The pathophysiology of stroke in individuals with SCD involves multiple mechanisms.33 The chronology and hierarchy, however, are not well established.34 The two main mechanisms responsible for stroke in individuals affected by the disease are: (1) occlusive vasculopathy characterized by the proliferation of smooth muscle cells and increased fibroblasts in the intima layer of artery walls and, consequently, progressive segmental narrowing of the distal internal carotid artery and proximal branches of the main intracranial arteries (circle of Willis) and (2) sickled RBC aggregation, and consequent occlusion of small vessel lumen.35 Some previously proposed models11,33,35,36 provide an overview of the pathophysiology of cerebral vasculopathy in SCD.

Sickled RBCs firmly adhere to the vascular endothelium of intramural small vessels of the arteries of Willis circle through several RBC-endothelial bridges.37,38 It is believed that the triggering factor in the pathogenesis of stroke in individuals with SCD is the adhesion of sickled RBCs and/or reticulocytes to the vascular endothelium, generating endothelial activation and damage.36 High expressions of endothelial adhesion molecules and RBC adhesion molecules have been observed in individuals with SCD, including integrins,39,40 endothelial selectins,41,42 soluble adhesion molecules,43 and immunoglobulin superfamily members.44 The adhesion of RBCs in the microvasculature causes the entrapment of denser and less deformable cells, decreasing the blood flow, and increasing the capillary transit time.45 This favors further polymerization of Hb S and causes vaso-occlusion.

Adhesion molecules, cytokines, and chemoattractants attract white blood cells (WBCs) to the site of damaged endothelium, causing microvascular obstruction and ischemia.46,47 The activated state of WBCs caused by the chronic inflammatory state characteristic of SCD causes a significant number of WBCs exhibiting high levels of molecules that can bind to the endothelium. Abnormal adhesion of RBCs and WBCs occurs mainly in the post-capillary venules where shear stress is sufficiently low to allow blood cell adhesion to the vascular wall. Cell adhesion to endothelium may also occur in larger arteries. However, it is improbable that the abnormal adhesion of RBCs and WBCs to the endothelium occurs in large cerebral arteries of individuals with SCD, resulting in stroke.11 Alternatively, abnormal adhesion happens in the venules of large arteries and the pathophysiological process occurs from the wall into the lumen of large arteries.

The role of hemolysis is evident in the pathophysiology of CVD.11 Chronic and acute hemolysis results in free hemoglobin that interferes with the nitric oxide (NO) metabolism. The release of arginase derived from hemolysis consumes l-arginine, a substrate for the production of NO. Free Hb, heme, and heme iron catalyze the production of oxygen radicals, potential NO scavengers and endothelium activators.48 The reduction in NO bioavailability and increase of oxygen free radical formation lead to endothelial dysfunction, increasing inflammation and contributing to a hypercoagulable state associated to SCD. Furthermore, reduced NO bioavailability limits smooth muscle relaxation and increases vascular resistance.11,49

Reduced NO bioavailability reduces vasodilation and impairs the inhibition of platelet activation and aggregation mediated by NO, and also inhibits the repression of cell adhesion molecule transcription.49 The role of vascular tone regulation on the pathophysiology of stroke is not clear, but there is evidence supporting its involvement, such as the reduction of plasma free Hb levels and hemolysis secondary to chronic transfusion therapy.50 The co-inheritance of alpha-thalassemia also decreases hemolysis51 and preserves the benefit of higher NO bioavailability.34

Adherent platelets aggregate at the site of endothelial injury, forming a web of cells together with WBCs and RBCs.33,36,52–54 Coagulation system abnormalities have been reported in individuals with SCD, generating a hypercoagulable state. Coagulation and fibrinolysis markers are elevated at steady state and are more elevated during vaso-occlusive crises.55,56 Low concentrations of protein C and S were associated with history of stroke,57 as well as high concentrations of the D-dimer.58 However, the relationship of coagulation and fibrinolysis markers with stroke has not been fully elucidated.

