|Home | About | Journals | Submit | Contact Us | Français|
This review will focus on the strengths and limitations associated with the current standard of care for primary prevention of ischaemic strokes in children with sickle cell anaemia (SCA) - transcranial Doppler ultrasound (TCD) screening followed by regular blood transfusion therapy when TCD measurement is above a threshold defined by a randomized clinical trial (RCT). The theoretical basis for potential alternative strategies for primary prevention of neurological injury in SCA is also discussed. These strategies will include, but will not be limited to: immunizations to prevent bacterial infections, particularly in low income countries; management of elevated blood pressure; and targeted strategies to increase baseline haemoglobin levels with therapies such as hyroxycarbamide or potentially definitive haematopoietic stem cell transplant.
Stroke is one of the most serious complications of sickle cell anaemia (SCA), permanently affecting cognition. Given the importance of reaching developmental milestones, and the necessity of formal education in childhood, the lifelong adverse impact of strokes in children is tremendous. Of note, the term SCA in this review refers to children with SCA (HbSS) and HbS-β0 thalassaemia (HbSβ0). Children with SCA have the one of the highest rates of ischaemic stroke of any group of children, with an incidence of 240 per 100,000 person-years in a retrospective cohort in the Baltimore-Washington area(Earley, et al 1998). This rate compares to an incidence of 2.3 per 100,000 in an unselected population of children from a California hospital discharge database(Fullerton, et al 2003) and 7.9 per 100,000 children from a prospective stroke registry in Dijon, France in which children with stroke were of Caucasian or Asian ethnicity(Giroud, et al 1995). By six years of age, at least 25% of children with SCA will have a covert or silent cerebral infarction (SCI)(Kwiatkowski, et al 2009) and this may reach 37% by 14 years of age(Bernaudin, et al 2005). By definition, SCI is not clinically apparent, but is detected by magnetic resonance imaging (MRI) of the brain via increased signal intensity that is 3 mm or larger, seen on at least 2 views by T2-weighted or fluid attenuated inversion recovery (FLAIR) MRI sequences, with no history or physical examination findings to suggest a focal neurological deficit that corresponds to the location of the lesion(Casella, et al 2010). Only one study to-date has reported SCI incidence in children. In the Cooperative Study of Sickle Cell Disease (CSSCD), the incidence of SCI in children who had normal MRI of the brain at study entry was 1.01 (95% confidence interval (CI): 0.4–2.7) per 100 patient-years(Pegelow, et al 2002). In children who had pre-existing SCI at study entry, the incidence of new SCI was higher, 7.06 (95% CI: 4.2–11.8) per 100 patient-years. Additional data on SCI incidence will be available when the final results of the Silent Infarct Transfusion (SIT) Trial are published(Casella, et al 2010). SCI is associated with lower intelligence quotient (IQ) when compared to children with normal MRI examinations of the brain and sibling controls, but higher when compared to children with overt strokes and SCA(Armstrong, et al 1996, Schatz, et al 2001). The most important series of studies regarding primary prevention of overt strokes in children with SCA involves the use of transcranial Doppler ultrasound (TCD) to identify the subgroup of children most likely to have overt strokes, followed by chronic blood transfusion therapy(Adams, et al 1998, Adams and Brambilla 2005). This review focuses on the strengths and limitations associated with the established strategy for primary prevention of ischaemic stroke in children with SCA – TCD screening followed by regular blood transfusion therapy -and discusses the basis for alternative strategies for primary stroke and SCI prevention that are associated with fewer complications and burden for patients and their families.
The underlying pathophysiology of stroke in children with SCA is probably altered cerebral haemodynamics, secondary to anaemia and intrinsic properties of sickle red blood cells and their interactions with the endothelium, resulting in an insufficient supply of oxygen to select regions of the brain to meet the demand(DeBaun, et al 2006). Multiple studies using different techniques have demonstrated hyperaemia of the brain in patients with SCA(Herold, et al 1986, Prohovnik, et al 2009). This increased blood flow to the brain is a form of cerebral autoregulation, to increase oxygen delivery and maintain a constant perfusion pressure in all patients with anaemia, not just individuals with SCA(Kuwabara, et al 1990, Prohovnik, et al 1989, Strandgaard and Paulson 1984). Oxygen delivery is primarily dependent on the cardiac output and oxygen content of the blood. Thus, the chronic haemolytic state in persons with SCA results in a corresponding low haemoglobin concentration and oxygen content; this places all individuals with SCA at increased risk for cerebral ischaemia. Clinical evidence to support this observation is based on the association between overt strokes and baseline haemoglobin levels; the lower the baseline haemoglobin level, the higher the risk of developing a stroke(Ohene-Frempong, et al 1998). Further, acute drops in haemoglobin levels are temporally associated with both overt(Wierenga, et al 2001) and silent cerebral infarcts (Dowling, et al 2010a). In both of these studies, the drop in haemoglobin was rather dramatic. In 10 children with parvovirus and neurological symptoms, Hb dropped an average of 33 g/l, the largest drop in Hb was 80 g/l and lowest 10 g/l (Wierenga, et al 2001). Four of the 10 cases of acute SCI reported had acute anaemia with Hb of 22–34 g/l (Dowling, et al 2010a).
