Radiosurgical Treatment of Arteriovenous Malformations

It has been estimated that as many as 500,000 people in the United States and Canada have AVMs.^8,9 During the last 40 years, the overall role of stereotactic radiosurgery has been established in the management of selected vascular malformations. Together with advances in microsurgical techniques, postoperative neurointensive care management, and advances in intravascular microcatheter techniques for AVM characterization and embolization, radiosurgery alone or in combination represents a moderately low-risk and effective strategy for properly selected patients.^11,12

A cure for every patient treated with radiosurgery has not yet been achieved. The obliteration rates at 3 years (“angiographic cure”) range from 70% to 85%.^13–17 Several factors have been associated with the likelihood of obliteration, such as AVM volume, prescription dose, location, angioarchitecture, and patient age and lack of history of embolization. Radioresistance has been reported in some lesions that show no response, despite appropriate doses of radiation.^17,19

Small treatment volume, a 23- to 25-Gy prescription dose, and a small number of draining veins^18–23 are factors associated with a more favorable prognosis. Conversely, factors such as a large AVM’s volume, site, and the radiation dose to the exposed brain tissue, have been associated with the likelihood of complications, including bleeding during the latency period, cerebral edema, and/or radionecrosis.^23–34

Some researchers, including Chang et al.,³⁵ have suggested that other factors in addition to volume should be considered when predicting obliteration results, such as the AVM’s flow pattern and the number of draining veins. Furthermore, some lesions that appear to be similar in vessel architecture, location, and volume, respond differently or unexpectedly to treatment plans based on the dose–volume relationship. Some AVMs are obliterated within 2 years, whereas others do so after 3 or 4 years.

In terms of angiographic features associated with lower likelihood of obliteration, Meder et al.³⁶ reported that radiosurgery is less effective in AVMs with direct intranidal fistulas and single draining vessels and suggest that radiosurgery is more likely in patients with plexiform or compact-type AVMs.

Similarly, in a semiquantitative flow study using phase-contrast two-dimensional (2D) magnetic resonance angiography (MRA), Petereit et al.³⁷ reported a negative effect on the obliteration of AVMs by flow speeds greater than 100 ml/s. We reviewed our own experience by defining a critical zone for treatment, the high flow nidus excluding the perilesional draining veins in AVMs using dynamic analysis of circulation times through the AVMs.

Improved Imaging in Radiosurgery

The accuracy of radiosurgical treatment is dependent on the accuracy of target delineation by radiological studies performed with a stereotactic frame in place. In the management of vascular malformations, the primary imaging modality is angiography. Magnetic resonance imaging (MRI) examinations are also recommended for obtaining further information, to fully understand the topographic and tomographic anatomy of the malformation as well as the relationship between the targeted AVM and the non-targeted anatomical structures adjacent to the AVM. The gamma surgery planning program allows target outlines marked on the angiogram to be projected on the MRIs and vice versa (Fig. 105-1). Thus, both modalities can be used in the planning procedure.

FIGURE 105-1 A, DSA pre-embolization. B, DSA postembolization. C, Three-dimensional angiography-magnetic resonance imaging planning for key target volume definition. DSA, digital subtraction angiography.

With the goal of improving imaging, and therefore targeting of the AVM, our protocol focuses on the definition of the AVM’s critical zone, which corresponds to the least-resistant portion of the lesion with the predominant flow draining to the main vein. Our experience in measuring the pedicular pressure in feeders, when treating malformations with superselective embolization, was taken into account in a dynamic study of malformations.³⁸ Pedicular pressures lower than 25 mm Hg are observed in direct AV fistulas, whereas plexiform AVM pressures increase gradually to nearly 50 mm Hg (Table 105-1).

Because the direct pressure measurement protocol is both time consuming and costly, it is impractical for routine radiosurgery planning. We have chosen, therefore, a more practical way to measure the arterial, nidal, and venous circulation times dynamically in a digital subtraction angiogram, frame by frame, using the digital counter of the angiography equipment.

