Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Since the advent of advanced radiological modalities such as MRI and magnetic resonance angiography (MRA), dissections of cervical arteries are increasingly recognised as a common cause of stroke in young adults. Auer et al 1recently advocated MRA as the initial diagnostic tool for vertebral artery dissection. Conventional angiography might be avoided altogether in subjects with a suspicious history and MRA images suggestive of a dissection (double lumen or mural haematoma).1 The sensitivity of MRA for the diagnosis of vertebral artery dissection was only 20% in one study, but the specificity was excellent (100%).2 The sensitivity was considerably better in the hands of Auer et al,1 but in this study the specificity (true negative rate in subjects free of disease) was not considered because all patients had vertebral artery dissection. The following case report illustrates that care must be taken to avoid false positive results when using MRA for the diagnosis of vertebral artery dissection.
A 47 year old male pilot suddenly experienced clumsiness and slight loss of strength in the right arm and leg during a long distance flight, while he stooped forward. During the following hours, he developed a global headache without irradiation to the neck, but the other symptoms gradually diminished. Prior history was unremarkable, except for a 3 hour period of horizontal diplopia which suddenly developed 3 months earlier. He had never smoked. Family history was negative for cardiovascular disorders. The patient later confessed that he had recently picked up the habit of gargling his throat with toothpaste twice a day, always with his neck in extreme retroflexion.
General physical examination (8 hours after onset of symptoms) was normal. Neurological examination showed minimal paresis and impaired dexterity of the right hand, mild circumduction of the right leg, and an insecure tandem gait. An MRI (including T1 weighted spin echo images with and without fat suppression, and proton density and T2 weighted fast spin echo sequences, performed on a 1.5 Tesla whole body MRI system) performed several hours later visualised both a fresh and an old right sided cerebellar infarct (figure A). In addition, MRI showed an irregular right vertebral artery in which a patent lumen was partially surrounded by a semilunar area of high signal intensity on T1 and T2 weighted images. On fat suppressed images, this area’s high signal intensity persisted, excluding the possibility that it originated from perivascular fat. This image was suggestive of mural haematoma due to vertebral dissection (figure B). Because we were reluctant to base any treatment decisions (anticoagulants) merely on MRI findings, digital subtraction angiography was performed on the day of admission. This examination was normal (figure C). Shortly after this procedure, the patient developed vertigo and nystagmus which disappeared after 3 hours. Because we were puzzled by the discrepant findings on conventional angiography and MRI, we performed an MRA 4 days later. At this examination, the semilunar area of high signal intensity was found again (figure D), despite saturation of craniofugal and craniopetal flow respectively, which was applied to exclude the possibility that the high signal originated from flow in the periarterial venous plexus. Therefore, this examination was again suggestive of right vertebral artery dissection. An extensive search for other causes of stroke showed no abnormalities. Hence, due to the continuing discrepancy between conventional angiography and MRI/MRA, and due to the absence of any other cause of stroke, no certain diagnosis could be established.
In this patient, a diagnosis of right vertebral artery dissection was initially made given the clinical course with repeated episodes of ischaemia restricted to the vertebrobasilar system, as well as the suggestive MRI findings.1 We speculated that habitual gargling was a potential underlying cause, as neck retroflexion can cause cervical dissections. However, we had to reject this diagnosis in view of the normal conventional angiography, which remains the gold standard for diagnosing cervical artery dissection.3 In one series,2 conventional angiography was never falsely negative in patients with clinical signs or symptoms of vertebral artery dissection. The possibility that conventional angiography had nevertheless yielded a false negative result seems highly unlikely. In dissected arteries, MRI/MRA can detect intimal flaps, mural haematomas, or aneurysmal dilatations that are sometimes missed by conventional angiography, but even in such patients conventional angiography is never completely normal in the acute stage. Follow up examinations of patients with proven vertebral artery dissection indicate that the appearance of a dissected artery on conventional angiography can normalise in a substantial proportion of patients, but always after an interval of at least 1 to (usually) several weeks.1Conventional angiography in our patient was performed on the day of admission, directly after the “abnormal“ MRI and four days prior to the “abnormal“ MRA, hence spontaneous resolution of the dissection is very unlikely. Therefore, we consider our MRI/MRA examinations falsely positive, and we hypothesise that the area of semilunar high signal intensity originated from a perivascular venous plexus, in which we were unable to saturate inflow of blood completely, presumably due to extremely slow flow.
Our “pilot study“ illustrates the specificity problems of MRI/MRA for the diagnosis of vertebral artery dissection. Two anatomical structures surrounding vertebral arteries contribute to these problems. The first structure is the venous plexus that surrounds vertebral arteries. This structure may have a semilunar appearance, and slow flow in its lumen may give rise to high signal intensity on both MRI and MRA, creating an image suggestive of dissection.4 5 It has been suggested that saturation slabs in conjunction with MRA completely suppress flow related high signal, thus distinguishing it from high signal from an intramural haematoma which cannot be suppressed by saturation slabs.4 5The present case report illustrates that flow in this plexus cannot always be suppressed.
The second tissue that may falsely present as a dissection is fat that directly surrounds vertebral arteries. This fat also gives rise to high signal intensity, but using fat suppression techniques it can be readily differentiated from intramural haematoma. Furthermore, the usual diameter asymmetry of vertebral arteries, turbulence and magnetic susceptibility near sharp vessel turns can also cause false positive MRA results.2 In some patients, MRI cannot distinguish between intraluminal thrombus and intramural haematoma, leading to false conclusions.
Decisions based on false positive MRI/MRA results can be hazardous due to the sometimes severe side effects of anticoagulants, the treatment that is recommended by some to prevent further ischaemic events. Another danger of a false positive diagnosis of vertebral dissection is that it may preclude the search for other causes of stroke that could be amenable to secondary prevention.
MRI/MRA remains important because it helps visualise ischaemic lesions and, in some patients, provides complementary morphological information to cerebral angiography.1 Furthermore, it is a non-invasive procedure, an important advantage over cerebral angiography which carries a morbidity and mortality risk. Our patient, who developed transient neurological deficits shortly after angiography, underscores this. Therefore, MRA can play a part in the diagnosis of vertebral artery dissection, provided that the pitfalls mentioned above are recognised to avoid false positive results. In case of doubt, cerebral angiography remains the gold standard for vertebral artery dissection.