Photon Radiation

The most common form of ionizing radiation used in radiotherapy is the photon, which is a particle that has no mass, travels at the speed of light, and carries the energy present in microwaves, visible light, ultraviolet light, and x-rays. The nature of the interaction of a photon with matter depends upon the energy of the photon. In the kiloelectron volt (keV) range, which is used in most diagnostic x-ray units, photons interact with matter via the photoelectric effect.² A photon interacts with a tightly bound electron, which absorbs most of the energy and is ejected from the atom and is then free to interact with other atoms in the vicinity. This interaction is highly dependent on the atomic number of the material being irradiated. Diagnostic x-ray studies take advantage of this phenomenon, as bone, which is high in calcium, is much more likely to interact with these photons than soft tissue, which is mostly carbon, hydrogen, and oxygen. This differential interaction is the basis of diagnostic x-ray imaging.

At higher energies, which is the energy of most therapeutic radiation, the Compton effect predominates.² A high-energy photon in the megaelectron volt (MeV) range interacts with a loosely held orbital electron, which results in ejection of the electron, and scattering of the photon (with a change in energy of the scattered photon). This interaction is independent of the atomic number of the material being irradiated, but it is dependent on electron density. Thus, images generated from therapeutic energy x-rays are less useful for imaging than diagnostic x-rays, as there is little contrast between bone and soft tissue. However, this energy range is useful in radiation therapy, as it is highly penetrating, and is able to interact equally with all tissues.

There are two main sources of high-energy photons used in radiation therapy. The first source is radioactive decay, which is a natural process that occurs in elements with unstable nuclei and results in the emission of energy as the nucleus gains stability. Cobalt-60, the most common element used in radiation therapy units, undergoes a process known as beta decay to become nickel-60, which results in the emission of a high-energy photon known as a gamma ray (γ-ray). For ⁶⁰Co, the average energy of the photons generated in the radioactive decay is 1.25 MeV. Radiotherapy devices that take advantage of this phenomenon use a shielded radioactive source, and a small opening or openings result in shaping (known as collimation) of the beam to the desired size and shape. As will be discussed later, this is the source of radiation used in devices such as the Gamma Knife.

The other main source of high-energy photons is a linac, which uses microwaves to accelerate electrons to a high energy. These electrons are directed to collide with a high-molecular-weight target, and the resulting interactions between nuclei and electrons result in the production of high-energy photons (X-rays) via a process known as bremsstrahlung, or “braking” radiation. The difference between X-rays and γ-rays is simply the site of origin: X-rays are produced by electron interactions but γ-rays are produced by nuclear decay. Unlike the γ-rays produced by ⁶⁰Co, the energy of the X-rays generated in a linac is based on the characteristics of the machine, and most commercial units offer multiple energy options. These X-rays are produced in the head of the unit (known as the gantry), and can be shaped by a system of controllable leaves, known as a multileaf collimator (MLC), placed between the source of the X-rays and the patient (Fig. 45.1).

FIGURE 45.1 Linear accelerator diagram. Left, A schematic drawing of a linear accelerator. Electrons are generated in the electron gun, and accelerated using microwave energy in the waveguide. The electrons are bent 270 degrees to hit a high-molecular-weight target, which generates high-energy x-rays that are then shaped using the collimator. Right, An actual linear accelerator in clinical use. The gantry is capable of 360-degree rotation to deliver radiation from every angle.

The physical characteristics of photons determine their biological effects. Photons deposit dose in a characteristic pattern, in which there is an initial buildup of energy after entry into the patient, followed by a steady loss of energy as they pass through the tissue according to the law of exponential decay. The energy of the photon is lost in inverse proportion to the square of the distance traveled. Thus, photons spare the skin to some degree, but deposit the majority of their dose upon entrance into tissue, and continue to deposit decreasing amounts of energy as they exit the body (Fig. 45.2). The shape of this depth-dose curve is dependent on the energy of the photon involved. Higher energy photons exhibit greater skin-sparing effect, but have a slower rate of attenuation as they pass through tissue. In general, photons in the 1- to 6-MeV range are used for SRS, as the desired target depth in the cranium is relatively shallow, and higher energies result in unnecessary increased dose delivered beyond the target point.

