Reproducibility of adenosine stress cardiovascular magnetic resonance in multi-vessel symptomatic coronary artery disease

PURPOSE First-pass perfusion cardiovascular magnetic resonance (CMR) is increasingly being utilized in both clinical practice and research. However, the reproducibility of this technique remains incompletely evaluated, particularly in patients with severe coronary artery disease (CAD). The purpose of this study was to determine the inter-study reproducibility of adenosine stress CMR in patients with symptomatic multi-vessel CAD and those at low risk for CAD. METHODS Twenty patients (10 with CAD, 10 low risk CAD) underwent two CMR scans 8 +/- 2 days apart. Basal, mid and apical left ventricular short axis slices were acquired using gadolinium 0.05 mmol/kg at peak stress (adenosine, 140 micro/kg/min, 4 min) and rest. Myocardial perfusion was evaluated qualitatively by assessing the number of ischemic segments, and semi-quantitatively by determining the myocardial perfusion reserve index (MPRi) using a normalized upslope method. Inter-study and observer reproducibility were assessed--the latter being defined by the coefficient of variation (CoV), which was calculated from the standard deviation of the differences of the measurements, divided by the mean. Additionally, the percentage of myocardial segments with perfect agreement and inter- and intra-observer MPRi correlation between studies, were also determined. RESULTS The CoV for the number of ischemic segments was 31% with a mean difference of -0.15 +/- 0.88 segments and 91% perfect agreement between studies. MPRi was lower in patients with CAD (1.13 +/- 0.21) compared to those with low risk CAD (1.59 +/- 0.58), p = 0.02. The reproducibility of MPRi was 19% with no significant difference between patients with CAD and those with low risk CAD (p = 0.850). Observer reproducibility for MPRi was high: inter-observer CoV 9%, r = 0.93 and intra-observer CoV 5%, r = 0.94. For trials using perfusion CMR as an endpoint, an estimated sample size of 12 subjects would be required to detect a two-segment change in the number of ischemic segments (power 0.9, alpha 0.05). CONCLUSIONS Adenosine stress CMR, by qualitative and semi-quantitative normalized upslope analyses are reproducible techniques in both patients with multi-vessel CAD and those without known CAD. The robust inter-study reproducibility of perfusion CMR supports its clinical and research application.