Dural Arteriovenous Fistulas: Epidemiology and Clinical Presentation

Intracranial dural arteriovenous fistulas (DAVFs) are relatively rare lesions consisting of anomalous connections between dural arteries and venous sinuses and/or cortical veins. Their clinical presentation is quite variable, with symptoms dependent on their location and venous drainage pattern. Lesions with cortical venous drainage, however, have the highest risk of causing the most significant morbidity and mortality. This places an emphasis on promptly suspecting and diagnosing these lesions. This review highlights the etiology, epidemiology, clinical presentation, and clinical course of patients with intracranial DAVFs.

Dural arteriovenous fistulas (DAVFs) are anomalous connections between dural arteries and venous sinuses and/or cortical veins. These relatively rare arteriovenous shunts can occur within the spine or intracranially. Intracranial DAVFs can cause ischemic deficits and/or hemorrhage, which can lead to significant morbidity and mortality. These potentially severe consequences that are associated with a subset of these lesions underscore the need for understanding the natural history, prompt diagnosis, and treatment of these lesions to improve patient outcomes. This review aims to characterize the etiology, epidemiology, clinical presentation, diagnostic evaluation, and clinical course of patients with intracranial DAVFs.

Historical perspective

The concept of a DAVF has evolved over more than a century. In 1873, Rizzoli described the presence of a cranial DAVF. He described the presence of a dural-based “arteriovenous aneurysm” that passed through the wall of the skull in a 9-year-old girl with symptoms of seizures and pulsatile occipital swelling. The girl’s postmortem examination revealed direct communication between the hypertrophic branches of the occipital artery and the transverse sinus. In 1931, Sachs and Tonnis were the first to describe the angiographic appearance of a DAVF ; they described the presence of direct connections between the meningeal arteries and the venous system within the dura.

These lesions were considered congenital, benign lesions until the 1970s. In the late 1970s, Castaigne and Djindjian proposed an acquired etiology, suggesting that DAVFs develop from the opening of microshunts and/or angiogenesis within the dura between meningeal arteries and veins. The putative benign nature of these lesions was challenged principally by Cognard and colleagues, who proposed that the nature of these lesions depended on the degree of cortical venous drainage. Since these early studies there have been numerous studies describing the angiographic and magnetic resonance imaging (MRI) features, clinical characteristics, and clinical course of DAVFs. The literature contains reports using dural arteriovenous malformation (AVM) as the terminology for these lesions, to distinguish them from pial AVMs. Here, the authors use the term DAVF rather than dural AVM or DAVM, as they have a distinct pathophysiology from parenchymal AVMs.

Epidemiology

DAVFs present at a mean age of 50 to 60 years, but individual presentation is highly heterogeneous. In the past, DAVFs were relatively uncommon, but are now being diagnosed with increased frequency. This increased frequency of incidentally diagnosed DAVFs has been attributed to the wide availability of MRI. These lesions account for approximately 10% to 15% of intracranial vascular malformations. Among supratentorial and infratentorial vascular malformations, they account for 6% and 35% of lesions, respectively. These lesions seem to have no gender preponderance, but several studies have reported an increased incidence of hemorrhage in men in comparison with women. No linkages to family history or genetics have been identified.

Etiology

The cause of DAVFs is not always clear. Most of these lesions are believed to initiate from thrombosis of a dural venous sinus. This occlusion causes venous congestion and subsequent venous hypertension. Over time, this increased venous pressure dilates small capillaries, which open direct shunts between dural arteries and veins. This creates DAVFs. These fistulas will initially drain into larger venous sinuses. However, with increased venous pressure, the veins will undergo remodeling with hyaline deposition and intimal proliferation. This remodeling will cause blood to reflux into the cortical veins instead of solely flowing into the venous sinuses. With progressive remodeling, drainage into the venous sinuses will be completely obstructed and will rely on cortical venous reflux for drainage.

Increased cortical venous reflux in a retrograde direction will lead to cortical venous hypertension and subsequent cortical venous remodeling. This process can result in intracranial hemorrhage (ICH) or parenchymal ischemia. ICH is thought to occur from rupture of fragile parenchymal veins. These veins become fragile because of the increased pressure from retrograde venous reflux. Parenchymal ischemia is thought to occur from venous congestion. This congestion and venous hypertension prevents adequate arterial delivery of oxygen and removal of metabolic byproducts within the surrounding parenchyma. In exceedingly rare circumstances, venous hypertension can cause venous stasis within the DAVF. This stasis can lead to secondary venous thrombosis, which can spontaneously obliterate the draining vein and resolve the DAVF.