There are also imaging studies that show that the response to anaemia, increased cerebral blood flow (CBF) and increased cerebral blood flow velocity (CBFV), are inversely correlated with neurocognitive function in children with SCA(Strouse, et al 2006a). In a recently published study, blood oxygenation level-dependent (BOLD) and cerebral blood flow-based functional magnetic resonance imaging were used to measure primary visual cortex responses to photic stimulation in 23 children with SCA. Interestingly, BOLD signal amplitude was positively associated with Wechsler Abbreviated Scale of Intelligence scores (Zou, et al 2011). The authors postulate that, although preliminary, this work suggests that patients with SCA are unable to mount a good haemodynamic response to support increased neuronal activity and that this may be the basis for cognitive dysfunction in SCA. Haemoglobin concentration is the primary determinant of, and inversely correlated with, TCD velocity(Brass, et al 1991, Pavlakis, et al 2010) and an elevated TCD is the strongest known risk factor for stroke in SCA(Adams, et al 1992a). Together, these studies strongly suggest that chronic anaemia, coupled with acute drops of baseline low haemoglobin levels are potentially modifiable risk factors for primary stroke prevention. See table 1.
Another risk factor for strokes in SCA is the presence of intracranial cerebral vasculopathy that compromises cerebral blood flow and probably compounds the risk of chronic anaemia. The cause of cerebral vasculopathy in children with SCA is not known, but pathologically, endothelial damage with intimal thickening and/or smooth muscle proliferation are seen within the larger arteries, with fibrin deposition, inflammatory changes and thrombus formation(Merkel, et al 1978, Rothman, et al 1986). Pathologically, this is quite different from the typical causes of cerebral vasculopathy in adult stroke, where there is intracranial atherosclerosis due to cholesterol-containing plaques(Qureshi, et al 2009). The most common locations for the vascular changes seen in SCA are branch points in the cerebral circulation, such as the supraclinoid (distal) ICA as it branches into the anterior cerebral and middle cerebral arteries(Kandeel, et al 1996).
The onset of the most severe form of cerebral vasculopathy, moyamoya, seems to precede overt strokes in about 2/3rd of the cases(Dobson, et al 2002, Gebreyohanns and Adams 2004, Hulbert, et al 2011). Conceivably, prevention of the abnormalities, such as occlusion of the internal carotid arteries, which lead to the development of moyamoya, would decrease the absolute risk of stroke in this vulnerable population. Determinants of severe cerebral vasculopathy in SCA are not known; this is an understudied area in SCA. Data are needed on the prevalence of cerebral vasculopathy in SCA, as well as the sensitivity and specificity of vasculopathy in predicting stroke risk. Further studies of vasculopathy and moyamoya in SCA, including their genetics, are needed. Now that improved bio-repositories are available, genetic questions may be more approachable.
Several case series have identified extracranial vasculopathy as a potential risk factor for stroke in some children with SCA. Prospective, controlled studies have not been done so it is not clear whether all children with SCA should be screened or if extracranial vasculopathy is a risk factor for stroke in SCA. Gorman et al (2009) studied 131 children with SCA with a standard TCD protocol and additional submandibular views to image the extracranial carotid arteries and concluded that mean flow velocities >160 cm/s indicated extracranial internal carotid artery (ICA) stenosis. Four children were identified with abnormal velocities in this region that were confirmed as stenosis on MRA of the neck or cerebral angiography. Interestingly, all 4 children had SCIs in a watershed distribution and evidence of cognitive delays. Deane et al. (2010) performed a retrospective analysis of extracranial ICA velocities in 236 children with SCA with and without a history of stroke. Fourteen children (5.9%) had tortuous extracranial internal carotid arteries and 13 (5.4%) had stenosis or occlusion. None of the children with tortuous vessels but 8 of those with stenosis had previously had a stroke; the presence of stenosis was strongly associated with overt clinical stroke (odds ratio [OR] 35.9, 95% CI 9.77–132, P<0.001). In 6 children, extracranial stenosis was part of extensive intracranial vasculopathy, but in 2 there was no evidence of intracranial disease. Stenosis seemed to be more common in older children. The authors concluded that extracranial ICA stenosis was strongly associated with stroke in SCA and might explain some cases of stroke in the absence of overt intracranial vasculopathy. Finally, Telfer et al. (2011) reviewed cervical MRA studies performed prospectively in 67 patients (55 children), for clinical indications including intracranial TCD abnormalities, acute or previous arterial ischaemic stroke (AIS). Cervical ICA lesions were seen in 10 patients including 2 adults (15%). Radiological features in 7 cases were consistent with arterial dissection. Of these, 4/7 patients presented with AIS.
The relevance of patent foramen ovale (PFO) as a risk factor for stroke in children with and without SCA, the role of PFO in paradoxical embolization, and how this should affect treatment decisions is not clear, based on a recent systematic review(Dowling and Ikemba 2011). A recent retrospective case series also looked at the presence of intracardiac shunt in children with SCA in stroke and showed that 25% (10 of 40) had PFO compared with 10–15% of the general population(Dowling, et al 2010b). This suggests the presence of PFO may be a risk factor for stroke in children with SCA, but further investigation is needed.
