Subtypes of ischemic stroke in children and young adults

Although the incidence of ischemic stroke (IS) increases with age,¹ stroke in children and young adults is not uncommon. The incidence of cerebrovascular disease (both ischemic and hemorrhagic) in children has been reported as 2.5 cases per 100,000 per year.² A more recent study in a metropolitan population of children aged ≤15 years found the incidence of IS to be 1.2 per 100,000 per year, slightly less than that for intracranial hemorrhage.³ The incidence of IS in young adults has been estimated at 14 to 62 cases per 100,000 per year.^4-7 Because it occurs during prime educational and work years, IS in children and young adults can be especially disabling. Before we can adequately address the issues of primary and secondary prevention of IS in the young, the causes of stroke in this age group must be accurately identified. In the present study, using accepted criteria, we undertook to determine and compare the subtypes of IS in children and young adults.

Methods. To identify children with IS, hospital records from 1980-1995 at the James Whitcomb Riley Hospital for Children (RH) were searched with use of the following ICD-9 codes for up to five discharge diagnoses: 342 (hemiplegia), 346 (migraine), 430 (subarachnoid hemorrhage), 431 (intracerebral hemorrhage), 432 (other and unspecified intracranial hemorrhage), 433 (occlusion and stenosis of precerebral arteries), 434(occlusion of cerebral arteries), 435 (transient cerebral ischemia), 436(acute but ill-defined cerebrovascular disease), 437 (other and ill-defined cerebrovascular disease), 438 (late effects of cerebrovascular disease), 767.0 (subdural and cerebral hemorrhage, fetus or newborn), 772.1(intraventricular hemorrhage, fetus or newborn), and 772.2 (subarachnoid hemorrhage, fetus or newborn). Records for all children from birth to age 18 years were reviewed; children aged 1 to 18 years with IS were included for this analysis. Children from birth to age <12 months, children with intracranial hemorrhage or hemorrhagic stroke, and children with preexisting cerebral palsy or developmental delay were excluded. Total number of admissions for 1981 to 1995 was used. We estimated the number of admissions for 1980 using the average number of hospital admissions from 1981 to 1985.

IS subtype was determined with use of the TOAST criteria.⁸ Based on this criteria, IS was classified as(1) large-vessel atherosclerosis (AT), (2) cardioembolic (CE), (3) small-vessel disease (SV), (4) other determined cause, or (5) unknown cause. IS could be classified as "unknown cause" if the cause was truly unknown or if there were more than one possible cause and the single most likely cause of IS could not be determined. Diagnostic evaluation of the children and young adults was at the discretion of the attending physician.

The young adults with IS were identified from the stroke registries at Northwestern University and Indiana University. The Northwestern University registry was compiled from August 1991 to June 1994. The Indiana University registry was begun in July 1994 and is ongoing. All young adults aged >18 to 45 years from these two registries made up the comparison group. Patients in the Indiana University registry are seen at one of the three adult hospitals (Indiana University Hospital, Roudebush Veterans Affairs Medical Center, or Wishard Memorial Hospital) on the Indiana University Medical School campus. All patients in both registries have the IS subtype recorded at entry into the registry.

Differences in subtypes of IS between children and young adults were compared using the χ² test or Fisher's exact test. Between-group age differences were compared using Student's t-test.

Results. From 1980-1995, there were 111,263 admissions to RH. Records of 1,311 children from birth to age 18 were reviewed; 113 were cases of stroke in children aged 1 to 18 years, and 92 of these were IS. Mean age of the children was 7.08 ± 4.67 years; 55% were boys, and 76% were white. One hundred sixteen young adults aged >18 to 45 years were identified from the registries; mean age was 36 ± 8 years; 52% were boys, and 86% were white (table 1).

Children had significantly fewer AT and more unknown causes of IS (seetable 1). Subtype distributions (percentages) in children/young adults were as follows: AT 0/16, CE 15/14, SV 0/3, other 49/44, and unknown 36/23. "Other" causes were divided into several categories based on frequently occurring diagnoses. In these common categories, children were more likely to have a prothrombotic state and less likely to have extracranial arterial dissection than were young adults. The majority of children with a prothrombotic state (18 of 23) had sickle hemoglobinopathy. None of the young adults with a prothrombotic state and IS had sickle-cell disease. IS secondary to SV was rare in both groups.

