32 Stereotactic Radiosurgery for Arteriovenous Malformations of the Basal Ganglia, Thalamus, and Brainstem

10.1055/b-0039-173923

32 Stereotactic Radiosurgery for Arteriovenous Malformations of the Basal Ganglia, Thalamus, and Brainstem

Or Cohen-Inbar and Jason P. Sheehan

Abstract

Management of arteriovenous malformations (AVMs) in the basal ganglia, thalamus, and brainstem is a formidable challenge because these lesions carry a higher risk of rupture and bleeding compared to AVMs in other locations, and significant morbidity and mortality are associated with such ruptures. Adding to the challenge is the fact that any of the available treatment modalities may yield neurologic deterioration. Stereotactic radiosurgery (SRS) has an important role to play in the management of AVMs in the basal ganglia, thalamus, and brainstem that are deemed too risky for surgical removal. AVMs, especially in the ventral midbrain, pons, and medulla oblongata, as well as in the lateral thalamus where total surgical removal is difficult, are good candidates for SRS. SRS offers acceptable obliteration rates with lower risks of hemorrhage during the latency period. Complex lesions and nidi require complex multidisciplinary treatment. Large AVMs or lesions partly involving the subpial or epipial areas of the diencephalon, dorsal midbrain, or cerebellopontine angle should be considered for a combination of endovascular embolization, micro-surgical resection, and SRS. Because incompletely obliterated lesions could cause lethal hemorrhages, additional treatment (e.g., repeat SRS and surgical resection) may be required. Continued clinical and radiologic observation and follow-up are mandatory, even when angiographic obliteration has been confirmed after SRS.

Introduction

Arteriovenous malformations (AVMs) are rare congenital vascular lesions that occur with equal incidence in both sexes and that are typically diagnosed by early adulthood (third decade of life). ¹ The incidence of AVMs, as determined by population-based studies, is on the order of 1:100,000. ¹ ^, ² The defining feature of these vascular malformations is direct arterial-to-venous connections without an intervening capillary network. ³ This abrupt transition from a high-pressure muscular arterial system to a low-pressure venous system leads to intracerebral venous dilatation, engorgement, and vessel wall arterialization, with resultant edema and irritation of the surrounding brain tissue. These processes and pressure challenges predispose the patient to parenchymal bleeding from the nidus or associated aneurysmal changes in arteries and veins. ¹ ^, ² ^, ³ ^, ⁴ ^, ⁵ ^, ⁶

The annual hemorrhage risk of AVMs in all locations is 2 to 4%. ¹ ^, ² ^, ³ ^, ⁴ ^, ⁵ ^, ⁶ The combination of a relatively early age of presentation and this nontrivial estimated annual hemorrhage risk leads to the expectation of a high lifetime risk of morbidity from untreated AVMs. ¹ ^, ⁴ In one report, annual hemorrhage rates ranged from a low of 0.9% for low-risk patients (defined as those without prior hemorrhagic AVM presentation, deep AVM location, or deep venous drainage) to a high of 34.4% for high-risk patients with these three risk factors. ⁵ Such deep AVMs in the basal ganglia, thalamus, and brainstem are believed to have a more aggressive natural history ⁶ and hemorrhage-associated mortality of up to 62.5%. ⁷

While the most common presentation of AVMs remains intracerebral hemorrhage (particularly for patients with deep-seated lesions who do not present with seizure), most AVMs are identified incidentally. This trend is attributed mainly to the wide availability of neuroimaging. ⁸ AVMs are also identified after patients present with seizures, headache, or focal neurologic deficits. Several observational studies have suggested that the risk of future morbidity associated with an unruptured AVM (i.e., its natural history) might differ from the risk associated with a previously ruptured one. ⁵ Much controversy remains about the management of unruptured AVMs, with some physicians stating that conservative medical management may be superior to intervention. ⁹ The ARUBA study (A Randomized Trial of Unruptured Brain AVMs) further supports this view. ¹⁰ This trial was an attempt to evaluate intervention versus best medical management for unruptured brain AVMs. Unfortunately, due to errors in study design and enrollment, the study’s conclusions should be viewed with great caution. ¹¹

