Radiosurgery

Published on 09/04/2015 by admin

Filed under Hematology, Oncology and Palliative Medicine

Last modified 22/04/2025

Print this page

This article have been viewed 4920 times

25 Radiosurgery

Dennis C. Shrieve, MD, Jay S. Loeffler, MD, FACR, Michael W. McDermott, MD, David A. Larson, MD, PhD

Stereotactic radiosurgery (SRS) is most often defined as the delivery of a single high dose of ionizing radiation to an intracranial target defined using stereotactic imaging.¹ Although the definition has been expanded by some to include up to five fractions delivered using stereotactic technology, this chapter is limited to a discussion of the single-dose approach and results. Extracranial SRS is discussed in Chapter 75.

The use of accelerated particles for highly conformal treatment of brain tumors developed following the proposal of physicist Robert Wilson and coworkers at the University of California, Berkeley.² The initial clinical trial using protons to treat pituitary disorders was begun in 1954 with 840 patients subsequently treated with protons or helium ions.³^,⁴ These focal-radiation delivery methods were further developed and the term radiosurgery was coined by Swedish neurosurgeon Lars Leksell and radiobiologist Borje Larsson of the Karolinska Institute in Stockholm, initially for treatment of functional disorders.⁵ Several different delivery techniques result in a dose distribution that is highly conformal to the target. Dose fall-off outside the target is precipitous and therefore only the target and immediately adjacent normal tissue are included in the high-dose region. After more than 40 years of development, radiosurgery has entered the mainstream of medical care. Since the 1960s, more than 1,000,000 patients have been treated with some form of radiosurgery, most within the past 10 years. Radiosurgery is now practiced at more than 250 facilities in the United States.

In North America, patients who undergo radiosurgery are usually selected, treated, and followed by a dedicated multidisciplinary radiosurgery team of experts that includes radiation oncologists, neurosurgeons, physicists, nurses, and technologists; the same radiosurgery team provides quality assurance. This multidisciplinary approach to therapy has received support from the radiosurgery task forces of societies representing both neurosurgery and radiation oncology, and is considered the standard of care in the United States.^1,⁶^–⁸

Techniques

Stereotaxis

The term stereotactic (from the Greek stereo and taxis, ordered in three dimensions) refers to techniques in which intracranial targets are localized precisely in three-dimensional space for introduction of a probe for biopsy or for delivery of localized radiotherapy. The basis for construction of the stereotactic space is the rigid application of a head frame to the patient’s skull. Stereotactic imaging is then performed with a localizer fitted over the patient’s head and attached to the frame. Fiducial markers in the localizer box define x, y, and z coordinates for any location in the defined space above the frame and appear on each image obtained. Imaging may be done with either computed tomography (CT) or magnetic resonance imaging (MRI) technology, or both with “fusion” of the two imaging modalities (Fig. 25-1).

FIGURE 25-1 • A, Leksell stereotactic frame with MR-compatible localizer. Note N-shaped fiducial markers on localizer and scale on frame for patient positioning during treatment. B, Computer rendering of computed tomography slice depicting stereotactic space defined by N-shaped fiducials (arrow) from linac radiosurgery planning system. C, Stereotactic magnetic resonance image (MRI) of patient with right acoustic schwannoma (left) with overlying “fused” MRI (right). Note excellent anatomic fusion of images and fiducials indicating level within stereotactic space.

Stereotactic Radiosurgery Delivery Systems

At present, radiosurgery is performed by using one of three methods of delivery of high-energy radiation. The radiosurgery technology most commonly available around the world uses high-energy x-rays generated from a clinical linear accelerator, or “linac.” Gamma ray radiosurgery with the gamma knife, although not as widely available as linac radiosurgery, is performed at an increasing number of centers and has been used more often than the other radiosurgery technologies. Radiosurgery with charged particles (e.g., protons or heavy ions) produced by a cyclotron or synchrotron is the technology least frequently used, but may be most suitable for very large target volumes.⁹ Any radiosurgery treatment-planning system should provide rapid, three-dimensional visualization of the target and isodose distribution superimposed on CT scans, MRIs, or angiograms.

Gamma Knife

More than 250 gamma knife units are now located around the world, including more than 100 in the United States. As of the year 2006, more than 400,000 patients had been treated with this technology. The gamma knife unit consists of a central core containing either 192 or 201 cobalt 60 (⁶⁰Co) sources surrounded by cast-iron or tungsten shields and a steel entrance door (Fig. 25-2). The sources are situated in individual beam channels in the central core directed convergently toward a single target point (isocenter) at the hollow center of the radiation unit (Fig. 25-3). Collimation of the beams is provided by collimators projecting to 4, 8, 14, or 18 mm diameter (Model 4C) or 4, 8, 16 mm (Perfexion) at isocenter (Fig. 25-4).

FIGURE 25-2 • Gamma knife unit.

FIGURE 25-3 • Schematic of gamma knife unit showing patient in treatment position, sources, and beams directed to isocenter through primary and secondary collimators.

(From Lutz W: Radiation physics for radiosurgery. In Alexander E, Loeffler JS, Lunsford LD (eds): Stereotactic Radiosurgery. New York, McGraw-Hill, 1993.¹⁸⁵)

FIGURE 25-4 • A, Close-up of gamma knife helmet. The “8” designates this helmet as having collimators projecting to 8 mm at isocenter.

B, A patient positioned in a gamma knife helmet prior to entry into the gamma knife core.

After stereotactic frame placement and imaging, treatment planning consists of selection of a number of individual “shots” or radiation doses to small volumes within the target, each with a unique isocenter coordinate. Treatment plans are evaluated and optimized using dose-volume histograms (DVH). Once treatment planning is completed, the patient lies on the treatment couch, the position of which is adjusted so that the patient’s head, fixed within the stereotactic frame, is moved into the treatment position in the helmet (see Fig. 25-4). The frame is oriented within the helmet in the x, y, and z directions so that the radiation beams align with the target site according to the treatment plan. Several different isocenters may be treated in sequence during the treatment session to make the isodose surface conform closely to any irregular three-dimensional target contour (Fig. 25-5). Historically, it has required 10 to 15 minutes to place the patient in position and treat each isocenter. Modifications of the gamma knife have shortened these times considerably. Because ⁶⁰Co has a half-life of 5 years, treatment times must progressively be increased.

FIGURE 25-5 • A, Magnetic resonance imaging scan of right parietal glioblastoma in 39-year-old man with focal recurrence 15 years following diagnosis and treatment with 60 Gy external-beam chemotherapy and brachytherapy boost. B, Gamma knife treatment plan: 13 isocenters were used to obtain conformity of 50% isodose line to tumor volume. Three different collimator helmets were employed (18 mm × 3, 14 mm × 7, 8 mm × 3). The target volume was 20.2 cc and a dose of 11 Gy was prescribed to the 50% isodose line representing the target periphery.

Figs. 25-6 to 25-8 demonstrate examples of gamma knife plans for arteriovenous malformation (AVM), metastatic disease, and a benign acoustic schwannoma, respectively. In each, the dose was prescribed at the 50% isodose line (yellow), which was required to conform tightly to the target configuration (red or blue). The 25% isodose line (green) is also shown. The crosses represent projections of isocenter locations onto the MRI plane.

FIGURE 25-6 • A, Magnetic resonance imaging scan of 41-year-old woman with left parietal arteriovenous malformation (AVM). B, Four isocenters were used (14 mm × 1, 8 mm × 3) and 18.5 Gy delivered to the 50% isodose line. Angiographic orthogonal views of the AVM with isodose distribution superimposed on anterior-posterior (C) and sagittal (D) angiograms.

FIGURE 25-7 • A, Magnetic resonance image of a dural-based mass at the right superior frontoparietal convexity, immediately adjacent to the falx. The patient was a 65-year-old man who had been previously diagnosed with adenocarcinoma of the colon. B, 11 isocenters (18 mm × 2, 14 mm × 5, 8 mm × 4) were used to cover the 13-cc target volume with the 50% isodose line. 16 Gy was prescribed.

FIGURE 25-8 • A, Magnetic resonance image of a left acoustic neuroma in a 72-year-old man who had developed sudden onset of left sensorineural hearing loss 2 years earlier. Magnetic resonance imaging scan showed a small acoustic neuroma and the patient was observed. His hearing deteriorated, however, and a repeat scan at 2 years showed tumor progression. B, The gamma knife isodose plan. The patient received 14 Gy at the target periphery, delivered with three isocenters (8 mm × 3). The target volume was 0.4 cm².

Linear Accelerator Radiosurgery

Linac radiosurgery is widely available, in part because existing linacs may be modified for radiosurgery with relatively little expense. It is likely that more than 100,000 patients have been treated with linac radiosurgery. Although many treatment centers use a linac unit that has been modified for radiosurgical use, dedicated stereotactic linac units that are specially designed and have excellent mechanical accuracy are now available. “Home-grown” linac planning systems have been largely replaced by commercially available systems that are approved by the U.S. Food and Drug Administration for this purpose.

