Comparisons between Carotid Duplex Sonography and Cerebral Angiography in Assessing the Degree of Carotid Stenosis

In the article Overestimation of Carotid Stenosis on Angiography: Potential Carotid Doppler Bias in this issue of the AJNR (page 639), Dix and Skrocki compare a number of aspects of carotid stenosis quantification. The angiographic interpretation of percent stenosis on initial assessment performed between 1993 and 1998 was compared with an organized retrospective look at those same studies. The second look was done with the readers blinded to the previous angiographic and sonographic findings. The initial angiography had been performed with knowledge of the sonographic results, which formed part of the clinical information leading to the angiography. Additionally, percent stenosis estimations were assigned from duplex data retrospectively according to the criteria of Bluth (1).

Various comparisons of new and old readings of both angiography and duplex sonography form the body of results. The conclusions are interesting: if the duplex study suggested more trivial stenosis, the angiographic interpretations were reasonably concordant. If the duplex study suggested greater than 60% stenosis, however, the original angiographic interpretation assigned a stenosis degree about 16% (P < .05) greater than that assigned during the recent blinded rereading. The authors concluded this was due to some bias prompted by the knowledge of the sonographic interpretation at the time of the original angiography. The final conclusion—radiologists must use strict criteria for measurement of stenosis to avoid inherent bias.

It is difficult to criticize a study that comes to the important conclusion that one should beware and use precise criteria for quantification, especially where carotid stenosis is concerned. Nonetheless, there are possible explanations for these results in addition to those offered by Dix and Skrocki. In the years since the first North American Symptomatic Carotid Endarterectomy Trial (NASCET) publication, the ambiguities and controversies of carotid stenosis measurement have had a great deal of airing in the radiologic, neurologic, and surgical communities and literature. Much of the problem stems from the anatomic aberration of the carotid bulb, the site of so much occlusive disease and almost twice the diameter of its outflow artery, the cervical internal carotid. Prior to these discussions in the literature, the manner of stenosis calculation by various practitioners and groups received little attention. Since then, published discussions have tried to put forth evidence regarding which of numerous ways is best to measure stenosis, and how to translate accurately the results of important studies such as those of NASCET into new ways of assessing carotid disease.

It is likely that the original angiographic interpretations, upon which this study by Dix and Skrocki is partly based, were done with less consideration for the way NASCET assessed stenosis from angiograms. In other words, for the formal retrospective blinded reading, performed specifically for this current investigation, there must have been more awareness of both the specific methods used for NASCET and the debates in the literature than there would have been at the time of the first clinical reading. If this is correct, then a different set of scores closer to those from NASCET would have been expected the second time. This potential bias in the study is not addressed, and is certain to be present to some degree. The flurry about stenosis measurements in the literature, started at the time of the NASCET report in 1991, has grown over the years and continues today. Therefore, have the angiographic scores changed because the angiographers did not have the duplex results to bias them, or have the readers actually changed their own paradigms to try to emulate the measuring techniques of NASCET? Perhaps they have always ignored duplex results when performing angiography, but now the paradigms for stenosis measurement have changed.

It is not surprising that there were differences between the duplex results and the angiographic assessments for these cases. The authors list many reasons for this in their discussion. Perhaps the most important reason for the inconsistent results was the criteria used for the designation of percent stenosis by duplex sonographic parameters, the criteria of Bluth (1). In 1988, Bluth performed an extremely diligent, thorough examination and integration of different aspects of both Doppler and B-scan image parts of duplex sonography. There are multiple ways of deriving percent carotid stenosis; Bluth compared the narrowest part of the stenosis to the diameter of the adjacent bulb wall on transverse sonographic images. This method is very different from that of NASCET, and will produce very different numbers. NASCET compared the diameter of the stenosis to the diameter of the normal internal carotid well beyond the bulb, where the walls are parallel and often half the diameter of the bulb. We cannot expect the final numbers to be the same when the measurements of different structures are used in the calculation.

It is interesting that many sonographic machines have been loaded with software to calculate percent stenosis based on various important, correlative series of the 1980s that helped to establish duplex sonography as an important test for patients with this condition. Whereas there is logic to each of the ways of measuring carotid stenosis, the results of both NASCET and the Asymptomatic Carotid Atherosclerosis Study (ACAS) are attached to one particular approach to measurement. Therefore, all those dealing with carotid stenosis who wish to use those results must be aware of how their numbers were derived. It is not surprising that the percentages for many duplex carotid studies do not match NASCET or ACAS measurements, especially if the percent stenosis numbers of sonography have been derived from calculations like those in the 1988 study (1), which used completely different ratio calculations for the original derivations.

Near occlusions of the carotid are recognized by Dix and Skrocki as a case when the distal carotid artery should not be measured for stenosis calculation, because the disease has caused a decrease in its size (2). Whereas high-level, modern color duplex studies can recognize near occlusions rather well (3), reliance on Doppler velocity parameters alone can yield misleading results as a stenosis progresses in severity beyond the high peak velocities of severe stenosis. Such worsening degrees of stenosis, through a decrease in pressures and flow that occur simultaneously with the decreasing carotid diameter in near occlusion, will produce decreasing levels of peak Doppler velocity.

Topics concerning how best to determine appropriate candidates for stroke prevention treatments are extremely important and timely. This study by Dix and Skrocki addresses some of the ways to calculate percent stenosis, in order to relate to the results of clinical trials. This article adds another dimension to this topic, maintains interest in the problem of consistency of carotid stenosis measurements, and sets the scene for the continuing interest that carotid angioplasty and stenting add to reducing the risks of stroke disease.