PEER-REVIEW REPORTS. various signs and symptoms such as headaches, tinnitus, bruit, neurologic deficits, venous hypertensive encephalopathy with

Transcription

1 Multimodality Treatment of Intracranial Dural Arteriovenous Fistulas in the Onyx Era: a Single Center Experience Sabareesh K. Natarajan 1, Basavaraj Ghodke 1,2, Louis J. Kim 1,2, Danial K. Hallam 1,2, Gavin W. Britz 1,2, Laligam N. Sekhar 1,2 Key Words Dural arteriovenous fistula Embolization Multimodality treatment Onyx Radiosurgery Surgery Abbreviations AVF: Arteriovenous fistula CT: Computed tomography CVR: Cortical venous reflux DAVF: Dural arteriovenous fistula DCF: Dural cavernous fistula ECA: External carotid artery ICA: Internal carotid artery IPS: Inferior petrosal sinus MMA: Middle meningeal artery VA: Vertebral artery l Department of Neurological Surgery and 2 Department of Interventional Neuroradiology, Harborview Medical Center, University of Washington, Seattle, Washington, USA Portions of this work were presented in abstract form and as proceedings at the American Society of Neuroradiology 47th Annual Meeting & Neuroradiology Education and Research Foundation Symposium at Vancouver, Canada, on May 20, To whom correspondence should be addressed: Laligam N. Sekhar, M.D., FACS [ lsekhar@ u.washington.edu] Citation: World Neurosurg. (2010) 73, 4: DOI: /j.wneu Journal homepage: Available online: /$ - see front matter 2010 Elsevier Inc. All rights reserved. INTRODUCTION Dural arteriovenous fistulas (DAVFs) are abnormal arteriovenous connections between an arterial feeder and a dural venous sinus or leptomeningeal vein with the nidus located withintheduralleaflets. Theyaccountfor10% to 15% of all intracranial arteriovenous malformations (44). Although their etiology is not clearly understood, DAVFs have been found in association with venous thrombosis, intracranial surgery, tumor, puerperium, and trauma; (53) and some of these lesions are BACKGROUND: The results of treatment of intracranial dural arteriovenous fistulas (DAVFs) since Onyx became available as an embolic agent at our institution is reported. An algorithm is presented for treatment of DAVFs with Onyx, and the role of endovascular transvenous, surgical, and radiosurgical approaches are presented. METHODS: Thirty-two patients with DAVFs treated between November 2005 and November 2008 by endovascular embolization, surgery, or radiosurgery were identified by a retrospective chart review. Treatment strategies were based on the location or complexity of the fistula and the patient s clinical status. Data collected included DAVF characteristics, obliteration rates, complications, and outcomes. The results were analyzed and correlated with the treatment modality. RESULTS: Presenting symptoms were as follows: hemorrhage (n 12 patients), headaches (n 12), tinnitus (n 5), orbital symptoms (n 7), and seizures (n 1). Thirty patients were treated by endovascular embolization (transarterial only with Onyx-21, transvenous only with platinum coils-6, transarterial [Onyx] and transvenous [coils]-3). Five patients (4 after incomplete/failed embolization) had surgical excision of the fistula. Three patients were treated with Gamma Knife radiosurgery (primary-1, 2 after incomplete/failed embolization). The locations of the fistulas were transverse sigmoid (10 patients), petrotentorial (7 patients), indirect carotid cavernous fistula (7 patients), parasagittal/falcine (3 patients), middle fossa dura (3 patients), torcula (1 patient), and anterior fossa dura (1 patient). The distribution of patients according to Borden classification was I-6, II-13, and III-13. Complete obliteration of the fistula was achieved in 26/32 (81%) patients after multimodal treatment. All surgical cases had complete obliteration. In the high-risk group with cortical venous reflux, 23/26 (89%) patients were cured. Endovascular complications included a stuck microcatheter tip with fracture of the tip in two patients and cranial nerves V and VII palsies in one patient. At last follow-up (range 1 36 months), 24 patients had modified Rankin score of 0 2, 5 patients had modified Rankin score of 3 5, and 3 patients were dead. Two patients died during admission due to the insult of the hemorrhage, and one died after an accidental fall with subsequent traumatic subdural hematoma. CONCLUSIONS: Multimodality treatment of DAVFs has high success rates for cure at our center. Transarterial embolization with Onyx has become the primary treatment for intracranial DAVFs at our center and is associated with high safety profile and efficacy. Transvenous coil embolization is still preferred in DAVFs with supply from arterial branches supplying cranial nerves, predominant internal carotid artery feeders and potential extracranial intracranial collateral anastomosis. In our series, patients with incompletely treated DAVFs were treated with surgery and those with partially treated type I fistulas had radiosurgery for palliation. congenital (21). Depending on their location and the venous drainage pattern, the clinical manifestations range from asymptomatic to various signs and symptoms such as headaches, tinnitus, bruit, neurologic deficits, venous hypertensive encephalopathy with WORLD NEUROSURGERY 73 [4]: , APRIL

2 cognitive dysfunction, and intracranial hemorrhage (3, 33). The venous drainage pattern is the most important predictor of the clinical behavior, and DAVFs with cortical venous reflux (CVR) exhibit a much higher incidence of hemorrhage or venous infarction. The annual mortality rate for DAVFs with CVR may be as high as 10.4%, whereas the annual risk for hemorrhage or nonhemorrhagic neurologic deficits during follow-up are 8.1% and 6.9%, respectively, resulting in an annual event rate of 15% (64). In addition, rebleeding rates may be as high as 35% during the first 2 weeks after the initial hemorrhage (18). Thus, these formidable lesions need careful evaluation to determine the best therapeutic option. Many treatment approaches to these lesions have been proposed (4, 12, 19, 22, 23, 28, 51). With the advent of Onyx (ev3, Irvine, CA), most intracranial DAVFs can be successfully managed with endovascular techniques. Transarterial access to these lesions through their supply from the