Edited by: Andrei V. Alexandrov, University of Alabama Birmingham, USA
Reviewed by: Ru-Lan Hsieh, Shin Kong Wu Ho-Su Memorial Hospital, Taiwan; Bin Jiang, Beijing Neurosurgical Institute, China
*Correspondence: Lauren A. Beslow, Division of Neurology, The Children’s Hospital of Philadelphia, Faculty Fellow, Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine of the University of Pennsylvania, Colket Translational Research Building, 3501 Civic Center Boulevard, 10th Floor Room 10011, Philadelphia, PA 19104, USA. e-mail:
This article was submitted to Frontiers in Stroke, a specialty of Frontiers in Neurology.
This is an open-access article distributed under the terms of the
Arterial ischemic stroke (AIS) affects between 1.2 and 13 pediatric patients per 100,000 person-years in developed countries (Giroud et al.,
A recent study of neonates with perinatal AIS used a modification of the Alberta Stroke Program Early Computed Tomography Score (ASPECTS) on acute MRI for use in children (modASPECTS) to assess semiquantitatively infarct volume (Wusthoff et al.,
The study was performed with institutional review board approval.
This is a cross sectional study comparing modified pediatric ASPECTS on acute MRI to infarct volume as a percent of supratentorial brain volume (SBV) in a prospectively identified cohort of neonates and children with acute supratentorial AIS.
Subjects were identified from a single tertiary care center stroke registry. In the current study, perinatal subjects were ≥37 weeks gestational age with acute AIS within the first 28 days of life and presented between January 1, 2004 and October 31, 2009. Childhood subjects were age >28 days to 18 years inclusive presenting with acute AIS between January 2005 and November 2008. All subjects’ infarcts were confirmed by MRI demonstrating restricted diffusion within an arterial vascular territory conforming to the localization of acute neurologic deficit. For inclusion in the study, an axial diffusion weighted image (DWI) with apparent diffusion coefficient (ADC) map and an axial T2 image were required. The perinatal subjects and childhood subjects were previously described in two papers (Beslow et al.,
Infarct location was defined by the arterial territory or territories affected. Additionally, watershed infarcts were defined as infarcts in the borderzone between anterior cerebral artery-middle cerebral artery (ACA-MCA), middle cerebral artery-posterior cerebral artery (MCA-PCA), and internal deep MCA borderzone territories. Some children had watershed ischemia in other regions in addition to their defined vascular territory infarcts.
Infarct volume was measured on axial DWI for both perinatal and childhood subjects by Lauren A. Beslow. DWI was chosen since T2-weighted imaging may not demonstrate the full extent of the lesion if acquired soon after stroke onset, and lesions in the neonates are seen less well on T2 due to incomplete myelination and high water content of the brain. ADC maps were used to confirm areas of acute infarction. Previous adult studies have also used DWI to measure acute infarct volume (Lovblad et al.,
ASPECTS was designed to provide quick and reliable semiquantitative estimation of early ischemic middle cerebral artery territory infarction on CT in adults (Barber et al.,
For each hemisphere there are 15 regions yielding a maximum of 15 points per hemisphere, and the maximum total modified pediatric ASPECTS score is 30. During scoring, a region was scored as positive even if a small portion of a modified pediatric ASPECTS score region was involved, including watershed areas or punctate foci.
The raters were a board-certified pediatric neuroradiologist (Arastoo Vossough), a pediatric neuroradiology fellow (Hisham M. Dahmoush), and a pediatric stroke neurologist (Sabrina E. Smith). The three raters performed practice scoring on eight non-study subjects together but performed their ratings on study subjects independently and blinded to the other rater’s scores as well as to the volumetric measurements. Arastoo Vossough and Hisham M. Dahmoush used a clinical image viewing program (iSite, Philips, Netherlands) for viewing the images for their ASPECTS ratings while Sabrina E. Smith used ITK-SNAP for viewing the images for her ratings. Raters were permitted to confirm areas of infarction on DWI for scoring by inspecting the ADC maps.
