Stroke Risk and Its Management in Patients With Sickle Cell Disease
Fenella J. Kirkham
Clinical Presentation of Stroke in Sickle Cell Disease
There is a broad
spectrum of acute presentation with CVA and other neurological complications in patients with SCD ( Table 1 ). Clinical stroke, with focal signs lasting more than 24 hours (Figure 1A), is 250 times more common in
children with SCD than in the general pediatric population, and commonly presents 'out of the blue' in an apparently well child. Patients with SCD also have transient ischemic attacks with symptoms and signs resolving
within 24 hours, although many of these individuals are found to have had recent cerebral infarction or atrophy on imaging (Figure 1B). The insidious onset of 'soft neurological signs', such as difficulty in tapping
quickly, is usually associated with cerebral infarction.[6,7] In addition, seizures (Figure 1C), coma (Figure 1D) and headache (Figure 1E) are common
presentations of stroke and cerebrovascular disease in children with SCD. Hyperventilation, for example during an electroencephalogram, occasionally provokes transient or permanent neurological deficits, usually involving the posterior circulation territory.
Altered mental status—with or without reduced level of consciousness, headache, seizures, visual loss or focal signs—can occur in numerous contexts, including infection (Figure 1D), shunted hydrocephalus (Figure 1E), acute
chest syndrome (ACS)[13,14,15] (Figure 1F), aplastic anemia secondary to parvovirus,[16,17] after surgery,[10,18,19]
transfusion or immunosuppression[21,22] (Figure 1G), and apparently spontaneously (Figure 1H). In one large series of patients with ACS, 3% of children had neurological
symptoms at presentation, and such symptoms developed in a further 7-10% as a complication of ACS. These patients are classified clinically as having had a CVA,[2,3] although there is a wide differential
of focal and generalized vascular and nonvascular pathologies—often distinguished using acute magnetic resonance techniques (Figure 1)—with important management implications.[11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27]
Axial T2-weighted MRI (unless otherwise stated). (A) Large middle cerebral
territory infarct in a previously asymptomatic girl with sickle cell anemia (SCA). (B) Left occipital infarct with moderate lobar atrophy (arrow) in a boy with HbSβ0-thalassemia who presented with two episodes of reversible ischemic neurological deficit
(left and right hemiparesis). (C) Covert ('silent') infarctions (arrow) in a child with SCA and focal tonic seizures. (D) Infarction in both anterior and posterior arterial border zones in an 8-year-old boy with previously uncomplicated SCA who developed seizures
and coma after surgery to drain a painful swelling of his left cheek, which was sustained after a fall and was associated with fever. (E) CT scan of a child with hemoglobin sickle cell disease who had a shunt placed in infancy. The scan was performed 10 days
after presentation with headache, and shows definite cerebral edema; the dense appearance of the straight sinus (arrow) raises the possibility of venous sinus thrombosis, but is not diagnostic. (F) Signal change in the left posterior occipital (arrow) and
bilateral high parietal regions ('reversible posterior leukoencephalopathy') with hippocampal involvement in a woman with previously uncomplicated SCA who presented aged 18 years with acute chest syndrome. (G) Signal change in the grey and white matter ('reversible
posterior leukoencephalopathy'; arrows) in a 9-year-old boy with SCA who had been treated with ciclosporin for nephrotic syndrome. (H) Intracerebral hemorrhage (arrow) in a previously well child with SCA who presented with acute headache.
addition to those who have had obvious acute neurological events, by adolescence up to 25% of children with homozygous sickle cell anemia have covert or 'silent' infarction on MRI, characteristically in the anterior or posterior border zones[28,29,30,31]
(Figure 1C), without having had a clinical stroke. Voxel-based morphometry (Figure 2) shows that, compared with controls, there is evidence of damage in the white matter of the border zones, even in patients with sickle cell anemia who have normal T2-weighted
MRI. Neurological examination is usually normal,[6,7] although these patients might have had subtle transient ischemic attacks, headaches or seizures. Cognitive difficulties,[33,34]
which commonly affect attention and executive function, are common in SCD, sometimes from infancy; they can be progressive,[37,38] and are associated
with covert brain damage.[31,33,37,39]
Voxel-based morphometry comparison of white matter density between controls and patients with sickle cell anemia and covert infarction. Regions of reduced white matter density in patients
with sickle cell anemia and covert infarction are displayed on the mean white matter segment. The white matter abnormality distribution is similar to that in the patient with overt stroke shown in Figure 1D.