external carotid artery (ECA) branches, which can be safely embolized, is more often used with Onyx. Their location close to dural venous sinuses also facilitates access and occlusion through the sinus. The combination of transarterial and transvenous embolization results in higher obliteration rates than previously reported in series with only transvenous embolization. However, surgery is still required in certain subsets of patients with 1) complex anatomic features, 2) in case of failure of both transarterial and transvenous embolization, or 3) in lesions for which direct endovascular access is not possible (28). Radiosurgery is mainly used for palliation of type I fistulas or if other modalities fail. In this study we reviewed our experience with a consecutive series of 32 patients subjected to a multimodality approach for intracranial DAVF treatment. On the basis of our initial experience with Onyx and insights into the management of a variety of complex DAVFs, we propose our current algorithm and detailed consideration of the surgical, radiosurgical, and endovascular treatment options for DAVFs at various locations in the present Onyx era. METHODS Figure 1. Patient 1: (A) Noncontrast head computed tomographic scan showing subarachnoid blood. (B) Cerebral angiogram showing an anterior fossa dural arteriovenous fistula (Borden type III) with anterior meningeal feeders from the ophthalmic artery and venous drainage into a frontal leptomeningeal vein that drains into the superior sagittal sinus. Patients The study was approved by the institutional review board at the University of Washington, Harborview Medical Center (Seattle, WA) and conducted in compliance with Health Insurance Portability and Accountability Act regulations. All symptomatic intracranial DAVFs were evaluated and treated by a neurovascular team (LNS, BG, LJK, DKH, and GWB). The team determined whether patients were candidates for preoperative embolization and potential endovascular cure. If a fistula was not amenable to endovascular management (eg, a lack of appropriate arterial or transvenous access) or could not cured by embolization, the patient underwent surgical or radiosurgical treatment. Type I fistulas were treated only if the patient had disabling symptoms, and the goal of treatment was not to cure the fistula but palliation of the symptoms. A review of the records of the cerebrovascular service at University of Washington, Harborview Medical Center identified 32 consecutive patients with DAVFs treated by endovascular embolization (transarterial Onyx, transvenous coil embolization, or both), surgery, or radiosurgery between November 2005 and November The clinical and angiographic features were retrospectively reviewed (by LNS, BG, SKN) from a prospectively maintained endovascular registry. Data collected included age, sex, presenting symptoms, DAVF location, arterial supply, venous drainage, type of fistula Borden et al. (6), treatment modality, outcome, and complications. Presentation The average age at presentation was 52 years (range, 1 86 years). Fifteen patients were men (47%) and 17 were women (53%). Presenting symptoms included hemorrhage (12 patients; 38%), headaches (12; 38%), tinnitus (5; 16%), orbital symptoms (7; 22%), and seizures (1 patient; 3%). Of the 12 patients who presented with hemorrhage, 3 patients had subarachnoid hemorrhage and 1 had concurrent intraparenchymal hematoma. Eight patients presented with intraparenchymal hematoma; one hematoma had an intraventicular extension. One patient had only intraventricular hemorrhage. Because of the high risk of rebleeding, all patients who presented with hemorrhage were treated on an urgent basis. Embolization Procedure All endovascular procedures were performed under general anesthesia and in a biplane angiographic unit with three-dimensional rotational angiographic capability (Philips Medical Systems, Best, The Netherlands). After transfemoral cannulation, diagnostic angiography was performed including the intracranial (internal carotid artery [ICA], vertebral artery [VA]) and extracranial arteries (ECA). In some cases, cervical branches of the subclavian artery were included. The decision to embolize, the route, and the material to be used WORLD NEUROSURGERY, DOI: /j.wneu

3 Figure 2. Patient 1: Angiogram obtained 8 months after surgical resection showing complete obliteration of the fistula. were chosen based on the angiographic findings. Transarterial Approach. After transfemoral access and sheath placement, the patient was heparinized, with 5000 units intravenously and 1000 units per hour of intravenous heparin, to maintain an activated coagulation time of between 250 and 300 seconds. A 6F guiding catheter (MPD; Cordis Neurovascular, Miami Lakes, FL) or a triple coaxial guiding system that allows more distal intracranial placement of the guide (Neuron-2; Penumbra, Inc., Alameda, CA) was navigated into the target vessel. A dimethyl-sulfoxide compatible microcatheter (Marathon or Echelon; ev3) with a microwire (Mirage or X-Pedion; ev3) was navigated into the feeding vessel of the DAVF as close to the nidus as possible and allowing for an adequate distance for reflux of the embolic agent. When there were multiple feeders, the feeder with maximum reflux distance from the skull base was chosen to decrease the chances of cranial nerve palsies. Embolization of the DAVF was then performed with Onyx-34 or Onyx-18 in the manner previously described (43). The end point of the embolization was the angiographic obliteration of the fistula, with some filling of the draining veins. When the fistula was very complex, more than one session of Onyx embolization through multiple feeders was needed. The average interval between such sittings was 2 to 4 weeks. Transvenous Route. The transvenous route was mainly used for indirect ICA cavernous sinus fistulas and some tentorial DAVFs. After diagnostic angiography, the femoral Figures 3 and 4. Patient 2: Cerebral angiogram showing a petrotentorial Borden type I dural arteriovenous fistula supplied by multiple external carotid artery feeders including the middle meningeal trunk (3A,B,C), internal maxillary artery (3D), superficial and deep temporal arteries (3D), occipital artery (4A,B), posterior auricular artery (4C), and the ascending pharyngeal artery (4D). The venous drainage was into the superior petrosal sinus and transverse sigmoid junction with antegrade flow toward the jugular vein and retrograde flow into the transverse sinus toward the torcula and controlateral transverse sinus. WORLD NEUROSURGERY 73 [4]: , APRIL

