The Development of Neuroimaging Biomarkers for Cognitive Decline in Sickle Cell Disease

SCD is an autosomal recessive hemoglobinopathy caused by a mutation in the beta globin gene that results in hemoglobin with a decreased affinity for oxygen and a propensity for polymerization in the setting of hypoxia.¹ Sickled red blood cells flow abnormally through the vasculature, obstruct the microcirculation, and hemolyze, with resultant inflammation and endothelial activation.² While this physiologic process impacts all organ systems, patients with SCD are at risk for a host of neurologic complications, including vasculopathy, overt hemorrhagic or ischemic stroke, silent cerebral infarcts (SCIs), and cognitive dysfunction, all of which result in great morbidity.

Cognitive Dysfunction in the Absence of Stroke

People with SCD, even in the absence of vasculopathy or stroke, have impaired cognitive function in comparison to people without SCD.^3–5 The effects of SCD on global intelligence in the absence of stroke or SCI were demonstrated in a cross-sectional study performed by Steen et al. who found a significant decrease in full-scale, verbal and performance intelligence quotient (FSIQ, VIQ, and PIQ, all with p < 0.01) in children with hemoglobin SS (HbSS) disease and normal brain MRIs compared to non-sickle cell controls that were matched for age, sex and race. Mean FSIQ, VIQ, and PIQ were 90.8, 92.4, and 91.0, respectively, in non-sickle cell controls in comparison to 77.9, 80.3, and 79.3 in children with sickle cell anemia (SCA) and normal brain MRIs, with an average 12-point decrease in each IQ measure. Similar findings were demonstrated when the authors controlled for parental education as a surrogate for socioeconomic status.⁴ To support these cross-sectional data, a meta-analysis performed by Schatz et al. compared cognitive testing in children with SCD and normal brain MRIs versus healthy, demographically-matched or sibling-matched controls. While the included studies varied in control population and definition of infarction (MRI versus clinical history), the meta-analysis found a mean 4 to 5-point decrease in IQ among SCD patients without silent or overt stroke in comparison to controls. Additionally, 10 of the 14 studies included in the meta-analysis found deficits within the specific cognitive subdomains of attention and executive function, verbal and language function, or memory.³ All together, these data demonstrate the detrimental effects of SCD on cognition even in the absence of stroke, and stress the need for early intervention to mitigate neurocognitive sequelae.

The neurodevelopmental effects of SCD are often identified within the first years of life^6–9 and worsen with age.^6,7,10,11 In a cohort of 80 children with SCD at an average age of less than 2 years, mean cognitive performance per the Bayley Scales of Infant Development was 1 SD below the population mean, with 17.5% having significant neurodevelopmental delay.⁹ Longitudinally, infants with SCD that were evaluated at 3, 9, and 12 months of life using the Bayley Infant Neurodevelopmental Screener (BINS, a screening tool that assesses neurologic status, cognitive function and language with lower scores suggestive of higher risk for neurodevelopmental delay) had higher risk BINS scores in comparison to healthy controls. Of note, the prevalence of high-risk scores increased between 3 to 9 months of age, and higher risk scores correlated with higher TCD velocities.⁷ Relatedly, infants and toddlers with SCD that were evaluated at 6, 12, 24, and 36 months of age were found to have a significant decrease in cognition between 12 and 24 months⁶ supporting that deficits are present early in life with an increased prevalence as children age.

Not only does the prevalence of cognitive dysfunction increase over time, but the cognitive deficits associated with SCD accumulate and worsen with age.^3,4,10,12 While children with SCD and normal brain MRIs in the Cooperative Study of Sickle Cell Disease (CSSCD) cohort did not have a significant decline in FSIQ or PIQ, VIQ scores decreased by approximately 0.5 points per year on the Weschler Intelligence Scale for Children (WISC). Furthermore, these children were found to have a significant decline in coding performance, a measure of visual motor speed and coordination, by 0.2 points per year (p <0.01) and mathematics by 0.9 points per year (p <0.01),¹⁰ suggestive of the cumulative cognitive effects of this disease in the absence of any cerebrovascular accidents. The Silent Cerebral Infarct Multi-Center Clinical Trial (SIT) corroborated these longitudinal analyses through the multivariate modeling of full-scale IQ in 150 children with SCD (107 children with SCIs) showing an approximate decline of 1 point per year of life while controlling for hemoglobin oxygen saturation, presence of SCI, household income and parental education.¹² As expected with prior literature showing cognitive deficits that first manifest in childhood and progress over time, Vichinsky et al. reported that the Weschler Adult Intelligence Scale-III (WASI-III) performance IQ score in adults with SCD without history of stroke (silent or overt) or end-organ damage was significantly lower than that of healthy controls without SCD. Additionally, adults with SCD were found to have marked deficits in attention and processing speed.⁵

