Stereotactic Radiosurgery for Brain Arteriovenous Malformations

25 Stereotactic Radiosurgery for Brain Arteriovenous Malformations

Or Cohen-Inbar, Dale Ding, and Jason P. Sheehan

Abstract

Brain arteriovenous malformations (AVMs) are rare, angioarchi-tecturally diverse vascular malformations that most frequently present with hemorrhage, seizure, headache, and focal neurological deficit. Left untreated, an AVM’s rupture risk is approximately 2 to 4% annually, although certain factors, such as prior hemorrhage, deep location, deep venous drainage, and associated arterial aneurysms, have been shown to predispose AVMs to rupture. The management of AVMs is challenging and multifactorial, and a number of single-modality or multimodality therapies have been devised to treat these lesions. The primary goal of AVM treatment is complete obliteration of the nidus, which eliminates the risk of hemorrhage.

Stereotactic radiosurgery (SRS) is a minimally invasive alterative to resection and curative embolization for AVM intervention, and acts by inducing progressive endoluminal occlusion of nidal vasculature. Obliteration after SRS is achieved in approximately 70 to 80% of AVMs within 3 to 5 years of treatment, and is intimately related to nidus volume and SRS margin dose. Adverse radiation effects (ARE), defined as T2-weighted perinidal hyperintensities on magnetic resonance imaging, are radiologically evident in approximately one-third of patients after SRS. The rates of symptomatic and permanent ARE are approximately 10 and 2 to 3%, respectively. The risk of AVM hemorrhage persists during the latency period between SRS and nidal obliteration, and is comparable to or more benign than an untreated AVM’s natural history. Large AVMs (diameter > 3 cm or volume > 12 cm³) can be partially occluded with partial embolization prior to SRS, or treated with dose- or volume-staged SRS techniques. Pre-SRS embolization can also selectively occlude flow-related arterial aneurysms or intranidal arteriovenous fistulas.

Keywords: Gamma Knife, intracranial arteriovenous malformation, intracranial hemorrhages, radiosurgery, stroke, vascular malformations

Key Points

Stereotactic radiosurgery (SRS) is a minimally invasive alterative to microsurgical resection and curative embolization for the treatment of brain arteriovenous malformations (AVMs).
The primary goal of SRS is obliteration of the AVM nidus; this can be achieved in approximately 70 to 80% of cases within 3 to 5 years of treatment, and obliteration is directly related to AVM volume and SRS margin dose.
Adverse radiation effects (ARE) manifest as T2-weighted perinidal hyperintensities on magnetic resonance imaging, and they are radiologically evident in approximately one-third of patients after SRS; the rates of symptomatic and permanent ARE are approximately 10% and 2 to 3%, respectively.
The risk of AVM hemorrhage persists during the latency period between SRS and nidal obliteration.
Pre-SRS embolization can be employed to reduce the volume of large AVMs (diameter > 3 cm or volume > 12 cm³)or to occlude associated arterial aneurysms or intranidal arteriovenous fistulas; however, embolized AVMs may have reduced the probability of SRS-induced obliteration.

25.1 Introduction

Cerebral arteriovenous malformations (AVMs) are rare congenital vascular malformations, consisting of an abnormal tangle of blood vessels within the intracranial space. AVM blood vessels shunt blood directly from arteries to veins without an intervening capillary bed, exposing the venous structures to abnormal hemodynamic forces.¹ The resultant abrupt transition from a high-pressure muscular arterial system to a low-pressure venous system results in venous dilatation and engorgement. Eventually, this process causes rupture of the AVM nidus, which is associated with significant morbidity and mortality.² Secondary processes evolving in time include vessel wall arterialization, with resultant edema and inflammation of the surrounding brain tissue.³^,⁴^,⁵ Many patients suffer from the clinical manifestations of these pathological changes, such as focal neurological deficits and seizures.⁶ The incidence of AVMs is similar for both genders, and they are typically diagnosed by early adulthood (third or fourth decades of life).³ The incidence of cerebral AVMs is estimated to be in the order of 1.12 to 1.34 per 100,000 person-years.⁷ Cerebral AVMs account for approximately 10% of subarachnoid hemorrhages and 1 to 2% of all strokes.⁸ In the absence of treatment, the overall annual risk of a spontaneous hemorrhage from a cerebral AVM is approximately 2 to 4%, although this rate has been found to vary substantially depending on a number of patient- and AVM-specific factors.³ In one study, the annual AVM hemorrhage rates ranged from 0.9% for low-risk patients (no history of prior AVM hemorrhage, superficial AVM location, AVM with a component of superficial venous drainage) to as high as 34.4% for high-risk patients (a history of prior AVM hemorrhage, deep AVM location, AVM with exclusively deep venous drainage).⁵^,⁶ The combination of a relatively young age at presentation and a nontrivial annual hemorrhage risk leads to a substantial lifetime risk of morbidity and mortality from untreated AVMs.³^,⁴