4 Surgical Technique Preoperative angiographic studies were carefully reviewed for precise location of the draining vein and the fistula. The ideal approach was determined on the basis of the angiographic findings. During surgery, the retrograde leptomeningeal draining vein was interrupted as it exited the fistula and the nidus was excised, with surgical packing of the sinuses in some cases. Intraoperative angiography was performed to verify obliteration of the fistula and to identify any accessory fistulae that were not appreciated before surgery. Radiosurgery High-resolution biplane, stereotactic cerebral angiography that included selective injection of the VA, ECA, or ICA, when appropriate, was used for radiosurgical targeting and dose planning. In complicated cases or when the fistula nidus was thought to be near cranial nerves or the brainstem, stereotactic computed tomography (CT) or magnetic resonance imaging was also used to assist with dose planning. The mean radiation dose to the margin of the fistulas was 19.2 Gy (range, Gy) at the 50% to 60% isodose line using one to eight irradiation isocenters (mean, 4 isocenters). Figure 5. Patient 2: Postembolization Onyx cast showing progress of Onyx embolization during five sessions and the residual fistula that was later treated with radiosurgery. vein was punctured (usually on the contralateral side) and, using road maps obtained from an intra-arterial injection, a 6F guiding catheter was placed in the internal jugular vein. Through this guiding catheter, the microcatheter was advanced through the inferior petrosal sinus (IPS) or the superior petrosal sinus (or one of the clival veins) into the cavernous sinus, placing it as far anteriorly as possible. Occasionally, the superior ophthalmic vein or one of the facial veins (through the pterygoid plexus) and the contralateral IPS were used to access the cavernous sinus. For tentorial DAVFs, the microcatheter was advanced through the draining vein, past the fistula, to the arterial side. Embolization was then performed with platinum coils or Onyx, using the appropriate microcatheter. Follow-up Clinical follow-up data were supplemented by telephonic interviews and graded ac- Figure 6. Patient 3: Cerebral angiogram showing an indirect carotid cavernous fistula (Borden type II) with bilateral meningohypophyseal trunk feeders (A,B) and left external carotid artery feeders draining into the cavernous sinus, superior ophthalmic vein, and inferior petrosal sinus (C,D) WORLD NEUROSURGERY, DOI: /j.wneu

5 by endovascular embolization in 38 sessions: transarterial only with Onyx (21 patients), transvenous only with platinum coils (6), and transarterial and transvenous (3 patients). Seven patients among these 30 received multimodality treatment that included embolization followed by surgery (n 4; 13%) or embolization followed by radiosurgery (n 3; 10%). All seven patients with DCF were treated by tranvenous Guglielmi detachable coil embolization, with one patient receiving additional transarterial Onyx therapy. Figure 7. Patient 3: (A) Tranvenous access to the fistula site through the left superior petrosal sinus. (B) Postembolization cerebral angiogram showing complete obliteration of the fistula. cording to the modified Rankin scale. Most patients had follow-up angiograms approximately 6 months after the procedure or therapy. A neurosurgeon performed a clinical assessment during the postoperative period and at the time of follow-up angiography. All the angiograms were reviewed by the senior neurosurgeon (LNS) and an experienced neuroradiologist (BG) for the Borden grading and the outcome (as measured by modified Rankin scale). The angiographic results of treatment were classified as incomplete or complete, based on the evidence of remnant fistula on the posttreatment angiogram. RESULTS Anatomic and Angiographic Features Thirteen lesions (41%) were classified as Borden type III, another 13 lesions (41%) as Borden type II, and 6 lesions (18%) as Borden type I. Seven of the type II fistulas were dural cavernous sinus fistula (DCFs) in patients who presented with visual symptoms and had no hemorrhage. Ten DAVFs (31%) were located at the transverse/sigmoid sinus region, 7 (22%) were located at the tentorial region, 3 (9%) were located at the falcine/superior sagittal sinus region, 7 (22%) were DCFs 3 (9%) were in the middle fossa, and 1 each (3%) located at the torcular/transverse sinus and anterior fossa region. The fistula was supplied exclusively by branches of the ECA in 11 patients (34%), including the occipital artery, posterior auricular artery, and ascending pharyngeal artery in lesions of the transverse sigmoid sinus and from the middle meningeal artery (MMA), accessory meningeal artery, artery of foramen rotundum, and ascending pharyngeal artery in lesions of the cavernous sinus. Nineteen fistulas (59%) were also supplied from cavernous or intradural ICA branches, and two (6%) had feeders from the dural branches or cerebellar arteries arising from the vertebrobasilar system. The main ICA branches supplying the DAVF were the meningohypophyseal trunk and the tentorial branch. In cavernous sinus lesions, draining veins were mainly the superior ophthalmic vein and the IPS, and leptomeningeal venous drainage was through the superficial sylvian vein and the basal vein of Rosenthal. In transverse sigmoid sinus lesions, the drainage system was the ipsilateral jugular vein and the contralateral transverse sinus internal jugular vein. The main leptomeningeal venous drainage was through the veins of Labbé or the occipital veins. Leptomeningeal venous drainage was present in 24 patients (75%), with venous stenoses and ectasia identifiable in most of these patients. Treatment and Angiographic Cure All fistulas initially had an attempt at treatment by endovascular means except in two patients. Of those two patients, one was treated with