Neuroimaging and Cognitive Function in Sickle Cell Disease: A Systematic Review

Sickle cell disease (SCD) is the most common inherited single-gene disease. Complications include chronic anaemia, reduced oxygen-carrying capability, and cerebral vasculopathy, resulting in silent cerebral infarction, stroke, and cognitive dysfunction with impairments in measures of executive function, attention, reasoning, language, memory, and IQ. This systematic review aims to investigate the association between neuroimaging findings and cognition in children with SCD. Searches of PubMed and Embase were conducted in March 2022. Studies were included if participants were <18 years, if original data were published in English between 1960 and 2022, if any genotype of SCD was included, and if the relationship between cognition and neuroimaging was examined. Exclusion criteria included case studies, editorials, and reviews. Quality was assessed using the Critical Appraisal Skills Programme Case Control Checklist. A total of 303 articles were retrieved; 33 met the eligibility criteria. The presence of overt or silent strokes, elevated blood flow velocities, abnormal functional connectivity, and decreased fMRI activation were associated with neuropsychological deficits in children with SCD when compared to controls. There is a critical need to address the disease manifestations of SCD early, as damage appears to begin at a young age. Most studies were cross-sectional, restricting the interpretation of the directionality of relationships. Future research employing longitudinal neuroimaging and neuropsychological assessments could improve our understanding of the cumulative consequences of SCD on the developing brain.


Introduction
Sickle cell disease (SCD) refers to a group of genetic disorders characterised by misshapen red blood cells caused by the polymerisation of abnormal haemoglobin in hypoxic conditions, with approximately 300,000 babies born each year worldwide suffering from these disorders [1,2]. SCD-related complications result in anaemia, cerebrovascular disease [3], brain infarction [4], ischaemia-reperfusion injury [5,6], and increased risk of cognitive impairments [7] (Figure 1). Research has suggested most individuals with SCD experience cognitive impairments as a result of overt strokes or clinically silent cerebral infarctions (SCIs) [8]. For instance, specific impairments in writing and reading abilities have been observed in children with SCD who have experienced strokes [9], and receptive and expressive language developmental delays are frequent [10]. Studies have also shown that children with SCD and SCIs performed worse on tests of vocabulary, coordination, visual-motor speed, and arithmetic when compared to children with SCD with no imaging abnormalities [11,12]. The most common SCD genotype is haemoglobin SS (HbSS) [13]. The relationship between cerebrovascular complications and cognitive function ( Figure 2) can be investigated using transcranial Doppler ultrasonography (TCD), and, in addition   resting cerebral blood flow (CBF) reduces the BOLD response due to a diminished capability to increase blood flow in response to an increase in brain activity [41].
Given the importance of cerebrovascular disease and brain structure and function in cognition and the contributing role of neuroimaging in diagnosing and identifying those at greater risk for complications in SCD, the purpose of this review was to examine how quantitative imaging modalities relate to cognition in children with SCD, identify measures that relate to neuropsychological functioning, and ascertain key regions for future studies. Further insight into the pathophysiology of SCD will also allow for the implementation of different targets for intervention ( Figure 1).   Vaso-occlusion arises when sickled red blood cells obstruct blood flow to the extent that tissues are oxygen-deprived [14]. In response, an inflammatory reaction is triggered, which may lead to the narrowing of vessels, including intracranial and extracranial arteries ( Figure 2). This increases the risk of vascular nitric oxide release, endothelial activation, and adherence of red and white cells, platelets and microparticles, especially as contemporaneous reduction of protein S and C shifts the blood towards a more prothrombotic condition [15][16][17], increasing the risk of arterial, as well as venous, thrombosis [18].
Inclusion and exclusion criteria for articles were created and are summarised in Table 2. References of excluded articles were examined to identify any appropriate articles for inclusion. The search was conducted from 1960 to 2022 to provide a complete and exhaustive list of the neuroimaging research in paediatric SCD. No limits were applied for foreign articles, but studies were only included if they had been translated into English due to time constraints. collected neuroimaging data and cognitive performance in SCD. The remaining papers' full texts were obtained, and eligibility was assessed using inclusion and exclusion criteria ( Table 2).

Quality Assessment
To assess quality, the Critical Appraisal Skills Programme Case Control Checklist was used. Articles were evaluated, which included questions on the suitability of controls (data from normative databases, community/sibling control) and the validity of neuropsychological assessments and neuroimaging used in articles of interest. The checklist included 3 categories with quality scores of 0 (No) or 1 (Yes), from which a total score was calculated. Articles were classified into good quality (66% or higher), satisfactory quality (36-65%), and poor quality (0-35%).