The goal of treating AVMs with stereotactic radiosurgery (SRS) is to obliterate the nidus and thereby eliminate any future hemorrhage risk. Regardless of the SRS tool used (Gamma Knife [Elekta], Cyberknife [Accuray], or other linear accelerator [LINAC]–based systems), the mechanisms of AVM obliteration after radiosurgery include progressive intimal thickening, thrombosis of irradiated vessels, and eventual occlusion of the vascular lumen. ¹² The obliteration of an AVM nidus with SRS depends on several factors, including the size of the lesion nidus (and its compactness) and the maximum safe dose tolerated by the adjacent tissue. ¹ The reported success rates for complete AVM obliteration with SRS are as high as 80% at 3 years for lesions measuring less than 3 cm in largest diameter. ¹ ^, ¹³ Even with larger AVMs, lesion reduction does occur and additional treatment is effective in most cases. ¹³ Karlsson et al ¹⁴ reported an 80% overall obliteration rate in 1,319 patients treated at the Karolinska Institute. The likelihood of obliteration was calculated to be approximately 90%, 80%, and 70% for AVMs given margin doses of 20 Gy, 18 Gy, and 16 Gy, respectively.

One should note that AVMs in different locations pose different clinical challenges and can behave quite differently. SRS of deep-seated AVMs in the basal ganglia, thalamus, and brain-stem is associated with significant complications and a high incidence of hemorrhage during the latency period, ⁶ ^, ¹⁵ ^, ¹⁶ ^, ¹⁷ as well as with obliteration rates lower than those for AVMs in other locations. ⁷ ^, ¹⁶ All-location 3-year obliteration rates have been reported to range from 57% to 81%. ¹⁵ ^, ¹⁶ ^, ¹⁷ ^, ¹⁸ ^, ¹⁹

Brainstem AVMs constitute only 2 to 6% of all intracranial AVMs. ²⁰ ^, ²¹ ^, ²² The natural history of such untreated vascular anomalies indicates a high risk of major morbidity or mortality rates from hemorrhage. ¹ ^, ² ^, ¹⁹ ^, ²³ ^, ²⁴ AVMs located in the posterior fossa carry a higher risk of hemorrhage than AVMs in other locations. ²⁴ ^, ²⁵ Kiran et al ¹⁶ reported an 81% incidence of hemorrhage in 53 patients with deep-seated AVMs in the basal ganglia, thalamus, and brainstem. This bleeding risk was higher than that in patients with AVMs in other locations (67%). In addition, because of the critical location in proximity to vital neuronal pathways and nuclei, there is a high risk of morbidity and mortality when brainstem AVMs rupture. As detailed in this chapter, no single management option fits all, and the best option for the management of a specific brainstem AVM is often tailored.

Thalamic AVMs represent another special group within the cerebral vascular anomalies, constituting 4.3 to 11% of all AVMs. ⁴ ^, ²⁶ Their natural history similarly tends to be more aggressive, with an annual hemorrhage risk approaching 10%, ⁵ ^, ⁶ ^, ⁷ ^, ²⁷ compared with 2 to 4% for AVMs in general. ¹ ^, ⁴ Additionally, the age at diagnosis in these patients seems to be younger than that of patients with AVMs in other locations. ⁷ Deep-seated thalamic AVMs having nidi with deep venous drainage, as well as AVMs with a history of prior hemorrhage, are more likely to rupture. ⁵ Most (72–91%) patients with thalamic AVMs present with hemorrhage, compared with up to 50% of all patients with AVMs. ⁶ ^, ⁷ ^, ²⁶ Fleetwood et al ⁶ reported a 9.8% annual bleeding rate for 96 patients with 500 patient-years of clinical follow-up before surgical management. Sasaki et al ⁷ described an annual hemorrhage rate of 11.4% (overall rate 71.4%) in 14 conservatively managed patients and a mortality rate of 42.9% during a mean follow-up period of 6.6 years. Thalamic AVMs not only result in an increased risk of morbidity and mortality due to hemorrhage but also may cause significant steal phenomenon and mass effect because of the critical neuronal pathways and nuclei in these locations.