microsurgery as the primary option and the other had radiosurgery. Thirty of 32 patients were treated Immediate and Follow-up Embolization Results Of the 30 patients who had endovascular treatment, 21 (70%) had complete obliteration of the fistula after a single or a multiple embolization session. All seven DCFs were completely obliterated. Of the nine patients in whom endovascular treatment/embolization was incomplete, four underwent subsequent surgery and three had radiosurgery. In the remaining two patients, the residual fistula remained stable at the time of follow-up angiogram, and no further treatment was performed. Surgery Surgery achieved complete occlusion of four fistulas after failed or incomplete endovascular treatment. One patient with an anterior fossa DAVF supplied by ethmoidal branches of the ophthalmic artery was cured by direct surgery; in this case, embolization was considered unsafe with significant risk of complications involving the visual system. Radiosurgery Three patients underwent Gamma Knife radiosurgery and will have follow-up angiography 3 years post-treatment. One patient had transient blindness after diagnostic angiography, which was probably secondary to vasospasm, and was referred for primary radiosurgery (26, 58). Obliteration Rates After Multimodal Treatment The goal of complete fistula obliteration was accomplished in 23 of 26 patients (89%) with WORLD NEUROSURGERY 73 [4]: , APRIL

6 Figure 9. Patient 4: Postembolization angiogram showing Onyx cast (A) and complete obliteration of the fistula (B). Figure 8. Patient 4: (A) Noncontrast head computed tomographic scan showing an intracerebral hemorrhage. Cerebral angiogram showing a convexity osteodural arteriovenous fistula with feeders from the bilateral occipital artery, middle meningeal trunk (B,C), and right posterior cerebral artery (D). Venous phase of cerebral angiogram showing partial narrowing of the superior sagittal sinus and with venous congestion on bilateral internal carotid artery injections (E,F). Borden type II/III fistula, including 7 DCFs. Twenty-six (21 by primary embolization; 4 in the embolization and surgery group; 1 by surgery) of 32 (81%) patients had their fistulas completely obliterated after multimodality treatments at 1 to 36 months of follow-up (mean, 19.5 months). Only three of six type I fistulas (in which the goal was palliation) were obliterated. Clinical Outcome Twenty-four of 32 patients (75%) (including all those with DCF) had a favorable outcome (modified Rankin scale 0 2) during the follow-up period. Five patients were disabled (modified Rankin scale 3 5), and three patients died during the study period. Of the five patients with modified Rankin scale 3 5, the cause for the poor outcome was the initial insult of the hemorrhage in three patients, recurrent intractable seizures in one, and transient cognitive deficits with word-finding difficulty in another patient. Two patients died during their admissions: one due to the initial insult of the hemorrhage and another due to hepatorenal failure secondary to alcoholic cirrhosis. Another patient died 2 months after discharge as a result of traumatic subdural hematoma. There was no procedure-related mortality in our series. Complications Procedural complications were noted in three patients treated by endovascular techniques. The microcatheter was glued inside the artery after embolization of the fistula and was left in place in 2 of 24 (8%) patients after Onyx embolization without any dangerous consequence. Embolization of the feeding vessels of the trigeminal and facial nerves (petrous branch of the MMA) associated with Onyx reflux oc WORLD NEUROSURGERY, DOI: /j.wneu

7 Figure 11. Patient 5: Cerebral angiogram with selective left occipital artery injection (A). Postembolization angiogram showing complete obliteration of the fistula (B,C) and a plain anteroposterior x-ray of the neck showing the retained stuck catheter in the common carotid artery (D). curred in one patient with hemifacial hypesthesia and hemifacial palsy. The resultant damage to cranial nerves V and VII was permanent and did not show any improvement on follow-up at 28 months. In our study, the overall complication rate was 18.8% (6 of 32 patients), although only one (3.2%) patient developed a permanent neurologic deficit. There were no long-term complications related to surgery or radiosurgery. Illustrative Cases Case 1: Surgical obliteration of anterior fossa DAVF (type III). This 61-year-old man presented with severe headache (Hunt-Hess grade I) and had subarachnoid hemorrhage on CT scan (Figure 1A). Cerebral angiogram showed an anterior fossa DAVF (Borden type III) with anterior meningeal feeders from the ophthalmic artery and venous drainage into a frontal leptomeningeal vein that drained into the superior sagittal sinus Figure 10. Patient 5: (A) Noncontrast head computed tomographic scan showing posterior fossa hemorrhage. Cerebral angiogram showing a infratentorial cortical dural arteriovenous fistula supplied by feeders from bilateral occipital artery (B,D) and left posterior meningeal branches. The venous drainage was into a cerebellar vein that had a venous aneurysm and drained into the transverse sinus (C). Figure 12. Patient 6: (A,B) Computed tomographic scan showing left occipital hematoma before and after evacuation. WORLD NEUROSURGERY 73 [4]: , APRIL