Results
The search retrieved 303 articles. A total of 68 duplicates were removed, and 235 articles were assessed by title and abstract. A total of 165 articles were excluded as they did not meet the eligibility criteria. Full texts of 70 articles were analysed, of which 33 met the inclusion criteria. The PRISMA flow diagram ( Figure 2) outlines the specific selection process.

Data Extraction
The methodological characteristics of included studies presented by imaging modality are summarised in Table 3  Quality appraisal scores for each paper are summarised in Table 4. Thirteen articles (39.3%) were of good quality, and 20 articles (60.6%) were of satisfactory quality.

Characteristics of Articles in Review
Most studies used cross-sectional designs (n = 29; 87.8%), with two longitudinal studies and two retrospective cohort studies. Publication years ranged from 1993 to 2022, with seven (21.2%) from the 1990s, fourteen (42.4%) from the 2000s and twelve (36.3%) from 2011-2022. Articles' sample sizes varied from 14 to 373, with mean ages spanning TCD conditional 10%; abnormal 5% Association between impaired language, specifically syntactical and possibly phonological processing, and elevated TCD velocities, but no effect on semantics or language or attention.   Found that children with silent and overt strokes scored lower on tests of executive function and sustained attention when compared to children with normal MRI scans.
There were no statistically significant differences on tests of academic achievement, visuomotor skills, or global intellectual function; however, participants with overt stroke did perform worse than children with silent infarcts and normal MRI. No significant relationship was found between level of neuropsychological performance and MRI abnormalities. There was a relationship between increased variability in performance on assessments and imaging anomalies (i.e., perfusion anomalies in five cases, structural brain anomalies in two cases and both perfusion and structural anomalies in three cases).     Found the CC was reduced for children with overt or silent infarctions compared to those without visible strokes and controls. The size of the CC was associated with working memory, speed production, and distractibility.  Found an association between performance on the verbal subtest of the WASI and BOLD signal amplitude within the visual cortex. Patients with SCA who exhibited lowered fMRI activation also had lower WASI scores.

Study Outcomes
A. TCD CBFV measured using TCD was investigated in 11 (33.3%) studies of infants, toddlers, preschoolers, and school-age children and adolescents with SCD (Table 3 (A)). Hogan et al. [43] and Schatz et al. [44] reported that in 9-month-old infants with SCD and in 26-month-old children with SCD, high CBF velocities were related to a moderate to high risk of developmental delay when compared to controls. Although narrowing of the intracranial arteries plays a role, the increase in CBF velocity is mainly related to the anaemia and may affect brain development over a longer period. However, Aygun et al. [45] found no relationship between TCD velocities and scores on an academic screening measure in children with SCD (mean age of 3.5 years). Differing family and environmental factors may partially explain these contradictory TCD findings.
Sanchez et al. [46] reported that higher TCD velocities were negatively related to phonological processing and syntactical ability in children with HbSS genotypes of SCD. Kral et al. [49] found that children with abnormal velocities showed greater verbal intelligence impairments than those with lower conditional TCD values. Conversely, Kral et al. [50] reported most subjects (n = 22) displayed elevated TCD velocities and found a positive relationship between better verbal memory and raised TCD values. The abnormal TCD group's performance in both studies may have been influenced by the chronic blood transfusions patients underwent. Prussien et al. [52] found participants not receiving chronic transfusion with higher TCD velocity had poorer performance on tests of executive function and perceptual reasoning.
Onofri et al. [51] and Strouse et al. [53] included younger participants (mean age 8-9 years) and did not find any relationship between cognition and TCD velocities. Similarly, Hijmans et al. [48] found no correlation between neuropsychological measures (including sustained attention, IQ, and inhibition) and TCD values. Hijmans et al. reported children with asymmetries in CBF scored higher on 8 of 13 neuropsychological assessments than those without asymmetries. The researchers found only one test of sustained attention statistically significant; children with right-left asymmetries had lower mean reaction times than those without asymmetries [48]. Lastly, Bernaudin et al. [47] found that elevated TCD measures were related to poorer neuropsychological outcomes and lower IQ scores in children with SCD. However, there were no significant differences between the children with abnormal or normal TCD values after the patients with stroke were excluded.
Seven (33.3%) studies did not find a significant association between MRI abnormalities and impaired cognition [48,51,[57][58][59]63,65] (Table 3 (B)). Grueneich et al. [57] suggested these null findings were related to MRI abnormalities being associated with greater variability in cognition. The authors posited that children with pathopsychological changes develop a pattern of strengthening certain cognitive abilities to accommodate for weak abilities that have been negatively impacted by these changes. This results in uneven neuropsychological profiles while maintaining an overall level of functioning within normal limits on assessments [57]. Hijmans et al. [48] and Montanaro et al. [59] found there were no significant differences in neuropsychological function and intelligence between children with SCI on MRI and children with normal-appearing MRI. The discrepancy between these results and previous studies may be due to confounding variables (e.g., socio-cultural) as well as technical MRI differences. Onofri et al. [51] found no relationship between impaired IQ and SCI, despite 40% of the patients presenting with SCI on MRI. However, participants were younger than in the other studies.
C. Structural MRI: volume (Table 3 (C)) Chen et al. [67] reported that, compared to their counterparts in the high-IQ group, children with SCD in the low-IQ group had reduced grey matter volume in the frontal, parietal, and temporal lobes (Table 3 (C)). Schatz and Buzan [68] found that the size of the CC was associated with working memory, speed of production, and distractibility. D. Structural MRI: diffusion tensor imaging for microstructure (Table 3 (D)) Scantlebury et al. [69] found the structural integrity of WM pathways was affected in children with SCD and reported increased apparent diffusion coefficient values in several brain areas associated with deficits in processing speed and working memory. Stotesbury et al. [70] found widespread WM anomalies were significantly correlated with slower processing speed, even when there was no evidence of infarct on MRI.
Colombatti et al. [71] reported that patients with SCD exhibited greater connectivity in the precuneus than controls, which was pronounced in children with poorer neuropsychological functioning. The connectivity within the default-mode network may be reduced in patients with severe cognitive deficits, as it is involved in speculative processes such as planning and memory. Zou et al. [41] found children with SCD had diminished BOLD responses within the visual cortex during black-and-white visual stimulation tests compared to non-SCD children with posterior fossa tumours, which was associated with lower scores on a test of IQ [41].

