MR appearance of an intracranial dural arteriovenous fistula leading to cervical myelopathy

Intracranial dural arteriovenous fistula (DAVF) is a rare cause of cervical myelopathy. MR signal abnormalities of the cervical spinal cord without parenchymal cord enhancement in conjunction with enlarged perimedullary vessels are highly suggestive of such a lesion. We report the MR imaging and angiography findings as well as the clinical and radiologic time course of an intracranial DAVF draining into spinal medullary veins.

Intracranial DAFVs account for approximately 10% of all intracranial arteriovenous malformations.¹ They result from an abnormal communication between meningeal arteries and veins or venous sinuses.² Common symptoms of intracranial DAVFs are bruit, dizziness, and headache¹; other manifestations are intracranial hemorrhage, cerebral ischemia, and cranial neuropathies.² Cervical myelopathy resulting from intracranial DAVFs is extremely uncommon. We describe the MRI, myelographic, and angiographic findings as well as the clinical and radiologic time course of an intracranial DAVF draining via the superior petrous sinus into spinal medullary veins.

Case report. A 67-year-old man was admitted with subacute progressive quadriparesis (table, Case 13). T2- and proton density (PD)-weighted MRI revealed swelling of the cervical spinal cord, high intramedullary signal extending up into the medulla oblongata, and a linear, hypointense structure at the ventral surface of the cervical spinal cord, which was suspected of being a flow void representing an enlarged vessel (figure 1, A and B). Contrast-enhanced T1-weighted MRI (see figure 1C) showed linear enhancing structures at the ventral surface of the cervical cord suggestive of enlarged veins, which was verified by contrast-enhanced MR angiography (MRA; see figure 1D). No parenchymal contrast enhancement of the spinal cord was noted on T1-weighted MRI. Myelography and postmyelographic Ct showed enlargement of the cord. Intraarterial digital substraction angiography (DSA) of the right common carotid artery, the right innominate artery, and the right and left vertebral arteries was unremarkable. DSA of the left common carotid artery, however, showed an arteriovenous fistula fed by a branch of the pharyngeal ascending artery and by another branch of the occipital artery (see figure 1E); the lesion drained via the superior petrous sinus, the petrous vein, and the anterior pontomesencephalic vein into the anterior and posterior spinal veins (see figure 1F). After superselective embolization of these feeders using a solution of Natriumamidotrizoate/Zein (Ethibloc, Ethicon, Hamburg, Germany), additional feeders from the distal portion of the occipital artery and from the middle meningeal artery became visible. These feeders were embolized as well, resulting in complete occlusion of the fistula (figure 2A); enlarged or early filling spinal medullary veins no longer opacified on injection of contrast material. At follow-up DSA 7 weeks after embolization, the fistula was still completely occluded. A repeat MR study 10 weeks after embolization showed marked reduction of the signal abnormality in the cervical cord with only a minimal persisting hyperintensity at the medullocervical junction without cord swelling (see figure 2B). Enlarged vessels at the ventral surface of the cord were absent. At this time the patient had clinically improved considerably (see table).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Sagittal fast spin-echo T2-weighted MRI (repetition time [TR], 4000; echo time [TE], 96; 2 excitations) (A) and sagittal fast spin-echo proton density (PD)-weighted MRI (TR, 4000; TE, 48; 2 excitations) (B) show abnormal high signal extending from the C-3 level up to the medulla oblongata, swelling of the cervical spinal cord, and a linear, hypointense structure at the ventral surface of the cervical spinal cord, which aroused suspicion of flow void representing an enlarged vessel (arrows). (C) Sagittal spin-echo T1-weighted MRI (TR, 650; TE, 20; 1 excitation) after application of contrast material shows linear enhancing structures at the ventral surface of the cervical cord suggestive of representing enlarged veins (arrows). (D) Midsagittal maximum intensity projection image after postcontrast MR angiography (three-dimensional fast low-angle shot; TR, 6; TE, 2.1; 1 excitation; slice thickness 2 mm) shows an enlarged vessel ventral to the brainstem and the cervical cord (small arrows), basilar artery (arrowheads), and straight sinus (large arrows). Overlying vessels outside the midline were made invisible by segmentation. (E and F) Digital subtraction angiography. (E) Lateral view of the left common carotid artery angiogram shows an arteriovenous fistula (large arrow) near the mastoid condyle supplied by an enlarged pharyngeal ascending artery (small arrows) and a feeder from the proximal occipital artery (arrowheads). (F) Angiogram after selective injection into the pharyngeal ascending artery shows the arterial feeder (arrows) and the drainage pathway via superior petrous sinus (arrowheads) and the petrous vein into anterior and posterior spinal veins.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. (A) Postembolization angiogram of the common carotid artery shows complete occlusion of the fistula. (B) Sagittal fast spin-echo T2-weighted MRI 10 weeks after embolization shows only a small persisting hyperintensity at the medullocervical junction. No cord swelling is present.

