| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Pediatric Imaging |
1 From the Depts of Diagnostic Imaging (R.G.S., T.E., G.M.H., K.J.H.), Hematology (W.C.W.), and Biostatistics (X.X.), St Jude Children’s Research Hosp, 332 N Lauderdale St, Memphis, TN 38105-2794; and Depts of Pediatrics (R.G.S., L.W.W., W.C.W.) and Radiology (R.G.S., K.J.H.), Univ of Tennessee School of Medicine, Memphis. Received Jul 29, 2002; revision requested Sep 20; final revision received Nov 25; accepted Jan 14, 2003. Supported by American Lebanese Syrian Associated Charities. R.G.S. supported by grant RO1 HL60022 from the NHLBI. Address correspondence to R.G.S. (e-mail: grant.steen@stjude.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Magnetic resonance (MR) imaging and MR angiographic findings for 185 patients with SCD examined at St Jude Children’s Research Hospital since 1993 were reviewed. At least two readers independently reviewed images. Standard MR imaging criteria were used to identify lacunae, loss of white matter volume, encephalomalacia, or leukoencephalopathy. Patients were assigned grades to indicate limited or extensive abnormalities. Standard MR angiographic criteria were used to identify arterial tortuosity (limited vasculopathy) and stenosis or occlusion (extensive vasculopathy). Findings were evaluated as a function of patient clinical status (including stroke) and diagnosis. Recent methods (T1- and T2-weighted MR imaging plus fluid-attenuated inversion recovery [FLAIR] at 3-mm section thickness) were compared with older methods (T1- and T2-weighted MR imaging without FLAIR at 5-mm section thickness).
RESULTS: At mean age of 10 years, overall prevalence of infarction, ischemia, or atrophy in patients with SCD was 44% (82 of 185), and prevalence of vasculopathy was 55% (102 of 185), without evidence of a significant referral bias. Twenty-six of 27 patients with clinical stroke had abnormal findings at imaging, but even if patients with stroke were excluded, 35% (56 of 158) had a "silent infarction" (MR imaging–visible injury without clinical stroke), and 49% (78 of 158) had abnormal findings at MR angiography. Patients with clinically severe disease had more abnormalities at MR imaging (P < .001) and MR angiography (P < .004) than did patients with milder disease. Severe vasculopathy was more prevalent in patients with hemoglobin SS than in those with hemoglobin SC (P < .001). Recent imaging methods showed more abnormalities than did older methods (P < .01). With newer methods, 43% (29 of 67) of patients had extensive abnormalities, whereas with older methods, 28% (33 of 116) had extensive abnormalities.
CONCLUSION: Prevalence of ischemic brain injury in pediatric patients with SCD is substantially higher than was previously reported, in part because of improvements in imaging methods.
© RSNA, 2003
Index terms: Brain, infarction, 10.78 • Cerebral angiography, 17.12142 • Cerebral blood vessels, 17.651 • Children, central nervous system, 10.651, 17.651 • Magnetic resonance (MR), vascular studies, 17.12142 • Sickle cell disease (SS, SC), 10.651, 17.651
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Findings of the CSSCD evaluation of patients (1,2) caused problems in several ways. Cranial vasculopathy was not assessed in this cohort because of limitations in MR angiographic methods. Patients in the CSSCD were evaluated at 0.6 T or at 1.0 T, and all imaging was performed by using 5-mm-thick sections (1). Both low-field-strength and thick imaging sections would be expected to reduce sensitivity to abnormalities. The CSSCD investigators themselves note that their estimate of the prevalence of imaging abnormalities would be affected by improvements in imaging technology (6).
