Magnetic Resonance Imaging versus Computed Tomography in Transient Ischemic Attack and Minor Stroke: The More Υou See the More You Know

Multiple studies have shown that over half of the patients with transient ischemic attack (TIA) or minor stroke have evidence of acute ischemic tissue injury on early diffusion-weighted imaging (DWI) [1,2,3]. The sensitivity of DWI is highest if imaging is obtained earlier, preferably within the first 24 h after symptom onset [4]. Magnetic resonance imaging (MRI) has shown a superior sensitivity over CT in identifying ischemic lesions in the setting of suspected acute ischemic stroke [5]. Furthermore, in one study, it was shown to identify acute ischemic lesions in a third of patients with normal CT after TIA [6]. However, not all patients in this study completed both modalities, and a direct prospective comparison between MRI and CT in a population of patients with TIA or minor stroke is lacking.

The aim of this study was to directly compare the diagnostic yield of MRI and CT imaging in a TIA and minor stroke population.

Patients were selected from the CATCH study; detailed inclusion criteria have previously been described [7]. Briefly, CATCH is a prospective cohort study of high-risk (focal weakness or speech disturbance lasting ≥5 min) TIA and minor stroke (NIHSS <4) patients enrolled between April 2008 and September 2010. Consecutive patients aged 18 years or older presenting with minor stroke or TIA were examined by a stroke neurologist within 24 h of symptom onset. Detailed baseline clinical and outcome information was prospectively collected from each patient. All patients had undergone a computed tomography (CT) within 24 h of symptom onset. In a subset of patients, we performed MRI of the brain at baseline and within 24 h of symptom onset. In this substudy, we included patients who had completed both CT and MRI within 24 h of symptom onset for their presenting event.

Patients were further classified as ‘final diagnosis of TIA' (focal neurologic symptoms that completely resolved within 24 h of symptom onset and TIA mimics excluded upon review of all investigations and follow-up data at 90 days), ‘final diagnosis of minor stroke' (nondisabling, focal neurologic symptoms lasting more than 24 h and minor stroke mimics excluded) and ‘stroke or TIA mimics'. Overall data for the entire cohort includes all patients from the three categories.

All MRI images were acquired on a 3.0-T GE scanner. MRI sequences included DWI (slice thickness 3.5 mm and slice spacing 0.0 mm) and FLAIR. MRIs were evaluated by a neuroradiologist blind to the CT results. The classification of acute ischemic lesions required hyperintense signal on DWI with associated hypointensity or isointensity on apparent diffusion coefficient. If a DWI hyperintense lesion was also hyperintense on the apparent diffusion coefficient map, it was classified as a chronic lesion due to T2 shine-through phenomenon. Measurements of the planimetric DWI lesion volume were performed using Quantomo software [8]. DWI hyperintense lesion borders were defined using a semiautomated threshold intensity technique. In this population, we have previously shown a high intra- and interrater reliability in the measurement of small acute DWI lesions [9].

All CT imaging was performed on a Siemens 64-slice scanner. A standard axial CT of the whole brain was performed with a sequential (nonhelical) technique at 5-mm slice thickness. Imaging was evaluated by a neuroradiologist who was given clinical information on the presenting event. CT scans were assessed blind to the results of the MRI. Acute ischemic lesions were defined on the CT scans as hypodensity or loss of gray-white matter differentiation using variable window width and center level settings increasing the gray-white matter contrast to optimize sensitivity [10]. Hypodensities darker than the normal white matter were not rated as acute ischemic lesions, since the time from symptom onset was always less than 24 h.

Statistical analysis was completed with Stata (version 11). Standard descriptive statistics were used for continuous or binomial outcomes as appropriate. The rates of acute ischemic lesions on CT and MRI were compared using McNemar's test, since MRI and CT represent matched pairs for each patient. Distributions of lesion volume were compared using the Wilcoxon rank-sum (Wilcoxon-Mann-Whitney) test.