AVMs continue to represent a significant clinical challenge, and expert opinions differ regarding the optimal management of AVMs.⁹ Treatment goals also can vary among patients, and they might be directed toward reducing seizure activity, ameliorating symptomatic chronic “vascular steal,” or alleviating neurological deficits caused by perinidal cerebral edema. However, the main goal of any intervention for AVMs is the complete obliteration of the nidus, thereby eliminating subsequent risk of hemorrhage. There currently exists significant controversy regarding the management of unruptured AVMs, with recent prospective studies reporting poorer short-term outcomes after intervention for unruptured AVMs compared to conservative management.¹⁰^,¹¹ A randomized trial of unruptured brain arteriovenous malformations (ARUBA) reported significantly higher rates of symptomatic stroke and death in patients assigned to undergo intervention (31%) compared to those assigned to conservative management (10%) at a mean follow-up of 33 months.¹⁰ The Scottish Audit of Intracranial Vascular Malformations prospective AVM cohort study similarly found that unruptured AVM patients who underwent intervention had significantly higher rates of death or sustained morbidity (defined as Oxford Handicap Scale ≥ 2 for ≥ 2 years at 4 years’ follow-up and significantly higher rates of death or symptomatic stroke secondary to an AVM, associated aneurysm, or intervention at 12 years’ follow-up).¹¹ A widely accepted view for the management of incidentally diagnosed AVMs is that the intervention is only justifiable if the risks of treatment-associated morbidity and mortality are less than those of the AVM’s natural course.

Given its minimal invasiveness and favorable therapeutic profile, stereotactic radiosurgery (SRS) has emerged as an effective treatment for AVMs without immediate risk of hemorrhage.¹² Many small- or medium-sized (diameter < 3 cm or volume < 10– 15 cm³) AVMs that are deemed too risky for resection can be safely and completely obliterated by radiosurgery. Regardless of the SRS platform used (e.g., Gamma Knife, CyberKnife, Linear accelerator [LINAC] based system), AVM obliteration following radiosurgery ensues through progressive intimal thickening, thrombosis of irradiated vessels, and eventual occlusion of the vascular lumen.¹³ The rate of obliteration, shown either on catheter cerebral angiography or magnetic resonance imaging (MRI), has been consistently reported to range from approximately 70 to 80% within 3 to 5 years of SRS in large, unselected cohorts.⁹ Even for larger AVMs, some degree of AVM volume reduction typically occurs after SRS, thereby facilitating additional definitive treatment of the remaining nidus.¹⁴ Additionally, some studies have suggested that SRS may confer partial protection from AVM hemorrhage during the latency period prior to complete obliteration.¹⁵ The progressive thickening of vessel walls is postulated to reduce the tension within the vascular wall, acting to protect the nidus from rupture.¹⁵

25.1.1 Role of Pre-SRS AVM Embolization

Pre-SRS embolization is usually performed when an AVM nidus is too large (maximum diameter > 3 cm, volume > 10–15 cm³)to treat with single-session SRS alone.¹⁶ However, pre-SRS embolization has previously been reported to decrease the rate of AVM-nidus obliteration.¹⁴^,¹⁷ We found, in multivariate analysis, that pre-SRS embolization is an independent negative predictor of obliteration (p < 0.001).⁶ One possible explanation is recanalization of the embolized portion of the nidus, which is not targeted with SRS.¹⁸ Embolization can also make SRS more difficult if it treats parts of the nidus in multisegmented fashion rather than walling off a specific sector, transforming a compact nidus into a more diffuse one.

Some preclinical data suggest that embolic agents can scatter or absorb radiation, thereby reducing the effective radiosurgical dose to the targeted nidus.¹⁹ However, Bing et al recently provided experimental data that contradict the notion of radiation beam scattering or absorption by embolic agents.²⁰ Another mechanism by which embolized AVMs are more susceptible to SRS treatment failure is embolization-induced angiogenesis, although the contribution of this biomolecular phenomenon to macroscopic outcomes is unknown.²¹ In a matched cohort analysis of embolized and nonembolized AVMs treated with SRS, we found that, while the obliteration rate for the embolized AVM cohort was significantly lower than the nonembolized AVM cohort, the effect of prior embolization on post-SRS outcome was confounded by nidal angioarchitectural complexity (defined as the sum of the number of major feeding arteries and draining veins).²² Finally, one should consider that inherent baseline differences between embolized and nonembolized AVMs may account for the disparities in outcomes after SRS.