Stroke Further Exacerbates Cognitive Dysfunction Associated with SCD

Children with SCD are at increased risk of both overt stroke and SCI, defined as ischemic lesions visualized on neuroimaging in the absence of clinical history or neurologic deficits associated with stroke. Historically, overt stroke impacted approximately 9% of patients with sickle cell anemia by the age of 20 years.^13–15 Implementation of screening with transcranial Doppler ultrasound (TCD) and chronic transfusion therapy (CTT)^16,17 has decreased the risk of overt stroke to 1.9%.¹⁸ However, children with SCD remain at high risk for SCIs as TCD is not an effective screening tool for this neurologic complication.^18,19 SCIs occur in very young children, with 13.7% of patients having a SCI by 2 years of age,^18,20,21 32.4–37% by 14 years of age, 39% by 18 years of age and extending into young adulthood without plateau.^18,22,23

The cognitive implications of overt stroke and SCIs have been widely examined, with several studies demonstrating that children with SCD affected by stroke have poorer global intelligence.^{7,10,15,24–26} Armstrong et al. found that children with SCD and a history of overt stroke had significantly lower FSIQ and PIQ than those with a history of SCI, and that those with SCIs had significantly lower VIQ (with a strong trend in FSIQ) in comparison to those with SCD and normal brain MRIs.²⁴ Similar findings were demonstrated by Hogan et al., who compiled data from numerous studies and found that ethnically-matched non-sickle cell controls had the highest mean FSIQ, followed by those with SCD and no evidence of stroke, then those with SCD and MRI evidence of SCI, and lastly those with SCD and a history of overt stroke.²⁷ The CSSCD examined a cohort of pediatric SCD patients over a 10-year time period with standardized neuropsychiatric testing and brain MRIs, permitting cross-sectional comparisons of cognitive outcomes between patients with overt stroke, SCIs, and normal MRIs. The CSSCD found that those with SCIs had significantly lower FSIQ and VIQ scores in comparison to those with normal MRIs. At an average age of 12 years, the average FSIQ score for those with evidence of SCI was 77.2 in comparison to 84.8 in children with SCD and normal brain imaging, both of which are over a standard deviation below the expected normative average.¹⁰

While measures of IQ provide a global cognitive assessment, such comprehensive measurements may be less sensitive to the impact of SCD than evaluating specific cognitive subdomains. Infarction has been associated with specific deficits in vocabulary,^15,24 verbal comprehension,¹⁵ processing speed,^26,28,29 visual motor speed and coordination (coding),^24,28 attention,²⁶ and executive function.²⁶ Thus, the cognitive sequelae of SCD are likely underrepresented when examining measures of global intelligence alone.

Imaging Biomarkers for the Neurologic Complications of SCD

The two most common imaging modalities studied in SCD are TCD and MRI. TCD is currently used to identify patients with sickle cell anemia at highest risk for cerebral infarction, and relatedly, those who would benefit from initiation of CTT for primary stroke prevention.^16,17 While TCD continues to be used as a safe, non-invasive, and relatively low-cost means of identifying those at highest risk of overt stroke, it has several limitations. First, TCD is unable to accurately predict susceptibility to SCIs.^18,19,22 Wang et al. specifically explored the relationship between TCD and MRI results in a cross-sectional cohort study, reporting that only 29% of the participants with SCIs had conditional or abnormal TCD results. The authors concluded that TCD and MRI findings are often discordant with regard to SCIs and there is a need to develop additional biomarkers for early neurologic complications of SCD.¹⁹ Furthermore, patients with SCD that receive care per current guidelines for monitoring with TCD and initiation of chronic transfusion therapy continue develop SCIs.^18,22 Lastly, overt strokes still occur in those with TCD velocities <200 cm/second, and alternatively, do not always occur in patients with TCD velocities >200 cm/second.¹⁶

Guidelines from the American Society of Hematology also recommend a structural brain MRI in school age children to screen for the presence of SCIs,³⁰ potentially prompting consideration for escalation of primary disease modification with CTT for secondary stroke prevention if infarction was identified.³¹ Aside from recommended developmental screening,³⁰ laboratory or imaging biomarkers do not currently exist that are specific to the risk of cognitive dysfunction in patients with SCD. Thus, recent studies have focused on the utilization of advanced neuroimaging techniques to better understand the pathophysiology of cognitive dysfunction in patients with SCD, with the ultimate goal of developing alternative neuroimaging biomarkers to be used in risk prediction algorithms and to assess the efficacy of treatment options for patients with SCD. The remainder of this review focuses on introducing different physiologic metrics that can be quantified with MRI, and the utilization of these advanced MRI sequences to understand the impact of SCD on brain development.