In response to chronic severe anemia, cerebral blood flow and cerebral blood velocity are increased in individuals with SCD.59,60 Any decrease in cerebral blood flow by physiological or pathological reasons leads to a risk of imbalance between demand and supply of oxygen in the brain, increasing the risk of stroke.11 Increased hypoxia and the inability of the brain vasculature to dilate lead to ischemia.61 Furthermore, hypoxia and ischemia may increase the expression of several adhesion molecule receptors in the vascular endothelium of the human brain. Increased transcription of several genes involved in angiogenesis, inflammation, vascular tone regulation, cell proliferation, apoptosis, and coagulation in hypoxic situations could contribute to cerebral vasculopathy.11

After temporary hypoxia followed by re-oxygenation in order to induce reversible sickling, transgenic mice exhibit an excessive inflammatory response characterized by increased WBC adhesion and extravasation in the microvasculature, and evidence of oxidant production in the vascular endothelium.62 This pro-inflammatory state leads to intimal hyperplasia and proliferation of smooth muscle cells and fibroblasts, progressive stenosis of the affected intracranial artery and, finally, occlusion.63,64

Transcranial Doppler ultrasound measures the blood flow velocity in the brain arteries and is an important tool to detect the risk of stroke in children with SCD. Results of Transcranial Doppler assist physicians to include high-risk individuals on chronic transfusion or hydroxyurea therapy to prevent the occurrence of the first stroke (primary prevention).12,14,30

The Stroke Prevention Trial in Sickle Cell Anemia (STOP), a randomized clinical trial performed in children with SCD and high-risk of stroke detected by TCD [time averaged maximum mean velocity (TAMMV) in the internal carotid or middle cerebral artery ≥200cm/s], showed a 92% decreased difference in risk of the first stroke in the group treated with chronic transfusion therapy compared with the standard treatment (observational). Children with TAMMV ≥200cm/s have a 10% risk per year of developing stroke, which may be reduced to less than 1% with chronic transfusion therapy.12 The STOP study was halted earlier than planned after this clear-cut evidence was found. Based on these results, the National Institutes of Health (NIH) recommended TCD screening of children with SCA to assess the risk of stroke development, and chronic transfusion therapy to reduce risk of stroke in children at high risk. Consequently, there was a 45% reduction in the incidence of hospitalizations due to stroke in the USA, as well as a 45% reduction of hospital stays and 24% decrease in hospital fees attributable to stroke, when comparing the pre- and post-STOP published data.65

After the publication of STOP, several studies have been released reporting reduction of stroke incidence in children with SCD.7,66–69 Data from the Centre Hospitalier Intercommunal in France showed that TCD and intensification therapy (chronic transfusion therapy, bone marrow transplantation, or hydroxyurea therapy) reduced the cumulative risk of stroke before 18 years of age from 11% to 1.9%.7

The STOP-2 trial evaluated the possibility of interrupting chronic transfusion therapy after TAMMV normalization in children with SCA and high risk of stroke development detected by TCD. STOP-2 was also halted early after some evidence that discontinuing chronic transfusion therapy results in reversion to high-risk TCD or stroke in many children.70

Recently, the NIH also halted early the Transcranial Doppler with Transfusions Changing to Hydroxyurea (TWiTCH) trial. The aim of the study was to evaluate whether hydroxyurea therapy would lower the TCD TAMMV in children with SCD to a similar degree as chronic transfusion therapy. Children on chronic transfusion therapy to prevent the first stroke (at least one year; mean time of 4.6 years) were randomized to keep receiving transfusions or discontinuing chronic transfusion therapy and initiating hydroxyurea therapy. Hydroxyurea therapy was not inferior to chronic transfusion therapy in reducing TCD TAMMV in children with SCD at high-risk for stroke development and no magnetic resonance angiography (MRA)-defined severe vasculopathy at the entry of the study.14

Several studies have been conducted to establish risk factors for the development of stroke in individuals with SCD. However, the association of the underlying vasculopathy of stroke with these risk factors is not well set. Table 1 summarizes characteristics and results of studies evaluating the influence of genetic, laboratory, and clinical factors on the occurrence of stroke in individuals with SCD. Data from these studies suggest that certain characteristics are mainly responsible for the occurrence of stroke in SCD (Table 2). Studies that evaluated some risk factors with no significant association with stroke were not included in Table 1 due to the limited space. Studies aimed to evaluate risk factors for silent infarcts and hemorrhagic stroke alone were not included because they were out of the scope of this review.

