Determination of coronary flow reserve by parametric imaging

Abstract

Nine mongrel dogs were instrumented with electromagnetic flow probes (EMF) to measure coronary blood flow through the left anterior descending (LAD) and left circumflex (LCx) coronary arteries at rest and after maximal coronary vasodilation (1 mg/kg/min adenosine). Relative coronary blood flow was determined by parametric imaging in the left posterior oblique projection using digital subtraction angiography (DSA). Transmural myocardial perfusion of the LAD and LCx beds was determined with tracer-labeled microspheres. Coronary flow reserve (maximal coronary blood flow divided by resting blood flow) was calculated under control conditions and after constriction of the proximal LAD or LCx by a screw occluder. Heart rate decreased significantly from 140 beats/min at rest to 122 beats/min after adenosine (p<0.001) and from 134 (rest) to 120 beats/min (adenosine; p<0.05) after coronary constriction. Peak systolic pressure was kept constant with an aortic constrictor. Left ventricular end-diastolic pressure increased significantly from 18 mm Hg at rest to 23 mm Hg (p<0.05) after coronary constriction. At baseline, coronary flow reserve was 4.2 with DSA, 3.8 with EMF, and 3.7 with microspheres; after coronary constriction, it was 2.6 (DSA), 1.9 (EMF), and 1.5 (microspheres) (all p<0.001 versus baseline). Coronary blood flow showed a good correlation between EMF and microspheres (r=0.87, p<0.001), with a standard error of estimate (SEE) of 0.78 ml/g/min. Coronary flow reserve also showed a good correlation between EMF and microspheres (r=0.82, p<0.001), with an SEE of 0.93. There was a moderate correlation between EMF and DSA (r=0.68, p<0.001), with an SEE of 1.35 (40% of mean coronary flow reserve). The correlation coefficient between microspheres and DSA was 0.54 (p<0.01), with an SEE of 1.46 (39% of mean coronary flow reserve). The mean difference (accuracy) and standard deviation of difference (precision) were 0.2±1.0 between EMF and microspheres, -0.1±1.4 between EMF and DSA, and -0.6±1.7 between microspheres and DSA. We conclude that determination of coronary flow reserve by parametric imaging is associated with large variations that are greater than variations also inherent in the two reference techniques. Parametric imaging allows relatively accurate assessment of coronary flow reserve (small mean difference), but precision is low (large standard deviation of mean differences). This low precision probably is due to the superposition of different cardiac structures in the two-dimensional display of a three-dimensional perfusion zone, potentially inhomogeneous contrast distribution, poor temporal resolution of the once-per-cycle imaging, inadequate displacement of blood by contrast material, and perturbations of flow caused by contrast material. Thus, in this animal model, only changes in coronary flow reserve that are clearly greater than 40% can be expected to be measured correctly with the currently available technique.