Magnetic resonance angiography is an accurate imaging adjunct to duplex ultrasound scan in patient selection for carotid endarterectomy

Purpose: The purpose of this study was to evaluate the accuracy of magnetic resonance angiography (MRA) for categorizing the severity of carotid disea se relative to duplex ultrasound scan and cerebral contrast arteriography ( CA) to determine if MRA imaging could replace the need for cerebral angiogr aphy in cases of indeterminate or inadequate duplex scan imaging. Methods: Seventy-four carotid bifurcations in 40 patients undergoing 45 car otid endarterectomies from 1996 to 1998 were imaged with duplex ultrasound scan; MRA (two-dimensional neck and three-dimensional intracranial, time-of -flight technique); and biplanar, digital subtraction cerebral arteriograph y. Studies were blindly reviewed by one reader who used established thresho ld velocity criteria for the duplex scan and the North American Symptomatic Carotid Endarterectomy Trial method for MRA and CA to determine the percen tage of diameter reduction of the internal carotid artery (ICA). Disease se verity was grouped into four categories (< 50%, 50%-74%, 75%-99% stenosis a nd occlusion), and the results of MRA and duplex ultrasound scan were compa red with CA. Results: Sensitivity, specificity, positive predictive value, and negative predictive value for detection of > 50% ICA stenosis were 100%, 96%, 98%, a nd 100% for MRA and 100%, 72%, 88%, and 100% for duplex ultrasound scan, re spectively; similarly, for detection of > 75% ICA stenosis values were 100% , 77%, 76%, and 100% for MRA and 90%, 74%, 72%, and 91% for duplex ultrasou nd scan, respectively. Both MRA and duplex ultrasound scan accurately diffe rentiated all cases of > 95% stenosis (n = 7) from occlusion (n = 4). Short length ICA flow gaps were present on MRA in all cases of 75% to 99% stenos is and one half of cases of CA-defined 50% to 74% stenosis. In patients wit h 50% to 74% stenosis, the mean angiographic stenosis was significantly gre ater when a flow gap was present on MRA (64% +/- 6%) versus no flow gap (57 % +/- 7%) (P = .04). There was overall agreement among duplex ultrasound sc an, MRA, and CA in 73% of carotids imaged. Of the 24% discordant results be tween MRA and duplex ultrasound scan, MRA correctly predicted disease sever ity in all cases, and inaccurate duplex ultrasound scan results were due to overestimation in 83% of cases. The operative plan was altered by CA findi ngs in only one patient (2%) after duplex ultrasound scan and MRA. Conclusions: MRA can accurately categorize the severity of carotid occlusiv e disease. Duplex ultrasound scan facilitates patient selection for carotid endarterectomy in most cases, but adjunct use of MRA improves diagnostic a ccuracy for > 75% stenoses and may obviate the need for cerebral arteriogra phy when duplex scan results are inconclusive or demonstrate borderline dis ease severity.