Magnetic resonance imaging and histologic findings of experimental cerebral fat embolism

RATIONALE AND OBJECTIVES: The purpose of this study was to determine whether cerebral fat embolism demonstrated reversible or irreversible findings in magnetic resonance (MR) imaging over time and to compare the features in MR images with histologic findings in a cat model. MATERIALS AND METHODS: MR images were obtained serially at 2 hours, 1 and 4 days, and 1, 2, and 3 weeks after embolization with 0.05 mL of triolein into the internal carotid artery in 19 cats. Any abnormal signal intensity and change in the signal intensity were evaluated on T2-weighted images, T1-weighted images, diffusion-weighted images (DWIs; including apparent diffusion coefficient ADC] maps), and gadolinium-enhanced T1-weighted images (Gd-T1WI) over time. After MR imaging at 3 weeks, brain tissue was obtained and evaluated for light microscopic (LM) examination using hematoxylin-eosin and Luxol fast blue staining. For electron microscopic examination, the specimens were obtained at the cortex. The histologic and MR findings were compared. RESULTS: The embolization lesions showed hyperintensity on T2-weighted images, hyperintensity, or isointensity on DWIs, hypointensity, or isointensity on ADC maps and contrast enhancement on Gd-T1WIs at 2 hours. The T2-weighted hyperintensity extended to the white matter at day 1 and decreased thereafter. Contrast enhancement decreased continuously from day 1, and hyperintensity on DWI decreased after day 4. Hypointensity on ADC maps became less prominent after day 4. By week 3, most lesions had reverted to a normal appearance on MR images and were correlated with LM findings. However, small focal lesions remained in the gray matter of 8 cats and in the white matter of 3 cats on MR images, and this correlated with the cystic changes on LM findings. Electron microscopic examination of the cortical lesions that reverted to normal at week 3 in MR images showed that most of these lesions appeared normal but showed sporadic intracapillary fat vacuoles and disruption of the endothelial walls. CONCLUSIONS: The embolized lesions of the hyperacute stage were of 2 types: type 1 lesions, showing hyperintensity on DWIs and hypointensity on ADC maps, have irreversible sequelae, such as cystic changes; whereas type 2 lesions, showing isointensity or mild hyperintensity on DWIs and ADC maps, reverted to a normal appearance in the subacute stage.