The theoretical underpinning of our approach is that AVMs with fistulous patterns have shorter circulation times and those with plexiform patterns have longer circulation times. The dynamic study also allows us to characterize the AVM from the portion that is least radioresistant and has the fastest flow, ruling out the recruitment vessels, especially in large malformations. Theoretically, with time, an AVM behaves as a waterfall, with the recruitment siphon effect of the vessels normally found around the original AVM. This secondary recruitment phenomenon is seen in perilesional tortuous normal vessels in an angiogram, which change their appearance as they participate in the steal phenomenon.

The recruitment phenomenon is more clearly seen once the flow in the dominant shunt zone decreases with embolization. This is the perinidal portion, with a diffuse appearance, and we have defined it as a “pseudonidus,” which we do not believe it is absolutely necessary to include in the prescription dose. If the pseudonidus is not included, the treatment volume will be smaller and the dose to the critical zone may therefore be increased.

In our protocol, once the Leksell stereotaxic frame (Elekta) is in place on the patient’s head, three-dimensional (3D) volume MRIs are acquired in T1 simple and gadolinium-enhanced sequences, together with axial T2 and coronal sequences for 3D images of the AVM. For digital subtraction angiography (DSA), Axiom Artis equipment is used with software for automated outline correction (Artis&Syngo, Siemens). Currently, images are obtained at 10.5 frames/seconds with a digital contrast injector (Mark V Pro Vis, Medrad), with contrast injected at 7 ml/s in the proximal portion of the carotid and vertebral arteries at the skull base. With superselective contrast injection, the volume is reduced to 3 ml/s. Several circulation times (Fig. 105-2) are then measured frame by frame in the angiography console: nidal filling time (NFT), time from the first frame with nidal staining to the corresponding frame at maximum arterial enhancement; nidal time (NT), includes NFT up to the last frame outlining the arterial nidus; draining time (DT), from the time of the first or predominant draining vein image, when venous draining is multiple, to the time when the abnormal draining vein or veins are no longer seen; total circulation time (TCT), time elapsed from the first frame of the nidal filling time to the last draining frame; and early shunt time (EST), time from the first frame of nidal filling to the first frame of venous draining time.

FIGURE 105-2 Circulation times.

As the circulation times are studied and recorded, the various AVM components are defined, including the single or multiple compartments (Fig. 105-3), the type of venous drainage, and several flow patterns, depending on the circulation time compatible with the three types of circulation: I-A direct fistulae, I-B predominantly fistulous, II-A mixed fistulous-plexiform without stenosis of the draining vein, II-B mixed with draining vein stenosis, and III predominantly plexiform (Table 105-2).

FIGURE 105-3 A to C, Thirty-eight-year-old man, arteriovenous malformation with two carotid and vertebral compartments, flow pattern type II-A, prescription treatment volume 16.1 ml,16 Gy/50% isodose line, radiosurgery-based grading system 2.37.

Associated factors that increase the risk of bleeding are also identified, such as perinidal and proximal aneurysms with high hyperflows, as well as aneurysms in the draining system and stenosis of the main draining vein. The dynamic anatomic definition of the nidus allows us to exclude the recruiting vessels from the pseudonidus.

Nidal circulation times (NTs) less than 2 seconds and a total circulation of less than 6 seconds correlate with the I-A and I-B fistulous flow patterns. Shunt (EST) times of less than 2 seconds and nidal filling times (NFTs) of 2 to 4 seconds and total circulation times of up to 8 seconds correlate with the mixed fistulous-plexiform flow pattern, and nidal times greater than 4 seconds and total circulation time greater than 9 seconds reflect predominantly plexiform flow. It is important to note that a patient with a II-B flow pattern with an early shunt time less than 1.5 seconds and a slow draining time greater than 7 seconds, resulting from stenosis of the predominant venous drainage, should be considered a candidate for embolization to reduce the risk of bleeding.