Like the gamma knife, linacs have a defined isocenter. The linac isocenter is the point of the intersection of the axis of rotation of the gantry and the treatment couch and the central axis of the photon beam (Fig. 25-9). Targets may be treated with single or multiple isocenter plans. Initial linac systems used special circular collimators of diameters from 5 to 60 mm to reduce the penumbra of the beam and the radiation dose to normal tissues. However, the use of circular collimators greatly limits the potential for isodose shaping, especially when using single-isocenter plans. Excellent conformity is now possible using micro-multileaf collimation (Fig. 25-10) and either multiple fixed fields or dynamic-arc radiosurgery. In the latter, the field shape is constantly changing to conform to the beam’s-eye view outline of the target volume (see Fig. 25-10). This treatment strategy has been shown to produce isodose distributions with conformity indices similar to those reported for multiple isocenter gamma knife plans.¹⁰

FIGURE 25-9 • A, Schematic of linac isocentric set-up demonstrating patient position for radiosurgery on table extension. Note rotation of table and gantry about isocenter. B, Linear accelerator radiosurgery. The patient is positioned using a localizer so that the plan isocenter precisely coincides with the isocenter of the linear accelerator.

(A, From Lutz W: Radiation physics for radiosurgery. In Alexander E, Loeffler JS, Lunsford LD (eds): Stereotactic Radiosurgery. New York, McGraw-Hill, 1993.¹⁸⁵)

FIGURE 25-10 • Dynamic conformal arc (DCA) technique for linac radiosurgery. Right acoustic schwannoma was treated with 5 DCA. The micro-multileaf collimation (mMLC) conforms to the tumor periphery without a margin. The shape of the mMLC changes with each degree of arc. The resulting plan is shown in Fig. 25-11.

Figs. 25-11 to 25-13 demonstrate linac-based radiosurgery plans for acoustic neuroma, recurrent pituitary adenoma, and AVM. All plans employed a single isocenter.

FIGURE 25-11 • A, Acoustic neuroma treated as shown in Fig. 25-10. Pretreatment magnetic resonance imaging (MRI) scan (left). The tumor volume was 4.1 cm³. B, A dose of 13 Gy was prescribed to the tumor margin corresponding to the 85% isodose line (red). The prescription isodose volume also contained 0.96 cm³ of normal tissue resulting in a conformity index of 1.3. Also shown is the isodose line corresponding to 50% of the prescription dose (green). C, A follow-up MRI scan 18 months after stereotactic radiosurgery shows smaller tumor volume and loss of central enhancement.

FIGURE 25-12 • Recurrent hemangiopericytoma treated with linac dynamic conformal arc stereotactic radiosurgery. Tumor volume was 1.26 cm³. A dose of 20 Gy was delivered to the tumor margin corresponding to the 83% isodose line (red). The prescription dose volume also included 0.38 of normal tissue resulting in a conformity index of 1.3. Also shown is the isodose line corresponding to 50% of the prescription dose (purple).

FIGURE 25-13 • Arteriovenous malformation (AVM) treated with dynamic conformal arc stereotactic radiosurgery. The tumor volume was 2.54 cm³ and was defined using magnetic resonance imaging (MRI), angiogram and computed tomography–angiogram (shown). A dose of 20 Gy was prescribed to the margin corresponding to the 87% isodose line (red). The isodose line receiving 50% of the prescribed dose is also shown (green). A 1-year follow-up MRI showed significant reduction in volume of AVM.

Further improvement in conformity of linac radiosurgery plans may be obtained using intensity-modulated delivery methods. Coupled with inverse treatment planning systems, avoidance of normal structures adjacent to the target volume is also possible.¹¹ Clearly the future of linac-based stereotactic radiotherapy and radiosurgery lies in the ability to produce conformal treatment plans entirely covering target while using intensity modulation to avoid adjacent critical structures.¹² Such treatment may often be delivered through a single isocenter and with prescription to the 80% isodose line or higher, thereby reducing dose inhomogeneity within the target.¹⁰

To align the patient’s head with respect to the linac beam, some groups have the patient lie on the treatment couch with the head supported and positioned on a separate, floor-mounted, stereotactic fixture, whereas others use only the treatment couch together with the standard alignment lasers associated with linac technology. Most groups recommend that target positioning be verified radiographically before each individual treatment. With most systems, patients are treated while they lie supine. In most cases, beam-entrance patterns trace a series of non-coplanar arcs on the scalp, one arc for each of several positions to which the table is turned.¹³

Particle-Beam Radiosurgery

Charged particle-beam radiosurgery has been used predominantly in North America, Europe, and the former Soviet Union during the previous 40 years, and during that time more than 6000 patients have been treated, some receiving therapy in as many as four fractions rather than one.¹⁴ In contrast to x-ray or gamma knife techniques, which require a large number of convergent radiation beams to produce a highly focal dose distribution, charged-particle beams produced by cyclotrons or synchrotrons produce a highly focal dose distribution with only a few beams because of the Bragg ionization peak produced by each individual beam. The width of the Bragg peak and its depth in relation to the scalp are adjusted to match the target’s dimensions and depth in brain. Patients usually are treated with two to six stationary, intersecting beams, each with its own irregularly shaped collimator that conforms to the contour of the target (Fig. 25-14). To ensure an acceptable three-dimensional dose distribution, compensators may be interposed in the beams to compensate for skull curvature and tissue inhomogeneities. As in linac or gamma knife radiosurgery, the three-dimensional dose distribution produced by a few intersecting particle beams falls off rapidly with distance from the target periphery, but it is usually more homogeneous within the target. Although proton radiosurgery is used to treat the common intracranial targets of photons techniques, a disproportionate number of patients with large irregular shaped lesions are referred to proton centers to achieve a greater conformality of dose. For example, at the Massachusetts General Hospital more than 50% of the AVMs treated are larger than 10 cc. This experience has provided unique data concerning dose-volume analyses for larger targets that is otherwise not available from photon radiosurgery experiences.¹⁵

FIGURE 25-14 • The stereotactic alignment for radiosurgery (STAR) device used at the Massachusetts General Hospital for intracranial radiosurgery. The robotic patient positioner allows for a diverse number of field angles using a fixed beam. For more complicated geometric targets, patients are treated on the proton gantries.

Radiobiologic Effect

Usually, radiosurgery delivers little or no clinically important radiation dose beyond the target volume, in part because of the steep dose gradient at the target’s margins and in part because the minimum target isodose contour (usually 50% to 80%) conforms closely to the three-dimensional configuration of the target. Therefore, radiosurgery doses are high and inhomogeneous. The biologic effectiveness of radiation depends on the total dose of radiation, the dose per fraction, and the type of tissue irradiated. For late-responding tissues like normal central nervous system or vasculature tissue, there is a strong dependence on dose per fraction. With radiosurgery, total dose and dose per fraction are high, and variations in both according to isodose distributions produce sharply changing biologic effects. Within the target periphery, dose inhomogeneity (sharply rising dose from target margin to isocenter) has a disproportionate radiobiologic effect and may substantially elevate the potential for a cure or complication. Beyond the target’s perimeter, the radiobiologic effect declines rapidly as a result of decreasing dose and dose per fraction, a characteristic that may spare normal tissue, but that may permit the survival of potential infiltrative disease just outside the radiologically apparent target.¹⁶

Intracranial Target Lesions

Intracranial targets for radiosurgery fall into one of four categories based on two criteria: (1) whether the target tissue’s radiobiologic response to radiosurgery is early or late; and (2) whether the target volume contains any normal tissue.¹⁶

Category A

Target volumes containing an AVM are late-responding in both normal and abnormal tissue because the target volume contains some normal brain.

Category B

Target volumes containing a benign tumor, such as meningioma, pituitary adenoma, and acoustic neuroma are also late-responding and show a marked radiobiologic effect, but predominantly in abnormal tissue, because the target contains no normal tissue in most cases.

Category C

Target volumes containing a low-grade glial tumor also contain some normal brain, but show a more substantial effect in late-responding normal tissue than in early-responding tumor tissue. Radiosurgery is recommended for these tumors only occasionally.

Category D

Target volumes containing a metastasis show an early response and a moderately large effect, but predominate in early-responding abnormal tissue because the target volume contains no normal tissue. Glioblastomas may be considered to be in category D, although they may contain a small amount of normal tissue.