8 ward the torcula and controlateral transverse sinus (Figures 3 and 4). The patient s symptoms were disabling enough to warrant treatment. Transarterial embolization of multiple ECA feeders with Onyx was performed in five separate sessions during 3 months and this relieved the symptoms by reducing the shunt through the fistulae (Figures 4 and 5). There were no complications or cranial nerve palsies. The remnant fistula has been treated by two sessions of Gamma Knife radiosurgery. Her tinnitus has significantly decreased. Follow-up angiography after 1 year has shown a considerable diminution of the dural arteriovenous fistula (AVF), but a new cortical venous drainage (into the vein of Labbé) from the fistula, which was now located predominantly around the superior petrosal sinus (conversion from a type I to a type III fistula). Due to a thrombosis of one jugular vein, some stenosis of the other, and the persisting fistula, the venous drainage of the head was significant through the peterygoid venous plexus and orbital veins. A further embolization procedure was recommended to the patient, but because of significant symptomatic improvement, the patient has elected to have repeat angiography after 3 months, before making a decision. Case 3: Transvenous embolization of DCF. This 49-year-old man presented with headache and proptosis and was found to have a DCF (Borden type II) with bilateral meningohypophyseal trunk feeders and left ECA feeders draining into the cavernous sinus, superior ophthalmic vein, and inferior petrosal sinus (Figure 6). Complete obliteration of the fistula and relief of all symptoms was achieved after transvenous coil embolization of the cavernous sinus through the left superior petrosal sinus (Figure 7). Figure 13. Patient 6: (A,B) Angiography showing meningeal feeders with venous aneurysm close to the transverse sinus. Postembolization showing only partial obliteration due to extravasation of Onyx through divided middle meningeal artery. (C) Extravasation through a middle meningeal branch divided at surgery. (D) Complete excision after surgery. (Figure 1B). Surgical obliteration of the origin of the draining vein was done by a fronto-orbital approach. An angiogram performed 8 months later showed complete obliteration of the fistula (Figure 2). Case 2: Gamma Knife radiosurgery for palliation of a type I fistula. This 32-year-old woman presented with severe headache and disabling tinnitus in the left ear and a petrotentorial Borden type I AVF supplied by multiple ECA feeders, including the occipital artery, posterior auricular artery, middle meningeal trunk, ascending pharyngeal artery, internal maxillary artery, and superficial and deep temporal arteries. The venous drainage was into the superior petrosal sinus and transverse sigmoid junction with antegrade flow toward the jugular vein and retrograde flow into the transverse sinus to- Case 4: Onyx embolization of convexity DAVF (Borden type III). This 60-year-old man presented with headache and was found to have an intracerebral hemorrhage (Figure 8A). Cerebral angiogram showed a convexity osteodural AVF with feeders from the bilateral occipital artery, middle meningeal trunk, and right posterior cerebral artery (Figure 8B,C,D). The venous drainage was predominantly into the cortical vein and into the middle third of superior sagittal sinus. There was partial narrowing of the superior sagittal sinus and venous congestion on bilateral ICA injections (Figure 8E,F). Transarterial embolization of Onyx was performed in three sessions through the bilateral middle meningeal trunks and right occipital artery. No complications or cranial nerve palsies occurred, and the fistula was completely obliterated (Figure 9). Case 5: Onyx embolization of tentorial DAVF (Borden type III). This 85-year-old woman presented with headache and was found to have a posterior fossa hemorrhage on CT scan (Figure 10A). Cerebral angiogram showed a infratentorial cortical vein dural AVF supplied by feeders from the bilateral occipital artery and left posterior meningeal branches. The venous drainage was into a cerebellar vein that had a venous aneurysm and drained into the transverse sinus (Figure 10B,C,D). Compete obliteration of the WORLD NEUROSURGERY, DOI: /j.wneu

9 Table 1. Review of Literature for Multimodality Treatment Results in DAVFs Before and After Onyx Series (ref. no.) No. cases Modality of treatment Obliteration Complications (No. of patients) Pre-Onyx Era Barnwell et al.,1989 (4) 16 Embolization Surgery 75% Inadvertent venous occlusion (2); Hydrocephalus (1), treated by shunt placement Lewis et al., 1994 (35) 8 Embolization Radiosurgery 43% Brainstem radiation injury (1) Duffau et al., 1999 (18) 20 Embolization Surgery Radiosurgery 75% None Kim et al., 2002 (27) 40 Embolization Surgery 37.5% Rebleed due to draining vein thrombosis (1) Koebbe et al., 2005 (29) 18 Embolization Radiosurgery 50% CN palsy (2) Onyx Era Nogueira et al., 2008 (46) 12 Embolization 91.6% None Lv et al., 2008 (38) 31 Embolization 61.3% CN palsy (5), microcatheter gluing (1); posterior fossa infarction (1); trigeminocardiac reflex (2) Cognard et al., 2008 (12) 30 Embolization Surgery Radiosurgery 87% Transient CN deficit (1) Rebleed due to draining vein thrombosis (1) Chew et al., 2009 (8) 12 Embolization Surgery 83.3% None Our study 32 Embolization Surgery Radiosurgery 81% CN deficit (1) Microcatheter gluing (2) CN, cranial nerve; DAVF, dural arteriovenous fistula; No., number. fistula was achieved after transarterial embolization with Onyx through the left occipital artery (Figure 11A,B,C). The microcatheter stuck in the Onyx cast and fractured; a fragment of the distal microcatheter was left in the occipital artery down to the common carotid artery (Figure 11D). Case 6: Onyx embolization followed by surgery for tentorial DAVF (Borden type III). This 54- year-old woman presented with occipital intraparenchymal hemorrhage and had an operation for removal of blood clot (Figure 12A, B). At surgery, arterialized veins were seen and angiography revealed a DAVF located near the transverse sinus on the left side and predominantly fed by branches of the ECA, particularly the MMA, branches of the occipital artery, and the posterior meningeal artery arising from the ascending pharyngeal artery (Figure 13A,B). Embolization was attempted but not successful because of persistent extravasation of contrast through a middle meningeal branch that had been divided during the operation (Figure 13C). During re-exploration of the occipital craniotomy, she underwent surgical occlusion of multiple arterialized veins draining into the transverse sinus with complete excision of the fistula (Figure 13D). The patient did not recover from her initial hemorrhage, and care was withdrawn in accordance with the wishes of the family. DISCUSSION Etiology, Clinical Features, and DAVF Location The pathogenesis of DAVFs is controversial, with a congenital and an acquired etiology having been proposed (34). Lasjaunias et al. (34) proposed that a primary structural weakness of the dura coincides with a trigger factor and results in the formation of a DAVF. The common predisposing factor appears to be venous sinus thrombosis, with the body compensating for this by attempts at recanalization (32). In our series, sinus occlusion was observed in two patients. There were 3 patients with multifocal complex infantile DAVFs; and in 27 patients, no specific cause could be identified. The DAVF is more commonly diagnosed in perimenopausal women, suggesting an underlying hormonal influence; however, it tends to have a more aggressive course in men (31, 44, 62). The demographics in our series reflected this finding in that Figure 14. Transvenous obliteration of a fistula draining into the right transverse sigmoid junction with the two-catheter technique. (A,B) Anteroposterior and lateral views of placement of two catheters in the sinus, respectively. WORLD NEUROSURGERY 73 [4]: , APRIL

