Article Text

Download PDFPDF

Myelopathy from intracranial dural arteriovenous fistula
  1. G P Kalamangalam1,
  2. J Bhattacharya2,
  3. E Teasdale2,
  4. M Thomas1
  1. 1Department of Neurology, Institute of Neurological Sciences, Southern General Hospital, Glasgow, UK
  2. 2Department of Neuroradiology, Institute of Neurological Sciences
  1. Correspondence to:
 Dr A M Thomas, Department of Neurology, Southern General Hospital, Glasgow G51 4TF, UK;


Dural arteriovenous fistulas arising intracranially are an uncommon cause of progressive myelopathy. This report is of a patient in whom the diagnosis of the condition was confounded by coexisting small vessel cerebrovascular disease.

  • myelopathy
  • arteriovenous fistula

Statistics from

Request Permissions

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.

A 68 year old right handed man was admitted to his referring hospital in March 2001 with a three month history of deteriorating mobility, lower limb weakness, and recurrent falls. He had initially presented to his general practitioner with a sudden onset of unsteadiness, and a stroke was diagnosed. He failed to improve, and following a gradual but marked deterioration he was admitted to hospital. Over the course of a month in hospital his condition worsened, with recurrent collapses caused by a postural fall in blood pressure, increasing leg weakness and unsteadiness, loss of sphincter control, and difficulty in swallowing.

His past medical history consisted of a previous right sided stroke 10 years before, from which he had made a complete recovery. He had suffered two myocardial infarcts. He had essential hypertension, diet controlled hypercholesterolaemia, and was an ex-smoker. Drug treatment on admission to hospital was aspirin, dipyridamole, and various antihypertensive agents. The latter were discontinued when his hypotensive episodes were noticed, and fludrocortisone was added.

Magnetic resonance imaging (MRI) of the brain and cervical spine at the referring hospital suggested florid diffuse small vessel cerebrovascular disease involving the deep white matter of the hemispheres and brain stem, with extension to the upper cervical spinal cord. The radiological changes, in the context of his known significant vascular history, were initially thought sufficient to explain his clinical syndrome.

Examination on transfer to our unit revealed a cogent, articulate man. A significant postural fall in systemic blood pressure (an average difference of 50/30 mm Hg between the supine and sitting positions) was noted; general examination was otherwise unremarkable. Fundoscopy and eye movements were normal. Facial movements were full and symmetrical. His cough was weak, with a bovine character, but palatal movement appeared symmetrical and the gag reflex was intact. A formal swallowing assessment was satisfactory. The upper limbs were neurologically normal. The lower limbs revealed mild wasting with a moderate pyramidal weakness, more noticeable proximally. All deep tendon reflexes were brisk without clonus, and the plantar responses were flexor. A urinary catheter was in situ. There was no sensory disturbance. He was unable to sit unsupported because of truncal weakness.

Neuroradiological review of his original cranial MRI confirmed widespread, confluent areas of T2 signal prolongation in the hemispheric white matter (fig 1A); these were unusually extensive in the brain stem (fig 1B). The latter changes were noted to extend into the cervical cord on both sagittal and axial T2 sequences (not shown). A sagittal T1 weighted view of this region (fig 2A) revealed an irregular contour of the medulla and cervical cord, suggestive of vascular flow voids. These were confirmed on a contrast enhanced examination (fig 2B), which revealed a network of enlarged vessels enmeshing the lower brain stem and cervical cord. Angiography of the carotid and vertebral arteries was performed. This revealed a dural arteriovenous fistula of the clivus fed by clival and meningeal branches of the intracranial left internal carotid artery (fig 3). There was venous drainage into enlarged and tortuous perimedullary veins on the anterior aspect of the brain stem, and thence to the anterior and posterior perimedullary venous plexuses of the cervical spinal cord (not shown). The fine calibre of the feeding vessels precluded endovascular treatment of the arteriovenous fistula.

Figure 1

T2 weighted axial sequences through the hemispheres (A) and medulla (B) showing confluent areas of high signal change. The florid abnormality in the latter (which was continuous with similar appearances in the pons and cervical cord) is unusual for small vessel cerebrovascular disease.

Figure 2

T1 weighted sagittal sequence (A) through the posterior fossa and cervical cord. Note the irregular appearance of the anterior surface (arrow), suggesting vascular flow voids. Contrast enhanced examination (B) confirmed these; a network of enlarged vessels is seen enveloping the lower brain stem and cervical cord.

Figure 3

Cerebral angiography: left internal carotid injection. A cluster of fine calibre abnormal vessels (arrow) is seen arising from the terminal portion of the carotid in the early phase. Late phase (not shown) revealed drainage caudally from veins on the anterior brain stem into the anterior and posterior perimedullary venous plexuses.

Following neurosurgical referral, the arterialised draining vein was clipped just beyond the arterial feeders. The patient was subsequently discharged to a rehabilitation unit, where his initial progress was promising: four months after operation (October 2001) he was able to sit unaided and was starting to walk with support. The ferromagnetic character of the surgical clip has precluded postoperative MRI.


Progressive myelopathy from an intracranial arteriovenous fistula with spinal perimedullary venous drainage is a rare entity—our patient appears to be only the 15th case to be reported. Anatomically, the condition seems to arise as an exaggeration of the normal anastamoses between the pontomesencephalic and the anterior and posterior spinal veins.1 This may be provoked by the thrombotic occlusion of a draining intracranial venous sinus, although there was no history of this in our patient, nor any suggestive angiographic evidence. Thus, while a dural spinal arteriovenous fistula would usually obtain its arterial supply from that of the spinal cord itself, it is occasionally possible for the nidus to arise intracranially. The arterialised spinal veins lead to congestive oedema and subsequent ischaemia of the spinal cord, causing the symptoms.

Fourteen other patients with myelopathy caused by intracranial dural arteriovenous fistulae have been reported.2,7 There is a male preponderance (male to female ratio, 13:2), and the mean age of onset is 55.5 years. In most previous cases there was considerable difficulty and delay in establishing a definitive diagnosis before contrast enhanced MRI and cerebral angiography were undertaken. Cranial MRI was either not performed or was normal in all cases. Endovascular embolisation, either alone or followed by surgery, was the treatment used. The delay before diagnosis has perhaps skewed the prognosis of the condition—only four patients are documented as having made a recovery to independent ambulation. Two patients died: one from the series of five reported by Gobin et al,5 who died postoperatively, and one from a series of three reported by Partington et al,4 who died of unrelated causes. The remaining eight are reported to have been left disabled, though treatment presumably halted any further decline.

Patients with a suspected spinal arteriovenous malformation will usually have spinal angiography. However, as the authors of the above case series emphasise, angiographic studies must include the cranial vasculature when spinal studies are normal or if the abnormality on MRI is maximal in the upper spinal cord. Apart from reiterating this diagnostic pitfall, our main objective has been to highlight the potential for neurologists to be led astray by the common MRI appearances of “diffuse small vessel disease.” While this label was appropriate for the supratentorial imaging changes in our patient, reflecting his age, background vascular history, and risk factors, it was clearly not appropriate for the brain stem and cervical cord ischaemia. Such a radiological distinction is admittedly challenging for a clinician to make. In this case it was the atypical, and rapidly progressive, nature of the clinical picture that prompted us (GPK, AMT) to seek the expert neuroradiological assistance that finally led to his correct diagnosis.


We thank Mr Kenneth Lindsay for helpful discussion.