- Email: [email protected]
Radiosurgery for dural arteriovenous malformations
Handbook of Clinical Neurology, Vol. 143 (3rd series) Arteriovenous and Cavernous Malformations R.F. Spetzler, K. Moon, and R.O. Almefty, Editors http://dx.doi.org/10.1016/B978-0-444-63640-9.00012-6 © 2017 Elsevier B.V. All rights reserved
Chapter 12
Radiosurgery for dural arteriovenous malformations CONOR GRADY1, CAROLINA GESTEIRA BENJAMIN1, AND DOUGLAS KONDZIOLKA1,2* 1 Department of Neurosurgery, NYU Langone Medical Center, New York, NY, USA 2
Center for Advanced Radiosurgery, NYU Langone Medical Center, New York, NY, USA
Abstract Intracranial dural arteriovenous malformations (DAVFs) are relatively uncommon vascular lesions characterized by the direct connection of dural arteries into dural venous sinuses or leptomeningeal veins. Strategies for the treatment of these complex lesions have evolved significantly over the past three decades, and include open surgical disconnection, endovascular embolization, stereotactic radiosurgery (SRS), or a combination of these approaches. Radiosurgical intervention is unique in offering significant benefits to patients while exposing them to few of the risks associated with more invasive interventions. In this chapter we provide an overview of DAVFs and discuss the features of these lesions that affect management. We focus, in particular, on radiosurgical management of these lesions, describing present treatment paradigms, the procedure for the treatment of DAVFs with SRS, and expected clinical outcomes using SRS.
INTRODUCTION
PATHOGENESIS
Intracranial dural arteriovenous fistulas (DAVFs) are relatively uncommon vascular lesions characterized by the direct connection of dural arteries into dural venous sinuses or leptomeningeal veins. They are believed to constitute 10–15% of all intracranial vascular lesions (Houser et al., 1972), and are distinct from other intracranial arteriovenous malformations (AVMs) in their pathogenesis and presentation. Strategies for the treatment of these complex lesions have evolved significantly over the past three decades, and include open surgical disconnection, endovascular embolization, stereotactic radiosurgery (SRS), or a combination of these approaches. Radiosurgical intervention is unique in offering significant benefits to patients while exposing them to few of the risks associated with more invasive interventions. In this chapter we will provide an overview of DAVFs and discuss the features of these lesions that affect treatment. We will focus, in particular, on radiosurgical management of these lesions, and expected clinical outcomes using SRS.
In contrast to the majority of intracranial AVMs, DAVFs are primarily believed to be acquired lesions (Houser et al., 1979; Chaudhary et al., 1982; Nishijima et al., 1992). Over the past four decades, clinical observation, histologic examination of resected specimens, and animal models have led to multiple hypotheses of the cardinal event leading to their formation. Houser et al. (1979) were amongst the first to define the angiographic characteristics of these lesions, and noted the relationship of their formation to the presence of dural sinus thrombosis. Subsequent studies continued to support the association of DAVF with sinus thrombosis, though controversy existed regarding the etiologic relationship of sinus thrombosis to fistula formation (Chaudhary et al., 1982; Kutluk et al., 1991). The present leading hypothesis of DAVF formation posits a repetitive cycle of anatomic and molecular events triggered by venous hypertension. Venous hypertension itself may result from any number of factors, including, but not limited to, trauma and venous
*Correspondence to: Douglas Kondziolka, MD, MSc, FRCSC, FACS, Center for Advanced Radiosurgery, NYU Langone Medical Center, 530 First Avenue, Suite 8R, New York NY 10016, USA. Tel: +1-646-501-2360, E-mail: [email protected]
126
C. GRADY ET AL.
thrombosis (Nishijima et al., 1992). Anatomically, venous hypertension leads to dilation of dural veins and enlargement of microscopic arteriovenous shunts normally present in dura. The enlarging of these shunts further contributes to venous hypertension, as increased arterial flow reaches new venous territories. Blood flow through these microscopic fistulae become increasingly turbulent as they enlarge, leading to stasis and promoting thrombosis, further contributing to venous hypertension. The outcome of this cascade of events has been demonstrated by histologic examination of resected DAVFs, which reveals dilation and thickening of microscopic arterial and venous vessels present in the dura adjacent to fistula sites, as well as partially canalized sinus thromboses (Houser et al., 1979; Nishijima et al., 1992). This theory has been additionally supported by animal studies in which both carotid to external jugular anastomoses and experimentally induced sinus thromboses have first resulted in venous hypertension, followed by the formation of DAVFs (Terada et al., 1994; Herman et al., 1995). On a molecular level, venous hypertension is believed to cause local ischemia and endothelial injury, leading to the increased expression of angiogenic growth factors which contribute to the histologic findings described above. Immunohistochemical analysis of resected DAVFs has demonstrated markedly increased expression of basic fibroblast growth factor and vascular endothelial growth factor (Uranishi et al., 1999). Lawton et al. (1997) also successfully demonstrated increased expression of angiogenic growth factors and DAVF formation in a rat model of venous hypertension.
