Abstract WMP16: Improved Prediction of Hemorrhagic Transformation in Acute Stroke Patients Using Lowered Cerebral Blood Volume on MRI and Clinical Severity by NIHSS
Objectives: Hemorrhagic transformation of the ischemic lesion is a common thrombolytic complication when treating acute stroke patients with standard IV-tPA. In a retrospective study we demonstrated that patients that have HT post-thrombolysis have a significantly lowered cerebral blood volume (CBV) on their pre-thrombolysis MRI compared to non HT patients. The objective of this study was to determine prospectively whether reduced CBV in combination with stroke severity can predict the occurrence of HT in patients post-thrombolysis. Methods: Patients were selected from the NINDS Stroke Registry if they: had an acute ischemic stroke located in the MCA territory, were treated with standard IV-tPA, had a pre-treatment MRI with evaluable DWI and PWI, and had post-treatment MRI evaluation for HT. A rater calculated CBV maps in PMA™ (ASIST-Japan) and performed image registration and region of interest analyses in MIPAV™ (NIH). The rater repeated the CBV analysis without image registration using DICOM software available on the scanner. Multinomial regression with covariates of baseline NIHSS, DWI lesion size, and CBV ratio was performed. All MRI scans were reviewed by expert readers blinded to the CBV analyses to determine the presence of HT using the ECASS-II criteria. Results: Seventy-six patients met the study criteria with a mean (SD) age of 68.1 (±14.1) years, median baseline NIHSS of 12 (IQR25-75: 5-18) and median onset to first MRI of 109 minutes (IQR25-75: 82-157). Thirty-six percent of patients (27/76) were positive for HT post-thrombolysis. The mean CBV ratio was 0.25 (STD ± 0.23) in the positive HT patients compared to 0.55 (STD ±0.24) in the negative HT patients. Multinomial regression demonstrated that the CBV ratio < 0.5 (p<0.006) and baseline NIHSS > 15 (p<0.034) significantly predicted the occurrence of any HT. The CBV ratio < 0.5 was the only independent predictor of severe HT, PH1 or PH2 (p<0.008). DWI lesion size using > 100 ml involvement of the MCA territory was not significant in predicting any or severe HT. Conclusions: Pre-treatment CBV ratio used in combination with baseline NIHSS are promising predictors of HT after standard IV-tPA. The ability to calculate CBV ratios directly on the scanner supports usage in acute intervention decision making.