The pathogenesis of stroke in SCD probably involves a combination of the βS mutation of the HBB gene, genetic modifiers, and environmental factors. Genetic predisposition to stroke in SCD has been suggested by the fact that the occurrence of stroke among siblings and twins is increased.71,72

Stroke risk in individuals with SCD probably involves many genes, within and outside of the HBB locus. Genes involved in the process of inflammation, immune response, coagulation, cell adhesion, lipid metabolism, blood pressure regulation, hypoxia, among others, are candidates involved in the development of stroke. These genes have been associated with increased susceptibility or protection against the occurrence of the event.73 Evidence has been shown that inflammation is the most important process in the arteries of the Willis circle related to the development of stroke.74

The protective effect of alpha-thalassemia against the development of stroke is well established in the literature.7,8,51,75–81 This association has been attributed to both hematologic and rheologic factors. Due to the lower production of alpha globin chains, the co-inheritance of alpha-thalassemia reduces the intracellular concentration of Hb and, consequently, reduces polymerization, modulating other hematological characteristics. Most studies show an increase in the number of RBCs, total Hb and hematocrit levels, and decreased levels of mean corpuscular volume, mean corpuscular Hb, mean corpuscular Hb concentration, reticulocytes, leukocytes and hemolysis. Currently, a high reticulocyte count is considered the most important risk factor for stroke, as analyzed later in this review. A high reticulocyte count was a risk factor for acute cerebral ischemia and high-risk TCD in a multivariate analysis.82,83 Thus, the reduction of reticulocyte count caused by the co-inheritance of alpha-thalassemia51 may be crucial to protect children with SCA from stroke.

The deformability of sickled cells in individuals who also have alpha-thalassemia is greater than that of patients without it.84,85 In addition, individuals with alpha-thalassemia have reduced numbers of dense cells and irreversibly sickled cells.86–88 Furthermore, the presence of alpha-thalassemia could reduce the adhesion of sickled cells to the endothelium in vivo.39,40 Thus, the improvement in rheological features of RBCs conferred by the co-inheritance of alpha-thalassemia may also contribute to reducing the risk of stroke in children with SCA.

In the last few years, some studies have suggested a role of other genes in the pathogenesis of stroke in children with SCA. In a candidate gene association study, Sebastiani et al. used Bayesian network modeling that tested 108 single nucleotide polymorphisms (SNPs) in 39 genes and found that 31 SNPs in 12 genes interact with Hb F to modulate the risk of stroke. The predictive value of the model was assessed in an independent validation set of patients with SCA and it predicted the occurrence of stroke with a 100% true positive rate and a 98.14% true negative rate giving an overall predictive accuracy of 98.2%. Polymorphisms in the ADCY9, ANXA2, BMP6, CCL2, CSF2, ECE1, ERG, MET, SELP, TEK and TGFBR3 genes were associated with stroke and had a major statistically independent effect on the risk of event. Furthermore, polymorphisms in genes previously associated with stroke in populations without SCA, such as transforming growth factor (TGF)-beta pathway genes, were also included in the network.89

In a genome-wide association and exome study using a discovery cohort of 677 children and an independent validation cohort of 288 children, two mutations in the GOLGB1 (Y1212C) and ENPP1 (K173Q) genes were significantly associated with a decreased risk for stroke. The GOLGB1 Y1212C mutation was also associated with protection from silent infarcts and abnormal TCD.90 In a recent longitudinal study by our group, the ENPP1 K173Q was associated with an increased risk of stroke and trends toward increased risk for high-risk TCD.91 The ENPP1 is a class II glycoprotein known to influence insulin sensitivity, binding, and, consequently, inhibiting the insulin receptor. The variant ENPP1 173Q is a more potent inhibitor of the insulin receptor than the wild variant, and it has been associated with insulin resistance and diabetes type II.66,92 Insulin stimulates the activation of protein kinase B and this protein signals the release of NO from endothelial cells. This pathway is impaired in individuals with the ENPP1 173Q variant. This variant has also been associated with increased blood pressure, cardiovascular events, and reduced activity of enzyme nitric oxide synthase.93,94 We have proposed that NO pathway impairment is the possible mechanism for the ENPP1 K173Q modulation of stroke in children with SCA.91 Further studies are needed to shed additional light on the role of ENPP1 K173Q in the pathogenesis of stroke in pediatric patients with SCA.

