History

The first case of AVM treated with radiotherapy was reported by Magnus.³ In a patient he operated on with an AVM at the motor cortex, he did not attempt surgical removal because of the high possibility of neurological deficits. After decompressive craniectomy, he treated the patient with radium therapy and reported that the patient was seizure-free 2 years after radiotherapy. No imaging or histological studies were available. Cushing and Bailey reported the first successful surgery on an AVM.⁴ Cushing explored a vascular tumor and felt that the lesion could not be attacked without fatal hemorrhage. The patient was treated with radiotherapy. He reexplored the lesion 3 years later and described that the tangle of pulsating vessels previously encountered was largely thrombosed and transformed into a multitude of small bloodless shreds which could be easily separated from the adjacent normal cortex.

There was intense interest in the use of radiation for AVMs following Cushing’s discovery, but the initial results were not encouraging. Although some studies^5,6 with small numbers of cases provided evidence of the possible utility of radiation in the treatment of AVMs, most did not provide imaging or histological proof of the efficacy of radiotherapy. This led to an almost unanimous consensus in the assessment of radiation as being worthless in the management of AVMs.

With the introduction of Gamma Knife, the potential value of irradiation in vascular malformations was reassessed. Contributing factors included an increasing body of evidence that the cells constituting the vessel wall were responsive to ionizing radiation. Long-term angiographic follow-up of a small series of AVMs treated with fractionated conventional radiation by Johnson in the 1950s revealed that the AVMs were obliterated in 45% of cases.⁷ In 1970, the first radiosurgical treatment for an AVM was performed by Ladislau Steiner and associates at Karolinska Institute in Stockholm. The patient refused surgery, and given the patient’s renal insufficiency, the risk of surgery was considered too high. Although the intention was to deliver focused radiation to the nidus, because only small collimators were available, the feeding arteries were targeted and 25 Gy was given as the prescription dose to the 50% isodose line. On angiography 19 months after the treatment, the feeding vessels were obliterated, and the malformation no longer filled. Subsequently, larger collimators were available and could cover the AVM nidi, and more patients were treated successfully. Today more than 50,000 AVM patients have been treated with Gamma Knife radiosurgery, which has been proved to be a safe and effective treatment alternative for AVMs.

Modality

The goal of radiosurgery for AVMs is to deliver a high absorbed radiation dose to the AVM nidus, typically in a single session, while largely sparing the surrounding brain significant dose and thereby minimizing undesirable effects from the treatment. As the number of modalities for delivering radiosurgery have increased over time, so have the numbers that have been applied to the treatment of AVMs. Reports in the literature exist for AVM treatment with various radiosurgical modalities including the Gamma Knife, isocentrically mounted linear accelerators,^8–11 and robotic linear accelerators (CyberKnife).¹² Regardless of modality, radiosurgical devices achieve the desired characteristics of small fields, fast-dose falloff, and highly accurate targeting through the use of two basic principles: superposition of beams and stereotactic targeting.

Radiosurgery achieves highly conformal dose distributions by spreading the total energy delivered over a targeted volume, either through the superposition of many small beams on the target or through the use of noncoplanar arcs (where each arc can be thought of as a large number of small beams). The energy of any given single beam is too low to cause a significant biological effect. However, the superposition of all the beams at the target delivers a substantial amount of radiation, sufficient to cause biological changes resulting in the occlusion of the AVM nidus.