25.1.2 Evaluation of Arteriovenous Malformation Obliteration

Catheter cerebral angiography continues to serve as the gold standard for confirming AVM obliteration after SRS, albeit its invasiveness makes it less ideal for routine follow-up. Consequently, most neurosurgeons prefer MRI for routine post-SRS follow-up, and use angiography to confirm obliteration after a lack of flow voids is observed on MRI. The authors previously conducted a study evaluating the sensitivity and specificity of MRI in evaluating AVM obliteration after SRS, compared to angiography.²³ The sensitivity was reported as 85 and 77% and the specificity was reported as 89 and 95% by two independent observers.²³ Therefore, MRI predicts AVM obliteration after SRS in the majority of patients and can be appropriately used in their follow-up.²³ Kano et al²⁴ suggested that the potential slight overestimation of obliteration rates as determined by MRI is balanced by the tendency to underestimate long-term obliteration rates based only on early follow-up examinations.²⁴ We advocate routine follow-up using MRI every 6 months for the first 2 years after SRS, unless there exists a clinical indication for more frequent evaluations (i.e., new or worsening neurological symptoms). Afterward, MRIs can be performed annually until obliteration is achieved, at which point we recommend angiography to confirm the lack of a residual nidus.

25.2 Results

25.2.1 Predicting Outcomes after SRS for AVMs

Arteriovenous Malformation Obliteration after SRS

The primary goal of SRS for AVMs is complete angiographic obliteration of the nidus. Nidal obliteration confers durable protection from future AVM hemorrhage. Flickinger et al performed a dose-response analysis for 197 AVM patients treated with SRS who had at least 3 years of post-treatment angiographic follow-up.²⁵ The median target volume was 4.1 cm³, and the median treatment parameters were a minimum target dose (i.e., margin dose) of 20 Gy, maximum dose of 36 Gy, and isodose line of 50%, and two isocenters. AVM obliteration was achieved in 72% of cases. Of the 55 AVMs that did not undergo obliteration, 35 patients (64% of patent AVMs, 18% of overall study cohort) were determined to have a residual lesion due to persistent filling of untargeted portions of the original nidus. In the multivariate analysis for in-field obliteration, only margin dose (p = 0.04) was found to be an independent predictor.¹ Additionally, the authors constructed a sigmoid dose-response curve for obliteration, with margin doses of 13, 16, 20, and 25 Gy corresponding to obliteration rates of 50, 70, 90, and 98%, respectively. Margin dose was not significant in the multivariate analysis for overall AVM obliteration, in which only nidus volume (p < 0.001) was found to be predictive.

Spetzler-Martin Grade

Although the Spetzler–Martin (SM) grading system was originally devised with the intention of predicting outcomes after surgical resection of AVMs, it has also been shown to predict outcomes after SRS.²⁶ Andrade-Souza et al evaluated the outcomes of 136 AVMs treated with LINAC SRS and reported excellent outcome (complete AVM obliteration without new or worsening neurological deficit) in 89% of SM grade I AVMs, 70% of grade II AVMs, 62% of small grade III AVMs (diameter < 3 cm), and 45% of large grade III (diameter ≥ 3 cm) and grade IV AVMs.²⁷ Koltz et al analyzed a cohort of 102 AVM patients who underwent single session or dose-staged SRS and had at least 5 years of follow-up.²⁸ After a mean follow-up of 8.5 years, the obliteration rates, stratified by SM grade, were 100, 89, 86, 54, and 0% for grades I, II, III, IV, and V, respectively. The combined rate of major morbidity and mortality for SM grades I, II, III, IV, and V were 20, 11, 9, 18, and 75%, respectively.

In a cohort of 502 patients with SM grade I and II AVMs (median volume of 2.4 cm³) treated with SRS (median margin dose of 23 Gy) at our center with median radiologic and clinical follow-up durations of 48 and 62 months, obliteration was achieved in 76%, with actuarial rates of 41, 66, and 80% at 3, 5, and 10 years, respectively.²⁹ Radiologic, symptomatic, and permanent ARE developed in 37, 8, and 1%, respectively, and the annual rate of AVM hemorrhage during the post-SRS latency period was 1.4%. Kano et al reported the outcomes of 217 SM grade I and II AVMs (median volume of 2.3 cm³) treated with SRS (median margin dose of 22 Gy).³⁰ After a median follow-up of 64 months, the actuarial obliteration rate was 58, 90, and 93% at 3, 5, and 10 years, respectively. Symptomatic ARE developed in only 2% of patients, and all cases were transient. The annual post-SRS hemorrhage rate was 2.3%, and was significantly higher in AVMs with an associated arterial aneurysm.