Specific causes of stroke in children and young adults are listed intable 2. In the children with CE stroke, 9 of 14 had cyanotic heart disease. Young adults had more heterogeneous causes of CE stroke, with 5 patients with right-to-left atrial shunting and 3 patients each with endocarditis and left atrial or ventricular thrombus. Eighteen of 23 children with prothrombotic disease had sickle-cell anemia, and none had antiphospholipid antibodies. No young adults had sickle-cell anemia, and 10 had antiphospholipid antibodies.

Nearly three-fourths of the children (24 of 33) with unknown cause of stroke had no other medical diagnoses. Other conditions in this group included migraine (2), polycystic kidney disease (2), and one patient each with aseptic meningitis, remote history of encephalitis 9 years before, dysgerminoma, medulloblastoma, and treated acute lymphoblastic leukemia with normal blood counts. In the young adults, 16 of the 27 with unknown cause of stroke had no other diagnoses. Possible causes of stroke in the remainder included: estrogen use (4), migraine (3), mitral valve prolapse (2), stroke during pregnancy with no laboratory evidence of hypercoagulability or venous thrombosis (1), and positive HIV serology (1).

To better define the age boundaries for defining childhood and young adult-stroke, we compared the causes of stroke in patients aged 15 to 18 years with those aged 1 to <15 years and those aged >18 to 45 years(table 3). None of the children aged 15 to 18 years had cyanotic heart disease. The three cervicocephalic dissections in children all occurred in the older age group. One patient each from the groups with sickle-cell anemia and moyamoya disease had IS at age 15 to 18 years.

Discussion. Few large studies of childhood stroke have been published,^2,9-14 and even fewer have been collected since the advent of MRI.^10-14 The Strokes in Children Study Group reported a series of patients with acute hemiplegia in 1973.⁹ Of the 60 patients, 27% had occlusive vascular disease, 17% had cardiac disease, 8% had sickle-cell anemia, 7% had a vascular malformation, and 42% were of unknown cause. A 10-year review (1965 to 1974) at the Mayo Clinic found only 38 cases of ischemic stroke, 4 of whom had preexisting hemiparesis and seizures or mental retardation.² Over 50% of these children had preexisting heart disease. Lanska et al.¹⁰ reported a series of childhood stroke patients from 1976-1988, including 17 patients aged 1 to 18 years. This series is also weighted toward CE causes of stroke: 11 of the 17 patients older than 1 year had either cyanotic heart disease (9) or coarctation of the aorta (2). Satoh et al.¹¹ reported a series of 54 children with ischemic cerebrovascular disease (TIA, reversible ischemic neurologic deficit, or stroke); 38 had completed stroke. Fifty-one of these children had no risk factor for stroke. Conditions possibly associated with stroke included 10 with infection or fever of unknown origin and 10 with head trauma. All children in this series had head CT, and those with moyamoya disease were excluded. Another series¹² was reported of 14 Japanese children aged 11 months to 9 years with basal ganglia infarcts on MRI, but no definitive information about causes of the infarcts was provided. A French series¹³ of 35 children aged 2 months to 17 years found 17 cases (49%) with specific causes, including 7 with sickle-cell disease, 4 with a hypercoagulable state, 4 with preexisting hypertension, 2 with moyamoya, 2 with AIDS, 1 with mitochondrial encephalopathy with lactic acidosis and stroke (MELAS), and 1 with cyanotic heart disease. Methods of case ascertainment were not specified in this report.

One prior report¹⁴ compared the causes of ischemic stroke in patients aged 6 months to 39 years. This study described 22 children (6 months to 18 years) and 78 young adults (19 to 39 years). Causes of stroke were categorized as structural/anatomic, presumed hypercoagulable, infectious/inflammatory, no cause, or migraine. This series reports no prothrombotic causes of stroke; the authors note that no children had sickle-cell anemia or stroke in their record review. As with our patients, more children than young adults (36% versus 27%) had no identified cause of stroke.

The young adults in our registry appear similar to other reported series of young adults with IS. These patients were seen at two different institutions yet did not have significantly different causes of stroke.¹⁵ Our young adults had causes of ischemic stroke in similar proportions to those in other large published series.^16-21 In a recent analysis of the Iowa Young Adults Stroke registry,¹⁸ TOAST subtyping of all cases of IS showed 22% AT, 21% CE, 9% SV, 31% other (including 7% hematologic causes), and 16% unknown. Carolei et al.¹⁷ reported a series of 333 patients aged 15 to 44 years with IS (58%) or TIA (42%). They classified IS or TIA etiology as AT (33%), CE (19%), other (8%), unknown or undetermined(35%), and "mixed" (5%). Some patients with associated medical diseases (for example, syphilis and diffuse intravascular coagulation) were classified in the AT and CE groups; this may partly explain the smaller number of`other' causes identified. Patients with stroke were more likely than those with TIA to have CE and mixed causes of ischemia, whereas TIA patients were more likely to have an unknown cause of ischemia. No patients with moyamoya were reported in this series. In case series published since 1979, the causes of IS in young adults have been classified by Biller^22-24 as atherosclerotic (7 to 69%), cardiac (5 to 35%), dissection (1 to 22%), and vasculitic (1 to 6%).

