15

Chapter 15
Stereotactic Radiosurgery and Radiotherapy
John C. Flickinger
Ajay Niranjan
Stereotactic radiosurgery and radiotherapy are techniques to administer precisely directed, high-dose irradiation that tightly conforms to an intracranial target to create a desired radiobiologic response while minimizing radiation dose to surrounding normal tissue. These techniques exploit the fact that the radiation tolerance of normal tissue is volume-dependent. With these techniques the complication risks for any radiation dose delivered are reduced by minimizing or eliminating the margin of normal tissue otherwise included in the radiation treatment volume with conventional radiotherapy techniques. In the case of radiosurgery, all of the irradiation is done in a single session or fraction, while in stereotactic radiotherapy (SRT), more than one fraction of irradiation is administered. Table 15.1 lists the key requirements for successful stereotactic irradiation. Advances in imaging, computers, and treatment planning in the last 2 decades have led to the development of a variety of different stereotactic radiosurgery/radiotherapy techniques and their wider applications. Successful clinical experience with intracranial radiosurgery for a variety of applications has led to a re-examination of radiobiology and exploration of both fractionated approaches and extracranial applications. Margin reduction with radiosurgery and fractionated stereotactic irradiation techniques makes target definition accuracy more critical. Drawing contours from different imaging techniques or by different physicians are approaches to reduce target definition error (Fig. 15.1).
Terminology
Stereotactic refers to using a precise three-dimensional mapping technique to guide a procedure. The terminology used in stereotactic irradiation can be confusing. The term radiosurgery or stereotactic radiosurgery (SRS) is used for stereotactically guided conformal irradiation of a defined target volume in a single session. Radiosurgery or SRS can be delivered with Gamma Knife (Elekta Inc., Norcross, GA) modified LINAC radiosurgery systems (including Cyberknife (Accuray Inc., Sunnyvale, CA) and image-guided radiotherapy systems), tomotherapy, or proton beam systems. The term stereotactic radiation therapy (SRT) refers to stereotactically guided delivery of highly conformal radiation to a defined target volume in multiple fractions, typically using noninvasive positioning techniques. The term fractionated stereotactic radiosurgery (FSR) is limited to stereotactically guided high-dose conformal radiation administered to a precisely defined target volume in two to five sessions. Although it would have been less confusing to refer to this as hypofractionated SRT and reserve the term radiosurgery for single-fraction irradiation, the terminology is already in use. Adding intensity-modulated radiation therapy (IMRT) to the nomenclature can further complicate or confuse the terminology. Any radiation treatment plan that uses individual treatment beams that irradiate only part of the target at a time is IMRT. Strictly speaking, multiple isocenter radiosurgery (of a single target volume) meets the criteria for IMRT or stereotactic IMRT (SIMRT), but the term SIMRT is usually only used when multileaf collimators are employed. The terminology is useful to distinguish when the same linear accelerator is equipped to treat with either fixed circular collimators for radiosurgery (SRS or SRT mode), or to deliver IMRT using multileaf collimators (SIMRT mode).
Radiobiologic Considerations
Prior to the introduction of radiosurgery, essentially all clinical irradiation was administered with radiation dose fractions between 1.2 and 3 Gy for intracranial targets. Extracranial targets were usually treated with 1.2- to 4-Gy fractions with 6- to 8-Gy fractions used occasionally for treatment of bone metastases or malignant melanoma. Before the rapid adaption of radiosurgery into the clinic in the late 1980s, most radiation oncologists and radiation biologists believed that fractionating radiation treatment lessens the relative risk of injury to normal tissue compared with tumor in essentially all circumstances. Radiobiologic analysis of a limited number of malignant tumors in cell culture and clinical experience with conventionally fractionated radiotherapy of fast-growing malignant tumors established this radiobiologic dogma. Increasing the fractionation of radiotherapy for slow-growing benign tumors may not necessarily improve the balance between tumor control and radiation complication. Slow-growing benign tumors are difficult to study in either cell culture or animal models, so the effect of fractionation has not been well delineated. Stereotactic radiosurgery allowed clinicians to administer high single doses of radiation to intracranial targets with relative safety, thereby leading to a new appreciation of the underlying radiobiology. Laboratory studies that suggest that the radiation response for the high-dose single fractions used in radiosurgery is predominantly related to the supporting endothelial cells (25). Pathology studies of benign and malignant tumors treated by radiosurgery also support a vascular response (109).
Analysis of clinical data from radiosurgery to delineate dose-response relationships and define radiobiologic parameters is fraught with difficulties. Typical radiosurgery treatment plans use inhomogeneous dose distributions with the prescription isodose covering anywhere from 90% to 100% of the target volume. The absolute minimum dose to the target typically is 5% to 30% lower than the prescription dose. Contours of the same tumor/target volume or critical structures may vary slightly from one clinician to another. Using the linear-quadratic formula to extrapolate from experience with conventional radiotherapy experience with low-dose fractions to high-dose single fractions for radiosurgery appears problematic. Using single-fraction radiosurgery dose-response curves for arteriovenous malformation (AVM) obliteration, radiation injury to brain parenchyma and cranial nerves to calculate α/β ratios yields values of negative 30 to negative 60 rather than 2 to 3 as expected from conventional fractionated radiotherapy data (19,21). Comparing dose responses for fractionated SRT to radiosurgery is hampered by limited data with dose-response curves that have insufficient slopes for accurate comparison.
