Comparison of CT and catheter arteriography for evaluation of peripheral arterial disease

Computed tomographic arteriography (CTA) has emerged as a promising technique for less invasive imaging of the lower extremity arteries. The aim of this study was to determine the concordance between CTA and catheter arteriography (CA) in patients with peripheral arterial disease (PAD). Twenty-five patients underwent both CTA and CA, and each set of images was interpreted independently by 3 readers. The infrarenal arteries were divided into 16 segments, and each segment was scored as: 1 = stenosis <50%; 2 = 50-99% stenosis; 3 = occlusion. Modal scores from 3 readers were used to compare results for each segment, with CA assumed to represent true arterial anatomy. Agreement between CTA and CA readings was defined as: concordance (modal scores were identical); moderate discrepancy (MD) (modal scores differed by 1); or severe discrepancy (SD) (modal scores differed by 2). In total, 718 segments were assessed by both CTA and CA. For all segments, the sensitivity and specificity of CTA for <50% stenosis was 86% and 90%; for 50-99% stenosis, sensitivity and specificity were 79% and 89%; and for occlusion, 85% and 98%. Above-knee (AK) CTA scores had slightly better concordance of 86.1% than below-knee (BK) readings (82.3%) (p=0.104). Severe discrepancies between AK CTA and CA scores were observed in 1.8% of segments compared to 5.4% of BK segments (p=0.038). Poor CTA image quality was the cause in 20% of AK segments and 28% of BK segments. Poor CA image quality was the cause in 8% of AK and 7% of BK discrepancies. Registration disagreement (stenosis observed in a level in 1 study attributed to a different level in the other) accounted for 18% of AK and 17% of BK discrepancies. In 54% of AK and 48% of BK discrepancies, neither image quality nor registration errors were identified, indicating that inherent differences in the depiction of stenosis by CA and CTA were responsible. When discrepancies caused by registration error were excluded, SD observed in BK segments (4.0%) remained significantly higher than in AK segments (1.25%) (p=0.029), and poor CTA quality image was the most common cause (76%) of severe BK discrepancies. In AK discrepancies without an identifiable technical cause, CTA uniformly showed more stenosis, suggesting greater CTA diagnostic precision in larger vessels. In general, agreement between CTA and CA was moderately good. Compared to CA, CTA may be better at depicting stenosis in large, proximal vessels owing to the superior accuracy of cross-sectional images in the measurement of stenosis. There appeared to be poorer CT resolution and higher frequency of severe discrepancies between CTA and CA in BK arteries.