Acute infection can also be temporally associated with strokes in SCA. There is a link between infection and stroke in children without SCA as well, and this is an active area of paediatric stroke research in children without SCA. Infection may elevate stroke risk via increased blood viscosity, in association with increased white blood cell counts, fever, dehydration, and perhaps an increase in oxygen demand by the brain and direct vascular injury. Stroke after significant infections, such as varicella, sepsis and meningitis has been recognized for many years in children without SCA(Fullerton, et al 2007, Sebire, et al 1999). In children with SCA, there is a clear relationship between infection (bacterial meningitis) and strokes(Chang, et al 2003, de Montalembert, et al 1993, van der Plas, et al 2011). In countries where infants receive standard immunizations, such as conjugated and unconjugated pneumococcal vaccines and conjugated H. Influenza type B vaccine, along with penicillin prophylaxis(Bjornson, et al 1996), the prevalence of bacterial meningitis with subsequent strokes has dropped precipitously(Hord, et al 2002, Thigpen, et al 2011). However, in patients in developing countries where standard care does not include penicillin prophylaxis and these immunizations, and where malaria is endemic, bacterial meningitis is more common and cerebral malaria may be associated with strokes(Idro, et al 2005). For low income countries, the easiest strategy to prevent many strokes may be to implement a targeted immunization programme for children with SCA(Chintu, et al 1983, Serjeant 1985) and to provide penicillin prophylaxis(Riddington and Owusu-Ofori 2002). Recommended immunizations in children with SCA, which may prevent bacteremia, meningitis and stroke, include: conjugated 13-valent pneumococcal vaccine, conjugated H. Influenza type B vaccine, and for children over 2 years of age, unconjugated 23-valent pneumococcal vaccine and the meningococcal vaccine(Reinert, et al 2007, Vernacchio, et al 1998). Other standard childhood immunizations are also recommended. In addition, patients with acute infections that result in precipitous drops in haemoglobin or the development of acute chest syndrome are also at increased risk of strokes(Ohene-Frempong, et al 1998, Vichinsky, et al 2000).
A nested case-control study from the Kaiser Paediatric Stroke Study found an association between minor infections and ischaemic stroke in children without SCA (Hills et al 2009). The authors compared non-neonatal cases with paediatric arterial ischaemic stroke (n=97), with 3 controls matched for birth year and primary care facility(Hills, et al 2009). Exposure was defined as an outpatient visit for a minor infection up to 4 weeks prior to an ischaemic stroke or a visit over the same time period in controls. Visits for minor infection were documented in 26.6% of cases and 13% of controls (OR 2.93, 95% CI: 1.51– 5.68, p>0.001). An additional 9 cases and no controls had serious infections, such as meningitis or sepsis. The OR for any infection (major or minor) was 5.01, 95% CI: 2.64–9.54, p<0.0001). The multicentre Vascular Effects of Infection in Paediatric Stroke (VIPS) cohort study in children 2–17 years with arterial ischaemic stroke will test the hypotheses that (1) infection can lead to childhood arterial ischaemic stroke by causing vascular injury and (2) resultant arteriopathy and inflammatory markers predict recurrent stroke(Fullerton, et al 2011). Hopefully, the insights from this study can ultimately be generalized to other populations, including SCA.
Hypertension in SCA is associated with both overt stroke and SCI. Hypertension in SCA is difficult to classify, as, after adjusting for age and gender, children with SCA have lower blood pressure than children with normal haemoglobin levels(Pegelow, et al 1997). Rather than hypertension, the more appropriate term is relative high blood pressure, where blood pressure is elevated above baseline in a patient with low blood pressure, or relative to others with SCA. Relative high blood pressure in children with SCA is a risk factor for overt stroke(Pegelow, et al 1997, Rodgers, et al 1993) and SCI(DeBaun, et al 2011). As with all individuals with anaemia, in order to maintain a constant cerebral perfusion pressure, patients with SCA will increase cerebral blood flow and systolic blood pressure (SBP) in response to decreased cerebral perfusion(Strandgaard and Paulson 1984); therefore, relative high blood pressure may indicate decreased cerebral perfusion or perhaps cerebral vasculopathy. In the CSSCD, Pegelow et al. (1997) examined the relationship between an elevated SBP measurement and multiple clinical outcomes, including overt stroke. In this prospective study of over 4,000 individuals, the research team demonstrated that risk of stroke was related to the magnitude of the SBP, although no absolute threshold or percentile was identified that conferred an increased risk of an overt stroke and the presence or absence of cerebral vasculopathy was not assessed. In the Silent Cerebral Infarct Transfusion (SIT) Trial, investigators sought to identify risk factors associated with SCI. In this cross-sectional study, children with SCA between the ages of 5 and 15 years with no history of overt stroke or seizures were screened with MRI for SCI; 814 children with complete data were included. SCI was diagnosed in 30.8% (251 of 814) participants(DeBaun, et