13N-Ammonia PET/MR myocardial stress perfusion imaging early experience

Objectives Simultaneous acquisition Positron Emission Tomography/Magnetic Resonance (PET/MR) is a new technology with potential use in cardiac perfusion imaging. Stress perfusion imaging with 13N-ammonia-PET and MR has been previously validated separately for detection of coronary artery disease (CAD). In this pilot study, we optimize a protocol for cardiac PET/MR stress perfusion imaging, and evaluate its diagnostic accuracy for CAD when compared to GSPECT-MPI.

Methods 6 patients with reversible ischemia on GSPECT-MPI for whom coronary angiography (CA) was planned were recruited. Patients received 400mcg Regadenoson, followed 30 seconds later by simultaneous 13N-ammonia-PET (396 ± 26 MBq) and gadolinium (0.075 mmol/Kg) contrast MR perfusion imaging. The procedure was repeated at rest. PET attenuation correction µ-map was a dual echo VIBE Dixon sequence. PET images were reconstructed with 3D-OSEM with 3 iterations, 21 subsets and post-Gaussian filter of 4mm. Areas of decreased perfusion on MR were correlated to PET myocardial blood flow (MBF) by 2-compartment modeling analysis.

Results PET/MR showed excellent concordance with CA findings. Compared to GSPECT-MPI, PET/MR demonstrated comparable sensitivity (100% vs. 100%), superior specificity (100% vs. 60%) and diagnostic accuracy (100% vs. 60%). Areas of decreased perfusion on stress MR correlated with decreased MBF on stress PET (1.80 ± 0.02 vs 2.24 ± 0.26 mL/g/min for normal perfused areas, p < 0.01).

Conclusions Early experience with 13N-PET/MR perfusion imaging showed excellent diagnostic accuracy, sensitivity and specificity for CAD detection. Perfusion PET/MR offers a comprehensive ischemic evaluation with potential benefits of shorter exam time than SPECT, lower radiation dose, and internal validation between PET and MR. In addition, the ability to quantify absolute MBF, the use of cine MR to assess regional wall motion and late gadolinium enhancement to assess fibrosis, may improve diagnostic accuracy for CAD.

Research Support This work is supported by a grant from Astellas Pharma, Inc.