Effect of Outflow Tract Contributions to 82Rb-PET Global Myocardial Blood Flow Computations

Abstract

Algorithms are able to compute myocardial blood flow (MBF) from dynamic PET data for each of the 17 left ventricular segments, with global MBF obtained by averaging segmental values. This study was undertaken to compare MBFs with and without the basal–septal segments. Methods: Data were examined retrospectively for 196 patients who underwent rest and regadenoson-stress 82Rb PET/CT scanning for evaluation of known or suspected coronary artery disease. MBF data were acquired in gated list mode and rebinned to isolate the first-pass dynamic portion. Coronary vascular resistance (CVR) was computed as mean arterial pressure divided by MBF. MBF inhomogeneity was computed as the ratio of SD to mean MBF. Relative perfusion scores were obtained using 82Rb-specific normal limits applied to polar maps of myocardial perfusion generated from myocardial equilibrium portions of PET data. MBF and CVRs from 17 and 14 segments were compared. Results: Mean MBFs were lower for 17- than 14-segment means for rest (0.78 ± 0.50 vs. 0.85 ± 0.54 mL/g/min, paired t test P < 0.0001) and stress (1.50 ± 0.88 vs. 1.67 ± 0.96 mL/g/min, P < 0.0001). Bland–Altman plots of MBF differences versus means exhibited nonzero intercept (−0.04 ± 0.01, P = 0.0004) and significant correlation (r = −0.64, P < 0.0001), with slopes significantly different from 0.0 (−7.2% ± 0.6% and −8.3% ± 0.7% for rest and stress MBF; P < 0.0001). Seventeen-segment CVRs were higher than 14-segment CVRs for rest (159 ± 86 vs. 147 ± 81 mm Hg/mL/g/min, paired t test P < 0.0001) and stress CVR (85 ± 52 vs. 76 ± 48 mm Hg/mL/g/min, P < 0.0001). MBF inhomogeneity correlated significantly (P < 0.0001) with summed perfusion scores, but values correlated significantly more strongly for 14- than 17-segment values for rest (r = 0.67 vs. r = 0.52, P = 0.02) and stress (r = 0.69 vs. r = 0.47, P = 0.001). When basal segments were included in MBF determinations, perfusion inhomogeneity was greater both for rest (39% ± 10% vs. 31% ± 10%, P < 0.0001) and for stress (42% ± 12% vs. 32% ± 11%, P < 0.0001). Conclusion: Averaging 17 versus 14 segments leads to systematically 7%–8% lower MBF calculations, higher CVRs, and greater computed inhomogeneity. Consideration should be given to excluding basal–septal segments from standard global MBF determination.