A total of 347 patients were included in this study. Baseline characteristics are shown in table 1. Overall, acute ischemic lesions were more common on DWI (193/347: 56%) than on CT (41/347: 12%, p < 0.0001, table 2). We found out that 155 of the 306 (51%) CT-negative patients were positive on DWI-MRI. Considering DWI-MRI as the gold standard, in this study, CT has a sensitivity of 20% (38 CT-positive patients vs. 193 MRI-positive patients) and a specificity of 98% (151 CT-negative patients vs. 154 MRI-negative patients). There were 151 MRI-negative patients versus 306 CT-negative patients (49% negative predictive value for CT). In addition, 38 of 41 patients who had an acute lesion on CT had a concurrent lesion on DWI (93% positive predictive value for CT).

For the final diagnosis of TIA, the yield of both modalities was lower but a significant difference remained (table 3). Furthermore, 54 of the 155 (35%) CT-negative patients were positive on DWI (fig. 1). For the final diagnosis of minor stroke, the yield is higher for both modalities and the added yield of MRI is even greater (table 4). A total of 100 of the 120 (83%) CT-negative patients were positive on DWI.

On DWI, the infarct volume was larger in CT-positive patients (median 5.07 ml, IQR 10) compared to CT-negative patients (median 0.68 ml, IQR 1.31, p < 0.0001). The median infarct volume (0.89 ml, IQR 2.76) and mean (2.92 ml, 95% CI 0.09-12.82) of all positive DWI lesions were small and not statistically different for TIA (median 1.13 ml, IQR 2.97) and minor stroke (median 0.845 ml, IQR 2.53) patients.

The current study represents the first prospective direct comparison between acute CT scan findings and urgent MRI imaging. We found out that MRI imaging has a much higher yield than CT in identifying acute ischemic lesions in the setting of a clinical presentation of TIA or minor stroke. We also observed that the lesions missed by CT, but identified by MRI had a smaller infarct volume.

Previous work has shown that DWI-MRI identifies acute ischemia in approximately 35-50% of patients presenting with TIA or minor stroke [1,11,12,13,14,15,16,17], and that, in this population, acute ischemic changes are less frequent (10%) on CT. In this study, we found out that 50% of patients who were negative on CT had evidence of an acute lesion on MRI. Forster et al. [6] recently reported that 33% of TIA patients with a normal baseline CT had evidence of acute infarction on MRI. We showed results comparable to those of Forster et al. [6] for the TIA subgroup and demonstrated an even larger proportion of acute infarction on MRI after negative CT in the minor stroke population. The yield of CT and MRI imaging for detection of acute ischemic changes is much higher in patients who present with disabling ischemic stroke [5]. This is thought to be mainly due to the larger infarct size [18]. Our results support this finding, as the sensitivity of CT alone in identifying ischemic lesions less than 1 ml in volume was extremely poor.

A limitation of our study is that the CT and MRI studies were not conducted at the same time, with the time from symptom onset to imaging being longer for MRI compared to CT. Since early ischemic changes become more conspicuous over time, on imaging this potentially introduces bias in favor of MRI. The relatively short time span (median of 8 h) between the two studies reduces the likelihood of recurrent events that might affect the results.

This study supports the use of MRI instead of CT for a patient presenting acutely with a TIA or minor stroke because of the superior yield of ischemia in small volume lesions. Previous literature has shown that the finding of an acute ischemic lesion on MRI adds useful clinical information. Firstly, the presence of an ischemic lesion in the brain causes symptoms in the patient. Secondly, location and distribution have diagnostic value in relation to the stroke mechanism [19], but localization on clinical grounds alone in the setting of transient symptoms is far from perfect [20]. Finally, an acute ischemic lesion on DWI in the investigation of TIA and minor stroke has prognostic value as a strong predictor of recurrent stroke [1,15,16,21,22,23].

S.B.C. received salary support from the Alberta Innovates-Health Solutions and the Heart and Stroke Foundation of Canada's Distinguished Clinician Scientist award, supported in partnership with the Canadian Institute of Health Research (CIHR), the Institute of Circulatory and Respiratory Health and AstraZeneca Canada Inc.

The study received funding from the Canadian Institute of Health Research (CIHR) and a Pfizer Cardiovascular research award.

The authors have no conflicts of interest to disclose.