Use of MRI to predict symptomatic haemorrhagic transformation after thrombolysis for cerebral ischaemia.

BACKGROUND AND OBJECTIVE: Predictors of symptomatic haemorrhagic transformation (s-HT) of cerebral ischaemia after intravenous recombinant tissue-plasminogen activator (rt-PA) were identified in studies using CT scans. We evaluated whether MRI can identify other predictors. METHOD: We analysed predictors of s-HT in a cohort of consecutive patients who received intravenous rt-PA for cerebral ischaemia after MRI at baseline. We used receiver operating characteristic curves considering an area under the curve (AUC) of 0.70 or higher as indicating acceptable discrimination. RESULTS: Of 944 patients, 49 patients (5.2%) developed s-HT. Clinical factors independently associated with s-HT were age (adjusted OR (adjOR) 1.03 for 1 year increase; 95% CI 1.01 to 1.05), excessive alcohol consumption (adjOR 3.13; 95% CI 1.32 to 7.42), recent transient ischaemic attack (adjOR 2.88; 95% CI 1.04 to 7.95) and baseline national institutes of health stroke scale score (adjOR 1.06 for 1 point increase; 95% CI 1.02 to 1.10). MRI predictors were vascular hyperintensities (adjOR 3.89; 95% CI 1.50 to 10.08), old infarcts (adjOR 2.01; 95% CI 1.11 to 3.66) and volume of diffusion-weighted imaging (DWI) abnormality (adjOR 1.02 for 1 cm3 increase; 95% CI 1.01 to 1.03). The only variable with an acceptable discrimination was volume of DWI abnormality (AUC 0.72; 95% CI 0.64 to 0.79), a value of 4 cm3 predicting s-HT with a 78% sensitivity and 58% specificity. Variables that can be assessed only with MRI did not predict s-HT. CONCLUSION: Although the volume of DWI abnormality predicts s-HT, other imaging characteristics that can only be assessed with MRI were not significantly associated with s-HT. Trial registration number NCT01614080.

Remote brain hemorrhage after IV thrombolysis: Role of preexisting lesions.

OBJECTIVE: To test the hypothesis that remote intracerebral hemorrhages (r-ICHs) after IV thrombolysis occur in preexisting brain lesions. METHOD: We prospectively collected baseline data from consecutive patients treated with IV thrombolysis for cerebral ischemia and reviewed their baseline MRI scans to identify preexisting lesions in those who developed r-ICH. We evaluated outcomes with the modified Rankin Scale (mRS) and defined good outcomes as scores of 0 to 2 or similar to the preexisting mRS score. RESULTS: Of 944 patients, 24 (2.5%) had r-ICH: lobar in 14, deep in 7, and both in 3. Sixteen of them (1.7% of all patients, 66.7% of those with r-ICH) were asymptomatic. Of the 41 r-ICHs found in these patients, 17 (41%) occurred within a lesion present before thrombolysis: 6 cerebral microbleeds (CMBs), 6 old and 1 recent infarct, and 4 areas of white matter hyperintensity. Patients with r-ICH were more likely to have strictly lobar CMBs (p = 0.049). They were 10 years older (p = 0.007), had a 16-mm Hg higher systolic blood pressure (p = 0.035) at baseline, and had more CMBs (p = 0.007). r-ICHs were better predicted by clinical (age, baseline systolic blood pressure) than imaging (purely lobar CMBs and having >5 CMBs) variables. r-ICHs tended to be associated with worse outcomes. CONCLUSION: We identified preexisting brain lesions in nearly half of the patients with r-ICH. All were of vascular origin, supporting the hypothesis that r-ICHs occur in preexisting brain lesions. Higher-field machines could help identifying preexisting lesions in those who developed r-ICH in an apparently normal area.