Correlation of coronary artery stenosis evaluation with left heart structure and function by multi-slice computed tomography.
UNLABELLED Because of the ultrashort tracer half-life and high positron energy of (82)Rb, PET images acquired with this tracer are noisier and of lower resolution than those obtained with other PET tracers. The validity of electrocardiographic gating using (82)Rb for assessment of left ventricular (LV) function is not well established. To support feasibility, we compared functional parameters from gated (82)Rb PET with simultaneous high-resolution contrast-enhanced CT ventriculography, obtained as a byproduct a CT coronary angiography during hybrid cardiac PET/CT. METHODS A total of 24 patients underwent PET/CT, consisting of rest and dipyridamole (82)Rb perfusion studies and contrast-enhanced CT angiography, using a 64-slice scanner, for the workup of coronary artery disease. From gated PET images, LV ejection fraction (EF), end-diastolic volume (EDV), and end-systolic volume (ESV) were calculated using 2 commercial products. For functional CT analysis, commercial software using endocardial contour detection was applied. RESULTS Inter- and intraobserver agreement was good for all methods. On CT, EF was 66% +/- 13%, ESV was 41 +/- 29 mL, and EDV was 115 +/- 36 mL. On PET, EF during dipyridamole was 56% +/- 15% and 52% +/- 15% using the 2 commercial products (P < 0.05 vs. CT), ESV was 36 +/- 28 and 47 +/- 35 mL (P = not significant vs. CT), and EDV was 75 +/- 30 and 91 +/- 33 mL (P < 0.05 vs. CT). Correlations with CT were 0.85 and 0.87 for EF using commercial software, 0.76 and 0.88 for ESV, and 0.60 and 0.68 for EDV (P < 0.01 for all). Bland-Altman analysis confirmed systematic underestimation of EF and EDV by PET versus CT but did not show a significant deviation from linearity. CONCLUSION Global LV function can be measured reproducibly from gated (82)Rb PET, using different available software products. However, underestimation of EF by (82)Rb PET, compared with CT ventriculography, is present, which is a result of underestimation of EDV from count-poor ED frames. This underestimation needs to be considered for clinical interpretation of (82)Rb PET.