P.379
Radiosurgical Techniques
Radiosurgery was originally envisioned to treat intracranial lesions by delivering a high dose of ionizing radiation in a single-treatment session using multiple beams precisely collimated to the target inside the cranium. Advances in both imaging and computer technologies resulted in wider applications of radiosurgery. There are now a variety of different radiosurgery and SRT techniques available for intracranial and extracranial use.
Gamma Knife Radiosurgery
After prior experience with stereotactic treatment using orthovoltage radiotherapy and proton beam irradiation, Leksell and Larson created the first prototype of the gamma knife in 1967. The gamma knife uses a relatively hemispherical array of multiple fixed cobalt-60 beams (201 in most models) that are sharply collimated to create small, relatively spherical treatment volumes of varied diameter with sharp dose falloff. The earlier model originally was referred to as the U-style and contained cobalt sources arranged in hemispherical array,
P.380
including sources at the pole of the hemisphere. These units present challenging loading and reloading issues with the cobalt-60 sources, particularly with radiation protection. To eliminate this problem, the B-unit (Elekta, Inc., Norcross, GA) (after Bergen, Norway, the first site) was redesigned so that sources were arranged in an annular section of a hemisphere, similar to the Northern Hemisphere with the Arctic Circle excluded. In 1999, the model C version of the gamma knife was introduced with the option to use robotic positioning to set treatment coordinates. This expedites execution of multiple-isocenter treatment plans. Manual positioning is still needed for some targets far from the center of the head. The model 4-C, introduced in 2005, was equipped with enhancements designed to improve workflow, increase accuracy, and provide integrated imaging capabilities. The Perfexion model introduced in 2006 uses a larger patient aperture and internally mounted secondary collimators. Because of the larger patient aperture, it is able to treat all intracranial and even cervical spine targets quickly and efficiently with robotic positioning.

Rotating Gamma System
A radiosurgery device called the rotating gamma system (RGS) was developed in China. The rotating gamma system (OUR International Inc., Shenzen, China) employs 30 cobalt-60 radiation sources in a revolving hemispherical shell. The secondary collimator is a coaxial hemispheric shell with six groups of five different collimators to produce spherical treatment volumes of different diameter. The experience with this system is somewhat limited.
Proton Radiosurgery
The chief advantage of charged proton radiosurgery is that the beams stop at a depth related to the beam's energy. Electron beams also use charged particles but lack the sharp beam edge of the proton beam. The lack of an exit dose and the sharp beam profile of protons allow target irradiation with lower integral doses than are delivered with photon (Linac x-ray or cobalt-60 gamma) irradiation. An unmodified proton beam irradiation deposits increased energy in the last couple of millimeters of the path length. This area of increased ionization, where cell killing is even higher because of an increased radiobiologic effect, is termed the Bragg peak or Bragg-Gray peak. To allow homogeneous irradiation of targets greater than a millimeter or two, the Bragg peak is normally modulated or spread out throughout the target, essentially eliminating its effect. The first treatment of a malignant tumor by irradiation with a proton beam Bragg peak was carried out in 1957 and followed by functional neurosurgery for advanced Parkinson's disease in 1958. Presently, proton beam irradiation is available at a limited number of centers because of the high cost of equipment and maintenance. If technical improvements and increased competition lead to continued cost reductions, proton beam irradiation will become increasingly used because of its dosimetric advantages
LINAC Radiosurgery
The pioneering work of many researchers in 1980s led to the gradual modifications of linear accelerators (LINACs) designed for conventional radiotherapy to be used for radiosurgery. LINAC technologies were modified by incorporating improved guiding (stereotactic) devices and methods to measure and improve accuracy of various components. Unmodified LINACs for conventional radiotherapy tend to deviate from alignment with the isocenter at different gantry angles. Most early LINAC-based radiosurgery techniques used multiple radiation arcs with circular secondary collimators to create spherical dose distributions for stereotactically defined three-dimensional targets. Improved hardware and advanced dose-planning software have been developed to enhance conformity. These include beam shaping with micromultileaf collimators, intensity modulation with inverse treatment-planning algorithms. Many LINAC-based systems such as Xknife (Radionics Inc., Burlington, MA), Novalis (BrainLAB, Heimstetten, Germany), the Peacock System (NOMOS Corp., Sewickley, PA), and Cyberknife (Accuray Inc., Sunnyvale, CA) are commercially available. The Peacock system uses inverse planning and multileaf wedge-generated intensity-modulated beams to obtain target conformity. The Cyberknife combines a miniaturized LINAC mounted on an industrial robot with a system for target tracking and beam realignment. This system uses a 6-MeV LINAC with different-sized circular collimators attached to a six-axis robotic manipulator. Stereotactic frames are not normally used for targeting. Cyberknife plans use multiple fixed-beam positions and multiple isocenters. Before the radiation is delivered from any beam position, the target position is tracked using an integrated x-ray image processing system, consisting of two orthogonal diagnostic x-ray cameras and an optical tracking system. During treatment, the image processing system acquires x-ray images of the patient's body multiple times throughout the treatment, while stealth tracking software compares the actual images with the target images to correct alignment of the beam.
