Magnetic Resonance Imaging Correlates of Transient Cerebral Ischemic Attacks
Background and Purpose MRI of patients with a transient ischemic attack (TIA) may provide more detailed morphological insights than CT. We therefore studied the frequency and type of TIA-related infarcts shown by MRI, examined the utility of intravenous contrast material, and searched for potential predictors of infarct occurrence.
Methods We performed 1.5-T MRI of the brain on 62 patients (age range, 28 to 93 years; mean, 61 years) with a hemispheric TIA. Contrast material (Gd-DTPA) was given to 45 individuals. We recorded type, number, size, and location of ischemic brain lesions and related the presence of acute infarction to features of clinical presentation and probable causes for the TIA.
Results MRI showed focal ischemic lesions in 50 patients (81%), but an acute TIA-associated infarct was seen in only 19 subjects (31%). In patients with an acute lesion, the infarcts were smaller than 1.5 cm in 13 (68%), purely cortical in 11 (58%), and multiple in 7 (37%) individuals. Contrast enhancement contributed to the delineation of an acute lesion in only 2 of 45 patients (4%). Acute infarction was unpredictable by clinical TIA features, but the frequency of identifiable vascular or cardiac causes was significantly higher in those patients with TIA-related morphological damage (odds ratio, 5.2 [95% confidence interval, 1.6 to 17.3]).
Conclusions More than two thirds of TIA patients showed no associated brain lesion even when MRI and contrast material were used, but the overall frequency of ischemic damage was high. TIA-related infarcts on MRI were mostly small and limited to the cortex and tended to consist of multiple lesions. A positive MRI underscores the need for comprehensive diagnostic workup since evidence of infarction appears to be associated with a higher frequency of significant vascular or cardiac disorders.
Cerebral ischemic events with a focal neurological deficit of less than 24-hour duration have been arbitrarily called TIAs.1 Even these short-lasting symptoms may be associated with overt structural damage to the brain in some patients.2 This has led to the term “cerebral infarction with transient signs.”2 It was also suggested that the prognosis of TIA patients with related infarction differed from that of patients without associated ischemic lesions.3 Subsequent data on the frequency and type of TIA-related infarcts and their predictability by clinical variables have been almost exclusively drawn from studies using CT.3 4 5 6 Studies on the long-term prognosis of TIA patients in relation to associated morphological damage were also based on CT-verified infarcts.7 8
The higher sensitivity of MRI for cerebral ischemic damage might lead to different results by providing a more detailed insight into TIA-related cerebral damage. Two studies on a small number of TIA patients suggest a much higher rate of brain lesions shown by MRI than CT.9 10 In a study of 22 patients with TIA, Awad et al10 found focal changes in 17 (77%) when using MRI. Focal CT abnormalities were seen in only 7 (32%) of these individuals. However, there was no clear relationship between clinical symptoms and the majority of MRI lesions.9 10
In the absence of a more detailed analysis of the diagnostic contribution of MRI in TIA, we investigated a series of 62 patients to determine (1) the frequency and type of TIA-related brain lesions shown by MRI, (2) the utility of a contrast-enhanced study, and (3) variables in the patients’ clinical presentation and diagnostic workup that might predispose to the development of infarction.
Subjects and Methods
Over a period of 2 years, 79 patients of the Department of Neurology had been referred to MRI because of an acute focal cerebral neurological deficit that had lasted less than 24 hours and included no signs of brain stem involvement. The clinical course resembled an attack of migraine with aura in 14 individuals, and these were excluded from the analysis. MRI was nonspecific in all of them. Three more patients were excluded because of other brain lesions (metastases in 2 and an arteriovenous malformation in 1). Our study thus included 62 patients with a hemispheric TIA (mean age, 61 years; range, 28 to 93 years; 42 men and 20 women).
