Cranial dural arteriovenous fistulas (DAVFs) represent an important class of cranial vascular lesions. The clinical significance of these lesions is highly dependent on the pattern of venous drainage, with cortical venous reflux being an important marker of an aggressive, high-risk fistula. For asymptomatic benign fistulas, conservative management, consisting of observation with follow-up, is a reasonable option. For symptomatic benign fistulas or aggressive fistulas, treatment is recommended. A variety of treatment modalities are available for DAVF management, including endovascular techniques, open surgery, and radiosurgery. A multimodality approach is often warranted and can offer improved chances of achieving a cure.
Cranial dural arteriovenous malformations, commonly referred to as dural arteriovenous fistulas (DAVFs), are abnormal shunts between dural arteries and a dural venous sinus or a cortical vein. DAVFs occur throughout the intracranial space, but are most often located near and involve 1 or more dural venous sinuses. Depending on the extent of the lesion and its associated venous drainage, DAVFs are either managed conservatively by observation or treatment is recommended. Various treatment options for DAVFs are available, including endovascular embolization, open surgery, and radiosurgery. The clinical decision pathways for managing patients with DAVFs are discussed in this article.
Review of DAVF classification schemes
To appropriately triage a patient with a DAVF, it is essential to first understand the anatomy of the fistula and the involved venous pathways. Various classifications schemes have been devised to describe these lesions, and the Borden and the Cognard schemes are the most widely used. The key feature for both of these grading systems is the direction and severity of venous drainage, and these features ultimately determine the clinical significance of the lesion and influence the decision to treat.
The Borden classification system groups DAVFs into 3 main types based on the location of venous drainage (dural sinus and/or cortical vein) and on whether cortical venous reflux (CVR) is present or absent. Borden type 1 DAVFs have anterograde venous drainage into a dural sinus and no CVR. Type 2 lesions drain retrogradely into dural sinuses and also into cortical veins, with associated CVR. Type 3 lesions drain directly into a cortical veins, causing significant CVR.
The Cognard system groups DAVFs into 5 main types based on the direction of venous sinus drainage (anterograde or retrograde), the presence or absence of CVR, and the type of venous outflow (eg, nonectatic vs ectatic cortical veins). A Cognard type 1 DAVF has anterograde flow with no CVR. A type II DAVF drains into a sinus with retrograde flow in the sinus (type IIA), retrograde flow in a cortical vein (type IIb), or retrograde flow in the sinus and cortical veins (type II A+B). Type III DAVF drains retrogradely into a nonectatic cortical vein, and type IV drains into an ectatic cortical vein. Lastly, Cognard type V DAVFs drain into a spinal perimedullary vein.
Both of these classification systems highlight the importance of CVR; however, Zipfel and colleagues proposed modification of these classic schemes to include clinical information about whether or not the CVR is symptomatic. This modification is based on studies by Soderman and colleagues and Strom and colleagues that show that patients presenting with intracranial hemorrhage (ICH) or nonhemorrhagic neurologic deficit (defined as symptomatic CVR) have a greater neurologic risk and death than patients who present incidentally or solely with symptoms of increased dural sinus drainage (defined as asymptomatic CVR). The new scale proposed by Zipfel and colleagues is as follows: 1) type 1 DAVFs drain only into a dural sinus and have no CVR; 2) type 2 DAVFs drain into a dural sinus and have either asymptomatic or symptomatic CVR; and 3) type 3 DAVFs drain directly into cortical veins and have either asymptomatic or symptomatic CVR. Neurologic risk increases from type 1 lesions to type 2 or 3 lesions with asymptomatic CVR (1.4–1.5% annual risk of ICH) to type 2 or 3 lesions with symptomatic CVR (7.4%–7.6% annual risk of ICH). Zipfel and colleagues recommend that patients with symptomatic CVR (type 2 or 3 DAVF) undergo immediate, definitive treatment of the fistula by an endovascular or open surgical approach. Stereotactic radiosurgery is not recommended in these patients because of the time delay required for occlusion of the fistula.