Predicting Stereotactic Radiosurgery Outcome

The results of radiosurgery, unlike those of microsurgery, may not fully manifest for many years after treatment. The risk of hemorrhage from an SRS-treated AVM persists to some degree during the latency period (typically 3 years) between radiosurgery and obliteration. Furthermore, conditions such as radiation-induced necrosis, edema, and cysts may develop in a delayed fashion, even after obliteration. In light of the fundamental difference between radiosurgical and microsurgical approaches, conventional grading scales characterizing AVMs may seem insufficient for prognostic purposes. Grading scales specific to radiosurgery have thus been created to predict outcome in patients who have undergone such treatment. ²⁸

The Spetzler-Martin grading system ²⁹ does little more than differentiate deep thalamic and brainstem AVMs by size, because these AVMs are typically eloquent in location and drain deeply. ¹⁵ Thus, these lesions are automatically classified as at least a grade III in the old system and as a grade B in the new system. ³⁰ The supplementary grading system of Lawton et al ³¹ adds an important anatomical dimension, namely nidal compactness. This feature is especially critical in the evaluation of SRS targets in the diencephalon or brainstem. Furthermore, the supplementary grade helps classify patients, concisely quantitating their natural history risk, neurologic condition, and recoverability with scores for hemorrhagic presentation and age. We previously reported good predictive value for the Spetzler-Martin scale for patients undergoing Gamma Knife radiosurgery (GKRS). ³² Among its advantages are its inherent simplicity and practicality. However, the Spetzler-Martin scale lacks radiosurgery-specific variables, which limits its predictive value in patients undergoing SRS.

Many grading scales have been proposed throughout the years in an attempt to predict patient outcome after radiosurgery. ¹⁴ ^, ²⁸ ^, ³² ^, ³³ ^, ³⁴ These scales are limited to predicting obliteration based solely on treatment parameters (radiation dose). ¹⁴ ^, ²⁸ Most of these scales fail to account for the fact that an increased radiation dose increases the likelihood not only of obliteration but also of radiation-induced complications; thus, the scales are not predictive of overall patient outcome. ³³ The potential benefit of increasing the radiation dose (higher obliteration rates) must be weighed against the increased morbidity that comes with it. ³⁵ Flickinger et al ³⁶ demonstrated an increasing rate of permanent radiosurgery-induced deficits as the volume of tissue receiving 12 Gy increased. The same group reported a lower rate of radiographic radiation-induced changes (RICs) but comparable rates of symptomatic RICs. ³⁵

Pollock and Flickinger ³³ developed the radiosurgery-based AVM scale (RBAS) in an attempt to provide an estimate of the rates of nidus obliteration, hemorrhage, and radiation-related complications ( Table 32.1 ³³ ^, ³⁴ ). The RBAS was based on an outcome analysis of 356 patients, 56% (n = 199) of whom harbored deep AVMs in the basal ganglia, thalamus, or brainstem. It is important to note that this system was created to predict patient outcome after a single radiosurgery procedure and not the overall results of radiosurgical management. In addition, because complications after AVM radiosurgery, such as cyst formation, may occur many years after an angiographically verified obliteration, the final results of AVM radiosurgery may change with this additional information. These complications are not accounted for in the RBAS.

