Transient ischemic attack with infarction: A unique syndrome?

International Congress Series 1290 (2006) 45 – 55

www.ics-elsevier.com

Transient ischemic attack with infarction: A unique syndrome?B Hakan Ay a,b,*, Walter J. Koroshetz b, Thomas Benner a, Mark G. Vangel a,c, Ona Wu a, Lee H. Schwamm b, A. Gregory Sorensen a,d a

A. A. Martinos Center for Biomedical Imaging and Stroke Service, Department of Radiology, United States b Stroke Service, Department of Neurology, Massachusetts General Hospital, Harvard Medical School, United States c GCRC Biomedical Imaging Core, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, United States d Division of Health Sciences and Technology, Harvard-Massachusetts Institute of Technology, Boston, MA, United States

Abstract. It is debated whether transient symptoms associated with infarction (TSI) are best considered a minor ischemic stroke, a subtype of transient ischemic attack (TIA), or a separate ischemic brain syndrome. We studied clinical and imaging features to establish similarities and differences among ischemic stroke, TIA without infarction, and TSI. 87 consecutive patients with TIA and 74 patients with ischemic stroke were studied. All underwent diffusion-weighted imaging on admission. Symptom duration and infarct volume were determined in each group. Thirty-six patients (41.3%) with TIA had acute infarct(s). Although TIA-related infarcts were smaller than those associated with ischemic stroke (mean: 0.7 vs. 27.3 ml; p b 0.001), there was no lesion size threshold that distinguished ischemic stroke from TSI. In contrast, the symptom duration probability density curve was not broad, but instead peaked early with only a few patients having symptoms for longer than 200 min. The probability density function for symptom duration was similar between TIA with or without infarction. The in-hospital recurrent ischemic stroke and TIA rate was 19.4% in patients with TSI and 1.3% in those with ischemic stroke. TIA with infarction

B Received Dec 1, 2004, and in revised form Feb 18, 2005. Accepted for publication Mar 1, 2005. Published online Apr 25, 2005 in Wiley InterScience (http://www.interscience.wiley.com). DOI: 10.1002/ana.20465. * Corresponding author. A. A. Martinos Center for Biomedical Imaging and Stroke Service, Departments of Neurology and Radiology, Massachusetts General Hospital, CNY149-2301, Charlestown, MA 02129, United States. E-mail address: [email protected] (H. Ay).

0531-5131/ D 2006 Published by Elsevier B.V. doi:10.1016/j.ics.2006.04.007

46

H. Ay et al. / International Congress Series 1290 (2006) 45–55

appears to have unique features separate from TIA without infarction and ischemic stroke. We propose identifying TSI as a separate clinical syndrome with distinct prognostic features. D 2006 Published by Elsevier B.V.

Transient ischemic attack (TIA) is a clinical syndrome defined as an acute focal neurological deficit presumed to be of vascular origin that lasts less than 24 h [1]. TIAs are not necessarily transient at the tissue level; they are associated with variable rates of infarction, ranging from 4% to 77% with computed tomography (CT) and conventional magnetic resonance imaging (MRI) [2–5] and 35% to 67% with diffusion-weighted imaging (DWI) [6–10]. Imaging is currently the only mode to link a transient neurological event to an ischemic mechanism with certainty. A number of groups have already proposed imaging-based definitions that reserve the diagnosis of TIA only for patients with a brief episode of neurological dysfunction caused by focal brain or retinal ischemia and without evidence of acute infarction on neuroimaging; all other episodes with transient symptoms and relevant cerebral infarction on brain imaging are called ischemic stroke (IS) [11,12]. However, the concept of transient neurological symptoms associated with irreversible ischemic brain injury challenges such a definition of IS: if the presence of infarction is the common denominator to label a neurological event as an IS, then it remains to be elucidated how TIA-related infarcts differ from those associated with IS such that one patient completely recovers within minutes and another patient does not. A comprehensive characterization of infarcts associated with TIAs and their relation with symptom duration not only may help to clarify a meaningful definition of IS but also might help to understand the features of TIA that allow for rapid recovery. Therefore, we sought to identify clinical and imaging characteristics of TIA-related infarctions and to determine whether these features differ from those associated with imaging-negative TIA, as well as IS. 1. Patients and methods Over a period of 2 years between 2000 and 2002, a total of 99 consecutive patients who were admitted with the diagnosis of TIA and who had an acute MRI scan were identified. TIA was defined as an acute focal neurological deficit lasting less than 24 h that was attributable to ischemia [1]. Patients with isolated transient monocular blindness were not included. We excluded 12 patients because an MRI study could not be obtained: 6 had cardiac pacemakers, 1 had a cerebral aneurysm clip, 1 was severely obese, and 4 could not be scheduled. The head CT scan was normal in all of these patients. The time of onset and duration of transient symptoms were established blinded to the MRI findings through detailed interviews with patients and reliable observers. Each patient underwent an MRI study including DWI on the day they arrived at the hospital. MRI was performed on a 1.5-T whole-body scanner (GE Signa; GE Medical Systems, Milwaukee, WI). DWI was obtained using single-shot echo planar imaging with the following characteristics: repetition time, 7500 ms; TE, 99.3 ms; field of view, 22  22 cm2; image matrix, 128  128; slice thickness, 5 mm with 1 mm gap; 23 axial slices;

