The trusted source for
healthcare information and
Abstract & Commentary
Source: Okahara M, et al. Diagnostic accuracy of magnetic resonance angiography for cerebral aneurysms in correlation with 3D-digital subtraction angiographic images: A study of 133 aneurysms. Stroke. 2002;33:1803-1808.
Digital subtraction angiography (DSA) is considered the gold standard in the diagnosis of cerebral aneurysms. It is invasive, however, and has potential risks, particularly in elderly patients. Noninvasive options for aneurysm detection include MR angiography (MRA) and CT angiography (CTA). Given its widespread availability, MRA has become a popular screening tool for aneurysms, and is believed to be at least 90% sensitive for these lesions. On the contrary, these data from Okahara and colleagues suggest that unless an MRA is interpreted by an experienced reader, its accuracy may be significantly lower.
Okahara et al studied 82 patients with 133 cerebral aneurysms, 36 with subarachnoid hemorrhage (SAH), and 64 without SAH. All patients underwent DSA and MRA. DSA was enhanced using 3D rotational reconstructions generated by a computer workstation. This significantly increases sensitivity and specificity over that of standard DSA. MRA was analyzed using 3D time of flight (TOF) images. Source MRA data were not examined and gadolinium contrast was not given. MRAs were examined by 4 readers: a senior neuroradiologist (NR), an experienced neurosurgeon (NS), a general radiologist (GR), and a radiology resident (RR). Aneurysm detection overall was: NR 79%, NS 75%, GR 63%, and RR 60%. Detection rates for aneurysms < 3 mm ranged from 38-55%, significantly lower than for larger > 3 mm aneurysms, which were detected 68-89% of the time. Only 60% of the 10 ruptured aneurysms < 3 mm were detected by the 4 readers.
Rates of missed diagnosis on MRA were highest in patients with more than a single aneurysm. Lesions at the internal carotid artery (ICA) or anterior cerebral artery (ACA) were most difficult to visualize, producing detection rates as low as 14%. A significant number of false-positive aneurysms were diagnosed by MRA: 29 by NR, 18 by NS, 31 by GR, and 30 by RR. Diagnostic accuracy was generally lower for MRA in the setting of SAH than in asymptomatic patients, but this did not reach statistical significance.
The majority of aneurysms missed by MRA were simply overlooked. Others were not seen due to technical factors such as: inhomogeneous signal intensity from atherosclerotic changes or turbulent flow, small aneurysm size, overlap of artery and aneurysm, or proximity to an artery branch point. MRA readers were blinded to the clinical history. With such history provided, only the RR significantly improved, at least in part due to a further accrual of experience in interpretation.
In the setting of SAH, there is no substitute for DSA in any standard algorithm for the diagnosis and treatment of critically ill patients. For asymptomatic patients being screened for aneurysms, however, MRA may present a viable alternative. Patients may undergo MRA due to a positive family history of SAH, endocarditis with risk of a mycotic lesion, or perhaps in the setting of non-specific headaches. Because pretest probabilities may vary widely among these scenarios, MRA may be appropriate for some, but not all, instances.
The poor performance of MRA in the hands of Okahara et al may be due to a number of factors. They did not use gadolinium and used 3D Maximal Image Projection (MIP) images rather than source data, compromising sensitivity. This methodology, however, actually strengthens their study and widens its applicability. Gadolinium is not commonly used for MRA in most centers and many MRA readers in the community use only MIPs, without access to source images. Okahara et al compared MRA to 3D rather than standard DSA. This may have further handicapped MRA. 3D-DSA affords a significant advantage over standard DSA, which would otherwise be considered the gold standard. MRA may, therefore, have fared a bit better if compared against merely ordinary DSA.
CTA represents a further noninvasive option and offers accuracy roughly equal to MRA (White PM, et al. Radiology. 2001;219:739-749). CTA may be considered as a viable alternative to MRA for claustrophobic or agitated patients, but it does require a significant contrast load, which must be properly infused, and is prone to similar interpretive pitfalls. Choosing a screening method is, therefore, a decision that must be individualized based on the patient, the quality of the diagnostic study available, and the skill of the radiologist.
Finally, it is important to consider whether the detection of aneurysms measuring < 3 mm bears any true importance. Some authors argue that such aneurysms have an exceedingly low rupture rate, in the range of 0.01% (1/10,000), and therefore are acceptable to overlook. Others, however, quote rupture rates as high as 0.7% and emphasize that a significant number of SAHs (including 10/35 in this study) result from aneurysms in this tiny size range. When one moves beyond issues of diagnosis and onto the question of treatment for these aneurysms, this controversy becomes even more heated. —Alan Z. Segal
Dr. Segal, Assistant Professor, Department of Neurology, Weill-Cornell Medical College, Attending Neurologist, New York Presbyterian Hospital, is Assistant Editor of Neurology Alert.
Readers are Invited. . .
Readers are invited to submit questions or comments on material seen in or relevant to Neurology Alert. Send your questions to: Neill Larmore, Neurology Alert, c/o American Health Consultants, P.O. Box 740059, Atlanta, GA 30374. For subscription information, you can reach the editors and customer service personnel for Neurology Alert via the internet by sending e-mail to firstname.lastname@example.org. We look forward to hearing from you.