Comparison of Tc-99m sestamibi and Tl-201 gated perfusion SPECT

To determine the interpretability of gated thallium-201 perfusion SPECT compared with that performed by use of technetium-99m sestamibi (MIBI), 33 patients with prior myocardial infarction were studied. Patients received 22 to 30 mCi (814 to 1110 MBq) MIBI at peak stress, and a 15-minute gated SPECT acquisition was begun 30 to 40 minutes thereafter. On a subsequent day gated Tl-201 SPECT was acquired for 15 minutes, 4 hours after a resting 3.5 mCi (130 MBq) injection. SPECT was performed over a 180-degree arc by use of a 90-degree angled 2-detector camera.

Gated studies were interpreted independently by 4 experienced physicians. Study quality was graded (0 = uninterpretable to 4 = excellent). Wall motion (0 = normal to 2 = akinetic/dyskinetic) and wall thickening (0 = normal to 2 = absent) were graded for each of 10 segments viewed in orthogonal planes. Left ventricular ejection fraction (LVEF) was calculated by use of software thus far validated only for MIBI. The average count density of mid-ventricular end-diastolic short axis tomograms with sestamibi was 3.47 times greater than with thallium. Mean study quality was 3.4 for MIBI and 1.8 for thallium (P < 10(-6)). No gated MIBI SPECTs, but 2 gated thallium studies (6%) were judged uninterpretable. Among interpretable scans, interobserver agreement (Kendall statistic) in assessing wall motion was 0.73 for MIBI and 0.66 for thallium (P = .01). For assessing wall thickening, the Kendall statistic was 0.73 for MIBI and 0.69 for thallium (P = .05). Correlation (r) of LVEFs was 0.91, SEE = 6.4.

We conclude that gated thallium SPECT is inferior to MIBI because of much poorer image quality and somewhat poorer interobserver agreement among experienced physicians. However, LVEF can be determined reliably from gated thallium SPECT.