PRESENTATION DAVFs present with a variety of symptoms which are related to the anatomic location of the fistula itself. The majority of DAVFs present with local symptoms arising from altered hemodynamics. In one of the early large case series looking at the presenting symptoms of DAVFs, Lasjaunias et al. (1986) reported that 83% of patients with cavernous sinus DAVFs presented with proptosis or visual changes, while 70% of patients with torcular, transverse sinus, or sigmoid sinus DAVFs presented with an audible bruit. More recently, Gross et al. (2016), in a retrospective review of a prospectively maintained database of 260 patients with DAVFs, found that 92% of patient with cavernous sinus lesions presented with ocular findings, while 75% of patients with transverse sinus, sigmoid sinus, marginal sinus, or torcular DAVFs presented with pulsatile tinnitus. DAVFs can also present with more malignant symptoms, such as seizure, neurologic deficit, or hemorrhage. Fistula location is again one of the factors correlated with the risk for the development of these symptoms. Awad
et al. (1990) observed that only 10–20% of cavernous or transverse sigmoid sinus DAVFs presented with hemorrhage, while greater than 90% of tentorial DAVFs had hemorrhage at presentation. The discrepancy of risk for hemorrhage of fistulae in different locations has been further illustrated by more recent case series, which have reported a 6–11% risk of hemorrhage as the presenting symptom for DAVFs of the cavernous and transverse sigmoid sinuses compared to a 54–79% risk for presentation with hemorrhage in DAVFs of the tentorium or petrosal sinuses (Singh et al., 2008; Gross et al., 2016). Although this indicates that there is a correlation between the location of the fistula and the likelihood of presenting with hemorrhage or other malignant symptoms, it is important to note that fistulae at any location may exhibit aggressive behavior. Aside from location, other factors have been identified as predictors of presentation with hemorrhage. In a review of 402 patients with DAVFs, Singh et al. (2008) identified cortical venous drainage (CVD), focal neurologic deficit, male sex, location in the posterior fossa, and age greater than 50 at presentation as independent risk factors for hemorrhage. Of note, amongst these risk factors, the presence of CVD was the strongest predictor of presentation with hemorrhage, with an odds ratio of 10.5, more than twice the odds ratio of the next highest predictor amongst the identified risk factors. The unique role of CVD in determining risk of hemorrhage at presentation was further elucidated by Gross and Du (2012), who identified male sex and tentorial or petrous location of fistulae as strongly associated with the presence of cortical drainage. These studies reinforce the findings of previous case series and literature reviews (Awad et al., 1990), and lend support to the long-accepted practice of sorting DAVFs into high- or low-risk categories based on the presence of CVD.
CLASSIFICATION Given its significance in determining risk for malignant presentation, multiple classification schemes have been proposed for DAVFs, categorizing lesions primarily based on angiographic evidence of CVD. The two most widely applied of these criteria were formulated by Borden et al. (1995) and Cognard et al. (1995). The Borden classification system divides lesions into three categories based on the type of arteriovenous anastomosis (i.e., arterial drainage into a dural sinus or vein versus direct anastomosis with a cortical vein) as well as the presence of CVD. Borden type I lesions are a direct anastomosis between a meningeal artery and a dural sinus or vein with purely anterograde drainage. Type II lesions are also a result of anastomosis between a meningeal artery and a dural vein or sinus, but with reversal
RADIOSURGERY FOR DURAL ARTERIOVENOUS MALFORMATIONS of flow and reflux into cortical veins. Type III lesions occur when meningeal arteries directly drain into cortical veins. Type I lesions are considered benign, and their spontaneous remission has been well documented (Newton and Hoyt, 1970; Vinuela et al., 1984; Kutluk et al., 1991). Both type II and type III lesions are considered aggressive, and warrant evaluation for intervention. The classification system proposed by Cognard and colleagues (1995) takes several other features of a DAVF into account. Based on a retrospective analysis of a single center’s experience with 205 patients presenting with DAVF, the Cognard system divides lesions into five categories, corresponding with their risk for malignant presentation. Type I lesions drain directly into a dural sinus, with only anterograde flow. Type II fistulae drain into dural sinuses with resulting retrograde flow, and are subdivided based on whether the retrograde flow involves only dural sinuses (subtype a), only cortical veins (subtype b), or involves both sinuses and cortical veins (subtype a + b). Type III lesions are direct anastomosis of meningeal arteries with cortical veins without ectasia, while type IV lesions are direct cortical vein anastomoses with ectasia. Type V lesions drain directly into the spinal venous system. As in the Borden classification system, those types featuring CVD (i.e., types IIb–V) are considered to be aggressive and should be considered for intervention. Though amenable to classification by the Borden and Cognard systems, DAVFs involving the cavernous sinus can additionally be classified by the Barrow system (Barrow et al., 1985). In this system, type A lesions are direct, high-flow fistulae between the internal carotid artery and the caverous sinus. Type B, C, and D lesions all represent low-flow fistulous connections between dural arteries and the cavernous sinus, and are distinguished by whether the dural vessels arise from the internal (type B), external (type C), or both (type D) carotid arteries. Barrow type A fistulae are traumatic lesions which are distinct from other forms of DAVFs, and their natural history and management are beyond the scope of this chapter. Type B–D lesions are analogous to DAVF occurring at other locations.
NATURAL HISTORY Attempts to precisely characterize the natural history of DAVFs have been hindered by the relatively low incidence of these lesions. A prospective, populationbased cohort estimated a detection rate of only 0.16 per 100 000 adults per year (Al-Shahi et al., 2003). Consequently, a recent literature review and meta-analysis discovered that all literature describing the natural history of DAVFs is comprised of small, hospital-based cohorts, the vast majority of which are retrospective in
127
design, and which seldom account for the treatment of lesions that are symptomatic at presentation when reporting subsequent outcomes (Jolink et al., 2015). In the absence of high-quality data, quoted risks of morbidity and mortality vary widely. Brown et al. (1994), in reviewing the records of 54 patients with untreated DAVF with average follow-up of 6.6 years, found that 5 patients had a DAVF-related hemorrhage, resulting in an overall 1.8% per year risk of hemorrhage. This is similar to the annual risk of rupture of 1.5% reported by Soderman et al. (2008), in their evaluation of 53 patients presenting with unruptured DAVFs. In the same study, patients presenting with hemorrhage were calculated to have a 7.4% per year risk of re-hemorrhage. This is significantly lower, however, than the 20–35% re-hemorrhage rate demonstrated by Duffau et al. (1999) in their review of 20 patients with previous hemorrhage from DAVF. Likewise, following 20 patients with long-term CVD, van Dijk et al. (2002) calculated an annual mortality rate of 10.4%, as well as a 15% annual rate of hemorrhage or focal neurologic deficit, rates significantly higher than reported in previous studies. Looking at an endpoint other than symptomatic presentation, Hetts et al. (2015), found that only 3.2% of 574 patients with angiogram-confirmed DAVF demonstrated progressive or de novo fistulae. Of note, these progressive lesions demonstrated 26%, 32%, and 47% risk for hemorrhage, focal neurologic deficit, or cranial neuropathy, respectively, during the period of the study. Taken in sum, the available literature on the natural history of DAVFs indicates that these lesions pose a variable risk to affected patients. In general, aggressive features, such as CVD or formation of de novo lesions, indicate that patients are at an increased risk for hemorrhage or significant neurologic symptoms compared to patients with stable lesions not featuring CVD.