Cerebral Blood Flow and Cerebrovascular Reserve

MRI can be used to measure cerebral blood flow (CBF, Table 1), a measure of arterial blood delivery to brain parenchymal tissue. Cerebral oxygen delivery is the product of CBF and the oxygen content of the blood (CaO₂, Table 1). CBF increases in the setting of anemia, or reduced CaO₂, to maintain adequate cerebral oxygen delivery.^32–39 As a result, the brain is left with reduced ability to respond to additional hemodynamic stressors, given that its vasculature is nearing maximal vasodilation. Cerebrovascular reactivity (CVR, Table 1), defined as the remaining vasodilatory capacity in response to vasoactive challenge, is consequently reduced in the setting of elevated CBF.³⁹

Table 1.

MR Metrics Evaluating Cerebral Pathophysiology in Sickle Cell Disease

Category	MR-measured metric	Definition	Typical response in sickle cell disease
Hemodynamic Measures	Cerebral Blood Flow (mL/100g/min)	Arterial blood delivery to brain parenchymal tissue	Cerebral blood flow increases in the setting of anemia, or reduced CaO2, to maintain adequate cerebral oxygen delivery.32–39
	Cerebral Oxygen Delivery	Product of cerebral blood flow and the oxygen content of the blood (CaO2)
	Cerebral Vascular Reactivity	Increase in blood flow in response to vasodilatory challenge	Cerebral vascular reactivity is reduced in the setting of elevated cerebral blood flow.39
	Oxygen Extraction Fraction (%)	The fraction of oxygen that the brain parenchyma extracts from the blood	Oxygen extraction fraction increases in people with sickle cell disease,44,63–67 and regional elevation of oxygen extraction fraction co-localizes with regions of the brain at highest risk for microstructural impairment and stroke.44,67 Utilizing a TRUST sequence, there are also reports of unchanged69 and decreased54,68 oxygen extraction fraction in people with sickle cell disease.
Microstructure	Diffusion Tensor Imaging	MRI sequence that characterizes the diffusion properties of water molecules	As water diffusion is typically restricted within the direction of healthy myelin, increased diffusivity and reduced anisotropy suggest damage if water diffusion is not contained within typical confines. Increased diffusivity and reduced anisotropy have been appreciated across different brain regions in people with sickle cell disease.77–83
	Mean/Radial/Axial Diffusivity (10−3mm2s−1)	The restriction of water diffusion
	Fractional Anisotropy	The directionality of water diffusion
Function	Functional connectivity MRI	Directly measures cerebral hemodynamics and indirectly assesses neural activity by correlating the BOLD MR signal between brain regions	Altered functional connectivity has been reported in patients with sickle cell disease compared to healthy controls, evaluating both specific regions and large-scale functional networks with a variety of analysis methodologies.95–102
	Resting-state Functional Networks	Temporally associated fluctuations in the BOLD signal between functionally-related regions of the brain

Resting cerebral blood flow is increased^34,39–45 and cerebrovascular reactivity is decreased^39,42,45 in patients with SCD compared to healthy age-matched controls. While not different between patients with hemoglobin SS versus hemoglobin S beta thalassemia zero (HbSβ⁰) disease,⁴⁶ CBF is significantly higher in those with HbSS and HbSβ⁰ disease compared to patients with hemoglobin SC or S beta thalassemia plus disease.³⁹ Various studies have reported age,^34,45 hemoglobin or oxygen content,^{34,39,42,45,47} hemoglobin F,^39,48 and grey matter volume³⁴ to be predictors of CBF. However, the relationship between CBF and ongoing hemolysis, as measured by biomarkers such as LDH, reticulocyte count and bilirubin, remains unclear.^39,45 The relationship between CBF and these known risk factors for the neurologic complications of SCD in addition to Kim et al.’s report of an association between decreased cerebrovascular reactivity and global cortical thinning in children with SCD⁴⁹ suggest that chronically elevated CBF may impact the brain’s structural development. As such, CBF and CVR have potential for imaging biomarkers of cognitive risk prediction.

SCIs most commonly occur in the border zone regions of the brain where CBF nadirs.^44,50,51 In the analysis of infarct density maps that were created from the SIT trial, Ford et al. demonstrated increased SCI density in the deep white matter of the frontal and parietal lobes that corresponded with regions that had the lowest CBF.⁵⁰ These findings suggest that 1) white matter potentially does not have the same compensatory vasodilatory mechanisms as the remainder of the brain and 2) it is these regions in which SCIs, a neurologic complication that is associated with cognitive dysfunction in patients with SCD, disproportionately occur.