Key questions arose from this information: What is the prevalence of imaging abnormalities in children with SCD? What factors affect the prevalence estimate? Is the higher prevalence of abnormalities in recent studies an artifact, perhaps of accrual bias or of a small sample size, or is it a result of progressive improvements in imaging methods? The purpose of our study was to determine the clinical and methodological factors that influence the estimate of prevalence of brain injury in pediatric patients with SCD.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Overall, 74% (136 of 185) of patients were enrolled in a protocol that became available in 1997, but all patients received the same imaging examination, even if they were enrolled prior to 1997. The protocol was broadly written so that any patients with SCD could undergo imaging, irrespective of their clinical condition. There were no patient exclusion criteria, since we were interested in the natural history of the disease. Patients were recruited from a pool of patients at the Pediatric Hematology Clinic of Memphis, and the only selection criterion was that of self-selection in keeping the MR imaging appointment.
All patients evaluated were enrolled in an institutional review board–approved protocol. Parents or guardians of all children signed an informed consent after a brief description of the protocol was provided to them. Patients younger than approximately 8 years of age often were sedated with intravenous administration of 50 mg of meperidine hydrochloride (Demerol; Abbott Laboratories, North Chicago, Ill) per square meter and of 100 mg of pentobarbital sodium (Nembutol Sodium; Abbott Laboratories) per square meter to help them complete their examinations.
The protocol enables us to follow up patients longitudinally for as long as 4 years, and it also enables us to enroll healthy siblings of patients. We specifically targeted accrual of young children (4–7 years old), since we had National Institutes of Health funding for a longitudinal study of young patients. We report findings of only the most recent imaging examination for each patient, since longitudinal follow-up data were not available for all of our patients. None of the data regarding healthy siblings are reported here.
MR Imaging
All patients underwent an MR imaging examination of the brain at 1.5 T with an imaging unit (Vision; Siemens Medical Systems, Iselin, NJ), and all patients were evaluated in the same imager by using a standard quadrature head coil. Weekly quality assurance monitoring of field homogeneity and eddy current compensation was performed, and image quality was monitored daily as a part of the clinical imaging program.
The protocol required that a T1-weighted image set be acquired transversely, with parameters as follows: repetition time msec/echo time msec/inversion time msec, 8,000/27/300; field of view, 17 x 23 cm; flip angle, 180°; matrix, 196 x 256; sections, 19; acquisitions, one in 3 minutes 53 seconds. These parameters were used to increase the contrast between white matter and gray matter in patients who were often quite young. A T2-weighted turbo spin-echo image set was acquired in the same orientation by using parameters as follows: repetition time msec/echo time msec, 3,500/17, 102 (three echoes per repetition time for each effective echo time); field of view, 17 x 23 cm; matrix, 190 x 256; sections, 19; acquisitions, one in 3 minutes 35 seconds.
In March 2000, we also added imaging with fluid-attenuated inversion recovery (FLAIR) in the same orientation, with parameters as follows: 9,000/119/2,470; field of view, 17 x 23 cm; flip angle, 180°; matrix, 154 x 256; sections, 19; acquisitions, one in 3 minutes 27 seconds.
Prior to March 2000, all patients were examined with a 5-mm section thickness and a 1-mm section gap. At the same time that we added FLAIR imaging to our standard patient examination, we also began to use a thinner imaging section to image 100% of the brain volume for image segmentation and classification. Each sequence (ie, T1 weighted, T2 weighted, and FLAIR) was performed with 3-mm section thickness and a 3-mm section gap; then the sequence was repeated with a 3-mm offset. Overall, 36% (67 of 185) of patients were evaluated with the newer method of FLAIR imaging combined with a 3-mm section thickness.
Evaluation of MR Imaging Findings
All images were read and reports were dictated by one of several neuroradiologists, often working with a neuroradiology fellow. All images were reviewed by an experienced reader (R.G.S.) with 8 years of experience in the evaluation of pediatric patients with SCD who sought to verify the dictation. In those patients with dictation that indicated that they had an abnormality or in those patients with questionable cases, images were evaluated by another neuroradiologist (K.J.H.) who reconciled each questionable case, usually in consultation with the first radiologist. All MR images were therefore evaluated by at least two readers, and all abnormal findings of examinations were evaluated by at least three readers.