Although the proportion of CE and other causes were similar in our children and young adults, the specific etiologies were different (seetable 2). Most of the children with CE stroke had cyanotic heart disease. The two most common conditions causing stroke of"other determined" etiology in children were sickle-cell disease and moyamoya. The majority of young-adult CE strokes were caused by right-to-left atrial shunting, usually through a patent foramen ovale. Young adults with"other determined" causes of stroke most frequently had carotid or vertebrobasilar dissection or antiphospholipid antibodies. No children with antiphospholipid antibodies or protein-C deficiency were identified.

The appropriate age division between childhood and young-adult stroke has not been consistently defined. Most studies of childhood stroke include children up to age 18 years,^2,9,10,12,14 whereas other reports of stroke in young adults include patients as young as 15,^3,15,16 16,^11,20 or 18 years.²⁵ In our patients, we found children aged 15 to 18 years have causes of stroke that are more similar to those in young adults (seetable 3). In the CE subtype, all children with cyanotic heart disease and stroke had their stroke before age 15. In the "other determined" subtype, all three dissections in children were in the 15- to 18-year-old group, and 17 of 18 children with sickle-cell disease had their stroke before age 15. Although our series includes only 11 children aged 15 to 18 years, it appears that a lower age boundary of 15 years in young-adult stroke series may be clinically appropriate.

No prior study has attempted to systematically subtype, using validated clinical criteria, the causes of IS in children. We used the TOAST criteria to estimate the proportion of cases in each subtype from the previously published series (table 4). The Mayo Clinic,² and Case Western Reserve¹⁰ series both report unusually high proportions with CE stroke. This may reflect a referral or ascertainment bias for children with congenital heart disease at these centers. The relatively smaller proportion of other prothrombotic causes likely results from either few African Americans in the study population² or exclusion of patients with sickle-cell anemia from the series.¹⁰ Other, identified causes of stroke were present in 0 to 46% of these cases; our finding of 49% is comparable with the findings of 46%¹³ and 41%,¹⁴ and is slightly higher than the 33%³ reported in the three most recent series. The proportion of stroke with unknown cause varies from 34 to 94%. Our finding of 36% with unknown cause in our larger and more recent series is thus within the range of previously published reports.

Children were less likely than young adults to have a specific cause of stroke identified. This may result in part from the longer time period over which our childhood stroke cases were collected. The potential role of antiphospholipid antibodies in stroke was unknown at the beginning of the childhood stroke study period (1980). The lack of childhood cases of antiphospholipid antibody-related strokes may thus either reflect the earlier study period or suggest that this disorder is more strongly associated with stroke in young adults. Activated protein-C resistance and MELAS were also unrecognized early in the period of ascertainment of childhood cases. We attempted to minimize any potential bias of different neuroimaging techniques in the children and young adults by choosing a time period for review of childhood cases in which both CT and MRI were available.

Although RH is the only pediatric hospital in Indiana, there are other hospitals with pediatric neurologists and pediatric intensive care units in the greater Indianapolis area; therefore, an incidence rate of ischemic stroke in children cannot be determined from our cases. Like any academic hospital-based series, our cases are likely to have a referral bias and thus may not be representative of the entire population of children with ischemic stroke.

In summary, the cause of most childhood stroke can be categorized into CE, other, or unknown subtypes. The TOAST criteria can be successfully applied to children, but the proportion of children with large-vessel atherothromboembolic or small-vessel disease is likely quite low. The cause of stroke is less often identified in children than it is in young adults. Specific prothrombotic diseases such as sickle-cell anemia are more likely to cause stroke in children than in adults; cervicocephalic dissection, antiphospholipid antibodies, and large-vessel atherothromboembolic disease are the more likely causes of stroke in young adults. Based on the similarity of stroke etiologies in the 15- to 18-year age group and the >18- to 45-year age group, the age range of 15 to 45 years should be used for studies of young adults with ischemic stroke. Causes of stroke that are more typical in young adults should be considered in children aged 15 years and older who have an ischemic stroke.