The effects of elevated CBF on cognitive function in SCD were evaluated by Prussien et al., who performed brain MRIs on 54 children and adults (6–31 years of age) with SCD. On multivariate analysis, hematocrit was positively correlated with fluid cognition, processing speed, and inhibitory control, and CBF was inversely related to working memory and inhibitory control (Table 2).⁵²

Table 2.

Investigations Relating MRI Biomarkers to Cognitive Outcomes in Patients with Sickle Cell Disease

Study	Cohort	Age (Years)	Overt Stroke	Silent Infarct	Chronic Transfusion	CBF	OEF	DTI	Functional Connectivity	Cognitive Tests Obtained
Prussien et al. 202152	49 HbSS 5 HbSB0	SCD 6–17 HC 18–31	5 (9.3%)	20 (37.0%)	17 (31.5%)	pCASL	TRUST	-	-	Wechsler Working Memory Index NIHTB - CB
Scantlebury et al. 201176	11 HbSS 3 HbSC 1 HbSD 10 HC	SCD 11.7 HC 10.2	0	5 (33.3%)	Not stated	-	-	ADC	-	WISC/WAIS
Sun et al. 201278	30 HbSS 15 HC	SCD 15.2* HC 15.1	0	0	Not Stated	-	-	FA MD RD LD MO EAR	-	WASI D-KEFS CCPT
Kawadler et al. 201579	25 HbSS 14 HC	SCD 13.1 HC 13.7	0	0	2 (8%)	-	-	FA MD AD RD	-	WASI D-KEFS BRIEF
Stotesbury et al. 201884	82 HbSS 1 HbSB0 32 HC	SCD -SCI 14.6 SCD +SCI 16.3 HC 15.3	0	37 (45%)	5 (6%)	-	-	FA MD RD ICVF ODI	-	WASI/WISC/WAIS
Chai et al. 202181	26 SCD 19 ACTL 21 HC	SCD 24.2 ACTL 26.1 HC 22.6	0	SCD 14 (53.8%) ACTL 8 (42.1%) HC 7 (33.3%)	SCD 4 (15.4%) ACTL 14 (73.7%)	-	-	FA	-	WISC/WAIS D-KEFS
Colombatti et al. 201695	39 HbSS 1 HbSB0 16 HC	SCD 8.1 HC 10.0	0	20 (50%)	9 (22.5%)	-	-	-	DMN	WISC/WPSSI
Sun et al. 201796	10 HbbSS 10 HC	SCD 15.3 HC 14.5	0	0	0	-	-	-	DMN	WASI/WISC D-KEFS CCPT Tower of London BRIEF WRAML Children’s Memory Scale
Coloigner et al. 201798	17 HbSS 2 HbSB0 1 HbSC 12 ACTL 19 HC	SCD 19.7 ACTL 20.8 HC 22.1	0	SCD 4 (20%) ACTL 1 (8.3%)	12 (60%) ACTL 11 (91.7%)	PC	-	-	ALFF	WASI/WISC/WAIS D-KEFS CVLT REY-O BASC2
Fields et al. 2020101	35 HbSS 5 HbSB0 20 HC	SCD 10.0 HC 11.5	0	15 (39.5%)	0	-	ASE	-	SM Auditory Visual Cerebellum DAN VAN Memory DMN Salience CO FP Subcortical	WASI NIHTB-CB

Amplitude of low frequency flucturations (ALFF), Non-sickle cell anemic control participants (ACTL), Apparent diffusion coefficient (ADC), Axial diffusivity (AD), Behavior Assessment System for Children (BASC), Behavior Rating Inventory of Executive Function (BRIEF), California Verbal Learning Test (CVLT), Cingulo-Opercular Network (CO), Conners’ Continuous Performance Test (CCPT), Dorsal Attention Network (DAN), Default mode network (DMN), Delis-Kaplan Executive Functioning System (DKEFS), Diffusion weighted imaging (DWI, Fractional anisotropy (FA), Fronto-Parietal Network (FP), Healthy control (HC), Hemoglobin (Hb), Hb NIH Toolbox Cognitive Battery (NIHTB-CB), Longitudinal diffusivity (LD), Mean diffusivity (MD), Phase contrast (PC), Radial diffusivity (RD), Rey Osterrieth Complex Figure Test (Rey-O), Sensory Motor Netowrk (SM), Sickle cell disease (SCD), Ventral Attention Network (VAN), Wechsler Abbreviated Scale of Intelligence (WASI), Wechsler Adult Intelligence Scales (WAIS), Wechsler Intelligence Scale for Children (WISC), Wechsler Preschool Scale Intelligence (WPSSI), Wide Range Assessment of Memory and Learning (WRAML)