H. Ay et al. / International Congress Series 1290 (2006) 45–55

47

and b values of 0 and 1000 s/mm2 with 6 gradient directions and 3 averages. The in-plane resolution was 1.71  1.71 mm2. The original DWIs were corrected for motion and eddycurrent distortions using FMRIB’s Linear Image Registration Tool (FLIRT 5.0; FMRIB Image Analysis Group, Oxford, United Kingdom) [13]. DWIs, as well as maps of isotropic apparent diffusion coefficient (ADC), were computed from the corrected images. All infarcts characterized by increased signal intensity on DWI and decreased or normal signal intensity on ADC images were visually identified (small DWI bright lesions often are undetectable on ADC maps). All infarcts on DWI were outlined using a commercial image display and analysis program (ALICE; Hayden Image Processing Solutions, Denver, CO), and lesion volumes were computed. An infarct seen on a single slice was assumed to extend the entire thickness of the slice including the 1 mm gap between slices. This convention leads to overestimation of infarct volume because of partial volume averaging. In patients with multiple acute infarcts, volumes for each individual lesion, as well as the total volume of all lesions (infarct load), were recorded. The control group included 83 patients with IS who were consecutively admitted to the hospital over a 3-month period in 2000. IS was defined as neurological deficit that lasted longer than 24 h and was associated with imaging evidence of infarction. All but nine control patients underwent an MRI study at the time of admission. These nine patients included four with mechanical heart valve, two with claustrophobia, two with severe obesity, and one with clipped intracranial aneurysm. Infarcts were outlined and volumes were calculated in the remaining 74 patients using the same software described earlier. Ischemic lesions were classified into the following groups with respect to location: (1) isolated deep-brainstem penetrator infarction; (2) isolated cortical infarction; and (3) subcortical with or without cortical infarction (i.e., infarcts that involve both cortical and subcortical structures or isolated subcortical infarcts alone). Infarcts within the territory of circumflex branches of the basilar and vertebral arteries, as well as a penetrator infarct plus another cortical or subcortical infarct(s), were deemed to be subcortical with or without cortical infarction. All numerical variables were expressed as mean F standard deviation. Two-sided t tests and Fisher’s Exact Tests, whenever appropriate, were used to identify differences in baseline characteristics between TIA patients with or without infarction and to compare infarct volumes between patients with or without TIA. Logistic regression models were used to relate infarct volume and location to clinical phenotype. Spearman’s rank correlation coefficients were calculated to describe associations between infarct volume and symptom duration, time to MRI, and age in patients with TIA. Illustration of the distributions of duration of symptoms for TIAs with and without infarction was obtained using Gaussian kernel density estimation. Two-sample Kolmogorov–Smirnov test was used to compare the distribution of symptom duration in patients with positive and negative imaging results for TIA. p b 0.05 was considered statistically significant. 2. Results Table 1 presents the baseline characteristics of the study population. DWI identified a focal region of increased signal intensity consistent with acute infarction in 36 of 87

48

H. Ay et al. / International Congress Series 1290 (2006) 45–55

Table 1 Baseline characteristics of the study patients Characteristic

TSI (N = 36)

TIA without infarction (N = 51)

All TIA (N = 87)

Ischemic stroke (N = 74)

Age (mean F S.D.)a Female/Male ratio Symptom duration Time to MRI (mean F S.D.)

