Simplified Estimation of Myocardial Flow Reserve using Stress/Rest Activity Ratios with Nitrogen-13-Ammonia and Rubidium-82 PET

Background: Myocardial blood flow (MBF) quantification is a powerful technique that provides incremental diagnostic and prognostic information over relative perfusion images. This technique requires that tracer injection happens while the patient lies in the PET camera, which preclude PET myocardial perfusion imaging with treadmill stress testing. The current study examines the application of a simplified estimate of MFR that could potentially be applied to exercise ¹³NH₃ and ⁸²Rb PET studies without the need for dynamic time-series imaging.

Methods: Rest and dipyridamole stress dynamic PET imaging was performed in consecutive patients using ⁸²Rb PET in one cohort (n=50) and ¹³NH₃ PET in another cohort (n=50). Reference standard MFR was quantified using a standard one-tissue-compartment model and tracer-specific extraction fractions. Stress/rest myocardial activity ratio (MAR) was calculated as late tracer uptake, and as activity product (AP) corrected for area under the arterial blood input curve. Optimized inverse functions were used to compute unbiased MFR estimates from the MAR and AP values.

Results: MFR, MAR, and AP values were 2.4±0.7, 1.2±0.2, and 1.4±0.2 in the ⁸²Rb PET cohort and 2.3±0.7, 1.5±0.3, 1.9±1.0 in the ¹³NH₃ cohort. In the ⁸²Rb PET cohort, the functions MFR = (MAR/0.82)^2.3 and MFR = (AP)^2.3 produced unbiased estimates of MFR_MAR and MFR_AP (2.39±0.78 vs 2.37±0.78, p=0.776) with good correlation (R=0.63). MFR_MAR (AUC = 0.830, p<0.001) exhibited similar accuracy to MFR_AP (AUC = 0.906, p<0.001) to identify patients with impaired flow reserve (MFR<2.0). In the ¹³NH₃ cohort post-hoc optimization of the inverse power functions MFR = (MAR/0.96)^1.9 and MFR = (AP/0.57)^0.7 also produced unbiased estimates of MFR_MAR and MFR_AP (2.26±0.80 vs. 2.28±0.82 p=0.873). MFR_MAR exhibited good accuracy (AUC = 0.891, p<0.001) to identify patients with impaired flow reserve (MFR<2.0) while MFR_AP exhibited relatively poor accuracy (AUC = 0.659, p=0.062). Conclusion: A simplified approach was demonstrated to estimate MFR with good accuracy using MAR or AP for ⁸²Rb PET and MAR for ¹³NH₃, potentially removing the need for dynamic imaging and tracer kinetic modelling.

• Download figure
• Open in new tab
• Download powerpoint