For small targets in categories A and B, it is to be anticipated that the ratio of patients cured to those with complications (therapeutic ratio) will bear little relation to the fractionation scheme used, whether single or multiple. For AVMs (category A), clinical data on fractionated treatments to a dose sufficient to test this prediction are not available. The clinical data available so far on radiosurgery for meningioma (category B) are not dissimilar to the best reported results of fractionated radiation therapy planned with CT or MRI.¹⁷^–¹⁹ For metastases (category D), it might be thought that fractionated treatments would be preferable to single-fraction treatments because hypoxic tumor cells might reoxygenate and become more radiosensitive between fractions. The therapeutic ratio appears to be better with radiosurgery for metastatic lesions than with fractionated treatment, although a valid comparison is not possible because no studies have used sufficiently high fractionated doses. Why fractionation is not more important for these targets is not entirely understood, although the explanation may relate to the treatment volume. Usually, for small targets, a single fraction does not produce enough damage to normal tissue to have clinical consequences. In any case, most patients with lesions in category C and many in category D, except those with recurrent tumors, undergo fractionated irradiation as part of their therapy.

Single-fraction SRS results in tumor control rates for malignant tumors higher than predicted by modeling based on in vitro and in vivo killing of tumor cells by radiation.²⁰ It has been proposed that, in the setting of single, high doses of radiation, the vascular endothelial cells become an additional important therapeutic target. Radiation-induced apoptosis of the endothelial cells may only become significant in single doses larger than 8 to 11 Gy.²¹ Similar effects likely come into play in the treatment of benign neoplasms with SRS.

Target Volume Definition

The hallmark of SRS is conformity of the prescription dose volume to the target volume. In conventional radiotherapy the gross target volume (GTV) defines radiographically visible tumor; there is often a “margin” added to this for microscopic extension, lymph nodes at risk, and so on, to form the clinical target volume (CTV); a further expansion is made to account for small, day-to-day errors in patient set-up to form the planning target volume. In SRS, the GTV is defined on stereotactic CT or MRI scan. In principle there is no need for expansion to cover subclinical disease or for set-up error. Set-up errors for all radiosurgery delivery systems are in the sub-millimeter range. Therefore, target volume definition is a crucial first step in the SRS planning process. Definition of AVM nidus or tumor volume requires the use of the appropriate imaging modality or modalities. Definition of normal structures at risk requires equal care. DVH analysis is highly recommended for evaluation of SRS plans in terms of target coverage and dose to normal structures. Various indices have been described to evaluate the degree conformity of the prescription isodose to the target volume.¹⁰^,²²

Radiation Dose

In all radiosurgery techniques, the volume of normal brain tissue beyond the target that receives a moderate to high dose of radiation increases approximately as the cube of the average diameter of the target volume. As complications of irradiation affecting normal tissue may correlate with the volume of normal tissue receiving a dose of a specified magnitude, at least to some extent, isoeffect dose levels may be expected to follow a power law, by which a log-log plot of isoeffective dose versus target diameter approximates a straight line with negative slope. The empirical power law dose-diameter isoeffect curve developed by Kjellberg and colleagues ²³ for proton radiosurgery is thought to correspond to a 1% risk for radiation necrosis of human brain tissue. The theoretical integrated logistic model dose-diameter curve developed by Flickinger²⁴—obtained from a mathematical model that predicts the relationship of dose to volume for a 3% risk for radiation necrosis—closely resembles Kjellberg’s 1% curve for both gamma knife and linac radiosurgery. In each case, a log-log plot of isoeffective dose versus diameter approximates a straight line with negative slope. Although insufficient clinical data are available to confirm the level or slope of these curves, no doubt several variables, in addition to dose and volume, influence the probability of radiation damage. Still, Flickinger’s model can be helpful in comparing treatment plans to determine, for each particular patient, which plan has the least potential to cause complications (Fig. 25-15).

FIGURE 25-15 • Plot of isoeffective dose versus target diameter. The solid line represents the empirical power law dose-diameter isoeffect curve developed by Kjellberg [Kjellberg, 1983 #1726] for proton radiosurgery, thought to correspond to a 1% risk for radiation necrosis of human brain tissue. The open (linac) and closed (Gamma Knife) circies represent individual calculation points corresponding to a 3% risk for radiation necrosis, based on the theoretic integrated logistic model developed by Flickinger [Flickinger, 1989 #517]. A log-log plot of these data and the solid line would approximate a straight line with negative slope.

With the possible exception of AVMs and trigeminal neuralgia (see following), there are no clear dose-response data for targets treated with SRS. Most centers base doses on tumor volume and any restriction based on adjacent normal structures such as optic chiasm. Radiation Therapy Oncology Group (RTOG) 90-05 determined the maximum tolerated dose (MTD) of SRS in previously irradiated patients being treated for recurrence.²⁵ Dose escalation was performed with stratification of patients into three categories based on maximum tumor diameter: less than 20 mm, 21 to 30 mm, and 31 to 40 mm. As expected, patients with smaller tumors tolerated higher doses. For the group with tumors smaller than 20 mm, 24 Gy was established as safe. Further dose escalation in this group was not possible because of lack of patient accrual at the 27-Gy level (presumably because of the reluctance of treating physicians to treat patients at this dose.) MTDs of 18 Gy and 15 Gy were established for patients with tumors between 21 and 30 mm and between 31 and 40 mm in diameter, respectively.

Although these results are extremely valuable and may serve as guidelines, several factors should be kept in mind when choosing SRS dose.²⁶ First, patients in RTOG 90-05 were previously irradiated patients. Patients receiving SRS as a boost, immediately following whole-brain radiation therapy (WBRT), for example, for metastatic disease, may neither require nor tolerate the doses established as MTD in the protocol. Some clinicians decrease doses by approximately 10% when treating as a boost compared with treatment with SRS alone. Other special circumstances may also apply. For example, it is clear that acoustic neuroma is well treated with doses well below the RTOG MTDs, and that these doses would result in unnecessary toxicity (see the following text).

As complications of radiosurgical treatment have been documented more thoroughly, there has been a general reduction in dose levels prescribed. Central to the practice of radiosurgery are four general principles: (1) Slowly proliferating tissues may not manifest a response for months or years after treatment; (2) a slow response does not necessarily indicate that the lesion is radiation-resistant; (3) the latent interval can be decreased by increasing the dose delivered; and (4) response often reflects proliferative activity rather than intrinsic radiation sensitivity.²⁷ Radiosurgical targets that are slowly proliferative include AVMs and benign tumors (meningioma, pituitary adenoma, and acoustic neuroma). Those lesions and other slowly growing tumors usually do not require a radiation dose that unduly risks damage to normal tissue to produce a rapid response or tumor necrosis.

Results of Radiosurgical Treatment

Arteriovenous Malformations

The natural history of AVMs represents a potentially life-threatening intracranial process, as is evident from the retrospective study of 160 patients with untreated AVMs (mean follow-up, 23.7 years) in which Ondra and associates ²⁸ found an annual rate of spontaneous hemorrhage of 3.9% and annual morbidity and mortality rates of 2.4% and 1.0% respectively. It appears that AVMs are cured, with little further risk of hemorrhage, when total microsurgical resection or total radiosurgical obliteration can be angiographically demonstrated.²⁹ Because there seems to be a persistent risk of hemorrhage during the latent interval following radiosurgery, complete microsurgical excision of the AVM is the preferred therapy whenever it can be accomplished safely.³⁰ When this is not possible, radiosurgery can obliterate both superficial AVMs and deep-seated, relatively inaccessible AVMs, with low morbidity and mortality rates and minimal hospitalization (Table 25-1). Radiosurgery induces a progressive thickening of the vascular wall and luminal thrombosis, which take months to years to complete. Reported obliteration rates are 74% to 92% at 2 to 3 years after radiosurgery (see Table 25-1).³¹^–³⁵ In most series, the obliteration rates at 2 years after radiosurgical treatment of small AVMs are higher (80% to 100%) than those for larger AVMs (40% to 70%).³⁶ This has been shown to be due to the higher minimum doses delivered to smaller AVMs.³⁷^,³⁸ Minimum doses of 13 and 25 Gy have been associated with in-field obliteration rates of 50% and 98%, respectively.³⁸^,³⁹ The “gold standard” for success following SRS for AVM is angiographic evidence of complete obliteration. Patients may initially be followed with MRI until the nidus is no longer evident, at which time complete obliteration may be confirmed on angiogram.⁴⁰ Repeat SRS for residual nidus demonstrated 2 to 3 years after SRS may result in subsequent obliteration in up to 85% of cases with low rates of morbidity.^35,^41,⁴² Likewise, an approach of repeat SRS for large AVMs may lead to improvement in obliteration rates, compared with a single session of low dose SRS.⁴³

Table 25-1 Results of Radiosurgery for Arteriovenous Malformation

The rate of permanent residual neurologic injury after radiosurgery for AVM is 3% to 4%.^35,^42,⁴⁴^–⁴⁶ In about one third of patients, follow-up MRIs show transient increased T₂ signal changes surrounding the target that develop about 6 to 18 months after radiosurgery and resolve a year or so thereafter. These changes depend on the location and size of the target.⁴⁷ Approximately one third of patients who develop these T₂ signal changes also have transient neurologic symptoms.

