Variability of Magnetic Resonance Angiography and Computed Tomography Angiography in Grading Middle Cerebral Artery Stenosis
Background and Purpose Magnetic resonance angiography and computed tomography angiography are new, noninvasive methods to provide images of the cerebral vasculature. The reliability of magnetic resonance angiography and computed tomography angiography when used to grade middle cerebral artery stenosis remains to be established. We sought to study the interobserver and intraobserver variabilities of magnetic resonance angiography and computed tomography angiography in grading middle cerebral artery stenosis.
Methods A total of 50 middle cerebral arteries in 25 patients were studied with magnetic resonance angiography and computed tomography angiography. All patients had a history of ischemic stroke. The films were read independently by two observers on separate occasions. Films were shown again to the same observer 4 weeks after the first reading. The degree of middle cerebral artery stenosis was categorized into four grades: normal/mild, moderate, severe, and occluded. The interobserver and intraobserver variabilities were calculated by the κ statistic method.
Results Interobserver variability for grading middle cerebral artery stenosis was good (κ=0.78) for magnetic resonance angiography and moderate (κ=0.51) for computed tomography angiography. There was perfect agreement between two observers in 86% of the vessels shown in magnetic resonance angiography and in 76% of the vessels shown in computed tomography angiography. Intraobserver variability for both imaging methods was good, with the κ value in the range of 0.70 to 0.76.
Conclusions Our results suggest that according to our protocol, magnetic resonance angiography is more reliable than computed tomography angiography in grading middle cerebral artery stenosis.
The natural history and prognosis of different grades of carotid stenosis are well documented.1 However, similar data for MCA stenosis are scarce. This disparity can be explained in part by the fact that conventional angiography, which is required for the diagnosis of MCA stenosis, is associated with significant hazards while no specific treatment is available for this condition. However, lack of such background information renders the study of the treatment of MCA stenosis more difficult.
With the recent advances in noninvasive investigation of the intracranial vasculature, it is now possible to study large numbers of patients safely. Imaging methods that are available currently include TCD, MRA, and CTA. To categorize patients with different degrees of stenosis in clinical trials or in cohort studies, a measure should be not only safe but also reliable. We performed a prospective study to compare the interobserver and intraobserver variabilities of MRA and CTA when used to grade MCA stenosis.
Subjects and Methods
We studied consecutive patients who presented to our neurology service with stroke or transient ischemic attack within a 6-month period and who had any evidence or suggestion of intracranial stenosis on TCD examination. For each patient, MRA and CTA were performed within 7 days of the TCD examination. The timing and sequence of examination were variable and depended on the availability of the MR and CT machines.
All patients were scanned on a Siemens Magnetom Impact IT system with a head coil. The three-dimensional time-of-flight images were acquired in the axial plane with a repetition time of 47 ms, echo time of 10 ms, flip angle of 20°, 20-cm field of view, partition of 64, 192×256 acquisition matrix, and one signal average for a total imaging time of 9 minutes 40 seconds.
CTA was performed with a General Electric HiSpeed Advantage helical scanner. A total of 135 mL iohexol (Omnipaque 300, Nycomed Imaging) was given by a power injector at the rate of 3 mL/s. Thirty-five helical scans were undertaken, starting at the floor of the sella turcica, with 1-mm collimation and 1 mm/s table speed (pitch of 1:1). The helical scanning was started after a 20-second delay from the commencement of contrast agent injection. The 35 axial CTA source images were then transferred to an Advantage Windows workstation (General Electric Medical Systems) to produce maximum-intensity projection and shaded surface display images.
The MRA and CTA films were read independently by two observers (W.W.M.L., E.L.) on separate occasions. A moderator was present on each occasion to record the results. All observers were blinded to the medical history of the patients, clinical diagnosis, and results of other investigations. For the testing of intraobserver variability, films were shown to the same observer 4 weeks after the first reading. MCA stenosis was graded according to the following criteria: grade 1, normal or mild stenosis (0% to 29% diameter stenosis); grade 2, moderate stenosis (30% to 69% diameter stenosis); grade 3, severe stenosis (70% to 100% diameter stenosis); and grade 4, occlusion (100% diameter stenosis with rarefaction of the MCA distal to the stenosis).
The percentage of MCA stenosis was measured by visual inspection, according to the principles established by the NASCET study.2
We calculated interobserver and intraobserver variabilities by the κ statistic3 using Statistical Package for the Social Sciences computer software (SPSS for Windows, Version 6.0).
A total of 50 MCAs in 25 patients were studied. There were 15 men and 10 women whose mean age was 57 (range, 29 to 83) years. Seventeen patients (68%) had non–insulin-dependent diabetes mellitus and 13 (52%) had hypertension. All patients had significant hemiparesis that lasted >24 hours. No patient had significant carotid stenosis as shown by Doppler examination.
The interobserver variabilities of CTA and MRA are shown in Tables 1⇓ and 2⇓, respectively. The κ statistic was good for MRA (κ=0.78) and moderate for CTA (κ=0.51). There was perfect agreement between the two observers on the grading of stenosis in 86% of the vessels shown in MRA and in 76% of the vessels shown in CTA.
The intraobserver variability of both MRA and CTA was good, with the κ value in the range of 0.70 to 0.76 (Tables 3 through 6⇓⇓⇓⇓). Fig 1⇓ shows an example of severe MCA stenosis, and Fig 2⇓ shows an example of an occluded MCA.
