Background: The accuracy of quantitative gated single photon emission computed tomography (SPECT) (QGS) and the potential limitations for estimation of left ventricular ejection fraction (LVEF) have been extensively evaluated. However, few studies have focused on the serial variability of QGS. This study was conducted to assess the serial variability of QGS for determination of LVEF between 2 sequential technetium 99m sestamibi-gated SPECT acquisitions at rest in both healthy and unhealthy subjects.
Methods and results: The study population consisted of 2 groups: group I included 21 volunteers with a low likelihood of CAD, and group II included 22 consecutive patients with documented CAD. Both groups underwent serial SPECT imaging. The overall correlation between sequential images was high (r = 0.94, SEE = 5.3%), and the mean serial variability of LVEF was 5.15% +/- 3.51%. Serial variability was lower for images with high counts (3.45% +/- 3.23%) than for images with low counts (6.85% +/- 3.77%). The mean serial variability was not different between normal and abnormal high-dose images (3.0% +/- 1.56% vs 3.9% +/- 2.77%). However, mean serial variability for images derived from abnormal low-dose images was significantly greater than that derived from normal low-dose images (9.6% +/- 2.22% vs 3.1% +/- 2.12%, P <.05).
Conclusions: Although QGS is an efficacious method to approximate LVEF values and is extremely valuable for incremental risk stratification of patients with coronary artery disease, it has significant variability in the estimation of LVEF on serial images. This should be taken into account when used for serial evaluation of LVEF.