73 F 14 16/20 131 F 235 min 25 F 28 h

73 F 13 25/26 55 F 113 min 20 F 25 h

73 F 13 41/46 87 F 178 min 22 F 26 h

66 F 16 29/45 F24 h 25 F 33 h

TSI = transient symptoms associated with infarction; TIA= transient ischemic attack; S.D. = standard deviation; MRI = magnetic resonance imaging. a p b 0.05 for age between patients with TIA and ischemic stroke.

patients (41%). All but one patient with TIA was scanned after their symptoms resolved. There were only two patients each in DWI-positive and -normal groups with time to MRI less than 3 h, and seven patients with less than 6 h (three in DWI-positive and four in DWI-normal groups). Thus, most of our patients with TIA were scanned late enough that subsequent DWI reversibility was unlikely. For patients with TIA with no infarction, TIA with infarction (transient symptoms associated with infarction [TSI]), and IS, the mean length of hospital admission was 4.5, 5.5, and 9.1 days, respectively. During this period, Table 2 Imaging findings and infarct volumes in the study groups Finding

TSI (N = 36)

Ischemic stroke (N = 74)

Acute infarcts (no. of patients) Multiple infarcts Isolated penetrator infarction Brainstem Internal capsule or corona radiata Thalamus Caudate nucleus Isolated cortical infarctions Premotor cortex Motor cortex Primary sensory cortex Sensory and motor cortices Subcortical with/without cortical infarction Infarct volume (ml) (mean F S.D.)a Range Isolated penetrator infarctionb Isolated cortical infarctionsa Subcortical with or without cortical infarctionb Infarct load (ml) (mean F S.D.)b Range Isolated penetrator infarctiona Isolated cortical infarctionsa Subcortical with or without cortical infarctionb

17 10 2 3 4 1 6 2 3 1 0 20 0.66 F 1.20 0.07–10.47 0.70 F 0.48 0.54 F 0.30 0.67 F 1.25 1.50 F 1.87 0.17–10.47 0.70 F 0.48 0.54 F 0.30 1.80 F 2.11

25 16 6 9 1 0 5 2 2 0 1 53 27.32 F 61.55 0.28–380.43 1.94 F 1.35 3.26 F 2.90 29.87 F 64.21 50.22 F 80.13 0.68–380.43 1.94 F 1.35 8.09 F 6.82 63.54 F 80.13

TSI = transient symptoms associated with infarction; S.D. = standard deviation. a p b 0.001. b p b 0.05.

H. Ay et al. / International Congress Series 1290 (2006) 45–55

49

one TIA and no ISs occurred in the TIA with no infarction group, three ISs and four TIAs occurred in the TSI group, and one IS and no TIAs occurred in the IS group. 2.1. Spatial characteristics of transient ischemic attack-related infarctions Table 2 summarizes imaging findings in patients with TIA and IS. A total of 104 individual foci of infarct were counted in all 36 TIA patients with positive DWI results. At least one infarct in each patient was in a vascular territory implicated by symptoms. Infarcts in patients with TIA often were small. There were infarcts as small as 0.07 ml. A total of 96% of all infarcts were smaller than 1 ml. The smallest single lesion that was associated with TIA was 0.17 ml in volume. All other infarcts smaller than this threshold did not occur in isolation, but rather with another larger infarct or group of infarcts. To overcome the limitations associated with measuring multiple infarcts, we calculated the infarct load as an estimate for the total burden of tissue damage. Infarct load was less than 1 ml in 50%, less than 2 ml in 83%, and less than 3 ml in 94% of patients with TIA. No correlation between infarct load and symptom duration (r = 0.21; p = 0.22), age (r = 0.19; p = 0.27), and time to MRI (r = 0.24; p = 0.15) existed in the TIA population. There were a total of 173 foci of acute infarct in all 74 patients with IS. Mean infarct volume and infarct load were greater in patients with IS than in patients with TIA ( p b 0.001 for both volume and load). Fig. 1 indicates that there is no separation between TIA and IS over a continuum of infarct volume. As the volume became larger, the