^*Mean age for SCD cohort with normal structural MRI (N = 15)

Different MRI sequences can be used to acquire measurements of CBF, such as phase contrast (PC) MRI and arterial spin labeling (ASL) MRI. Several groups have evaluated sequence parameters that can be utilized to more accurately quantify CBF in the setting of SCD. In-vivo measured or SCD cohort-specific blood relaxation times (T1), utilization of a two-compartment versus single compartment model, and consideration of the labeling efficiency can be used to improve CBF quantification.^53–55 While this review does not focus on the technical specifics of each imaging biomarker, accurate and non-invasive measurement of CBF will be required for successful utilization of this neuroimaging biomarker in risk prediction algorithms.

Oxygen Extraction Fraction

In the setting of significant anemia with a resultant decrease in CaO₂ and/or compromised cerebral perfusion, CBF increases via autoregulatory arteriolar dilation. Furthermore, oxygen extraction fraction (OEF; the fraction of oxygen that the brain parenchyma extracts from the blood, Table 1) can be altered to maintain the cerebral metabolic rate of oxygen utilization (CMRO₂).^56–58 Ischemic infarction occurs if CBF and OEF are inadequate to maintain CMRO₂.⁵⁹ Thus, prognostic markers of cerebral hemo-metabolic disequilibrium, such as OEF, can potentially be utilized to gain insight into tissue-level hemodynamic relationships and may also predict neurocognitive complications in SCD.

Historically, OEF was primarily measured via ₁₅O positron emission tomography, but newer methods utilizing MRI have allowed for OEF to be quantified more readily and with less risk to patients.^60–62 Children and adults with SCD have higher OEF in comparison to healthy controls.^44,63–67 Elevated OEF has been correlated to the severity of SCD, as there is a negative relationship between OEF and hemoglobin.^44,64 Jordan et al. demonstrated that OEF was highest in adult SCD patients with a greater degree of clinical impairment, as defined by the extent of vasculopathy, prior overt stroke, and/or chronic SCD-related pain necessitating routine blood transfusions.⁶³ Moreover, regions of higher OEF co-localize with regions of the brain at greatest risk for SCI.^44,67 Relatedly, Wang et al. demonstrated a spatial relationship between increased OEF and elevated mean diffusivity in adults with SCD, suggestive of the potential role of tissue-level metabolic stress in the underlying microstructural injury.⁶⁶

Utilizing MRI, OEF can be quantified at either a regional tissue level with an asymmetric spin echo sequence (ASE; voxel-wise measurement of OEF-dependent MR measurements of T2) ^60,61 or a summed whole brain level with sampling of a large draining vein by T2-relaxation-under-spin-tagging (TRUST) sequence.⁶² The T2 signal measured with TRUST is sensitive to both hematocrit and oxygenation. If hemoglobin and T2 are quantified in-vivo, oxygenation can be calculated via a predetermined calibration curve. However, the most appropriate calibration curve for patients with SCD remains unclear. In contrast with aforementioned studies, OEF was found to be decreased in participants with SCD in comparison to healthy controls when utilizing a hemoglobin S -specific calibration curve^54,68 while there was not a significant difference in OEF between participants with SCD and healthy controls when using an individualized calibration curve per participant.⁶⁹ With the TRUST sequence, a single T2 measurement is obtained in the superior sagittal sinus, providing a global OEF measurement. The presence of microvascular cerebral shunting in the setting of increased transit time that prevents hemoglobin from fully offloading oxygen⁷⁰ could potentially explain a lower OEF in patients with SCD as measured by TRUST, which is supported by Juttukonda et al.’s results showing a lower OEF in SCA patients with venous hyperintense signals on ASL due to increased transit time compared to those without venous hyperintensities.⁶⁵ All together, these studies highlight the need for future investigations to determine how these different quantification methods of OEF can be used to complement each other.

Chronic transfusion therapy is widely accepted as a means to mitigate stroke risk in sickle cell disease.^17,31 Guilliams et al. reported that both CBF and OEF decreased following exchange transfusion in children with SCD.⁶⁴ The effects of transfusion on cerebral hemodynamics were further evidenced in a study conducted by Juttukonda et al., which showed that adults with SCD had significantly lower OEF following transfusion (with greater reductions in OEF appreciated following simple transfusion in comparison to exchange transfusion), whereas children with SCD had reductions in both CBF and OEF following exchange transfusion.⁷¹ Hood et al. reported higher executive abilities immediately following transfusion in patients with SCD.⁷² All together, these results suggest that transfusion decreases the risk of neurocognitive complications by mediating tissue-level cerebral metabolic stress.