The crucial stage in the definition of the critical zone is when the angiographic images are transferred to the workstation and correlated with 3D multiplanar MRIs, including topographic and volumetric target analysis related to functional perilesional eloquence, for proper adjustment of the prescription dose.

Our embolization protocol for radiosurgery depends on the volume of the AVM, the corresponding flow pattern, and the risk factors for bleeding. If the size of the pre-embolization target is less than 10 ml, the treatment volume should include all the original volume. If the pre-embolization volume is greater than 10 ml, the prescription dose is delivered to the residual volume in the critical zone. In terms of the flow pattern, it is advisable to use embolization to reduce the flow in patterns I and II-B. Endovascular therapy is also used to eliminate high-flow aneurysms or reduce volume in prospective staged protocols (Fig. 105-4).

FIGURE 105-4 A, Staged radiosurgery. Thirty-eight-year-old woman, large arteriovenous malformation with two compartments, flow pattern type II-A and recruitment phenomena (pseudonidus). First GK treatment-embolization same day, PTV 14.9 ml off recruitment vessels, 18 Gy/50%, RGBS 2.25. B, First 3D GK plan. C, Second GK treatment 1 year after magnetic resonance imaging-3D treatment plan, PTV 10.7 ml, 18 Gy/57%, RGBS 1.87. D, Digital subtraction angiography showing complete obliteration 3 years after second treatment. GK, gamma knife; PTV, prescription treatment volume; RGBS, radiosurgery-based grading system.

Patients Characteristics

From 1995 to 2008, a total of 250 patients with AVMs underwent gamma knife (Elekta) radiosurgery. Of these, 75 patients were treated after dynamic measurement planning to accurately define the critical zone of the AVM, including the dynamic angiographic criteria that categorize the type of AVM: fistulous versus plexiform or mixed, and Spetzler-Martin (S-M) classification into grade III (55), IV (18), or V (2). According to radiosurgery-based grading system (RBGS) proposed by Pollock and Flickinger,³⁹ the minimum score was 0.6 and the maximum was 2.68, with a mean score of 1.64.

The clinical presentation was bleeding in 50% of patients, headache in 30%, seizure in 40%, and neurologic deficit in 10%. Most (70%) AVMs were located in the hemisphere and the rest were deepseated. Two patients had a lesion in the posterior fossa.

Endovascular therapy was performed in 56% of patients and superselective angiography without embolization in 29%. Radiosurgery was administered as a single treatment in 64 patients, repeat radiosurgery in 7 patients, or prospective staged radiosurgery in four patients (Tables 105-3 and 105-4). The AVM original volume was up to 120 ml, with a critical zone target volume of up to 15 ml in the single treatment group, a median target volume of 8.5 ml, and 28 ml in the prospective staged radiosurgery group. The marginal dose of radiation to the AVM was 18 to 22 Gy, with a median dose of 20 Gy.

Table 105-3 Dynamic Protocol for Arteriovenous Malformations in 75 Patients

Results

In 55 patients, DSA was used to review the treatment response (Table 105-5). In 75 patients, the evaluation was performed with MRA at 12 to 40 months (mean follow-up 28 months). Obliteration was observed in 90%, 67%, and 50% of patients in groups with S-M grades III, IV, and V, respectively. In the single treatment group, obliteration was 84% and obliteration was complete in 5 out of 7 patients who underwent repeat radiosurgery. In 4 patients who underwent prospectively staged radiosurgery, the lesion was obliterated in 3. In 2 patients, this was confirmed by DSA, and in the other one by MRA. One patient displayed subtotal obliteration after 40 months of follow-up and was completely obliterated at 5 years. The overall MRA obliteration rate was 84%.

Complications

Permanent neurologic complications were found in 7.0% (5 of 75) of patients, and in 4 of 42 patients embolized during preradiosurgery endovascular therapy. Two patients had nonfatal bleeding during the latency period, 1 had a minor visual field defect, and 1 a lacunar and occipital infarct after embolization. There was no mortality.