RADIOSURGERY FOR THE MANAGEMENT OF DAVFS Symptomatic lesions or high-grade lesions at increased risk for malignant presentation warrant intervention. Although endovascular embolization, surgical disconnection, or radiosurgery can each be used alone, singlemodality treatment may not be effective, and thus many DAVFs require multimodality approaches (Kobayashi and Al-Shahi Shalman, 2014). Endovascular embolization, especially via the transvenous route, has emerged as the most commonly employed solitary treatment of DAVFs. However, studies have shown that embolization alone is not sufficient for complete obliteration of the fistula in 50% of patients (Yang et al., 2013). This is largely due to limitation in visualizing all the feeding
128
C. GRADY ET AL.
vessel with angiography and the technical challenges of superselective catheterization of small feeding vessels. Therefore it is frequently combined with radiosurgery for higher obliterations rates (Kobayashi and Al-Shahi Salman, 2014). SRS can also be employed as a solitary approach, but is more frequently used as a valuable adjunct to endovascular or open approaches to complex lesions. Scenarios in which radiosurgery would be used as a single modality of therapy include patients whose lesions may not be amenable to other intervention, patients with significant medical comorbidities, or patients with low-grade, minimally symptomatic lesions. Similar to the experience with parenchymal AVMs, radiosurgery is expected to lead to progressive obliteration of DAVFs over 1–3 years following irradiation. Initial SRS followed by transarterial embolization either the same day or soon after has been shown to be an effective treatment paradigm, with embolization providing more immediate symptomatic relief and decreased flow through the lesion, and SRS providing long-term, durable closure with lower risk for recanalization (Yang et al., 2013). If treatment with both SRS and endovascular embolization is planned, SRS should be performed first to prevent embolization material from obscuring any portion of the nidus during planning and irradiation.
Procedure In the majority of cases cited in the literature, SRS for the treatment of DAVFs is performed using gamma knife technique (Brown et al., 2005; Chen et al., 2015). The procedure begins with the application of a Leksell stereotactic head frame under conscious sedation with local scalp anesthesia. Given the peripheral location of many noncavernous sinus DAVFs, eccentric head frame placement used to be critical to prevent patient collision with the collimator helmet. This, however, has become less of a concern with new models of gamma knife, which feature fully automated patient positioning via movement of the table on which the patient is introduced into the collimator. Once the head frame is applied, target identification is performed using stereotactic angiography, often requiring injections of bilateral internal and external carotid arteries as well as the bilateral vertebral arteries. High-resolution stereotactic magnetic resonance imaging is then performed to further aid target acquisition and to identify adjacent structures which must be spared from the dose plan. All imaging is exported to a computer workstation, where dose planning is performed. In the published experience of Yang et al. (2013), the median prescription dose delivered to the margin of DAVFs undergoing SRS was 20 Gy (range,
15–25 Gy). The median isodose volume was 2.0 cm3 (range 0.2–8.6 cm3), with prescription isodose of 50% in all but two cases, in which the prescription isodose was 60%. Median maximum dose was 40 Gy, with a range of 30–50 Gy. This is similar to the features of all series contained in a recent meta-analysis of outcomes in DAVFs treated by SRS; across all studies, reported median marginal doses ranged from 18 to 25 Gy, treatment volume ranged from 1 to 9.3 cm3, and prescription isodose from 40 to 60% (Chen et al., 2015). Likewise, a large, single-center cohort of 105 patients not included in the review employed a median radiation dose delivered to the margin of 19 Gy (range, 15–25 Gy) with a maximum dose of 38 Gy (range, 22–50 Gy) (Yang et al., 2013). Figures 12.1 and 12.2 provide example dose plans for representative cases.
OUTCOMES FOLLOWING RADIOSURGERY SRS provides reasonable rates of complete occlusion with low risk for serious adverse events. In their systematic review and meta-analysis of 19 studies examining the outcomes of a pooled total of 729 patients having undergone SRS for treatment of 743 DAVFs, Chen et al. (2015) reported a complete obliteration rate of 63%, while calculating post-SRS rates of hemorrhage, neurologic deficit, or death at last follow-up as 1.2%, 1.3%, and 0.3%, respectively. This review further subdivided outcomes based on the location, angioarchitecture, and Borden classification of the lesions treated, given the relationship of these features to risk for hemorrhage and neurologic deficit. By location, 73% of cavernous sinus DAVFs treated with SRS demonstrated complete occlusion at follow-up, compared to 58% of DAVFs occurring at other locations. No patients in any study with a cavernous sinus DAVF experienced post-SRS hemorrhage, while 1.3% of patients with noncavernous sinus locations did. Of patients with noncavernous sinus DAVFs, variability in outcome was seen based on Borden classification, with 67% of type I lesions experiencing complete occlusion, compared to 43% of type II lesions and 50% of type III lesions. The rate of post-SRS hemorrhage for patients with Borden type I was 0%, compared to 2.6% and 3.7% in patients with Borden type II and Borden type III lesions, respectively. The presence or absence of CVD, regardless of location or specific classification type, also demonstrated a relationship with outcome. Of lesions without CVD, 74.5% demonstrated complete occlusion on follow-up imaging, and none experienced post-SRS hemorrhage, compared to 56% occlusion and 4.6% post-SRS hemorrhage rates in lesions with CVD.
RADIOSURGERY FOR DURAL ARTERIOVENOUS MALFORMATIONS
129
Fig. 12.1. A 46-year-old woman with memory dysfunction found to have a falcotentorial dural arteriovenous fistula. Radiosurgery was performed with 20 Gy to be followed by embolization to reduce venous hypertension.
Fig. 12.2. A 36-year-old woman with a falcotentorial arteriovenous fistula associated with 2 years of headache. The radiosurgery plan was created to delivery 18 Gy to the 50% isodose margin.
A significant limitation in reporting outcomes for DAVFs treated by SRS is the previous or concurrent treatment of many patients with endovascular embolization. In the meta-analysis by Chen et al. (2015), at least
17% of all patients included in the analysis were reported to undergo endovascular intervention, while many of the studies included provided insufficient information to allow the authors to extract whether patients had
130
C. GRADY ET AL.
previous treatment or were undergoing SRS alone. A recent retrospective series by Park et al. (2016) helps to address this, following 20 patients with DAVFs treated with SRS alone. The study included eight symptomatic lesions without CVD, 11 lesions with CVD but no symptoms, and one “high-risk” lesion which had both CVD and symptoms, but which was not amenable to endovascular or open intervention. Follow-up with thin-cut time-of-flight magnetic resonance angiography revealed complete occlusion in 18 of 20 lesions at a mean of 29 months’ follow-up. All patients with symptomatic lesions experienced complete resolution of their symptoms, with a mean time of 3.5 months from SRS to symptom improvement.
