Duplex ultrasound and magnetic resonance angiographic (MRA) studies ar
e the principal noninvasive methods for evaluation of extracranial occ
lusive disease in patients at risk for stroke, but each has limited ab
ility to diagnose aortic arch and arch vessel disease. Recent favorabl
e reports of the nonnephrotoxic contrast agent Gadolinium (Gd) being u
sed to enhance MRA images of the abdominal aorta prompted us to examin
e its utility for the aortic arch vessels. Prospectively, 28 patients
'uvith suspected carotid or arch vessel disease were imaged by contras
t arteriographic examination and MRA+Gd of the aortic arch within 30 d
ays of each other. One (for contrast arteriograms) or two (for MRA) bl
inded readers measured stenoses with the contrast arteriograms as the
standard. A total of 196 arch vessels containing 58 stenoses and four
occlusions (by arteriogram) were examined with each method. Interobser
ver agreement for interpretation of MRA studies was substantial (kappa
=0.68). MRA detected all anatomic anomalies (e.g., bovine arch). The c
orrelation of MRA with arteriographic scans for arch vessel stenoses >
50% was sensitivity, 73% (readers 1 and 2); specificity, 98% (reader 1
), 89% (reader 2); positive predictive value, 73% (reader 1), 89% (rea
der 2); negative predictive value, 98% (readers 1 and 2); accuracy, 97
% (reader 1), 98% (reader 2). MRA+Gd is an accurate new noninvasive im
aging method for detection of significant aortic arch disease. In its
current state of development, however, it cannot obviate the need for
contrast arteriographic examination.