The details of beam superposition vary by modality. In the case of the Gamma Knife, 192 (or 201, depending on the model of the unit) individual beams are precisely aimed at a focal point, or isocenter, to achieve this effect. The beams are collimated through individual beam channels. In older units, this was achieved with a combination of internal primary and secondary collimation and external “helmet”-based final collimation. In the case of the newer Perfexion model Gamma Knife, the collimator assembly is housed entirely within the main body of the unit.¹³

For early linear accelerators adapted for radiosurgery, finely collimated beams were achieved through the use of circular collimator “cones” that could be attached to the accessory tray of the accelerator.¹⁴ Many currently available accelerators are equipped with micro-multileaf collimators that can achieve irregularly shaped beams that can more precisely conform to the target morphology.¹⁵ Both approaches are often used with a non-coplanar arc technique, which directs the fields at the target while spreading out the overall delivered energy.¹⁶

The ability to create a focal, high-dose distribution does little good in itself if there is no way to precisely and accurately aim at the target. This problem is elegantly solved using the principles of stereotaxy. Traditionally in intracranial radiosurgery a rigid frame is fixed to the skull. This frame defines a coordinate system by which any point within the brain can be localized. Fiducial markers, which are visible in stereotactic imaging studies as part of the procedure, are directly related to this coordinate system; thus, the target can be visualized and localized in “stereotactic space.” Accuracy and precision of treatment are thus guaranteed by the precision and accuracy with which the target can be localized and the assurance that this target will not move during a treatment owing to the rigid head frame. More recent innovations in radiosurgery have included frameless stereotaxy.¹⁷ In these systems image-guidance plays a greater role both before and during the procedure. Less invasive restraint systems such as thermoplastic masks are used in place of the rigid head frame, and periodic imaging is used to track and correct for patient motion.

Histopathology

Several histological studies have described the changes of irradiated vessels with progressive narrowing and obliteration of the lumen.¹⁸ The earliest changes are endothelial damage and endothelial-intimal separation. These are followed by subendothelial and intimal-medial proliferation of smooth muscle cells with elaboration of extracellular matrix components. Cellular degeneration and hyaline transformation of vessel walls follow, and finally the vessels obliterate completely. The above-mentioned histopathological changes are correlated with time after radiosurgery and tend to occur in smaller vessels.

Gamma Knife Radiosurgery

At the University of Virginia, we perform AVM radiosurgery with the Gamma Knife. Patients are evaluated at least 1 day before the Gamma Knife surgery (GKS). Anticonvulsion medication is used for patients already on these medications and for patients with supratentorial AVMs.

The technique of GKS begins with the placement of the Leksell G-Frame on the patient’s head by the neurosurgeon. The Leksell G-Frame consists of a rectangular aluminum base ring to which four aluminum posts are attached. The frame is affixed to the patient through the posts using titanium pins, which are screwed to the outer table of the skull. In our center, we place the frame in the operating room using controlled sedation, local anesthesia, and strict attention to aseptic conditions.

The accuracy of a GKS is ultimately dependent on the visualization of three-dimensional views of the intended target in the brain. Stereotactic biplane angiography is the gold standard for delineating the nidus. Additionally, stereotactic magnetic resonance imaging (MRI) and magnetic resonance angiography source images provide information, especially in the axial plane, which can enhance visualization of the nidus for the purpose of treatment planning.

Early Adverse Effects

Patients might experience headache from frame placement and nausea or vomiting from pain or sedative medications. Few patients developed seizure in the immediate post-GKS period; most who did were those with a prior history of seizure.

Radiation-Induced Changes

Radiation-induced change is an increased T2 signal around the AVM seen on MRI (Fig. 47.2). Radiation damage of glial cells, endothelial cell damage followed by breakdown of blood-brain barrier, excessive generation of free radicals, or release of vascular endothelial growth factors have been proposed to explain this imaging finding. The severity of radiation-induced changes on images and associated neurological deficits varies, ranging from asymptomatic, being only a few millimeters of increased T2 signal surrounding the treated nidus to massive brain edema with symptoms and signs of increased intracranial pressure. From our 1500 Gamma Knife procedures performed for AVM patients with follow-up MRI available for analysis, 34.4% of patients developed radiation-induced changes. Among them, 60% had mild (a few millimeters of increased T2 signal surrounding the nidus), 33% had moderate (compression of ventricle and effacement of sulci), and 7% had severe (midline shift) radiation-induced changes. The mean time to the development of radiation-induced changes was 13 months after GKS, and the mean duration of the changes was 22 months. Larger nidus volumes, higher prescription doses, history with preradiosurgical embolization, and nidus without previous hemorrhage were associated with higher risk of radiation-induced changes.

