Accuracy of radionuclide ventriculography assessed by magnetic resonance imaging in patients with abnormal left ventricles

BACKGROUND: We compared gated blood pool single photon emission computed tomography (SPECT) (GBPS), planar gated blood pool imaging (planar GBP), and cardiac magnetic resonance (CMR) measurements of left ventricular (LV) end-diastolic volume (EDV) and ejection fraction (EF) in patients with abnormal left ventricles. METHODS AND RESULTS: LV functional parameters were measured for 40 subjects (age, 59 +/- 13 years; 85% male) by GBPS, planar GBP, and CMR. GBPS data were analyzed by use of count-threshold software (BP-SPECT) and surface gradient software (QBS). Limits of agreement with CMR for EF were -5% to +18%, -15% to +14%, and -15% to +16% for BP-SPECT, QBS, and planar GBP, respectively. However, limits of agreement with CMR for LV EDV were wide by both GBPS methods: -118 mL to +55 mL and -143 mL to +22 mL for BP-SPECT and QBS, respectively. Bland-Altman reproducibility limits for EF were -9% to +8%, -6% to +9%, and -7% to +7% by BP-SPECT, QBS, and planar GBP, respectively, and those for EDV were -46 mL to +48 mL and -31 mL to +35 mL by BP-SPECT and QBS, respectively. CONCLUSION: GBPS LV EF measurements agree with measurements by CMR and are as reproducible as planar GBP measurements. However, wide limits of agreement of radionuclide versus CMR values suggest that caution must be applied in interpreting GBPS LV volume results, especially for patients with markedly abnormal left ventricles.