Left ventricular mass measured by myocardial perfusion gated SPECT. Relation to three-dimensional echocardiography.

PURPOSE

The purpose of this investigation was to determine whether left ventricular mass (LVM) assessed from myocardial perfusion gated SPECT (GSPECT) data corresponds with echocardiographic estimates, and whether mass accuracy decreases as relative myocardial wall thickness increases.

MATERIALS AND METHODS

Myocardial perfusion tomograms were selected retrospectively for 37 patients, of whom 18 had Tl-201 and 19 had Tc-99m sestamibi GSPECT poststress data collections, which were subsequently processed using quantitative gated SPECT software (Cedars Sinai Medical Center, Los Angeles, CA). These patients also had clinically indicated echocardiograms for assessment of wall thickness and possible valvular involvement. In addition, LV internal diameter and posterior wall thickness were measured at end-diastole by two-dimensional guided M-mode echocardiography to assess relative myocardial wall thickness, and LVM was measured by three-dimensional echocardiography using an acoustic spatial locator device.

RESULTS

LVM values were not significantly different between GSPECT and three-dimensional echocardiography (153 +/- 39 g versus 146 +/- 35 g, respectively; P = NS). GSPECT correlated significantly (r = 0.63, P < 0.0001) with three-dimensional echocardiography, with a mean difference of 7 +/- 32 g but a substantial root mean squared error of 31 g. Results were similar for similar mass ranges when subgrouped by isotope and by the presence of significant myocardial perfusion defects. Results were independent of relative myocardial wall thickness determined by two-dimensional echocardiography. The two methods yielded similar results in the highest mass range of 400 to 500 g.

CONCLUSIONS

GSPECT and three-dimensional echo LVM correlated significantly, but given the large spread of statistical errors, these two techniques should not be considered interchangeable. Because gamma camera resolution is limited, GSPECT LVM should be viewed as an approximation.