Simultaneous PET/MRI - A new tool for translational brain imaging early after stroke

Objectives: MRI-guided systemic thrombolysis beyond the 4.5h window failed to improve clinical outcome in patients with ischemic stroke. Limited accuracy of MR-based perfusion measurement is considered a possible reason for this shortcoming. We wanted to assess acute stroke in a new large animal model and in patients by simultaneous [15O]H2O PET/MRI to cross-validate PET-based and MRI-based blood flow measurements. Methods: 10 stroke patients (duration of symptoms=11.9±6.5hrs) without indication for thrombolytic therapy and three merino sheep after permanent middle cerebral artery occlusion underwent combined PET/MRI. [15O]H2O PET data were simultaneously acquired with PWI/DWI, FLAIR, T2*, PASL, T1MPRAGE, and MR angiography. PerfusionPET maps were calculated employing invasive or image-derived arterial input functions and full kinetic modeling (one-compartment model). Maps of perfusionMRI (Tmax/TTPdelay/CBF) were obtained (software PMA). Results: [15O]H2O PET/MRI is feasible in the large animal stroke model and in the acute stroke setting in patients. Only a weak correlation was found between perfusionMRI (Tmax) and perfusionPET (r=-0.24; p< 0.001) and a Bland-Altman analysis of CBF from both modalities revealed large limits of agreement (-0.55 to 0.4ml/g/min) in humans. A moderate correlation between perfusionMRI (Tmax) and perfusionPET was observed in sheep (r=-0.48; p<0.001). Conclusions: PET/MRI has great potential in validating/adjusting MR-based multimodal stroke imaging. It can serve as a reliable tool for monitoring preclinical stroke therapies. The high variability of perfusionMRI compared to the perfusionPET gold standard can result in serious under- or overestimation of perfusion disturbances in MRI and may thus lead to incorrect patient selection for late thrombolytic therapy in ongoing MR-based studies.