Tomotherapy
Tomotherapy, literally “slice” therapy, is a new form of radiotherapy that modifies the design of a diagnostic computerized tomography (CT) scan into a treatment-delivery machine, thereby combining the precision of CT imaging with the radiation treatment. This is done by adding a LINAC megavoltage treatment beam to the rotating x-ray source and moving table design of diagnostic CT unit, which normally uses only a kilovoltage diagnostic x-ray beam. Unlike traditional radiation therapy systems with a slow-moving external gantry designed for positioning individual beams onto the tumor from a few different directions, tomotherapy rapidly rotates the beam around the patient (and inside the housing of the unit), thus allowing the beam to enter the patient from many different angles in succession. Beam intensity modulation (IMRT) is possible through the use of a multileaf collimator system. The inclusion of CT imaging technology within the tomotherapy unit allows precise localization of the target before and during treatment.
LINAC Image-Guided Radiotherapy
The combination of diagnostic three-dimensional imaging with highly conformal treatment delivery in a single unit to maintain accuracy is the basis of various treatment techniques and processes collectively known as image-guided radiation therapy (IGRT). Although tomotherapy (which is also a type of IGRT) accomplishes this by modifying a CT unit into a megavoltage radiotherapy machine, most other IGRT techniques add CT imaging capability to a LINAC radiotherapy unit equipped for stereotactic and IMRT use. Because any patient movement between image acquisition and treatment delivery (or during treatment delivery) can introduce error, these IGRT systems use noninvasive immobilization devices and patient position tracking systems. Several manufacturers currently offer IGRT using LINAC technology capable of delivering SRS and radiotherapy, including the Trilogy (Varian Medical Systems, Inc.) and the SynergyS (Elekta, Inc.) equipped with cone-beam CT imaging capability.
P.381
Normal Tissue Tolerance in SRS and SRT
Radiosensitivity
Estimating the risks of a proposed treatment plan with various doses is an essential part of treatment planning and dose prescription for SRS and SRT. The ability of normal tissue to tolerate radiation without injury depends on the radiation dose administered, the volume of tissue irradiated, the sensitivity of the tissue affected, history of any prior radiation treatment to the region, as well as any individual variation in radiation sensitivity between different people. At present, with the exception of patients with known increased radiation sensitivity, such as those with ataxia telangiectasia, there is usually no information available to modify treatment plans for individual differences in radiosensitivity. Prior fractionated radiotherapy appears to have limited effects on the risks of developing postradiosurgery parenchymal edema and neurologic sequelae after radiosurgery, but has been observed to effect optic nerve tolerance.
Location Effects
Analysis of postradiosurgery injury reactions in AVM patients revealed no difference in the likelihood of postradiosurgery injury imaging changes (increased signal developing in surrounding brain on long relaxation time or T2-weighted images) in different regions of the brain (15,19). Dramatic differences were seen in the rates of developing neurologic sequelae between different regions of the brain (as shown in Fig. 15.2).
Radiation Therapy Oncology Group (RTOG) Dose-Escalation Studies
The RTOG Radiosurgery Dose-Escalation Study established tolerance doses for radiosurgery of recurrent brain metastases and high-grade gliomas not involving the brainstem (103). They administered radiosurgery to 156 patients with brain metastases or primary brain tumors that recurred or progressed after conventional radiotherapy following a dose-escalation protocol. Starting with initial doses of 18, 15, and 12 Gy for diameters <20, 21 to 30, and 31 to 40 mm, respectively, they escalated prescription doses in 3-Gy intervals until irreversible toxicity was seen in >20% of patients within 3 months. The exception was with tumors <20 mm in diameter where dose-limiting toxicity was not reached and investigators were reluctant to escalate above 24 Gy to 27 Gy. The recommended tolerance doses from that protocol were 24, 18, and 15 Gy for diameters of <20 mm, 21 to 30, and 31 to 40 mm, respectively. The data with longer follow-up beyond 3 months to assess late toxicity were fitted to individual logistic dose-response curves shown in Figure 15.3. These tolerance doses have been widely used as dose guidelines for radiosurgery of malignant tumors, often with interpolation for tumors close to 20 and 30 mm in diameter (e.g., 20 Gy for 18- to 22-mm diameter and 16.5 Gy for 28- to 32-mm diameter treatment volumes).
Optic Nerve Tolerance for Radiosurgery
The first analysis of optic nerve tolerance to radiosurgery—a combined Harvard and University of Pittsburgh study of patients with cavernous sinus meningiomas, craniopharyngioma, and pituitary adenomas—recommended 8 Gy as a safe maximum dose limit for the optic nerves/chiasm (110). The lowest optic chiasm dose at which optic neuropathy developed in that study was reported as being 9.7 Gy. Optic nerve/chiasm doses were estimated from isodose distributions overlaid on CT images, unlike the present day when the entire optic system is usually outlined on detailed magnetic resonance (MR) images and maximum doses are assessed from dose-volume histograms. It is highly likely that the true maximum doses to the optic system were higher and lay in portions of the nerve that were poorly visualized.
Stafford et al. (108) reported a later analysis of four cases of optic neuropathy occurring out of 215 Mayo Clinic radiosurgery patients with a median dose of 10 Gy to the optic chiasm. One case developed after an optic nerve/chiasm dose of 12.8 Gy with radiosurgery alone, at which the risk level appeared to be approximately 3%. The other cases developed in patients with prior fractionated radiotherapy (7 Gy after 58.8 Gy, 9 Gy after 45 Gy, and two procedures delivering 9 and later 12 Gy to the optic system after 50.4 Gy).