**Table 32.1 Stereotactic radiosurgery scales for arteriovenous malformations ^a**
Variable for scoring	Virginia Radiosurgery AVM Scale			Radiosurgery-based AVM scale ^b
Variable for scoring	Points	Cumulative score	Favorable outcome, % ^c	Scoring components
AVM volume, cm³				0.1 × AVM volume
<2	0	0	83
2–4	1	1	79
>4	2	2	70
AVM in eloquent location ^d	1	3	48	0.3 × AVM location score
Frontal, temporal				Location score = 0
Parietal, occipital, intraventricular, corpus callosum, cerebellar				Location score = 1
Basal ganglia, thalamus, brainstem				Location score = 2
History of hemorrhage	1	4	39	NA
Patient age, y	NA	NA	NA	0.02 × patient age
Abbreviations: AVM, arteriovenous malformation; NA, not applicable. ^aData for the Virginia Radiosurgery AVM Scale (V-RAS) are from Starke et al 2013 ³⁴ and data for the radiosurgery-based AVM scale (RBAS) are from Pollock and Flickinger 2002. ³³ ^bRBAS score was defined as the sum of the scoring components. ^cFavorable outcome was defined in the V-RAS both as AVM obliteration and as no post-treatment hemorrhage or permanent symptoms associated with Gamma Knife radiosurgery. ^dFor the RBAS, when an AVM involves multiple sites, fractional values are used according to the number of sites (0.5 for two sites; 0.33 for three sites).

The Virginia Radiosurgery AVM Scale (V-RAS) ( Table 32.1 ), initially described by Starke el al, ³⁴ was based on an analysis of 1,012 patients treated with GKRS for AVMs in different locations. Basal ganglia, thalamus, and brainstem locations were affected in 26.1% (n = 264) of these patients. A favorable outcome was defined as including both AVM obliteration and no post-treatment hemorrhage or permanent GKRS-associated symptoms. As shown in Table 32.1 , in the V-RAS, the eloquent location entails a minimum score of 1, which was found to be associated with a 79% chance of a favorable outcome. The V-RAS is the radiosurgery analog to the Spetzler-Martin scale. In a recent multicenter study of 2,236 AVM patients conducted through the International Gamma Knife Research Foundation, the V-RAS scale proved the most predictive of AVM outcomes after SRS. ³⁷

The Role of Pre-Stereotactic Radiosurgery Embolization for Deep-Seated Arteriovenous Malformations

The decision to partially embolize an AVM prior to SRS usually is made when an AVM is too large (e.g., > 3 cm in its maximum diameter or 15 cm³ in volume) to treat with radiosurgery alone in a single session. Therefore, we rely on embolization to reduce the AVM nidus to a targetable size for single-session SRS. ³⁸ When counseling patients, we must keep in mind that AVM endovascular embolization carries a definite risk of complications that may exceed 10%. ³⁹ ^, ⁴⁰ ^, ⁴¹ Complications may include microcatheter entrapment in cyanoacrylate glue or Onyx (Medtronic) liquid embolic agent, microcatheter-induced arterial perforation, AVM hemorrhage due to premature draining vein occlusion, and stroke secondary to reflux of embolic agents into unintended vessels.

Embolization has previously been reported to decrease the radiosurgical obliteration rate of AVMs. ¹³ ^, ¹⁷ ^, ³³ While it stands to reason that AVMs that require pre-SRS embolization are, on average, larger than those not requiring embolization, we previously reported that embolization was found in a multivariate analysis to be an independent negative predictor of obliteration (p < 0.001). ³⁴ For previously embolized AVMs, the embolized portion of the nidus is generally not irradiated; rather, focus is placed on the remainder of the patent nidus. However, unfortunately, embolization does not guarantee flow cessation, and previously treated arterial feeding vessels may recanalize. ⁴² Embolization can further make radiosurgical planning more difficult if it fails to wall off a specific sector of the nidus, potentially transforming a compact nidus into a diffuse multisegmented nidus.