al 2011). In a multivariate logistic regression (MLR) analysis of SIT trial data, lower baseline haemoglobin concentration, (p < 0.001), higher baseline SBP, p = 0.013 and male gender, p = 0.026, were statistically significantly associated with an increased risk of a SCI. In this MLR analysis, the reference category for haemoglobin was > 85 g/l; and the reference category for SBP was < 104 mmHg. Odds ratios for SCI in the first tertile of haemoglobin (< 76 g/l) were 2.12 [95% confidence interval (CI) 1.45 to 3.10, p < 0.001]. Odds ratios in the second tertile (76 to 85 g/l) were 1.19 (95% CI 0.81 to 1.76, p = 0.371). Odds ratios for SCI in the second tertile for SBP (104 to 112 mmHg) were 1.55 (95% CI 1.06 to 2.27, p = 0.025); for the third tertile (> 112 mmHg) odds ratios were 1.73 (95% CI 1.18 to 2.54; p = 0.005). No interaction was detected between baseline haemoglobin and SBP (p = 0.90). Clearly, lower haemoglobin concentration and higher SBP are risk factors for SCI in children with SCA and may be therapeutic targets for decreasing the risk of SCI. Pegelow et al. (1997) demonstrated that the 90th percentile for SBP in children with SCA aged 5–14 years was approximately 110. The impact of treating relatively elevated SBP in patients with SCA is not known, and warrants a clinical trial to determine if this strategy is effective in primary stroke prevention. At the present time, we recommend further evaluation of children with SCA and systolic blood pressure above this level. At the very least, individuals with an absolute hypertension based on norms for the general population should be appropriately treated.
The potential benefit of TCD screening for cerebral vasculopathy and stroke risk in children with SCA gained evidence as a potential intervention in the early 1990s. Adams et al. (1992a) screened 190 asymptomatic children with SCA with non-duplex (non-imaging) TCD and found in clinical follow-up that a time-averaged mean of the maximum velocity (TAMMX) in the middle cerebral artery (MCA) of 170 cm/s or greater was associated with stroke. The same group compared TCD to cerebral angiography in 33 neurologically symptomatic patients and found five criteria for cerebrovascular disease: (1) TAMMX of >190 cm/s, (2) low velocity in the MCA, <70 cm/s, (3) side-to-side MCA velocity ratio <0.5, (4) anterior cerebral artery (ACA)/MCA ratio >1.2 on the same side, (5) inability to detect an MCA in the presence of a good ultrasound window(Adams, et al 1992b). Using duplex (imaging) TCD, Seibert et al. (1993, 1998) expanded on this work and found several additional factors that were significant in the identification of patients at risk. These included: (1) maximum velocity in the posterior cerebral artery (PCA), vertebral or basilar arteries greater than the maximum velocity in the MCA; (2) maximum velocity of the ophthalmic artery (OA) >35 cm/s; (3) resistive index in the OA of less than 50; (4) velocity in the OA greater than that of the ipsilateral MCA(Seibert, et al 1998, Seibert, et al 1993).
The seminal work by Adams et al. (1998) and others led to the Stroke Prevention Trial in Sickle Cell Disease (STOP), a RCT of primary stroke prevention in SCA. Children with HbSS and HbSB0 aged 2–16 years with no history of stroke were screened with TCD for an elevated TAMMX, which was felt to be a sign of cerebrovascular disease. An elevated TAMMX in either the terminal portion of the ICA or the proximal portion of the MCA of 200 cm/s was required for random allocation to standard care (no transfusions) or transfusion with regular red cell infusions sufficient to lower and maintain haemoglobin (Hb S) at <30% of total Hb. The primary endpoints of cerebral infarction or intracerebral haemorrhage were determined by an adjudication panel, without knowledge of the treatment status of the patient and the incidence of stroke in the two groups was compared using survival analysis. From 1,934 children screened at 14 sites, 130 children with a mean age of 8.3±3.3 years were randomized to transfusion (n=63) or standard care (n=67). Based on the intention-to-treat analysis, 1.6% (1 of 63) and 16% (11 of 67) in the treatment and observation groups, respectively, developed strokes. Regular blood transfusion therapy was associated with a relative risk reduction of 92% (P=0.0009) for stroke, and absolute risk reduction of 14%. Thus, for every one stroke prevented, seven children were required to receive regular blood transfusion therapy, corresponding to a number needed to treat of seven.
After the completion of this landmark study, the most pressing question was whether individuals with an elevated TCD measurement could stop blood transfusion. STOP II was a second RCT that aimed to assess when it might be safe to discontinue blood transfusions in children receiving them for primary stroke prevention, if their TCDs normalized on transfusions(Adams and Brambilla 2005). After approximately 30 months on transfusion and normalization of TCD velocities, 79 children were randomized to either continuing or halting transfusion. In the group that discontinued transfusion, 34% (14 of 41) developed high-risk TCD velocities again. An additional 5% (2 of 41) of the children had stroke; these 2 children had a single abnormal TCD velocity (>200 cm/s) but stroke occurred prior to confirmatory TCD. Neither of these problems occurred in the 38 children who continued on regular transfusion.