Leber et al. (56)_ analyzed optic nerve injury risks in 50 patients with 24- to 60-month follow-up (median, 40 months) who underwent gamma knife radiosurgery for benign skull base tumors. Their risks of optic neuropathy were 0% with <10 Gy, 27% with 10 to 15 Gy, and 78% with >15 Gy. They found no cavernous sinus nerve injury with doses of 5 to 30 Gy.

P.382
Considering these data and published risks of optic neuropathy for conventional fractionated radiotherapy, α/β ratios in the range of 0 to 1 seem reasonable for estimating fractionated radiotherapy dose-equivalents for radiosurgery doses for the optic nerve and probably other cranial nerves
Tolerance of Other Cranial Nerves
From clinical experience with fractionated conventional radiosurgery and radiosurgery, it appears that special sensory nerves (optic and auditory) are the most radiosensitive, followed by somatic sensory nerves (trigeminal) and finally the motor nerves (cranial nerves 2, 4, 6, 7, and 9 through 12). After acoustic schwannoma radiosurgery to doses of 12 to 13 Gy, FSR to 18 Gy in three fractions or SRT to 45 to 50 Gy in 25 to 28 fractions, decreased hearing develops in 30% to 50% of patients, facial numbness in 2% to 3%, and facial weakness in ≤0.5%. Radiosurgery with present techniques for meningiomas involving the cavernous sinus is associated with a risk of trigeminal neuropathy of approximately 3% of patients, with radiation injuries to cranial nerves III, IV, or VI more uncommon (11,12,13,14,15)
Spinal Cord Tolerance
The tolerance of the spinal cord to SRS or SRT depends on the volume of spinal cord irradiated, the distribution of that radiation (e.g., maximum dose), the dose of radiation previously administered to the spinal cord, and the time interval between initial radiation and retreatment. Spinal cord tolerance to SRS or SRT has not been well defined because of a fortunate paucity of radiation injury reactions in clinical experience so far. Gerszten et al. (28) reported on 125 patients who underwent Cyberknife spine radiosurgery (17 benign and 108 metastatic cases). Seventy-eight lesions had previously received external-beam radiotherapy. Treatment volumes varied from 0.3 to 232 mL, with a mean of 27.8 mL. Prescription doses varied between 12 and 20 Gy (mean, 14 Gy) to an 80% isodose treatment volume. Spinal cord volumes receiving >8 Gy varied from 0.0 to 1.7 mL, with a mean, 0.2 mL. They identified no acute radiation toxicity or new neurologic deficits after a median follow-up of 18 months (range, 9 to 30 months). Benzil et al. (6) reported the New York Medical College experience with spine radiosurgery in 31 patients using a Novalis LINAC unit. Two patients who received biologic equivalent doses of >60 Gy developed radiculitis.
Clinical Uses of SRS and SRT
Table 15.2 lists the most commonly used indications for radiosurgery, with representative references. Except for functional radiosurgery, there are varied levels of experience with SRT for each of these indications.
Functional Radiosurgery
The most widely used functional application for radiosurgery is in the management of typical trigeminal neuralgia refractory to medical therapy (18,19,20,21,22). Atypical or constant pain does not respond. Other alternatives for managing typical trigeminal neuralgia are medication, open surgery with microvascular decompression, and rhizotomy procedures using glycerol injection, balloon compression, or radiofrequency injury to the nerve. Typically, 4-mm collimators are used for radiosurgery to a maximum dose of 80 Gy. Response rates reach approximately 85%, typically 1 week to 4 months after the procedure, but can develop as late as 6 months later. Approximately 50% of typical trigeminal neuralgia patients remain pain-free and off medication 5 years following radiosurgery. A typical radiosurgery plan for trigeminal neuralgia is shown in Figure 15.4.
A small destructive lesion in the ventralis intermedius nucleus of thalamus can be created either invasively with a needle equipped with a radiofrequency generator or noninvasively with radiosurgery using 4-mm diameter collimators and a maximum dose of 130 Gy to alleviate medically refractory unilateral tremor in patients with essential tremor and/or Parkinson's disease (23,24,25,26). Nondestructive management through insertion of a deep brain stimulator is preferred in most patients, but not all patients are acceptable candidates. Limited experiences with radiosurgery of the globus pallidus (pallidotomy) to attempt to alleviate more generalized parkinsonian symptoms have not been as favorable (48,49,50). There is favorable limited experience with bilateral radiosurgical capsulotomy for managing severe, refractory obsessive-compulsive disorder (60). Radiosurgery to hypothalamic hamartomas may help control refractory gelastic seizures (94,112). The use of radiosurgery as an alternative to extensive surgery in medically refractory mesial temporal lobe epilepsy shows promise and continues to be investigated (97).
P.383
Vascular Malformations
Untreated intracranial AVMs have a bleeding risk of approximately 3% per year, or higher if prior bleeds have occurred (83,86). This results in an average of 1% of untreated AVM patients dying each year from hemorrhage. Management options include observation, surgical resection, embolization, and radiosurgery. Radiosurgery can dramatically reduce the risk of hemorrhage. Radiosurgery obliterates the AVM nidus in approximately 75% of patients within 3 years of the procedure (21,39,69). Individual obliteration rates vary from 50% to 88% depending on marginal dose administered, as shown in the dose-response curve illustrated in Figure 15.5. Although AVM obliteration rates appear to be optimized with marginal doses of approximately 23 Gy, lower doses are selected for most patients to minimize complications. The risk of neurologic sequelae from radiosurgery averages approximately 3% but varies with treatment volume, dose, and location (Fig. 15.2). The risk of hemorrhage while waiting for complete obliteration to develop seems unaltered (86). All of these risks and benefits of radiosurgery need to be considered together to optimize management of individual AVM patients.