Fig. 1. Relation between infarct load and clinical phenotype. The curves are three logistic regression models fit to all patients, as well as separately for patients with isolated deep-brainstem penetrator infarction and cortical and/or subcortical infarctions. Dots at the top and bottom of the graph represent lesion loads for individual patients with transient ischemic attack (TIA) and ischemic stroke (IS), respectively. Figure shows that lesion load occurs as a continuous function with overlap between TIA and IS. For a given probability of having TIA, the volume threshold is lowest for isolated deep-brainstem penetrator infarctions and highest for subcortical with or without cortical infarction. As lesion volume decreases, the probability of transient symptoms associated with infarction increases.

50

H. Ay et al. / International Congress Series 1290 (2006) 45–55

probability of IS increased, whereas smaller volumes were more likely associated with a TIA. The continuum between TIA and IS persisted when the infarct location was taken into account as isolated penetration or subcortical with or without cortical infarction (see Fig. 1 and Table 2). There were 32 patients with IS who had an infarct size overlapping with that which is seen in patients with TIA (infarct load between 0.17 and 10.47 ml; see Fig. 1). These 32 patients established a relevant control group for TSI to determine why some small infarcts are associated with symptoms lasting less than 24 h, whereas other small infarcts are not. The clinical and imaging features listed in Table 3 were included in a logistic regression model. Multivariate analysis showed that smaller infarct load ( p b 0.001), nonpenetrator location ( p b 0.01), and prior TIAs within the last month ( p b 0.05) were independently associated with TSI. A TIA-sized infarct that was associated with the clinical syndrome of IS was more likely to occur within the territory of penetrating arteries. Indeed, there were 15 patients with the clinical syndrome of IS and an infarct volume less than 3 ml. Of these 15 patients, 13 had a penetrator infarct. 2.2. Temporal characteristics of transient ischemic attack associated with infarction Symptoms lasted longer in TIA patients with a positive DWI than patients with normal DWI, yet this trend in mean symptom duration was not statistically significant ( p = 0.057). Fig. 2 demonstrates the probability density function of symptom duration. As shown in Fig. 2, symptom duration in TIA with infarction does not exhibit a continuum within the spectrum from TIA to IS (0–24 h); the probability density peaked early with a prompt return back to baseline. This is in contrast to the probabilistic distribution of infarct volume in Fig. 1, which illustrates that the transition from TSI to IS is smooth and uninterrupted. Table 3 Clinical and imaging determinants of clinical phenotype (TSI or IS) Determinant

TSI (N = 36)

IS with TSI-range infarction (N = 32)

Mean age (years) Sex (female) Hypertension Diabetes mellitus Coronary artery disease Atrial fibrillation Prior IS Prior TIA within the last montha Mean time to MRI (h) Mean infarct volume (ml)a Isolated deep penetrator infarction TOAST causative subtypes (no. of patients) Large artery atherosclerosisa Cardiac embolism Small artery occlusion Other/Undetermined

73 16 21 patients 10 patients 9 patients 11 patients 5 patients 18 patients 25.40 1.50 10 patients

69 13 21 patients 9 patients 8 patients 7 patients 8 patients 5 patients 30.30 3.80 16 patients

10 8 10 8

2 10 14 6

TSI = transient symptoms associated with infarction; IS = ischemic stroke; TIA= transient ischemic attack; MRI = magnetic resonance imaging; TOAST = Trial of Org 10172 in Acute Stroke Treatment. a p b 0.05.

H. Ay et al. / International Congress Series 1290 (2006) 45–55

51

Fig. 2. Temporal behavior of symptoms in patients with transient ischemic attack (TIA). The probability density function curve of symptom duration for transient symptoms associated with infarction (TSI) indicates the absence of continuity within the first 24 h. The probability density function is the probability that the variable takes a value in a given interval and is equal to 1 over its entire range of values. The area under curve is almost equal to 1 at around 200 min. Also note that the curves for TIA with or without infarction overlap ( p = 0.82). The distribution of duration of symptoms as seen here suggests that symptom duration is not a reliable feature to be used for predicting whether a transient neurological spell is associated with infarction. DWI = diffusion-weighted magnetic resonance imaging.