Series treating patients younger than 18 years of age with radiosurgery for AVM have shown results comparable with those found in adults when similar doses are used. Smyth and colleagues ⁴⁸ reported on 40 children treated with SRS for AVM. Results were significantly correlated to marginal dose. Obliteration rates were 7% and 63% for marginal dose of less than 18 Gy and more than 18 Gy, respectively. Similarly, others have reported obliteration rates of 77% to 80% for small AVMs (<3 cc) and 61% to 65% for AVMs 3 to 10 cc in volume.⁴⁹^,⁵⁰

Other related vascular lesions such as dural arteriovenous (AV) fistulae and cavernous malformations have also been treated with SRS. Dural AV fistulas treated in a fashion similar to AVMs (dose 10 to 28 Gy based on volume) demonstrated a 68% to 77% obliteration rate at 2-year follow-up. An additional 24% of shunts had significantly reduced flow, indicating a potential obliteration rate of up to 90% with further follow-up.⁵¹^,⁵² Cavernous malformations have also been treated with SRS in selected “high risk” patients. Hasegawa and colleagues⁵³ reported on 82 such patients who had experienced at least one major hemorrhage. During an average observation period of 4.3 years, the annual risk of hemorrhage was 33.9%. Following SRS (median dose 16.2 Gy), the risk of hemorrhage was reduced to 12.3% in the 2-year period after SRS and to 0.76% for years 3 through 12. Of these patients, 13% had new neurologic symptoms unrelated to hemorrhage following SRS; half of these were mild and transient. The authors recommend SRS for “high-risk” patients who have suffered at least one major hemorrhage.

Benign Tumors

For most benign intracranial tumors, fractionated radiation therapy has been the preferred treatment, and it produces few significant complications, especially when performed with three-dimensional planning and delivery. Whether fractionated radiation therapy or radiosurgery is performed, it is not necessary to cause significant tumor necrosis or to obtain a complete radiographic response to achieve tumor control. Rather, the goal is to halt the tumor’s growth permanently and, in the case of a hormonally active pituitary tumor, to halt abnormal hormone production permanently as well. For most benign tumors, neither fractionated radiotherapy nor SRS is likely to rapidly decompress mass effect from tumor.

Benign tumors commonly treated by radiosurgery are acoustic neuromas, meningiomas, and nonsecreting pituitary adenomas or those secreting somatotropin, adrenocorticotropin, or prolactin. Clinical observations on the radiosurgery of these lesions have provided fundamental information about the tolerance to single-fraction treatment on the part of the pituitary gland, the cranial nerves of the cavernous sinus, and the mesial temporal lobes.⁵⁴^–⁵⁸

Acoustic Neuroma

The role of radiosurgery for acoustic neuroma as an alternative to microsurgery is a topic of lively debate. The radiosurgical progression-free survival (PFS) rate with acoustic neuromas has been reported to be more than 90%.⁵⁹ The early use of high doses in the range of 20 Gy at the tumor margin resulted in high rates of transient cranial nerve V and VII neuropathies. There was also hearing loss in approximately half of patients with useful hearing at the time of SRS. Because of these toxicities and the high rate of tumor control, doses have decreased during the past 10 years (Table 25-2). Chopra and colleagues recently reported a 98.3% 10-year control rate in more than 216 patients treated with the gamma knife to marginal doses of 12 to 13 Gy.⁶⁰ No patient in this group developed a new facial neuropathy. Eight patients (3.7%) developed some degree of trigeminal neuropathy At 3 years there was a 78% probability of maintaining initial Gardner-Robertson hearing level in 110 patients with testable hearing. Similar results have been obtained by other institutions,⁶¹^–⁶⁴ and marginal doses in the range of 12 to 13 Gy have become the standard.

Table 25-2 Results of Radiosurgery for Acoustic Schwannoma

Follow-up imaging following SRS for acoustic schwannoma may show early enlargement of the tumor. Pollock reported a 14% rate of enlargement at a median of 9 months following SRS.⁶⁵ The majority of these tumors (86%) did not subsequently demonstrate serial progression.

Fractionated stereotactic radiotherapy or staged radiosurgery has been used to treat acoustic neuromas with very similar results as those reported for SRS. Tumor control rates of more than 90% for treatment regimens of 50 to 54 Gy in standard fractions or 20 to 35 Gy in fractions of 4 to 5 Gy have been reported with virtually no incidence of cranial neuropathy and excellent preservation of hearing rates similar to low-dose SRS.⁶⁶^–⁶⁹

Meningioma

SRS is increasingly becoming the sole treatment or a component of the treatment of intracranial meningioma. During the previous decade there have been a number of reported series with substantial follow-up demonstrating excellent local control of benign meningioma following SRS (Table 25-3).

Table 25-3 Results of Radiosurgery for Benign Meningioma

Stafford and colleagues reported on 184 World Health Organization (WHO) grade-1 meningiomas in 168 patients treated with SRS.⁷⁰ Local control at 5 years was 93% following marginal doses of 12 to 36 (median 16 Gy). These authors found no correlation between marginal dose and tumor control. Kreil and colleagues reported on 200 patients treated with a median dose of 12 Gy (7-25 Gy) SRS for meningiomas ranging in volume from 0.38 cc to 89.8 cc.⁷¹ Local control probability was 97.2% at 10 years with a complication rate of 2.5%, including temporary symptoms relating to peritumoral edema. Kollova and colleagues from the Czech Republic recently published their series of 325 patients with a median meningioma volume of 4.4 cc (0.11-44.9 cc).⁷² The median marginal tumor dose was 12.6 Gy (6.5-24 Gy). Five-year actuarial tumor control was 97.9%, with 69.7% of tumors showing some regression and 27.8% remaining stable in size. A strict dose limit was set for the optic pathways at 8 Gy for previously unirradiated patients and 3 Gy for previously irradiated patients. Temporary treatment-related toxicity was reported in 10.3% of patients and permanent effects were suffered by 5.7%. However, no patient was reported to have an optic nerve-related complication. Tumor control was significantly compromised when the marginal dose was less than 12 Gy. Patients treated with marginal doses in excess of 16 Gy had a significantly increased risk of peritumoral edema, temporary or permanent sequelae of radiosurgery. The authors conclude a dose range of 12 to 16 Gy for meningiomas is optimal. The University of Florida group treated 181 patients with WHO grade I meningioma with linac-based radiosurgery. A median marginal dose of 13.14 Gy (10-20 Gy) was delivered to a median treatment volume of 7 cc (0.3-48.6 cc). The local control rate was 96% at 5 years.⁷³

Often meningiomas are treated without biopsy, based on characteristic appearance on MRI scan. Flickinger and colleagues ⁷⁴ reported on a series of 219 meningiomas diagnosed by imaging criteria. Marginal doses of 8.9 to 20 (median 14) Gy were prescribed. The actuarial local control rate was 93.2% at 5 and 10 years. The probability of misdiagnosis was 2.3% at 10 years. The historical trend toward using lower marginal doses after 1991 (14 Gy versus 17 Gy) did not decrease the local control rate. However, in the later period, when stereotactic MRI was used for planning, the rate of complication was significantly lower: 5.3% versus 22.9%, when CT alone was used. This confirmed a previously reported excellent local control of 93% for meningiomas diagnosed by imaging compared with 60% for those previously resected.⁷⁵

Cavernous sinus involvement with meningioma, in which complete surgical resection is not usually considered an option, presents a special situation because the optic nerve or chiasm may lie very close to tumor and patients are at high risk for either tumor- or treatment-related morbidity. The Gamma Knife group in Milan recently reported on their experience in treating 123 patients with cavernous sinus meningiomas with a mean tumor volume of 7.99 cc (0.7-30.5 cc). Tumor control was achieved in 121 (98.4%) patients with a median follow-up of 3 years. Of 73 patients with cranial nerve deficits prior to SRS, 23 (31%) improved following treatment. Five patients (4.1%) were reported to have treatment-related adverse effects. Of these, only one was permanent (CN VI palsy) occurring 8 months after radiosurgery. Optic-nerve dose was limited to 10 Gy and no patient experienced visual deterioration.⁷⁶ The Pittsburgh Gamma Knife group reported on 155 patients with benign cavernous sinus meningioma, most treated with marginal doses of 12 to 14 Gy. Local control was 93.1% at 10 years with 34% of tumors regressing on follow-up imaging. Eight patients experienced adverse effects, including three patients with visual deterioration who received 12 Gy to the optic nerve.⁷⁷

Pollack compared radiosurgical to surgical outcomes for patients with small to medium (<35 mm) meningiomas.⁷⁸ SRS resulted in a probability of local control at 7 years of 95%, equivalent to that found following a Simpson grade 1 resection (96%). Less extensive surgery resulted in poorer outcome—82% and 34% 7-year local control for Simpson grade 2 and grades 3 to 4 resections, respectively. Complications were lower in the SRS group (10%) than in the surgical group (22%). It was suggested that SRS will likely become the preferred treatment for patients with meningiomas in this size range and without mass effect.