When an AVM nidus fails to completely obliterate by 3 years after radiosurgery, irradiation can be repeated with acceptable morbidity (87). Although some residual radiation injury effect would be expected within the previously irradiated, unobliterated AVM nidus vasculature, retreatment appears to require similar, if not higher, doses to achieve similar rates of complete obliteration as initial radiosurgery (87).
Management of large AVMs is presently difficult because radiosurgery may be associated with high complication risks and low obliteration rates. Recent improvements in embolization with liquid glue or (ev3, Inc., Plymouth, MN) polymer can sometimes help reduce the target volume, but adds to the total risks of the overall management (65). Another promising approach is staged radiosurgery, in which large AVMs are treated in two or three sections separated by 4- to 6-month intervals to reduce acute toxicity. Whether there is any benefit to fractionating stereotactic irradiation of AVMs is presently unclear (106).
Cavernous malformations do not show detectable flow on angiography but nevertheless are vascular lesions with annual hemorrhage risks of 0.5% per year with no prior bleed, 4.5% with one prior hemorrhage, and approximately 32% per year after a history of two or more hemorrhages (61,62,63,64,65). Lower pressures in these lesions lead to smaller bleeds than are typically seen with AVM. Repeated bleeds from brainstem cavernous malformations can cause considerable neurologic morbidity. Symptomatic, surgically accessible lesions should be resected. Radiosurgery of brainstem cavernous malformations with a history of two or more prior hemorrhages appears to reduce the
P.384
risk of subsequent bleeds to approximately 1% per year with acceptable morbidity (63,66,67,68,69).

Benign Tumors
Most small benign intracranial tumors are well managed with radiosurgery, FSR, or SRT. Radiosurgery control rates are high with radiosurgery, with prescription doses on the order of 12–14 Gy (11,12,13,33,34,35,36,37). Kondziolka et al. (51) evaluated long-term tumor control in 285 consecutive patients who underwent radiosurgery for benign intracranial tumors between 1987 and 1992, with a median follow-up period of 10 years. This included 157 patients with vestibular schwannomas, 10 with other cranial nerve schwannomas, 85 with meningiomas, 28 with pituitary adenomas, and 5 with craniopharyngiomas. Forty-four percent of the patients had prior surgical resection and 5% had prior fractionated radiotherapy. They found that 95% of the 285 patients had imaging-defined local tumor control (63% had tumor regression and 32% had no further tumor growth). The crude tumor control was 95% (271/285 patients) with a 15-year actuarial tumor control rate of 93.7%. In 5% of the patients, delayed tumor growth was identified. Resection was performed after radiosurgery in 13 patients (5%) for tumor growth.
Vestibular Schwannomas
Vestibular schwannomas, also known as acoustic neuromas, are benign tumors arising from Schwann cells. They are associated with loss of genetic information on chromosome 22 (78). Vestibular schwannomas either occur on one side as spontaneous mutations or bilaterally as the hallmark of type 2 neurofibromatosis (NF-2). Vestibular schwannomas usually arise within the internal auditory canal and later extend intracranially into the cerebellar pontine angle. Because these tumors lack the ability to invade bone, the portion of tumor outside the canal in the cerebellar pontine angle eventually grows into a globular extension that is larger than the intracanalicular portion (Fig. 15.1). The differential diagnosis of a cerebellar-pontine angle tumor includes vestibular schwannoma (90%), meningioma (close to 10%), cholesteatomas, facial or trigeminal schwannoma, and rare primary or metastatic malignant tumors. Cerebellar pontine angle meningiomas (which can be managed similarly to vestibular schwannomas) also may involve the internal auditory canal, but are usually distinguished by a broad, flat, dural attachment that is lacking in vestibular schwannomas.
Observation or surgical resection were essentially the only management strategies offered to acoustic schwannoma patients until favorable experiences with gamma knife radiosurgery were reported in the 1980s. Observation may be appropriate in selected NF-2 patients and some elderly patients with small, minimally symptomatic vestibular schwannomas, but early intervention appears to be the best strategy for long-term hearing preservation for most patients (99,105). Surgery appears to be the best initial strategy in patients with vestibular schwannomas large enough to cause symptomatic brainstem compression with obstructive hydrocephalus. For small-to-medium vestibular schwannomas, tumor control rates with radiosurgery or SRT are comparable to those of surgical resection (71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94).
