Feasibility of myocardial dual-isotope perfusion imaging combined with gated single photon emission tomography for assessing coronary artery disease.

The clinical feasibility of both dual-isotope single photon emission tomography (SPET) and gated SPET have been described. The present study evaluates the feasibility of combining gated SPET with exercise 201Tl/rest 99mTc-tetrofosmin dual-isotope SPET corrected for scatter. Ninety-one patients with known or suspected coronary artery disease underwent cardiac catheterization and coronary angiography. Twenty-nine of them underwent exercise 201Tl/rest 99mTc-tetrofosmin dual-isotope SPET with a second 201Tl injection 3 h after the initial 201Tl injection (protocol 1). We then segregated a Bull's eye polar map into three coronary artery territories and quantified the relative regional uptake. The remaining 62 patients underwent exercise 201Tl/rest 99mTc-tetrofosmin dual-isotope SPET combined with gated SPET. We visually evaluated exercise and rest images from the three coronary artery territories. Left ventricular (LV) function was assessed globally by means of the LV ejection fraction and regionally by means of visual scoring analysis, compared with left ventriculography (LVG). The correlation between rest 99mTc-tetrofosmin and 201Tl reinjection images in 87 areas of coronary artery territory (r=0.89, P<0.01) and in 13 infarcted areas (r =0.94, P<0.01) was very close in protocol 1. The overall values for vessel-related sensitivity, specificity and accuracy were 88%, 79% and 82%, respectively, in protocol 2. The correlation between gated SPET and LVG was significant and linear with respect to the LV ejection fraction (r=0.77, P<0.01). The wall motion score from visual evaluation in gated SPET revealed a close overall agreement with LVG (concordance rate, 88%; kappa, 0.670). Exercise 201Tl/rest 99mTc-tetrofosmin dual-isotope SPET with scatter correction for assessing the coronary artery disease offers excellent diagnostic accuracy and the additional gated SPET provides useful information about LV function similar to that for LVG. This sequential protocol requires only 2 h to generate much useful clinical information.