Abstract 3220: Does Multimodality MRI have Added Benefit in the Diagnosis and Management of “Classic” Hypertensive Intracerebral Hemorrhage?

Abstract

Introduction: It is unclear whether MRI adds benefit to non-contrast head CT (CT) in the diagnosis and management of patients with “classic” hypertensive intracerebral hemorrhage (HICH) on CT, and whether this additive information justifies MRI costs. We sought to answer this question using a prospective cohort of 159 consecutive patients with spontaneous ICH who systematically underwent CT and multimodality MRI. Methods: ICH etiology was classified in one of 12 predefined categories along with diagnostic certainty (highly probable, likely and possible), by two blinded neuroradiologists based on CT review (“CT diagnosis”). Two other blinded neuroradiologists reviewed in addition the MRI and determined the most likely diagnosis (“MRI diagnosis”). The “final” diagnosis was used as the reference standard and was assigned by two external, independent and blinded ICH clinician-experts who evaluated all clinical and imaging data including the initial and a 3 month MRI, CT angiography, contrast angiography, pathology and follow-up clinic visits, as available. Results: Of 159 patients, 86 (54%) had HICH as the final diagnosis. CT and MRI correctly identified 63 (73%) and 78 (91%) of these patients, respectively (P=0.005). Notably, MRI correctly classified eight of nine patients (89%) who had an “unknown” diagnosis by CT. Conversely, 74 patients were classified as HICH by the blinded CT review, and 64 (86%) of these patients had HICH as the final diagnosis. If the CT diagnosis was categorized as highly probable HICH (n=42), then it was almost always correct (98%). If the CT was categorized as likely (n=18) or possible (n=14) HICH, then it was correct in only 72% of cases. MRI increased diagnostic yield in the likely and possible categories by identifying one cavernous malformation (3%) and by improving diagnostic confidence in 19 patients (59%). MRI was wrong in three instances, all in the highly probable category (4%) by incorrectly classifying the etiology of intracerebral hemorrhage due to a possible vascular malformation (n=2) or coagulopathy (n=1). Conclusions: MRI is more accurate in correctly classifying hypertensive intracerebral hemorrhage than non contrast CT. MRI has significant additive yield over non contrast CT by improving diagnosis and diagnostic confidence in patients for whom the CT diagnosis is of intermediate or low certainty, but not for those diagnosed with a hypertensive intracerebral hemorrhage with high confidence.