10 segment of the sinus that has the dural arteriovenous fistula. In these patients, transarterial and surgical skeletonization are the only two options. Rarely, it may be possible to gain access to the compartmentalized segment through the transvenous route. (C) The figures on the right (above and below) show type III fistulae without and with venous aneurysms. The fistula in most of these patients is not close to the sinus and hence sinus occlusion cannot cure the fistula. Rarely, the fistula may be very close to the sinus, but proximal on a small short vein draining into the sinus (left image). It is very important to recognize this type III fistula, as these can easily be mistaken for a type I or II fistula. Sinus occlusions in these patients can be very dangerous. Figure 15. Borden types I (A), II(B), III (C) arteriovenous fistulas with different types of vascular anatomy and drainage patterns, and the preferred choice of treatment modalities for each type. (A) The left image shows a case where there could be flow stagnation in the sinus, but the flow is still antegrade and there could be normal brain parenchyma that is dependent for its venous drainage on that sinus. This is very important to recognize because occluding this sinus by the transvenous route could be dangerous. On the other hand, if there is retrograde flow in the sinus (right image), transvenous occlusion is the first choice for treating the fistula. (B) In these figures, the right and center image replicate the same physiology as in A, except that these patients also have cortical venous reflux. The options remain the same as in A for these patients. The left image is a type where there is compartmentalization or isolation of the 75% of the patients presenting with hemorrhage were men. The incidence of DAVF by location, as reported in the literature, is as follows: transverse sigmoid sinus, 50%; cavernous sinus, 16%; tentorium, 12%; superior sagittal sinus, 8%, and anterior fossa, 2% (3, 6, 10, 28). Classification From a clinical viewpoint, it is useful to classify DAVF according to their venous drainage pattern (17). We used the Borden classification (6) in our series. Certain angiographic features of these lesions are associated with aggressive behavior. These include CVR, variceal or aneurysmal venous structures, thrombosis of venous outflow, distant location from major sinus, and galenic venous drainage (3, 41, 54). A patient s first hemorrhage from a DAVF is associated with 20% to 30% mortality, emphasizing the importance of early recognition and treatment of patients at highest risk (7). Natural History The natural history of DAVFs has been poorly documented until recently and has been based on retrospective reviews (3, 10). Prospective series (15, 16, 64) support the need for active and curative treatment of DAVF when associated with leptomeningeal or CVR and conservative management in those DAVF that do not have such venous drainage. The majorityofourpatients(82%) belongedtothe high-risk category (Borden types II or III) and required active intervention to prevent risk of hemorrhage. Recent studies demonstrate evidence that the risk of bleeding for a DAVF with CVR is less when the patient does not present with a hemorrhage or a nonhemorrhagic neurologic deficit (56, 57). Strom et al. (57) report that asymptomatic versus symptomatic DAVFs (in patients presenting with hemorrhage or neurologic deficit) have annual hemorrhage rates of 1.4% versus 19%, respectively. Although small in size, these studies make the important point that the naturalhistorymaydependonthetypeofpresentation. Nevertheless, lower annual hemorrhage rates of asymptomatic type II and III DAVFs (up to 1.5% per annum) (56) still pose significant long-term risk to patients, thereby justifying treatment in appropriately selected patients. At our institution, we favor aggressive treatment of all symptomatic type II/III WORLD NEUROSURGERY, DOI: /j.wneu

11 Figure 16. Treatment algorithm followed at our institution. DAVF, dural arteriovenous fistula; ICA, internal carotid artery; EC-IC, extracranial-intracranial; ACF, anterior cranial fossa fistula. DAVFs and recommend treatment in asymptomatic type II/III DAVFs if the patient is a good candidate on the basis of physiological age, lifetime hemorrhage risk profile, and willingness or unwillingness to accept the natural history. Management decision-making for patients in the latter group (those with asymptomatic type II/III DAVFs) is not unlike that for patients with unruptured intracranial aneurysms. Multimodality Management Strategies Type II and III DAVFs have a high risk of hemorrhage, and the goal of treatment in these patients is cure. Type I DAVFs have a very low risk of hemorrhage, and patients with these lesions are normally observed without treatment unless they have disabling symptoms like tinnitus or develop a new neurologic deficit or CVR during follow-up. The therapeutic armamentarium includes conservative monitoring, arterial embolization, transvenous occlusion, surgical excision, and radiation therapy. In the majority of patients, a combination of treatment offers the best chance for success. Several series confirm the importance of a multimodality approach for the treatment of DAVFs (Table 1) (4, 5, 8, 12, 19, 20, 27, 29, 30, 35, 39, 46, 47, 50). Transarterial Embolization Several series have now established