28 Treatment of Inoperable Cerebral Arteriovenous Malformations
This chapter on the treatment of “inoperable” arteriovenous malformations (AVMs) critically evaluates what constitutes an “inoperable” AVM and how to best manage patients with this pathology. Due to continued technological and technical advances, “inoperable” AVMs constitute a shrinking subset of an already uncommon pathology. There are two types of inoperable AVMs. Type 1 lesions are not amenable to microneurosurgical resection because of a prohibitively high-risk profile, but they may be readily treatable by radiosurgery and/or endovascular neurosurgery. Type 2 lesions are too risky for treatment by any modality unless the patient is experiencing severe neurologic dysfunction or has suffered a hemorrhage. This chapter explores not just the evidence, but also the nuances involved in making safe decisions about how to optimize the care of these patients.
Keywords: inoperable arteriovenous malformations, radiosurgery, endovascular therapy, arteriovenous malformation treatment, lesion anatomy
- The successful treatment of “inoperable” arteriovenous malformations (AVMs) requires that a neurosurgeon shift away from the classic paradigms applied in a traditional, open microneurosurgical approach.
- Continual advances in technology and surgical technique, as well as the dynamic nature of lesions, necessitate constant reevaluations of patients with “inoperable” AVMs to ensure that their lesion has not shifted into the classification of a treatable AVM.
- Multimodality treatment of an “inoperable” AVM may provide a cure without subjecting a patient to the same risk that they would incur with treatment by microneurosurgical techniques or an isolated endovascular or radiosurgical approach.
Open neurosurgical resection has been the gold standard for durable obliteration of arteriovenous malformations (AVMs) since inception of the specialty. Although open techniques have evolved somewhat with technological innovations, including the surgical microscope, improved noninvasive and angiographic imaging, and neuronavigation, the tenets have remained relatively constant. It is the standard for which newer treatment modalities (i.e., radiosurgery and endovascular therapy) are compared—with regard to both risk profile and treatment outcomes. These newer therapies have armed neurosurgeons with less invasive treatment options. The decision points for the surgeon then become more complex: not only should an AVM be treated, but also how.
The Spetzler–Martin AVM grading scale uses anatomical considerations to assess risk for surgical treatment of AVMs (► Table 28.1).1 Although widely used to describe AVMs in general, its applicability to treatment with radiosurgery and/or endovascular surgery is less clear, and the morbidity and mortality considerations associated with open resection do not apply with non-microsurgical options.2,3
Table 28.1 Spetzler–Martin classification of cerebral arteriovenous malformations
Small (< 3 cm)
Medium (3–6 cm)
Large (> 6 cm)
Eloquence of adjacent brain
Venous drainage pattern
Source: Adapted from Spetzler and Martin.1
Note: AVM Grade = [size] + [eloquence] + [venous drainage]; that is, [1, 2, or 3] + [0 or 1] + [0 or 1].
What truly constitutes an “inoperable” AVM? Is there a subset of patients who might be appropriate for treatment (by any modality) despite what might classically be seen as having a prohibitive risk profile? Certain AVMs might be considered “inoperable” upon first evaluation. However, patients suffering devastating or recurrent hemorrhage or developing neurologic deficits can quickly force the issue and tip the balance toward moving forward with treatment in spite of a high-risk profile. Therefore, it behooves the vascular neurosurgeon to have a sense of how to manage these difficult AVMs. This chapter presents the evidence for considering treatment of AVMs that might otherwise be considered “inoperable” and describes less conventional approaches to treatment that may be particularly germane to these lesions.
28.2 Materials and Methods
When approaching an intricate topic such as the treatment of “inoperable” AVMs, one must incorporate a number of aspects into their knowledge base to ensure an adequate understanding of the subject. The information provided in this chapter is based on published articles, editorials, and case-specific examples. The remainder of this chapter is devoted to providing a concise summary of this information and how it pertains to the management of “inoperable” AVMs.
28.3.1 Developing a Frame of Reference for “Inoperable” AVMs
Simply proceeding to write about how “inoperable” AVMs should be treated is putting the cart before the horse. “Inoperable” is a hazy categorization that would likely create significant discussion among even the most experienced surgeons. Neurosurgeons are acutely aware that almost any surgery could be performed if outcomes were to be disregarded, but we also recognize that to do so would be an entirely unacceptable reality. Therefore, we must reconcile what we consider an “inoperable” AVM before discussing how to treat the pathology.