In a structured interview we assessed the duration and distribution of symptoms and recorded major cerebrovascular risk factors and previous cerebrovascular events. A complete diagnostic workup served to identify potential vascular and cardiac causes for the TIA. Neurosonographic findings were considered significant if they showed more than 75% stenosis or occlusion of the carotid artery contralateral to the transient neurological symptoms.11 The classification of sources of cardioembolism was based on electrocardiography and transthoracic echocardiography findings and followed the recommendations of the Trial of Org 10172 in Acute Stroke Treatment (TOAST) investigators.12
MRI was performed on a 1.5-T machine (Gyroscan S15 or ACS, Philips), and spin-echo pulse sequences were used to obtain axial proton density (TR, 2200 to 2600 milliseconds; TE, 20 to 30 milliseconds) and T2-weighted (TR, 2200 to 2600 milliseconds; TE, 80 to 120 milliseconds) images. T1-weighted (TR, 600 milliseconds; TE, 20 milliseconds) spin-echo scans were performed in case of focal abnormality and after the application of contrast material (0.1 mmol/kg Gd-DTPA, Schering). Contrast-enhanced MRI studies were routinely performed in the latter part of the study and were obtained on 45 patients. Slice thickness was uniformly 5 mm, and the interslice gap was 0.5 mm.
All scans were interpreted by one investigator with knowledge of the pattern of TIA symptoms but blinded to any other clinical information. Infarcts were considered related to the preceding TIA if they (1) were located in a vascular territory corresponding to the patient’s symptoms, (2) had signal characteristics of an acute ischemic lesion, ie, were hyperintense on both proton density- and T2-weighted scans and isointense to only minimally hypointense on T1-weighted scans, and (3) showed regional swelling and/or contrast enhancement. We recorded the location, number, and size of these infarcts. Other coexisting types of vascular lesions were separated into old infarcts and lacunes (sized <6 mm), which had to show at least partial parenchymal destruction (signal isointense to cerebrospinal fluid), and into white matter hyperintensities. The severity of white matter hyperintensities was graded as punctate (grade 1), early confluent (grade 2), and confluent (grade 3).13 Grade 3 white matter hyperintensities also included irregular periventricular hyperintensity extending into the deep white matter, as suggested from a correlation of MRI and histopathologic findings.14
Statistical analysis consisted of a comparison of demographic, clinical, etiologic, and imaging findings between TIA patients with and without an acute lesion on MRI. The Mann-Whitney U test served to compare continuous variables. The odds ratio and its 95% CI were calculated to estimate differences in the frequency of findings between groups.15
The type and frequency of ischemic brain lesions noted in our study participants are shown in Table 1⇓. Overall, 50 patients (81%) demonstrated some MRI evidence of a focal cerebral lesion. However, acute infarcts related to the TIA symptomatology were detected in only 19 patients (31%). Old infarcts and/or lacunes were present in 15 of 24 patients (63%) with previous cerebrovascular attacks and were also seen in 17 of 38 (45%) previously asymptomatic individuals.
The lesion characteristics of acute infarction are shown in Table 2⇓. Cortical and small lesions prevailed, and multiple areas of acute ischemic damage were noted in more than one third of TIA patients with a relevant MRI (Figs 1 through 3⇓⇓⇓). TIA symptoms could be fully related to the extent and location of the acute MRI lesions in 12 patients but were only partly explained by visible ischemic damage in the remaining 7 individuals. One subject with TIA-related infarction also showed another acute lesion in the contralateral hemisphere that had remained clinically silent and was not included in the analysis of infarct characteristics.
Enhancement of lesions was noted in 5 of 45 patients (11%) with a contrast study. In 2 patients contrast uptake clearly helped to delineate the sites of ischemic damage (Fig 2⇑), while it only supported the acuteness of a lesion in the other 3 individuals. In 8 patients the acute ischemic lesions were not enhanced after Gd-DTPA injection. These patients had been studied earlier (1.9±1.9 days) than those with a positive contrast study (4.4±3.9 days), but this difference did not reach statistical significance.
Patients with and without acute infarction on MRI had a similar mean age (61±14 years versus 60±16 years, respectively), and the mean intervals between TIA and the imaging study were comparable (3±3 days versus 3.5±3 days, respectively). The duration of symptoms in patients with a TIA-related lesion ranged from 2 minutes to 23 hours (mean, 120 minutes; median, 30 minutes). In comparison, neurological symptoms of patients without acute MRI lesion lasted from 2 minutes to 23.5 hours (mean, 175 minutes; median, 60 minutes), which was not significantly longer. Table 3⇓ compares the distribution of clinical symptoms, cerebrovascular risk factors, previous cerebrovascular events, and potential causes for the TIA in relation to the presence or absence of an acute ischemic lesion on MRI. A higher frequency of major vascular or cardiac causes for cerebral ischemia in patients with TIA-related infarction was the only finding to reach a statistically significant difference.