Brain Pathology in Patients
With Sickle Cell Disease
In parallel with the wide range of clinical neurological presentations of CVA in patients with SCD, a number of apparently discrepant findings have been described in the pathological literature. In patients who
have presented with acute stroke, large infarcts in an arterial territory or in the border zones between anterior and middle cerebral arteries have been described,[40,41] in association with endothelial hyperplasia, fibroblastic reaction, hyalinization and
fragmentation of the internal elastic lamina, and thrombi in large and small vessels. Not all patients who die after developing neurological symptoms have large-vessel disease, however. In addition to the typical small necrotic lesions in the border
between the cortex and the subcortical white matter, acute demyelination and venous sinus thrombosis[42,43] have also been documented.
When a child with SCD presents with acute neurological problems, the
priority is to conduct an exchange transfusion. As these patients are often admitted to a peripheral hospital without facilities for emergency imaging under general anesthesia, the literature on acute imaging is scanty. CT often does not show abnormalities
within the first 24 hours after the onset of neurological symptoms. By contrast, diffusion-weighted MRI[22,23,44] (Figure 3A) can show ischemic regions within minutes—that is, before irreversible
infarction has occurred—and T2-weighted MRI is usually abnormal within a few hours.
Diffusion and perfusion MRI. (A) Diffusion MRI shows abnormalities (arrows) in the patient with ciclosporin toxicity (Figure 1G). (B) Perfusion MRI abnormality throughout the right hemisphere and posteriorly on the left in an
11-year-old boy with sickle cell anemia who developed headaches after a chest crisis and was found to have high internal carotid artery and middle cerebral artery velocities on transcranial Doppler ultrasound (right vessels 200 cm/s; left vessels 161 cm/s).
In patients presenting with clinical stroke, neuroimaging might reveal a large infarct in the distribution of the middle cerebral artery, or an abnormality in the basal ganglia, or deep white or grey matter of the border zones.
The involvement of these territories indicates proximal arterial pathology, whereas parieto-occipital or thalamic involvement should suggest venous sinus thrombosis. Subarachnoid and intracerebral hemorrhage also occur, often in the
context of acute hypertension, or as a result of venous sinus thrombosis, rupture of aneurysms (usually located at the bifurcations of major vessels, particularly in the vertebrobasilar circulation)
or rupture of fragile collaterals.
Although it was assumed in the past that stroke in SCD was the result of sickling of red blood cells in the small cerebral vessels, this does not seem to be the direct
cause in most cases. Between 60% and 90% of children with SCD and acute stroke in an arterial distribution have abnormal findings on conventional angiography or magnetic resonance angiography (MRA), and there is also an association
with seizures (Figure 4). Typical abnormalities include stenosis (Figure 5A,B) or occlusion (Figure 5C) of the distal internal carotid or middle cerebral arteries,[10,50,51,52]
which can be diagnosed as high time-averaged maximum velocity on transcranial Doppler ultrasound (TCD; Figure 4), a clinically important risk factor for stroke (see below). Vertebral or internal carotid dissection,
moyamoya syndrome (bilateral severe stenosis or occlusion of the internal carotid arteries with collateral formation; Figure 5D) and, occasionally, rarer patterns such as small-vessel vasculitis have also been documented. Moyamoya syndrome
is associated with an increased risk of stroke recurrence.[48,54] Venous sinus thrombosis[24,25,42,43] (Figure 1E, Figure 6) is probably
underdiagnosed because many patients do not undergo acute vascular imaging; if emergency MRA is available and the results are found to be normal, magnetic resonance or CT venography should be considered.
Transcranial Doppler velocities. Abnormal right middle cerebral artery (220 cm/s) and normal left middle cerebral artery (130 cm/s) in a 10-year-old girl with sickle
cell anemia, focal tonic seizures (right side), transient ischemic attacks and headaches, on an exchange-transfusion program. Magnetic resonance angiography shows right proximal stenosis of the middle cerebral artery (bottom panel).