CONCLUSION DAVFs are complex lesions that pose variable risks to the patients who harbor them. SRS is a safe and effective treatment modality for these lesions, both as a solitary treatment and in conjunction with endovascular embolization or open surgical disconnection. SRS is especially effective in lesions of the cavernous sinus, and in lesions without CVD.
REFERENCES Al-Shahi R, Bhattacharya J, Currie DG et al. (2003). Prospective, population-based detection of intracranial vascular malformations in adults: the Scottish Intracranial Vascular Malformation Study (SIVMS). Stroke 34: 1163–1169. Awad I, Little J, Akarawi W et al. (1990). Intracranial dural arteriovenous malformations: factors predisposing to an aggressive neurological course. J Neurosurg 72: 839–850. Barrow D, Spector R, Braun I et al. (1985). Classification and treatment of spontaneous carotid-cavernous sinus fistulas. J Neurosurg 62: 248–256. Borden J, Wu J, Shucart W (1995). A proposed classification for spinal and cranial dural arteriovenous fistulous malformations and implications for treatment. J Neurosurg 82: 166–179. Brown R, Wiebers D, Nichols D (1994). Intracranial dural arteriovenous fistulae: angiographic predictors of intracranial hemorrhage and clinical outcome in nonsurgical patients. J Neurosurg 81: 531–538. Brown R, Flemming K, Meyer F et al. (2005). Natural history, evaluation, and management of intracranial vascular malformations. Mayo Clin Proc 80: 269–281. Chaudhary M, Sachdev V, Cho S et al. (1982). Dural arteriovenous malformation of the major venous sinuses: an acquired lesion. AJNR Am J Neuroradiol 3: 13–19. Chen C, Lee C, Ding D et al. (2015). Stereotactic radiosurgery for intracranial dural arteriovenous fistulas: a systematic review. J Neurosurg 122: 353–362. Cognard C, Gobin YP, Pierot L et al. (1995). Cerebral dural arteriovenous fistulas: clinical and angiographic correlation
with a revised classification of venous drainage. Radiology 194: 671–680. Duffau H, Lopes M, Janosevic V et al. (1999). Early rebleeding from intracranial dural arteriovenous fistulas: report of 20 cases and review of the literature. J Neurosurg 90: 78–84. Gross B, Du R (2012). The natural history of cerebral dural arteriovenous fistulae. Neurosurgery 71: 594–602. Gross B, Albuquerque F, Moon K et al. (2016). Evolution of treatment and a detailed analysis of occlusion, recurrence, and clinical outcomes in an endovascular library of 260 dural arteriovenous fistulas. J Neurosurg Sep2: 1–10. http://dx.doi.org/10.3171/2016.5.JNS16331; E pub ahead of print. Herman J, Spetzler R, Bederson J et al. (1995). Genesis of a dural arteriovenous malformation in a rat model. J Neurosurg 83: 539–545. Hetts S, Tsai T, Cooke D et al. (2015). Progressive versus nonprogressive intracranial dural arteriovenous fistulas: characteristics and outcomes. AJNR Am J Neuroradiol 36: 1912–1919. Houser O, Baker H, Rhoton A et al. (1972). Intracranial dural arteriovenous malformations. Radiology 105: 55–64. Houser O, Campbell J, Campbell R et al. (1979). Arteriovenous malformation affecting the transverse dural venous sinus – an acquired lesion. Mayo Clin Proc 54: 651–661. Jolink W, van Dijk J, van Asch C et al. (2015). Outcome after intracranial haemorrhage from dural arteriovenous fistulae; a systematic review and case-series. J Neurol 262: 2678–2683. Kobayashi A, Al-Shahi Salman R (2014). Prognosis and treatment of intracranial dural arteriovenous fistulae: a systematic review and meta-analysis. Int J Stroke 9: 670–677. Kutluk K, Schumacher M, Mironov A (1991). The role of sinus thrombosis in occipital dural arteriovenous malformations – development and spontaneous closure. Neurochirurgica 34: 144–147. Lasjaunias P, Chiu M, ter Brugge K et al. (1986). Neurological manifestations of intracranial dural arteriovenous malformations. J Neurosurg 64: 724–730. Lawton M, Jacobowitz R, Spetzler R (1997). Redefined role of angiogenesis in the pathogenesis of dural arteriovenous malformations. J Neurosurg 87: 267–274. Newton T, Hoyt W (1970). Dural arteriovenous shunts in the region of the cavernous sinus. Neuroradiology 1: 71–81. Nishijima M, Takaku A, Endo S et al. (1992). Etiological evaluation of dural arteriovenous malformations of the lateral and sigmoid sinuses based on histopathological examinations. J Neurosurg 76: 600–606. Park K, Kang D, Park S et al. (2016). The efficacy of gamma knife radiosurgery alone as a primary treatment for intracranial dural arteriovenous fistulas. Acta Neurochir 158: 821–828. Singh V, Smith W, Lawton M et al. (2008). Risk factors for hemorrhagic presentation in patients with dural arteriovenous fistulae. Neurosurgery 62: 628–635. Soderman M, Pavic L, Edner G et al. (2008). Natural history of dural arteriovenous shunts. Stroke 39: 1735–1739.
RADIOSURGERY FOR DURAL ARTERIOVENOUS MALFORMATIONS Terada T, Higashida R, Halbach V et al. (1994). Development of acquired arteriovenous fistulas in rats due to venous hypertension. J Neurosurg 80: 884–889. Uranishi R, Nakase H, Sakaki T (1999). Expression of angiogenic growth factors in dural arteriovenous fistula. J Neurosurg 91: 781–786. van Dijk J, ter Brugge K, Willinsky R et al. (2002). Clinical course of cranial dural arteriovenous fistulas with long-term persistent cortical venous reflux. Stroke 33: 1233–1236.
131
Vinuela F, Fox A, Debrun G et al. (1984). Spontaneous carotid-cavernous fistulas: clinical, radiological, and therapeutic considerations. Experience with 20 cases. J Neurosurg 60: 976–984. Yang H, Kano H, Kondziolka D et al. (2013). Stereotactic radiosurgery with or without embolization for intracranial dural arteriovenous fistulas. Prog Neurol Surg 27: 195–204.
