Magnetic Resonance Imaging and Clinical Factors Associated With Ischemic Stroke in Patients Suspected of Cervical Artery Dissection

Background and Purpose— Cervical artery dissection is a major cause of ischemic stroke in the young and presents with various imaging findings, including stenosis and intramural hematoma (IMH). Our goal was to determine the relative contribution of lumen findings and IMH to acute ischemic stroke and whether a heavily T1-weighted sequence could more reliably detect IMH. Methods— Institutional review board approval was obtained for this retrospective study of 254 patients undergoing magnetic resonance imaging/magnetic resonance angiography for suspected dissection. Imaging included standard turbo spin–echo (TSE) T1-fat saturation and heavily T1-weighted flow-suppressed magnetization-prepared rapid acquisition gradient–recalled echo sequences. Subjects with stents (1) or atherosclerotic disease (26) were excluded, leaving 227 subjects. Kappa analysis was used to determine IMH interrater reliability on magnetization-prepared rapid acquisition gradient–recalled echo and T1-fat saturation in 4 vessels per subject. Lumen findings, cardiovascular risk factors, medications, and nondissection stroke sources were recorded. Mixed-effects multivariate Poisson regression was used to determine the prevalence ratio of each factor with acute ischemic stroke, accounting for 4 vessels per patient with backward elimination to a threshold P value of 0.10. Results— Patients were 41.9% men, mean age of 47.3±16.6 years, with 114 dissections and 107 strokes. IMH interrater reliability was significantly higher for magnetization-prepared rapid acquisition gradient–recalled echo (&kgr;=0.83; 95% CI, 0.78–0.86) versus T1-fat saturation (0.58; 95% CI, 0.57–0.68). The final acute stroke prediction model included magnetization-prepared rapid acquisition gradient–recalled echo-detected IMH (prevalence ratio, 2.0; 95% CI, 1.1–3.9; P=0.034), stenosis, pseudoaneurysm, male sex, current smoking, and nondissection stroke sources. The final model had high discrimination for acute stroke (area under the curve, 0.902; 95% CI, 0.872–0.932), compared with models without stenosis (0.861; 95% CI, 0.821–0.902), and without stenosis and IMH (0.831; 95% CI, 0.783–0.879). All 3 models were significantly different at P<0.05. Conclusions— Along with stenosis, IMH detection significantly contributed to acute ischemic stroke pathogenesis in patients with suspected cervical artery dissection. In addition, IMH detection can be made more reliable with heavily T1-weighted sequences.