In a careful analysis of MRI in hemispheric TIA, we were able to detect acute ischemic lesions in only 19 of 62 patients (31%). Given the high sensitivity of MRI for ischemic changes, this may seem an unexpectedly low number compared with previous CT studies. With the exclusion of the results of first-generation scanners, the reported frequency of acute CT lesions in TIA ranges from 12% to 48%, as summarized in a review by Toole.3 More recent large CT series also showed frequencies of 17%7 and 28%8 of patients with TIA-related lesions. Otherwise, coexisting ischemic damage was much more frequent in our MRI study than has been reported by CT. Old ischemic lesions with parenchymal destruction, ie, infarcts and lacunes, were seen in 17 of 38 previously asymptomatic TIA patients (45%) and 15 of 24 patients with a cerebrovascular history (63%). This compares with a 14% rate of silent strokes noted on CT of 2329 patients with TIA or minor stroke.16 Patient characteristics cannot serve to explain the relatively low number of acute infarcts observed by MRI. Individuals with transient brain stem symptoms whose lesions may be more difficult to visualize were not included, and the mean duration of neurological symptoms in our patients was even longer than that in most CT series.4 8 It may therefore be concluded that the majority of TIA is in fact not associated with a visible morphological lesion and that CT interpreters may tend to overestimate the proportion of TIA-related brain infarction.
Some other factors also have to be kept in mind when the number of acute infarcts found in this study is interpreted. First, we based our analysis on consecutive referrals from a neurological department of a university hospital to its associated MRI center. Consequently, the proportion of very old patients and those with multiple preceding vascular events is likely to have been underrepresented. If we assume that the etiology of TIA and extent of coexisting ischemic damage vary with these variables, our study certainly is not representative of all TIA patients. However, it covers that segment of the TIA population in whom MRI information may be most valuable for treatment decisions. Second, aware of the patients’ TIA symptoms, we chose to search for acute MRI lesions. This was intended to minimize the uncertainty about the relationship between an infarct and the TIA. In view of the frequently small, multiple, and nonenhancing lesions observed, the danger of both overreporting and underreporting of associated infarcts appeared high if MRI interpretation was performed blinded to any neurological information.
Application of contrast material added little to the visualization of acute infarcts. The acute lesion would have been missed in only 2 of 45 patients if a contrast-enhanced study had not been added. On the other hand, a large proportion of acute infarcts in our series showed no uptake of contrast material. Previous reports have emphasized the need of contrast material for identifying recent lacunar lesions among multiple small old infarctions.17 Evidently, in TIA patients the disruption of the blood-brain barrier often may not be severe enough to cause enhancement. Therefore, lack of contrast uptake should not rule against an acute ischemic lesion in the presence of focal signal hyperintensity and swelling. We also noted a tendency for enhancement to occur more frequently with a longer time interval between MRI and the TIA.
MRI may not strikingly increase the number of visible TIA-related lesions, but it appears to show a different distributional pattern of infarction than reported with CT. In the group of 19 patients with acute lesions, purely cortical infarcts were present in 14 individuals (74%) and multiple acute lesions were seen in 7 (37%) of them. These characteristics appear to fit well with the pathophysiological mechanisms possibly invoked during a TIA. First, transient ischemia of a large vascular territory should preferentially cause cortical damage since gray matter has a much higher energy demand than white matter. Second, TIAs may result from intermittent embolic occlusion of a main arterial branch followed by thrombolysis and upstream transportation of the remaining particles toward the periphery. This should give rise to both cortical and multiple infarcts. Such a mechanism may have also been the reason that neurological symptoms often were not fully explained by the detectable lesions. In contrast to our findings, deep and subcortical infarcts prevailed in CT studies on TIA, and multiplicity of lesions was less frequently noted.6 8 A lower sensitivity of CT for small cortical lesions than for hypodense areas within the brain is a likely explanation. In view of the overall similar number of TIA-related lesions reported by CT and MRI, this may also imply that specifically the association of subcortical lesions with transient ischemic symptoms tends to be overreported on CT.