The mental hurdle vascular neurosurgeons must overcome when managing “inoperable” AVMs relates to the juxtaposition of the terms treatment and inoperable. It is critical to stress that observation, although not a proactive form of treatment, should be considered a valid management option; and expectant management of lesions that are felt to have more risk associated with intervention (by any modality) than with observation is entirely reasonable. We urge readers to keep this management option in mind as they proceed through this chapter.
The A Randomized trial of Unruptured Brain Arteriovenous malformations (ARUBA) trial4 has created considerable dissent among physicians regarding the best course of treatment for patients with AVMs by reporting that medical management is superior to intervention. However, it is important to point out that this trial’s results should not be extrapolated to the patient with an “inoperable” AVM for several reasons. Firstly, the trial was stopped prematurely, and as such, long-term results are as of yet unknown. Many feel that the report of medical superiority may erode with time, and it is possible that quite different conclusions may become apparent upon completion of the 10-year follow-up period. Secondly, several issues have been raised with the study’s methodology. These issues have been outlined in editorials and evaluated by further studies. Some of the major points are as follows: the disease process is very heterogeneous, which stresses the rigid constraints of the purported randomized controlled trial; there was a bias toward nonsurgical therapy and no data published on cure rates. Furthermore, among patients undergoing intervention, only a small minority of cases involved surgical resection, with the remainder split among alternatives that are known to be noncurative in the short term. This suggests that the proportion of incompletely obliterated AVMs was high and, hence, even “treated” lesions continued to contribute to ongoing ruptures; centers only had to manage 10 patients with AVMs per year to be eligible for enrollment, and there were no minimum requirements for the treating neurosurgeons, which calls into question the microneurosurgical expertise applied to this population.5,6,7 ARUBA included no patients with a Spetzler–Martin grade higher than IV. In fact, most patients (62%) had scores of 2 or less, and, in accordance with their inclusion criteria, all patients had a modified Rankin scale score of 1 or lower.4 Finally, ARUBA, being a trial aimed at unruptured lesions, excluded AVMs with evidence (symptomatic or not) of hemorrhage. Hemorrhage is one of the factors that may shift the risk–benefit equation for patients with “inoperable” AVMs and lead the neurosurgeon to reconsider treatments that convey some risk. For these reasons, completely dismissing the initiation of the treatment of a patient with an “inoperable” AVM on the basis of the ARUBA trial would be a miscalculation.
We posit that AVMs with a severe treatment risk profile should only be considered for surgical intervention if they have failed conservative management in one of two ways:
- The patient is suffering severe, progressive symptoms attributable to the AVM that, when compared to the risk profile for treatment, make continued observation unacceptable.
- The patient experiences an AVM rupture that portends a riskier course for continued observation. There is significant nuance to this point that will be discussed below.
The fear of intracranial hemorrhage and its sequelae is a driving force behind decision-making for both patients and their treating surgeons. AVM-associated hemorrhage alone is not grounds for moving forward with treatment. There is evidence that patients who suffer AVM hemorrhage (in the form of either a primary or a recurrent bleed) may not have as poor a neurologic outcome as matched cohorts who experience spontaneous intracranial hemorrhage without an associated vascular anomaly.8,9 There is also evidence, though, that higher grade AVMs (such as our “inoperable” cases) may have a more aggressive course and catastrophic outcome when they bleed primarily or recurrently.10 However, we also know that patients who experience AVM ruptures have a higher risk of recurrent hemorrhage than those patients in whom the AVM remains unruptured.8,11 Therefore, we maintain that rupture is a valid reason to proceed with treatment of an “inoperable” AVM only if the surgeon has weighed the patient’s status and risk of future hemorrhage, neurologic morbidity, and mortality appropriately against the great risk that will be incurred with treatment of the AVM.
28.3.2 Salient Points for Radiosurgical or Endovascular Treatment
There are certain considerations that a treating physician must weigh if a patient is not a surgical candidate, but their AVM is felt to be amenable to endovascular therapies, radiosurgery, or a combination of the two. If intervention and/or radiosurgery are/is being considered, the surgeon must exquisitely understand the anatomy of the lesion. We feel that a vital part of this endeavor is performing a diagnostic cerebral angiogram before making any formal treatment decisions. At times, the lesion’s complexity may even justify microcatheter exploration (without treatment) at the time of the initial angiogram so that the anatomy of specific feeding pedicles can be elucidated. This anatomy can then be used to evaluate the risk of treatment based on the type of therapy that is being considered and provide time for the treatment team to formulate a thoughtful treatment plan that is tailored to the specifics of the patient and his/her lesion.
