had a lower oxygen content exhibited a greater reduction in CBF with 100% oxygen.

DISCUSSION

This is the first paper to systematically explore the determinants of resting cerebral blood flow in sickle cell disease patients. Not surprisingly, oxygen content was the strongest predictor of CBF, similar to previous findings in non-SCD, anemic patients^16,17. The increase in CBF in anemic subjects (both SCD and ACTL) fully restored their resting oxygen delivery to healthy, non-pathophysiological levels. In fact, cerebral oxygen delivery was even higher than observed in control subjects, but this difference disappeared after controlling for differences in patient age and brain composition.

Further support for importance of blood oxygen content in regulating CBF was observed by the CBF reduction produced by oxygen challenge. Placing patients on 100% oxygen increases dissolved oxygen transport, raising blood oxygen content by 1–2 ml O₂/dl blood. With higher oxygen content, one would expect CBF to decline because less blood flow is necessary to meet cerebral metabolic needs. The greater vasoconstriction observed in anemic patients represents the larger importance of dissolved oxygen when less hemoglobin is present.

We did not observe any association of CBF with hemoglobin S concentration, unlike that observed by Hurlet and colleagues¹⁸. However, their study design was fundamentally different, consisting of longitudinal CBF measurements in three patients with prior stroke who were weaned off chronic transfusion therapy. Our study was purely cross-sectional, in well controlled patients during steady state, so we cannot exclude the possibility that CBF correlates with hemoglobin S% in response to major, acute therapeutic changes. Within the chronically transfused patients, though, no trends in CBF were observed over S% ranging from 8.9% to 55.5%. CBF was also not correlated with hemoglobin F% in the nontransfused subjects, independently of total hemoglobin.

The second striking observation was the key importance in brain maturation in determining cerebral blood flow requirements. Grey matter is at least fourfold more metabolically active than white matter, so it is somewhat intuitive that total CBF was highly correlated with grey matter volume^19,20. The grey matter/white matter ratio peaks between 6 and 10 years of age, and declines monotonically until the early twenties through synaptic pruning^21,22. Although there is limited normative data, CBF in normal children appears to follow a similar pattern, rising sharply from two to five years of age and peaking at 6.5 years^23,24. Normal TCD velocities mirror this age distribution, supporting TCD as a CBF surrogate²⁵. Stroke risk in SCD patients also appears to mirror this age distribution¹. First stroke incidence is highest between two and five years of age and stroke rate peaks between six and nine years of age, suggesting that high resting CBF may be the key risk factor for stroke. This hypothesis is concordant with the success of TCD as a biomarker^3,26,27. Elevated TCDs are prognostic for stroke, even in the absence of stenosis, suggesting that high resting CBF is pathologic²⁸.

Why is high resting blood flow associated with stroke? The ability of the brain to augment brain blood flow under stress, known as the cerebrovascular reserve, is limited in SCD^6–8,29,30. The higher the resting CBF, the smaller the cerebrovascular reserve⁷. Using carbon dioxide inhalation as a vasodilatory challenge, Prohovnik and colleagues suggested cerebral perfusion peaks at levels about 150 ml/100g/min in SCD patients⁷. If that is the case, a patient with a resting CBF of 120 ml/100g/min will only be able to augment their oxygen delivery by 25% under stress, compared with 250% for a patient with a resting CBF of 60 ml/100g/min. Thus high resting flow and diminished vascular reserve leaves patients vulnerable to the ischemic strokes commonly found in SCD.

This model for stroke risk is consistent with known risks factors for silent strokes, including acute anemic events and hypoxia, acting to impair oxygen delivery, and fever or sepsis, which increases metabolic demands⁶. These triggers acutely worsen oxygen supply/demand balance, compromising brain regions in watershed areas such as the fronto-parietal white matter. Because of the reciprocal relationship between CBF and oxygen content (Equation 2), a three gram drop in hemoglobin is much worse if the starting hemoglobin is 8 g/dl than if it is 12 gm/dl. For someone with a hemoglobin of 8 gm/dl, CBF would have to increase 60% to preserved oxygen delivery, compared with only 33% if the starting hemoglobin were 12 gm/dl. Chronic transfusion therapy lowers the risk of silent stroke, but does not abolish it^6,31, suggesting more aggressive prophylaxis or treatment of these supply-demand modulators (anemia, nighttime hypoxia, fever) should also be explored.