Although several studies have reported that other genetic markers are associated with risk of stroke in individuals with SCD (Table 1), there is controversy over the different studies. Most of these genetic associations are still preliminary and require confirmatory studies.2 Few studies have been validated, mainly because the interpretation of these studies is hampered by the relatively small sample size and/or absence of cohorts to validate the results.95 In addition, there is a wide variation in the definition of the outcomes studied. Furthermore, stroke in children with SCD seems to be a complex multifactorial and polygenic disorder that is influenced by many characteristics, each with only modest effects. The small effect of each marker also contributes to the controversial results. The influence of the tumor necrosis factor-alpha (TNF-α) G-308A (rs1800629) polymorphism is a good example of these controversies. Some reports have indicated that homozygous for the −308G allele is associated with increased risk of stroke.96,97 On the other hand, the −308A allele was reported to be associated with increased risk of stroke in two recent reports from our group82,98 whereas two other studies reported no association.99,100 Similarly, some studies have identified an association of the Central African Republic (CAR) haplotype with stroke in subjects with SCD.79,101–103 However, reliable replication of this potential association has not been reached in other independent validation cohorts.78,104–107

Regarding laboratory parameters, the relationship of hemolysis markers, such as low Hb concentration, high lactate dehydrogenase (LDH), and high reticulocyte count, with the occurrence of stroke stands out in the literature.7,8,13,19,76,77,79,80,83,99,108,109 Recently, the reticulocyte count has gained prominence as possibly the most important laboratory risk factor for increased risk of stroke.7,82,83,110,111 In recently published reports, including two by our group, reticulocyte count was the most important predictor of stroke or high-risk TCD.82,83,110–112 The re-analyzed data from the Cooperative Study of Sickle Cell Disease report, designed to assess the impact of very early detection of reticulocytosis, anemia, or leukocytosis on prediction of future major adverse events, showed that a high reticulocyte count was significantly associated with increased risk of stroke and death during childhood.111 More recently, data from a French cohort of children demonstrated a substantial independent association of high reticulocyte count [Hazard ratio: 1.82 per 50×109/L increase; 95% confidence interval (95% CI): 1.10–3.01] with development of cerebral macrovasculopathy.112 In a Brazilian cohort of 395 children with SCA, multivariate analysis showed that for each percent increase in reticulocyte count, the mean risk of acute cerebral ischemia (stroke or transient ischemic attack – TIA) or high-risk TCD increased by approximately 1.3% (95% CI: 1.13–1.47%) and 1.5% (95% CI: 1.27–1.69%), respectively.82 Reticulocytes probably have an essential role in the pathogenesis of cerebral vasculopathy in children with SCA. Hyperhemolysis, as indicated by high steady state reticulocyte count, releases Hb, free heme, arginase, and other molecules from RBCs, which generate reactive oxygen species, scavenge NO, and inhibit NO production, promoting endothelial damage, platelet activation and induction of inflammation in the vascular endothelium.113–116 Endothelial dysfunction and inflammation stimulate selective release of mediators, which further promote expression of adhesion molecules on the endothelial and blood cells,117 contributing to a series of pathophysiological events that culminate in vasculopathy involving large cerebral arteries. Additionally, data from the French cohort showed that the serum LDH level and reticulocyte count were significant independent factors associated with stroke in multivariate analyses,7 suggesting that hemolysis is not the single event associated with the pathophysiology of stroke but high steady state reticulocyte count per se is also probably involved. The tendency of reticulocytes to adhere to endothelial cells, resulting in endothelial activation and damage, might be the first stage of a series of pathophysiological events resulting in cerebral vasculopathy.36 The benefits of hydroxyurea therapy in the prevention of stroke14,118 could be partially attributed to its effect on decreasing reticulocyte adhesion to endothelial cells.119