Two different TCD techniques are available, imaging (duplex) TCD and non-imaging (non-duplex) TCD. As the STOP study used non-imaging TCD, this technique is now the reference standard, though imaging TCD is often what is available in radiology departments(Adams 2005). The non-imaging pulsed Doppler TCD technique used in the STOP study has several advantages: low cost in comparison to duplex imaging systems, portability, and ability to document mean velocities greater than 300 cm/s, allowing the assessment of severe intracranial stenosis(Bulas 2005). Also, the small lightweight transducer is easier to manipulate, particularly with small temporal bone ultrasound windows in children. Limitations of the non-imaging TCD include inability to visualize intracranial anatomy, which can make vessel identification difficult and/or inaccurate. The non-imaging technique can also be hard to learn and is generally not the imaging modality of choice in paediatric radiology departments(Bulas 2005). The examination includes manually measuring the velocities, which is labour-intensive and might be less accurate when performed by inexperienced personnel. In contrast, imaging TCD is available on most ultrasound machines with 2-MHz transducers and colour Doppler capability. One advantage of this technique is that most radiology department ultrasonographers are already trained to use this equipment(Adams 2005). The intracranial vessels are visualized, providing increased confidence in vessel identification. The colour image can document vascular anatomy, flow direction, branch patterns, and vessel tortuosity.
There have been several studies correlating the same-day non-imaging TCD STOP protocol with velocity data from various imaging TCD systems to assess for comparability(Bulas, et al 2000, Neish, et al 2002). In most, but not all studies, TAMMX velocities in the MCA and ICA using imaging TCD were 10% lower than TAMMX velocities in similar studies using non-imaging TCD(Bulas, et al 2000, Neish, et al 2002). For imaging TCD to be comparable to the STOP non-imaging TCD classification of abnormal at 200 cm/s TAMMX velocity, some authors have recommended using a treatment threshold of 185 cm/s with imaging TCD machines(Jones, et al 2005). Despite several studies, the appropriate thresholds for treatment for imaging TCD remain a point of some uncertainty.
Based on the STOP study, the National Heart Lung and Blood Institute (NHLBI) recommends that children with SCA between 2 years and 16 years of age have TCD screening every 6 months (National Institutes of Health, National Heart, Lung, and Blood Institute [NHLBI], 2002). Alternatively, the American Heart Association guidelines state: “In children with SCA, it is reasonable to repeat a normal TCD (<170 cm/s) annually and to repeat an abnormal study (>200 cm/s) in 1 month. Borderline and mildly abnormal TCD studies may be repeated in 3 to 6 months”(Roach, et al 2008). Wang (2007) suggested a repeat TCD in 3 months for individuals with a blood flow velocity of 185 to 199 cm/s and every 6 months for those with flow velocities in the range of 170 to 184 cm/s. True evidence-based guidelines for the frequency of TCD screening are not available. This NHLBI recommendation is based on the STOP trial, which represents the best available data, but this study was not designed to test or determine the optimal interval for screening. As is the case after any clinical efficacy trial, the effectiveness of the new intervention should be assessed, in this case, specifically whether TCD measurement with subsequent blood transfusion therapy, in a setting other than a clinical trial, results in a decreased incidence rate of strokes. At least three studies have addressed the effectiveness of TCD measurement. In a retrospective cohort of Californian children, the stroke rate declined, from 0.44 per 100 person-years between 1993 and 1997 to 0.19 per 100 person-years post-1998(Armstrong-Wells, et al 2009). McCarville et al (2008) found a similar decline in stroke incidence after implementing TCD-screening in a single-centre cohort from 0.46 to 0.18 per 100 person- years. In a third single-centre retrospective cohort in Pennsylvania, the incidence of stroke declined 10-fold after TCD screening and prophylactic blood transfusions were implemented. In 475 children observed in the 8-year period before instituting TCD screening, the incidence of overt stroke was 0.67 per 100 patient-years. In the 8-year period after TCD screening began, in 530 children the overt stroke incidence rate was only 0.06 per 100 patient-years (P<0.0001)(Enninful-Eghan, et al 2010). In the only prospective study to demonstrate the effectiveness of routine TCD measurement at a paediatric haematology centre, Bernaudin et al. (2011) described their single centre cohort study. The group had employed routine TCD measurement since 1992 in a newborn SCA cohort in Créteil, France. These children (n = 217 HbSS or Hb Sβ(0)), were screened yearly with TCD beginning at age 12 to 18 months. MRI and magnetic resonance angiography (MRA) of the brain were performed every 2 years after age 5 years (or earlier, in case of abnormal TCD). A transfusion programme was recommended to patients with abnormal TCD and/or large vessel stenosis as detected by MRA (50% decrease in vessel diameter in the ICA, ACA, or MCA), hydroxycarbamide (HC) to symptomatic patients in absence of macrovasculopathy, and stem cell transplantation to those with a human leucocyte antigen (HLA)-geno-identical donor. Mean follow-up was 7.7 years (1609 patient-years). The cumulative risk of overt stroke by age 18 years was 1.9% (95% CI: 0.6%-5.9%). This is a dramatic reduction from the 11% reported in the Cooperative study(Ohene-Frempong, et al 1998). Also, in this very young cohort, 29.6% (95% CI: 22.8%-38%) had abnormal TCD, 22.6% (95% CI 15.0%-33.2%) had large vessel stenosis and 37.1% (95% CI: 26.3%-50.7%) had silent stroke seen on screening MRI/MRA by the age of 14 years. One limitation of this study is that MRA often over-estimates intracranial large vessel stenosis(Korogi, et al 1997). Thus, early TCD screening coupled with intensive preventative therapy seems to significantly reduce overt stroke-risk by 18 years of age for children with SCA, but silent infarcts and large vessel stenosis are still major issues.