High resting brain blood flow may also be more than just a biomarker. Systemic pulse pressures widen with worsening anemia³². Brain blood vessels dilate to accept the increased CBF, allowing these widened pressure fluctuations to propagate more distally in the vascular tree³³. Dynamic cerebral autoregulation is impaired in SCD patients, causing fluctuations in blood pressure to be more readily alter cerebral perfusion³⁴. Of note, systolic blood pressure is one of the stronger predictors of silent stroke risk, even with values in the normal range³⁵. Analogies may be drawn to elderly subjects, where pulse pressures steadily widen because of aortic stiffening. Silent strokes develop in a similar distribution as for SCD, with a rate correlated with middle cerebral artery pulsatility³³.

Increased CBF and cerebral blood velocity could also potentially accelerate vascular disease through altered wall shear stress mechanisms³⁶. Adverse vascular remodeling can occur in regions of either abnormally low or abnormally high wall shear stress³⁷. Viscosity is a key mediator between blood velocity and wall shear stress ³⁸. Intrinsic viscosity is increased in SCD subjects but is countered by the powerful blood thinning effect of anemia. Thus the relationship between increased CBF, wall shear stress, and adverse vascular remodeling is complicated and needs further study.

Age-related reductions in cerebral blood flow (Figure 2B) remained significant even after controlling for brain composition, driven by a more rapid rate of decline after 35 years of age. Similar findings have been described using multiple imaging modalities^39–42. Recent work by Lu and colleagues suggests that CBF reductions with age are not the result of declining cerebral metabolic rate, but instead reflect microvascular disease⁴². Changes were not specific to any single patient group but we did not have sufficient power to determine any interaction between patient group and vascular aging.

The association of LDH with cerebral blood flow in the normal population was puzzling. LDH and brain matter composition were weakly associated. LDH is present in all cells and may be acting as surrogate for endothelial turnover or some other nonspecific marker of metabolic rate. Neither LDH, nor any other markers of hemolytic rate, were associated with increased cerebral blood flow or oxygen delivery in the SCD population or entire cohort.

The statistical relationship between MCH(or MCV) and CBF was equally puzzling (albeit weak). The effect was not limited to sickle cell disease and was independent of transfusion status, hydroxyurea use, or hemoglobin subtypes. At this point, we cannot speculate on the physiological significance of this observation and await replication in independent cohorts.

Previously, accurate quantitation of cerebral blood flow could only be performed by nuclear methodologies that were either expensive, invasive, or not widely available. Phase contrast MRI measurements of carotid and vertebral artery flow is fast (~ 1 min), highly reproducible, accurate, and readily available on any MRI scanner in the world. Figure 1 suggests excellent concordance between MRI CBF measurements and those by Xenon inhaled computed tomography; the differences observed at the lower oxygen contents were simply due to the younger ages of our anemic patients (patient ages in the Brown manuscript were 55.6 ± 13.8 years old). Compared to transcranial Doppler, MRI is too expensive for routine screening purposes. However, it is routinely performed in SCD patients having abnormal TCD’s to evaluate baseline parenchymal and angiographic assessments. In the future, phase contrast MRI assessments could easily be added to standard MRI imaging protocols and might ultimately improve the discriminatory ability of TCD velocities, particularly with conditional velocity elevations. Resting phase contrast measurements could potentially also be combined with values collected after vasodilator stress such as acetazolamide to provide quantitative whole brain cerebrovascular reserve estimates.

Our study had some limitations. Our study populations were not well gender matched. However, after correction for sex differences in brain size, no sex differences were observed in CBF or oxygen delivery in any subgroup. Our sickle cell patients were heterogeneous with respect to transfusion status and hemoglobin electrophoresis. While this may decrease our sensitivity to detect some effects on cerebral blood flow, it increases the generalizability of our findings. We did not directly measure the p50 of patient blood which could potentially impact CBF. However, we failed to see any impact of hemoglobin S% and F% (two important modulators of p50) on CBF, nor differences in CBF between transfused and nontransfused SCD patients suggesting a small effect size of p50 on net CBF. Lastly, our study is cross-sectional which provides little information on the prognostic value of resting brain blood flow.

In summary, resting cerebral blood flow in sickle cell disease patients is determined by the same physiological principles as non-sickle cell disease subjects. According to basic supply-demand principles, total CBF increases inversely with hemoglobin levels and directly proportionally to grey matter mass such that resting oxygen delivery is normal. The linear dependence of CBF on grey matter mass explains the vulnerability of young SCD patients for stroke. Age related CBF decline is observed in both SCD and non SCD subjects. Phase contrast MRI assessment of total cerebral blood flow is a fast, easy, accurate and reproducible biomarker that could be added to current MRI imaging protocols in sickle cell disease.