Several antecedent events have been cited as causing the development of DAVFs. Besides incidentally, the most commonly referenced preceding event is head trauma, which can occur with or without skull fractures. Other preceding events that have been reported include craniotomy, acupuncture, cerebral infarction, hormonal alterations observed in pregnancy and menopause, increased systemic thrombotic activity, otitis, sinusitis, and tumors, especially meningiomas. Regardless of the actual antecedent event, decreased flow within a dural venous sinus or vein seems to play a critical role in the development of DAVFs.

Carotid-cavernous fistulas (CCFs) are typically considered distinct from most other intracranial DAVFs, and are discussed by Pradilla and colleagues elsewhere in this issue. CCFs are typically divided into direct or indirect fistulas. Direct or high-flow fistulas are caused by either trauma (blunt or penetrating) or rupture of an intracavernous carotid artery aneurysm. These events create a direct communication between the cavernous carotid artery and the cavernous sinus. Indirect or low-flow fistulas typically occur spontaneously or by occlusion of the cavernous sinus. These events create a connection typically between branches of the external carotid artery and the cavernous sinus via dural arteries and veins. This creates an indirect connection between the carotid artery and the cavernous sinus. Indirect CCFs behave similarly to the intracranial DAVFs discussed here.

Etiology

The cause of DAVFs is not always clear. Most of these lesions are believed to initiate from thrombosis of a dural venous sinus. This occlusion causes venous congestion and subsequent venous hypertension. Over time, this increased venous pressure dilates small capillaries, which open direct shunts between dural arteries and veins. This creates DAVFs. These fistulas will initially drain into larger venous sinuses. However, with increased venous pressure, the veins will undergo remodeling with hyaline deposition and intimal proliferation. This remodeling will cause blood to reflux into the cortical veins instead of solely flowing into the venous sinuses. With progressive remodeling, drainage into the venous sinuses will be completely obstructed and will rely on cortical venous reflux for drainage.

Increased cortical venous reflux in a retrograde direction will lead to cortical venous hypertension and subsequent cortical venous remodeling. This process can result in intracranial hemorrhage (ICH) or parenchymal ischemia. ICH is thought to occur from rupture of fragile parenchymal veins. These veins become fragile because of the increased pressure from retrograde venous reflux. Parenchymal ischemia is thought to occur from venous congestion. This congestion and venous hypertension prevents adequate arterial delivery of oxygen and removal of metabolic byproducts within the surrounding parenchyma. In exceedingly rare circumstances, venous hypertension can cause venous stasis within the DAVF. This stasis can lead to secondary venous thrombosis, which can spontaneously obliterate the draining vein and resolve the DAVF.

Several antecedent events have been cited as causing the development of DAVFs. Besides incidentally, the most commonly referenced preceding event is head trauma, which can occur with or without skull fractures. Other preceding events that have been reported include craniotomy, acupuncture, cerebral infarction, hormonal alterations observed in pregnancy and menopause, increased systemic thrombotic activity, otitis, sinusitis, and tumors, especially meningiomas. Regardless of the actual antecedent event, decreased flow within a dural venous sinus or vein seems to play a critical role in the development of DAVFs.

Carotid-cavernous fistulas (CCFs) are typically considered distinct from most other intracranial DAVFs, and are discussed by Pradilla and colleagues elsewhere in this issue. CCFs are typically divided into direct or indirect fistulas. Direct or high-flow fistulas are caused by either trauma (blunt or penetrating) or rupture of an intracavernous carotid artery aneurysm. These events create a direct communication between the cavernous carotid artery and the cavernous sinus. Indirect or low-flow fistulas typically occur spontaneously or by occlusion of the cavernous sinus. These events create a connection typically between branches of the external carotid artery and the cavernous sinus via dural arteries and veins. This creates an indirect connection between the carotid artery and the cavernous sinus. Indirect CCFs behave similarly to the intracranial DAVFs discussed here.

Clinical characteristics

Presentation

Patients with DAVFs typically present with symptoms at a mean age of 50 to 60 years. In recent years, with the widespread availability of MRI, there has been an increased frequency of incidentally discovered DAVFs. Nonetheless, most DAVFs are discovered when patients develop symptoms related to their fistula. Symptoms associated with DAVFs highly depend on the characteristics of the venous outflow. These symptoms can be due to either increased dural sinus drainage or the development of cortical venous hypertension.

Symptoms associated with increased dural sinus drainage depend on the location of venous drainage. Anterior fossa lesions are typically supplied by ethmoidal arteries and drain into the cavernous sinus. Because of their proximity to the orbit, these DAVFs therefore typically present with ocular symptoms including proptosis, chemosis, ophthalmoplegia, decreased visual acuity, or retro-orbital pain. Middle fossa lesions commonly drain into the transverse or sigmoid sinus, which is in close proximity to the auditory apparatus. These fistulas typically cause symptoms of pulsatile tinnitus. Fistulas that drain into the superior sagittal sinus or deep venous system produce symptoms of global venous congestion and long-term intracranial hypertension, and may manifest with symptoms of hydrocephalus, papilledema, seizures, or dementia. Brainstem DAVFs, though less common than the other locations, can present with cranial neuropathies and/or quadriparesis.

Besides location of sinus drainage, patient presentation highly depends on whether cortical venous hypertension is present. The presence of cortical venous hypertension typically causes more severe symptoms, including ICH and neurologic deficits. ICH typically occurs within the parenchyma and can also occur in the subarachnoid or subdural space. ICH is believed to result from rupture of fragile parenchymal veins that have been arterialized secondary to cortical venous reflux and hypertension. Blood will be distributed within the parenchyma, or subdural and/or subarachnoid spaces, depending on where the vein ruptures. The overall hemorrhage rate has been documented as being 2% per year but depends on the degree of cortical venous reflux. In addition to ICH, neurologic deficits can occur in patients with DAVF and cortical venous hypertension. These deficits include progressive dementia, seizures, parkinsonism, and other focal neurologic deficits including aphasia, alexia, weakness, paraesthesias, and ataxia. In more severe cases, signs of increased intracranial pressure may be present including headaches, papilledema, upgaze palsy, and mental status changes. These neurologic deficits are believed to result from the effects of venous congestion and cerebral ischemia from progressively arterialized cortical venous drainage. Iwama and colleagues and Kuroda and colleagues performed positron emission tomography (PET) to evaluate the hemodynamics and metabolic patterns in patients with DAVFs before and after treatment. Both these studies demonstrated decreased blood flow and increased oxygen extraction ratios in the regions surrounding the site of cortical venous drainage. These PET findings consistent with cerebral ischemia improved in most patients after either surgical or endovascular treatment of their DAVF.

Clinical Course

DAVFs typically follow a progressive clinical course, depending on the presence and development of progressive cortical venous reflux. DAVFs without cortical venous reflux usually present incidentally or with signs of increased dural drainage, such as tinnitus. These fistulas typically have a benign natural history. Satomi and colleagues performed the largest study thus far in patients without cortical venous drainage. Among 68 patients without cortical venous drainage who were conservatively followed, only 1 (1%) patient suffered an ICH and no patients developed neurologic deficits at a mean follow-up time of 27.9 months. In addition, 50 patients had follow-up catheter angiography at last follow-up, and only 2 (4%) developed cortical venous drainage. This and other studies have shown that DAVFs without cortical venous drainage on digital subtraction angiography at the time of diagnosis have a low risk of causing ICH or neurologic deficits. However, a small subset of these lesions, albeit extremely rarely, can develop cortical venous drainage and subsequent venous stenosis, thrombosis, increased arterial flow, and/or de novo fistula formation. This progression has a higher risk of ICH or neurologic deficits.

DAVFs with cortical venous drainage have a higher propensity to cause ICH or neurologic deficits ( Fig. 1 ). In 1999 Duffau and colleagues were the first to document a high risk of bleeding for patients with DAVFs and cortical venous drainage. These investigators evaluated 20 patients who presented with ICH secondary to DAVFs and venous reflux, and found that 35% of these patients incurred ICH at a mean time of 20 days after diagnosis. More recently in 2002, van Dijk and colleagues described 20 patients with partially treated or conservatively followed DAVFs with cortical venous drainage for a mean period of 4.3 years. At last follow-up, 16 (80%) patients developed ICH (25%) or ischemic deficits (66%). An annual risk of 8.1% was calculated for ICH and 6.9% for neurologic deficits. DAVFs that occur along the tentorium make up the greatest percentage of lesions with cortical venous reflux. Not surprisingly, these tentorially located DAVFs are also thought to have the highest risk of hemorrhage. Awad and colleagues performed a meta-analysis on 360 tentorial DAVFs, and 31 of 32 cases with cortical venous drainage presented with hemorrhage or nonhemorrhagic stroke. Besides cortical venous drainage and tentorial location, other risk factors predisposing to ICH or neurologic deficit include presence of a venous varix and lesions involving the deep drainage system.