In a cohort of 398 patients with SM grade III AVMs (median volume of 2.8 cm³) treated with SRS (median margin dose of 20 Gy) at our center with median radiologic and clinical follow-up durations of 54 and 68 months, obliteration was achieved in 69%, with actuarial rates of 38 and 60% at 3 and 5 years, respectively.³¹ Radiologic, symptomatic, and permanent ARE developed in 35, 12, and 4%, respectively, and the annual rate of AVM hemorrhage during the post-SRS latency period was 1.7%. Kano et al analyzed the outcomes of 474 SM grade III AVMs (median volume of 3.8 cm³) treated with SRS (median margin dose of 20 Gy).³² After a mean follow-up of 89 months, the actuarial obliteration rate was 48, 72, and 77% at 3, 5, and 10 years, respectively. Symptomatic and permanent ARE developed in 6 and 3% of patients, respectively. The annual post-SRS hemorrhage rate was 2.7%.

In a cohort of 110 patients with SM grade IV and V AVMs (median volume of 5.7 cm³) treated with SRS (median margin dose of 19 Gy) at our center with median radiologic and clinical follow-up durations of 88 and 97 months, obliteration was achieved in 44%, with actuarial rates of 10 and 23% at 3 and 5 years, respectively.³³ Radiologic, symptomatic, and permanent ARE developed in 47, 12, and 3%, respectively, and the annual rate of AVM hemorrhage during the post-SRS latency period was 3.0%.

Radiosurgery-Based Arteriovenous Malformation Score

Pollock et al evaluated the outcomes of 220 AVM patients who underwent SRS.³⁴ Multivariate analysis found that smaller AVM nidus volume (p = 0.003), fewer number of draining veins (p = 0.001), younger patient age (p = 0.0003), hemispheric AVM location (p = 0.002), and lack of prior AVM embolization (p = 0.02) were independent predictors of excellent outcome, which was defined in this study as complete AVM obliteration without new neurological deficits. Based on this analysis, Pollock and Flickinger developed and externally validated the radiosurgery-based AVM score (RBAS), which comprised the variables AVM volume, patient age, and nidus location, to predict excellent outcome after AVM SRS.³⁵ The nidus location component of the RBAS was subsequently simplified into a two-tiered variable (deep vs. superficial) in the modified RBAS, wherein deep location comprised basal ganglia, thalamus, and brainstem.³⁶ The most recent version of the modified RBAS was described by Wegner et al and utilizes the same patient and AVM characteristics as the original RBAS, except the coefficient multiplier for the AVM location component was changed from 0.3 to 0.5.³⁷

The RBAS has been shown to correlate with SRS outcomes in AVM cohorts from numerous centers.²⁷ Recently, Burrow et al analyzed the outcomes of 80 patients with an RBAS of 1 or lower (mean 0.76).³⁸ After a mean follow-up of 68 months, no patients experienced AVM hemorrhage, ARE, or a decline in modified Rankin scale (mRS) score. The obliteration rate for patients with at least 3 years of follow-up was 92%. The authors suggested that SRS may achieve comparable outcomes to resection for younger patients with superficial, small-volume AVMs who do not require surgical evacuation of a hematoma.³⁸

However, the negative effect of patient age on AVM SRS outcomes, as suggested by the RBAS, is inconsistent. We performed a matched cohort study of 66 elderly AVM patients (age > 60 years) who underwent treatment with SRS and compared their outcomes to those of nonelderly patients, matched in a 1:1 ratio, with comparable AVMs.³⁹ The elderly AVM cohort had a significantly higher mean age (67 vs. 36 years, p < 0.0001), RBAS (1.70 vs. 1.11, p < 0.0001), and number of SRS isocenters (2.9 vs. 2.7, p = 0.038) compared to the nonelderly AVM cohort. Elderly age was not significantly associated with obliteration, ARE, or AVM hemorrhage after SRS. In summary, advanced age may not adversely affect AVM SRS outcomes, which contrasts with its negative impact on AVM surgical outcomes.³⁹

Virginia Radiosurgery Arteriovenous Malformation Scale

We developed the Virginia Radiosurgery AVM Scale (VRAS) by analyzing our institutional Gamma Knife SRS experience of over 1,400 AVM patients treated over a period of 20 years.⁶ Patients with at least 2 years of radiologic follow-up or those with less than 2 years of follow-up who developed treatment-related complications were selected, yielding 1,012 patients for analysis. Favorable outcome was defined as AVM obliteration, no post-SRS hemorrhage, and no permanently symptomatic ARE. The mean age was 34 years, and 56% had prior AVM hemorrhage. The mean AVM volume was 3.5 cm³, including < 2 cm³ in 20%, 2 to 4 cm³ in 48%, and > 4 cm³ in 32%. AVM location was eloquent in 67%, and 52% had a component of deep venous drainage. The SM grade was III or higher in 48%, and the mean RBAS was 1.35. The mean SRS margin dose was 21 Gy.