Discussion. The most common causes of spinal cord signal abnormalities are tumor, myelitis, and trauma. Although intracranial DAVFs draining into spinal medullary veins are extremely uncommon, knowledge about their MR features is important. Patients with symptoms and signs of cervical myelopathy are typically examined first by MRI, and an improper diagnosis might result in delayed or incorrect treatment. Common MR findings in intracranial DAVFs include dilated cerebral cortical veins without detection of a parenchymal nidus, venous infarction, parenchymal and subdural hematomas, and cerebral vasogenic edema. The site of the nidus is often identifiable only on cerebral angiograms.³ As reported by Partington et al.,⁴ even sacral DAVM may cause severe myelopathy. By perusing the pertinent literature, we found that MRI of the cervical cord in patients with cervical myelopathy from intracranial DAVFs was performed in 12 cases (table); a detailed description of the MR features was given in 11 of 12 cases. MR findings in intracranial DAVFs draining into spinal medullary veins are similar to those of spinal DAVFs, including signal abnormalities of the spinal cord, enlarged perimedullary vessels, and cord enlargement.⁵ In spinal DAVFs, we had never found signal abnormalities extending from the cervical cord into the medulla oblongata, as had Ernst et al.² We thus believe that high signal on T2-weighted images in the cervical cord with extension into the medulla is more likely due to an intracranial than a spinal DAVF. Myelopathy from DAVFs is related to venous hypertension, edema, or hypoxia from a reduced arteriovenous pressure gradient.⁶ Bowen et al.⁵ found that contrast-enhanced MRA is superior to spin-echo MRI in detecting abnormal intradural vessels associated with spinal DAVFs. Although MRA was found to be superior to DSA in delineation of the anterior median spinal vein, the sensitivity of DSA in demonstrating spinal arteries and arterial feeders of DAVFs is higher than that of MRA.⁵ Enhancement of the enlarged veins or venous plexus after contrast administration on T1-weighted spin-echo images is a common finding in spinal DAVFs^5,7 and was also observed in our case and in Case 10 (see table). Terwey et al.⁷ (5/5 cases) and Bowen et al.⁵ (8/8 cases) found mild to marked parenchymal contrast enhancement of the spinal cord on contrast-enhanced T1-weighted images in spinal DAVFs. Slow accumulation of contrast material within enlarged intramedullary veins or disruption of the blood-brain barrier following chronic ischemia are discussed as probable causes for this kind of enhancement. Ernst et al.² found marked parenchymal contrast enhancement in a 4-year follow-up MR study of one case (Case 12) and suggested that variable parenchymal enhancement would be expected in intracranial DAVFs draining into spinal medullary veins. In the current case, however, parenchymal enhancement was absent, as was true in all other recorded DAVFs where contrast material was given at the initial MR examination (Cases 4 and 10). Feeders to intracranial DAVFs draining into spinal medullary veins most often originate from the occipital artery (Cases 1 to 3, 5 to 8, and 13), from meningohypophyseal or tentorial branches of the internal carotid artery (Cases 1, 2, 7, and 10), or from the pharyngeal ascending artery (Cases 3, 5, 6, 11, and 13). Common surgical strategies are coagulation or clipping of the petrous vein (Cases 1, 2, and 7), surgical obliteration of the arterial feeders (Case 9) or of the fistula (Case 3), or removal of the involved dural sinus (Case 8). Usually, after coagulation or clipping of the draining vein, this vessel is additionally divided as it enters the subarachnoid space. Endovascular embolization, however, is an alternative or a supplement to open surgical intervention (Cases 5, 6, 8, and 11 to 13). In our case, complete occlusion of the arterial feeders was performed by superselective endovascular embolization. As previously reported, follow-up MR after treatment may show persistent signal abnormalities or swelling of the spinal cord, but as in our case, none of the treated patients showed enlarged vessels.

We report the radiologic course of a patient with an intracranial DAVF fed by four branches of the external carotid artery and draining into spinal medullary veins. As in previously published cases, our MR findings included cervical cord swelling, hyperintensity of the spinal cord on T2- and PD-weighted MRI, a tubular signal void at the ventral surface of the cervical spinal cord on T2- and PD-weighted MRI, and an enlarged vessel at the ventral surface on contrast-enhanced MRA. To avoid an improper diagnosis or delayed or incorrect treatment, we believe that cerebral intraarterial angiography is indicated in cases of otherwise unexplainable acute or subacute myelopathy with absent parenchymal cord enhancement, especially if signal abnormalities extend from the cervical cord into the medulla oblongata or if the spinal angiogram is normal. Cerebral intraarterial angiography provides the option of complete or partial endovascular treatment of these fistulas.