Discussion

Inadequate characterization of the AVM nidus for radiosurgical treatment is one of the most common factors contributing to treatment failure. Therefore, we propose a protocol with dynamic characterization that includes the hemodynamic conditions of flow in a very close real-time collaboration between the neurosurgical and endovascular teams.

Some researchers, including Pan et al.,²³ Chang et al.,³⁵ Meder et al.,³⁶ and Petereit et al.,³⁷ among others, have suggested that flow pattern–related factors and even the fistulous type of AVM have a negative effect on obliteration. Frame-by-frame dynamic flow analysis and multiplanar MRI allow the identification of failure-related factors more objectively, including prenidal fistulas with early shunt times (<1 second; Fig. 105-5). The presence of a prenidal fistula could represent one of the factors related to the radiobiological resistance published by the Pittsburgh group.^17,19

FIGURE 105-5 A and B, Prenidal direct fistula (straight arrow) plus nidal component (curved arrow): 58-year-old woman, type I-B, prescription treatment volume 1.1 ml, 24 Gy/50%, radiosurgery-based grading system 1.57. C, Four-year follow-up: failed treatment (radiobiological resistance). D and E, Repeated digital subtraction angiography treatment plan; direct prenidal fistula superselective embolization and gamma knife treatment, same day.

The (Controversial) Role of Embolization

Luessenhop and Spence performed the first cerebral AVM embolization in 1959.⁴⁰ Flow-directed catheters have been most commonly used since 1990, and the most common complications are aberrant embolization, embolization of the normal proximal arteries, venous system occlusion, and arterial perforation.^41,42 Recommendations for embolization are supported by anecdotal case series and nonrandomized cohort studies.

The use of embolization to improve the results of radiosurgery has been controversial. Most publications concerning retrospective studies conclude that the procedure has a negative impact on outcome and that complications from endovascular therapy add to the complications of radiosurgery.

Andrade Souza et al.⁴³ addressed the question of whether preradiosurgical AVM embolization affects obliteration or clinical outcomes. They found that embolization reduced the obliteration rate and that outcomes were better in the nonembolized group, but the differences were not statistically significant. Obliteration was achieved in 70% of patients receiving radiosurgery alone and in 47% of patients treated with embolization before radiosurgery.

Unfortunately, there has been no standardization of the embolization technique. In reports of endovascular therapy, there are distinct outcomes from different groups. Some series have combined the use of particles and cyanoacrylate. There is also no clear information about results and more specifically in terms of the volume of the AVM, flow patterns, risk factors for bleeding, or nidus fragmentation. Furthermore, it is usually not specified whether an endovascular therapist worked together with the neurosurgeon. There is also no systemization of the two-dimensional versus 3D volume measurements,⁴⁴ or of quantitative flow measurements before and after embolization.⁴⁵

We are of the opinion that it is important in order to better understand the contribution of embolization, to define whether the radiosurgical treatment volume is based on the global initial volume before embolization or whether the radiation was targeted to the residual volume. We also recommend that neurosurgeons work very closely with endovascular therapists in real time. In some AVMs, stereotactically guided embolization and correlation with 3D topography of the pedicle on MRI using Gamma Plan (Elekta) can optimize the preradiosurgery results by estimating the objective volume and flow reduction, and by providing information about the hemodynamic factors that work against favorable radiosurgery outcomes (Fig. 105-6).

FIGURE 105-6 A and B, Forty-two-year-old woman, arteriovenous malformation type IIA, prescription treatment volume 5.4 ml, 21 Gy/50%, radiosurgery-based grading system 1.38. C and D, Digital subtraction angiography and magnetic resonance imaging, 2 years follow-up, complete obliteration, minor visual field impairment, seizure free, epileptic syndrome outcome Engel I.

