Relationships between blood pool and myocardial perfusion-gated SPECT global and regional left ventricular function measurements.

Abstract

Algorithms have been developed to quantify global and regional left ventricular (LV) function and asynchrony from myocardial perfusion (MP) and blood pool (BP)-gated single-photon emission computer-assisted tomography, but relationships between measurements from these two imaging modalities have not been documented. The objective of this investigation was to determine the degree to which automated BP and MP measurements agree with each other and are accurate, using cardiac magnetic resonance (CMR) as the reference standard. We also sought to determine the extent to which regions of abnormal phase correspond to segments exhibiting abnormal wall motion. We studied 20 patients with prior myocardial infarction (age 60+/-11 years; 95% males) who had BP, MP, and ECG-gated CMR data acquisitions. MP and BP measured parameters included global ejection fraction (EF) and volumes, regional contraction phases, and standard deviations and bandwidths of phase versus R-R histograms. CMR algorithms used manually drawn endocardial and epicardial contours to measure global and regional wall motion and wall thickening. Regional measurements were resampled for all three imaging modalities into 17 conventional LV territories. BP LV counts significantly exceeded MP counts with a ratio of 5.2 : 1. There were no differences among the three methods for global EFs or volumes (analysis of variance P=0.86 and 0.94). MP and BP correlated equally well (P=0.15) versus CMR for global EFs (MP: r=0.87 and BP: r=0.95) and volumes (r=0.91 for both). Phase histogram parameters correlated significantly for MP versus BP for phase standard deviation (r=0.79) and phase bandwidth (r=0.93). Detection of five patients with significantly extended phase bandwidth, indicative of asynchrony, showed 'good agreement' between MP and BP (kappa=0.73; McNemar's difference=0%, P=0.48). Abnormal regional BP EF predicted abnormal wall motion of specific LV segments (receiver-operating characteristic area=85+/-2%), and abnormal regional MP wall thickening predicted abnormal CMR wall thickening (receiver-operating characteristic area=87+/-3%). Abnormal MP phase was present in 25% of 67 dyssynergic segments and 64% of segments adjacent to dyssynergic segments, indicating that locations of phase abnormalities were more widely distributed in the LV than sites of depressed wall motion. MP and BP measures of LV global and regional function agreed well with each other and with independent CMR measurements. MP and BP phase measurements suggested that phase abnormalities were more widespread than localized wall motion abnormalities.