Findings of patient examinations were scored as normal or abnormal by using MR imaging with a combination of T1-weighted, T2-weighted, and FLAIR sequences. Small unifocal lesions (ie, < 1 cm in diameter) were sufficient to classify a patient’s findings as abnormal by using MR imaging. Lesions were broadly defined to include lacunar infarction, encephalomalacia, atrophy, or leukoencephalopathy. A lacuna was defined as a shelled-out volume, usually in white matter, that was visible with T1-weighted, T2-weighted, and FLAIR sequences. Encephalomalacia was defined as any other ischemic change (except atrophy) such as an infarct, which could be seen as high signal intensity by using a T2-weighted or a FLAIR sequence. Atrophy was defined as loss of brain tissue volume, even if no other abnormality was apparent; atrophy was apparent as lobar asymmetry, as open sulci (which are not expected in a child), or as ventricular dilatation. Leukoencephalopathy was defined as the degeneration or demyelination of white matter, which was seen as abnormally high signal intensity on T2-weighted or FLAIR images. The T2-weighted or FLAIR images were used primarily to identify focal areas of high signal intensity, which would be consistent with lacunae, or diffuse areas of high signal intensity, which would be consistent with leukoencephalopathy. The T1-weighted images were used to confirm lacunae, encephalomalacia, leukoencephalopathy, or atrophy.
A scoring system was used whereby a small (ie, ≤1 cm) focal abnormality or mild limited leukoencephalopathy (ie, ≤3 cm) was coded with a score of 1 and was tabulated as a "limited abnormality." A large (ie, >1 cm) or multifocal abnormality or extensive leukoencephalopathy (ie, >3 cm) or atrophy was coded with a score of 2 and was tabulated as an "extensive abnormality." A score of either 1 or 2 was sufficient for a patient to be considered to have "abnormal" findings.
MR Angiography
All 185 patients underwent MR angiographic examinations, performed by using a three-dimensional time-of-flight sequence, to obtain spoiled gradient-echo images in the transverse plane at 64 section levels (7–9). Magnetization transfer was used to suppress signal from brain parenchyma, and a variable flip-angle excitation (Tilted Optimized Nonsaturating Excitation, TONE; Siemens Medical Systems) pulse was used to avoid saturating signal from in-flowing blood. Since 1995, MR angiograms were optimized for patients with a high rate of blood flow by using the following parameters: 34/5; field of view, 20 cm; flip angle, 20°; matrix, 192 x 256; effective section thickness, 1 mm; and acquisitions, one in 7 minutes. MR angiograms were centered at the sella turcica so that major arteries arising from the circle of Willis could be visualized. The 64 sections were reconstructed into a transverse slab by using a standard maximum intensity projection algorithm. Transverse maximum intensity projection images were reconstructed to show the bilateral structure of the internal carotid arteries, the middle cerebral arteries, the anterior cerebral arteries, the posterior cerebral arteries, and the basilar artery (vertebral arteries were often truncated in the reconstruction).
Evaluation of MR Angiographic Findings
All MR angiograms were read, and reports were dictated and reviewed; questionable cases were reconciled, as with the MR images, by the same readers and with MR images for comparison. All MR angiograms were thus evaluated by at least two readers, and all abnormal findings of examinations were evaluated by at least three readers, with consensus achieved in conference. Criteria of abnormality by using MR angiography included stenosis or apparent occlusion of any vessel or arterial tortuosity. Tortuosity is a subjective clinical impression that is based on the presence of several features: dilatation (ectasia) of a vessel segment, abnormal increase in length of a vessel segment, and/or obvious bowing of an artery (7). These factors are interrelated in that ectatic vessels are likely both to bow and to be longer than normal vessels (7). Patients with ectasia of one vessel tend to have abnormalities elsewhere in the cranial vasculature (8), so obvious ectasia of the basilar or the middle cerebral artery was sufficient for a patient to have a diagnosis of arterial tortuosity.