Magnetic resonance angiography (MRA). (A) Grade 1 turbulence in an asymptomatic child
with sickle cell anemia and silent infarction. (B) Grade 2 turbulence in the right terminal internal carotid artery, A1 segment of the right anterior cerebral artery, right middle cerebral artery (M1 and M2 segments) and in the M1 segment of the left middle
cerebral artery in a boy with HbSβ0-thalassemia (Figure 1B). (C) Middle cerebral artery occlusion in the girl with the large middle cerebral artery territory occlusion (Figure 1A). (D) Moyamoya in a child with sickle cell anemia and recurrent transient
Magnetic resonance venogram showing obliteration
of the left transverse, sigmoid and straight sinuses in a patient with sickle cell anemia. The patient presented with headache, vomiting, drowsiness, papilledema and a sixth nerve palsy.
Risk Factors for Stroke in Sickle Cell
There seems to be a familial predisposition to stroke and to high blood flow velocities in SCD, indicating that genetic factors probably contribute to stroke risk. Siblings might, however, also share adverse environmental
conditions, including poverty, air pollution and poor nutrition. There is considerable current interest in epistatic polymorphisms as additional risk factors for stroke in SCD. The A3 and A4 alleles of the GT-repeat polymorphism of the angiotensinogen gene
seem to be associated with a fourfold increase in risk of clinical stroke in SCD, perhaps because of an effect on blood pressure. Factor V Leiden and the thermolabile variant of the methylenetetrahydrofolate reductase gene are rare in Afro-Caribbean populations,
and at present there is little evidence for a link between stroke and these or any other genetically determined or acquired prothrombotic disorders in patients with SCD.
In recent years, evidence for human leukocyte antigen (HLA)-linked
susceptibility for stroke has been found for the class I HLA-B and class II HLA-DRB loci in children with SCD. Hoppe and her co-workers used a candidate gene approach in patients screened with MRI as part of the US-based CSSCD. Dividing patients on the
basis of the MRI findings into three groups—those with normal imaging, those with infarction clearly involving the territories of the large vessels (with or without additional small-vessel involvement), and those with infarction not involving the territories
of the large vessels (presumed small-vessel disease)—the authors found that polymorphisms in specific HLA types and in genes involved in inflammation, cellular adhesion, blood pressure regulation and lipid metabolism were differentially associated with
the two stroke phenotypes. Another group has demonstrated that large-vessel disease in SCD is related to variations in genes that code for factors involved in the responses to inflammation, hypoxia, adhesion and coagulation.
erythrocytes adhere more avidly to vascular endothelial cells than do normal erythrocytes, and in SCD evidence has also accrued for neutrophil and platelet adhesion to the vascular endothelium, and activation of pathways leading to inflammation and thrombosis.
Vascular cell adhesion molecule 1 (V-CAM1) is of particular interest in SCD as this molecule seems to coordinate the inflammatory response by recruiting leukocytes, and mice depleted of V-CAM1 have a high leukocyte count—a risk factor for stroke in humans.
Interestingly, there is a high frequency of single nucleotide polymorphisms (SNPs) in the VCAM1 gene in people of Afro-Caribbean origin, and one of these variant alleles (Gly1238Cys) has been shown to associate with protection from stroke in the Jamaican population,
as compared with the wild-type allele.
Sebastiani et al. have recently used Bayesian modeling to examine the association of SNPs in candidate genes with sickle-cell-related stroke. They found 31 SNPs in 11 genes, including
bone morphogenetic protein 6 (BMP6), three genes involved in the transforming growth factor β (TGFβ) signaling pathway, and SELP (P-selectin; also known as granule membrane protein 140 and antigen CD62), which is known to be associated with stroke
in the general population; all of these factors seemed to directly affect stroke risk. They also identified SNPs in a further nine genes, including endothelin-1, which is close to BMP6 on chromosome 6 and is upregulated during acute hypoxia, that seemed to
be acting indirectly. When validated in a separate population, this combination of genes, interacting with the percentage of hemoglobin F, was found to have 98.2% accuracy for distinguishing patients with stroke from those without stroke, with all seven of
the studied strokes correctly classified.