Inherent in any successful trial is the tendency for clinicians to extrapolate the results to clinical situations that were not part of the actual trial. Given the paucity of RCTs in SCA, this tendency is understandable, but nevertheless, problematic. Participants in the STOP Trial were assessed at their baseline and not when ill. Specifically, TCD screening should not be performed during an acute illness to assess baseline stroke risk, as co-morbidities may alter the baseline measurements(Bulas 2005, Nichols, et al 2001). Hypoxia, fever, hypoglycaemia and worsening anaemia can also increase cerebral blood flow and flow velocity(Quinn, et al 2009). Although it is tempting to perform TCD examinations while a patient is in the hospital for a medical illness, the results might not be valid, if velocities fall in the abnormal or conditional range. Similarly, TCD measurements within 3 months after blood transfusion therapy are limited because increasing the haemoglobin with blood transfusion therapy will limit the meaning of the results.
The STOP TCD protocol required representative velocities (peak systolic, TAMMX, diastolic) in 15 arterial segments. Classification of TCD measurement is based on TAMMX velocities in only the MCA, bifurcation (BIF) and distal internal cerebral artery (dICA); clinical decision-making is based on the TAMMX velocities. Velocities in other arteries can be measured, but their clinical utility in predicting a future stroke, and the established benefit of providing regular blood transfusion therapy for abnormal velocities in vessels other than those used as criteria in the STOP Trial is unwarranted. Analysis of STOP data suggested that velocity elevation in the ACA did contribute to data from MCA/ICA arterial segments for risk prediction(Kwiatkowski, et al 2006). In addition, prior work suggests that elevated velocities in the posterior cerebral artery (PCA) may also identify children at risk. For example, if the maximum velocity in the PCA, vertebral or basilar arteries is greater than maximum velocity in the MCA, this may suggest a low or declining MCA velocity, such as may occur with progressive stenosis and occlusion of the MCA(Seibert, et al 1998). Given the nature of these studies, coupled with the absence of any prospective or systematic data that velocities that are above a pre-set threshold in the ACA and PCA confer definite risk of stroke, no reliable evidence supports use of these velocities as criteria for indefinite blood transfusion therapy.
Options for chronic transfusion therapy include simple transfusion, erythrocytapheresis, and exchange transfusion. Of these, it can be argued that erythrocytapheresis represents optimal therapy, as it reduces the degree of iron loading, in some cases averting even the need for chelation therapy(Sarode, et al 2011). Disadvantages include exposure to many more units of blood and donors and the need for reliable venous access, often mandating placement of temporary or semipermanent central lines and their attendant risks(Sarode, et al 2011). Except in specialized care situations, this option is not available to small children. Many centres have adopted partial exchange transfusion, with withdrawal of 5–10 μl/ml of the patient’s blood before transfusion of 14–18 ml/kg of packed blood cells(Lindsey, et al 2005). Given with additional fluid support as necessary, this procedure is generally well tolerated. While this makes theoretical sense in attempting to reduce iron burden, the degree of practical benefit from this procedure has not been rigorously demonstrated to date. In the setting of acute stroke, a retrospective study has suggested that exchange transfusion is associated with a reduced risk of recurrence of stroke long-term(Hulbert, et al 2006). As this is a retrospective study, it cannot be considered definitive; however, at this point, these data suggest that exchange transfusion or erythrocytapheresis should be considered the treatment of choice for acute stroke in SCA when feasible.
One of the major problems with transfusion as a method of primary stroke prevention in children with SCA is that regular simple transfusion inevitably causes iron overload in the absence of measures to counteract iron loading. An excellent review on iron-chelating therapy for transfusion-related iron overload has recently been published(Brittenham 2011). Briefly, when iron exceeds the body’s storage capacity and non-transferrin bound iron appears in the plasma, reactive oxygen species are generated and many tissues in the body are damaged. The major issues with iron chelation are high cost and, as confirmed by a recent systematic review(Lucania, et al 2011), few RCTs to demonstrate efficacy and safety of iron chelation have been performed. Table 2 lists treatment options for primary stroke prevention in SCA.