A scoring system was used whereby tortuosity, arterial ectasia, or indeterminate stenosis was coded with a score of 1 and tabulated as "tortuosity." Arterial stenosis or occlusion was coded with a score of 2 and tabulated as "stenosis/occlusion." A stenosis was defined as obvious narrowing or focal signal dropout in a major artery (ie, middle cerebral artery, anterior cerebral artery, posterior cerebral artery, internal carotid artery, or basilar artery), whereas occlusion was defined as signal loss from the distal portion of a major artery. Thus, we did not make a distinction between stenosis and occlusion, nor did we attempt to grade stenoses; any degree of stenosis was sufficient to be coded with a score of 2. A score of 1 or 2 was sufficient for a patient to be categorized as having abnormal findings.
Assessment of Potential Referral Bias in Our Sample
The eligibility criteria for our protocol enabled us to accrue patients referred for any of several neurologic symptoms, including stroke or declining school performance. It is potentially possible that our patient sample is not representative of the population of patients with SCD, as there could have been preferential referral of patients with serious neurologic illness. To assess this possibility, we evaluated a subset of our patients who were originally accrued for the CSSCD, and we compared the CSSCD cohort with the rest of our patients. The CSSCD cohort was originally a random sample of patients, since patients were identified by means of a newborn screening program and were accrued at younger than 6 months of age rather than being accrued on the basis of referral for symptoms. We evaluated all 18 CSSCD patients with hemoglobin SS (seven male patients, 11 female patients; mean age, 13.3 years ± 2.9; age range, 8.2–17.8 years), excluding those patients with hemoglobin SC, since many such patients have normal findings at imaging.
Effect of Clinical History on Imaging Findings
The varied clinical history of our patients was categorized by determining whether each patient was eligible for bone marrow transplantation (BMT) by using clinical criteria. We chose to focus on BMT, because this therapy is becoming more commonplace at our institution and because criteria for transplantation are fairly widely accepted in the transplantation community (9). The eligibility criteria for BMT included age older than 3 years and three vaso-occlusive events within 1 year, two episodes of acute chest syndrome within 2 years, a combination of two vaso-occlusive events and one episode of acute chest syndrome within 1 year, or clinical stroke.
Patient eligibility for BMT was determined by two of us (T.E., G.M.H.) for all patients with SCD, with a review of each patient’s chart for hospital admissions 2 years prior to the last imaging date. We reviewed medical records from LeBonheur Children’s Medical Center, the Pediatric Hematology Center of Memphis (PHECOM) database, and St Jude Children’s Research Hospital, all in Memphis, Tenn. More than 70% (130 of 185) of our patients resided in Memphis, so patient admissions would be at LeBonheur; we therefore reviewed insurance and billing records there to make sure that we captured all patient admissions. The parents of children who resided outside greater Memphis were contacted by one of us (G.M.H.), if possible, and we requested medical records for patients who may have been hospitalized in another city. We then categorized patients as eligible or ineligible for BMT on the basis of the clinical criteria just described.
Effect of Patient Diagnosis on Imaging Findings
The effect of diagnosis on imaging findings was assessed by stratifying patients according to hemoglobin phenotype. Patients with the severe form of the disease (hemoglobin SS) were compared with patients with hemoglobin SC; eight patients with other diagnoses were excluded from analysis, as there were too few patients for a meaningful comparison. Hemoglobin phenotype was determined by using high-performance liquid chromatography at a reference laboratory.
Effect of Patient Age on Imaging Findings
The effect of age on imaging findings was assessed by stratifying patients according to their age at MR imaging. For this analysis, we included only patients with hemoglobin SS so as not to confound an age effect with an effect of hemoglobin SC. We stratified patients into nine age classes (ie, <2 years old, 2–4 years old, >4–6 years old, >6–8 years old, >8–10 years old, >10–12 years old, >12–14 years old, >14–16 years old, >16 years old). Then we scored the proportion of patients in each age class with an extensive abnormality by using MR imaging (MR imaging score of 2, as described before in this article).
Effect of Recent Methods on Imaging Findings
We have routinely used FLAIR imaging and 3-mm section thickness in MR imaging examinations since March 2000, so 36% (67 of 185) of our patients were examined by using more recent imaging methods. To assess the effect of these methods on imaging findings, we compared patients who underwent imaging prior to March 2000 with the patients who underwent imaging since that date. All patients underwent imaging with the same magnet and with T1- and T2-weighted sequences that were identical, so the only difference was in the use of FLAIR imaging and the thinner imaging section.
Relationships between MR Imaging and MR Angiography
We evaluated the relationship between MR imaging and MR angiographic findings by determining the proportion of patients who had normal or abnormal findings by using each method. We evaluated all 185 patients in this analysis to determine whether vascular abnormalities could be primary to parenchymal abnormalities.
Statistical Analysis
We used a computerized statistical program (StatExact 3 for Windows; Cytel Software, Cambridge, Mass) for all statistical tests. The Kruskal-Wallis and the Fisher exact tests were used to compare observed and expected values for prevalence data. Power calculations were used to assess the likelihood that we could detect a significant difference between patient subpopulations in the prevalence data. A probability level of .05 was accepted as significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
Effect of Patient Age on Imaging Findings
Our results show that patients can have extensive imaging abnormalities at an early age. The proportion of patients with extensive abnormalities by using MR imaging tended to increase as a function of age in those patients older than 8 years of age (Fig 2). However, there was also an increased prevalence of imaging abnormalities in one group of children (ie, those >4–6 years old).
|
|
|
|
Relationships between MR Imaging and MR Angiography
If vascular injury is primary to brain injury, a patient with a vascular abnormality would be expected to have a parenchymal abnormality, whereas a patient with normal vasculature would be expected to have normal brain parenchyma. Although brain injury was associated with arterial stenosis or occlusion in 16% (30 of 185) of patients (Fig 4), brain injury was not associated with stenosis or occlusion in 28% (52 of 185) of patients. Thus, only 37% (30 of 82) of all brain injuries could be predicted on the basis of MR angiographic findings. Moreover, 2% (four of 185) of patients had abnormal findings at MR angiography, with no evidence of brain injury at MR imaging.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
In the largest study to date, a 22% (70 of 312) prevalence of infarction, ischemia, or atrophy was reported among patients with SCD (1), whereas we report a 44% (82 of 185) prevalence of infarction, ischemia, or atrophy among our patients. Thus, we report a prevalence of imaging abnormalities that is twice as high in a patient population that appears to be similar. Furthermore, the prevalence of silent stroke (MR imaging evidence of ischemic injury in the absence of a clinical history of stroke) reported in the early work (1) was 13% (39 of 312), whereas we report silent stroke in 35% (56 of 158) of our patients. Finally, the prevalence of MR imaging abnormalities we report among children who were clinically well (too well to be considered eligible for BMT) is 37% (35 of 96). Can this apparent increase in the prevalence of brain injury be explained as a result of improvements in imaging methods?
There are several potential reasons why we visualized more abnormalities in our patients. One possibility is that our patients were simply sicker. However, a comparison of the subset of patients accrued for the CSSCD with the remainder of our patients suggests that the CSSCD cohort of children is not significantly different from the rest of the children in our study. Thus, we believe that our sample of patients is representative of patients with SCD at large.
Another possible reason why we observed more abnormalities than were previously reported is that our patients were significantly older at imaging than were the patients reported before (1). The mean age of patients in the earlier report (1) was 8.3 years ± 1.8, whereas that in our patients was 9.9 years ± 4.9, and that in the patients evaluated for BMT eligibility was 10.9 years ± 4.1. Thus, the increased prevalence of imaging abnormalities that we report could be related to a longer period of exposure to effects of disease. Our data suggest that the prevalence of extensive imaging abnormalities does increase with patient age. This conclusion is consistent with CSSCD data, which showed that the number of children with overt stroke increased as a function of patient age. In the CSSCD study, the proportion of children with a documented cerebrovascular accident roughly doubled between 6 and 12 years of age (1).
A third reason why we report a higher prevalence of abnormalities than was shown before is that we used more recent methods for 36% (67 of 185) of our patients. We demonstrated that use of FLAIR and thin imaging sections significantly enhances the ability to visualize abnormalities. The enhanced sensitivity associated with recent methods could arise for many reasons. Section thickness could play a role, especially in visualizing subtle lesions. When a 5-mm section was used, a distance factor of 0.2 was used so that there was a 1-mm gap between sections. This eliminated radio-frequency cross talk between sections, but there is effectively only 83% (five of six) volume coverage of the brain. We recently began imaging with 3-mm sections to achieve 100% volume coverage for segmentation, and the thinner sections actually interrogate 20% more of the brain. Thinner sections also lead to greater conspicuity of small lesions, as subtle lesions are less obscured by partial-volume effects in a thin section. Extensive regions of leukoencephalopathy may also be more apparent if FLAIR imaging is used. Nevertheless, we saw more MR imaging abnormalities in our patients by using older methods than were seen in the CSSCD cohort (1). With a 5-mm section thickness without FLAIR, 35% (41 of 116) of our patients had limited or extensive abnormalities, whereas the CSSCD researchers reported that only 22% (70 of 312) of patients had abnormal findings with an equivalent examination. The high prevalence of imaging abnormalities that we report suggests that all pediatric patients with SCD should be screened by using MR imaging to identify patients with a brain injury, since silent infarction is a good predictor of the risk of stroke (6,10,11).
It seems likely that imaging methods have improved enough during the past decade to account for some of the increase in prevalence that we report, even without the advantages of FLAIR and a thin imaging section. The earlier study (1) began in 1989 and ended in 1993, so 74% (136 of 185) of our patients were enrolled at least 5 years after the close of the CSSCD study. This time represents several generations of hardware improvements and software upgrades. Imaging at 1.5 T has improved over the past decade because of advances in gradients and gradient-switching times, which result in higher spatial resolution and better temporal resolution (12). In addition, the CSSCD was a multi-institutional study, and some participating institutions performed imaging at 0.6 T or 1.0 T (1). All of our imaging was performed at 1.5 T, which should produce predictable increases in signal-to-noise and contrast-to-noise ratios, compared with these ratios at low-field-strength MR imaging (13,14).
There are several potentially substantial limitations to our study. One possible problem is that the 185 patients in our sample underwent a total of 321 MR imaging examinations. There were at least two examinations available for 81 patients, and one patient underwent a total of eight examinations for comparison, yet findings of only the most recent examination were reported here. Nevertheless, the availability of results of prior examinations may have improved the sensitivity of the last examination in certain cases, since multiple examinations give readers multiple opportunities to identify a subtle lesion. Thus, patients who undergo multiple examinations may be more likely to have a subtle lesion diagnosed than would a patient who undergoes only one examination. A second potential problem is that there may have been an accrual bias too subtle to be detected with our approach of comparing the available data in 18 CSSCD patients with the data in the other patients. Our patients could have somewhat more severe neurologic illness than is common among patients with SCD, even though our analysis did not reveal a major referral problem. We note that there could be, by now, a referral bias in the CSSCD sample itself, since CSSCD patients with serious neurologic illness may have been more likely to have undergone follow-up imaging since 1993. A third potential problem is that the clinical history of our patients was, in some cases, very difficult to determine, especially for patients outside the Memphis area. We made an exhaustive effort to obtain complete clinical histories for all of our patients, but we cannot exclude the possibility that some patients in the BMT-ineligible group may actually have been eligible for transplantation. However, this seems somewhat unlikely, since the proportion of patients who qualified for BMT was much higher than we anticipated. A fourth potential problem is that there were software upgrades to our MR imaging system during the course of the study, so we cannot exclude a possibility that patients who were evaluated recently underwent a more sensitive examination. This is a strong rationale for evaluation of findings of the last available examination, yet there were still six patients whose last examination was performed in 1993 or 1994.
