Scan reproducibility of magnetic resonance imaging assessment of aortic atherosclerosis burden

Subclinical atherosclerosis precedes the onset of clinical disease by many years. Noninvasive magnetic resonance imaging (MRI) offers the opportunity to visualize and quantify atherosclerotic plaque. However, the reproducibility of MRI measurements of abdominal and thoracic aortic atherosclerosis has not been reported. Electrocardiogram-gated, T2-weighted, turbo spin echo MRI of the descending thoracic and abdominal aorta was performed on 16 subjects, comprising 10 subjects with multivessel coronary artery disease (CAD) and 6 subjects without angiographic CAD. Three identical MRIs were performed on each subject, with subject repositioning between the second and third scans. Aortic anatomic and plaque measurements were performed in a blinded fashion. Fourteen subjects (88%) had MRI evidence of atherosclerotic plaque on at least one image. Slice plaque burden, plaque area, and plaque perimeter were greater in the CAD group (52% vs. 9%, p = 0.002; 264 vs. 18 mm2, p = 0.009; 159 vs. 15 mm, p = 0.006, respectively). Measurements of total aortic lumen area, lumen circumference, plaque area, and plaque perimeter correlated highly among the three scans (all r = 0.96, all p < 0.001). Measurements of slice-specific aortic lumen area and lumen circumference also correlated highly (all r = 0.98, all p < 0.001). Correlations of slice-specific plaque area and plaque perimeter were significant (all p < 0.001) but less robust (r = 0.62-0.85). These data demonstrate that MRI is a reproducible technique for assessing aortic anatomy and total aortic atherosclerosis, but increased slice density should be considered if serial evaluation of slice-specific data is desired.