Fig. 2 also shows that the curves for TIA with or without infarction almost overlap with each other. No statistically significant difference existed between these distribution curves for symptom duration, suggesting that symptom duration cannot differentiate the imaging phenotypes in TIA ( p = 0.82; two-sample Kolmogorov–Smirnov test). 3. Discussion These results provide quantitative evidence in four domains that are critical in understanding the clinical behavior of ischemic brain syndromes. First, whereas in general TSI has smaller infarct volumes than IS, there is no unique size that differentiates TSI from IS. Second, the time course of clinical symptoms of TSI substantially overlaps with the time course of symptoms of TIA without infarction. Duration of symptoms cannot distinguish TIA from TSI in our cohort; the positive predictive value of symptom duration longer than 60 min for TSI was only 48%. Third, symptoms in brain infarction last either minutes or more than 1 day. We rarely (~10%) noted recovery after 200 min. A history of recent prior TIA (within 30 days), non-penetrator location, and smaller infarct size are independent predictors of small lesions that recovered (TSI) compared with small ISs (which, by definition, did not recover within 24 h). Fourth, confirming published reports, in our cohort, TSI appears to have substantially greater risk for recurrent IS than TIA without infarction or IS.

52

H. Ay et al. / International Congress Series 1290 (2006) 45–55

Each of these four findings contributes to the validity of TSI as a specific entity. The concept of spatial continuity, meaning that the lesions have continuous sets of volumes in similar locations in TSI and IS patients, might appear to refute our contention that TSI and IS are separate. However, the symptom duration and the short-term prognosis are so different between TSI and IS that we believe this overlap in lesion volume does not negate the need for differentiating TSI as a separate syndrome from IS. The symptom duration, either lasting minutes (TIA) or more than 24 h (IS) (but not in between), has always led to at least one point of distinction between TIA (with or without infarction) and IS [1,11,12] and still holds some relevance. However, if it were not for the difference in patient prognosis, this distinction might be considered only one of taxonomy. All published data in patients with acute brain ischemia suggest that TIA without infarction, TSI, and IS differ in short-term prognosis. In our cohort, the in-hospital IS rate was 0%, 8.3%, and 1.3% in patients with TIA without infarction, TSI, and IS, respectively. An earlier report of 57 patients describes the IS rate in TSI during the period of hospital admission (3–5 days) as 14.8%, whereas no stroke was observed in patients with TIA and normal imaging results [6]. These observational data are further supported by a recent CTbased study of 322 patients with TIA [2]. The 90-day IS rate in patients with TIA and acute infarct on CT was 38%, whereas it was only 10% in TIA patients with normal CT. Most of these strokes occurred within the first week after TIA. These risk estimates for TSI are much greater than the 7-day recurrent stroke risk rate of 1.5% [14] or 14-day risk rate of 2.9% to 3.8% [15] reported in the literature for IS, or the in-hospital recurrent TIA/ stroke risk rate of 1.3% in our study. Given that at least 50% of all strokes with a preceding TIA occur within the first 48 h of the TIA [16], the short-term prognostic estimates summarized earlier are indispensable and underscore the need for an imaging-based clinical approach to patient management. One weakness of our study is the lack of long-term follow-up. Although we acknowledge this, we note that long-term outcome data after TSI in the literature is also relatively scarce. A recent study of 83 patients with TIA in whom DWI was obtained [17] with a mean follow-up of 389 days reported that the combined IS and TIA rate was 29.6% and 14.3% in patients with or without infarction on DWI, respectively. The Kaplan–Meier curves for event-free proportion continued to separate beyond 1 week, suggesting that TIA with infarction requires a more focused approach on the underlying stroke mechanism for long-term stroke prevention. Categorizations in cardiology to describe ischemic coronary syndromes take spatial (end-organ injury), temporal (progression of electrocardiographic changes), and prognostic information into account [18–20]. Our proposal adopts a similar approach for ischemic brain syndromes. By analogy, bTSIQ corresponds to non-Q-wave myocardial infarction (minor myocardial necrosis associated with typical rise and gradual fall of biochemical markers of myocardial necrosis but no development of pathological Q waves on the electrocardiogram), whereas bischemic strokeQ corresponds to Q-wave myocardial infarction (discrete and large regions of myocardial necrosis associated with typical rise and gradual fall of biochemical markers of myocardial necrosis and development of pathological Q waves on the electrocardiogram). bTIA with no infarctionQ is similar to angina pectoris (symptoms suggestive of an acute coronary syndrome but ischemia is not severe enough to cause sufficient myocardial damage to release detectable quantities of a