Chapter 12
Radiosurgery for dural arteriovenous malformations CONOR GRADY1, CAROLINA GESTEIRA BENJAMIN1, AND DOUGLAS KONDZIOLKA1,2* 1 Department of Neurosurgery, NYU Langone Medical Center, New York, NY, USA 2
Center for Advanced Radiosurgery, NYU Langone Medical Center, New York, NY, USA
Abstract Intracranial dural arteriovenous malformations (DAVFs) are relatively uncommon vascular lesions characterized by the direct connection of dural arteries into dural venous sinuses or leptomeningeal veins. Strategies for the treatment of these complex lesions have evolved significantly over the past three decades, and include open surgical disconnection, endovascular embolization, stereotactic radiosurgery (SRS), or a combination of these approaches. Radiosurgical intervention is unique in offering significant benefits to patients while exposing them to few of the risks associated with more invasive interventions. In this chapter we provide an overview of DAVFs and discuss the features of these lesions that affect management. We focus, in particular, on radiosurgical management of these lesions, describing present treatment paradigms, the procedure for the treatment of DAVFs with SRS, and expected clinical outcomes using SRS.
INTRODUCTION
PATHOGENESIS
Intracranial dural arteriovenous fistulas (DAVFs) are relatively uncommon vascular lesions characterized by the direct connection of dural arteries into dural venous sinuses or leptomeningeal veins. They are believed to constitute 10–15% of all intracranial vascular lesions (Houser et al., 1972), and are distinct from other intracranial arteriovenous malformations (AVMs) in their pathogenesis and presentation. Strategies for the treatment of these complex lesions have evolved significantly over the past three decades, and include open surgical disconnection, endovascular embolization, stereotactic radiosurgery (SRS), or a combination of these approaches. Radiosurgical intervention is unique in offering significant benefits to patients while exposing them to few of the risks associated with more invasive interventions. In this chapter we will provide an overview of DAVFs and discuss the features of these lesions that affect treatment. We will focus, in particular, on radiosurgical management of these lesions, and expected clinical outcomes using SRS.
In contrast to the majority of intracranial AVMs, DAVFs are primarily believed to be acquired lesions (Houser et al., 1979; Chaudhary et al., 1982; Nishijima et al., 1992). Over the past four decades, clinical observation, histologic examination of resected specimens, and animal models have led to multiple hypotheses of the cardinal event leading to their formation. Houser et al. (1979) were amongst the first to define the angiographic characteristics of these lesions, and noted the relationship of their formation to the presence of dural sinus thrombosis. Subsequent studies continued to support the association of DAVF with sinus thrombosis, though controversy existed regarding the etiologic relationship of sinus thrombosis to fistula formation (Chaudhary et al., 1982; Kutluk et al., 1991). The present leading hypothesis of DAVF formation posits a repetitive cycle of anatomic and molecular events triggered by venous hypertension. Venous hypertension itself may result from any number of factors, including, but not limited to, trauma and venous
*Correspondence to: Douglas Kondziolka, MD, MSc, FRCSC, FACS, Center for Advanced Radiosurgery, NYU Langone Medical Center, 530 First Avenue, Suite 8R, New York NY 10016, USA. Tel: +1-646-501-2360, E-mail: [email protected]
126
C. GRADY ET AL.
thrombosis (Nishijima et al., 1992). Anatomically, venous hypertension leads to dilation of dural veins and enlargement of microscopic arteriovenous shunts normally present in dura. The enlarging of these shunts further contributes to venous hypertension, as increased arterial flow reaches new venous territories. Blood flow through these microscopic fistulae become increasingly turbulent as they enlarge, leading to stasis and promoting thrombosis, further contributing to venous hypertension. The outcome of this cascade of events has been demonstrated by histologic examination of resected DAVFs, which reveals dilation and thickening of microscopic arterial and venous vessels present in the dura adjacent to fistula sites, as well as partially canalized sinus thromboses (Houser et al., 1979; Nishijima et al., 1992). This theory has been additionally supported by animal studies in which both carotid to external jugular anastomoses and experimentally induced sinus thromboses have first resulted in venous hypertension, followed by the formation of DAVFs (Terada et al., 1994; Herman et al., 1995). On a molecular level, venous hypertension is believed to cause local ischemia and endothelial injury, leading to the increased expression of angiogenic growth factors which contribute to the histologic findings described above. Immunohistochemical analysis of resected DAVFs has demonstrated markedly increased expression of basic fibroblast growth factor and vascular endothelial growth factor (Uranishi et al., 1999). Lawton et al. (1997) also successfully demonstrated increased expression of angiogenic growth factors and DAVF formation in a rat model of venous hypertension.
PRESENTATION DAVFs present with a variety of symptoms which are related to the anatomic location of the fistula itself. The majority of DAVFs present with local symptoms arising from altered hemodynamics. In one of the early large case series looking at the presenting symptoms of DAVFs, Lasjaunias et al. (1986) reported that 83% of patients with cavernous sinus DAVFs presented with proptosis or visual changes, while 70% of patients with torcular, transverse sinus, or sigmoid sinus DAVFs presented with an audible bruit. More recently, Gross et al. (2016), in a retrospective review of a prospectively maintained database of 260 patients with DAVFs, found that 92% of patient with cavernous sinus lesions presented with ocular findings, while 75% of patients with transverse sinus, sigmoid sinus, marginal sinus, or torcular DAVFs presented with pulsatile tinnitus. DAVFs can also present with more malignant symptoms, such as seizure, neurologic deficit, or hemorrhage. Fistula location is again one of the factors correlated with the risk for the development of these symptoms. Awad
et al. (1990) observed that only 10–20% of cavernous or transverse sigmoid sinus DAVFs presented with hemorrhage, while greater than 90% of tentorial DAVFs had hemorrhage at presentation. The discrepancy of risk for hemorrhage of fistulae in different locations has been further illustrated by more recent case series, which have reported a 6–11% risk of hemorrhage as the presenting symptom for DAVFs of the cavernous and transverse sigmoid sinuses compared to a 54–79% risk for presentation with hemorrhage in DAVFs of the tentorium or petrosal sinuses (Singh et al., 2008; Gross et al., 2016). Although this indicates that there is a correlation between the location of the fistula and the likelihood of presenting with hemorrhage or other malignant symptoms, it is important to note that fistulae at any location may exhibit aggressive behavior. Aside from location, other factors have been identified as predictors of presentation with hemorrhage. In a review of 402 patients with DAVFs, Singh et al. (2008) identified cortical venous drainage (CVD), focal neurologic deficit, male sex, location in the posterior fossa, and age greater than 50 at presentation as independent risk factors for hemorrhage. Of note, amongst these risk factors, the presence of CVD was the strongest predictor of presentation with hemorrhage, with an odds ratio of 10.5, more than twice the odds ratio of the next highest predictor amongst the identified risk factors. The unique role of CVD in determining risk of hemorrhage at presentation was further elucidated by Gross and Du (2012), who identified male sex and tentorial or petrous location of fistulae as strongly associated with the presence of cortical drainage. These studies reinforce the findings of previous case series and literature reviews (Awad et al., 1990), and lend support to the long-accepted practice of sorting DAVFs into high- or low-risk categories based on the presence of CVD.