Based on findings available to date, radiosurgery appears to be safe and effective in producing a durable arrest of growth of small, inoperable, recurrent, or residual meningiomas. Early experience indicated that treatment of large-volume tumors or the use of high marginal doses resulted in excessive risk of treatment-related sequelae.^24,^79,⁸⁰ Marginal doses in the range of 12 to 16 Gy appear to afford an excellent probability of tumor control and a low risk of complication. When the optic nerve or chiasm receive doses of less than 10 Gy, the risk of neuropathy is low. The risk of optic neuropathy when dose is limited to less than 8 Gy is likely less than 1%.⁷² Leber and colleagues reported that higher doses carry substantial risk of visual impairment that may not manifest for many years after treatment. A group of patients receiving optic nerve doses of 10 to 15 Gy had a 26.7% incidence of optic neuropathy, which increased to 77.8% in a group receiving more than 15 Gy.⁵⁸ Conformal radiotherapy delivered in standard fractionation (54 Gy in 30 fractions) has been shown to result in excellent tumor control, with low rates of complications and, in particular, an extremely low risk of optic neuropathy.⁸¹^,⁸² Fractionated therapy should be considered when optic-nerve tolerance is likely to be exceeded with SRS.⁸³ We do not recommend SRS for the treatment of optic-nerve-sheath meningioma.

Tumor volume has been shown to strongly correlate with local control and overall survival (OS) of patients with WHO grade 1 meningioma treated with SRS. DiBiase and colleagues found 5-year disease-free survival (DFS) and OS of 86.2% and 91.0%, respectively, for 162 patients treated with SRS.⁸⁴ Both DFS and OS were significantly better for patients with tumors smaller than 10 cm³ compared with those with larger tumors: 91.9% versus 68.0% (DFS) and 100% versus 59.7% (OS), respectively. Milker-Zabel and colleagues reported a 5-year local control rate of 95.7% for meningiomas of up to 60 cm³ in volume following fractionated stereotactic radiotherapy and standard fractionation (57.6 Gy in 32 fractions) compared with 84.5% for larger tumors.⁸⁵ A subset of patients undergoing neuropsychiatric testing showed no decrement in function following fractionated treatment.⁸⁶ So existing data indicate that, although control rates for SRS and fractionated radiotherapy are similar for smaller meningiomas, larger tumors may be better suited to fractionated radiotherapy.

Pituitary Adenoma

Nonfunctioning Tumors

Pituitary adenomas have been treated with radiosurgery for approximately 50 years using a variety of doses and techniques with excellent long-term local control.⁸⁷^,⁸⁸ Recent large series of nonfunctioning pituitary adenomas treated with SRS report control rates of more than 90% (Table 25-4).⁸⁹^–⁹¹ Response rates are 60% to 90% and may continue to increase with follow-up longer than 3 years.⁹⁰ Marginal doses used in these studies ranged from 5 to 35 Gy. Mingione and colleagues⁹⁰ demonstrated a significant reduction in tumor-control rates for marginal doses less than 12 Gy, compared with more than 12 Gy (68.7% versus 97.3%, respectively, P < 0.05). There was no apparent benefit when marginal dose exceeded 20 Gy.

Table 25-4 Results of Radiosurgery for Nonfunctioning Pituitary Adenoma

Tumor-control rates for nonfunctioning adenomas reported for SRS are equivalent to those reported following fractionated radiotherapy. Brada and colleagues reported PFS of 97% and 92% at 10 and 20 years, respectively, for 252 patients treated with conventional radiotherapy of 45 to 50 Gy.⁹² Milker-Zabel and colleagues⁹³ reported a 93% 5-year PFS in 60 patients with nonfunctioning pituitary adenomas treated to a median dose of 52.2 Gy in 29 fractions with stereotactic radiotherapy. The choice of one modality over the other should be based on patient and tumor characteristics. SRS may be the preferred treatment option for patients with discrete residual tumor located at least 5 mm from the optic apparatus. Fractionated radiotherapy, used successfully for more than 50 years to treat pituitary adenomas, should be recommended for tumors abutting optic nerve or chiasm and for diffuse or large tumors.

The actuarial risk of SRS-related hypopituitarism is dose- and axis-related.⁸⁹ Vladyka and coworkers reported that the mean pituitary dose is the best correlate to risk of hypopituitarism following SRS. A mean pituitary dose of 15 Gy was found to pose little risk of subsequent thyrotropic, gonadotropic, or adrenocorticotropic function. Mean doses of more than 17 Gy for gonadotropic and thyrotropic function and more than 20 Gy for adrenocorticotropic function resulted in a 50% rate of hypopituitarism at 5 years.⁹⁴ Dose to the pituitary stalk and hypothalamus may also contribute to hypopituitarism following SRS.

Secretory Tumors

Pituitary adenoma hypersecreting growth hormone (GH) and causing acromegaly is a serious, disabling disease associated with reduced life expectancy. The risk of death resulting from cardiovascular and respiratory disease and malignancy have been reported to be increased two- to threefold in the acromegalic patient population. Reduction of GH levels to near-normal levels is associated with reduced risk of death. Surgery is nearly always offered as initial management of these patients, but results in a biochemical cure in only 50% to 70% of patients.⁹⁵ A postoperative serum GH level in excess of 2 ng/ml is associated with a nearly threefold risk of death compared with lower levels.⁹⁶

SRS for residual GH-secreting adenoma to reduce GH levels to near normal has been reported in several recent studies (Table 25-5). The Mayo Clinic group reported on 46 patients treated for acromegaly with a median pretreatment GH level of 8.4 ng/ml (0.5–82 ng/ml) and a insulin-like growth factor-I (IGF-I) level 2.25 times the upper limit of normal.⁹⁷ Biochemical control was defined as a GH level of 2 ng/ml and a normal age- and sex-adjusted IGF-I with no suppressive medications. The median marginal dose was 20 Gy (14.4-30 Gy). At a median of 36 months (6-63 months) following SRS, 50% of patients achieved biochemical control. The actuarial rates of biochemical remission were 11% and 60% at 2 and 5 years, respectively. The rate of remission was strongly correlated with a pretreatment IGF-I of less than 2.25 times normal and the absence of pituitary suppressive medications at the time of SRS.

Table 25-5 Results of Radiosurgery for Acromegaly

Attanasio and colleagues ⁹⁸ reported on 30 acromegalic patients treated with gamma knife SRS to a median dose of 20 Gy (15-35 Gy). The 5-year probability of normalized GH (<2.5 ng/ml) was 50%. IGF-I levels were normal for age and sex in 30% at 5 years. Strict biochemical cure (both GH and IGF-I normalized) was achieved in 30% of patients at 5 years. There was no case of tumor growth and 79% of tumors were decreased in size at 4 years.

Jezkova and colleagues ⁹⁹ reported on 96 patients with acromegaly treated with 10 to 42 Gy (35 Gy, median) SRS. For the strictest criteria of control (GH < 1.0 ng/ml on oral glucose suppression and normal IGF-I for sex and age), control rates were 14.6%, 28.6%, and 44.2% at 1, 2, and 5 years, respectively. Corresponding probabilities of reducing GH to a “safe” level (<2.5 ng/ml) were 23.9%, 47.1%, and 67.4%, respectively.

It has been suggested that SRS results in more rapid normalization of hormone hypersecretion in patients with pituitary tumors.¹⁰⁰ Minniti and colleagues¹⁰¹ treated 47 acromegalic patients with fractionated radiotherapy of 45 to 50 Gy in 25 to 28 fractions. Normal suppression of GH during oral glucose tolerance test (<1 ng/ml) was demonstrated in 9%, 29%, 52%, and 77% of patients at 2, 5, 10, and 15 years, respectively. Likewise Powell and colleagues¹⁰² reported on 32 acromegalic patients treated with 45 to 54 Gy of “conventional” radiotherapy. A normal IGF-I level was found in 43.7% of these patients with a mean follow-up of 5.6 years. Long-term follow-up of 884 acromegalic patients treated with conventional radiotherapy was reported by Jenkins and colleagues and the UK Acromegaly Register Group.¹⁰³ GH normalization (<2.5 ng/ml) was achieved in 22%, 36%, and 60% of patients at 2, 5, and 10 years, respectively. Corresponding IGF-I normalization was seen in 38%, 50%, and 63% of patients, respectively. These data suggest a similar rate of normalization with fractionated radiotherapy compared with SRS.