Ideally, the embolization procedure should be concentric to simulate the usual surgical approach, thus avoiding inconvenient postembolization fragmentation of the target. Furthermore, it should be recorded whether the treated area belongs to the early filling and dominant flow zone, which is related to predominantly venous drainage, while excluding pseudonidus (Fig. 105-7). Special attention must be paid to postembolization fragmentation, because the outcomes and complications are operator-dependent.

FIGURE 105-7 A, Sixty-eight-year-old male, volume 10.2 ml, Pollock scale (RBGS) 2.68, obliteration rate probability <70%, demential syndrome, steal phenomenon. Pre-embolization: flow pattern type I-B. Early shunt time <1 sec: fast fistulae plus nidal component, which is why embolization is indicated. B, Postembolization. Modified flow pattern: type III plexiform slow-flow nidal filling time > 4 sec. Embolization plus gamma knife same day, prescription treatment volume 9.7 ml; 20 Gy/60%, RBGS 2.63. C, Digital subtraction angiography at 14 months follow-up. RBGS, radiosurgery-based grading system.

Embolization should be considered to reduce the bleeding risk in AVM with I-A, I-B, and II-B flow pattern, with an early shunt time less than 1.5 seconds and a slow draining time of more than 6 seconds, resulting from stenosis of the predominant venous drainage. The greatest benefit of embolization is obtained when the flow and concentric volume are reduced.

In our series, there was no negative effect of performing embolization and radiosurgery on the same day in 10 patients, although more intense asymptomatic signal changes were seen in the T2 sequences at 3 months in 60% of these patients, 2 of whom required temporary steroids as outpatients, without permanent sequelae.

Contrary to our expectations, 2 patients with complex AVMs and steroid treatment who underwent embolization and radiosurgery on the same day remained obliterated 10 years later (Fig. 105-8A and B). We cannot rule out a synergistic effect between radiation damage and the inflammatory response due to embolization. Furthermore, in this series the mean RBGS of 1.64 would represent an expected 60% to 70% mean obliteration rate.

FIGURE 105-8 A, Seventeen-year-old female, epileptic syndrome, pre-embolization volume 120 ml, flow pattern pre-embolization II-A. Two embolizations preradiosurgery, last one the same radiosurgery treatment day. Arteriovenous malformation flow pattern, postembolization type III plexiform dominant component. Critical zone: prescription treatment volume 14.9 ml, 18 Gy/50%, radiosurgery-based grading system 1.83 B, Digital subtraction angiography at 10 years follow-up, epileptic syndrome outcome Engel I. The fiducials on follow-up studies mean noninvasive stereotactic volume protocol.

Conclusion

Critical decisions are required before therapy is undertaken for a patient with an AVM. It is reasonable to state that the total morbidity and mortality of embolization averages 9% in experienced hands. The benefits of embolization are yet to be established, particularly with regard to lowering surgical morbidity and mortality, as well as improving the cure rate of radiosurgery for large lesions and fistulous flow patterns.

The additional therapeutic benefits and financial considerations of multiple embolization procedures have not been addressed. These issues and others remain future challenges in defining the benefits of embolization in the treatment of AVMs.

In order to improve the outcomes of AVM radiosurgery, our protocol is designed to be easily followed by any radiosurgical center. Instead of retrospectively, as has usually been suggested, our method can be prospectively and systematically used to evaluate hemodynamic variables of AVMs undergoing radiosurgery. Using digital quantitative methods to measure flow and volume, it is possible to obtain an objective measurement of the flow pattern and volume modifications in an embolized AVM, which will be a helpful tool in endovascular therapy. Building a clear and reliable database that includes data from other radiosurgery centers should help to standardize indications, and to analyze and share review and research results of endovascular therapy for AVMs, which until now have been anarchic and heterogeneous. A dynamic definition of the AVM critical zone might increase the obliteration rate and decrease the treatment volume in big lesions by removing the pseudonidus from the target.