CLASSIFICATION Given its significance in determining risk for malignant presentation, multiple classification schemes have been proposed for DAVFs, categorizing lesions primarily based on angiographic evidence of CVD. The two most widely applied of these criteria were formulated by Borden et al. (1995) and Cognard et al. (1995). The Borden classification system divides lesions into three categories based on the type of arteriovenous anastomosis (i.e., arterial drainage into a dural sinus or vein versus direct anastomosis with a cortical vein) as well as the presence of CVD. Borden type I lesions are a direct anastomosis between a meningeal artery and a dural sinus or vein with purely anterograde drainage. Type II lesions are also a result of anastomosis between a meningeal artery and a dural vein or sinus, but with reversal
RADIOSURGERY FOR DURAL ARTERIOVENOUS MALFORMATIONS of flow and reflux into cortical veins. Type III lesions occur when meningeal arteries directly drain into cortical veins. Type I lesions are considered benign, and their spontaneous remission has been well documented (Newton and Hoyt, 1970; Vinuela et al., 1984; Kutluk et al., 1991). Both type II and type III lesions are considered aggressive, and warrant evaluation for intervention. The classification system proposed by Cognard and colleagues (1995) takes several other features of a DAVF into account. Based on a retrospective analysis of a single center’s experience with 205 patients presenting with DAVF, the Cognard system divides lesions into five categories, corresponding with their risk for malignant presentation. Type I lesions drain directly into a dural sinus, with only anterograde flow. Type II fistulae drain into dural sinuses with resulting retrograde flow, and are subdivided based on whether the retrograde flow involves only dural sinuses (subtype a), only cortical veins (subtype b), or involves both sinuses and cortical veins (subtype a + b). Type III lesions are direct anastomosis of meningeal arteries with cortical veins without ectasia, while type IV lesions are direct cortical vein anastomoses with ectasia. Type V lesions drain directly into the spinal venous system. As in the Borden classification system, those types featuring CVD (i.e., types IIb–V) are considered to be aggressive and should be considered for intervention. Though amenable to classification by the Borden and Cognard systems, DAVFs involving the cavernous sinus can additionally be classified by the Barrow system (Barrow et al., 1985). In this system, type A lesions are direct, high-flow fistulae between the internal carotid artery and the caverous sinus. Type B, C, and D lesions all represent low-flow fistulous connections between dural arteries and the cavernous sinus, and are distinguished by whether the dural vessels arise from the internal (type B), external (type C), or both (type D) carotid arteries. Barrow type A fistulae are traumatic lesions which are distinct from other forms of DAVFs, and their natural history and management are beyond the scope of this chapter. Type B–D lesions are analogous to DAVF occurring at other locations.
NATURAL HISTORY Attempts to precisely characterize the natural history of DAVFs have been hindered by the relatively low incidence of these lesions. A prospective, populationbased cohort estimated a detection rate of only 0.16 per 100 000 adults per year (Al-Shahi et al., 2003). Consequently, a recent literature review and meta-analysis discovered that all literature describing the natural history of DAVFs is comprised of small, hospital-based cohorts, the vast majority of which are retrospective in
127
design, and which seldom account for the treatment of lesions that are symptomatic at presentation when reporting subsequent outcomes (Jolink et al., 2015). In the absence of high-quality data, quoted risks of morbidity and mortality vary widely. Brown et al. (1994), in reviewing the records of 54 patients with untreated DAVF with average follow-up of 6.6 years, found that 5 patients had a DAVF-related hemorrhage, resulting in an overall 1.8% per year risk of hemorrhage. This is similar to the annual risk of rupture of 1.5% reported by Soderman et al. (2008), in their evaluation of 53 patients presenting with unruptured DAVFs. In the same study, patients presenting with hemorrhage were calculated to have a 7.4% per year risk of re-hemorrhage. This is significantly lower, however, than the 20–35% re-hemorrhage rate demonstrated by Duffau et al. (1999) in their review of 20 patients with previous hemorrhage from DAVF. Likewise, following 20 patients with long-term CVD, van Dijk et al. (2002) calculated an annual mortality rate of 10.4%, as well as a 15% annual rate of hemorrhage or focal neurologic deficit, rates significantly higher than reported in previous studies. Looking at an endpoint other than symptomatic presentation, Hetts et al. (2015), found that only 3.2% of 574 patients with angiogram-confirmed DAVF demonstrated progressive or de novo fistulae. Of note, these progressive lesions demonstrated 26%, 32%, and 47% risk for hemorrhage, focal neurologic deficit, or cranial neuropathy, respectively, during the period of the study. Taken in sum, the available literature on the natural history of DAVFs indicates that these lesions pose a variable risk to affected patients. In general, aggressive features, such as CVD or formation of de novo lesions, indicate that patients are at an increased risk for hemorrhage or significant neurologic symptoms compared to patients with stable lesions not featuring CVD.