As mentioned earlier, the classically applied Spetzler–Martin grading scale, although more widely understood and therefore useful for discussion, is not ideal for evaluating the treatment risk of an AVM if the intended modality is radiosurgery or endovascular embolization. To address this need, several modality-specific scores have been developed.
The Pollock–Flickinger score3,12 evaluates factors that are more specific/critical to radiosurgery than those evaluated in the Spetzler–Martin grade. Volume, age, and location (superficial vs. deep) can be used to predict the likelihood of AVM obliteration as well as functional decline.12 In a more contemporary approach, the University of Virginia AVM grading system has also been shown to be very predictive of stereotactic radiosurgery outcome of AVM patients. Stereotactic radiosurgery has been applied to the treatment of large, inoperable AVMs by using either hypofractionated dosing or volume staging, with the literature suggesting that the latter technique may be more effective.13,14,15,16 Another technique describes the use of volume-staged radiosurgery to obliterate or downsize AVMs so that the residual lesion can be operated in a safe manner.16
Radiosurgery is an attractive option for inoperable AVMs, but it is not without its own risks. It is well known that the risk of hemorrhage remains during the time that it takes for the occurrence of AVM obliteration. The literature raises several other concerns in regard to stereotactic radiosurgery. A recent metaanalysis of large AVMs treated with radiosurgery found that the mortality rate from hemorrhage was 40.08% with a 6.1% annual risk of re-hemorrhage for patients who presented with rupture.17 Furthermore, the medical community’s understanding of secondary harms from stereotactic radiosurgery continues to evolve, and yet, for younger patients being considered for treatment, physicians do not fully know the risks they incur over a long expected lifespan. One such issue is that of secondary neoplasms. The supportive evidence is far from certain, but even the possibility of an association between radiation and secondary tumorigenesis should be enough to at least create a moment of pause when contemplating proceeding with therapeutic radiation in children and young adults.18,19,20,21,22,23 Other more clearly defined sequelae of stereotactic radiosurgery include radiation necrosis, perilesional ischemic strokes, and cyst development.24,25,26 For AVM treatment in particular, occlusion of draining veins may increase the likelihood of post-radiosurgery hemorrhage.27,28 This speaks to the need to understand anatomy and consider whether a large AVM with isolated drainage is appropriate for radiosurgery, and, if it is deemed appropriate, ensure that the treatment plan reflects this understanding. Overall, reported rates for radiosurgical cures are encouraging, but cure is not certain. Therefore, we feel that initiating a treatment plan that involves radiation should be weighed particularly carefully.
Endovascular practitioners have similarly developed grading schemes specific to their treatment modality. It is now understood that the Spetzler–Martin grade is not a good predictor of complications when undertaking endovascular treatment. Two recently published scales used to predict outcome for endovascular treatment of AVMs are the Buffalo score2 and the arteriovenous malformation embocure score (AVMES;29 ► Table 28.2). Neither system has been externally validated, and they have significant differences.
Table 28.2 Summary of the Buffalo score and the arteriovenous malformation embocure score (AVMES)
The AVMES is designed to stratify the risk associated with a curative embolization of cerebral AVMs. It assigns numerical values to the size of the nidus, the number of feeding pedicles, the number of draining veins, and vascular eloquence. The last variable is a corollary to the concept of eloquence evaluated by the Spetzler–Martin grade that attempts to define risk incurred by embolizations performed near critical structures. The AVMES defines vascular eloquence as a pedicle that is less than 20 mm from the internal carotid artery or the first segments of cerebral arteries that are too small for catheterization. The authors report that the AVMES has good discriminative ability for complications and complete obliteration, with increasing scores showing higher complication rates and lower obliteration rates.29
The Buffalo score also stratifies the risk of endovascular treatment of AVMs regardless of whether the intent is complete obliteration by endovascular or multimodality treatment. It scores the number of arterial pedicles, the diameter of the pedicles, and the eloquence of the nidus location. The authors directly compared the abilities of the Buffalo score and Spetzler–Martin grade to predict complication rates and found that increasing Buffalo score (but not increasing Spetzler–Martin grade) was strongly correlated with an increased complication rate. The Buffalo score does not correlate with complete endovascular obliteration, which is an important differentiation with the AVMES.2
If the goal of treatment is to achieve complete obliteration of an AVM in the safest possible way, stand-alone embolization is often not the correct decision. Although the AVMES does correlate with complications, untoward outcomes still occurred in a patient population that underwent endovascular treatment with the stated goal of cure rather than as an adjunctive therapy. This potentially leads to more aggressive embolizations than those that might be undertaken if the goal is not necessarily an isolated embocure but rather a multimodality cure. In this sense, the Buffalo score is more nimble and may provide a more practical way of assessing the portion of risk associated with AVM embolization when a multimodality treatment is planned from the outset.