Low hemoglobin, high white cell count, previous transient ischemic event, hypertension and history of chest crisis all seem to be risk factors for overt ischemic stroke in SCD ( Table 2 ).[3,27]
For hemorrhagic stroke, only low hemoglobin and high white cell count were found to be predictors in the CSSCD ( Table 2 ). A recent study comparing hemorrhagic with ischemic stroke in children, however, did not confirm these two variables as risk factors,
and instead highlighted the importance of recent transfusion and corticosteroids, perhaps in relation to acutely increased blood pressure. Large-vessel disease might be associated with markers of increased hemolysis, such as lactate dehydrogenase and
reticulocyte count. Patients presenting with neurological symptoms after chest crisis are more likely to have an atypical stroke syndrome or stroke mimic—such as hemorrhagic stroke, posterior circulation stroke,[14,63,64] symptomatic or asymptomatic
(covert) single or multiple small infarcts or 'lacunae', reversible posterior leukoencephalopathy, global or focal edema, laminar necrosis or acute necrotizing encephalomyelitis—than they are to have a typical infarctive stroke; thrombocytopenia
was the only predictor for neurological symptoms in a large series of patients with ACS. Covert infarction is also associated with high pocked red cell count in infancy, compatible with early splenic infarction, as well as low hemoglobin and high white
cell count; interestingly, covert infarction is less common in individuals with frequent pain, but more common in those with a past history of seizures ( Table 2 ). Progressive covert infarction is more common in patients with a persistently high white
cell count, suggesting a role for chronic inflammation. Factors associating with low IQ and other cognitive problems in SCD include hematological variables such as low hemoglobin,[29,65,66] high white cell count[66,67] and thrombocytosis, in addition
to parenchymal brain damage,[32,34,37,39] large-vessel disease,[63,68] perfusion abnormality (Figure 3B) and nutritional factors (see below).
In African Americans, there is a high prevalence of individuals with only two or three α
globin genes (compared with the normal complement of four). This α thalassemia seems to reduce the risk of stroke, probably through an increase in hemoglobin levels. High hemoglobin F levels seem to ameliorate the risk of overt stroke and silent
infarction, at least in childhood. The β globin haplotypes might also alter risk, probably by influencing hemoglobin F levels, although the data are conflicting. Increasing hemoglobin F levels provides the rationale for the use of hydroxyurea to ameliorate
disease severity ( Table 2 ).
Recent data indicate that nocturnal hypoxemia, related in part to anemia, might increase the risk of CNS events in patients with SCD. Platelet and leukocyte activation, which might affect endothelial function,
are inversely related to mean overnight oxygen saturation in children with SCD, and there is evidence for increased levels of markers of cellular and endothelial adhesion. The mean overnight oxygen saturation seems to be associated with the severity
of intracranial vasculopathy determined by turbulent flow on MRA,[63,73] which is in turn related to regional cerebral perfusion. Brain areas with low perfusion are at risk of ischemia, which might be exacerbated if there is also hypoxemia. It is not yet
clear if this effect is secondary to sustained hypoxemia or obstructive sleep apnea, which have been documented in reports of children with arterial ischemic stroke. Although the numbers of patients studied to date have been small, adenotonsillectomy—commonly
used to treat obstructive sleep apnea or recurrent tonsillitis—might not reduce the risk of CNS events, perhaps because hypoxemia commonly persists after surgery. There is some evidence that asthma predisposes to chest crisis and CVA, and the
complex relationships between these comorbidities are currently under intense investigation. A pilot feasibility and safety trial of overnight respiratory support has commenced.
Acute and chronic infections have long been recognized
as precipitants for the neurological complications of SCD; however, there are few data on the effects of penicillin prophylaxis or vaccination against Streptococcus pneumoniae or Haemophilus influenzae, which are now the standard of care. For the neonatally
screened East London SCD cohort, the introduction of penicillin prophylaxis from infancy was associated with 95.2% stroke-free survival at 15 years, which seems to be an improvement compared with previous reports.[3,77] Aspirin might prevent stroke, silent
infarction and cognitive impairment by mechanisms that include reduction of inflammation and the antiplatelet effect. A pilot trial has commenced; however, until a phase III randomized trial has been completed, aspirin must be used with caution in patients
with existing ischemic stroke and cerebrovascular disease because of the risk of hemorrhage. Blood pressure is relatively low in patients with SCD in comparison with controls, but relative hypertension is associated with increased risk of ischemic stroke.