a high degree of safety and efficacy of Onyx in the treatment of intracranial DAVFs (2, 8, 12, 46, 48, 52, 61). Due to the properties and increased predictability of Onyx, prolonged injections through a single feeder can be performed, with retrograde embolization of multiple feeders, and the catheter can be positioned in the feeder more proximally than is possible with N-butyl cyanoacrylate. This allows for more controlled injections with better distal nidus or fistula penetration, compared with cyanoacrylate. We found that it was far more efficient to gain access through the MMA, even when the diameter was very small, than through other more dilated feeders (eg, occipital and posterior auricular arteries) due to the superior flow control and reliability of target vessel embolization. The petrous branch of the middle meningeal trunk is often the main supply for DAVFs in the petrotentorial region. It also supplies the cranial nerves at the skull base. We avoid this branch and select another distal tentorial branch to embolize the fistula, even if that is not a direct feeder. The patient who had permanent cranial nerve palsy had embolization through the petrous branch, and we have since tried to avoid embolizing this branch. With ECA branch embolization, the possibility of anastomosis to the intracranial circulation, either through the DAVF itself or along pre-existing collateral vessels (eg, C1 occipital to VA), must be considered, even if such a connection is not visualized on the initial angiogram, as small collaterals may only become apparent after the embolization is initiated and changes in hemodynamic occur. Onyx offers the possibility of venous sinus packing/occlusion from an arterial side approach, which may be quite helpful in cases with previous occlusion of the draining venous structures. Our success rate of 70% compares favorably with published results of 50% to 90% (8, 9, 12, 46). Our complication rate was 9.4%, although the rate of permanent neurologic deficit was only 3.4% (1 patient). Postembolization complication rates of approximately 2% to 10% have been reported in most series (23, 24, 38, 45). In our series we had four patients who has difficulty in catheter retrieval after Onyx embolization. This has been previously reported with Onyx in arteriovenous malformations (43) and may be related to the formation of an Onyx plug and the reflux of Onyx that occurs around the catheter during embolization. Avoiding longer reflux distance around the catheters and selecting pedicles that are not tortuous and straight could decrease the chances of a stuck catheter. The use of microcatheters with detachable tips (Sonic, BALT, France) could allow leaving behind the glued catheter tip in the nidus. Transvenous Embolization The transvenous approach may be difficult or even impossible if the affected sinus is stenosed, compartmentalized, or even occluded downstream or upstream of the fistula. Recanalization of an occluded sinus is sometimes possible. There is a chance of isolating a segment of the sinus communicating with the fistula when trying to occlude a large venous sinus. This can lead to conversion of a DAVF into a worse grade with higher chance for cortical venous reflux and hemorrhage. One of the authors (BG) prefers to use the double-catheter technique in which two catheters are placed in the sinus and embolization is done through the more proximal catheter. WORLD NEUROSURGERY 73 [4]: , APRIL

12 Table 2. Summary of Individual Patient Demographics, Treatment, and Outcome Data Pt no Age (year) Sex Borden type Location Hemorrhage on presentation Treatment modality Endovascular route Complications Angio follow-up (months) Angio outcome Clinical follow-up (months) Clinical outcome (mrs) 1 36 F I TS Nil EV TA Nil 18 CO F I TS-Torcula Nil EV TA Microcatheter 30 CO 30 2 detached in OA 3 30 F I TS Nil EV TA Nil Nil CO Nil M II CCF Nil EV TV Nil 2 CO F II CCF Nil EV TV Nil 2 CO F II CCF Nil EV TV Nil 40 CO M II CCF Nil EV TA/TV Nil Nil CO M II CCF Nil EV TV Nil 6 CO F II CCF Nil EV TV Nil 2 CO F II CCF Nil EV TV Nil 18 CO F II TS Nil EV TA Nil 7 CO F II TS IPH EV TA Nil 18 CO M II TS IPH EV TA Nil 9 R M II Parasagittal IVH EV TA Stuck catheter 26 R M III Tentorial SAH EV TA Nil 2 CO F III Tentorial IPH (cerebellar) EV TA Stuck catheter with 26 CO 26 0 distal fracture M III Parasagittal SAH/IPH EV TA Nil 11 CO F III Middle fossa Nil EV TA/TV Nil 6 CO M III Tentorial IPH EV TA Nil 6 CO 6 0 (frontoparietal) M III Tentorial IPH/IVH EV TA Nil 18 CO 40 1 (occipitotemporal) M III Tentorial IPH EV TA Nil 5 CO F III Tentorial IPH (occipital) EV TA Nil 7 CO M III Middle fossa Nil EV TA Nil 8 CO F I TS Nil EV GK TA/TV Nil 7 R M I TS Nil EV GK TA Nil 12 R F I TS Nil GK Nil Nil 7 R F II TS-Torcula Nil EV S TA Nil 6 R M III Ant fossa SAH S Nil Nil 7 CO M II TS Nil EV S TA Nil 12 CO F III Ant Falcine Nil EV S TA Status epilepticus 10 CO F III Tentorial IPH (occipital) EV S TA Fractured catheter tip Nil CO Nil M III Middle fossa Nil EV S TA CN palsies V/VII 9 CO 9 3 angio, angiographic; Ant, anterior; CCF, carotid cavernous fistula; CN, cranial nerve; CO, complete obliteration; EV, endovascular; GK, Gamma knife; IPH, intraparenchymal hemorrhage; IVH, intraventricular hemorrhage; mrs, modified Rankin score; Nil, no complication or hemorrhage; no, number; OA, ophthalmic artery; R, residual; S, surgery; SAH, subarachnoid hemorrhage; TA, transarterial; TS, transverse sigmoid; TV, transvenous. Isolated sinus pockets can then be occluded with liquid embolic agents delivered through the distally placed catheter (Figure 14). Placement of embolic agents into a fistulous segment of a sinus that is being used as a conduit for venous drainage of the remaining brain carries a high risk of venous infarction or hemorrhage. Hence, it is imperative to clearly outline the venous out WORLD NEUROSURGERY, DOI: /j.wneu