Descriptive statistics were performed using frequency distributions and proportions for categorical variables and means with standard deviations or medians with interquartile ranges (IQRs) for continuous variables. The perinatal and childhood cohorts were analyzed separately since they represent distinct subgroups of pediatric stroke. Spearman rank correlation coefficient was used as a measure of criterion validity between infarct volume as a percent of SBV and the modified pediatric ASPECTS. A correlation coefficient of >0–0.3 was considered weakly positive, >0.3–0.7 moderately positive, and >0.7–1.0 strongly positive. Pre-specified subanalyses were done with the exclusion of subjects with only multifocal punctate infarcts who were likely to represent a subgroup of subjects in which the modified pediatric ASPECTS was not well correlated with stroke volume. Interrater reliability of the total modified pediatric ASPECTS was determined by the calculation of an intraclass correlation coefficient (ICC) using one-way analysis of variance. An ICC was considered moderate agreement if 0.41–0.60, substantial agreement if 0.61–0.80, and almost perfect (excellent) if 0.81–1.00 (Landis and Koch,
We further evaluated whether modASPECTS can reliably classify subjects as high-risk for poor outcome due to large infarct volume as, previously described, in which infarct volume ≥5% of SBV was predictive of HT and poor outcome (Beslow et al.,
Analyses were conducted using STATA version 11.1 (STATA Corporation, College Station, TX, USA).
There were 46 potential perinatal subjects with isolated supratentorial stroke previously described in our study evaluating later seizure occurrence after acute perinatal stroke (Wusthoff et al.,
There were 71 subjects (31 perinatal, 40 childhood). The median age of the childhood subjects was 3.7 years (IQR; 0.3–12 years). Forty-two subjects (59%) were male. Forty-four (62%) were white (2 Hispanic), 20 (28%) were black, 3 (4%) were mixed race, and 4 (6%) were unknown race or other race. Twenty-eight perinatal subjects (90%) presented with seizures. Thirty-nine childhood subjects (97.5%) presented with focal deficits, and 15 (37.5%) had seizures. HT was present in nine (29%) perinatal subjects and in five (12.5%) childhood subjects. HT was petechial in all perinatal subjects (seven punctate, two confluent), petechial in four childhood subjects (three punctate, one confluent), and small parenchymal in one childhood subject. Presence of HT did not affect the volumetric measurements or the modASPECTS ratings.
Of the perinatal subjects, 5 (16%) had a cardioembolic stroke, 1 of whom also had an intracranial vasculopathy, 3 (10%) had a thrombophilia, 1 (3%) had
The median absolute infarct volume on DWI for the perinatal and childhood stroke subgroups was 24.5 ml (IQR 7.8–62.9 ml) and 24.6 ml (IQR 4.5–106.2 ml), respectively. The median infarct volume as a percent of SBV in the perinatal subjects was greater than that for the childhood subjects (6.9%, IQR 2.4–17.2 vs. 2% IQR 0.4–9.8%,
Characteristic | Perinatal ( |
Childhood ( |
---|---|---|
Left | 14 (45.1%) | 20 (50%) |
Right | 7 (22.6%) | 14 (35%) |
Bilateral | 10 (32.3%) | 6 (15%) |
Isolated MCA | 20 (64.5%; 2 watershed*) | 29 (72.5%; 3 watershed*) |
Isolated PCA | 1 (3.2%) | 4 (10%; 1 watershed*) |
Isolated ACA | 0 (0%) | 1 (2.5%) |
ACA + MCA | 2 (6.5%; 1 watershed*) | 0 (0%) |
MCA + PCA | 5 (16.1%; 1 watershed*) | 1 (2.5%; 1 watershed*) |
ACA + MCA + PCA | 2 (6.5%) | 1 (2.5%) |
Multifocal punctate ischemia | 1 (3.2%) | 2 (5%) |
Watershed only | 0 (0%) | 2 (5%) |
Two childhood subjects and one perinatal subject had multifocal punctate ischemia without large or medium sized vessel stroke (Figure
The correlation between the modified pediatric ASPECTS and the quantitative infarct volume was strongly positive for the neonates and moderately positive for the childhood subjects when all subjects were included. The correlation was strongly positive in both groups when excluding the subjects with multifocal punctate ischemia without large or medium sized vessel stroke. For the perinatal subjects, the Spearman rank correlation coefficient (ρ) was 0.76,
The median total modified pediatric ASPECTS for perinatal and childhood subjects for the three raters was not significantly different. The median total modified pediatric ASPECTS for the perinatal subgroup for raters 1, 2, and 3 was 6 (IQR 4–10), 6 (IQR 3–12), and 6 (IQR 4–11), respectively (
In a prior study of children with AIS, infarct volume ≥5% of SBV was predictive of poor outcome and of HT (Beslow et al.,