Evaluation of Arteriovenous Malformation Obliteration

Cerebral angiography is the best method to confirm complete obliteration of an AVM after SRS. Yet, as an invasive procedure, angiography can be difficult to use as a routine follow-up tool in some patients after SRS. Thus, most neurosurgeons prefer magnetic resonance imaging (MRI) or magnetic resonance angiography (MRA) for follow-up. Lee et al ⁴³ reported on the sensitivity and specificity of MRI and MRA in evaluating AVM nidus obliteration compared with digital subtraction angiography. The sensitivity (the probability that MRI and MRA correctly demonstrate obliteration as confirmed by angiography) was 84.9% and 76.7% for two independent observers. The specificity was 88.9% and 95.2% for two independent observers. The authors concluded that MRI and MRA predicted AVM obliteration after SRS in most patients and can be used in their follow-up. However, because the specificity of MRI and MRA is not perfect, digital subtraction angiography should still be performed to confirm AVM nidus obliteration after SRS and in cases where suspicion for residual is high. ⁴³ Kano et al ²³ demonstrated that MRI and MRA significantly overestimate the obliteration rate compared to digital subtraction angiography, increasing the 5- and 10-year obliteration rates by 5% (from 64% to 69%). The authors concluded that the potential overestimation of obliteration rates as determined by MRI is balanced by the tendency to underestimate long-term obliteration rates that are based only on early follow-up examinations. ²³

Many authors currently use MRI evidence of obliteration to evaluate the outcomes of SRS. ¹⁵ ^, ¹⁸ ^, ²³ ^, ³² ^, ⁴⁴ ^, ⁴⁵ Actuarial calculations for AVM obliteration defined by MRA or formal angiography using corrected obliteration methods provide the most accurate portrayal of AVM obliteration. In our institute, routine follow-up is performed using MRA. During the first 2 years after SRS, we recommend biannual clinical and radiologic follow-up, unless clinical indications (e.g., patient complaints or development of symptomatic RICs) mandate more frequent follow-up. Upon MRI proof of obliteration, coinciding with a reasonable latency period from the time of SRS (depending on the size of the nidus and treatment parameters), confirmatory MRA is performed.

Brainstem Arteriovenous Malformations

Microsurgery

Microsurgery remains the first-line preferred treatment for most AVMs. Despite advances in microsurgical techniques and intraoperative monitoring, lesions involving any part of the brainstem pose a challenge for neurosurgeons. Surgical trials spanning the past 25 years that were aimed at extirpating brainstem AVMs resulted in generally poor outcomes. These reports deterred even confident neurosurgeons with extensive experience in operating on deep-seated AVMs from performing microsurgery for brainstem nidi. One important exemption is the ruptured subpial nidus because the hemorrhage may facilitate the creation of a dissection plane, resulting in fewer deficits after ablation. ²⁰ ^, ²⁵ Microsurgery for brainstem intraparenchymal AVMs thus remains a formidable challenge. ²⁰ ^, ²¹ ^, ²⁵ This topic is discussed in greater detail in the chapter on endovascular management of aneurysms of the posterior circulation (Chapter 27).

Drake et al ²⁰ reviewed a series of 15 patients, of whom only 2 had AVMs with the main nidus in an extrapial location where it could be safely resected. Four patients died of postoperative hemorrhage. Nozaki et al ²⁵ reported the outcomes for 25 patients with brainstem AVMs. Microsurgical extirpations were attempted in 19 patients, with total resection of the AVMs achieved in 14 patients (6 in dorsal midbrain, 2 in pons, and 6 in the cerebellopontine angle). The authors reported a surgical morbidity rate of 25% and no mortality. They noted that dorsal midbrain and cerebellopontine angle AVMs, located mainly in the subpial or epipial spaces, were suitable for resection and that previous hemorrhage facilitated the microsurgical dissection. Solomon and Stein ²¹ reported a 75% (9 of 12 patients) rate of complete resection of brainstem AVMs, with a 22% morbidity rate. In contrast, Lawton et al ⁴⁶ reported no serious neurologic deterioration in eight patients with brainstem AVMs that were successfully resected, which likely indicates careful patient selection.

Only gold members can continue reading. Log In or Register to continue