Should modeling methodology suppress anatomic excellence?

See related article, pages 3511–3517. Wardlaw and colleagues1 use modeling to validate noninvasive imaging instead of intra-arterial angiography (IAA) for carotid stenosis. They conclude that duplex ultrasound (DUS) is best. There are not sufficient data with CT angiography (CTA) and MR angiography. Yet carotid CTA gives excellent anatomic depiction, has no stroke risk, and takes a few seconds more than simple CT.2–4 It adds to CT for emergency cases in acute stroke protocols and is done 24 hours a day as a regular CT service. CTA already replaced much of diagnostic IAA. Contrast CT is a common CT technique and CTA is merely scanning during contrast infusion. Its anatomic exactness for stenosis quantification is accepted as anatomically correct,2–12 yet published data are insufficient for modeling needed for Wardlaw and colleagues.1 IAA had no scientific verification of arterial anatomic depiction; instead, there was excitement and acceptance that arteries look like arteries. CTA now gives anatomically correct images without stroke risk. Sites expert in CTA face ethical dilemmas against continuing stroke-risk IAA to quantify stenosis for clinical or research purposes. CTA includes inherent millimeter measurements, not part of standard IAA digital subtraction angiography measured in pixels.13 Because CTA studies were not adequate to plug into Wardlaw and colleagues’ models, CTA was only considered in a limited way for Wardlaw and colleagues’ model. In the 1990s, some stopped doing IAA for carotid stenosis before surgery, using DUS alone.14–17 Inferior anatomic depiction was deemed acceptable by removing IAA stroke risks, thereby enhancing surgical results. DUS is not anatomic, but physiological, measuring velocity within carotid lumen.17–19 There is operator variability of DUS probe position. External carotid artery can be misidentified as internal carotid artery (ICA) with common anatomic transposition of ICA medial to external carotid artery if pulsation patterns are distorted by stenoses of both external carotid artery and ICA. The latter is well known to angiographers but not always to those who perform DUS, because external carotid artery and ICA are only truly indentified by images showing where they go, beyond the limited DUS neck view. Although DUS reports give ranges of percent stenosis, anatomic diameters are not measured for percent calculation. In fact, distal ICA with parallel walls as the denominator for percent stenosis calculation20 is beyond the DUS neck window view. DUS machines have built-in tables for percent stenosis ranges. In-house calibrations were recommended for DUS against one’s own local angiography.15,17,21 Published percent stenosis tables for DUS are available, originally measured from IAA for local reference, yet not always derived from IAA using the North American Surgical Carotid Endarterectomy Trial (NASCET) percent stenosis method.20 There is the pitfall of decreased distal ICA lumen (near occlusion),10,20,22 fallacious to calculate as percent stenosis with NASCET methods. There is also wide variability of measuring the distal normal ICA for percent stenosis calculation, not always with compliance as “well beyond the bulb where walls are parallel.”20 It is difficult to know how some did percent stenosis to match with DUS. Self-calibrated IAA to DUS dissipated over time, because many dropped digital subtraction angiography in favor of DUS. Today most DUS laboratories have percent stenosis correlations from someone else with variations of expressed percent. There are multiple DUS publications that fit modeling criteria. Anatomic inferiority of DUS meant that percent stenosis ranges are needed to show consistency. DUS differs from anatomic stenosis quantified for IAA, and DUS saw many studies to show stenosis degree ranges. These studies form the bulk of Wardlaw and colleagues’ model 1 with the conclusion to dismiss IAAs (and CTAs) anatomic exactness in favor of DUSs less accurate stenosis quantification with no stroke risk. This result would have been perfect for the mid-1990s when many abandoned IAA for DUS. Yet currently, with CTA’s exacting anatomy, no stroke risk, and taking seconds more than simple CT, it is passé to accept inferior anatomic depiction. The conclusion to omit anatomic exactness for no stroke risk is obsolete with CTA substituting for IAA.23 Conclusions from the modeling results1 omit that CTA has already replaced much IAA in many locales. Once CTA’s capability is understood, ethical problems intercede against continued stroke-risk IAA for pure research purposes. The “Catch 22” is that, for Wardlaw and colleagues, there are not studies of CTA good enough for their modeling method, because CTA’s anatomic exactness substitutes for IAA. Anatomic exactness is not a problem for The opinions in this editorial are not necessarily those of the editors or of the American Heart Association. The opinions expressed in this editorial are not necessarily those of the editors or of the American Heart Association. From the Sunnybrook Health Sciences Centre (A.J.F., S.P.S., R.I.A., P.H., R.Y.), University of Toronto, Toronto, Ontario, Canada; and the Princess Margaret Hospital (E.S.B.), University of Toronto, Toronto, Ontario, Canada. Correspondence to Allan J. Fox, MD, Professor, Medical Imaging, University of Toronto, Neuroradiologist, Sunnybrook Heath Sciences Centre, 2075 Bayview Avenue, Room AG31b, Toronto, Ontario M4N 3M5, Canada. E-mail [email protected]; [email protected] (Stroke. 2009;40:3411-3412.) © 2009 American Heart Association, Inc.