FIGURE 47.2 Pre-GKS (Gamma Knife surgery) lateral (A) and frontal (B) projection angiograms of a 52-year-old male patient with a history of seizure. The nidus obliterated 6 years following GKS (C and D). The patient developed radiation-induced changes as shown on the T2-weighted magnetic resonance imaging (MRI) scan 8 months (E) and 2 years (F) after GKS. A cyst formation with increased T2 signal was seen 6 years after GKS (G).

One hundred twenty-two (8.7%) patients developed headache, worsening or new seizures, or neurological deficits associated with radiation-induced changes. Patients with severe radiation-induced changes and nidus at eloquent areas were more likely to develop symptoms. Twenty-six patients (1.8%) had permanent neurological deficits.

Cyst Formation

Cyst formation is a rare complication following GKS. Cysts that develop after resolution of previous hemorrhages or fluid cavities from encephalomalacia after surgeries should not be considered as complications related to GKS. Direct radiation injury to the perilesional brain tissue, increased permeability of the blood-brain barrier with accumulation of the exudative fluid, hemodynamic perturbations during gradual obliteration of the nidus with subsequent ischemic tissue damage, and tissue destruction due to subclinical perilesional hemorrhages have been proposed as the possible mechanisms of cyst formation. From our 1272 patients with follow-up MRI available, we found a total of 20 patients (1.6%) developing a cyst after a mean of 8.1 years after GKS. Four cysts were found in 710 patients with follow-up shorter than 5 years, eight cysts were found in 302 patients with follow-up between 5 and 10 years, and another eight cysts were found in 260 patients with follow-up between 10 and 20 years. Of the 20 patients, 18 had regular MRI follow-up and 14 (78%) of them had radiation-induced changes before the development of cysts. Six patients had large cysts and three of them were symptomatic, requiring surgery. Two patients underwent craniotomy and drainage of the cyst. The cyst wall showed no evidence of neoplasia.

Radiosurgery-Induced Neoplasia

We found two meningiomas from 1333 AVM patients treated with GKS; however, follow-up imaging was performed over a period of at least 10 years in 288 of these patients. If we conservatively estimate that radiosurgery-induced lesions would be evident within a 10-year time interval, then our incidence of radiosurgery-induced neoplasia is 2 in 2880 person-years or 69 in 100,000 person-years. Thus, there is a 0.7% chance that a radiation-induced tumor may develop within 10 years following GKS. This is less than the 1.9% risk detailed by Brada and colleagues,²⁷ but our results encompass a follow-up period of only 10 years. The long latency and relative rarity of these lesions following radiosurgery may defy a conclusive determination of the true incidence.

Incompletely Obliterated Arteriovenous Malformations

Most reported studies state that the risk of hemorrhage persists as long as the AVM nidus is still patent. These data provided the rationale for re-treatment of still-patent AVMs following the initial GKS.

In our experience, 74 males and 66 females with a mean age of 33 years underwent repeat GKS for still-patent AVM nidi following initial GKS from 1989 to 2007. Causes of initial treatment failure included inaccurate nidus definition in 14, failure to fill part of the nidus due to hemodynamic factors in 16, recanalization of embolized AVM compartments in 6, and suboptimal dose (less than 20 Gy) in 23 patients. Nineteen patients had repeat GKS for subtotal obliteration of AVMs. In 62 patients, the AVM failed to obliterate in spite of correct target definition and adequate dose. At the time of re-treatment, the nidus volume ranged from 0.1 to 6.9 cm³ (mean 1.4 cm³) and the mean prescription dose was 20.3 Gy. Clinical follow-up time ranged from 15 to 220 months with a mean of 84.2 months after repeat GKS.