At this time, only haematopoietic stem cell transplantation (HSCT) offers potentially curative therapy for patients with SCA. Current myeloablative treatment protocols allow the cure of 72–96% with SCA, depending on disease status at the time of transplantation(Hsieh, et al 2011). For the best chance at transplant success, a HLA-matched family donor is needed, but is only available in 14–18% of children with SCA. The major risks of transplant include infection, graft-versus-host disease (GVHD), failure to engraft, and death(Hsieh, et al 2011). Early concerns about central nervous system (CNS) complications, such as seizures, associated with transplantation appear to have been ameliorated by aggressive treatment with platelets and anti-convulsant medication. Families and patients also report serious concerns about potential long-term side effects, such as sterility and growth failure, limiting the acceptability of this procedure(Kodish, et al 1991, van Besien, et al 2001).
The indications for HSCT are controversial. Criteria for inclusion in clinical trials have been stringent. One landmark multicentre study utilized the following criteria: stroke or CNS events lasting more than 24 h, acute chest syndrome with recurrent hospitalizations, acute painful crises or priapism, stage I or II sickle lung disease or sickle nephropathy, bilateral retinopathy or major visual impairment in at least one eye, osteonecrosis of more than one joint, or red cell alloimmunization(Walters, et al 1996). Some physicians believe that transplant should be offered prior to the occurrence of end-organ damage, such as stroke. While transplants have been demonstrated to be effective in secondary stroke prevention, scant data exist concerning the role of transplant for primary stroke prevention in patients with abnormal TCD(Bernaudin, et al 2011, Walters, et al 2010); the relative merits versus risks that include not only death but chronic GVHD have not been systematically studied. Patients have been offered transplant on the basis of arterial stenosis and abnormal arterial velocities as detected by TCD and confirmed by MRA of the brain. The largest series is from France, where 87 children received HSCT between 1988 and 2004(Bernaudin, et al 2011). Of these children, 21 underwent stem cell transplant for what could be considered primary prevention of overt stroke, 6 had arterial stenoses detected by TCD that were confirmed by conventional arteriography or MRA of the brain, 2 had persistently abnormal high arterial velocities despite regular transfusion, and 13 received transplants because of severe anaemia associated with cognitive deficiency and/or silent infarcts on MRI. Vasculopathy without overt stroke has not been an indication for HSCT in other studies, but this may be logical extension of primary stroke prevention in SCA(Bernaudin, et al 2011). In the future, it is possible that cerebral vasculopathy as defined by MRA may be an indication for primary stroke prevention. At present, optimal treatment remains unclear. Ultimately, the success of HSCT for SCA will depend on the development of reliable procedures, low toxicity and mortality, and the ability to make HSCT available to individuals without HLA-matched compatible donors. The ability to rely on immunotolerance rather than cytotoxicity to establish bone marrow replacement or successful chimerism could be considered the “holy grail” of HSCT.
Hydroxycarbamide (HC, also known as hydroxyurea) increases haemoglobin F levels in SCA, possibly because it is cytotoxic, and this toxicity causes regeneration of erythrocytes. HC metabolism may also lead to downstream increases in HBG gene expression(Cokic, et al 2003). In a multicentre trial of adults with SCA, HC reduced the incidence of acute chest syndrome and painful vaso-occlusive crisis by almost 50%(Charache, et al 1995). In follow-up studies, although random assignment was not maintained, cumulative mortality was reduced by 40% in patients who continued to take HC, and there was little risk of cancer or serious infection over 9 years of observation(Steinberg, et al 2003). In a study of HC in more than 100 children with homozygous SCA, Hb F increased to almost 20% and treatment effects were present for nearly seven years with continuous treatment, without clinically significant toxicity(Zimmerman, et al 2004). In a systematic review of HC in children, new and recurrent neurological events were decreased in 3 observational studies of HC compared with historical controls(Strouse, et al 2008). Recently, the results of the BABY HUG placebo controlled trial of HC demonstrated that this agent could be given to infants 9–18 months of age at initiation of therapy, with a reduction of painful episodes, acute chest syndrome, transfusions and hospitalization similar to that seen in adults, with no significant toxicities(Wang, et al 2011). Children were followed for 2 years for evidence of organ dysfunction and clinical complications.
Interestingly, although no differences was seen in mean velocities for the two groups pre- and post-treatment, in the group treated with HC, intrapatient TCD velocities in the MCA increased by 16%, whereas in the placebo group, TCD velocities increased by 27%, p=0.0002(Wang, et al 2011). This suggests that HC slows the progression of cerebral vasculopathy in very young children. Short-term effects from HC include transient myelosuppression that is dose-dependent, typically causing mild neutropenia(Zimmerman, et al 2004). The major long-term concern is that HC is mutagenic and potentially carcinogenic. Leukaemia and cancer have been reported in HC-treated SCA patients, but there is no evidence that the incidence is higher than in the general population(Steinberg, et al 2010). In a single centre cohort, a maximum tolerated dose of approximately 25 mg/kg/day of HC, sustained for a mean of 45 months, did not result in an increase in the number of DNA mutations(Zimmerman, et al 2004). Finally, a recent paper looked at chromosome breakage and repair in 51 children with SCA who were on HC for 3–12 years, compared to 28 children with SCA prior to any treatment with HC (McGann, et al 2011). Chromosome damage in peripheral blood mononuclear cells was less for children receiving hydroxycarbamide than untreated patients. There were no differences in repairing chromosome breaks after in vitro radiation. This work also provides support for the idea that the risk of mutagenicity due to HC is low.