The effect of controlling blood pressure on stroke risk has not, however, been examined systematically.
There are few data on the role of nutrition in determining the risk of overt and silent stroke in SCD, although there is increasing
evidence for an effect in elderly adults. If nutritional factors do indeed contribute to stroke risk in patients with SCD, this might explain certain geographical variations that have been observed. For example, although Greek patients with HbSβ0-thalassemia
have silent infarction, high TCD velocities and cognitive problems seem to be rare; this rarity might be related to the Mediterranean diet. Short stature was a risk factor for low IQ in the Jamaican cohort, but these patients did not undergo MRI, so
the existence of any interaction with covert infarction could not be determined. In the East London cohort, low IQ was associated with infarction on MRI and with cerebrovascular disease and, in addition, was independently predicted by short stature from the
age of 3 years. The data on the relationships between anthropometric measurements and the risks of large-vessel disease or overt stroke are limited.
There is considerable evidence that high levels of homocysteine, related to genetic
factors as well as vitamin intake, predispose to cerebrovascular disease, including accelerated atherosclerosis, dissection, thrombosis, embolism and venous sinus thrombosis. Some studies have shown high homocysteine levels in children with SCD,[81,82] which
seem to be correctable with vitamin B supplementation, although there is controversy over the relevance of these findings to vascular disease. Homocysteine levels might be reduced by folate, vitamin B12, pyridoxine (vitamin B6) and riboflavin (vitamin
B2) supplementation. The importance of diagnosing pernicious anemia, particularly in patients on folate supplementation, was underscored by a recent case report. Iron deficiency is a common comorbidity of SCD in young children in the developing world.
This deficiency is likely to have a complex effect on phenotype, reducing the degree of hemolysis and potentially its associated complications, but possibly having a detrimental effect on cognitive function in infancy.
might improve endothelial function in patients with SCD. Morris et al. have shown that pulmonary pressures were reduced by arginine supplementation in patients with pulmonary hypertension; however, there are currently no data on cerebrovascular disease
or neurological complications. Intake of antioxidants such as aged garlic, ascorbic acid and vitamin E might affect intermediate risk factors, such as oxidative damage,[86,87] hematology,[87,88]blood pressure and the hypoxic hyperventilatory response;[63,89]
however, there are no data on the effects of supplementation on endothelial function, cerebrovascular disease or the risk of neurological events. Although increased hemolysis is a theoretical concern with high-dose vitamin C, there is no evidence that
it is an actual side effect. Zinc supplementation, which reduces red blood cell dehydration, might be associated with reduced hemolysis and other crises, but there are no data on its effects on neurological disease. There are very few data on other
nutritional supplements, such as fish oil, which might be of benefit to the general population at risk of vascular disease. Trials of appropriate nutritional supplementation from infancy would be justified as there is evidence for low levels of key components
in individuals with SCD despite adequate dietary intake as in the general population, indicating that consumption needs to be increased in this group of individuals. Currently, as there are no known nutritional risk factors, health professionals caring for
patients with SCD should encourage consumption of a wide variety of foodstuffs as part of a healthy diet, including at least five portions of fruit or vegetables per day.
Recent data from the CSSCD showed that covert infarction
seen on MRI was associated with an increased risk of overt stroke (1.03 per 100 patient-years) and progression of covert infarction (7.06 per 100 patient-years). The Silent Infarct Transfusion Trial, in which children with covert infarction seen on MRI
will be randomized to blood transfusion or observation, is currently enrolling patients and will report after 2011.
TCD is a safe, noninvasive, well tolerated, relatively low-cost procedure in which the velocity of blood
flow can be measured in intracranial vessels using an ultrasound probe placed over the temporal bone to screen for cerebrovascular disease. In a comparison with conventional angiography, TCD showed a sensitivity of 90% and specificity of 100% for the diagnosis
of abnormality. Follow-up studies have indicated that blood velocities over 200 cm/s (abnormal) and between 170 cm/s and 200 cm/s (conditional) in the internal carotid or middle cerebral artery of children with SCD predict stroke risks of 40% and 7%, respectively,
over the subsequent 3 years, presumably because high velocities indicate the development of severe middle cerebral artery narrowing secondary to turbulence or fixed stenosis. TCD might become abnormal before MRA does; however, although it is more expensive
and can require general anesthesia in young children, MRA is very useful in confirming the presence and extent of cerebrovascular disease.[10,50,51,52,94]
Stroke Prevention in Patients With Sickle Cell Disease
Royal College of Physicians (UK) has recently published guidelines on the management of stroke in childhood, which address the issues of primary and secondary prevention of stroke in SCD ( Box 1 ).