13 flow of the normal brain parenchyma before the procedure (28, 53). A transvenous approach with coil occlusion of the affected sinus is an option only when the brain is well drained by other venous channels (42). Transvenous treatment of DAVFs with angioplasty and stent deployment within the involved dural sinus has also been proposed (36). Transvenous coil embolization was done in six patients with Guglielmi detachable coils in our series and is a preferred option when the main arterial pedicles in the DAVF originate from the ICA or VA, potential extracranial intracranial collateral anastomosis exists, or when the risk of occlusion of nutrient vessels of cranial nerves is high, and also in DCF (59). Surgery Surgery is preferred for high-risk DAVFs with leptomeningeal retrograde venous drainage when embolization is technically difficult or has resulted in incomplete occlusion. Hoh et al. (25) described a technique of clipping the draining vein together with extensive cauterization of the surrounding dura. Surgical strategies must be adapted to the type of venous drainage (13, 63). In cases of sinus drainage with cortical venous reflux (type II), transarterial embolization combined with surgical excision of the fistulous sinus segment or direct surgical sinus packing (if the sinus is nonfunctional) may represent a more definitive treatment option (14). The DAVFs with cortical venous drainage alone without dural sinus involvement (type III) are best treated by interruption of the draining vein close to the fistula (49, 63). There are a few reports of the use of Onyx for embolization of anterior fossa DAVFs (1, 40, 60), although these particular lesions are easily obliterated by surgery. In our series, an open surgical procedure was performed to treat one patient with an anterior fossa DAVF supplied by the ophthalmic artery and resulted in a cure. Another patient required surgical packing of the sinus along with excision of the fistula, after embolization. Gamma Knife Radiosurgery Radiosurgery is inefficient for DAVFs, and the delay for obliteration carries a risk of bleeding, rebleeding (20, 29, 55), or conversion to a more dangerous type of lesion, as was seen in our series (11, 54). Nevertheless, radiosurgery may be tried where embolization has failed and surgery appears to carry a high risk. Radiosurgery is a good option in benign residual DAVFs (37). Wu et al. (65) reported a 75% angiographic obliteration rate at 24 months. In our series Gamma Knife radiosurgery was used for palliation in three patients with complex type I fistula. The goal of treatment in these type I fistulas is palliation and not immediate obliteration after partial endovascular obliteration, but the long-term results are awaited. Treatment Philosophy and Algorithm Our experience with Onyx has led to our current algorithm for the management of DAVFs. Figures 15 and 16 illustrate the algorithm used for deciding when and how a lesion needs to be treated. Table 2 summarizes the patient, treatment, and outcome characteristics. Limitations Limitations of this study are that it is a retrospective review. This is a single center experience, and the treating physicians reviewed the outcomes. There is definitely a need for longer clinical follow-up to ensure that there is no clinical or radiologic recurrence. Outcome in the radiosurgical group is still awaited, and this modality has not been evaluated in this series. In conclusion, our results have shown high success rates for cure with a multimodality approach for the treatment of DAVFs. At our institute, endovascular therapy was effective in achieving a durable angiographic cure and clinical resolution of symptoms. Transarterial embolization with Onyx is the primary treatment for intracranial DAVFs with high safety profile and efficacy. Transvenous coil embolization is still preferred in DAVFs that receive supply from arterial branches that supply cranial nerves, predominant ICA feeders, and potential extracranial intracranial collateral anastomosis. Incompletely treated DAVFs or anterior fossa DAVFs can be cured with surgery and partially treated type I fistulas can be treated by radiosurgery, if surgery is not an option. REFERENCES 1. Abrahams JM, Bagley LJ, Flamm ES, Hurst RW, Sinson GP: Alternative management considerations for ethmoidal dural arteriovenous fistulas. Surg Neurol 58: (discussion: 416), Arat A, Inci S: Treatment of a superior sagittal sinus dural arteriovenous fistula with Onyx: Technical case report. Neurosurgery 59:ONSE (discussion: ONSE170), Awad IA, Little JR, Akarawi WP, Ahl J: Intracranial dural arteriovenous malformations: Factors predisposing to an aggressive neurological course. J Neurosurg 72: , Barnwell SL, Halbach VV, Dowd CF, Higashida RT, Hieshima GB, Wilson CB: A variant of arteriovenous fistulas within the wall of dural sinuses. Results of combined surgical and endovascular therapy. J Neurosurg 74: , Barnwell SL, Halbach VV, Higashida RT, Hieshima G, Wilson CB: Complex dural arteriovenous fistulas. Results of combined endovascular and neurosurgical treatment in 16 patients. J Neurosurg 71: , Borden JA, Wu JK, Shucart WA: A proposed classification for spinal and cranial dural arteriovenous fistulous malformations and implications for treatment. J Neurosurg 82: , Brown RD Jr, Wiebers DO, Nichols DA: Intracranial dural arteriovenous fistulae: Angiographic predictors of intracranial hemorrhage and clinical outcome in nonsurgical patients. J Neurosurg 81: , Chew J, Weill A, Guilbert F, Raymond J, Audet ME, Roy D: Arterial Onyx embolisation of intracranial DAVFs with cortical venous drainage. Can J Neurol Sci 36: , Chong WKW, Holt M: Endovascular therapy for intracranial dural arteriovenous fistulas. Neuroradiol J 19: , Cognard C, Gobin YP, Pierot L, Bailly AL, Houdart E, Casasco A, Chiras J, Merland JJ: Cerebral dural arteriovenous fistulas: Clinical and angiographic correlation with a revised classification of venous drainage. Radiology 194: , Cognard C, Houdart E, Casasco A, Gabrillargues J, Chiras J, Merland JJ: Long-term changes in intracranial dural arteriovenous fistulae leading to worsening in the type of venous drainage. Neuroradiology 39:59-66, Cognard C, Januel AC, Silva NA Jr, Tall P: Endovascular treatment of intracranial dural arteriovenous fistulas with cortical venous drainage: New management using Onyx. AJNR Am J Neuroradiol 29: , Collice M, D Aliberti G, Arena O, Solaini C, Fontana RA, Talamonti G: Surgical treatment of intracranial dural arteriovenous fistulae: Role of venous drainage. Neurosurgery 47:66-67, (discussion: 66-57), WORLD NEUROSURGERY 73 [4]: , APRIL