Early radiosurgery series including patients treated during the 1980s with higher doses (14 to 18 Gy) and less conformal treatment plans had higher rates of postradiosurgery cranial neuropathies (15% to 20% trigeminal and/or facial and 67% with a drop in their Gardner-Robertson hearing level) (49). This led some groups to pursue SRT for vestibular schwannomas, while others pursued radiosurgery with refined techniques and lower doses. Both approaches led to improved results (71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94). The University of Pittsburgh reported on 313 previously untreated unilateral acoustic schwannoma patients who underwent gamma knife radiosurgery doses of 12 to 13 Gy between February 1991 and February 2001 (18). Median follow-up was 24 months, maximum follow-up was 115 months, and 36 patients had >60 months of follow-up. The actuarial clinical tumor control rate, free of surgical intervention, was 98.6% at 7 years. One patient's growing tumor was subsequently completely resected. The only other failure required a partial resection because of an enlarging adjacent subarachnoid cyst, despite control of the irradiated tumor. The 7-year actuarial rates for unchanged facial strength, unchanged facial sensation, unchanged hearing level, and useful hearing preservation were 100%, 95.6%, 70.3%, and 78.6%, respectively. Eight patients developed new trigeminal neuropathy, six of whom developed numbness (7-year actuarial rate = 2.5%), and the other two developed new typical trigeminal neuralgia (7-year actuarial rate = 1.9%). The risk of developing postradiosurgery trigeminal
P.385
neuropathy was associated with increasing tumor volume (p = .038). Similar results with low-dose radiosurgery of vestibular schwannomas were reported by Iwai et al. (38), Paek et al. (84), Muacevic et al. (77), and Rowe et al. (98).
Various fractionation schemes (18 Gy/3, 20 Gy/4 to 5, 25 Gy/5, 45 to 50 Gy/25, and 54 Gy/30) have been used with vestibular schwannoma with minor differences in results. After accounting for length and quality of follow-up, treatment results seem similar to those of radiosurgery with 12 to 13 Gy, but an advantage for fractionation cannot be excluded entirely (14). Combs et al. (12) from Heidelberg reported seemingly better useful hearing preservation (94%) in 106 acoustic schwannoma patients managed with FSRT to a median dose of 57.6 Gy with 1.8 Gy fractions. They reported 98% actuarial hearing preservation for non-NF2 patients compared to 68% for NF-2 patients. These numbers were based on telephone questioning rather than audiograms, making comparison to radiosurgery series difficult. Their 5-year actuarial tumor control rate was 93%, and postradiation trigeminal and facial neuropathy rates were 3.4% and 2.3%, respectively. The University of California, Los Angeles, also reported unusually good hearing preservation (93%) for their experience with 50 unilateral acoustic schwannoma patients irradiated to 54 Gy in 30 fractions to a 90% isodose treatment volume including a 1- to 3-mm margin around gross tumor (59). All tumors were controlled with a median follow-up of 36 months (range, 6 to 74 months). They defined useful hearing preservation as the ability to talk on the telephone and listen with the affected ear. New facial numbness developed in one patient (2%) and facial weakness also developed in one patient (2%) after radiotherapy.
Andrews et al. (3) analyzed the Jefferson vestibular schwannoma experience, comparing 69 radiosurgery patients with 50 SRT patients who received 50 Gy in 25 fractions. Their first 25 vestibular schwannoma patients were treated using a linear accelerator, and later radiosurgery patients were treated with gamma knife radiosurgery. They authors had similar facial and trigeminal neuropathy rates for the radiosurgery and fractionated radiotherapy groups but the rate of hearing loss was significantly higher in their radiosurgery group. There were only a small number of patients with serviceable (useful) hearing in each group prior to irradiation (12 in the radiosurgery and 21 in the FSRT groups) and follow-up was limited. Meijer et al. (73), from Amsterdam, also reported a single-institution comparison of radiosurgery (LINAC to 10 or 12.5 Gy) and SRT (20 Gy/4 to 5 fractions) for vestibular schwannoma. They selected 49 edentulous patients (mean age, 63 years) for radiosurgery and 80 patients (mean age, 43 years) with intact dentition for SRT. They found a higher rate of trigeminal neuropathy following radiosurgery (8%) than SRT (2%) at 5 years (p = .048), but similar hearing loss with radiosurgery (25%) than SRT (39% FSRT; p >.05), similar, but higher than usual rates of new facial neuropathy (7% radiosurgery vs. 3% FSRT; p >.05), and similar 5-year actuarial tumor control rates (100% with radiosurgery vs. 94% with SRT).
Nonacoustic Schwannomas
Schwannomas may occasionally involve other cranial nerves, particularly V, VII, and XI through XII in the jugular foramen. Tumor control rates are similar, but postradiosurgery neuropathies seem less common as somatic sensory and particularly motor nerves seem less sensitive to radiation injury than special sensory nerves like VIII (95,96,97,98).
Meningiomas
Radiosurgery and SRS are both excellent management options for most small benign meningiomas, with in-field tumor control rates well above 90%, as has been seen with most other benign tumors (17,51). Marginal recurrences rates as high as 25% may develop because of the tight margins used for radiosurgery or SRT treatment volumes limited to small recurrences or residual tumor after resection of large parasagittal meningiomas (11,33,99,100,101). Marginal recurrences are far less of a problem with unresected (and usually unbiopsied) meningiomas. A University of Pittsburgh study (17) analyzed 219 imaging-diagnosed meningiomas (unbiopised) managed with gamma knife radiosurgery to a prescription dose of 8.9 to 20 Gy (median, 14 Gy), treatment volumes of 0.47 to 56.5 mL (median, 5.0 mL) with 2 to 164 months of follow-up (median, 29 months). Tumors progressed in seven patients; two of the tumors proved to be different ones (metastatic nasopharyngeal adenoid cystic carcinoma and chondrosarcoma). Another patient with local control of the lesion developed a subsequent brain metastasis, changing the diagnosis of the first lesion to the same. The actuarial tumor control rate was 93.2% at both 5 and 10 years. The actuarial rate of identifying a diagnosis other than meningioma was at both 5 and 10 years. No pretreatment variables, including dose, correlated with tumor control in univariate or multivariate analysis. The actuarial rate for developing any postradiosurgical injury reaction was 8.8% at 5 and 10 years. The risk of postradiosurgery sequelae was lower (5.3%) after 1991 (with stereotactic MR imaging and lower doses; p = .0104).
Atypical and malignant (anaplastic) meningiomas have higher rates of local and marginal recurrence after therapeutic intervention. Complete surgical resection is advocated whenever possible, followed by a full course of conventional radiotherapy with at least 1-cm margins around the tumor volume. Radiosurgery has been recommended to improve local control of unresectable tumor (101,102,103,104). Malik et al. (67) reported 5-year actuarial control rates of 87% for typical meningiomas, 49% for atypical, and 0% for malignant meningioma in the Sheffield gamma knife experience. Harris et al. (30) reported on the Pittsburgh gamma knife experience in 12 malignant and 18 atypical meningiomas. Their 5-year local tumor control was 72% for malignant and 83% for atypical meningiomas; however, 10-year actuarial survival rates were only 59% and 0%, respectively. Katz et al. (40) could not substantiate that either accelerated fractionated radiotherapy or a radiosurgery boost improved tumor control or survival in their analysis of 27 atypical and 9 malignant meningioma patients managed at the University of Florida.
Pituitary Adenoma
Management of pituitary adenomas requires a multidisciplinary approach to properly select which patients are suitable for different approaches with medical therapy, surgery, fractionated radiotherapy, and radiosurgery, or combinations of these. Most patient with visual compromise, particularly with a hemianopsia or greater, will do better with initial surgical decompression. Prolactinomas are usually initially managed with medical therapy (92). Most other small pituitary adenomas, where the target volume can be separated from the optic nerves, are reasonable candidates for radiosurgery (11,34,35,36).
Sheehan et al. (104) performed a review of 35 peer-reviewed reports of radiosurgery for pituitary adenoma that included 1,621 patients. Most studies reported >90% control of tumor size (range, 68% to 100%). The weighted average tumor control rate for all published series (encompassing 1,283 patients) was 96%. In eight published series with mean or median patient follow-up periods of ≥4 years, tumor growth control rates varied from 83% to 100%.
Twenty-two series have published radiosurgery results for 314 Cushing's disease patients. The mean radiosurgical prescription (margin) doses for these series varied from 15 to 32 Gy. In those series with at least 10 patients and a median follow-up
P.386
of 2 years, endocrinologic remission rates range from 17% to 83%. Many of the patients in older series were treated in the pre-CT and MR imaging era of radiosurgery, sometimes as often as many as four times before their Cushing's disease went into remission.
Malignant Tumors
Brain Metastases
Brain metastases are the most common and best studied of the indications for radiosurgery (71). Early clinical investigations found impressive tumor control with radiosurgery for brain metastases that progressed after prior whole-brain radiotherapy (WBXRT). The RTOG phase I dose-escalation trial in recurrent brain tumors to some degree standardized dose prescription for brain metastasis radiosurgery (21). Because of the success of radiosurgery in controlling brain metastases after whole-brain radiotherapy and the high rate of eventual local tumor progression in brain metastases after conventional WBXRT, radiosurgery has been increasingly used in initial management of brain metastases (21). The subsequent phase III randomized trial, RTOG 95-08, established that radiosurgery immediately following standard WBXRT (37.5 Gy in 15 fractions) improves local control and quality of life for patients with one to three brain metastases while also improving overall survival for patients with solitary metastasis, all compared with patients initially managed with WBXRT only (2).
Although RTOG 95-08 established the role of radiosurgery after WBXRT in managing one to three brain metastases, questions remained about managing brain metastases with radiosurgery alone, preserving full-dose WBXRT as an option for later managing cases with subsequent progression. Aoyama et al. (4) recently published the outcome of a prospective randomized controlled trial to evaluate whether initial WBXRT provides better outcomes when added to SRS compared to using SRS alone. The 11-hospital study done by Aoyama et al. (4) randomized 132 patients with one to four brain metastases <3 cm in diameter to radiosurgery either with or without initial WBXRT. They found that the median survival time and the 1-year actuarial survival rates were not significantly different with or without WBXRT. The 1-year brain tumor “recurrence rate” (corresponding to the development of additional brain metastases) was higher in the SRS-alone group compared with the patient group that received both WBRT and SRS. Earlier retrospective studies had similar observations. The most common primary site in these studies was lung. Separate analyses of radiosurgery of brain metastases of different histologies with and without WBXRT found that WBXRT reduced subsequent development of brain metastases in lung cancer patients but not patients with melanoma or renal cell carcinoma. Administering initial WBXRT and waiting a month before radiosurgery for subsequent tumor shrinkage is a reasonable strategy for limiting radiation injury reactions and/or to improve tumor control for brain metastases >3 cm in diameter and for brainstem metastases >2 cm in diameter.
There is no clear limit as to how many metastases and what total volume of metastases can or should be treated by radiosurgery. RTOG 9508 was limited to one to three metastases, while the trial of Aoyama et al. (4) and a smaller University of Pittsburgh trial included patients with up to four brain metastases (52). Bhatnagar et al. (7) analyzed 205 patients who underwent radiosurgery for 4 to 18 brain metastases (median, 5). They reported a median survival of 8 months after radiosurgery and found that survival correlated with the total volume of metastases, age, and RTOG-RPA class, but not the total number of brain metastases. Presently, many centers use WBXRT alone to initially manage patients with five or more metastases, and subsequently consider radiosurgery for patients who are unable to be withdrawn from steroid medication and for patients whose brain metastases progress after WBXRT.