Carotid cavernous fistulas (CCFs) are another important group of DAVFs. The most widely used classification system for these fistulas is the Barrow system. CCFs are categorized into direct (Barrow type A) and indirect (Barrow types B–D). Indirect CCFs are DAVFs with arterial feeders from the internal carotid artery (Barrow type B), external carotid artery (Barrow type C), or both (Barrow type D).
The natural history of DAVF s
Insight into the natural history of DAVFs has been instrumental in the development of clinical decision pathways for managing patients with DAVFs. The natural history of DAVFs is known to be highly variable, and the potential of the lesion to cause morbidity and mortality needs to be weighed against the risks of the intervention used for treatment. Many studies have tried to identify independent factors predictive of the natural history of DAVFs to develop informed treatment plans and decrease mortality and morbidity. Many patients with DAVFs have a benign course, but 7% to 20% present with ICH. Mortality and morbidity of a hemorrhagic presentation of a DAVF has been estimated to be 20% to 30% with poor long-term prognosis. Individual factors related to the natural history of DAVFs should be carefully considered when making clinical decisions regarding treatment versus observation.
Location
Awad and colleagues analyzed 377 cases of DAVFs in a meta-analysis. They defined aggressive neurologic behavior as hemorrhage or progressive focal neurologic deficit other than ophthalmoplegia. The investigators found that DAVFs located at the tentorial incisura (8.4% of cases) were most often associated with aggressive neurologic behavior, and 96.9% of these tentorial lesions (31 out of 32) were associated with a hemorrhagic or nonhemorrhagic stroke. Conversely, DAVFs located at the transverse-sigmoid sinuses (62.6% of cases) and at the cavernous sinus (11.9% of cases) were least likely to be associated with aggressive neurologic behavior. Only about 10% of these lesions were associated with aggressive behavior. However, the investigators found that there was no location of DAVFs that completely precluded aggressive neurologic behavior.
In an analysis of 402 patients, Singh and colleagues found that patients were more likely to present with ICH if they had a DAVF in ethmoidal or posterior fossa locations. Twelve patients in their series had ethmoid DAVFs, and 9 of these patients (75%) presented with ICH. Thirty-two patients had DAVFs in the posterior fossa, and 17 (53%) of these patients presented with ICH. The odds ratio (OR) for a hemorrhagic presentation of a DAVF in the posterior fossa is 4. Patients harboring DAVFs in cavernous, marginal, and transverse-sigmoid sinus locations were least likely to present with ICH. Hemorrhage rates were 6.1% (10 of 163 patients) for cavernous sinus lesions, 5.9% (1 of 17 patients) for fistulas in the marginal sinus, and 15.3% (19 of 124 patients) for DAVFs of the transverse-sigmoid sinus.
The anatomic location of a DAVF likely has no direct correlation with aggressive clinical course. The more reasonable explanation for the findings mentioned earlier is that certain cranial locations, secondary to local venous anatomy, have a higher likelihood of developing CVR. The severity of the CVR then dictates the clinical course.
Venous Drainage
It is generally accepted that the presentation of a DAVF is dictated by its venous drainage pattern. The pattern of venous drainage is used to classify DAVFs into high-risk and low-risk groups. In the meta-analysis of 377 cases of DAVFs, Awad and colleagues analyzed the relationship between angiographic features of DAVFs and aggressive neurologic behavior. Aggressive neurologic behavior was correlated with leptomeningeal retrograde venous drainage, variceal or aneurysmal venous dilatations, and galenic venous drainage. Less than 1% of nonaggressive DAVFs were found to exhibit these 3 features simultaneously. However, every aggressive DAVF exhibited 1 or more of the 3 angiographic features. The presence of high-flow shunting or contralateral arterial contribution did not correlate with aggressive neurologic behavior.
In a series of 402 patients with DAVFs, Singh and colleagues found that patients presenting with ICH were more likely to have angiographic evidence of venous sinus thrombosis and CVR. Eighty-five percent of patients with DAVFs presenting with ICH had angiographic evidence of CVR, compared with 22% of patients with nonhemorrhagic DAVFs ( P <.001, OR 10.5). Venous sinus thrombosis was found in 33% of patients with DAVFs presenting with ICH and only 18% of patients with a nonhemorrhagic presentation ( P = .004). Ninety-five percent of DAVFs were classified as Borden type II and III compared with 45% in the nonhemorrhagic group ( P <.001).
Strom and colleagues sought to differentiate between symptomatic CVR and asymptomatic CVR to further risk stratify patients diagnosed with type 2 and 3 DAVFs. Patients with symptomatic CVR were defined as presenting with symptoms of cortical venous hypertension, namely ICH or nonhemorrhagic neurologic deficits (progressive dementia, seizures, parkinsonism, cerebellar symptoms, or other focal deficits). Patients with asymptomatic CVR presented incidentally or with isolated symptoms of increased dural sinus drainage (ie, pulsatile tinnitus for lesions of the transverse-sigmoid sinus). The investigators followed 28 patients with persistent cortical venous drainage (Borden type 2 and 3), and they found that the frequency of ICH or nonhemorrhagic neurologic deficit was significantly lower in patients with asymptomatic CVR (5.9%) versus symptomatic CVR (45.5%, P = .022). The annual event rate for either ICH or nonhemorrhagic neurologic deficit in patients with symptomatic cortical venous drainage was 19.0% compared with 1.4% in patients with asymptomatic cortical venous drainage. These data suggest that DAVFs with asymptomatic CVR are less aggressive than those with symptomatic CVR.
Demographics and Symptoms
Singh and colleagues found DAVFs to be more prevalent in women compared with men but found that men were more likely to present with ICH (74% vs 36%, P <.01, OR 3.4). A history of smoking was more common in patients presenting with ICH (23% vs 12%, P = .008), and Hispanic patients were more likely to present with ICH than a nonhemorrhagic presentation (15% vs 8%, P = .05).
Singh and colleagues also analyzed specific symptoms and their association with ICH. Headache, pulsatile tinnitus, and visual changes were the most common symptoms on presentation; however, headache, focal neurologic deficit, and seizure were significantly associated with ICH at presentation. Focal neurologic deficit was found to have an OR of 4.7 for hemorrhagic presentation in patients with DAVFs.
The natural history of DAVF s
Insight into the natural history of DAVFs has been instrumental in the development of clinical decision pathways for managing patients with DAVFs. The natural history of DAVFs is known to be highly variable, and the potential of the lesion to cause morbidity and mortality needs to be weighed against the risks of the intervention used for treatment. Many studies have tried to identify independent factors predictive of the natural history of DAVFs to develop informed treatment plans and decrease mortality and morbidity. Many patients with DAVFs have a benign course, but 7% to 20% present with ICH. Mortality and morbidity of a hemorrhagic presentation of a DAVF has been estimated to be 20% to 30% with poor long-term prognosis. Individual factors related to the natural history of DAVFs should be carefully considered when making clinical decisions regarding treatment versus observation.
Location
Awad and colleagues analyzed 377 cases of DAVFs in a meta-analysis. They defined aggressive neurologic behavior as hemorrhage or progressive focal neurologic deficit other than ophthalmoplegia. The investigators found that DAVFs located at the tentorial incisura (8.4% of cases) were most often associated with aggressive neurologic behavior, and 96.9% of these tentorial lesions (31 out of 32) were associated with a hemorrhagic or nonhemorrhagic stroke. Conversely, DAVFs located at the transverse-sigmoid sinuses (62.6% of cases) and at the cavernous sinus (11.9% of cases) were least likely to be associated with aggressive neurologic behavior. Only about 10% of these lesions were associated with aggressive behavior. However, the investigators found that there was no location of DAVFs that completely precluded aggressive neurologic behavior.
In an analysis of 402 patients, Singh and colleagues found that patients were more likely to present with ICH if they had a DAVF in ethmoidal or posterior fossa locations. Twelve patients in their series had ethmoid DAVFs, and 9 of these patients (75%) presented with ICH. Thirty-two patients had DAVFs in the posterior fossa, and 17 (53%) of these patients presented with ICH. The odds ratio (OR) for a hemorrhagic presentation of a DAVF in the posterior fossa is 4. Patients harboring DAVFs in cavernous, marginal, and transverse-sigmoid sinus locations were least likely to present with ICH. Hemorrhage rates were 6.1% (10 of 163 patients) for cavernous sinus lesions, 5.9% (1 of 17 patients) for fistulas in the marginal sinus, and 15.3% (19 of 124 patients) for DAVFs of the transverse-sigmoid sinus.
The anatomic location of a DAVF likely has no direct correlation with aggressive clinical course. The more reasonable explanation for the findings mentioned earlier is that certain cranial locations, secondary to local venous anatomy, have a higher likelihood of developing CVR. The severity of the CVR then dictates the clinical course.
Venous Drainage
It is generally accepted that the presentation of a DAVF is dictated by its venous drainage pattern. The pattern of venous drainage is used to classify DAVFs into high-risk and low-risk groups. In the meta-analysis of 377 cases of DAVFs, Awad and colleagues analyzed the relationship between angiographic features of DAVFs and aggressive neurologic behavior. Aggressive neurologic behavior was correlated with leptomeningeal retrograde venous drainage, variceal or aneurysmal venous dilatations, and galenic venous drainage. Less than 1% of nonaggressive DAVFs were found to exhibit these 3 features simultaneously. However, every aggressive DAVF exhibited 1 or more of the 3 angiographic features. The presence of high-flow shunting or contralateral arterial contribution did not correlate with aggressive neurologic behavior.
In a series of 402 patients with DAVFs, Singh and colleagues found that patients presenting with ICH were more likely to have angiographic evidence of venous sinus thrombosis and CVR. Eighty-five percent of patients with DAVFs presenting with ICH had angiographic evidence of CVR, compared with 22% of patients with nonhemorrhagic DAVFs ( P <.001, OR 10.5). Venous sinus thrombosis was found in 33% of patients with DAVFs presenting with ICH and only 18% of patients with a nonhemorrhagic presentation ( P = .004). Ninety-five percent of DAVFs were classified as Borden type II and III compared with 45% in the nonhemorrhagic group ( P <.001).
Strom and colleagues sought to differentiate between symptomatic CVR and asymptomatic CVR to further risk stratify patients diagnosed with type 2 and 3 DAVFs. Patients with symptomatic CVR were defined as presenting with symptoms of cortical venous hypertension, namely ICH or nonhemorrhagic neurologic deficits (progressive dementia, seizures, parkinsonism, cerebellar symptoms, or other focal deficits). Patients with asymptomatic CVR presented incidentally or with isolated symptoms of increased dural sinus drainage (ie, pulsatile tinnitus for lesions of the transverse-sigmoid sinus). The investigators followed 28 patients with persistent cortical venous drainage (Borden type 2 and 3), and they found that the frequency of ICH or nonhemorrhagic neurologic deficit was significantly lower in patients with asymptomatic CVR (5.9%) versus symptomatic CVR (45.5%, P = .022). The annual event rate for either ICH or nonhemorrhagic neurologic deficit in patients with symptomatic cortical venous drainage was 19.0% compared with 1.4% in patients with asymptomatic cortical venous drainage. These data suggest that DAVFs with asymptomatic CVR are less aggressive than those with symptomatic CVR.
Demographics and Symptoms
Singh and colleagues found DAVFs to be more prevalent in women compared with men but found that men were more likely to present with ICH (74% vs 36%, P <.01, OR 3.4). A history of smoking was more common in patients presenting with ICH (23% vs 12%, P = .008), and Hispanic patients were more likely to present with ICH than a nonhemorrhagic presentation (15% vs 8%, P = .05).
Singh and colleagues also analyzed specific symptoms and their association with ICH. Headache, pulsatile tinnitus, and visual changes were the most common symptoms on presentation; however, headache, focal neurologic deficit, and seizure were significantly associated with ICH at presentation. Focal neurologic deficit was found to have an OR of 4.7 for hemorrhagic presentation in patients with DAVFs.
Management of benign DAVF s
The decision to treat an intracranial DAVF or to manage it conservatively with observation is based on both lesion-specific factors and patient-specific factors. Lesion-specific factors include the flow velocity (high flow vs slow flow), location, and venous drainage direction and anatomy, including presence of CVR or venous ectasia. Patient-specific factors include the primary presentation, severity of symptoms, and associated patient comorbidities. Various treatment modalities are available for DAVFs, and these include endovascular embolization (transarterial, transvenous, or both), open surgery, radiosurgery, or a combination of these strategies.
Borden type 1 DAVFs and Cognard type I and type IIa fistulas without cortical venous drainage are generally considered benign secondary to absence of CVR. In some institutions, these fistulas are treated with conservative management by observation unless the associated symptoms are severe. Although the primary goal of DAVF treatment is cure, in some cases therapy should be directed primarily at palliation of symptoms by reduction of flow and not necessary at cure. Attempted complete angiographic cure of a benign fistula might be associated with increased risk of procedural complications, and therefore might not be warranted. Endovascular treatment is often the best option for palliative therapy, and the risks of endovascular treatment include arterial dissection, stroke, pulmonary embolism, infarction of cranial nerves, and redirection of venous drainage toward cortical veins.
However, fistulas that are classically labeled benign must still be monitored if cure is not achieved because these lesions have the propensity to transform into higher grade, aggressive lesions. Cognard and colleagues reported a series of 7 patients with DAVFs on the transverse or sigmoid sinuses that were initially benign (5 patients with Cognard type I lesions and 2 patients with type IIa lesions) but were later found to have changes in venous drainage patterns on follow-up angiography (from 1 month to 20 years). Five of the patients developed CVR secondary to either stenosis or thrombosis of the draining veins, increased arterial flow, the appearance of a new fistula, or extension of the initial fistula shunt. Based on these results, Cognard and colleagues recommended close clinical observation of type I and IIa fistulas that are treated conservatively with observation or with an incomplete intervention. Repeat angiography should be performed for any change in clinical condition. If stenosis of the venous drainage pathway is present, increased vigilance is necessary because these lesions are likely at greater risk for progression.
Satomi and colleagues reported a series of 117 patients with benign cranial DAVFs, only 3 of which were asymptomatic at the time of presentation. Of these 117 patients, 73 (62%) underwent conservative management with observation, 43 (37%) underwent palliative embolization for treatment of symptoms, and 1 (1%) patient had surgical treatment. Follow-up was available for 112 patients (95.7%), with a median follow-up period of 27.9 months, and repeat angiography was performed in 50 patients because of a change in symptoms. The investigators reported that tolerable, stable disease was achieved in 98.5% of the patients with observational management. Of the 50 patients with follow-up angiography, changes in venous drainage occurred in 5 cases, and each of these cases was associated with progressive thrombosis in the affected sinus. Two patients who were managed with observation had new development of CVR. The investigators therefore concluded that most patients with benign DAVFs can be managed satisfactorily with either observation or palliative intervention. However, patients with benign DAVFs have a 2% risk of developing CVR and therefore require close clinical follow-up and repeat angiography for changes in symptoms.

Stay updated, free articles. Join our Telegram channel

Full access? Get Clinical Tree