RADIOSURGERY FOR THE MANAGEMENT OF DAVFS Symptomatic lesions or high-grade lesions at increased risk for malignant presentation warrant intervention. Although endovascular embolization, surgical disconnection, or radiosurgery can each be used alone, singlemodality treatment may not be effective, and thus many DAVFs require multimodality approaches (Kobayashi and Al-Shahi Shalman, 2014). Endovascular embolization, especially via the transvenous route, has emerged as the most commonly employed solitary treatment of DAVFs. However, studies have shown that embolization alone is not sufficient for complete obliteration of the fistula in 50% of patients (Yang et al., 2013). This is largely due to limitation in visualizing all the feeding
128
C. GRADY ET AL.
vessel with angiography and the technical challenges of superselective catheterization of small feeding vessels. Therefore it is frequently combined with radiosurgery for higher obliterations rates (Kobayashi and Al-Shahi Salman, 2014). SRS can also be employed as a solitary approach, but is more frequently used as a valuable adjunct to endovascular or open approaches to complex lesions. Scenarios in which radiosurgery would be used as a single modality of therapy include patients whose lesions may not be amenable to other intervention, patients with significant medical comorbidities, or patients with low-grade, minimally symptomatic lesions. Similar to the experience with parenchymal AVMs, radiosurgery is expected to lead to progressive obliteration of DAVFs over 1–3 years following irradiation. Initial SRS followed by transarterial embolization either the same day or soon after has been shown to be an effective treatment paradigm, with embolization providing more immediate symptomatic relief and decreased flow through the lesion, and SRS providing long-term, durable closure with lower risk for recanalization (Yang et al., 2013). If treatment with both SRS and endovascular embolization is planned, SRS should be performed first to prevent embolization material from obscuring any portion of the nidus during planning and irradiation.
Procedure In the majority of cases cited in the literature, SRS for the treatment of DAVFs is performed using gamma knife technique (Brown et al., 2005; Chen et al., 2015). The procedure begins with the application of a Leksell stereotactic head frame under conscious sedation with local scalp anesthesia. Given the peripheral location of many noncavernous sinus DAVFs, eccentric head frame placement used to be critical to prevent patient collision with the collimator helmet. This, however, has become less of a concern with new models of gamma knife, which feature fully automated patient positioning via movement of the table on which the patient is introduced into the collimator. Once the head frame is applied, target identification is performed using stereotactic angiography, often requiring injections of bilateral internal and external carotid arteries as well as the bilateral vertebral arteries. High-resolution stereotactic magnetic resonance imaging is then performed to further aid target acquisition and to identify adjacent structures which must be spared from the dose plan. All imaging is exported to a computer workstation, where dose planning is performed. In the published experience of Yang et al. (2013), the median prescription dose delivered to the margin of DAVFs undergoing SRS was 20 Gy (range,
15–25 Gy). The median isodose volume was 2.0 cm3 (range 0.2–8.6 cm3), with prescription isodose of 50% in all but two cases, in which the prescription isodose was 60%. Median maximum dose was 40 Gy, with a range of 30–50 Gy. This is similar to the features of all series contained in a recent meta-analysis of outcomes in DAVFs treated by SRS; across all studies, reported median marginal doses ranged from 18 to 25 Gy, treatment volume ranged from 1 to 9.3 cm3, and prescription isodose from 40 to 60% (Chen et al., 2015). Likewise, a large, single-center cohort of 105 patients not included in the review employed a median radiation dose delivered to the margin of 19 Gy (range, 15–25 Gy) with a maximum dose of 38 Gy (range, 22–50 Gy) (Yang et al., 2013). Figures 12.1 and 12.2 provide example dose plans for representative cases.
OUTCOMES FOLLOWING RADIOSURGERY SRS provides reasonable rates of complete occlusion with low risk for serious adverse events. In their systematic review and meta-analysis of 19 studies examining the outcomes of a pooled total of 729 patients having undergone SRS for treatment of 743 DAVFs, Chen et al. (2015) reported a complete obliteration rate of 63%, while calculating post-SRS rates of hemorrhage, neurologic deficit, or death at last follow-up as 1.2%, 1.3%, and 0.3%, respectively. This review further subdivided outcomes based on the location, angioarchitecture, and Borden classification of the lesions treated, given the relationship of these features to risk for hemorrhage and neurologic deficit. By location, 73% of cavernous sinus DAVFs treated with SRS demonstrated complete occlusion at follow-up, compared to 58% of DAVFs occurring at other locations. No patients in any study with a cavernous sinus DAVF experienced post-SRS hemorrhage, while 1.3% of patients with noncavernous sinus locations did. Of patients with noncavernous sinus DAVFs, variability in outcome was seen based on Borden classification, with 67% of type I lesions experiencing complete occlusion, compared to 43% of type II lesions and 50% of type III lesions. The rate of post-SRS hemorrhage for patients with Borden type I was 0%, compared to 2.6% and 3.7% in patients with Borden type II and Borden type III lesions, respectively. The presence or absence of CVD, regardless of location or specific classification type, also demonstrated a relationship with outcome. Of lesions without CVD, 74.5% demonstrated complete occlusion on follow-up imaging, and none experienced post-SRS hemorrhage, compared to 56% occlusion and 4.6% post-SRS hemorrhage rates in lesions with CVD.
RADIOSURGERY FOR DURAL ARTERIOVENOUS MALFORMATIONS
129
Fig. 12.1. A 46-year-old woman with memory dysfunction found to have a falcotentorial dural arteriovenous fistula. Radiosurgery was performed with 20 Gy to be followed by embolization to reduce venous hypertension.
Fig. 12.2. A 36-year-old woman with a falcotentorial arteriovenous fistula associated with 2 years of headache. The radiosurgery plan was created to delivery 18 Gy to the 50% isodose margin.
A significant limitation in reporting outcomes for DAVFs treated by SRS is the previous or concurrent treatment of many patients with endovascular embolization. In the meta-analysis by Chen et al. (2015), at least
17% of all patients included in the analysis were reported to undergo endovascular intervention, while many of the studies included provided insufficient information to allow the authors to extract whether patients had
130
C. GRADY ET AL.
previous treatment or were undergoing SRS alone. A recent retrospective series by Park et al. (2016) helps to address this, following 20 patients with DAVFs treated with SRS alone. The study included eight symptomatic lesions without CVD, 11 lesions with CVD but no symptoms, and one “high-risk” lesion which had both CVD and symptoms, but which was not amenable to endovascular or open intervention. Follow-up with thin-cut time-of-flight magnetic resonance angiography revealed complete occlusion in 18 of 20 lesions at a mean of 29 months’ follow-up. All patients with symptomatic lesions experienced complete resolution of their symptoms, with a mean time of 3.5 months from SRS to symptom improvement.