Glioblastomas
During the late 1980s and 1990s, many centers that had been using brachytherapy for recurrent high-grade gliomas and as boosts after conventional radiotherapy switched to radiosurgery (72,107). Although retrospective series appeared to show that initial brachytherapy or radiosurgery boosts after conventional radiotherapy improved survival of glioblastoma patients, prospective randomized trials of both modalities used prior to conventional radiotherapy of glioblastoma patients were negative (72,107). Radiosurgery appears to be a reasonable option for small, well-circumscribed, high-grade gliomas that recur after prior conventional large-field radiotherapy and chemotherapy.
Radiosurgery of Spinal Metastases
Radiosurgery has been used to treat spinal tumors, mostly metastases, either as initial treatment or for recurrence after prior fractionated radiotherapy (6,26,27,28). By limiting spinal cord radiation dose with radiosurgery techniques, higher doses can be safely given to the tumor target volume with the hope of achieving greater local tumor control and higher response rates. Spinal cord tolerance in the experience of Gerzsten et al. (28) with 125 Cyberknife spine radiosurgery procedures in 17 benign and 108 metastatic cases was previously discussed in this chapter. Gerzsten et al. (26) separately reported results for single-fraction Cyberknife radiosurgery of 68 breast carcinoma metastases to the spine in 50 patients after 6 to 48 months of follow-up (median, 16 months). Pain was the most common indication for radiosurgery (in 57 lesions). Radiosurgery was delivered for radiographic tumor progression, as a postsurgical boost and for a progressive neurologic deficit in one case each. Radiosurgery was used as primary management in eight patients. Target volumes varied from 0.8 to 197 mL (mean, 27.7 mL). Maximum tumor doses were 15 to 22.5 Gy (mean, 19 Gy). No radiation-induced toxicity occurred during the follow-up period (6 to 48 months). Cyberknife radiosurgery achieved long-term pain improvement in 55 of the 57 patients (96%) who were treated primarily for pain. Long-term radiographic tumor control was seen in all patients who underwent primary radiosurgery as well as those treated for radiographic tumor progression after radiotherapy or as a postsurgical treatment. Similar results were seen in separate reports for spine radiosurgery of melanoma metastases (27). Randomized trials are needed to prove that SRS or hypofractionated SRT improve results compared with conventionally fractionated radiotherapy or IMRT with conventional immobilization (8).
Stereotactic Irradiation of Lung Tumors
Hypofractionated SRT has been used for treatment of small, medically inoperable non–small lung cancer primary tumors and to manage lung metastases in patients with limited metastatic disease (112,113,114). Beitler et al. (5) reported the Staten Island experience with SRT using five fractions of 8 Gy each in 75 patients (67 SRS alone and 8 with SRS boost after conventional radiotherapy) with 1 to 92 months of follow-up (median, 17 months). Treatment volumes varied from 0.26 to 1197 mL, with a median of 26.8 mL. Complete responses developed in 7 patients, tumor shrinkage in 25, stable disease in 22, and tumor progression in 9; 12 lacked follow-up. Radiation pneumonitis was reported in two patients. Patients with tumors <65 mL had a median survival of 26 months compared with 10 months for those with tumors >65 mL.
P.387
Ernst-Stecken et al. (13) reported results with hypofractionated stereotactic irradiation of 39 primary or secondary lung tumors in 21 patients. They delivered five fractions of either 7 Gy (n = 21) or 8 Gy (n = 18) to median tumor volumes of 2.9 mL (0.15 to 67.9 mL) using planning target volumes of 7.2 to 124.0 mL (median, 25.8 mL). They reported completed remission in 51%, partial in 33%, no change in 3%, and progressive disease in 13%. Most patients experienced grade 1 toxicity; none developed grade 2 or 4, but one patient developed grade 3 dyspnea 6 months after SRT.
Schefter et al. (100) reported a phase 1-2 dose escalation study of stereotactic body radiotherapy for lung metastases in 25 patients. Using 5-mm radial and 10-mm cranial-caudal margins, they administered three fractions of 16, 18, or 20 Gy while restricting the percentage of normal lung receiving >15 Gy to under 35%. Fourteen patients were in the phase 1 part of the study: six at 48 Gy, four at 54 Gy, and four at 60 Gy. Afterward, 14 patients were enrolled in the phase 2 part and received 60 Gy in three fractions, but follow-up was insufficient to report toxicity. Dose-limiting toxicity, defined as grade >3 lung, esophageal, or spinal toxicity occurring in any single patient, never developed in this study. Grade 1-2 esophagitis developed in three-fourths of the patients in the lowest dose group only. Grade 1 dermatitis developed in one fourth of the patients receiving 18 Gy × 3 and in one fourth of the patients with 20 Gy × 3. Grade 1 pain developed in one fourth of the patients within each dose level.
Miscellaneous Uses
Radiosurgery and hypofractionated SRT have been used as a substitute for brachytherapy in the management of recurrent head and neck tumors (50,113). Liver metastases can also be irradiated with these techniques (101). Hypofractionated prostate SRT is also being explored (43,64).