Discussion
This systematic review found that children with SCD showed poorer cognition when compared to controls. The following characteristics related to poorer neuropsychological functioning in children with SCD: elevated blood flow velocities, presence of overt stroke or SCI, decreased fMRI activation, and abnormal functional connectivity. The consistent finding of neuropsychological deficits is notable considering the variability across articles in neuroimaging modalities, patient criteria for inclusion, and cognitive assessments. The discrepant findings reported by some MRI studies [57] might be explained by differences in their study samples and methodology. For example, some studies excluded participants based on chronic transfusion [48], others had small samples [51], and some included mixed immigrant cohorts from various socio-economic backgrounds including bilingual or multilingual speakers [48,59], making generalisation more difficult.
Findings indicated that haemoglobin may be a marker for decreased oxygen delivery and that reduced oxygen saturation in children with SCD may indicate cerebral hypoxia, resulting in neuropsychological deficits. Hijmans et al. [48] reported haemoglobin was a significant predictor of verbal memory and that anaemia was a greater predictor of neuropsychological impairments than SCI on MRI. Similarly, Steen et al. [28] reported that severity of chronic anaemia accounted for 23% of the variance in full-scale IQ in children with SCD without strokes.
A review of the research on cognition and TCD velocities in SCD also revealed mixed findings. For example, Kral et al. [50] found that children with SCD with the highest TCD velocities receiving chronic blood transfusions had better verbal memory than those with moderately high velocities who remained untreated. These results imply that blood transfusions may affect cognition by improving oxygen saturation and haemoglobin. Conversely, some TCD studies had null findings but provided limited information related to potential confounding variables (e.g., medications and blood transfusion), which limits the ability to interpret neuropsychological results [45,48,51,53].
Research has also shown children with SCD have higher resting CBF velocity than controls [72]. Therefore, as CBF velocities in these patients may be at maximum during resting states, the need for increased CBF during executive function and working memory processes in the prefrontal cortex may go unfulfilled [52]. According to interactive specialisation theory, executive function development requires interconnectivity between multiple brain areas [73]. This theory may be especially relevant to SCD. Even in the absence of stroke, blood flow and WM integrity are impaired in children. WM pathways and sufficient blood flow are critical for integrating information between different brain areas in circuits connected to executive functioning [70,73,74]. Increasing global CBF may be a risk factor and a response to cerebral hypoxia; however, these compensatory strategies may be inadequate when there is a further demand for an increase in CBF, raising the risk of cognitive decline.
Few studies combine resting-state fMRI with task-based results, partly because the assumptions used to calculate the BOLD response may not be accurate in anaemia; despite this, differences between children may still be clinically significant. Zou et al. [41] reported that children with SCD had reduced visual cortical (V1) activity, which is associated with lower IQ [41]. Because of the small sample, these results must be carefully interpreted; however, patients with SCD not receiving a disease-modifying treatment exhibited no V1 activity under the same stimuli and poorer IQ scores, suggesting an association between neuropsychological deficits and untreated SCD.
Increases in the mean diffusivity of several brain areas, indicating damage to myelin integrity and neural microstructures, are correlated with poor processing speed in patients with SCD [69,70]. Recently, Chai et al. [74] speculated that the maintenance of CBF to the GM is prioritised over WM to maintain the health of neurons essential for survival, as infarcts in GM are immediately devastating. In contrast, infarcts in the WM are 'silent' because they merely inhibit fast information processing. However, WM strokes can substantially hinder crucial aspects of the day-to-day lives of patients with SCD, even if they do not result in major motor impairments.

