Statistics from Altmetric.com
In the paper by Wardlaw et al (this issue, pp155–160) the authors consider the question as to whether or not the severity of carotid bifurcation stenosis has some bearing on reader accuracy when interpreting magnetic resonance angiography.1 The answer is yes, assessment becoming less accurate with increase in severity of the stenosis. If we accept that there is an increased risk of complication from conventional angiography in patients with more severe stenoses then the importance of accurate non-invasive diagnosis to triage patients before carotid endarterectomy is realised.
The fact that there are so many different ways of performing magnetic resonance angiography2 implies that there is no one single best technique. As the authors point out, even the widely held “gold standard” of intra-arterial angiography is flawed.
Despite all the above variables, the attempt of Wardlawet al to report interobserver variability in the detection of signal gap and % stenosis in a Doppler prescreened symptomatic population, raises some interesting questions and recommendations for radiologists. Quality assurance issues are raised in this paper in addition to stressing the importance of familiarity with variations in technology, leading to a previously published recommendation to calibrate different systems against conventional angiography.3 This paper hints at the importance of knowing local between reader variability and suggests that consensus opinion may be helpful when assessing the degree of carotid stenosis; having first decided which method—CCA (preferred by the authors), NASCET, or ECST—to use when interpreting magnetic resonance angiography. There are definite recommendations for the more widespread availability of picture archival systems (PACS), in the United Kingdom, for the electronic transmission of magnetic resonance angiography images which constitute large data sets.
The paper confirms an association, previously described in the literature, between signal drop out and a greater degree of stenosis,3 most likely on the basis of turbulent flow—with the caveat that signal drop out may also be seen with tortuous vessels alone. Taken in association with distal flow signal in the cervical internal carotid artery signal drop out may be used to help distinguish stenosis from occlusion.
The authors again show that magnetic resonance angiography consistently overestimates the percentage stenosis and that further studies still need to be done to see if magnetic resonce angiography is more accurate, if more expensive, than say a second ultrasound examination, in identifying those patients with an 80%–99% stenosis who might most benefit from carotid endarterectomy.4