FIGURE 45.2 Depth-dose curve for photons and protons. The x-axis represents depth in tissue, and the y-axis represents the percentage of the total dose delivered (where 100% is the dose at 14.5 cm deep in tissue, a typical depth for cranial radiosurgery). Photon radiation (here shown as a 6-MeV beam) increases upon entrance into tissue until it reaches a maximum, and then decreases in a manner that is inversely proportional to the square of the distance traveled. Proton radiation (pristine proton peak) shows a low entrance dose, and a very sharp buildup known as the Bragg peak. For larger lesions, a spread-out Bragg peak (SOBP) can be used to ensure that the entire target volume is covered by the high-dose region, although this results in higher entrance dose.

(Image courtesy of Marc Bussiere, MSc.)

Proton Radiation

Another form of ionizing radiation used in SRS involves charged particles, most commonly protons. Proton production starts by stripping the electron from molecular hydrogen gas, and the resulting protons are then accelerated to a therapeutic energy level using alternating magnetic fields in a cyclotron or synchrotron. The physical properties of protons result in a different dose distribution from that of photon radiation. Protons have a defined distance of travel, known as the range, which is dependent on their energy. Protons release their energy primarily at the end of their range, which is known as the Bragg peak, after William Henry Bragg, the Nobel Prize–winning physicist. The consequence of the Bragg peak is that protons deposit relatively lower doses of radiation prior to the end of their range compared to photons, and have no exit dose, that is, no dose delivered past their range (see Fig. 45.2). By modulating the energy of protons, a spread-out Bragg peak (SOBP) can be generated to cover wider areas, with the tradeoff of increased entrance dose. Because of their sharp range and lack of exit dose, protons provide a favorable dose profile for use in SRS. The disadvantage of protons is that their availability is limited to a few centers due to the cost and complexity of maintaining such facilities for clinical use.

Fractionation versus Radiosurgery

Conventional radiotherapy is typically fractionated into daily doses, which is thought to result in reduced effects of radiation on normal tissue. Fractionation allows for DNA repair to occur, which is predicted to favor normal cells that retain the full complement of DNA repair proteins.¹⁹ Furthermore, fractionation allows for reoxygenation of hypoxic areas, resulting in increased sensitivity of malignant cells that were previously hypoxic.²⁰ Additionally, fractionated therapy allows for reassortment of cells in the cell cycle, as cells are most sensitive to radiation during the G2/M phase of the cell cycle and resistant during the late S phase and G1. Thus, giving radiation in daily fractions allows those cells that are in resistant phases of the cell cycle to move to more sensitive phases of the cell cycle during subsequent fractions.²¹ The downside of fractionation is that it allows for repopulation of tumor cells during the therapy. Reoxygenation, reassortment, repair, along with repopulation, are known as the “four Rs” of radiobiology, and explain the radiobiological basis for daily fractionated therapy.

The value of fractionation is more apparent for certain tissues, which typically have high rates of proliferation and show relatively less capability to repair sublethal DNA damage. This concept is quantified as the α/β ratio, which is a radiobiological concept, based on a model of radiation response, that attempts to explain the differential sensitivity of tissues to fractionation.²² Tissues with a high α/β ratio respond quickly to radiotherapy, and are sensitive to smaller fraction sizes. Examples of these so-called “early” responding tissues include the gastrointestinal tract, lymphocytes, and skin. Tissues with a low α/β ratio respond more slowly to radiation, are typically less proliferative, and show a high capacity for DNA repair. Examples of late responding tissues include neural tissue and the lung.

The α/β ratio can be used to calculate a biologically equivalent dose (BED), which attempts to equalize total doses that are given in different fraction sizes. The equation is BED = (number of fractions ∗ fractional dose) ∗ (1 + (fractional dose/α/β ratio)).²² Although there is debate over the validity of this model at the high doses used in radiosurgery, it allows one to approximate the effect of doses delivered in fractionated sessions and those given in radiosurgery. For example, if one assumes a tumor to have a high α/β ratio of 10, a dose of 20 Gy in a single fraction is biologically equivalent to a dose of 40 Gy in 8 fractions, or 50 Gy in 25 fractions. However, for normal brain tissue, which has an α/β ratio closer to 3, the schedule 20 Gy × 1 has a BED of 153.3, while 5 Gy × 8 has a BED of 106.7, and 2 Gy × 25 has a BED of 83.3. Thus, the preceding regimens have the same BED for a tissue with a high α/β ratio, but the larger fraction sizes have a much higher BED for tissues with a low α/β ratio. What this means is that although the tumor control would be predicted to be equivalent for 20 Gy in 1 fraction and 50 Gy in 25 fractions, the radiosurgical dose would produce a more profound effect on normal brain tissue. Although the BED is merely an approximation of biological effect and not a real quantity, it can be useful when considering the dose of radiation to be used when the target is close to critical normal structures such as cranial nerves.

Given these advantages to fractionation, the radiobiological rationale for radiosurgery is not immediately obvious. However, proponents of radiosurgery argue that there are some principles that make SRS radiobiologically favorable. First, radiosurgery relies on precise immobilization and localization of the tumor to be treated, with inclusion of minimal margin of normal tissue. The design of the treatment machine, as outlined later in this chapter, creates a high dose of radiation in a very precisely defined space, with a very rapid dose falloff outside the target volume. Thus, even if critical normal structures are very close to the target tissue, the dose received is much lower than that received in the target itself. Thus, the physics of radiosurgery can compensate for suboptimal radiation biology.

Another potential explanation for the excellent clinical effect of radiosurgical doses in malignant disease lies in the possibility that high radiation doses are required to kill radiation-resistant cells. Emerging data suggest that clonogenic cells within tumors may have an intrinsically higher radioresistance due to increased expression of antioxidant genes.²³ It may be that very high radiation doses are able to overcome this effect, resulting in death of those clonogenic cells, and leading to tumor control.

Finally, as described earlier, at high doses of radiation, the primary mode of cell death may be due to endothelial damage, which may explain why the radiobiological rationale of fractionation may not be applicable when considering the observed clinical efficacy of SRS, especially for such targets as arteriovenous malformations. If radiation-induced apoptosis of the endothelium is a response to high dose per fraction, then fractionation may result in less normal tissue toxicity but also less control of the targeted lesion. All of these mechanisms are speculative and remain to be proved in a rigorous scientific fashion.

The radiobiological concerns outlined here suggest some situations when it may be advantageous to choose fractionated treatment over single-fraction radiosurgery. Given that normal neural tissue, including cranial nerves, are late-responding, with an α/β ratio of approximately 3, they will be more sensitive to large radiation doses compared to small ones. If the target to be treated is in close proximity to one of these structures (e.g., optic nerve sheath meningioma²⁴ or a pituitary tumor very close to the optic chiasm²⁵), it may be more prudent to fractionate treatment to reduce the risk of injury to the cranial nerves. Conversely, for hormone-secreting tumors, there is biological evidence that high-dose radiosurgery is associated with more rapid normalization of hormone levels.²⁶ Thus, the choice between radiosurgery and fractionated treatment must weigh a number of radiobiological and clinical parameters.

Gamma Knife Radiosurgery

The Gamma Knife was pioneered by Dr. Leksell in the late 1960s, and became the first available commercial radiosurgery unit. The unit, as currently constructed, contains 201 fixed sources of ⁶⁰Co distributed in a hemisphere, each of which is a thin rod with the long axis oriented along the radius of the sphere, converging on a single point, called the treatment isocenter (Fig. 45.3). These sources are surrounded by a primary collimator, which directs the beams. A series of secondary collimators, or “helmets,” are available with either 4-, 8-, 14-, or 18-mm diameter holes, which can be chosen based on the desired size of the beams to be produced (Fig. 45.4). This secondary collimation determines the width of the radiation beams at the isocenter. By occluding certain collimator holes, the dose distribution can be altered to produce the desired shape and to protect normal structures. Newer models of the Gamma Knife allow for intensity modulation, which can produce more conformal dose distributions.

FIGURE 45.3 Schematic depiction of Gamma Knife radiation of an intracranial target. Sources in the primary collimator generate γ-rays, which penetrate through holes in the secondary collimator helmet to converge on a target from multiple directions. This convergence maximizes dose at the isocenter.

(Adapted from National Regulatory Commission file photo.)

FIGURE 45.4 A. The Gamma Knife Perfexion unit. The radioactive sources are contained within the shielded housing. B. One of the helmet collimators that is used to create the desired dose distribution.

(Image courtesy of Elekta AB.)

For radiosurgery, the Leksell frame is attached to the patient, and then a planning CT scan is performed and is uploaded to the treatment planning software. Once a plan is developed, the frame is attached to the treatment couch. The patient is then inserted into the unit, and the frame is positioned relative to the collimator helmet so that the treatment isocenter is aligned with the planned target isocenter. Because there are no moving parts during treatment, there is a high degree of setup accuracy. The dose to be delivered, along with the activity of the sources, determines the length of the treatment time. If multiple isocenters are used, the patient is repositioned relative to the secondary collimator, and the rest of the isocenters are treated. The half-life of ⁶⁰Co is 5.3 years, and the sources need to be replaced periodically so that treatment times do not become impractical for patients. Dose rates fall 1% per month with ⁶⁰Co treatment devices.

The Gamma Knife produces a very sharp dose gradient, with the highest dose at the treatment isocenter. Because the prescription dose is at the surface of the volume to be treated, which usually corresponds to the 50% isodose line, the result is that the dose at the treatment isocenter is usually at least 50% higher than the prescription dose. This dose heterogeneity can be helpful if it is present at the center of a malignant lesion, but can be troublesome if there are concerns about tissue necrosis. Although the dose distribution for a single isocenter is spherical, multiple isocenters can be used in an overlapping fashion to create an irregularly shaped volume. The drawback to this approach is that the dose heterogeneity is increased across the lesion.

Linac-Based Radiosurgery

As described earlier, the linac uses microwave energy to accelerate electrons into a therapeutic energy range, and those electrons are used to generate high-energy photons. For radiosurgery, the linac has a secondary collimator attached to the treatment head with a cone-shaped aperture of a specific size, which can be varied based on the size and shape of the lesion to be treated. To produce a dose distribution similar to that of the Gamma Knife, an approach using converging noncoplanar arcs of radiation was adopted³⁸ (Fig. 45.5).

FIGURE 45.5 Example of a linac-based stereotactic radiosurgery (SRS) treatment plan. Five noncoplanar arcs are depicted converging on the treatment isocenter. The eyes and the optic chiasm are also seen as three-dimensional (3D) reconstructed objects so that the treatment planner is aware of these critical normal structures during treatment planning.

(Image courtesy of Kevin Beaudette, MSc.)

A stereotactic frame, most commonly the Brown-Roberts-Wells (BRW) frame, is placed on the patient, with four pins securing the frame to the skull. The frame has three orthogonal axes—coronal, lateral, and axial—which intersect at the center of the circular frame. The CT localizer frame has nine fiducial rods, which appear in each axial CT slice, and provide a precise correlation between the axes on the frame and any anatomical point on the CT scan. The frame attaches to the CT simulation couch in a similar manner to the linac treatment couch, so that patient setup is identical in the simulator and the treatment unit. The target is outlined on the CT scan, along with any critical normal structures. MRI scans are often fused with the planning CT scan to facilitate target and normal tissue identification.³⁹

Treatment planning software is used to define 3 to 6 arcs converging on the treatment isocenter, which will result in a high dose at the convergence point, with rapid falloff in any given direction away from treatment isocenter (Fig. 45.6). Because arcs are used, the high-dose volume is usually more ellipsoid than a similar sized volume treated by Gamma Knife. As the number of arcs and the total degrees subtended by each arc increase, the dose to normal tissue decreases, as it is spread out over a larger volume.

FIGURE 45.6 Example of linac-based stereotactic radiosurgery (SRS) dosimetry. This image shows a representative axial computed tomography (CT) scan slice with a metastatic lesion in the upper cerebellum. Each colored line, known as an isodose line, outlines the area receiving at least the indicated dose. The dose is highest at the center of the target, and falls off rapidly with increasing distance from the isocenter. The directions of less rapid falloff depend on the configuration of the arcs used in the treatment plan.

(Image courtesy of Kevin Beaudette, MSc.)

When the plan is formed, the patient, with the frame still attached to the skull, is brought to the treatment room, and the frame is attached to the treatment couch. The isocenter of the treatment machine is aligned to the origin of the axes of the stereotactic frame, with an accuracy of less than 1 mm. The treatment isocenter is then aligned with the machine’s isocenter. The gantry rotates during the treatment to produce an arc. After the arc is delivered, the couch can be moved and a new arc created. The end result is that multiple arcs converge on a single point. For irregular volumes, multiple isocenters can be used, similar to those used by the Gamma Knife. When all of the planned arcs are treated, the frame is removed, and the patient can usually return home.

One of the more recent advances in SRS is the “frameless” radiosurgery system, which does not require an invasive head frame to be fixed to the patient’s cranium. Because conventional SRS required frame placement at the beginning of planning, the patient would often wear the head frame for the entire 5 to 8 hours required for planning and treatment. A number of centers have switched to noninvasive systems that are much more comfortable for the patient, including a thermoplastic mask combined with a dental mold with infrared fiducial markers,⁴⁰ or use of orthogonal diagnostic energy x-rays to verify patient setup and positioning. The other advantage of the frameless approach is that the planning and treatment can be done on separate days, which facilitates patient and physician scheduling. These approaches have been adopted at a number of centers, and in some preliminary studies appear to be of similar accuracy and clinical efficacy as conventional radiosurgery.⁴¹ However, there are some concerns that it may not be as accurate as conventional frames, and a planning margin of 1 to 2 mm is sometimes used to ensure target coverage.⁴² Thus, this approach has not been universally adopted, and is still under investigation.

Other recent developments in linac-based radiosurgery include the development of special radiosurgical linacs, of which the Novalis Tx is the most common. These linear accelerators can use cone-shaped collimators like conventional linac-based radiosurgery, but also have a multileaf collimator with small leaf size, allowing for very fine shaping of the radiation beam. These MLCs can be used for intensity-modulated radiation therapy (IMRT), in which the leaves attenuate the radiation beam to varying amounts in different areas of the radiation portal. The radiation is then delivered using a set of static (i.e., not arc) fields, with each field having a different intensity pattern. This technique allows for a much more conformal dose distribution, and can be especially useful in the treatment of irregular lesions. Furthermore, using IMRT, these specialized units can use a single isocenter to treat multiple lesions, which reduces treatment time for the patient. Initial planning studies suggest that the use of IMRT for radiosurgery rather than convergent arcs may allow for increased sparing of normal tissue receiving 50% of the prescription dose, and allows for more precise shaping around critical structures.⁴³ The downside of the IMRT approach is that there is more low-dose radiation given to normal brain, and the long-term consequences of this low-dose irradiation are still unknown.

Newer technologies for linac-based radiosurgery are on the horizon, including tomotherapy and volumetric arc therapy (VMAT), which deliver radiation in continuous arcs, similar to a CT scan. In these types of radiation therapies, the radiation is delivered in continuous arcs on a slice-by-slice basis as the patient is moved through the unit. Planning studies suggest that both tomotherapy and VMAT offer improved conformal dose delivery, but may result in larger areas of low-dose radiation to normal tissue,⁴⁴ although this area is still under investigation.⁴⁵

Robotic Radiosurgery

The last major development in photon radiosurgery is the development of the Cyberknife, in which a miniaturized linac is mounted on a robotic arm with 6 degrees of rotational freedom (Fig. 45.7). The linac produces 6-MeV x-rays, and is equipped with either swappable collimators of specific sizes or with a variable collimator that is adjustable to different sizes during the course of treatment. The linac and robotic arm combination can deliver multiple small beamlets of radiation from many different angles to produce a conformal dose plan. The Cyberknife is combined with an in-room diagnostic x-ray unit that takes images of the patient during the treatment, and can be used to modify the treatment plan in real time if there is any change in patient positioning. Thus, there is no need for an invasive frame, and immobilization with a thermoplastic mask is sufficient to achieve the required accuracy.

FIGURE 45.7 A Cyberknife robotic radiosurgical unit. The linear accelerator is mounted on a robotic arm, allowing for multiple beam angles.

The treatment plan is created from a large selection of possible robotic arm positions and possible locations within the tumor. Because the individual beamlets are small, each spot within the tumor can receive a different amount of radiation from each arm position, resulting in a conformality similar to that with IMRT. With a variable collimator, even more treatment options are available from each possible angle. Thus, the Cyberknife offers a very conformal dose plan, but the use of multiple small beams can result in lengthy treatment times for large, irregularly shaped lesions.

Proton Beam Radiosurgery

All of the methods outlined previously use photon radiation, but recently there has been a growing interest in using proton radiotherapy. As discussed earlier, protons provide a dosimetric advantage relative to photons owing to the concentration of energy at the Bragg peak and the lack of exit dose. The result is that the integral dose to the normal brain is lower than that for photon radiosurgery. The early radiosurgical approach adopted by Dr. Leksell in the 1950s used a cross-fire technique, in which opposed beams were directed at the target, but the Bragg peak fell outside the patient.³⁴ This approach was used because it was felt that the uncertainty of the dose at the Bragg peak could result in failure to provide adequate dose to the target. However, the team at the Harvard Cyclotron Laboratory, led by Dr. Kjellberg, began to perform Bragg peak radiosurgery beginning in 1961.⁴⁶ Over the past two decades, there has been a growth in the number of dedicated hospital-based proton therapy sites, beginning with Loma Linda University in 1990, followed by the Massachusetts General Hospital (MGH) in 2001. More recently, a number of centers are opening across the country as interest has increased in proton therapy.

For proton SRS, the immobilization and localization are similar to those used in various other radiosurgical setups. At the MGH, patients receiving proton SRS undergo placement of three radiopaque fiducial markers in the outer table of the skull with local anesthesia as an outpatient procedure. These markers, along with key points of bony anatomy of the skull, create a fixed axis system similar in concept to that created by BRW head frames used in photon radiosurgery.

For treatment planning, if the patient has good dentition, the patient is immobilized in either a modified Gill-Thomas-Cosman (GTC) frame, which consists of a custom occipital headrest and a custom bite block connected by a ring similar to a conventional head frame. If the patient has poor dentition, or has a low-lying tumor, a thermoplastic mask with polyfoam is created for immobilization. This is a marginally less accurate form of immobilization, but pretreatment imaging can correct for initial setup misalignments. With either form of immobilization, a CT scan is obtained with the patient immobilized, and the target and critical normal structures are contoured.

Creation of a proton radiosurgery plan typically uses 1 to 4 conformal portals. Because protons do not have an exit dose, fields can enter the body through a wider range of angles compared to photon arcs (e.g., from the vertex), because the protons will not exit through the rest of the body. The proton range is very sensitive to the tissue density through which the beam passes, and thus attempts are made to avoid sinuses or other air cavities. The energy of the proton beam is modulated to create an SOBP, which can be used to cover targets that are larger than the width of the single Bragg peak. For each field, a custom brass aperture is created to define the field edges, and a Lucite compensator is created to define the distal edge of the Bragg peak (Fig. 45.8).

FIGURE 45.8 Brass apertures and Lucite range compensator produced for proton radiotherapy. The brass aperture (on the left) defines the edges of a given stereotactic radiosurgery (SRS) field. The range compensator (on the right) determines the distal edge of the Bragg peak. Custom milling of the compensator creates areas of varying thickness of Lucite. The thicker areas of the compensator result in a Bragg peak closer to the surface, and the thinner areas result in a deeper Brag peak. Thus, the milling pattern determines the shape of the distal edge of the field.

(Image courtesy of Marc Bussiere, MSc.)

In the treatment room, the setup and positioning are verified with diagnostic x-rays that confirm the position of the fiducial markers in relation to the bony anatomy of the skull. Some patients are treated using a rotational gantry similar to that of a linac. At the MGH, where one of the treatment rooms has a fixed beamline, some patients are treated in a device known as STAR (stereotactic alignment for radiosurgery), which has greater ability to move and rotate the patient around a fixed beamline (Fig. 45.9). The fixed beam allows for treatment with a slightly higher degree of accuracy as compared to a rotational gantry because the weight of the gantry can result in some positional uncertainty.

FIGURE 45.9 Stereotactic alignment for radiosurgery (STAR) unit used for proton radiosurgery with a volunteer positioned for treatment. This unit immobilizes and positions the patient, and can be used for stereotactic radiosurgery (SRS) with a fixed proton beamline. The nozzle is the exit point for the protons from the beamline. The modified Gill-Thomas-Cosman (mGTC) frame is used for patient immobilization.

In general, proton radiosurgery results in a very conformal plan, with much less dose inhomogeneity compared to photon-based radiosurgery (Fig. 45.10). These homogeneous plans are desirable for heterogeneous lesions that contain normal tissues such that excessive dose is not wanted. This can be more advantageous than photon plans that produce unacceptable hot spots either within or outside the intended target. The second major advantage of proton radiosurgery is the lack of exit dose, which means that the treatment results in less radiation to the normal brain. Although the risks of low-dose irradiation of normal brain cannot yet be quantified, avoiding this possible complication is desirable.

FIGURE 45.10 Representative dose plan of proton stereotactic radiosurgery (SRS) compared to photon SRS. This is a coronal image of a patient with three atypical meningiomas being treated with SRS. The plan on the left shows a proton plan, where one can see that there is no exit dose from the beams used. The plan on the right shows a photon SRS approach, where there is a larger area of low dose given to normal brain tissue.

(Image courtesy of Marc Bussiere, MSc.)

Evaluation and Comparison of Radiosurgery Techniques

The evaluation of radiosurgical plans relies on a few parameters, which have been defined by ASTRO.⁴⁷ Key components to evaluate include the ratio between the maximum dose (MD) and the peripheral dose (PD), which gives a sense of the dose homogeneity within the treatment volume. The second key parameter is the ratio between the prescription isodose volume (PIV), which is the total volume that receives at least the prescription dose, and the target volume (TV), which is the volume of the target defined on planning CT scan. This PIV/TV ratio is a measure of target coverage and dose conformality.

Analysis of the Radiation Therapy Oncology Group (RTOG) 90-05 study, a multicenter phase II dose escalation study for SRS of brain metastases, revealed some interesting data regarding both Gamma Knife and linac-based radiosurgery. For Gamma Knife, 94% of the plans had an MD/PD ratio greater than 2, versus only 7% of the linac-based plans.⁴⁸ These data suggest that the Gamma Knife plans are less homogeneous and more prone to hot spots at the center of the treatment volume. However, it was noted that increasing the number of isocenters used for a given Gamma Knife treatment plan resulted in MD/PD ratios that were more similar to those with linac-based radiosurgery. Current RTOG protocols use a guideline of an MD/PD ratio of less than 2 to qualify as treatment “per protocol.”⁴⁹

In contrast, the results for the PIV/TV ratio were more favorable for Gamma Knife radiosurgery compared to linac-based radiosurgery. Seventy-seven percent of Gamma Knife plans had a PIV/TV ratio between 1 and 2, with no plans less than 1 or greater than 4. Conversely, only 51% of linac plans had a ratio between 1 and 2, and 10% of plans had a ratio less than 1 and 7% had a ratio greater than 4.^48,50 These data suggest that the conformality of plans with Gamma Knife were superior to those produced with linac-based radiosurgery. The RTOG currently recommends a PIV/TV ratio between 1 and 2 as a standard for SRS.⁴⁹

Although there are differences between plan characteristics, it is important to recognize that there do not appear to be clinical differences in efficacy or toxicity between radiosurgical modalities. Initially, during the development of linac-based radiosurgery, there was some concern that it would prove to be inferior to the Gamma Knife. Multivariate analysis of the RTOG 90-05 study showed that treatment with a linac was associated with an increased risk of local failure, with a hazard ratio of 2.85. The explanation put forward in the paper was that the higher MD/PD ratio and better PIV/TV ratio seen with the Gamma Knife plans resulted in better tumor control compared to linac-based plans. However, proponents of linac-based SRS suggested that this difference could also be due to the fact that Gamma Knife radiosurgery accounted for only 30% of the patients, and was limited to two clinical centers with significant radiosurgical experience, which is difficult to compare to the 70% of treatments performed in 15 institutions with varying degrees of experience with SRS. This hypothesis gained support with data from the RTOG 95-08 study, which randomized patients with one to three brain metastases to receive whole-brain radiotherapy with or without an SRS boost. In this study, there were very few major deviations from the MD/PD and PIV/TV ratios specified by the RTOG with either linac-based radiosurgery or with Gamma Knife, and there was no survival difference based on the type of radiosurgical unit used.⁵¹ Thus, the data suggest that when appropriate radiosurgical guidelines are followed, these two modes of SRS are functionally equivalent.

Fractionated Stereotactic Radiotherapy

Although the radiosurgical literature centers on single fraction radiosurgery for intracranial targets, there has been a growing interest in fractionating therapy based on the radiobiological advantages outlined here. Thus, several groups became interested in applying the immobilization and localization devices used in radiosurgical procedures with fractionated radiation to reduce the risk of late complications. The biggest challenge in fSRT is the reproducibility of positioning the patient daily such that the targeting accuracy is maintained. Most of the immobilization methods described here, especially the fixed head frame, were not suitable for multiday usage, although some groups have tried this approach.⁵² One of the most common solutions to this problem was to replace the pins in the head frame with a bite block/dental mold, and an occipital plate to hold the head in position, which became known as the Gill-Thomas-Cosman (GTC) frame.⁵³ This approach is well tolerated for both children and adults with intracranial tumors.⁵⁴ These approaches have been adopted with all of the technologies described earlier, including IMRT, Cyberknife, and proton radiotherapy. However, because of technical challenges, reports of fractionated therapy with Gamma Knife are much rarer.