Further solidifying that impaired tissue-level cerebral hemodynamics are implicated in the cognitive dysfunction appreciated in SCD, Prussien et al. reported that increased OEF was associated with decreased processing speed after controlling for age, income, and history of silent or overt infarction (Table 2).⁵² These findings highlight the potential use of OEF as a biomarker for cognitive dysfunction independent of its utility for stroke risk prediction.

Diffusion Tensor Imaging

Diffusion-tensor imaging (DTI) is an MRI sequence that characterizes the diffusion properties of water molecules and provides insight into white matter microstructure. The utility of this imaging modality is based on the premise that various forms of tissue damage result in the degradation of the cellular barriers that typically function to restrict water diffusion.^73,74 As a result, the degree and directionality in which water is able to diffuse through various regions of the brain can be suggestive of underlying microvascular and microstructural injury.⁷⁴ Commonly used DTI metrics include mean diffusivity (MD), radial diffusivity (RD), axial diffusivity (AD), and the apparent diffusion coefficient (ADC), which measure the extent to which water diffusion is restricted, as well as fractional anisotropy (FA), which measures the directionality of water diffusion (Table 1).^74,75 Given that white matter injury is common in SCD, DTI could be useful in evaluating microstructural changes in otherwise normal appearing white matter that can predispose to further neurologic sequelae and result in subsequent cognitive dysfunction.

Increased diffusivity (MD, RD, and ADC) and reduced anisotropy (FA) are suggestive of microstructural white matter injury, and have been appreciated across various regions of the brain in children and adults with SCD.^76–82 The seven studies published thus far utilizing DTI to compare white matter microstructure in children with SCD compared to healthy controls are limited by small sample sizes (with 30 or less SCD participants) and utilize different analysis techniques (Table 3).

Table 3.

Investigations Utilizing Diffusion Tensor Imaging in Patients with Sickle Cell Disease

Study	Participants	Age (Years)	Overt Stroke	Silent Infarct	Chronic Transfusion	DTI Metrics	Impacted Regions (SCD without infarct vs. HC)	Impacted Regions (SCD vs. HC)	Impacted Regions (SCD with vs. without infarct)
Scantlebury et al. 201176	11 HbSS 3 HbSC 1 HbSD 10 HC	SCD 11.7 HC 10.2	0	5 (33.3%)	Not stated	ADC	Frontal lobe Cerebellum	Frontal lobe Cerebellum Pons Vermis	No differences identified
Balci et al. 201277	16 HbSS 14 HC	SCD 25.4 HC 26.1	0	6 (37.5%)	Not stated	ADC FA Fiber Count	-	CC segment I, IV Frontal lobe Centrum semiovale Periventricular WM Brainstem Caudate Thalamus Pons	CC segment IV Centrum semiovale Superior frontal white matter Pons Occipital Temporal white matter
Sun et al. 201278	30 HbSS 15 HC	SCD 15.2* HC 15.1	0	0	0	EAR FA MD RD LD MO	Corpus callosum Centrum ovale	Corpus callosum Centrum ovale	Corpus callosum*
Kawadler et al. 201579	25 HbSS 14 HC	SCD 13.1 HC 13.7	0	0	2 (8%)	FA MD	Peduncles Cerebellum Frontal lobe Parietal lobe CC Subcortex Internal capsule ATR CST	-	-
Kapustin et al. 201980	27 HbSS 12 HC	SCD -HU 15.3 SCD +HU 13.2 HC 14.0	0	15 (55.6%)	0	MD	-	-	-
Chai et al. 202081	26 SCD 19 ACTL 21 HC	SCD 24.2 ACTL 26.1 HC 22.6	0	SCD 14 (53.8%) ACTL 8 (42.1%) HC 7 (33.3%)	SCD 4 (15.4%) ACTL 14 (73.7%)	FA	CC CST ILF IFO SLF UNC	CC CST ILF IFO SLF UNC	SLF
Costa et al. 202182	28 SCD 26 HC	SCD 14.5 HC 15.0	11 (39.3%)	Not stated	17 (60.7%)	FA MD RD AD	-	All white matter	-

Non-sickle cell anemic control participants (ACTL), Apparent diffusion coefficient (ADC), Axial diffusivity (AD), Anterior thalamic radiations (ATR), Corpus callosum (CC), Cortical spinal tract (CST), Fractional anisotropy (FA), Healthy control (HC), Hemoglobin (Hb), Inferior fronto-occipital fasciculus (IFO), Inferior longitudinal fasciculus (ILF), Longitudinal diffusivity (LD), Mean diffusivity (MD), Radial diffusivity (RD), Sickle cell disease (SCD), Superior longitudinal fasciculus (SLF), White matter (WM), Uncinate (UNC)