This study demonstrates that a modified pediatric ASPECTS on acute MRI estimates arterial ischemic infarct volume as a percent of SBV for perinatal and childhood infarction with excellent interrater reliability and validity. The method is attractive since the modified pediatric ASPECTS can be performed in just a few minutes compared to manual segmentation image analysis which takes up to several hours and requires post-processing of the images. Additionally, certain image formats may not be compatible with manual segmentation programs, while the modified pediatrics ASPECTS can be performed on routine clinical viewing stations as long as DWI with ADC is available. The score can also be used to classify infarct volume as large versus small according to the threshold of ≥5% SBV with high sensitivity and specificity. The threshold of 5% of SBV is important for childhood stroke because it predicts both HT of stroke and poorer outcome at follow-up on the Pediatric Stroke Outcome Measure (Beslow et al.,
Our study has several limitations. A region was scored as positive even if a small part of a modified pediatric ASPECTS score region was infarcted. Furthermore, each involved region in the modified pediatric ASPECTS is given equal weight (one point) even though areas like the M1–M6 represent more volume than smaller areas like the internal capsule, insula, or caudate. Both of these factors may cause the modified pediatric ASPECTS to overestimate the infarct volume as a percent of SBV in some patients. An extreme scenario for which the modified pediatric ASPECTS overestimates stroke volume is multifocal punctate ischemia, as demonstrated by the great improvement of the Spearman rank correlation coefficient when excluding such patients (one subject in perinatal group, two subjects in childhood group). Conversely, a subject with stroke affecting fewer than five of the MCA regions could still have a stroke volume ≥slant5% SBV if the entire regions are affected, thereby causing the modified pediatric ASPECTS to underestimate the true stroke volume as a percent of SBV. About 15% of subjects were classified incorrectly as large versus small infarcts, and misclassifications were evenly distributed in either direction. Therefore, one should consider scenarios in which classification could be incorrect when applying the score. Despite these limitations, both the sensitivity and the specificity of the method were high. However, we do not recommend the use of the modified pediatric ASPECTS for use in children with isolated multifocal punctate ischemia. A strength of the study is that the raters had different backgrounds and experience, including radiologists of differing clinical training levels and a pediatric neurologist. Moreover, the raters used two different software packages for viewing the images for the ratings with a high degree of reliability, indicating that modified pediatrics ASPECTS can be performed on a variety of image viewers. Replication of the study involving raters who have other training such as intensivists or research assistants might be useful in order to expand the use of the modified pediatric ASPECTS.
The ease of use of the modified pediatric ASPECTS, along with excellent interrater reliability and validity suggest it could be a useful tool for both researchers and clinicians. For example, use of the modified pediatric ASPECTS could expedite eligibility determination in intervention trials that require stratification by infarct size, such as the planned dose finding and safety study of intravenous recombinant tissue plasminogen activator in pediatric stroke (Amlie-Lefond et al.,
The easily performed modified pediatric ASPECTS on acute MRI represents a useful tool for estimating AIS volume as a percent of SBV in neonates and children with acute AIS. Prospective studies are required to confirm its use for identifying children at risk for HT or poor long term outcome. In the future, this could be a valuable tool for easily and reliably stratifying subjects according to infarct volume for clinical trials and for outcome prediction.
R. N. Ichord: Consultant/Advisory Board, Berlin Heart Clinical Event Committee.
Lauren Beslow: NIH-T32-NS007413, L. Morton Morley Funds of The Philadelphia Foundation, NIH-K12-NS049453. Arastoo Vossough: NIH 1R15CA115464-01A2, NIH 5R01NS042645-07, NIH 5R01HL090615-02, NIH 1R21EY020662-01A1, American Society of Pediatric Neuroradiology. Sudha Kilaru Kessler: NIH-K12-NS049453. Daniel Licht: NIH R01-NS072338, K23-NS52380, Dana Foundation, June and Steve Wolfson Family Fund for Neurological Research. Rebecca Ichord: NIH-R01-NS050488, K23-NS062110. Sabrina Smith: NIH-K12-NS049453.