Methodological Issues and Future Directions
There are several methodological issues in the SCD literature. First, apart from two longitudinal studies, all articles investigating neuroimaging and cognition were cross-sectional, limiting any interpretation of the directionality of relationships. Moreover, because of the challenges of determining when an SCI has occurred, few studies can explore the role of time since cerebrovascular injury is a potential mediating variable related to neuropsychological impairments in children. Studies including neuroimaging rarely assess other possible contributors such as sleep apnoea and numerous school absences on cognition. Research shows environmental factors, in addition to biological risks, are significantly related to executive function and IQ in SCD [22]. Children with SCD frequently grow up in low socio-economic status households, and factors including home environment, parental education, and family income are associated with neuropsychological functioning [22]. More studies with larger sample sizes are required to evaluate the mechanisms and risk factors for neuropsychological deficits in SCD in the presence and absence of overt stroke and SCI. For these reasons, it is crucial to differentiate between socio-economic and psychological aspects, the indirect impact of chronic illness, and the disease-related effect on brain function [75]. This necessitates the use of both neuropsychological studies and longitudinal neuroimaging, as well as clinical and demographic data.

Clinical Implications
This review aims to raise awareness of the need for more research focusing on the mediating role of environmental/psychosocial factors in the relationship between neuroimaging and neuropsychological deficits in SCD. In clinical settings, objective, standardised assessments should be given to children, caregivers, and teachers to ensure multiple sources of information are collected, followed by carefully formulating targeted interventions for children with SCD and their families. This information is critical in identifying vulnerable children with SCD and families who require additional support. The paediatric neurocognitive interventions model is a comprehensive framework that could aid in the development of appropriate treatments for children with SCD [76]. The model emphasises the importance of a developmental, multi-dimensional treatment that focuses on behaviour, emotions, and cognition. The foundation of this approach involves tackling the psychosocial and systemic requirements of children. Therapies are individually tailored to identify neuropsychological impairments, progressing from externally supported compensatory tactics to self-directed compensatory approaches. In children with SCD, there is a critical need for holistic treatments that address foundational needs that are not yet met due to living with a chronic illness (e.g., mental health, social isolation, hospitalisations), as well as targeted interventions for cognitive deficits.
A significant problem within this patient group is that decreased cognitive functioning may have a negative impact on medication adherence, self-care, and clinical attendance, particularly as children with SCD get older [77], with executive function deficits predicting hydroxyurea non-adherence [78]; clinician monitoring of adherence is, therefore, essential [79]. Moreover, recall barriers (forgetfulness) were identified by Fogarty et al. [80] as the biggest challenge of medication adherence for Irish teenagers, who are, however, amenable to the use of mobile apps to monitor, be notified of, or obtain information on SCD treatments [80].

Limitations
There are some limitations to this review. Non-English language published studies were not reviewed, and only two databases were searched. One reviewer carried out the review and quality appraisal. As with other reviews of retrospectively published findings, the results may be biased as significant correlations are more likely to be reported than nonsignificant relationships [81]. Another limitation of this review is the exclusion of other MRI measures, including cerebral blood flow and oxygen extraction fraction (OEF), and their relationship with cognitive function, recently reviewed by Ramos et al. [27]. Lastly, although articles used standardised neuropsychological assessments, they differed significantly, making direct comparisons challenging. Therefore, it was difficult to determine how certain neuroimaging findings were differentially related to neuropsychological domains.

Conclusions
This review highlights the prevalence of cerebrovascular disease in children with SCD and finds that more research is needed as neuroimaging findings do not fully explain cognitive deficits. Results suggest macro-and microscopic brain damage occurring due to insufficient blood supply and energy-intensive processes, with decreased blood flow leading to atrophy of cortical areas and neuronal cell death. This review also emphasises the need to address the disease manifestations of SCD at a young age. Finally, this review indicates there is great promise in utilising neuroradiological markers for early risk stratification so that children with SCD at the highest risk for cognitive deficits can be targeted early for interventions. With successful early risk stratification, interventions and treatments can also be tailored to specific patterns of strengths and weaknesses.