H. Ay et al. / International Congress Series 1290 (2006) 45–55

53

marker of myocardial injury) and comprises a heterogenous group including both ischemic and probably nonischemic spells (TIA mimics). The diagnostic precision of the definitions proposed in this study largely depends on availability and timing, as well as sensitivity of the techniques used to establish the presence of ischemia or infarction. DWI is well suited to identify small infarcts in TIA, because, unlike CT and conventional MRI, DWI can mark a lesion as acute on a background full of chronic (and likely) silent infarcts [21]. This advantage, together with its greater sensitivity for small infarcts [22] and specificity for ischemic brain injury [23] compared with conventional MRI and CT, makes DWI the preferred choice of imaging in TIA. However, the timing of MR scanning appears to be important to accurately translate these findings into clinical practice, because DWI typically underestimates the final infarct volume when obtained early (during or soon after symptoms) [24] and late (months after symptoms) [7]. Therefore, we examined all patients within the first 24 h of hospital admission and all but one patient after their symptoms improved. This is indeed the typical clinical practice because most TIA patients seek medical attention after their symptoms fully resolve; only 0% to 7% of patients with TIA could be admitted and scanned at the height of their symptoms [6–8,25]. For this same reason, the small DWI lesions observed in patients with TIA most likely represent permanent brain injury, because the probability of DWI reversibility decreases as the time from symptom onset to imaging increases [7]. Examining only those patients with infarct size that occur in both IS and TSI, we find that the volume associated with TIA is small and is not constant throughout the brain; it varies with respect to lesion location. Importantly, this location dependency of the TIAvolume relation is not simply a feature of the clinical measurement method. In our data, TIA-related infarcts do not preferentially occur only in silent brain regions, rather they also occur in critical structures such as precentral gyrus, posterior limb of internal capsule, or brainstem, together with other brain regions with less clinical importance. We assume that there would be a similar phenotype-specific volume gradient in each brain region; however, a greater number of patients and more precise brain parcellations would be needed to show this. A comprehensive approach that takes various clinical and imaging characteristics into account suggests that it might be of more clinical relevance to keep bTIA with no infarctionQ, bTSIQ, and bISQ as separate phenotypes. Each phenotype is associated with distinct imaging, clinical, and prognostic features. TSI appears to be the most unstable phenotype; it is associated with the greatest risk for imminent stroke, implying that an underlying stroke mechanism is active. In contrast to the idea that bTIA with infarctionQ is just a minor IS, an increasing amount of data suggest that physicians should be alert that patients with such transient events require urgent and as much, if not more, intensive care to prevent recurrent ischemic events as patients with the syndrome of IS. Acknowledgement This study was supported by grants from the Agency for Health Research and Quality (R01-HS11392-02, W.J.K.) and the NIH (National Institute of Neurological Disorders and Stroke, R01-NS38477-04, A.G.S.; P41-RR14075, A.G.S.).