Aspirin is an appealing option for primary ischaemic stroke prevention, as it is effective in adult stroke, inexpensive and widely available. The proposed mechanisms of action for stroke prevention in SCA are that aspirin impairs platelet function and exerts anti-inflammatory and anti-thrombotic effects. The roles of each of these mechanisms in the aetiology of stroke in SCA are not well defined. Thrombocytosis is common in SCA and in one series was associated with lower intelligence quotient(Bernaudin, et al 2000). Furthermore, platelet activation is associated with hypoxia, hypercoagulability, and inflammation(Inwald, et al 2000). Aspirin may have a beneficial effect on these and other mechanisms in children with SCA(Osamo, et al 1981).
Aspirin use and trials in children with SCA pose special concerns. For example, the risk of haemorrhagic stroke in children with SCA is elevated(Strouse, et al 2006b), and theoretically, aspirin use could increase this risk further or increase the severity of the cerebral haemorrhage. In addition, chronic ibuprofen use has been associated with diminished aspirin efficacy in adults(MacDonald and Wei 2006), and use of ibuprofen is routine in children to combat vaso-occlusive crisis pain. Finally, Reye syndrome, though rarely reported(Schror 2007), is a theoretical concern. There are no studies of primary prevention of CNS complications in children with SCA using aspirin. A formal trial assessing the benefits of aspirin use against the risks and complications, most notably cerebral haemorrhage, should be considered before aspirin is used routinely as standard practice for primary stroke prevention.
All studies to date that have evaluated revascularization procedures in SCA have done so as secondary stroke prevention in children. Children in these series typically had surgical revascularization with either indirect cerebral bypass methods, such as pial synangiosis, (Smith, et al 2009), or encephaloduroarteriosynagiosis(Hankinson, et al 2008, Hulbert, et al 2011). In the series by Smith et al. (2009), 11 of 12 children had evidence of prior ischaemic stroke on MRI. One child had a transient ischaemic attack (TIA), but no evidence of radiographic stroke(Smith, et al 2009). Hankinson et al. (2008) reported that 1 child in their series of 12 surgically treated children had moyamoya identified via TCD screening and had surgery for primary stroke prevention. The remaining 11 children underwent surgical revascularization after overt stroke (n=10) or TIA (n=1). As a part of larger, multicentre study of SCI in children with SCA on regular blood transfusion for secondary stroke prevention, Hubert et al. (2011) reported 4 children with SCA who had revascularization procedures for either progressive silent or overt strokes or TIAs despite blood transfusion. Of these 4 surgically-treated children, one had an overt stroke 5 days after the revascularization, one child continued to have TIAs but no overt stroke and the other 2 children remained free clinical or MRI evidence of cerebral ischaemia over 16 months and 4 years of follow up.
Currently, we do not have evidence that revascularization can prevent first stroke. Thus, there is no evidence for primary stroke prevention with revascularization in children with SCA and additional studies will be needed to define the role of surgery in preventing stroke in SCA.
The currently accepted standard of care for primary stroke prevention in SCA is TCD screening, followed by long-term transfusions when TCD indicates a high stroke risk; however, transfusion is costly and carries risk of iron overload and alloimmunization. Furthermore, the number needed to treat for abnormal TCD remains high, resulting in potentially unnecessary transfusion in a sizable number of children; therefore, additional and more refined methods of detection and treatment for the primary prevention of stroke are still required. Our personal opinion is that primary prevention needs to be studied further, and that additional clinical trials are warranted. True primary stroke prevention would prevent vasculopathy in children with SCA before it occurs, and this should be a future goal. At this time, further studies of HC for prevention of overt stroke, SCI and abnormal TCD are warranted, and RCTs in children are ongoing. Aspirin is appealing for its ease of use and low cost, though it is untested and may increase haemorrhage risk; at present, it cannot be recommended for primary prevention of stroke in SCA. Stem cell transplant holds promise as a cure, but may never be available to all patients; risks appear to be minimized with the use of a HLA-matched sibling donor, which has been estimated to be present in 14–18% of children with SCA. Also, families may not be accepting of the risks of stem cell transplantation, including infertility. In the future, we hope to see further refinement of stem cell transplantation and, perhaps, gene therapy. In the meantime, it may also be necessary to test combinations of less intense regimens, such as HC and aspirin, particularly for use in resource-poor nations, where long-term blood transfusion or stem cell transplantation are simply not feasible. TCD screening and transfusion to prevent stroke represents a major step forward in the treatment of stroke in SCD, but there is still much work to do.
LCJ is supported by NINDS K23NS062110. JFC is supported by NHLBI R34HL108756, R01HL091759, and U54HL090515. MRD is supported by NINDS U01-NS-042804 and NHLBI R01HL079937.
Author ContributionsDrs. Jordan, Casella, and DeBaun each drafted sections of the manuscript, all revised it critically, and approved the submitted and final versions.