^*Mean age for SCD cohort with normal structural MRI (N = 15)

^**SCD with mild gliosis versus SCD with a normal structural MR

Microstructural white matter injury is also present in SCD in the absence of SCI (Table 3).^76–79,81 While all of Kawadler et al.’s analyses were limited to children with SCD unaffected by SCI,⁷⁹ Scantlebury et al. showed that increased regional ADC persisted in subgroup analyses comparing only chidren with SCD without SCI against healthy controls.⁷⁶ Utilizing two different analysis techniques, Sun et al. reported increase diffusivity in the left centrum ovale with tract-based spatial statistics and increased MD in the corpus callosum with regional analyses in children with SCD unaffected by SCI compared to healthy controls.⁷⁸ Moreover, the severity of SCD has been positively correlated with the extent of white matter injury (Table 3).^77,78,81,83 These findings were expanded upon more recently by Jacob et al. in a larger cohort in Tanzania, which not only compared children with SCD with and without SCIs, but also showed that patients with SCD and mild/moderate vasculopathy had significantly reduced FA across widespread areas (particularly within the anterior regions of the brain) in comparison to those without vascular abnormalities.⁸³

While the physiology of white matter injury in SCD in the absence of SCI remains unknown, associations have been described between markers of disease severity and disrupted white matter integrity. In a cohort of children with SCD that had no radiographical evidence of SCI, Kawadler at al. observed a significant negative relationship between RD and oxygen saturation (p <0.05) in the genu of the corpus callosum, with a trend toward a negative relationship between RD in the midbody and posterior of the corpus callosum and hemoglobin (p <0.1).⁷⁹

Decreased white matter microstructural integrity has been associated with cognitive deficits in patients with SCD (Table 2).^76,81,84 Scantlebury et al. reported a significant negative correlation between processing speed and ADC in the right frontal lobe in all participants (SCD and control participants), but this relationship ceased to exist when the cohort was included in the multivariate model predicting processing speed.⁷⁶ In a study performed by Stotesbury et al., decreased FA and increased MD within specific regions were significantly associated with decreased processing speed in patients with SCD.⁸⁴ Most recently, Chai et al. investigated the relationship between FA in 11 different white matter tracts and three cognitive subdomains: processing speed, working memory and executive function. They reported that a significant association between FA within the corpus callosum and processing speed, working memory and executive function measures, all of which are known to be impacted in patients with SCD.⁸¹ All together, these studies highlight the potential utility of DTI as a neuroimaging biomarker to evaluate for microstructural changes that are not identified by standard-of-care screening modalities, especially in the white matter where the majority of SCIs occur,^44,50 and are suggestive of a relationship between structural connectivity and clinical cognitive outcomes in patients with SCD that warrant further investigation.

Functional Connectivity

Functional connectivity MRI (fcMRI) is a non-invasive imaging modality that facilitates the exploration of neuronal connectivity via the blood-oxygen-level-dependent (BOLD) MR signal, which is dependent on the paramagnetic properties of deoxyhemoglobin. While neural activity increases local oxygen consumption, there is also a regional increase in cerebral blood flow with a resultant change in the ratio of oxy- to deoxyhemoglbin and BOLD signal. Hence, the BOLD signal directly measures cerebral hemodynamics while providing an indirect measure of neural activity (Table 1).⁸⁵

Various brain regions have known fluctuations in the BOLD signal at rest that are temporally associated with functionally-related regions of the brain and have been described as resting-state functional networks (RSNs, Table 1).^85–88 These functional networks have been examined at rest and during administration of cognitive tasks to define the associated neurocognitive function for each RSN. The default mode network (DMN), comprised of the posterior cingulate, precuneus, ventromedial prefrontal cortex, and the inferior-lateral parietal cortices, is one of the most widely studied RSNs.^88,89 The DMN represents a baseline or ‘default’ mode of brain function that is activated during internally-focused tasks and deactivated during externally-focused task-oriented behaviors.^88,90 In addition to the DMN, there are primary RSNs associated with neurocognitive functions such as sensation and movement, vision, hearing, memory, attention and executive function.^88,91 Disrupted functional connectivity of these networks has been implicated in altered cognition in other disease entities,^92,93 and can potentially be utilized as a biomarker for the neurocognitive complications of SCD.