54

H. Ay et al. / International Congress Series 1290 (2006) 45–55

We thank T. Reese for providing the echo planar imaging pulse sequences and B. Oner for helping with image analysis. References [1] National Institute of Neurological Disorders and Stroke, Special report from the National Institute of Neurological Disorders and Stroke: classification of cerebrovascular diseases, III, Stroke 21 (1990) 637 – 676. [2] V.C. Douglas, et al., Head computed tomography findings predict short term stroke risk after transient ischemic attack, Stroke 34 (2003) 2894 – 2899. [3] F. Fazekas, et al., Magnetic resonance imaging correlates of transient cerebral ischemic attacks, Stroke 27 (1996) 607 – 611. [4] J. Bogousslavsky, F. Regli, Cerebral infarct in apparent transient ischemic attack, Neurology 35 (1985) 1501 – 1503. [5] I. Awad, et al., Focal parenchymal lesions in transient ischemic attacks: correlation of computed tomography and magnetic resonance imaging, Stroke 17 (1986) 399 – 403. [6] H. Ay, et al., bFootprintsQ of transient ischemic attacks: a diffusion-weighted MRI study, Cerebrovasc. Dis. 14 (2002) 177 – 186. [7] C.S. Kidwell, et al., Diffusion MRI in patients with transient ischemic attacks, Stroke 30 (1999) 1174 – 1180. [8] S.T. Engelter, et al., Diffusion MR imaging and transient ischemic attacks, Stroke 30 (1999) 2762 – 2763. [9] R.H. Bisschops, et al., Hemodynamic and metabolic changes in transient ischemic attack patients: a magnetic resonance angiography and (1)H-magnetic resonance spectroscopy study performed within 3 days of onset of a transient ischemic attack, Stroke 33 (2002) 110 – 115. [10] A. Rovira, et al., Diffusion-weighted MR imaging in the acute phase of transient ischemic attacks, AJNR Am. J. Neuroradiol. 23 (2002) 77 – 83. [11] G.W. Albers, et al., TIA Working Group, Transient ischemic attack – proposal for a new definition, N. Engl. J. Med. 347 (2002) 1713 – 1716. [12] B. Ovbiagele, C.S. Kidwell, J.L. Saver, Epidemiological impact in the United States of a tissue-based definition of transient ischemic attack, Stroke 34 (2003) 919 – 924. [13] M. Jenkinson, et al., Improved optimization for the robust and accurate linear registration and motion correction of brain images, Neuroimage 17 (2002) 825 – 841. [14] The Publications Committee for the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) Investigators, Low molecular weight heparinoid, ORG 10172 (danaparoid), and outcome after acute ischemic stroke: a randomized controlled trial, JAMA 279 (1998) 1265 – 1272. [15] International Stroke Trial Collaborative Group, The International Stroke Trial (IST): a randomised trial of aspirin, subcutaneous heparin, both, or neither among 19 435 patients with acute ischaemic stroke, Lancet 349 (1997) 1569 – 1581. [16] S.C. Johnston, et al., Short-term prognosis after emergency department diagnosis of TIA, JAMA 284 (2000) 2901 – 2906. [17] F. Purroy, et al., Higher risk of further vascular events among transient ischemic attack patients with diffusion-weighted imaging acute ischemic lesions, Stroke 35 (2004) 2313 – 2319. [18] J.S. Alpert, et al., Myocardial infarction redefined – a consensus document of The Joint European Society of Cardiology/American College of Cardiology Committee for the redefinition of myocardial infarction, J. Am. Coll. Cardiol. 36 (2000) 959 – 969. [19] A. Barbagelata, et al., Thrombolysis and Q wave versus non-Q wave first acute myocardial infarction: a GUSTO-I substudy. Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries Investigators, J. Am. Coll. Cardiol. 29 (1997) 770 – 777. [20] P.W. Armstrong, et al., Acute coronary syndromes in the GUSTO-IIb trial: prognostic insights and impact of recurrent ischemia. The GUSTO-IIb Investigators, Circulation 98 (1998) 1860 – 1868. [21] J. Oliveira-Filho, et al., Diffusion-weighted magnetic resonance imaging identifies the bclinically relevantQ small-penetrator infarcts, Arch. Neurol. 57 (2000) 1009 – 1014. [22] A. Gass, et al., Diffusion-weighted MRI for the bsmall stuffQ: the details of acute cerebral ischaemia, Lancet Neurol. 3 (2004) 39 – 45.

H. Ay et al. / International Congress Series 1290 (2006) 45–55

55

[23] M.E. Mullins, et al., CT and conventional and diffusion-weighted MR imaging in acute stroke: study in 691 patients at presentation to the emergency department, Radiology 224 (2002) 353 – 360. [24] H. Ay, et al., Normal diffusion-weighted MRI during stroke-like deficits, Neurology 52 (1999) 1784 – 1792. [25] R.A. Crisostomo, M.M. Garcia, D.C. Tong, Detection of diffusion-weighted MRI abnormalities in patients with transient ischemic attack: correlation with clinical characteristics, Stroke 34 (2003) 932 – 937.