We found no clear patterns of clinical presentation that might allow us to predict infarction as the basis of transient neurological signs. We also failed to confirm a significant relationship between the duration of the TIA and the occurrence of an acute lesion, but the number of patients for such comparison was relatively small. Infarction-related clinical differences observed in larger patient series studied with CT were also minor and could not serve to predict an acute ischemic lesion.6 8 In agreement with the findings of Bogousslavsky et al,4 we observed a significantly higher proportion of probable vascular and cardiac causes for the TIA in those patients with an acute lesion on MRI. This was due to a higher frequency of both high-grade stenosis and occlusions as well as cardioembolic sources. This association may explain the high rate of CT-verified brain lesions associated with a TIA in the North American Symptomatic Carotid Endarterectomy Trial.8
In conclusion, MRI does not detect a dramatically higher number of acute TIA-related infarcts than that reported with CT. However, MRI allows visualization of a different lesion pattern, with a high proportion of cortical ischemic damage and frequent multiplicity of lesions. Routine administration of contrast material does not seem to be warranted. MRI evidence of TIA-related infarction underscores the need for a comprehensive diagnostic workup since it appears to be associated with a higher probability of significant vascular or cardiac disorders.
Selected Abbreviations and Acronyms
|TIA(s)||=||transient ischemic attack(s)|
We thank Erich Flooh, PhD, for performing the statistical analysis.
- Received October 9, 1995.
- Revision received November 28, 1995.
- Accepted January 5, 1996.
- Copyright © 1996 by American Heart Association
The Ad Hoc Committee on the Classification and Outline of Cerebrovascular Disease, II. Stroke. 1975;6:566-616.
Waxman SG, Toole JF. Temporal profile resembling TIA in the setting of cerebral infarction. Stroke. 1983;14:433-437.
Toole JF. The Willis lecture: transient ischemic attacks, scientific method, and new realities. Stroke. 1991;22:99-104.
Shuaib A, Hachinski VC, Oczkowski WJ. Transient ischemic attacks and normal cerebral angiograms: a follow-up study. Stroke. 1988;19:1223-1228.
Koudstaal PJ, van Gijn J, Lodder J, Frenken CWGM, Vermeulen M, Franke CL, Hijdra A, Bulens C, for the Dutch Transient Ischemic Attack Study Group. Transient ischemic attacks with and without a relevant infarct on computed tomographic scans cannot be distinguished clinically. Arch Neurol. 1991;48:916-920.
Evans GW, Howard G, Murros KE, Rose LA, Toole JF. Cerebral infarction verified by cranial computed tomography and prognosis for survival following transient ischemic attack. Stroke. 1991;22:431-436.
Eliasziw M, Streifler JY, Spence JD, Fox AJ, Hachinski VC, Barnett HJM, for the North American Symptomatic Carotid Endarterectomy Trial (NASCET) Group. Neurology. 1995;45:428-431.
Awad IA, Modic M, Little JR, Furlan AJ, Weinstein M. Focal parenchymal lesions in transient ischemic attacks: correlation of computed tomography and magnetic resonance imaging. Stroke. 1986;17:399-403.
Adams HP, Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon DL, Marsh EE III, and the TOAST investigators. Classification of subtype of acute ischemic stroke: definition for use in a multicenter clinical trial. Stroke. 1993;24:35-41.
Fazekas F, Chawluk JB, Alavi A, Hurtig HI, Zimmerman RA. MRI signal abnormalities at 1.5 T in Alzheimer’s dementia and normal aging. AJNR Am J Neuroradiol. 1987;8:421-426.
Fazekas F, Kleinert R, Offenbacher H, Schmidt R, Kleinert G, Payer F, Radner H, Lechner H. Pathologic correlates of incidental MRI white matter signal hyperintensities. Neurology. 1993;43:1683-1689.
Sandercock P. The odds ratio: a useful tool in neurosciences. J Neurol Neurosurg Psychiatry. 1989;52:817-820.
Herderschee D, Hijdra A, Algra A, Koudstaal PJ, Kappelle LJ, van Gijn J, for the Dutch TIA Trial Study Group. Silent stroke in patients with transient ischemic attack or minor ischemic stroke. Stroke. 1992;23:1220-1224.
Miyashita K, Naritomi H, Sawada T, Nakamura M, Kuriyama Y, Ogawa M, Imakita S. Identification of recent lacunar lesions in cases of multiple small infarctions by magnetic resonance imaging. Stroke. 1988;19:834-839.