Cushing syndrome is a constellation of signs and symptoms secondary to an overproduction of adrenocorticotropic hormone (ACTH). The most common cause of Cushing syndrome is an ACTH hypersecreting pituitary microadenoma, or Cushing disease. Features of Cushing syndrome include fatigue, muscle weakness, hypertension, depression, cognitive impairment, purplish skin, striae, easy bruising, hyperpigmentation, loss of libido, diabetes, hirsutism, acne, and menstrual disorders.¹⁰⁴ Transphenoidal resection of the microadenoma is the treatment of choice leading to remission in 65% to 90% of patients.¹⁰⁵ Morning serum cortisol levels of greater than 5 mcg/dl requires careful monitoring. The recurrence rate in these patients is up to 25% at 10 years, and may be higher in patients younger than 25 years of age.¹⁰⁶ For patients with persistent or recurrent disease, based on serum cortisol or urine-free cortisol levels, treatment options may include repeat surgery, bilateral adrenalectomy, medical therapy with suppressive drugs, or radiation therapy.¹⁰⁵

Radiosurgery has been used to treat patients who have failed surgery for Cushing disease (Table 25-6).¹⁰⁷^–¹¹⁰ Jagannathan and colleagues reported on 90 patients treated with gamma knife radiosurgery for Cushing disease after failed surgery. Of these, 49 patients (54%) achieved remission, defined as a normal 24-hour urine-free cortisol. New hormonal deficits appeared in 20 patients (22.3%) following SRS. The Gamma Knife group in Marseilles reported on 40 patients with Cushing disease.¹⁰⁷ At a median follow-up of 52 months, 42% of patients were in remission and 15% were reported to have new endocrinopathies. The Massachusetts General Hospital proton group recently reported their experience with single fraction SRS for 33 patients with Cushing disease.¹¹⁰ At a median follow-up time of 62 months, 52% of patients had achieved remission. No second tumors, optic neuropathy, or necrosis was noted. However, 52% of patients developed new pituitary deficits. Devin and colleagues¹⁰⁸ treated 35 patients with Cushing disease with linac-based SRS. Initially 17 (49%) of patients achieved remission. However, 4 patients subsequently relapsed 17 to 64 months following SRS. Of these patients, 40% developed new endocrine deficiency.

Table 25-6 Results of Radiosurgery for Cushing Disease

Results treating Cushing disease with conventionally fractionated radiotherapy are strikingly similar to those published for SRS.¹¹¹ Minniti and colleagues¹¹² treated 40 patients with 45 to 50 Gy in 25 to 28 fractions for Cushing disease. Normalization of plasma cortisol was achieved in 28%, 73%, 78%, and 84% of patients at 1, 3, 5, and 10 years following radiotherapy, respectively. Hypopituitarism was observed in 62% and 76% of patients at 5 and 10 years, respectively. Estrada and colleagues¹¹³ treated 30 patients with Cushing disease with 48 to 54 Gy (1.8-2.0 Gy daily). At a median follow-up of 42 months (18-114 months), 25 of 30 patients achieved remission with an average time to remission of 18 months. The probability of remission was 44% and 83% at 1 and 3 years. No patient who achieved remission subsequently recurred. No patient suffered visual impairment, second tumor or brain necrosis. Hypopituitarism was seen in 57% of patients following radiotherapy.

Radiosurgery is an effective adjunct to surgery or in the setting of recurrence for Cushing disease. Data does not seem to indicate a more rapid response nor a higher rate of response than that reported for conventionally fractionated radiotherapy.¹¹¹ The choice of radiotherapeutic modality should be made on the basis of tumor size and location. In the case of larger tumors or those abutting optic nerve or chiasm, fractionated radiotherapy is likely to be the therapy with the higher therapeutic ratio.

Studies reporting on SRS for prolactinomas report less successful outcomes. Landolt and Lomax ¹¹⁴ treated 20 patients with prolactinoma with gamma knife radiosurgery and achieved normalized prolactin levels in only 5 (20%). They found that the rate of normalization was higher (45%) in patients who were not receiving dopamine agonists at the time of SRS, compared with those who were (0%). Pouratian and colleagues¹¹⁵ recently reported on 23 patients treated with gamma knife SRS for prolactinoma. At a median follow-up time of 55 months, 26% had achieved normalized prolactin levels.

Brain Metastases

The median survival for patients with symptomatic metastases to the brain is about 1 month if they remain untreated and about 3 to 6 months if they undergo conventional WBRT, irrespective of the fractionation scheme used.¹¹⁶ As compared with patients receiving only WBRT, those treated with surgery and radiation therapy for solitary brain metastases have a significantly longer survival (40 weeks versus 15 weeks, p < 0.01) and improved quality of life, and significantly fewer patients have recurrence at the initial metastatic site (20% versus 52%, p < 0.02).¹¹⁷ Systemic chemotherapy has not been shown to influence the outcome to any great extent. When patients with extracranial metastases are considered for promising therapies with biologic response modifiers, those with brain metastases are usually excluded. Therapies such as SRS aimed at control of intracranial metastases are therefore of great potential value.

Because of their biologic and physical characteristics, metastatic lesions are well suited for radiosurgery. They are usually relatively small when diagnosed, and they are often nearly spherical and radiographically distinct. They are minimally invasive and displace normal brain parenchyma circumferentially beyond the target volume, reducing the risk of radiation injury to normal tissue. Although surgery for metastases has certain advantages in that it provides immediate resolution of mass effect, provides diagnostic information if needed, and entails no risk of radiation necrosis, radiosurgery has the substantial advantage that the entire extent of disease can usually be encompassed within the treatment volume because treatment is linked directly to three-dimensional visualization of the tumor. Moreover, radiosurgery avoids the risks of hemorrhage, infection, and tumor seeding, and, as it requires minimal or no hospitalization, the costs of therapy are less than those of surgery.¹¹⁸ Local control can be expected in 73% to 98% of patients treated with radiosurgery. However, because many patients with metastatic cancer may have limited survival, it is more informative to refer to actuarial local control following SRS in such patients. The results of radiosurgical treatment for single brain metastases appear to be at least equivalent to those of surgery followed by WBRT (Table 25-7).

Table 25-7 Freedom from Local and Distant Failure Following Stereotactic Radiosurgery With or Without Whole-Brain Radiation Therapy for Brain Metastases

The target volume for brain metastases treated with radiosurgery is most often the enhancing tumor defined on CT or MRI scan. It has been suggested, based on pathologic examination, that brain metastases infiltrate at least 1 mm beyond gross tumor borders.¹¹⁹ Noel and colleagues reported a significant increase in local control when a 1-mm CTV margin was added.¹²⁰ These authors routinely used a relatively low prescription dose of 14 Gy to the margin of the CTV. Therefore, the addition of a margin would have increased the dose to the GTV, likely at a steep portion of the dose-response curve. Recently, Nataf and colleagues¹²¹ found that the addition of a 2-mm margin did not improve local control or survival, but resulted in a significant increase in parenchymal complications.

Many centers use the RTOG 90-05 guidelines to assign prescription dose for brain metastases.²⁵ These guidelines recommend “safe” doses of 15 Gy, 18 Gy, and 24 Gy for tumors smaller than 20 mm, 21 to 30 mm, and 31 to 40 mm diameter, respectively. It is unclear if doses larger than 20 Gy for small metastases result in superior local control.¹²²^,¹²³

Actuarial local control rates 1 year following SRS alone in treating single or multiple metastases range from 68% to 91%.^122,¹²⁴^–¹²⁸ In a multicenter study reported by investigators from several gamma knife facilities, 116 patients were treated with gamma knife radiosurgery for single brain metastases (mean minimal dose, 17.9 Gy).¹²⁸ Local control of tumor was achieved in 99 patients (85%). Among all patients, the 1- and 2-year actuarial tumor control rates were 75% and 67%, respectively, with a plateau in the curve at 18 months. These findings suggested that, for most patients treated, local control rates were durable. Multivariate analysis showed that the patients with better local control were those who underwent WBRT in addition to radiosurgery. Local control rates at 1 year were 81.2% and 68% for patients receiving WBRT and those receiving SRS alone, respectively. Paradoxically, 42 patients with brain metastases of histologies reputed to be resistant to fractionated radiation therapy (melanoma and renal cell carcinoma) also had improved local control: 100% at 2 years compared with 51.5% for other histologies, a finding also reported by others.¹²⁹ Median survival from the time of radiosurgery was 11 months. Patients with breast cancer had a significantly superior median survival of 18 months.

In a single-institution, randomized, prospective trial Kondziolka and colleagues ¹³⁰ examined the relative efficacy of WBRT (30 Gy in 12 fractions) alone compared with the same WBRT plus SRS. Patients with two to four brain metastases with known histology were eligible. Twenty-seven patients were accrued before stopping rules were met. Local failure at 1 year was 100% for WBRT alone versus 8% with the addition of SRS. There was no significant difference in OS between the groups.