After a mean follow-up of 8 years, favorable outcome was achieved in 64% of patients. In the multivariate analysis of only patient and AVM variables (i.e., excluding SRS treatment parameters), age < 65 years (p = 0.041), smaller AVM volume (p < 0.001), noneloquent AVM location (p < 0.001), lack of prior AVM hemorrhage (p < 0.001), and lack of prior AVM embolization (p < 0.001) were found to be independent predictors of favorable outcome after SRS. Based on the significant factors in the multivariate model, the five-tiered VRAS was constructed, comprising AVM volume (<2 cm³ = 0 points, 2–4 cm³ = 1 point, > 4 cm³ = 2 points), eloquence of AVM location (noneloquent = 0 points, eloquent = 1 point), and history of AVM hemorrhage (unruptured = 0 points, ruptured = 1 point). The rates of favorable outcome for a VRAS of 0 to 1, 2, and 3 to 4 were 80, 70, and 45%, respectively.⁶

Since the VRAS is relatively new compared to the SM grading scale and RBAS, it has yet to be subjected to the same rigorous external testing as these two classification schemes. Recently, Huo et al analyzed a cohort of 162 patients with partially embolized AVMs who underwent SRS.⁴⁰ The VRAS was found be predictive of AVM obliteration (VRAS of 0–1, 2, 3, and 4 resulted in obliteration rates of 89, 68, 51, and 35%, respectively) and post-SRS complications, including hemorrhage, seizure, and headache (VRAS 0–2, 3, and 4 resulted in complication rates of 8, 24, and 29, respectively). We have performed an external validation of the VRAS in a multicenter cohort of over 2,000 AVM patients treated with SRS. The study was performed under the auspices of the International Gamma Knife Consortium, and the findings from this study demonstrate superiority of the VRAS over SM or RBAS systems in terms of predicting SRS outcomes in AVM patients. While SM grade and RBAS were significantly associated with AVM SRS outcomes, the VRAS was found to be the best predictor among the three grading systems.

25.2.2 Outcome of SRS for Different AVM Types

Primary Motor and Somatosensory Cortex AVMs

AVM nidus location carries tremendous prognostic significance when evaluating microsurgical or endovascular outcomes. Specific outcome data for primary motor or somatosensory cortex (PMSC) AVMs is limited for all treatment modalities, including the radiosurgical literature.⁴¹ We previously reported a series of 134 patients with PMSC AVMs who underwent SRS with median radiographic and clinical follow-up durations of 64 and 80 months, respectively.⁴¹ The most common presenting symptoms were seizures (40%) and hemorrhage (28%), and 34% underwent pre-SRS embolization. The median AVM volume was 4.1 cm³ (range: 0.1 –22.6 cm³), and the median margin dose was 20 Gy (range: 7–30 Gy).

The overall obliteration rate, determined by angiography or MRI, was 63%, stratified as 80% for small AVMs (volume < 3 cm³) versus 55% for larger AVMs (volume > 3 cm³). In the multivariate analysis, the lack of prior embolization (p = 0.002) and a single draining vein (p = 0.001) were found to be independent predictors of obliteration.⁴¹ The cumulative obliteration rate for our cohort of PMSC AVMs was comparable to those of other series in the SRS literature (61–70%).²⁸^,⁴¹^,⁴² The obliteration rates, stratified by nidus size, were also similar to those of prior studies (83–87% for smaller AVMs with volume < 3 cm³ and 50–56% for larger AVMs with volume > 3 cm³).²⁸^,⁴¹^,⁴² The annual post-SRS hemorrhage risk of our PMSC AVM cohort was 2.5%, and SRS-related morbidity was transient and permanent in 14 and 6%, respectively. When we compared these PMSC AVMs to a matched cohort of noneloquent, lobar AVMs, we did not find statistically significant differences between the obliteration rates and clinical outcomes of the two cohorts.⁴¹ SRS appears to offer a reasonable risk-to-benefit profile for these challenging lesions. Furthermore, eloquent location does not appear to confer the same negative prognostic value for SRS that it does for resection or embolization.⁴¹ A literature review is given in ► Table 25.1.