Measurement of left ventricular volumes and ejection fraction by quantitative gated SPET, contrast ventriculography and magnetic resonance imaging: a meta-analysis.

Abstract

All previous validation studies of quantitative gated single-photon emission tomography (QGS) have examined relatively few patients, and the accuracy of QGS thus remains uncertain. We performed a meta-analysis of data from 301 participants in ten studies that compared QGS using technetium-99m-labelled tracers with contrast left ventriculography (LVG), and from 112 participants in six studies that compared QGS with magnetic resonance imaging (MRI). Linear regression and Bland-Altman analyses were used to evaluate pooled data from individuals across the studies. The correlation between QGS and LVG for end-diastolic volume (EDV) (r=0.81, SEE=27 ml), end-systolic volume (ESV) (r=0.83, SEE=18 ml) and ejection fraction (EF) (r=0.79, SEE=8.3%) was good, as was that between QGS and MRI for EDV (r=0.87, SEE=34 ml), ESV (r=0.89, SEE=27 ml) and EF (r=0.88, SEE=7.2%). However, Bland-Altman plots indicated that LVG minus QGS differences for EDV generated a systematic and random error of 32+/-58 ml (mean+/-2SD), and that MRI minus QGS generated an error of 13+/-73 ml. In the subgroup of patients in whom ECG gating was set at eight intervals, QGS significantly underestimated EF by 7.6%+/-17.4% (mean+/-2SD) compared with LVG and by 6.3%+/-14.6% compared with MRI; no such underestimation was observed in the subgroup in whom ECG gating was set at 16 intervals. We conclude that in patients with ECG gating set at eight intervals, QGS systematically underestimates LV volumes and EF compared with both LVG and MRI. Since QGS also shows considerable variations around the systematic deviations, there remains uncertainty over whether an individual value determined with QGS approximates the true LV volumes and EF.