Despite the relatively large number of studies, the role of concomitant glucose-6-phosphate dehydrogenase (G6PD) deficiency on risk of stroke remains controversial. In our experience, the prevalence of stroke or high-risk TCD was not significantly different in the groups with and without G6PD deficiency in a retrospective cohort study.120 This absence of association has been reported in other studies that used molecular analysis as the diagnostic method for G6PD deficiency.78,121 However, Thangarajh et al.,122 showed that the presence of the 376G (rs1050829) or 202A (rs1050828) allele was a significant and independent risk factor for intracranial MRA-arteriopathy in males with SCA. The 376G allele leads to a very mild reduction in G6PD activity, as demonstrated in one study.120 Children with the 376G allele (isoform A in males or AA in females) showed 85.2% of the enzymatic activity of G6PD when compared to individuals with the wild 376A allele (isoform B in males or BB in females). On the other hand, studies that reported an association between G6PD deficiency and abnormal TCD or intracranial stenosis used the measurement of G6PD activity as the method to define G6PD deficiency.7,77 One plausible explanation for these divergent results is that transcriptional and epigenetic factors that influence G6PD expression123 may be involved in the modulation of stroke in children with SCA. Thus, children without pathogenic missense mutations, but with downregulation of G6PD expression might have a higher risk for CVD. Similarly, children with pathogenic missense mutations and upregulation of G6PD expression might have a lower risk. However, some studies108,120,121,124 did not detect any difference in the mean G6PD activity in groups with and without ischemic stroke or high-risk TCD. The association of G6PD deficiency and stroke seems to be unlikely. The most important laboratory predictor of stroke, high steady state reticulocyte count,83,111 is associated with raised G6PD activity, as demonstrated in one study.120 Another reason for the improbable association between G6PD deficiency and stroke is that G6PD deficiency is an X-linked inherited disease and the effects of deficiency should be more common in males. However, available data show no influence of gender.108 It is obvious that genetic background heterogeneity among populations may also lead to contradictory results. G6PD deficiency may be a risk factor for some children, but not for others from different ethnic backgrounds. Further large-scale prospective longitudinal studies controlled for ancestry are warranted to elucidate the relationship of G6PD deficiency with stroke susceptibility in children with SCA. Furthermore, G6PD seems to have a critical antioxidant role in endothelial cells.7 Further studies measuring the G6PD activity in circulating endothelial cells may provide a better understanding of the relationship between G6PD deficiency and cerebrovascular vasculopathy in SCA.

About clinical features associated to the development of stroke, factors that cause imbalance between demand and supply of oxygen in the brain are critical. For example, aplastic crisis secondary to erythrovirus B19 infection,125 nocturnal hypoxemia,126 and the occurrence of acute chest syndrome8 have been reported to be associated with stroke. Additionally, in a large study of 516 children with SCA, silent cerebral infarct was the single factor associated with intracranial MRA-vasculopathy in the final multivariate logistic regression model.122 Another study of 248 children with SCA reported a strong association between silent infarcts identified at age of six years or older and subsequent development of stroke.127 Although both events share some pathophysiological background, the reason for this association is not clear. More convincingly, patients with a history of TIA were much more likely to have an overt stroke.8,127 According to data from the Cooperative Study of Sickle Cell Disease (CSSD), children who have had a TIA have a 56-times higher chance to subsequently develop stroke when compared to those without prior TIA (95% CI: 12.0–285).8 In our experience,82 four children had TIA and subsequently developed strokes and another three children had TIA and did not evolve with strokes; of the latter group the TIA was followed by a high-risk TCD (two cases) or an inconclusive TCD (one case) due to difficulty of insonation in the presence of a ‘good’ transtemporal window, probably due to severe stenosis.128 Then, recognizing and treating TIA certainly reduces the risk of an overt stroke. Children with a TIA event should be intensively monitored by TCD and/or MRA to evaluate the indication of intensification therapy such as prophylactic blood transfusion program or hydroxyurea therapy. Another clinical factor that has been associated with the occurrence of stroke is nocturnal126 or daytime129–131 Hb desaturation. Hypoxemia stimulates and exacerbates pathophysiological events involved in the vasculopathy associated with stroke in SCD, including Hb S polymerization, platelet activation, endothelial adhesion, and adhesion of RBCs to the endothelium.132–134