Combination of T2*W and FLAIR Abnormalities for the Prediction of Parenchymal Hematoma Following Thrombolytic Therapy in 100 Stroke Patients

INTRODUCTION The objective of our study was to determine whether the combination of hypointense spots ("cerebral microbleeds," CMBs) with a leukoaraiosis is associated with the risk of parenchymal hematoma (PH) after thrombolytic therapy. PATIENTS AND METHODS We analyzed magnetic resonance imaging (MRI) scans acquired within 6 hours after symptom onset from 100 ischemic stroke patients. Multiparametric MRI including a T2*-weighted (T2*w) MRI and fluid attenuated inversion recovery (FLAIR) was performed before thrombolysis in all patients. Initial T2*w imaging was rated by two independent observers for the presence of CMBs smaller than 5 mm. White matter changes were evaluated using an adapted scale of Fazekas and Schmidt. PH was defined in follow-up imaging. FINDINGS A PH was observed in seven per 100 patients. CMBs were detected by observer 1 in 22 and observer 2 in 20 patients. We found a very low sensitivity (0.14) for prediction of PH by the presence of CMBs. We found a concordant increase in the rate of PH when the periventricular hyperintensity in FLAIR was larger than a thin lining. Sensitivity was good-to-perfect (0.86 and 1.00, observers 1 and 2) and specificity was substantial (0.65 and 0.66). Using the combination of a periventricular matter lesion (PVML)> 1 and the presence of CMBs did not improve the prediction of PH. DISCUSSION A marked periventricular hyperintensity in FLAIR imaging seems to be associated with a substantially increased risk of PH. A combination of CMBs with leukoaraiosis scores did not appear to be beneficial for prognosis. J Neuroimaging 2009;19:311-316.