RTOG 95-08 has also studied the benefit of WBRT (37.5 Gy in 15 fractions) compared with WBRT plus a boost with SRS in patients with one, two, or three brain metastases.¹³¹ At 1 year the local control rates were 82% and 71% for WBRT and SRS combined and WBRT alone (p = 0.01), respectively. For patients with single metastases, a significant survival benefit was demonstrated for the addition of SRS (6.5 versus 4.9 months). Similar significant survival advantages of 1.6 to 2 months were demonstrated for patients in RTOG recursive partitioning analysis (RPA) class I,¹³² patients younger than 50 years of age, and those with non–small cell lung cancer, regardless of number of metastases. Karnofsky performance status (KPS) was improved or stable at 6 months in 43% versus 27% in SRS-treated patients versus those receiving WBRT alone (p = 0.03), respectively. There was no statistical difference in the rate of grade 3 or 4 toxicities between the groups.

The addition of WBRT to SRS for treatment of brain metastases has universally been shown to improve local control ^122,^124,^126,^128,¹³³ and to decrease the risk of new metastases developing elsewhere in the brain.¹²⁴^,¹³⁶ Fuller and associates¹³³ reported significantly improved local control within the radiosurgical target region or elsewhere in the brain for patients undergoing SRS plus WBRT as compared with those who received only SRS, although there was no significant difference in survival rates for the two groups. Similarly, Flickinger and associates¹²⁸ showed that significantly fewer patients receiving whole-brain irradiation plus radiosurgery had recurrence at the radiosurgical target site.

The Japanese Radiation Oncology Group randomized 132 patients with one to four brain metastases smaller than 3 cm in diameter to SRS alone or WBRT (30 Gy in 10 fractions) and SRS.¹²⁴ SRS dose was determined by tumor size and treatment arm: 22 to 25 Gy for tumors smaller than 2 cm and 18 to 20 Gy for tumors 2 to 3 cm in the SRS alone arm and 30% less dose for the WBRT and SRS arm. OS was 8.0 months in the SRS alone group and 7.5 months in the group receiving WBRT (p = 0.42). Recurrence rates anywhere in the brain at 1 year were 46.8% for those receiving WBRT and 76.4% following treatment with SRS alone (p = 0.02).

These results are parallel to those found for the addition of WBRT to surgery for single brain metastases. Patchell and colleagues ¹³⁴ found that the addition of high-dose WBRT (50.4 Gy) decreased failure at the surgical site from more than 75% to less than 20% at 1 year and decreased failure elsewhere in the brain from more than 50% to less than 20%. However, the addition of WBRT to surgery or SRS has not been demonstrated to increase OS, given that effective salvage therapies are available (WBRT, further SRS or surgery or combinations thereof).¹²⁶^,¹³⁴ Lively debate continues as to whether the demonstrated benefits of WBRT outweigh the potential associated toxicity.¹³⁵ WBRT has been reported to reduce the risk of neurologic death following surgical resection of single brain metastases.¹³⁴ Patients with brain metastases have been shown to have impaired neurocognitive function before receiving WBRT.¹³⁶ A prospective study from Germany found similar results in patients studied prior to WBRT and also reported that after minimum follow-up of 9 months after WBRT, these patients showed no significant decline in any tested neurocognitive function.¹³⁷ Although Regine and colleagues¹³⁸ reported no significant change in neurocognitive testing among a cohort of patients with controlled brain metastases following WBRT; there was a dramatic and significant decline in patients treated on the same study in whom brain metastases were uncontrolled. Aoyama and colleagues¹²⁴ found no difference between the arms receiving SRS alone or with WBRT in survival, death from neurologic causes, neurocognitive function, or treatment-related toxicities.

However, it is argued that WBRT may be omitted without compromising patient survival or control of intracranial disease with close follow-up and salvage therapy.^125,^126,¹³⁹ Using this approach, more than 70% of patients initially managed with SRS alone may be spared WBRT.

In a series of 421 metastatic lesions in 248 patients who were treated at one center during a 7-year period, Alexander and colleagues ¹⁴⁰ reported that control rates for lesions with radiation-resistant histologic findings were statistically equal to those of other lesions. Several studies have found that “radioresistant” histologies are less well controlled with SRS than other histologies. A recent Eastern Cooperative Oncology Group Study of 31 patients with melanoma (14), renal cell carcinoma (14), or sarcoma (3) found a local control rate of 67.8% at 6 months following SRS. Chang and colleagues¹⁴¹ reported 1-year local control rates of 64%, 47%, and 0% for metastases from renal cell carcinoma (n = 77), melanoma (n = 103), and sarcoma (n = 9), respectively. Recently reported control rates at 1 year for melanoma range from 47% to 81%.¹⁴¹^–¹⁴⁴ For renal cell carcinoma, 1-year local control rates of 64% to 93% have been reported.^141,¹⁴⁵^–¹⁴⁷ Differences seen among studies may be due to SRS dose used or differences in definition of local failure in these histologies prone to hemorrhage.

Within the range of doses commonly used for radiosurgery, there appears to be no clear-cut dose-response relationship, and the range of doses used does not appear to correlate with outcome. Most radiosurgical teams prefer to use a lower dose for larger lesions to minimize the risk of complications. Therefore, the response rate can be expected to correlate with target size. Among reported series, complete response was achieved in 78% of patients with lesions with tumor volumes less than 2 cm³ and in 50% of those with tumor volumes 10 cm³ or greater, and subsequent progression occurred in only 4% of complete responders versus 20% of partial responders.¹⁴⁸ In another series,¹⁴⁰ tumor volume greater than 3 cm³ was significantly associated with higher rates of local failure in univariate analysis, although in multivariate analysis the association was not significant (p = 0.06).

Many groups omit WBRT and use SRS alone in many circumstances. Sneed and colleagues retrospectively reviewed the outcome of 62 patients treated with SRS alone versus 43 patients treated with SRS plus WBRT for single or multiple brain metastases.¹²⁶ Patients receiving SRS alone were at significantly greater risk for developing new brain metastases. Salvage therapy in these patients was either repeat SRS or WBRT. Overall, 74% of patients in the SRS-alone group were spared WBRT. There was no difference in OS between the groups. In general, most patients who undergo radiosurgery for lesions metastatic to the brain experience clinical improvement and reduced steroid requirements, and only 11% to 25% eventually die of neurologic causes.^124,^131,^133,¹⁴⁸

Glioblastoma Multiforme

Although surgical resection followed by irradiation and chemotherapy lengthens the survival of patients who have glioblastoma multiforme (GBM), most patients die as a result of persistence or local recurrence of their tumor.¹⁴⁹^,¹⁵⁰ Reirradiation of focally recurrent malignant gliomas using stereotactic brachytherapy has resulted in durable local control and median survival of 12 months.¹⁵¹^,¹⁵² Radiosurgery for focally recurrent GBM has resulted in median survival times of 8 to 10.2 months (Table 25-8).¹⁵²^–¹⁵⁵

Table 25-8 Results of Radiosurgery for Recurrent Glioblastoma

For highly selected patients with small primary glioblastomas—a minority of the patients affected with this tumor—a highly conformal brachytherapy boost was shown to provide a longer than expected median survival of 20 months following conventional fractionated radiation therapy in a single-institution trial.¹⁵⁶ Because conformal dose distributions of radiosurgery and brachytherapy are similar, it was proposed that radiosurgery may provide an effective therapeutic boost and may have a similar potential for improving survival. Single-institution studies have reported longer-than-expected survival in highly selected patients with glioblastoma treated with standard surgery and radiotherapy followed by a boost with SRS to the enhancing tumor volume (Table 25-9).¹⁵⁷^–¹⁵⁹ These studies reported median survivals of 16 to 19.9 months and 2-year survivals of 36% to 40%. A large multi-institutional review found a median survival of 22.1 months and 2-year survival of 45%.¹⁶⁰ These findings were especially striking when comparisons were made based on RTOG RPA class.¹⁶¹

Table 25-9 Results of Radiosurgery Boost for Primary Glioblastoma

However, a larger prospective RTOG study failed to find any survival benefit for patients treated with SRS boost before standard postoperative radiotherapy. The ROTG 95-03 randomized patients with unifocal GBM smaller than 4 cm in greatest dimension to standard postoperative radiotherapy and chemotherapy or SRS treatment to residual enhancing tumor followed by standard radiotherapy and chemotherapy.¹⁶² Median survival times were not different between these two randomizations: 13.6 months versus 13.5 months, respectively. No subgroup analyzed demonstrated a survival benefit from upfront SRS. Of first failures, 62% were local only and 25% of patients randomized to radiotherapy alone received SRS or reirradiation for salvage at the time of failure.

Larson and colleagues ¹⁶³ performed multivariate analysis of 189 patients with WHO grades 1 through 4 glioma treated with the gamma knife. These investigators explored a broad spectrum of criteria for patient selection and treatment. The main endpoints of their study were survival and complications, and the analysis was designed to determine factors that were associated with better or worse outcomes. Multivariate analysis identified five variables significantly associated with decreased survival: higher pathologic tumor grade, older age of the patient, lower KPS score, larger tumor volume, and multifocal versus unifocal tumor. It is likely that variations in published reports regarding survival rates after radiosurgery or brachytherapy for malignant glioma can be attributed, in part, to substantial differences in patient populations with respect to these five variables. Patients with unfavorable constellations of these variables are unlikely to benefit from radiosurgery.

