(Stroke. 1999;30:383-388.)
© 1999 American Heart Association, Inc.
Original Contributions |
Correspondence to R.A. Bhadelia, MD, Department of Radiology, TuftsNew England Medical Center, 750 Washington St, Boston, MA 02111. E-mail rafeeque.bhadelia{at}es.nemc.org
| Abstract |
|---|
|
|
|---|
MethodsParticipants of the Cardiovascular Health
Study, aged 65 years or more and without prior stroke, were studied
with brain MRI (n=3456). The prevalence of brain infarcts (
3 mm)
on MRI was determined in subjects with and without TIA. The
cardiovascular risk factors and clinical and
subclinical cardiovascular disease associated with MRI
infarcts were studied in subjects with TIA.
ResultsSubjects with TIA (n=100) had a higher prevalence of MRI infarcts than subjects without TIA (46% versus 28%; P<0.001). The unadjusted odds ratio for having MRI infarcts in subjects with TIA was 2.20 (95% CI, 1.47 to 3.30) and remained significantly elevated after adjustments for risk factors and cerebrovascular disease (odds ratio, 1.86; 95% CI, 1.23 to 2.83). In subjects with TIA, diastolic blood pressure (P=0.01) and internal carotid artery intima-media thickness (P=0.01) were the only factors predictive of the presence of MRI infarcts by stepwise logistic regression analysis.
ConclusionsMRI infarcts are imaging manifestations of clinically important cerebrovascular disease in subjects with a history of TIA, given their increased prevalence and positive association with increased diastolic blood pressure and internal carotid artery intima-media thickness.
Key Words: cerebral infarction cerebral ischemia, transient magnetic resonance imaging
| Introduction |
|---|
|
|
|---|
MRI is more sensitive than CT in detection of brain infarcts19 and in differentiation of infarcts from diffuse white matter changes or leukoaraiosis often seen on neuroimaging studies of elderly or hypertensive subjects.20 21 22 This results in frequent detection of brain infarcts in asymptomatic individuals.23 24 25 Previous reports from the Cardiovascular Health Study (CHS) have shown that 28% of community-dwelling elderly subjects without known prior stroke demonstrate brain infarcts on MRI scans, and these findings have a strong relationship with age and other cerebrovascular risk factors.25 26 This raises the question of whether older subjects with a history of TIA have a greater likelihood of demonstrating infarct on MRI scan than those without TIA, after adjustments for risk factors. It is also unclear whether MRI infarcts in TIA subjects are associated with risk factors or CVD. Presence of such an association may make MRI infarcts an important imaging manifestation of morphological brain changes from cerebrovascular disease.
Use of MRI in evaluation of TIA patients for brain infarcts has been limited to a few small hospital-based studies.19 27 28 In this study we performed a cross-sectional analysis of the data from a large population-based cohort (1) to compare the prevalence of MRI-demonstrated brain infarcts in elderly subjects with and without a history of TIA and (2) to determine which risk factors and measures of CVD are associated with the presence of brain infarcts in subjects with TIA.
| Subjects and Methods |
|---|
|
|
|---|
At entry into the study, participants reported any prior physician-diagnosed vascular event, including TIA and stroke. TIA or stroke events occurring after baseline were confirmed by an adjudication process.30 At the time of MRI examination, the history of TIA and stroke was determined for each participant. A history of TIA was defined as single or multiple episodes of TIA detected by baseline self-report, baseline event confirmed by medical record review, or an adjudicated TIA occurring before MRI. For the purpose of this analysis, we included participants with both carotid and vertebrobasilar TIA symptoms. Stroke history was defined as confirmed baseline self-report (confirmed by physical examination or medical records) or a postbaseline stroke event occurring before MRI. Participants with a documented history of stroke at the time of MRI examination were excluded from the analysis to eliminate this as a possible confounder.
Medical History and Clinical Variables
All CHS participants underwent a baseline clinical examination
and completed standardized questionnaires assessing demographic
factors, health status, and medication use. The details of baseline
examinations and postbaseline telephone interviews and clinical visits
have been described in detail previously.29
The cerebrovascular risk factors assessed were age, sex, blood pressure, history of hypertension or antihypertensive medication, diabetes, smoking, and serum lipids.29 The presence of clinical CVD was determined by a history of atrial fibrillation (self-reported or confirmed by ECG) and a confirmed history of coronary heart disease (defined as angina, myocardial infarction, angioplasty, or coronary bypass graft). Subclinical CVD was assessed by the ankle-arm ratio (ratio of supine brachial and tibial systolic blood pressure measurements), and vascular ultrasound examination of common (CCAs) and internal carotid arteries (ICAs) (to determine intima-media thickness [IMT] and presence or absence of stenosis or atherosclerotic plaque). The CCA IMT was determined as the mean of the maximum wall thickness for near and far walls on both the left and right sides. The ICA IMT was obtained the same way, but the mean value was obtained from the measurements of both sides on 3 different scan planes.31 32
All clinical information, including drug histories and laboratory findings, was derived from the annual examinations closest in time to the MRI. The ultrasound findings were obtained from the 19921993 examination. The results of baseline ultrasound from the 19891990 examination were used for participants who were missing ultrasound data from the 19921993 examination.
Magnetic Resonance Imaging
As part of the CHS protocol, cranial MRI was performed at 4
field centers with the use of 1.5-T (GE Medical Systems; Picker)
instruments at 3 sites and a 0.35-T (Toshiba, American Medical Systems)
instrument at 1 site. The MRI protocol consisted of a sagittal
T1-weighted localizing sequence. This was followed by axial T1-weighted
and axial spin-density and T2-weighted images. All axial sequences were
angled to the anterior/posterior commissure line with 5-mm scan
thickness without interslice gaps and at a field of view of 24 cm.
Imaging data were displayed on a high-resolution workstation and read at a single reading center by primary and secondary readers blinded to any demographic information. The information regarding readers and recording and display systems is detailed elsewhere.22 To be considered an infarct or infarctlike lesion, a focal brain abnormality was required to be a nonmass area in a vascular distribution, hyperintense on spin-density and T2-weighted images. Infarcts of the cortical gray matter and deep nuclear regions and capsule were defined as lesions bright on spin-density and T2-weighted images. However, the abnormalities in white matter were also required to be hypointense on T1-weighted images. These criteria, as well as the focal nature of infarcts, allowed their distinction from diffuse white matter changes or leukoaraiosis.21 22
Location and dimensions of each infarct were measured. Lesions were
mapped to 16 anatomic locations. For the purpose of this
analysis, we categorized these lesions as cortical,
subcortical/deep nuclear, or cerebellar/brain stem. Lesion dimensions
were assessed by measurement of the maximum anterior-to-posterior and
right-to-left diameter by using a manually applied caliper. The
superior-to-inferior dimension was also recorded as the
number of axial slices on which the lesion appeared. The volume of each
lesion was approximated by its dimensions. Total volume of all the
infarcts present in an individual was also calculated. The CHS
database for MRI contains information on infarcts
3 mm and
<3 mm in maximum dimension. In this report we focused the
analyses on MRI infarcts
3 mm in maximum dimension
because of high interobserver agreement for detection of these
lesions.25
Statistical Analysis
All analyses were conducted with SPSS/Windows version
6.1 (SPSS Inc). For comparisons involving continuous variables,
2-sample t tests were used, and for those involving
categorical or dichotomous variables,
2
tests were used. Because of asymmetrical distribution of the observed
MRI infarct volumes in subjects with and without TIA, comparisons were
done with the Mann-Whitney test. Crude and adjusted odds ratios for MRI
infarcts were determined by multivariate logistic
regression. Stepwise multiple logistic regression analysis was
performed to determine independent factors related to the presence of
1 MRI-demonstrated infarct in subjects with TIA. For comparisons of
subjects with and without TIA, only values of P
0.01 were
considered significant. However, for comparison of TIA subjects with
and without infarct, values of P
0.05 were considered
significant because of smaller sample size.
| Results |
|---|
|
|
|---|
Prevalence of MRI Infarcts in Subjects With and Without
TIA
The prevalences of MRI-demonstrated brain infarcts,
cerebrovascular risk factors, and clinical and subclinical CVD in
subjects with and without TIA are presented in Table 1
. The subjects with TIA had a higher
observed prevalence of brain infarcts than those without TIA (46%
versus 28%; P<0.001). The subjects with TIA were
older and reported history of hypertension and use of antihypertensive
medication more frequently than those without TIA. The subjects with
TIA also showed higher prevalence of clinical and subclinical CVD than
those without TIA.
|
Subjects with TIA were more than twice as likely to have MRI infarcts
than those without TIA (Table 2
). The
unadjusted odds ratio for having infarcts in presence of TIA was 2.20
(95% CI, 1.47 to 3.30). The increased odds for having MRI infarcts
with TIA remained significantly elevated after adjustments for risk
factors and CVD. Interaction terms between TIA and the risk factors and
CVD measures were analyzed, but none was significant.
|
The TIA subjects were not more likely to have multiple infarcts than subjects without TIA (average number of infarcts per person was 1.65 for TIA subjects and 1.60 for those without TIA). The subjects with TIA had higher prevalence of cortical infarcts than those without TIA (28% versus 10.8%; P<0.01). Infarcts associated with TIA were larger than those in subjects without TIA. The total infarct volume (total volume of all infarcts observed in an individual) was 6.22 mm3 for TIA subjects compared with 1.89 mm3 for subjects without TIA (P<0.01).
Associations of Risk Factors and Prevalent CVD With
MRI-Demonstrated Infarcts in Subjects With TIA
Diastolic blood pressure, CCA IMT, and ICA IMT were
significantly higher in the 46 TIA subjects with MRI infarcts than in
the 54 TIA subjects without infarcts (Table 3
). There were no significant differences
in age, sex, and prevalence of diabetes, hypertension, or
coronary heart disease in TIA subjects with and without
infarcts. Coronary heart disease was present in 38.9% of
those without infarcts as opposed to 28.3% with infarcts, but
the difference was not statistically significant. The prevalence of
atrial fibrillation and ICA stenosis (>50%) tended to be
higher in TIA subjects with infarcts than in those without infarcts,
but again the observed differences were statistically not
significant.
|
Stepwise multiple logistic regression analysis performed on
subjects with TIA, with MRI infarct as the dependent variable,
showed that the diastolic blood pressure and the ICA IMT
independently predicted MRI infarct after adjustment for age and sex
(Table 4
). The odds ratio was 1.79 for
each 10-mm increase in diastolic blood pressure. Each 1-mm
increase in ICA IMT was associated with a 2.57-fold increased risk of
MRI infarction. The odds ratio was 1.21 for each 0.2-mm increase in ICA
IMT.
|
| Discussion |
|---|
|
|
|---|
Previous MRI studies have reported a 70% to 84% prevalence of brain
infarcts in TIA subjects compared with a 46% prevalence in our
study.19 27 28 Awad et al19 found
focal brain lesions in 77% of the 22 TIA patients studied,and in a recent study by Fazekas et al,28 overall
frequency of focal ischemic brain lesions was 81%. Several
characteristics of our study design may have contributed to the lower
observed infarct prevalence. We used strict selection criteria to
consider a focal lesion as infarct or infarctlike. In our study, a
focal lesion was required to be hyperintense on spin-density and
T2-weighted images and hypointense on T1-weighted images to be
considered an infarct if located in the white matter region.
Furthermore, our analysis included only lesions
3 mm,
shown to have high interobserver agreement for
detection.25 Because widened perivascular spaces
are usually <5 mm and are usually isointense to the brain on
spin-density images, our criteria may have limited inclusion of these
lesions as infarcts in the analysis.25 33 Finally,
the diffuse white matter changes often seen on neuroimaging studies of
elderly individuals, which are considered to be somewhat different
pathologically than infarction, were also excluded from the
analysis.20 21 22 We believe that use of these
strict criteria, exclusion of subjects with prior stroke, and probable
inclusion of subjects with mild TIA symptoms may be the reasons for
lower observed prevalence of MRI infarcts in TIA subjects in our study
compared with the previous studies.
The distribution of MRI infarcts was slightly different for subjects with and without TIA, with more cortical-based infarcts in subjects with TIA. This finding is in agreement with the observations by Fazekas et al,28 who investigated the prevalence of both acute infarcts in the locations consistent with TIA symptoms and focal ischemic lesions in any part of the brain in TIA patients. It is possible that the cortical infarcts observed in our study represent the lesions in the regions related to a TIA event. However, since the determination of the duration and side of TIA may be less reliable several months or perhaps years after the initial event,3 we did not attempt to classify the MRI infarcts as related or unrelated to TIA event. Furthermore, the clinical significance of both of these types of infarcts may be similar,13 and therefore we evaluated all infarcts observed on MRI scans regardless of their location.
Brain infarcts without a prior history of stroke are often called silent infarcts. Several previous studies have evaluated the prevalence of silent cerebral infarction on CT and MRI in various disease states. The presence of silent infarction in the brain has been linked to age,23 24 25 26 hypertension,34 35 36 carotid stenosis,16 37 38 ulcerated carotid plaque,16 37 atrial fibrillation,39 40 coronary heart disease,41 and idiopathic dilated cardiomyopathy.42 Our results show that the odds of seeing infarcts on MRI images remained elevated in TIA subjects even after adjustments for age, sex, carotid stenosis, other cerebrovascular risk factors, and clinical and subclinical CVD. These observations suggest that subjects with TIA may be at an increased risk for brain infarcts independent of these factors. Plausible explanation for this finding include morphological changes secondary to prolonged hemodynamic disturbances resulting from TIA.3 43
TIAs can be mimicked by migraine, arthritis, and other nonspecific symptoms.3 It is possible that results obtained entirely from self-reports may inadvertently misclassify these nonspecific symptoms as TIA.44 Our results show that the subjects with TIA were clearly different from those not reporting TIA with regard to many cerebrovascular risk factors and measures of CVD. These findings suggest that misclassification was unlikely to be a major problem in our study.
In the past, clinical significance of brain infarction in TIA patients has been investigated extensively with CT used as the imaging modality.8 9 10 11 12 13 14 15 16 17 18 TIA patients with brain infarcts on CT have been reported to be older and more likely to have hypertension, carotid stenosis, and ulceration of carotid plaques than TIA patients without such infarcts.8 11 18 TIA patients with infarcts are believed to be at a greater risk of major vascular events.15 Poor collateral circulation and shorter survival times for these patients have also been reported.12 18 It has been suggested that this increased risk for subsequent vascular events applies to brain infarcts both in the distribution of the TIA and in unrelated parts of the brain.13 14 15 16 These observations led many authorities to believe that TIA patients with brain infarcts on neuroimaging study should be classified and treated differently from those without such infarcts.16 19 However, the investigators of the North American Symptomatic Endarterectomy Trial have opposed this rationale.45 They failed to observe a significant difference in the risk of stroke between TIA patients with and without brain infarction ipsilateral to severe carotid stenosis.
Our observations suggest that TIA subjects with MRI infarcts
3
mm in maximum diameter have significantly higher diastolic
blood pressure and carotid wall IMT than TIA subjects without infarcts.
This relationship was independent of age, sex, and other risk factors.
Moreover, increasing values of diastolic blood pressure and
ICA IMT were associated with higher risk of MRI infarcts in subjects
with TIA. Association between diastolic blood pressure and
brain infarcts has been demonstrated before by imaging and autopsy
studies.46 47 48 This relationship between
diastolic blood pressure and brain infarcts is believed to
be due to hypertension-induced increase in cerebral microvascular
tone.
Thickening of the carotid wall is a marker of atherosclerotic disease.31 32 Both ICA IMT and CCA IMT have been linked to coronary heart disease and to atherosclerotic disease in other vascular beds. This association is found to be stronger for ICA IMT than for CCA IMT.32 In a previously published report from the CHS, Polak et al49 found ICA IMT to be strongly associated with clinical manifestations of cerebrovascular disease. In a multivariate regression analysis that included several risk factors and measures of CVD, they found ICA IMT to be the best predictor of TIA and stroke. The association seen between ICA IMT and MRI infarcts in TIA subjects in this study further indicates that the association also applies to morphological brain changes. Whether their presence suggests an increased risk for incident stroke is a matter that needs to be addressed in the future by long-term follow-up studies of the CHS cohort.
In conclusion, this study has shown that MRI-demonstrated infarcts are imaging manifestations of clinically important cerebrovascular disease in subjects with a history of TIA, given their increased prevalence and positive association with increased diastolic blood pressure and ICA IMT.
| Acknowledgments |
|---|
Received September 11, 1998; revision received November 11, 1998; accepted November 11, 1998.
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
D. M. Yousem, R. N. Bryan, N. J. Beauchamp Jr., and A. M. Arnold A National Neuroimaging Database: A Call to Action AJNR Am. J. Neuroradiol., June 1, 2004; 25(6): 908 - 909. [Full Text] [PDF] |
||||
![]() |
Y. Inatomi, K. Kimura, T. Yonehara, S. Fujioka, and M. Uchino DWI abnormalities and clinical characteristics in TIA patients Neurology, February 10, 2004; 62(3): 376 - 380. [Abstract] [Full Text] [PDF] |
||||
![]() |
W. T. Longstreth Jr, P. Diehr, T. A. Manolio, N. J. Beauchamp, C. A. Jungreis, D. Lefkowitz, and for the Cardiovascular Health Study Collaborative Cluster Analysis and Patterns of Findings on Cranial Magnetic Resonance Imaging of the Elderly: The Cardiovascular Health Study Arch Neurol, April 1, 2001; 58(4): 635 - 640. [Abstract] [Full Text] [PDF] |
||||
![]() |
R. T.F. Cheung, M. Weih, and U. Dirnagl Ischemic Preconditioning and Tolerance in Human Brain Response Stroke, September 1, 2000; 31 (9): 2266 - 2278. [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Stroke Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 1999 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |