Introduction
By definition, dural arteriovenous fistulas (DAVFs) represent arteriovenous shunts from a dural arterial supply to a dural and/or subarachnoid venous drainage channel. Although the pathogenesis of DAVFs is still controversial, most authors agree that DAVFs are acquired lesions.3)5)9)10)12)18)21)24)26)29) While the incidence of cranial DAVFs is approximately 10~15% of intracranial vascular malformations, craniospinal DAVFs are very rare.25) The natural history of DAVFs is highly variable and the pathophysiology of DAVFs is still not well understood.22)23) While the majority of DAVFs drain into the transverse or sigmoid sinus (60%), drainage into the cavernous sinus comprises 15%, tentorial incisura 10%, the superior sagittal sinus and dural convexity 10%, followed by fistulas in the anterior and middle cranial fossa. The nature and severity of symptoms mainly depend on the location and the type of venous drainage. Most DAVFs are benign and some may spontaneously involute. However DAVFs may have an aggressive neurological course or present as a fatal cerebral hemorrhage due to dilatation of cortical veins.11)17)19) The outcome and prognosis of DAVFs also have abroad range. It has been noted that venous drainage patterns are the most important predictors of clinical course and prognosis. Therefore, a better understanding of the angiographic aspects of DAVFs, including the presence of leptomeningeal drainage, may help to forecast a patient's course or outcome. While the treatment goal of DAVFs is total obliteration of the lesion, elimination of DAVFs are difficult and may require many therapeutic approaches. Various treatment strategies for DAVFs have been described in the literature. Accepted treatment options include endovascular embolization, microsurgery, radiosurgery or a combination of these methods. We retrospectively reviewed 10 cases of DAVFs excluding carotid-cavernous fistula (CCF), analyzed clinical and radiological features especially focused on venous drainage pattern, and correlated them with treatment strategies and outcomes.
Materials and Methods
Patient Population During the period between June 2001 and January 2008, 10 patients with dural arteriovenous fistulas, excluding CCF of traumatic origin, were treated at our institution. Age of these patients ranged from 21 to 69 years (mean 52.6 years), and male to female ratio was 7 to 3, respectively. Follow-up ranged from 12 to 76 months (mean 33.0 months). Medical records and radiological studies were retrospectively reviewed to analyze clinical and radiological features, treatment modalities, and outcome.
Angiographic Classification We categorized DAVFs according to Cognard's classification (Table 1) which divided DAVFs into five types according to their venous draining patterns.7) TypeⅠ, located in the main sinus with antegrade flow; typeⅡ, in the main sinus with reflux into the sinus (Ⅱa), cortical veins (Ⅱb), or both (Ⅱa+b); typeⅢ, with direct cortical venous drainage without venous ectasia; typeⅣ, with direct cortical venous drainage with venous ectasia; typeⅤ, with spinal venous drainage.6)7)
Treatment modalities and Evaluation of Outcome We performed endovascular embolization, surgical interruption, or a combination of these treatment based on clinical symptoms, fistular type and drainage pattern, and pathophysiology. Outcomes of these patients were assessed both angiographically and clinically. Angiographic outcomes were classified as incomplete or complete, based upon the angiographic assessment after treatment. The result was classified as incomplete when there was evidence of remaining arteriovenous shunt even though there was significant reduction in the size and flow of the fistula. The obliteration was considered as complete when angiography showed no evidence of fistula after treatment. Clinical outcomes were classified according to the Glasgow outcome scale (GOS).
Results
Angiographic Types & Clinical Presentations Table 2 summarizes the patient's angiographic type, clinical symptoms, and other parameters of DAVFs on initial diagnosis. In case 4 and 9, DAVFs were located in the sagittal (Fig. 3C, 3D) and sigmoid sinus (Fig. 5C, 5D) with antegrade flow, respectively. Both belonged to typeⅠ based on Cognard's classification (Fig. 3). Case 9 had pulsatile tinnitus which is often associated with high flow fistulas in the sigmoid sinus. This happens because of the recruitment of arterial feeders and the close proximity to the middle ear. Case 4 had only mild headache. In cases 1, 3, and 5, DAVFs were located in the sigmoid (Fig. 1B) or sagittal sinus, with reflux into the cortical veins (typeⅡb). These patients usually presented with more aggressive clinical manifestations to include mental status changes with intracranial hemorrhage (Fig. 1A). In case 6, DAVFs were located in the sagittal sinus with reflux into the sagittal sinus and cortical veins (typeⅡ+b). This patient experienced seizure episodes. In case 2 and 8, DAVFs were engaged in direct occipital venous drainage without venous ectasia (Fig. 2C, 2D, type Ⅲ). While prior studies have usually demonstrated a more aggressive clinical manifestation, case 2 only complained of headache and visual disturbance, which may be related to occipital lobe edema due to venous hypertension (Fig. 2A, 2B) (Fig. 2). Case 8 presented with mental status changes with intracerebral hemorrhage which is typically fatal. In case 7 and 10, DAVFs showed spinal venous drainage (Fig. 4C, Fig. 6C, type Ⅴ). Case 7 presented with headaches and posterior neck pain associated with a subarachnoid hemorrhage (Fig. 4A) (Fig. 4). Case 10presented with quadriplegia and respiratory paralysis as a result of spinal venous hypertension and edema (Fig. 6A, 6B).30)
Treatments and Outcomes In our study, 6 cases (Case 5, 6, 7, 8, 9, 10) underwent endovascular embolization, 1 (Case 10) of which underwent a combined surgical procedure. These cases involved Cognard's classification typeⅠ (Case 9), typeⅡb (Case 5), typeⅡa+b (Case 6), typeⅡ (Case 8), and typeⅤ (Case 7, 10). Treatment modalities of typeⅠ utilize methods that diminish the vascular flow. On typeⅡband typeⅡa+b, because of the high risk of bleeding associated with cortical reflux, complete occlusion or at least suppression of the cortical venous reflux is mandatory. As a result, transarterial embolization was attempted on case 5, 6, 9. On type Ⅲ and typeⅤ, aggressive treatments are required because of the severe neurological deficits, intracranial hemorrhaging, and high possibility of rebleeding at the dural arteriovenous fistular site. However, on case 8, patient who presented with mental change with intracerebral hemorrhage (ICH), intraventricular hemorrhage (IVH), must have been operated, as the site of DAVF very close was transverse sinus, it was tried endovascular embolization for surgical high risk. On case 7 and 10, trans-arterial embolization was attempted, in effort to reduce flow, which may result in the disappearance of symptoms and later help if requiring surgery. Trans-arterial embolization was incomplete for case 10 and an operation was subsequently performed. Four cases (Case 5, 6, 7, 9) of those accomplished resulted in a GOS 5 clinical outcome. Case 8 had a GOS of 4, while case 10 had a GOS of 3. Five patients (Case 5, 6, 7, 8, 9) were deemed angiographically complete, and 1 patient (Case 10) was considered incomplete. The 4 cases (Case 1, 2, 3, 4) underwent a surgical procedure. These cases belong to typeⅠ (Case 4), typeⅡb (Case 1, 3), and typeⅢ (Case 2). On case 4, because of multiple feeding arteries, it was difficult to access the DAVF sites by endovascular procedure and as a result, surgical interventions were pursued. On case 1 and 3, because aggressive clinical manifestations such as mental status changes required aggressive therapeutic measures, surgical interventions were pursued to decrease intracranial pressures and removal DAVFs. On case 2, surgery was performed based on the patient's clinical presentation and difficulty to access the DAVF sites by endovascular procedure. In 2 cases (Case 2, 4), clinical outcomes were measured as a GOS 5. In case 1, the GOS was 4, while case 3 had a GOS of 3. All cases (Case 1, 2, 3, 4) were angiographically complete.
Illustrative Cases Case 9. A 21-year-old man presented with pulsatile tinnitus one year after a traffic accident. There were no other neurological deficits at the time of first admission. Cerebral angiography 14 months after the traumatic head injury demonstrated a DAVFs between the right occipital artery and right jugular bulb (Fig. 5C, 5D). The feeding artery was the right occipital artery and it drained into the ipsilateral sigmoid sinus with antegrade flow. The DAVF was categorized as a typeⅠ by Cognard's classification. Although his lesion was considered benign because there was no cortical venous reflux associated with increased hemorrhagic risks, treatment was considered because of his disturbing tinnitus symptoms. Treatment modalities could be adopted to diminish the vascular flow. Trans-arterial embolization with particles can reduce flow, which results in disappearance of symptoms and sometimes provides a complete cure. Follow up angiography 5 days after endovascular embolization, no evidence of recurred or residual abnormal arteriovenous (AV) connection was seen. (Fig. 5E, 5F) The patient was observed for 34 months and has had no abnormal neurological symptoms. His clinical outcome was 5 in GOS (Fig. 5). Case 1. A 63-year-old woman presented with mental status changes after sudden onset of headaches, nausea, and vomiting on admission. Brain CT scans showed an ICH involving the left temporo-occipito-parietal lobe, subarachnoid hemorrhage (SAH) and IVH involving bilateral lateral ventricles, third and fourth ventricles (Fig. 1A). Cerebral angiography on admission demonstrated DAVFs between the left occipital artery and left sigmoid sinus. The feeding artery was the left occipital artery and it drained into the ipsilateral sigmoid sinus. The venous drainage was accomplished to the sigmoid sinus with reflux into the cortical veins (Fig. 1B). The DAVF was categorized as a typeⅡb by Cognard's classification. Treatment was indicated for surgical intervention because there was intracranial hemorrhage (ICH, IVH) associated with cortical venous reflux, requiring both a decrease intracranial pressure and removal the of DAVFs lesion. Follow up angiography two months after initial angiography showed no evidence of recurrence or residual abnormal AV connection (Fig. 1C). The patient subsequently had a clinical outcome 4 in GOS (Fig. 1). Case 10. A 68-year-old man suffered progressive quadriparesis and quadriplegia over 3 days and with sudden respiratory arrest. Preoperative T2 weighted MRI demonstrated flow-related signal voids on the ventral and dorsal surface of the upper cervical cord. T2 weighted images indicated an increased signal from a swollen spinal cord between the medulla and upper thoracic spinal level (Fig. 6A). Selective spinal angiography demonstrated craniospinal DAVFs at the craniocervical junction which was supplied from radicular arteries at the C1 and C2 level and right occipital artery. The DAVF was noted to drained into the spinal perimedullary veins (Fig. 6C). Due to the myelopathy from perimedullary venous hypertension and venous engorgement at the spinal epidural space, the initial goal of treatment was urgent elimination of the spinal arteriovenous fistular site. Although trans-arterial embolization may significantly decrease flow through the DAVFs, it was unlikely to result in complete and permanent obliteration of fistula. Therefore, the procedure involved a combination of a retrograde trans-venous approach and surgical maneuvers.16) First, we performed a trans-arterial endovascular embolization with glue for fast relief of serious symptoms and to reduce bleeding during the operation. Postembolization angiography showed no evidence of craniospinal DAVFs. However, spinal angiography performed 50 day after endovascular embolization, showed evidence of residual or recurrent shunts (Fig. 6D). An operation for termination of residual craniospinal DAVFs was then successfully performed. He eventually had a clinical outcome 3 in GOS (Fig. 6).
Discussion
Angiographic Types and Presentations Various radiological characteristics were identified that might predict intracranial hemorrhage. The risk of hemorrhage was reported to be 1.5% per year by Brown and his colleagues.4) The presence of an draining vein aneurysmal dilatation or a venous varix frequently correlates with an increased intracranial hemorrhage risk.2)13)19) Many authors reported that the leptomeningeal retrograde venous drainage, venous ectasia, and galenic drainage were all noted to predispose the patient to an aggressive course. By multiple variance analysis, leptomeningeal venous drainage and venous ectasia were identified as the most significant risk factors for developing an aggressive clinical course and poor outcome. Cognard et al. also reported that the presence of an ectasia larger than 5mm in diameter or three times larger than the diameter of the draining vein of DAVFs statistically correlated with an increased frequency of hemorrhage.6) Therefore, DAVFs with a venous ectasia is an important risk factor for hemorrhage. However, it is difficult to ascertain whether the ectasias alone were always responsible for the hemorrhage or whether the ectasias existed because of insufficient venous drainage of DAVFs, resulting in an increased hemorrhage risk. There was no definitive correlation between location and the presence of aggressive symptoms. However, the venous drainage in the anterior cranial fossa and tentorium almost always drain into intradural veins, and as a result is associated with increased hemorrhagic conditions. For example, DAVFs in the tentorial incisura are frequently associated with drainage into the leptomeningeal vein, it might be related to intracranial hemorrhage, represent poor natural history. Otherwise, Cognard et al. asserted that age, gender, and flow velocity have no correlation with aggressive clinical symptoms.6) In our study, we reviewed 5 cases with intracranial hemorrhage; 3 cases (Case 1, 3, 5) in typeⅡb, 1 case (Case 8) in typeⅢ, the last case (Case 7) in typeⅤ. Cognard et al. reported approximately 20%, 40%, 46% hemorrhage risks in typeⅡb, typeⅢ, and typeⅤ DAVFs, respectively. All type Ⅱb DAVFs showed intracranial hemorrhage (ICH, IVH, SAH), but because of the small study size, this finding cannot be confidently correlated. Half of the cases with type Ⅲ, Ⅴ DAVFs demonstrated intracranial hemorrhage (ICH, IVH, SAH), which is similar to prior reports.
Treatment Modalities and Goals The purpose of DAVFs therapies are to prevent complications, reduce symptoms, and potentially cure the patient through the elimination of DAVFs. Therapeutic maneuvers include conservative management, endovascular embolization, microsurgery, radiosurgery, and a combination of these treatments.27)28) The therapeutic plan will be based on the pattern of venous drainage, the patient's clinical presentation, lesion location, and the natural history of the lesion. On angiography, if the venous reflux did not have an antegrade drainage into the main sinus or if it was not associated with a hemorrhage (Case 4, 9; typeⅠ of Cognard's classification), treatment modalities adopted methods that diminished the vascular flow. Conservative management is preferable in patients with mild local clinical symptoms because of the high probability of spontaneous occlusion or post angiography occlusion. Trans-arterial embolization can produce reduction of flow, which results in disappearance of symptoms and sometimes provides complete cure as seen in some cases. If there is antegrade reflux to the main sinus (typeⅡa), treatment is aimed to reduce flow and venous hypertension. First, trans-arterial embolization can be attempted. If it does not yield satisfactory results, more aggressive treatments such as embolization with glue and trans-venous occlusion may be considered. With typeⅡb (Case 1, 3, 5) and typeⅡ a+b (Case 6), because of the increased risk of bleeding associated with cortical reflux, complete occlusion or at least suppression of the cortical venous reflux is mandatory. This may sometimes be achieved by arterial embolization, but sinus occlusion may be necessary and can be accomplished by surgical resection of the sinus or trans-venous sinus occlusion. With regard to typeⅢ (Case 2, 8), typeⅣ, and typeⅤ (Case 7, 10), the patient who has retrograde leptomeningeal reflux must be treated aggressively. Because these patients usually have intracranial hemorrhages with neurological deficits, the possibility of rebleeding is very high at the dural arteriovenous fistula site. The management should proceed as soon as possible after a diagnosisis made. Because time is also a major factor affecting outcomes, endovascular embolization may initially be more suitable than surgery. However, partial obliteration does not protect against hemorrhages. The goal of treatment must be complete occlusion utilizing endovascular treatments, surgical management, radiosurgery, or combination of therapies.8)14)
Conclusion
The various clinical presentations, imaging characteristics, and natural history of DAVFs is now better understood through advancements in medical techniques. The purpose of therapy is to reduce symptoms, prevent complications, and potentially provide a cure by eliminating DAVFs. The therapeutic plan should be based on the pattern of venous drainage, lesion location, the natural history of the lesion, and the patient's clinical presentation. Aggressive management should be adopted for patients with intracranial hemorrhage or neurological deficits secondary to retrograde venous drainage into leptomeningeal veins. Treatment modalities are chosen based on a multidisciplinary approach depending on individual therapeutic goals. This work was supported by chung buk national university grant in 2007.
REFERENCES
-
Awad IA, Little JR, Akrawi WP, Ahl J. Intracranial dural arteriovenous malformations: factors predisposing to an aggressive neurological course. J Neurosurg 72:839-50, 1990 < |