Radiosurgery for Functional Disorders

Radiosurgery has been used for the treatment of functional disorders for nearly 5 decades.¹⁶⁴ Such treatment usually requires extremely high doses (up to 180 Gy) to small, anatomically defined volumes. Treatment of Parkinson disease–related tremor or dystonia, intractable chronic pain, epilepsy, or obsessive-compulsive disorder are areas of current investigation.¹⁶⁵

SRS for trigeminal neuralgia is by far the most commonly practiced and accepted form of functional radiosurgery. A multi-institutional gamma knife study demonstrated that doses of more than 70 Gy of SRS delivered to the trigeminal nerve root entry zone adjacent to the pons resulted in a significantly greater probability of complete pain relief than lower doses, 72% versus 9%, respectively.¹⁶⁶ Goss and colleagues¹⁶⁷ reported on 25 patients treated with 90 Gy using a linac-based system. All 25 patients treated experienced excellent or good pain relief at a median of 2 months from SRS. Relapse was experienced in 32% of patients 4 to 13 months following SRS. Longhi and colleagues¹⁶⁸ recently reported on a series of 160 patients treated with gamma knife SRS for trigeminal neuralgia. Of these patients, 90% experienced good or excellent pain relief at a median time of 45 days. Recurrence was noted in 18% of patients at a median time of 14.2 months. Complications were more common in patients treated with doses in excess of 90 Gy and durable pain relief was significantly associated with doses of more than 80 Gy.

Normal-Tissue Tolerance to Radiosurgery

With increasing international experience with SRS, guidelines for reducing the risk of normal-tissue toxicity have emerged (Table 25-10). In general, normal tissues at risk depend on the location of the target volume. These include the brain parenchyma (edema, necrosis), brainstem (edema, necrosis, neuropathy), cranial nerves (neuropathy), and hypothalamic-pituitary axis (hypopituitarism). The interaction of dose and volume irradiated has not been clearly defined for most structures at risk.

Table 25-10 Recommended Normal Structure Dose Constraints to Avoid Complications Using a Single Fraction

Structure	Single-Fraction Dose Constraint	Reference #
Intraparenchymal brain lesion ≤ 2 cm	≤24 Gy (MTD unknown)	25
Brain lesion > 2 cm and ≤ 3 cm	≤18 Gy
Brain lesion > 3 cm and ≤ 4 cm	≤15 Gy
Brainstem	≤16 Gy Dmax	176
Pituitary	≤15 Gy (mean)	94
Optic chiasm and optic nerves	≤8-10 Gy	57, 58, 172
Cranial nerves III, IV, VI	30 Gy MTD unknown

57, 58 Cranial nerve V (trigeminal) ≤12.5-13 Gy 64, 77, 181–183 Cranial nerve VII (facial) ≤12.5-15 Gy 182, 184 Cochlea ≤3.7 Gy (mean) 173 Cranial nerves IX-XII MTD unknown 174, 176

MTD, Maximum tolerated dose; WBRT, whole-brain radiation therapy.

Dose constraint is a maximum unless otherwise specified.

The risk of treatment-related edema and necrosis following radiosurgery has been correlated with SRS dose and volume irradiated. Kjellberg and Abe ¹⁶⁹ first introduced the concept of the risk of postradiosurgery necrosis being related to dose and volume. They reported a log-log relationship of dose and volume that could define levels of risk between 1% and 99%. This model did not account for dose-volume effects outside the treatment volume. Flickinger¹⁷⁰ described the integrated logistic formula, which accounted for the contribution of all isodose volumes to the risk for complication. Subsequently Flickinger and colleagues correlated the risk of asymptomatic or symptomatic T₂ MRI changes following SRS for AVM to the volume having received 12 Gy.¹⁷¹

The RTOG Protocol 90-05 was designed to determine the MTD of radiosurgery for patients with malignant brain tumors or brain metastases who had failed prior conventional radiotherapy. Recommended dose levels found to result in less than 20% risk of serious complication were 15 Gy, 18 Gy, and 24 Gy for tumors smaller than 20 mm, 21 to 30 mm, and 31 to 40 mm, respectively.²⁵

Cranial nerve damage following SRS can result in blindness, ophthalmoplegia, facial weakness, numbness, or hearing loss. The special sensory cranial nerves II and VIII are apparently more sensitive to radiation than the motor nerves. Tishler and colleagues ⁵⁷ found that 24% of patients whose optic nerves received more than 8 Gy suffered visual loss, compared with none who received lower doses. These results were largely supported by Leber and colleagues⁵⁸ who demonstrated a dose response for optic neuropathy following SRS, finding the actuarial risk to be 0%, 26.7%, and 77.8% for doses of less than 10 Gy, 10 to 15 Gy, and more than 15 Gy, respectively. Stafford and colleagues¹⁷² reported on patients treated for cavernous sinus meningioma and found a less than 2% risk of optic neuropathy for nerves receiving less than 10 Gy and 6.9% for those receiving more than 12 Gy.

Early experience with SRS for acoustic neuroma using high-prescription doses of 18 to 20 Gy resulted in a high incidence of hearing loss (49%), facial weakness (21%), and trigeminal neuropathy (27%).⁵⁹ The subsequent change in practice to reduce the marginal tumor dose to 12 to 13 Gy has resulted in no compromise of tumor control but much lower rates of hearing loss (26%), facial weakness (0%), and trigeminal neuropathy (5%).⁶⁰ Increased incidence of hearing loss has been associated with relatively low doses (5.3 Gy) to the cochlea.¹⁷³

Lower cranial nerves are less well studied. Zhang and colleagues ¹⁷⁴ reported on 27 patients treated for jugular foramen schwannoma using marginal doses of 9.8 to 20 Gy and maximum doses of 25.4 to 50 Gy. No patient suffered a new cranial neuropathy of cranial nerves IX through XII. Martin and colleagues¹⁷⁵ treated 36 patients with jugular foramen schwannomas with marginal SRS doses of 12 to 18 Gy (maximum dose 21.4-36 Gy). No patient suffered an SRS-related neuropathy.

Meeks and colleagues ¹⁷⁶ modeled the risk of cranial nerve dysfunction and brainstem DVH. They found that a maximum brainstem dose in excess of 16 Gy was the most important risk factor for delayed cranial nerve deficit. The model predicted a risk of cranial nerve deficit less than 5% if one third or less of the brainstem received less than 16 Gy. Tolerance is diminished for larger volumes.

Leber and colleagues ⁵⁸ reported no cranial neuropathies related to cranial nerves III through VI in 210 nerves receiving up to 30 Gy with long-term follow-up. Tishler and colleagues⁵⁷ reported an incidence of 12% neuropathy for cranial nerves III through VI receiving doses of 15 to 40 Gy without a demonstration of a dose response.

Pituitary tolerance to SRS has been described by Vladyka and colleagues.⁹⁴ A strong dose-response was demonstrated. For the gonadotropic axis, dysfunction was found in 60% of patients for mean pituitary doses of more than 17 Gy. No patient receiving less than 15 Gy demonstrated dysfunction with follow-up of up to 90 months. Similarly, thyrotropic dysfunction was found in 85% of patients with long-term follow-up following mean doses of more than 17 Gy and no patients receiving less than 15 Gy. For adrenocorticotropic function, the risk was found to be 85% for doses larger than 20 Gy, compared with 10% when the mean pituitary dose was less than 20 Gy. No incidence of adrenocorticotropic dysfunction was found when the mean pituitary dose was 18 Gy or less. In general, pituitary function was maintained in all axes when the mean dose did not exceed 15 Gy.

Summary

Radiosurgical facilities have been established in many community hospitals and most academic medical centers, and the number of patients treated with radiosurgery is increasing every year. The number of cooperative group trials of radiosurgical applications is also increasing, and changes in practice patterns are occurring and are likely to continue to evolve. Radiosurgery is being incorporated into staged treatment protocols, together with embolization, surgery, and radiation therapy; and it is being used as immediate adjuvant therapy to prevent recurrence, in preference to observation followed by treatment at recurrence.¹⁷⁷

Also resulting from such work are certain generalizations that define radiosurgery. It has been established that the larger the target volume, the smaller the dose that can be delivered safely,¹⁷⁸ and the lower the probability of achieving control or obliteration of the target lesion. It is known that the development of symptomatic radiation necrosis is related to the dose, volume, site, and histology of the target lesion, and that cranial nerves II and VIII are particularly vulnerable to damage by radiation. Radiosurgery is also especially damaging to late-responding tissue such as abnormal blood vessels, which accounts for the high overall response to radiosurgical treatment of vascular tumors and AVMs. Tentatively, certain conclusions seem justified: