Electromagnetic Radiation

EM radiation is composed of oscillating electric and magnetic fields. The EM spectrum spans a range from infrared waves, through visible light, to high-energy x-rays and gamma rays. As the wavelength of the waves decreases, the energy of the waves increases. In radiation therapy, it is the high-energy x-rays and gamma rays that are used for clinical effect. Because of the small wavelengths and high energy of x-rays and gamma rays, EM radiation exhibits a dual nature: it can be described as waves, and it can also be described in terms of small, discrete packets of energy called photons. High-energy photons are considered to be an indirectly ionizing form of radiation. When interacting with tissue, photons induce the liberation of charged particles (electrons), which then cause the majority of the ionization and, thus, the biologic effect.

X-rays and gamma rays differ only in their manner of production. X-rays are produced either as a result of the interaction between a high-speed electron and a nucleus (bremsstrahlung x-rays) or as a result of electrons in the outer shell of an ionized atom falling from a high- to a low-energy level to fill a vacancy created by an electron that has been ejected (characteristic x-rays). X-rays may be a by-product of radioactive decay or may be created by human intervention. For example, linear accelerators (LINACs) generate x-rays by accelerating electrons and directing them to strike a target composed of a substance with high atomic number. The electrons interact with the target nuclei and generate (primarily) bremsstrahlung and (secondarily) characteristic x-rays.²

In contrast, gamma rays are photons emitted from the recoiling nucleus of a radioactive atom when it decays. An example is cobalt 60 (⁶⁰Co), which as it undergoes beta decay converts a neutron to a proton and in the process emits a beta particle (an electron), an antineutrino, and gamma photons.

High-energy photons (>1 MV) are useful in radiation therapy because they deposit a significant amount of energy at depth in tissue, so they can be used to treat tumors deep within the body. In addition, high-energy photons exhibit a property called the build-up region when they enter tissue because the electrons liberated by the interacting photons near the skin surface are propelled in a mostly forward direction and deposit their energy deeper in tissue. This gives photons an advantage known as the “skin-sparing” effect (Fig. 249-1).

FIGURE 249-1 Comparison of dose deposition for photon beams, electron beams, and proton beams. Notice that photon beams have a “buildup region” that provides a measure of skin sparing at the surface of the patient. Electron beams exhibit less skin sparing and a sharper dose falloff. Proton beams deposit most of the dose at the end of their range, a phenomenon known as a “Bragg peak.” Photons also have a maximum dose range in tissue.

Particle Radiation

Particle radiation differs from photon radiation in that the energy of the radiation is propagated through the kinetic energy of the particle itself. High-energy particle radiation is directly ionizing; it has sufficient kinetic energy to ionize atoms as they interact in tissue. Unlike high-energy photons, which tend to sparsely interact with matter and can travel long distances before being completely attenuated, high-energy particles tend to have shorter, bounded ranges of penetration in tissue before they are completely attenuated. The particles routinely used for therapeutic purposes are electrons and protons, with heavy ions used less frequently. Neutrons, which have also been evaluated for use in radiation therapy, are uncharged particles with very different modes of interaction in tissue and are not described here.

High-energy electrons are usually produced in LINACs by replacing the high–atomic number target (usually tungsten), which results in x-ray production, with a foil that serves to scatter the electrons in a desired pattern. Electrons begin depositing appreciable dose near the surface of tissue, have a predictable range at which they deposit the majority of their energy, and exhibit rapid dose falloff. This gives electron therapy a particular advantage in the treatment of cutaneous or subcutaneous lesions (Fig. 249-1).

High-energy protons are produced in particle accelerators such as cyclotrons. Protons are much more massive particles than electrons. Therefore, at a particular velocity, protons have much greater kinetic energy and do not scatter as easily. Hence, protons can potentially cause less damage to surrounding tissue. In addition, most of the energy absorption from protons occurs at the distal end (over the last few millimeters) of the particle track. The precisely defined area of intense ionization at the end of the track after the passage of protons is called a Bragg peak (Fig. 249-1). After the Bragg peak the deposited energy falls off quickly, so protons have a defined range in tissue with essentially no exit dose. To treat the entire thickness of a tumor, the proton beam may be altered to spread the Bragg peak out to the desired range of depth. By taking advantage of the Bragg peak effect, as well as cross-firing of a number of proton beams, a well-localized volume of high radiation delivery can be produced and has been applied in a radiosurgical setting.^3,4

Free Radicals

When cells are irradiated with ionizing radiation, photons can interact with water molecules by stripping an electron from a hydrogen atom, which results in a fast electron and an ionized water molecule. The resulting fast electrons further interact with water molecules through additional ionizing events. The positively charged water molecule exhibits a short half-life before dissociating into an H⁺ ion and an OH⁻ free hydroxyl radical. The hydroxyl radical is reactive and has sufficient energy to break chemical bonds in nearby molecules. This indirect effect of radiation through a free radical intermediary is responsible for the majority of radiation-induced damage. The chemical reactions are enhanced by the presence of oxygen. Tissue hypoxia (PO₂ <30 mm Hg) will inhibit the production of free radicals and thereby lessen the damaging effects of radiation.

Radiation Injury to DNA

A variety of evidence suggests that DNA is the most important target for cellular damage by ionizing radiation. The reactive water derivative may interact with DNA and create the potential for permanent cell injury or death. Radiation can also interact directly with DNA.

Damage to DNA can take the form of single-strand breaks, where one of the two intertwined helixes are broken, or double-strand breaks, in which both are broken. Single-strand breaks (the more common form of damage) are potentially repaired by DNA repair enzymes because the undamaged strand can be used as a template for repair. Double-strand breaks are more difficult if not impossible to repair; they cause the most serious biologic damage. Double-strand breaks may be the result of a single particle or the interaction of two single-strand breaks caused by separate particles occurring at close temporal and spatial distances.

Studies on synchronously dividing cell cultures have demonstrated that cells in the G₂ and M phases of the cell cycle are most susceptible to double-strand DNA breaks.

Radiobiology of Conventional Radiotherapy

Radiation damages the DNA of tumor cells, as well as the DNA of normal cells in its path. Normal tissue, however, is generally more capable of DNA repair than tumors are, partly because of aberrant cell cycle control mechanism in tumors, as well as differences in genetic features that permit damage to the abnormal tumor phenotype. Abnormal metabolic patterns may also make tumors more susceptible than normal cells to increased oxidative stress.

Cells require time to repair DNA damage, and the normal cell response to irradiation is to delay the cell cycle. The length of the G₂ phase delay correlates with radiation resistance. Therefore, the radiobiology of differential cell repair is of paramount importance for conventional radiotherapy. Repair plays a less critical role as the number of fractions decreases.

The probability of cell survival after single doses of radiation is a function of the absorbed dose, measured in the unit gray (Gy). Typical mammalian cell survival curves obtained after single-dose irradiation in culture have a characteristic shape that includes a low-dose shoulder region followed by a steeply sloped region at higher doses.^5,6 The shoulder region is interpreted as an accumulation of sublethal damage at low doses, with lethality resulting from the interaction of two or more such sublethal events. As noted previously, single-strand breaks in DNA may be repaired and therefore represent sublethal damage to the cell. However, double-strand breaks may result in cellular changes, including cell death. Such a model can be described by the following probabilistic equation in which probability (cure or complication) = exp(−K*exp[−αD − βD²])] (“exp” represents exponential, K equals the number of clonogens, “α“ and “β“ are constants related to single-event cell killing and cell killing through the interaction of sublethal events, respectively, and “D” represents dose). The α/β ratio is the single dose at which overall cell killing is equally attributable to both components of cell killing (αD = βD² or D = α/β).⁷ The validity of the linear quadratic formula for single-dose radiosurgery has been questioned, however.⁸ Nevertheless, it still provides a meaningful method to relate radiosurgery to fractionated radiation schemes.

The α/β ratio varies depending on the tumor and normal tissue type (Fig. 249-2). Late-responding tissues such as the brain or spinal cord have an α/β ratio of approximately 2, whereas many tumors have an α/β ratio of nearly 10. The α/β ratio for skin or mucosa is approximately 5 to 8. Tumors with a low α/β ratio (i.e., a small α or single-hit component for radiation kinetics) will have less of a desired effect when a scheme involving a low radiation dose per fraction is used than when comparable tissues with a high α/β ratio are treated. The dose may be normalized to a scheme of 2 Gy per fraction (NTD2Gy) by using the following equation⁹:

FIGURE 249-2 Comparison of single-dose effect curves (A) and fractionated-dose effect curves (B) for low- and high-α/β tissue. The small advantage seen in the low-dose region in which low-α/β tissue is spared is amplified through dose fractionation.

Conventional fractionated radiation therapy relies on the four R’s of radiobiology: repair of nonlethal injury, reoxygenation of hypoxic tumor cells, repopulation of tumor cells, and reassortment of tumor cells into more susceptible phases of the cell cycle. There are advantages and disadvantages to fractionated radiation therapy and radiosurgery. Depending on the clinical scenario, one may prove superior to the other. Certainly, there seems to be little advantage to fractionation for functional cases (e.g., trigeminal neuralgia) or for the treatment of patients with arteriovenous malformations (AVMs).⁸

As mentioned earlier, hypoxia at a PaO₂ below 30 mm Hg reduces the development of damaging free radicals and thus the degree of radiation damage. Experimental studies and clinical observation have suggested that aerated cells become nonviable after irradiation and that the site of irradiation is dominated by hypoxic cells. From this observation it is thus important to note that substantial dose escalation of a tumor may prove to have diminishing returns because hypoxic cells are not adequately depopulated. However, evidence has demonstrated a phenomenon known as reoxygenation, whereby tumors may reestablish their oxygenated state between sessions if the radiation is delivered in fractions. Through reoxygenation, a higher percentage of tumor cells will be depopulated by fractionated irradiation. A variety of factors are involved in the process of reoxygenation, such as decreased oxygen consumption by dead cells and a reduction in the number of cells in relation to capillary blood supply.

Malignant tumors generally fall into the category of early-responding tissue containing hypoxic cells, whereas normal brain tissue consists primarily of late-responding tissue containing well-aerated cells. In malignant tumors, arguments can be made for and against fractionation. Fractionation increases the cellular depopulation of a tumor for a given total radiation dose because of the phenomenon of reoxygenation. At the same time, fractionation reduces the damage to critical late-responding normal tissue. However, fractionation allows malignant tumor cells to repopulate between fractions. This phenomenon is in contrast to the treatment of many benign tumors and AVMs, in which both targeted abnormal tissue and normal brain tissue consist of late-responding tissue of similar radiologic type. There is little to be gained by fractionation in these situations.

The standard approach to radiotherapy includes daily treatments preceded by a dose delivery simulation in which patient positioning relative to the treatment machine is confirmed to result in appropriate beam entrance and exit sites. The most commonly prescribed absorbed dose is 1.8 to 2 Gy, which has proved to be well tolerated in most areas of the body and can be repeated a specific number of times, depending on the region involved and the therapeutic target. For practical purposes, tolerance of the whole brain is considered to be 45 to 50 Gy in 20 to 25 fractions, although it is recognized that this dose may yield substantial dementia and memory loss with time.

Radiobiology of Radiosurgery

A therapeutic advantage may also be achieved by depositing more radiation dose in the tumor than in surrounding normal tissue. The use of cross-firing beams enables delivery of high doses to the region of the target while minimizing the dose to surrounding tissue. Although a single radiation beam entering a patient begins with a region of high dose (after the initial build-up region) and gradually decreases in dose with increased depth, with cross-firing beams the dose at depth progressively increases as the various beams intersect. Thus, as the dose is deposited along each beam, the total dose absorbed by normal tissue can be kept low while achieving a much higher dose at the intersection of the beams and a steep dose gradient between the low- and high-dose regions. This rapid falloff in radiation dose is the basic principle used to spare normal tissue in radiosurgery. Although cross-firing can also be used in conventional radiotherapy, its practical limits are two to four fields, and accuracy of delivery is on the order of 1 cm or greater in many anatomic sites. However in radiosurgery, because hundreds of beams are added together, the isodose lines begin to take on the configuration of the tumor, thus minimizing dose outside the target. This ability led Lars Leksell to propose the concept of radiosurgery in 1951. He described the use of radiation as a means of replacing the scalpel or electrode for functional neurosurgery. In so doing, the biology of differential repair was discarded, and the main biologic advantage became the ability to destroy focally identified areas and avoid normal brain tissue by physical means.

The initial radiosurgical concept of Leksell was intended for the treatment of functional neurological disorders, but it has now expanded to become a standard treatment option for numerous benign and malignant central nervous system pathologies. In radiosurgery, the surgeon does not attempt to spare some tissues and treat others but to achieve inactivation or destruction within the targeted volume. Obliteration of the vascular supply with accompanying endothelial damage of vessels to the tumors also seems to play a much more significant role in radiosurgery than in radiation therapy.

Conventional Radiotherapy

External beam radiation most commonly uses x-rays, gamma rays, or electrons and is the most common form of therapeutic radiation for cancer treatment. Early external beam units could generate x-rays with only relatively low energy and achieve just a superficial depth of penetration. These devices were not capable of treating deep-seated brain lesions. The introduction of ⁶⁰Co teletherapy machines and LINACs has allowed the routine use of megavoltage (>1 MV) high-energy beams. These beams have penetrating power sufficient to reach intracranial lesions.

LINACs are the most commonly used machines for the clinical delivery of radiotherapy. These machines work by accelerating electrons to high kinetic energy and directing them to strike a target with high atomic number (e.g., tungsten) to produce x-rays. In modern LINACs, acceleration of electrons is accomplished by using microwave-frequency waveguides whose oscillating electromagnetic fields interact with the electrons and cause them to gain energy. Once accelerated, systems of magnets direct the electron beam to the target. High-energy electrons interacting with the target material cause the creation of x-rays primarily by bremsstrahlung interactions. The resulting x-rays are then shaped with collimators and flattening filters to achieve a desired field size and beam characteristics.

LINACs have fixed-direction beams with a treatment head that is usually mounted on a rotating gantry. The axis of the rotating gantry is called the isocenter, and it is the point through which the beam will pass regardless of the position of the gantry.

Each treatment with a LINAC requires that the patient be properly positioned relative to the isocenter so that the appropriate dose is delivered to the target.

Brachytherapy

In contrast to external beam therapy, in brachytherapy a radioactive source (or array of sources) is placed within the patient and left for a predetermined period. This technique allows high local doses of radiation while minimizing damage to surrounding tissue, but it can be used only for sites that are accessible through either existing body cavities or surgical insertion.

Because brachytherapy deals with radioactive nuclides, safety in handling is a concern. Afterloading techniques with machines that automatically load the radioactive sources and the use of low-energy isotopes have greatly increased radiation safety. If considered from the point of view of traditional radiotherapy theory, lack of homogeneity is an intrinsic weakness of brachytherapy for solid tumors. Nevertheless, intracystic instillation of an isotope with limited penetration provides an excellent means of delivering an adequate dose to the wall of a cystic neoplasm (e.g., a craniopharyngioma).

Stereotactic Radiosurgery

Leksell published his paper on radiosurgery in 1951.¹ He described coupling of the stereotactic technique with narrow beams of radiant energy to target an area in the brain. The points of entry of the beams were distributed over the convexity of the patient’s skull. At the time, Leksell was using x-ray radiation as his energy source.

In the perspective of the past half century, Leksell’s paper turned out “to be a milestone in contemporary neurosurgery”¹⁰ that totally changed our way of viewing the management of neurosurgical pathology and forced “reluctant neurosurgeons to consider major changes in classic thinking about the proper care of many illnesses including vascular malformations, cavernous sinus meningiomas and vestibular schwannomas.”¹¹

Radiosurgery proved successful after development of the Gamma Knife and modification of LINACs and cyclotrons for conformal delivery of beams to intracranial targets, as well as targets in other parts of the body. Radiosurgery is a minimally invasive technique designed by Lars Leksell to deliver a destructive amount of radiation to intracranial lesions that may be inaccessible or unsuitable for open surgery. Undoubtedly, the experience of delivering ether anesthesia to neurosurgical patients for Dr. Olivecrona motivated Leksell to devise a technique associated with fewer complications than occur with open surgery. A passage from Leksell’s autobiography proved that the idea of a minimally invasive neurosurgical approach was on his mind for some time. At the first Scandinavian neurosurgical meeting held in Oslo, Leksell left the conference room during a less than exciting presentation and decided to walk in a garden. While on this walk, Leksell met Sir Hugh Cairns. Leksell confessed to Cairns his doubts concerning the state of neurosurgical techniques available at the time and was convinced that something new had to be developed. He explained his plans to mechanically direct a probe into the brain by using perhaps the brain’s own electrical activity and ablate pain pathways. He also mused about the idea of using narrow-beam x-rays or ultrasound as the physical agent and doing away with the probe entirely. His enthusiasm and ideas were given a warm reception by Cairns, and the encouraged young Leksell started work that led to the development of an “arc-radius” type of stereotactic system. Leksell wrote, “I was born under the sign of the ‘archer’ and looked forward to sharpshoot into the brain.”¹²

Leksell first built on the principles of a target-centered semicircular stereotactic arc. In the early 1950s, he replaced electrodes with ionizing radiation using x-rays. Building on the work of radiation therapists who were treating pituitary adenomas with higher dose, lower fraction protocols, Leksell postulated that a single fraction of radiation could be even more beneficial for intracranial pathology.

In the following 10 years, Leksell made considerable progress in the treatment of deep brain structures with a single heavy dose of radiation. He collaborated with physicists Kurt Liden and Borje Larsson to use a proton beam for radiosurgery.^13–15 The first stereotactic proton beam operation was performed at the Gustaf Werner Institute in Uppsala in 1960. Leksell found the synchrotron too awkward and expensive for widespread use and consequently developed a similar technique based on the LINAC. At the time, however, LINACs lacked the precision needed for use by neurosurgeons. Leksell also tried focused beam ultrasound, but this too lacked precision and required that a cranial defect be made before its use.

The next logical step was to look for another radiation source. Leksell turned to ⁶⁰Co. The first stereotactic Gamma Knife unit was installed in Sophiahemmet Hospital in 1968.¹⁶ The unit was originally intended for functional neurosurgery (Larsson and colleagues, 1974). However, the applications were quickly expanded to include AVMs and certain brain tumors.^17,18 The first Gamma Knife yielded fairly promising results, and an improved second Gamma Knife unit was built and installed at the Karolinska Institute in 1974. This unit proved to be both reliable and easy to use. Leksell wrote that, “Maybe the most important lesson learnt at the Karolinska is that the simplicity of using the Gamma Unit makes this integration possible and that the same individual can be a competent microsurgeon and also a stereotactic radiosurgeon. Someone competent in both techniques is best fitted to decide where the boundaries between the two methods should lie.”¹⁶

In 1983 at a hospital in Buenos Aires, Betti and Derechinsky introduced the concept of a modified LINAC for radiosurgery.^19,20 This system relied on a 10-MV LINAC and used a chair-based Talairach stereotactic frame.²¹ Other innovative developments in LINAC-based radiosurgical devices quickly followed from Hartman and colleagues in Heidelberg, Barcia-Salorio and associates in Valencia, Colombo and coworkers in Vincenza, Podgorsak and colleagues in Canada, and Friedman and Bova in Florida.^{19,20,22–26} At each of these centers, neurosurgeons played a critical role in the refinement of LINACs for radiosurgery. Similarly, the multileaf collimator for LINACs was designed in part to achieve conformality. Innovative work in the field of radiosurgery involving the use of heavy particles from cyclotrons was also conducted by Raymond Kjellberg, Jay Loeffler, and Jacob Fabrikant.

Superior results in radiosurgery can be achieved with several different types of devices. Although a precise and accurate device is critical to radiosurgery, the user of the device is equally if not more so important. To paraphrase Dr. Charles Wilson regarding a horserace, the jockey and not just the horse can make the difference in winning the race. In the words of Mark Twain, “a fool with a tool is still a fool.”

Radiosurgery is possible because of the synthesis of two concepts: stereotaxis as applied to delivery of radiation and three-dimensional visualization. Stereotaxis allows precise localization of a target point to be determined within a coordinate system known as stereotactic space. This is typically accomplished by attaching a rigid stereotactic head ring to the patient. The head ring is of known dimensions, and by affixing it to the patient’s head a physical correspondence between the stereotactic coordinate system and the volume of the head is established. A number of stereotactic systems have been developed, including the Spiegel-Wycis frame,²⁷ the Leksell stereotactic frame,²⁸ the Talairach frame,²⁹ and the Todd-Wells frame, which eventually became the CRW and BRW frames.³⁰

Accurate three-dimensional visualization is accomplished by using a system of three-dimensional fiducials that are attached to the head ring during imaging. The fiducials establish a correspondence in the images between the stereotactic coordinate system and the image coordinate system, thus making possible precise localization of each voxel in the resulting imagery.

The result is that a surgeon can create a plan that treats the target of interest by establishing the precise locations at which the radiation must be directed. Because the precise locations in stereotactic space are known, the treatment unit (which is calibrated to work in the same coordinate space) can precisely execute the treatment.

The main requirements for radiosurgery are (1) accurate determination of the target with stereotactic techniques; (2) direct superimposition of isodose distributions on images showing the anatomic location of the target; (3) accurate knowledge of the dose for a particular pathology; (4) steep dose falloff immediately outside the target; (5) low doses delivered to the skin, lens, and other critical intracranial structures; and (6) treatment accomplished in a reasonable amount of time. Various techniques have been used to fulfill the goals of radiosurgery and are discussed in the following sections.

Linear Accelerator–Based Radiosurgery

LINACs were first proposed as radiation sources for radiosurgery by Larsson and coauthors in 1974. The earliest reports on clinical LINAC-based radiosurgery were published in 1983 by Betti and Derenchinsky²⁰ and in 1985 by Colombo and coworkers³¹ and Hartmann and associates.²⁴ The LINACs used for radiosurgery are usually modified from machines used for routine cancer therapy to achieve smaller beam sizes and more precise positioning specifications.

A variety of approaches have been developed to use LINACs as a radiosurgical tool; however, most approaches follow the same basic technique. Combinations of treatment table, gantry rotation, and collimator rotation movement are used to direct the photon beams to the intracranial target from many different directions (instead of the one to five beams used for traditional radiation therapy, more than five beams are often used). By making use of two intersecting axes of rotation and putting the center of the target at this intersection point, beam entry points over the entire upper hemisphere of the skull can be accessed. Multileaf collimators are used to shape the treatment field at each location and may be modulated to achieve particular dose distributions. If x-rays are directed into the head while the gantry is rotating, the central line of the beam might trace out paths, called arcs. Combinations of arcs and modulating multileaf collimators can assist in achieving conformal dose distributions. Micro-multileaf collimators have further improved the accurary, precision, and conformality of LINAC-based radiosurgery to the brain and spine. Finally, a number of dedicated LINAC-based radiosurgical solutions are available on the market, including Varian’s Trilogy (Palo Alto, CA), Elekta’s Synergy (Elekta AB, Stockholm), Tomotherapy Hi-Art (Tomotherapy, Inc., Madison, WI), the CyberKnife (Accuray, Inc., Sunnyvale, CA), and Novalis (BrainLab, Germany).

Proton Beams

To make a charged particle beam usable for radiosurgery, several modifications are required. Because protons beams travel to different depths according to their energy, to provide the best match of Bragg peak effect to the target tissue, the energy of the beams and spreading of the Bragg peak need to be adjusted during treatment. This can be accomplished through the addition of variable-thickness absorbers to adjust the range of energy. If four beams from different directions are used to treat a target, it is likely that the target is shaped differently from each beam’s perspective. A beam-shaping aperture placed near the surface of the patient provides the cross-sectional conformation. To treat patients with charged particles, very detailed information about the target from the perspective of each beam is required. Each beam needs a customized range-modifying absorber, a variable-thickness rotating absorber, and a beam-shaping aperture. Charged particle radiosurgery can produce excellent dose distributions, but the treatment process is much more time-consuming, expensive, and difficult than the Gamma Knife or LINAC approaches.

Gamma Knife Radiosurgery

First introduced by Lars Leksell in 1951, radiosurgery combines stereotactic technique with highly focused radiation beams to deliver a large dose to the target while keeping a low dose of radiation to surrounding normal brain parenchyma.

Gamma Knife Description

The Gamma Knife unit (Fig. 249-3) itself consists of several main components: a large spherical shield (the bulk of the unit) that contains the array of cobalt sources and protects the patient and operational staff from gamma emissions, a central body that actually holds the source array and contains the primary collimation system that directs the gamma rays to the focus point, a treatment table that moves the patient’s head in and out of the unit (and in more recent units precisely positions the head so that the target is at the focus point), a control suite to allow operational control of the unit, and a treatment planning system that allows the neurosurgeon to create appropriate dose distributions.

FIGURE 249-3 The Leksell Gamma Knife PERFEXION.

Older Gamma Knife units (Models U, B, C, and 4C) use an external, helmet-based system for final beam collimation. Each helmet has a number of removable collimators machined to result in a particular field size (4, 8, 14, or 18 mm). Individual collimators may be replaced with solid “plugs” to achieve particular beam-shaping effects, which is used mainly for the protection of critical structures proximal to the target volume. Treatment plans that make use of more than one field size or use plugs require the operator to change helmets/plugs during the treatment.

In the most recent PERFEXION Gamma Knife, the external collimation system has been replaced by a single, internal collimating structure with precisely machined individual collimators (4, 8, and 16 mm). The ⁶⁰Co source array has been split into eight sectors with source holders that can slide on linear bearings driven by motors at the rear of the unit to align the sector with any of the available collimator sizes or a “blocked” position. Thus, each of the eight sectors may be configured independently of the others so that a “composite” isocenter can be created that is composed of multiple field sizes. Shielding with plugs (a highly manual process) has similarly been replaced by the fully automated process of setting a sector to the blocked position.³³

Central Nervous System Toxicity of Fractionated Radiation Therapy

A series of clinical syndromes could develop after fractionated radiation treatment of the central nervous system. These syndromes occur in a distinct chronologic order and have characteristic pathophysiologic changes.

Acute radiation necrosis occurs within hours or days after radiation exposure. This response usually requires an extremely high dose of radiation delivered in a short period. Patients may exhibit nausea and vomiting followed by disorientation, respiratory distress, seizure, coma, and death. This acute encephalopathy is due to disruption of the blood-brain barrier.

Early delayed complications occur a few weeks to months after radiotherapy. Early complications are caused by white matter injury characterized by demyelination and vasogenic edema. Injury may produce a somnolence syndrome in children, reappearance of the initial tumor’s symptomatology, a temporary decline in long-term memory, and encephalopathy. Patients may have increased edema and contrast enhancement on magnetic resonance imaging (MRI) that may resolve spontaneously over a period of a few months. Both acute and early delayed complications are responsive to steroids.

Radiation necrosis is a long-term complication that occurs 6 months to even decades after radiation treatment. Radiation necrosis is coagulative and predominantly affects white matter. This coagulative necrosis is due to small-artery injury and thrombotic occlusion. These small arteries demonstrate endothelial thickening, lymphocyte and macrophage infiltration, hyalinization, fibrinoid deposition, thrombosis, and finally occlusion. The vascular endothelial injuries lead to damage to oligodendroglia. As a result, white matter tissue is often affected more than gray matter tissue. In addition to vessel occlusion with resultant tissue necrosis, demyelination, axonal swelling, reactive gliosis, and disruption of the blood-brain barrier can also be observed.

Another late complication is diffuse cerebral atrophy. Diffuse cerebral atrophy is clinically associated with cognitive decline, personality changes, and gait disturbances. In children, radiation therapy, surgery, and the intracranial tumor contribute to intellectual decline. The radiation-associated decline in intelligence is generally in the area of performance, especially visuospatial integration.^34–36 Because myelination is not complete until the age of 2 to 3 years, young children are at greater risk.³⁷ A decline in verbal IQ has also been noted. As in adults, previous acquired knowledge is usually preserved, but there is a deficit in acquiring new skills.³⁸

Complications after Radiosurgery

The radiosurgical procedure is not generally associated with any immediate- or short-term side effects. Patients sometimes experience nausea. Rarely, seizures occur in the post-treatment period, usually in patients with supratentorial lesions and in those with a history of seizure disorders. Overall, radiosurgery does not appear to significantly alter the seizure threshold in patients with intracranial disease.

Radiation-induced changes are characterized by a bright signal on T2-weighted MRI. In cases in which it is associated with contrast enhancement on T1-weighted MRI, it presumably represents radiation-induced injury with an associated breakdown of the blood-brain barrier. Guo reported that radiation-induced changes can be observed in 47% of patients after Gamma Knife radiosurgery for AVMs. The onset of these changes occurred 3 to 15 months after treatment in the majority of patients (92%) and more than 26 months after treatment in 8%.³⁹ Progressive resolution of the radiation-induced effects is the usual course. The clinical manifestations included headache, symptoms of raised intracranial pressure, and focal neurological deficits. In a small percentage of patients, it is associated with focal damage to neural tissue. Neurological deficits were still present at the time of the last follow-up in 3% of patients.

Flickinger and associates evaluated follow-up imaging and clinical data in 307 patients with AVMs treated by Gamma Knife radiosurgery.⁴⁰ Radiation-induced changes developed in 30.5% of patients and were symptomatic in 10.7%. The changes resolved within 3 years and did so significantly less often in patients with symptoms (52.8%) than in asymptomatic patients (94.8%). The 7-year actuarial rate for the development of persistent symptomatic radiation-induced changes was 5.05%. Multivariate logistic regression modeling found that the 12-Gy volume was the only independent variable that significantly correlated with the changes whereas symptomatic radiation-induced changes were correlated with both 12-Gy volume and AVM location. They suggested that complications from radiosurgery for AVMs can be predicted with a statistical model relating the risk for development of symptomatic imaging changes after radiosurgery to 12-Gy treatment volume and location.

The nature of the radiation-induced changes remains to be fully elucidated. These changes presumably represent a whole gamut of pathologic processes ranging from gliosis to true necrosis. It is important to emphasize that the signal changes on MRI associated with clinical deterioration are often incorrectly interpreted as radionecrosis despite the fact that the changes are often transitory.