If no treatment was offered after
a first ischemic stroke in children with sickle cell anemia, the risks of recurrence documented by cohort studies have been between 20% and 92%.[27,96,97,98,99] Two studies found a high risk of recurrence in children who had arteriographic abnormalities.[99,100]
Ever since transfusion therapy was shown to be effective at reducing recurrent events,[99,100] it has been the mainstay of treatment for stroke in SCD. The aim of transfusion is to reduce the percentage of HbS and increase levels of HbA, thereby reducing
the deleterious effects of sickling while improving tissue oxygenation. For acutely sick patients, such as those with severe chest crisis or neurological symptoms, treatment is usually simple or partial exchange blood transfusion. Although there has never
been a controlled trial of transfusion, a chronic intermittent transfusion program, aiming to keep the HbS level below 30%, decreases the risk of stroke recurrence to between 10% and 20%.[2,101,102] There is ongoing controversy over whether allowing the HbS
percentage to drift up towards 50% with less-frequent blood transfusions, or switching to hydroxyurea[101,103,104] after a few years, is safe. Moyamoya syndrome is a risk factor for stroke recurrence despite regular transfusion,[48,54] although this risk
seems to be reduced after extracranial-intracranial bypass or indirect revascularization.[101,105] Bone marrow or cord blood transplantation is an option for children with a matched sibling donor.[101,106,107,108] These procedures might, however, result in
acute neurological complications including subarachnoid hemorrhage and border zone edema in the context of hypertensive encephalopathy, and serial MRI scans in individuals with pre-existing cerebral damage might show new lesions as well as extension
of existing abnormality. Further studies of these procedures are needed as some researchers have not found progression, and the cerebrovascular disease can stabilize as demonstrated on both MRA and TCD.
The peak age
for hemorrhage in SCD is 20-30 years, and these young adults usually have aneurysms, which are amenable to extirpation by coil embolization or surgery; these procedures reduce the risk of recurrence. By contrast, aneurysms and other cerebrovascular
diseases are rarely found in children with SCD and intracranial or subarachnoid hemorrhage—conditions that seem to be common in those exposed to corticosteroids and rapid blood transfusion. Care with corticosteroid use and judicious blood transfusion
to achieve small increases in hemoglobin without acute hypertension might reduce the incidence of hemorrhagic stroke.
There is Class A evidence for the role of blood transfusions in preventing first stroke in children
with sickle cell anemia and high blood flow velocities on TCD. The Stroke Prevention Trial in Sickle Cell Anemia (STOP) showed a clear benefit for prophylactic transfusion in children with blood flow velocities of more than 200 cm/s. A large proportion
of the treated patients apparently received unnecessary transfusion, however, as only 15% of the nontransfused children went on to have a stroke despite very high velocities. This is an important issue, as transfusion is associated with risks of infection
and alloimmunization, and many children and their families find regular blood transfusion and chelation therapy burdensome. The subsequent STOP II has, however, shown that it is essential to continue to transfuse long-term even if TCD velocities return
to the normal range, as the risks of stroke or reversion to high-risk TCD velocities are unacceptably high. Reductions in the incidence of stroke in SCD recorded in California and East London are probably related to the introduction of TCD screening
and prophylactic transfusion. TCD screening is currently not universally available, however, and cannot, therefore, be mandated at present.[117,118] In addition, there is evidence that adults with SCD do not have high TCD velocities, and programs of TCD screening
and prophylactic transfusion are probably not appropriate for older age groups.
Although it is now possible to prevent ischemic stroke in SCD with lifelong regular blood transfusions, this is
a heavy burden for a child and there is a need for further research to determine modifiable risk factors for this and other complications.p