Carotid stenosis is the most common lesion found in whites with thromboembolic stroke. However, in blacks, Hispanics, Japanese, and Chinese, intracranial stenosis is more common than extracranial stenosis,4 5 which makes it numerically perhaps the most important cause of stroke worldwide. Yet little is known about the prevalence, natural history, or treatment of this condition. A few studies have been performed on the prognosis of MCA stenosis, but the patients were either few in number6 7 or were subjects in a clinical trial for extracranial/intracranial bypass surgery.8 Before the treatment of MCA stenosis is studied, there is a pressing need to perform a prospective study on a sizable number of patients to document the prevalence, risk factors, and prognosis of this condition. Furthermore, recent evidence suggests that the treatment of patients with intracranial stenosis may differ from that for patients with extracranial stenosis.9 In a retrospective study of 151 patients with intracranial stenosis, the Warfarin-Aspirin Symptomatic Intracranial Disease Study Group9 found a favorable risk/benefit ratio for warfarin compared with aspirin for the prevention of major vascular events.
Despite the success of the NASCET and European Carotid Surgery Trial studies in establishing the efficacy of carotid endarterectomy, the ideal way to measure carotid stenosis in clinical practice remains controversial.10 11 Notwithstanding the unresolved issue of the relative accuracy of each method, the ideal method should be safe and reliable. With mortality and morbidity of conventional angiography in the range of 1% to 4%,12 noninvasive investigations will be more preferable in the study of large numbers of patients. TCD is widely available and is a low-cost method. Unfortunately, TCD cannot be performed in a significant number of patients because of insufficient temporal windows. MRA and CTA are relatively new ways to image the intracranial vasculature. MRA and CTA images can delineate the morphology of MCA and therefore are complementary to TCD, which provides data on cerebral hemodynamics. A number of current studies concern the use of MRA to study intracranial stenosis.13 14 We recently reported our initial experience in the use of CTA in assessment of the MCA in patients with acute ischemic stroke.15
We did not include conventional angiography in the present study because we aimed to measure the variability of MRA and CTA rather than to establish the validity or accuracy of these methods. In the latter case, comparison with a gold standard such as conventional angiography or, preferably, with a pathological specimen will be required.
In the present study, the interobserver variability of MRA (κ=0.78) in grading MCA stenosis is similar to that reported for the assessment of carotid stenosis by duplex ultrasonography (κ=0.67).16 The main reason for the lesser reliability of CTA was poor picture quality in a few cases. We found that in most CTA pictures, the picture quality was very good. In some cases, poor picture quality might have been due to the variable amount of time taken by the intravenous contrast to reach the circle of Willis or the need for different dosages of contrast material for different patients. In addition, when movement artifact occurred, CTA, unlike MR imaging, could not be repeated immediately because large volumes of contrast injection had to be given.
The κ statistic may be different for other pairs of readers, depending on their background and experience. Both of the readers in the present study are qualified radiologists. They have read intracranial MRA regularly for more than a year and have participated in three research projects that involved the use of CTA of the intracranial vessels. The κ statistic method does not differentiate a minor from a serious disagreement. All disagreements in the case of MRA were minor, although there were cases of serious disagreement for CTA.
In conclusion, our results suggest that according to our protocol, MRA is more reliable than CTA to grade MCA stenosis. However, it is likely that both imaging methods will improve rapidly in time, and our results may not be generalizable to other institutions with different protocols.
Selected Abbreviations and Acronyms
|CTA||=||computed tomography angiography|
|MCA||=||middle cerebral artery|
|MRA||=||magnetic resonance angiography|
|TCD||=||transcranial Doppler ultrasonography|
- Received December 4, 1995.
- Accepted February 27, 1996.
- Copyright © 1996 by American Heart Association
Brennan P, Silman A. Statistical methods for assessing observer variability in clinical measures. BMJ. 1992;304:1491-1494.
Caplan LR, Gorelick PB, Hier DB. Race, sex and occlusive cerebrovascular diseases: a review. Stroke. 1986;17:648-655.
Leung SY, Ng THK, Yuen ST, Lauder IJ, Ho FCS. Pattern of cerebral atherosclerosis in Hong Kong Chinese: severity in intracranial and extracranial vessels. Stroke. 1993;24:779-786.
Caplan L, Babikian V, Helgason C, Hier DB, DeWitt D, Patel D, Stein R. Occlusive disease of the middle cerebral artery. Neurology. 1985;35:975-982.
Bogousslavsky F, Barnett HJM, Fox AJ, Hachinski VC, Taylor W, for the EC/IC Bypass Study Group. Atherosclerotic disease of the middle cerebral artery. Stroke. 1986;17:1112-1120.
Chimowitz MI, Kokkinos J, Strong J, Brown MB, Levine SR, Silliman S, Pessin MS, Weichel E, Sila CA, Furlan AJ, Kargman DE, Sacco RL, Wityk RJ, Ford G, Fayad PB, for the Warfarin-Aspirin Symptomatic Intracranial Disease Study Group. The Warfarin-Aspirin Symptomatic Intracranial Disease Study. Neurology. 1995;45:1488-1493.
Ringelstein EB. Skepticism toward carotid ultrasonography: a virtue, an attitude, or fanaticism? Stroke. 1995;26:1743-1746. Editorial.
Alexandrov AV, Bladin CF, Maggisano R, Norris JW. Measuring carotid stenosis: time for a reappraisal. Stroke. 1993;24:1292-1296.
Seller RJ. Imaging blood vessels of the head and neck. J Neurol Neurosurg Psychiatry. 1995;59:537-539.
Wong KS, Liang EY, Lam WWM, Huang YN, Kay R. Spiral computed tomography angiography in the assessment of middle cerebral artery occlusive disease. J Neurol Neurosurg Psychiatry. 1995;59:537-539.
Sutton-Tyrrell K, Wolfson SK, Thompson T, Kelsey SF. Measurement variability in duplex scan of carotid atherosclerosis. Stroke. 1992;23:215-220.