Children and adults with SCD have alterations in functional connectivity compared to healthy controls.^94–101 Whole-brain analysis of the amplitude of low frequency fluctuations (ALFF) in BOLD signaling has demonstrated that SCD patients have significant differences in the insula, precuneus, anterior cingulate cortex, and medial superior frontal gyrus in comparison to healthy age and race-matched controls.⁹⁸ The impact of SCD on functional connectivity within the DMN remains unclear with some finding diminished connectivity,^94,96,97 unchanged connectivity¹⁰¹ and increased connectivity^94,95,97 in children with SCD compared to healthy controls. Additonal large-scale RSNs have been reported to be impacted by SCD, including the sensory motor,^97,101 auditory,¹⁰¹ executive control,⁹⁴ and salience networks.^94,97,101 While these studies report differences, there can be discrepant findings. For example, Case et al. reported increased connectivity within parts of the salience networks⁹⁴ while Fields et al. reported diminished connectivity¹⁰¹ within this specific RSN. Such differences are most likely secondary to discrepancies between study samples and analysis techniques. Lastly, some investigators have focused on connectivity involving specific brain regions versus large-scale networks, as evidenced by Bhatt et al. reporting greater connectivity between the left locus coeruleus and the left dorsolateral prefrontal cortex in patients with SCD,⁹⁹ and Karafin et al reporting changes in connectivity involving the peraqueductal gray.¹⁰⁰

Several recent studies have aimed to examine the implication of physiologic parameters on functional connectivity in patients with SCD.^95,96,98,102 Hemoglobin was associated with similarity in resting connectivity patterns between patients with SCD and healthy controls as assessed with a graphical lasso model¹⁰² and significantly modulated activation of the DMN in SCD.⁹⁶ Alterations in functional connectivity as measured by ALFF persisted after accounting for the degree of anemia.⁹⁸ Additionally, diminished functional connectivity is associated with increased metabolic stress, as demonstrated in Fields et al.’s report that nodes included within large-scaled RSNs with diminished connectivity in participants with SCD co-localized with regions of elevated white matter OEF.¹⁰¹

Importantly, recent investigations have shown associations between altered functional connectivity and neurocognitive abilities in patients with SCD (Table 2). In examining regional intensities of spontaneous fluctuations, decreased ALFF within the frontal lobe correlated with reduced verbal fluency.⁹⁸ Attenuated activation of the DMN was appreciated in a pediatric SCD cohort with significantly lower scores in measures of executive function, auditory attention, and working memory in comparison to healthy age, ethnicity, and background-matched controls.⁹⁶ Conversely, increased activity within the posterior precuneus of the DMN has been observed in SCD patients with decreased IQ, which the authors hypothesized to represent DMN recruitment as a compensatory strategy in the setting of mild cognitive impairment.⁹⁵ Disrupted functional connectivity is present prior to the onset of significant changes in IQ,^96,101 suggestive of the use of fcMRI as an early neuroimaging biomarker for cognitive decline.

Conclusion

Patients with SCD suffer from a host of neurologic complications of their disease, including cognitive dysfunction in the absence of concomitant stroke. While the pathophysiology of these cognitive deficits requires further investigation, recent studies utilizing MRI to evaluate metabolism, structure and function of the brain have provided insight into the impact of this disease on brain development. Patients with SCD experience cerebral hemodynamic adaptations to compensate for severe anemia, which are associated with microstructural and functional alterations in the brain. Future investigations will be required to explore the relationship between these different imaging biomarkers, and to determine if these metabolic, structural and functional changes in the brains of patients with SCD are related to long-term cognitive outcomes.

Glossary

AD

Axial diffusivity

ADC

Apparent diffusion coefficient

ALFF

Amplitutde of low frequency fluctuations

ASL

Arterial spin labeling

ASE

Asymmetric spin echo

BINS

Bayley Infant Neurodevelopmental Screener

BOLD

BBlood-oxygen-level-dependnet MR signal

CaO₂

Arterial oxygen content

CBF

Cerebral blood flow

CMRO₂

Cerebral metabolic rate of oxygen utilization

CVR

Cerebrovascular reactivity

CTT

Chronic transfusion therapy

CSSCD

Cooperative Study of Sickle Cell Disease

DMN

Default mode network

DTI

Diffusion tensor imaging

FA

Fractional anisotropy

fcMRI

Functional connectivity MRI

FSIQ

Full scale intelligence quotient

MD

Mean diffusivity

MRI

Magnetic resonance imaging

OEF

Oxygen extraction fraction

PC

Phase contrast

PIQ

Performance intelligence quotient

RD

Radial diffusivity

RSN

Resting state network

SCA

Sickle cell anemia

SCD

Sickle cell disease

SCI

Silent cerebral infarct

SIT

Silent Cerebral Infarct Multi-Center Clinical Trial

TCD

Transcranial Doppler ultrasound

TRUST

T2-relaxation-under-spin-tagging

VIQ

Verbal intelligence quotient

WASI

Weschler Adult Intelliggence Scale

WISC

Weschler Intelligence Scale for Children