CONCLUSION DAVFs are complex lesions that pose variable risks to the patients who harbor them. SRS is a safe and effective treatment modality for these lesions, both as a solitary treatment and in conjunction with endovascular embolization or open surgical disconnection. SRS is especially effective in lesions of the cavernous sinus, and in lesions without CVD.
REFERENCES Al-Shahi R, Bhattacharya J, Currie DG et al. (2003). Prospective, population-based detection of intracranial vascular malformations in adults: the Scottish Intracranial Vascular Malformation Study (SIVMS). Stroke 34: 1163–1169. Awad I, Little J, Akarawi W et al. (1990). Intracranial dural arteriovenous malformations: factors predisposing to an aggressive neurological course. J Neurosurg 72: 839–850. Barrow D, Spector R, Braun I et al. (1985). Classification and treatment of spontaneous carotid-cavernous sinus fistulas. J Neurosurg 62: 248–256. Borden J, Wu J, Shucart W (1995). A proposed classification for spinal and cranial dural arteriovenous fistulous malformations and implications for treatment. J Neurosurg 82: 166–179. Brown R, Wiebers D, Nichols D (1994). Intracranial dural arteriovenous fistulae: angiographic predictors of intracranial hemorrhage and clinical outcome in nonsurgical patients. J Neurosurg 81: 531–538. Brown R, Flemming K, Meyer F et al. (2005). Natural history, evaluation, and management of intracranial vascular malformations. Mayo Clin Proc 80: 269–281. Chaudhary M, Sachdev V, Cho S et al. (1982). Dural arteriovenous malformation of the major venous sinuses: an acquired lesion. AJNR Am J Neuroradiol 3: 13–19. Chen C, Lee C, Ding D et al. (2015). Stereotactic radiosurgery for intracranial dural arteriovenous fistulas: a systematic review. J Neurosurg 122: 353–362. Cognard C, Gobin YP, Pierot L et al. (1995). Cerebral dural arteriovenous fistulas: clinical and angiographic correlation
with a revised classification of venous drainage. Radiology 194: 671–680. Duffau H, Lopes M, Janosevic V et al. (1999). Early rebleeding from intracranial dural arteriovenous fistulas: report of 20 cases and review of the literature. J Neurosurg 90: 78–84. Gross B, Du R (2012). The natural history of cerebral dural arteriovenous fistulae. Neurosurgery 71: 594–602. Gross B, Albuquerque F, Moon K et al. (2016). Evolution of treatment and a detailed analysis of occlusion, recurrence, and clinical outcomes in an endovascular library of 260 dural arteriovenous fistulas. J Neurosurg Sep2: 1–10. http://dx.doi.org/10.3171/2016.5.JNS16331; E pub ahead of print. Herman J, Spetzler R, Bederson J et al. (1995). Genesis of a dural arteriovenous malformation in a rat model. J Neurosurg 83: 539–545. Hetts S, Tsai T, Cooke D et al. (2015). Progressive versus nonprogressive intracranial dural arteriovenous fistulas: characteristics and outcomes. AJNR Am J Neuroradiol 36: 1912–1919. Houser O, Baker H, Rhoton A et al. (1972). Intracranial dural arteriovenous malformations. Radiology 105: 55–64. Houser O, Campbell J, Campbell R et al. (1979). Arteriovenous malformation affecting the transverse dural venous sinus – an acquired lesion. Mayo Clin Proc 54: 651–661. Jolink W, van Dijk J, van Asch C et al. (2015). Outcome after intracranial haemorrhage from dural arteriovenous fistulae; a systematic review and case-series. J Neurol 262: 2678–2683. Kobayashi A, Al-Shahi Salman R (2014). Prognosis and treatment of intracranial dural arteriovenous fistulae: a systematic review and meta-analysis. Int J Stroke 9: 670–677. Kutluk K, Schumacher M, Mironov A (1991). The role of sinus thrombosis in occipital dural arteriovenous malformations – development and spontaneous closure. Neurochirurgica 34: 144–147. Lasjaunias P, Chiu M, ter Brugge K et al. (1986). Neurological manifestations of intracranial dural arteriovenous malformations. J Neurosurg 64: 724–730. Lawton M, Jacobowitz R, Spetzler R (1997). Redefined role of angiogenesis in the pathogenesis of dural arteriovenous malformations. J Neurosurg 87: 267–274. Newton T, Hoyt W (1970). Dural arteriovenous shunts in the region of the cavernous sinus. Neuroradiology 1: 71–81. Nishijima M, Takaku A, Endo S et al. (1992). Etiological evaluation of dural arteriovenous malformations of the lateral and sigmoid sinuses based on histopathological examinations. J Neurosurg 76: 600–606. Park K, Kang D, Park S et al. (2016). The efficacy of gamma knife radiosurgery alone as a primary treatment for intracranial dural arteriovenous fistulas. Acta Neurochir 158: 821–828. Singh V, Smith W, Lawton M et al. (2008). Risk factors for hemorrhagic presentation in patients with dural arteriovenous fistulae. Neurosurgery 62: 628–635. Soderman M, Pavic L, Edner G et al. (2008). Natural history of dural arteriovenous shunts. Stroke 39: 1735–1739.
RADIOSURGERY FOR DURAL ARTERIOVENOUS MALFORMATIONS Terada T, Higashida R, Halbach V et al. (1994). Development of acquired arteriovenous fistulas in rats due to venous hypertension. J Neurosurg 80: 884–889. Uranishi R, Nakase H, Sakaki T (1999). Expression of angiogenic growth factors in dural arteriovenous fistula. J Neurosurg 91: 781–786. van Dijk J, ter Brugge K, Willinsky R et al. (2002). Clinical course of cranial dural arteriovenous fistulas with long-term persistent cortical venous reflux. Stroke 33: 1233–1236.
131
Vinuela F, Fox A, Debrun G et al. (1984). Spontaneous carotid-cavernous fistulas: clinical, radiological, and therapeutic considerations. Experience with 20 cases. J Neurosurg 60: 976–984. Yang H, Kano H, Kondziolka D et al. (2013). Stereotactic radiosurgery with or without embolization for intracranial dural arteriovenous fistulas. Prog Neurol Surg 27: 195–204.