28.3.3 Defining and Classifying “Inoperable” Arteriovenous Malformations
On the basis of the points raised above, it behooves us to crystallize how we define an “inoperable” AVM so that readers fully understand our frame of reference when deciding on treatment. There are two types of AVMs that we will consider for the purposes of “inoperable” AVMs.
The first subtype is a lesion that, by any surgical standard and for any reason (such as location and complexity), is not amenable to treatment by microneurosurgical resection (either primarily or in concert with preoperative embolization) but that may be appropriate for treatment by endovascular embolization alone, radiosurgery alone, or a combination of embolization and radiosurgery. This subtype should not be viewed to include a lesion that may be reasonable for microneurosurgery but is felt to be better suited for embolization or radiosurgery—it should only include lesions that are felt to be truly unsuitable for open resection. Treatment may still be appropriate to undertake regardless of presentation (i.e., inclusive of both asymptomatic and symptomatic/hemorrhagic AVMs) if a non-microneurosurgical intervention provides a reasonable chance for cure. To simplify the ensuing discussion, we will refertothisentityasa “type 1 inoperable AVM.”
A “type 2 inoperable AVM” is a lesion that, at least if it presents in an asymptomatic fashion, would be judged unsuitable for treatment regardless of modality. In these cases, the patient would only receive treatment if his/her symptoms were to have failed conservative management in one of the two ways previously described—progressive, severe symptoms, or hemorrhage.
This classification relies on the premise that, despite a plethora of data and classification schemes, sometimes physicians are left to approach a treatment decision based on a gestalt of the patient’s status, the anatomy of the AVM, and the ability of one’s surgical/interventional skills. In the case of many patients’ AVMs, physicians’ choices for management could vary widely on the basis of the surgeon’s skills and biases. However, we still feel that there are lesions that would generally be judged to fall into the subtypes outlined earlier, regardless of which surgeon evaluates the patient.
In the following section, we present several case examples to outline lesions that we feel would almost uniformly be judged to fit into the described subtypes and that are managed with various non-microneurosurgical treatments. Where appropriate, we will also provide references and evidence for treatment modalities in regard to how they relate to the case examples.
28.3.4 Case Examples for the Treatment of “Inoperable” AVMs
Type 1 Inoperable AVMs
Case 1 —A 48-year-old man presented with seizures. Imaging workup revealed an unruptured left frontoparietal Spetzler–Martin grade V AVM with left middle cerebral artery (MCA), lenticulostriate, pericallosal, and callosal marginal arterial feeding pedicles. The lesion was deemed too risky for operative intervention due to proximity to the motor, premotor, and speech cortex. After discussion of treatment options, the patient decided to undergo multiple rounds of angiographic embolization (angioembolization). A total of 10 rounds of embolization with Onyx-18 (Covidien, Irvine, CA, United States) and Trufill (Trufill n-BCA [n-butyl cyanoacrylate] Liquid Embolic System; Codman Neuro, Raynham, MA, United States) were performed with near-complete obliteration of the lesion. (We sometimes choose to perform embolizations with n-BCA if the target vessels are too fine to allow distal catheterization.) After these embolizations, only a small anterior remnant fed by lenticulostriate pedicles remained (► Fig. 28.1). Over the course of the embolizations, the patient experienced transient episodes of sensory dysfunction in his right arm and leg, as well as dysarthria and hair loss, but these symptoms improved significantly with time.
Fig. 28.1 Case 1: This patient underwent 10 rounds of embolization of a Spetzler–Martin grade V left frontoparietal arteriovenous malformation (AVM). (a–c) Sequential images from a lateral angiogram demonstrating the residual AVM after four rounds of embolization with Onyx-18 and Trufill (Trufill n-BCA [n-butyl cyanoacrylate]). (d) Anteroposterior (AP) injection of middle cerebral artery (MCA) pedicle demonstrating residual AVM nidus following six rounds of embolization. (e,f) Sequential images of a microcatheterization and injection of a MCA feeding pedicle prior to round 8 of embolization. (g) AP angiogram demonstrating residual nidus after 10 rounds of embolization. (h) Intraoperative angiogram in lateral projection demonstrating complete obliteration of AVM nidus. Postoperative T2-weighted magnetic resonance imaging (i,j) with two slices through the AVM resection cavity and computed tomography (k,l) images at similar levels showing obliteration of AVM with no residual flow voids. Previously embolized vessels are noted posterior to the resection cavity where intraoperative mapping revealed eloquence.