Our findings imply that vasculopathy occurs earlier in the disease course than does brain injury, as tortuosity revealed by using MR angiography is common in children who have normal findings by using MR imaging. This is encouraging, since it suggests that it may become possible to use MR angiographic findings to identify children at risk for stroke. However, we found a relatively poor correlation between MR imaging and MR angiographic results; 28% (52 of 185) of patients had limited or extensive brain lesions but nonstenotic or nonoccluded arteries, whereas 2% (four of 185) of patients had stenosis or occlusion yet had normal findings by using MR imaging. With MR angiography, visualization of small vessels is not possible unless there is very rapid blood flow (7,9), so MR angiography is insensitive to emboli in small arteries (15). Hemodynamic impairment without occlusion can cause white matter infarction (16), and leukoencephalopathy often is associated with occlusion of small arteries (17). Even lacunar infarction can result from thrombosis in a perforating vessel too small to be seen by using MR angiography (18). Because brain injury can be associated with vasculopathy in arteries too small to be visualized by using MR angiography, findings at MR angiography cannot be used to predict all patients who are at risk of stroke.
Nevertheless, large arteries become ectatic as an adaptive response to anemia, and arterial ectasia is associated with tortuosity (7,8), so tortuosity could be a sign of chronic brain hypoxia. We demonstrate here that arterial tortuosity can be present before an abnormality is depicted at MR imaging. These findings suggest a general hypothesis that hypoxia can precede ischemia in the pathogenesis of brain injury (19). A large prospective study of patients with SCD, with long-term follow-up, is needed to rigorously test the hypothesis that an abnormality that is detected at MR angiography can be used to predict at least some patients at risk of brain injury. Screening patients for ischemic brain injury could also help to identify a group of patients at increased risk of recurrent stroke (20–23).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: BMT = bone marrow transplantation, CSSCD = Cooperative Study of Sickle Cell Disease, FLAIR = fluid-attenuated inversion recovery, SCD = sickle cell disease
Author contributions: Guarantors of integrity of entire study, R.G.S., K.J.H.; study concepts and design, R.G.S., W.C.W., K.J.H.; literature research, R.G.S., W.C.W.; clinical studies, T.E., G.M.H., L.W.W., W.C.W.; data acquisition, R.G.S., T.E., G.M.H., L.W.W., K.J.H.; data analysis/interpretation, R.G.S., K.J.H.; statistical analysis, R.G.S., X.X.; manuscript preparation, R.G.S., T.E., X.X.; manuscript definition of intellectual content, R.G.S., K.J.H.; manuscript editing and revision/review, all authors; manuscript final version approval, R.G.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. S. Silva, P. Vicari, M. S. Figueiredo, H. Carrete Junior, M. H. Idagawa, and A. R. Massaro Brain Magnetic Resonance Imaging Abnormalities in Adult Patients With Sickle Cell Disease: Correlation With Transcranial Doppler Findings Stroke, July 1, 2009; 40(7): 2408 - 2412. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
X. W. van den Tweel, A. J. Nederveen, C. B. L. M. Majoie, J. H. van der Lee, L. Wagener-Schimmel, M. A. A. van Walderveen, B. T. P. The, P. J. Nederkoorn, H. Heijboer, and K. Fijnvandraat Cerebral Blood Flow Measurement in Children With Sickle Cell Disease Using Continuous Arterial Spin Labeling at 3.0-Tesla MRI Stroke, March 1, 2009; 40(3): 795 - 800. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. R Deane, D. Goss, S. O'Driscoll, S. Mellor, K. R E Pohl, M. C Dick, S. E Height, and D. C Rees Transcranial Doppler scanning and the assessment of stroke risk in children with haemoglobin sickle cell disease Arch. Dis. Child., February 1, 2008; 93(2): 138 - 141. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. G. Steen and R. J. Ogg Abnormally High Levels of Brain N-Acetylaspartate in Children with Sickle Cell Disease AJNR Am. J. Neuroradiol., March 1, 2005; 26(3): 463 - 468. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. G. Steen, C. Fineberg-Buchner, G. Hankins, L. Weiss, A. Prifitera, and R. K. Mulhern Cognitive Deficits in Children With Sickle Cell Disease J Child Neurol, February 1, 2005; 20(2): 102 - 107. [Abstract] [PDF]
|
|
|
|
 |
|
 D. R. Powars Childhood stroke frequency decreased in California during 1999 and 2000 Blood, July 15, 2004; 104(2): 298 - 299. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
