Individuals who experience a transient ischemic attack (TIA) are at high risk of stroke [1–3]. The stroke incidence is ~10% at 90 d after TIA and ~15% at 1 year after TIA . The 1988 World Health Organization (WHO) criteria define TIA as focal neurological symptoms lasting less than 24 h with no clear non-vascular cause . Diffusion- weighted imaging (DWI) enables detection of acute ischemic lesions [5, 6]. The clinical significance of the presence of such lesions in patients with TIA has gained recent attention, as acute ischemic lesions are reportedly independently predictive of stroke [7–10]. Therefore, a tissue-based definition of TIA (i.e., focal neurological symptoms lacking acute ischemic lesions) is emerging [11–14].Clinical features of patients with lesion-positive DWI are unclear, and findings vary [15–21]. Clinical and etiological characteristics of patients with TIA and lesion-positive DWI have been examined. A systematic review of 19 studies (combined n = 1,242; individual studies, median n = 58, range n = 14–200) concluded that atrial fibrillation and ipsilateral carotid stenosis of >50% were associated with lesion-positive DWI, whereas age, sex, hypertension, and diabetes history were not . Symptoms of motor weakness, dysphagia, and dysarthria were associated with lesion-positive DWI, as was symptom duration >60 min . Conversely, the rate of lesion-positive DWI was not significantly related to the presence of atrial fibrillation in a Japanese study of 464 patients . Therefore, the significance of the lesionpositive DWI is unclear.
Here, we assessed the clinical characteristics, features, and outcomes of lesion-positive DWI in patients with TIA admitted to a stroke center.
Materials and Methods
AThis was a retrospective study on a prospectively collected stroke database. We evaluated 235 patients with TIA who visited our stroke center between January 2009 and December 2015. Vascular neurologists made diagnoses in accordance with the WHO criteria, which define TIA as an acute loss of focal cerebral function lasting less than 24 h . DWI was performed routinely in all patients with TIA upon admission (i.e., within 24 h of symptom onset) if the patient had no contraindications for magnetic resonance imaging (MRI). Lesion-positive DWI was defined as an increased signal on DWI with a reduced apparent diffusion coefficient.
Department of Neurology physicians recorded detailed baseline data for patients with TIA including demographics, clinical features, imaging features, and precise potential sources of cardioembolism, per the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification . Modified Rankin scale (mRS) scores were acquired at admission and at discharge from the hospital . All data collected were recorded prospectively in the medical center registry; the 3-month follow-up mRS score was determined via a telephone survey administered by a research nurse affiliated with the Department of Neurology . For categorical variables, Pearson’s chi-square test and Fisher’s exact test were used to determine correlations with lesion-positive DWI. Association of lesion-positive DWI with cardioembolic risk stratification according to TOAST classification was calculated by excluding the effect of atrial fibrillation using the Cochran–Mantel–Haenszel test. Relative risk associated with each variable for lesion-positive DWI was calculated. For continuous variables, Student’s t-test was used. Statistical analysis was performed using SPSS software.
The proportion of patients with lesion-positive DWI was 14.0% among 235 subjects with TIA. Continuous etiological variables are presented in Table 1, and categorized variables are presented in Table 2. Only male sex, history of previous stroke, atrial fibrillation, and medium- to high-risk source of cardioembolism (TOAST classification) were found to be significantly associated with the presence of lesion-positive DWI. Atrial fibrillation showed the highest relative risk of cardioembolism. Medium- to high-risk sources of cardioembolism were not significantly related to lesion-positive DWI when the presence of atrial fibrillation was excluded (p = 0.108; Cochran–Mantel–Haenszel test). Although not significant, a trend toward association of sources of high-risk for cardioembolism (TOAST classification) with lesionpositive DWI was noted if atrial fibrillation was excluded (p = 0.052; Cochran–Mantel–Haenszel test). No association between the presence of lesion-positive DWI and clinical features was found (Table 3). Whether the patients had lesion-positive DWI was not related to an increase in mRS score during the hospital stay or at the 3-month follow-up after discharge (Table 4).
We found that etiological factors associated with a high risk of stroke were also associated with the presence of lesion-positive DWI in patients with TIA. These factors included male sex, previous history of stroke, and presence of atrial fibrillation. A high risk of cardioembolism according to TOAST classification showed a trend toward association with lesion-positive DWI, but the relationship was not statistically significant when atrial fibrillation was excluded. No clinical features were associated with the presence of lesion-positive DWI. These findings differ from the previously reported large systematic review, which showed that traditional vascular risk factors (age, sex, diabetes, hypertension, and blood pressure) were not associated with lesion-positive DWI, while clinical features, such as motor weakness, dysarthria, aphasia, and time from onset > 60 min, were . Although the association with the time from onset was not assessed in this study, the admission route and symptom onset during sleep or waking were assessed (Table 2). Patients admitted via the outpatient clinic and those with symptom onset noted during sleep likely underwent DWI longer after onset than did inpatients and those who first experienced symptoms while awake. However, no significant association was detected.
The differential findings may be due to differences in ethnicity, as well as differences in the medical environment. The hospital environment affected the time from onset to radiological assessment in an observational study . Detection of ischemic lesions is difficult 1–12 h after TIA; therefore, different treatment and imaging protocols in different medical environments may affect detection of lesion-positive DWI in patients with TIA . On the other hand, in a study of multi-modal MRI in patients with TIA, new lesions were found in 46.7% of patients who initially showed normal DWI findings at a follow-up a few days after onset . Therefore, DWI findings in the acute TIA setting may not represent the presence of true tissue ischemia; variations in the time from onset and lack of DWI follow-up may be a limitation of current reports.
For prognosis, the mRS score was analyzed according to the presence of lesions on DWI, and no significant effect was noted. This is in accordance with a systematic review on outcomes of patients with non-disabling stroke, where the prognosis was similar between the DWI-negative and -positive groups , but contrary to previous reports where lesion-positive DWI was independently predictive of stroke recurrence [7–9].Several limitations of this study should be noted. First, this study was a retrospective analysis of patients with TIA. Second, this study had a small sample size, which limited the statistical power, and it was performed at a single center.
In conclusion, the etiological factors associated with lesion-positive DWI in patients with TIA were male sex, previous stroke history, and atrial fibrillation. Clinical features and prognosis were similar between lesion-positive and negative groups. Future studies should include multi-center samples with large numbers, considering each unique medical environment. Routine acquisition of follow-up DWI for proper evaluation of the tissue-based definition of TIA should also be considered.