Precision and History of Linear Accelerators

Over the years, radiation oncologists relied on the radiobiology of tissue response to moderate fractions of radiation and thereby decrease complications, hence the name radiotherapy. They did not stress precision of beam delivery. As specialists in cancer, they thought in terms of infiltrative disease, which needed large fields of radiation rather than concentration. Therefore, LINACs were not initially designed for high-precision delivery; the preoccupation was mostly with accuracy of the dose. When neurosurgeons tried to apply the LINAC beam to ablate lesions with a highly concentrated dose, precision of beam delivery became important.

The issue of precision was initially resolved with the description of a gantry correction device by two University of Florida scientists¹⁸ and lately by the development of LINACs dedicated to radiosurgery.¹⁹ This allowed LINAC radiosurgery to become a competitive technique, not only for lesions but also for functional disorders of the brain, such as trigeminal neuralgia.^20-24 LINAC radiosurgery is the most common form of radiosurgery performed today, and its versatility also allows application of the technique to the spine and other sites in the body (Fig. 255-1).^16,25,26 Another approach using a LINAC attached to a robot was advanced by the Stanford University group under leadership of the neurosurgeon John Adler.²⁷ This approach is gaining popularity, also because of its ability to reach extracranial targets and obviate the need for a stereotactic frame.

FIGURE 255-1 Example of precision capability and clinical application of linear accelerator radiosurgery in the spine. A, Sagittal T2-weighted magnetic resonance image of an arteriovenous malformation (AVM) treated at the C5 level. The patient recovered almost complete function after tetraparesis secondary to AVM bleeding. B, Radiosurgery performed with 12 Gy delivered to the 90% isodose line encompassing the AVM.

Dosimetry

The main dosimetric characteristic of radiosurgery is its ability to deliver a high dose of radiation to the target with a low dose to normal tissue in the periphery of the lesion. This can be accomplished when multiple fields converge on a point called an isocenter. The isocenter is generally placed by the LINAC planning software in the geometric center of the lesion. However, as a result of the pass point characteristics of converging beams, the maximum radiation point is usually situated slightly superior to the isocenter when planning intracranial radiosurgery because the multiple radiation fields enter from the top of the head, which brings the “hot spot” to a site slightly above the isocenter.

The limitation of the radiosurgery technique imposed by the beam is the intermediate volume. This is the area outside the lesion where the multiple fields (beams) partially overlap.¹⁹ As the target volume increases, the intermediate volume area also increases, which means that more volume of normal parenchyma is exposed to higher doses of radiation; consequently, the target volume in radiosurgery is suggested to be no greater than 3 cm (approximately 12.6 cm³). The target volume also has an impact on the shaping capabilities of LINAC radiosurgery. Modern LINAC conformality approaches, such as intensity-modulated radiosurgery (IMRS), allow the treatment of targets larger than 3 cm in largest diameter (Fig. 255-2).

FIGURE 255-2 Intensity-modulated plan to cover a large lesion close to the optic apparatus. A, Representation of the static fields. B, Representation of the beam modulation to achieve the dose distribution in C. Axial T1-weighted magnetic resonance image with gadolinium enhancement. Notice the sharp dose falloff to avoid the optic apparatus and the brainstem (arrow). Recurrence of this skull base meningioma was treated with stereotactic radiotherapy.

Functional Lesion Considerations

The dream of Dr. Lars Leksell of performing functional neurosurgery without violation of the cranium was realized. However, it did not become commonplace in functional neurosurgery. Radiation could not compete with radiofrequency heat generation as the approach of choice to create lesions in the brain⁴² despite the risks inherent in violating the skull and passing a probe to the depth of the brain. When Dr. Leksell conceptualized radiosurgery in 1951,² knowledge of the importance of electrophysiology in functional neurosurgery was very incipient and did not allow this perception at the time. Leksell and Larsson designed the first gamma unit with 179 cobalt 60 sources and slit collimators to generate an oval-shaped lesion and thereby mimic the heat lesion made with a radiofrequency probe.¹ This dose distribution shape, as used at the University of California, Los Angeles, in the 1980s with Dr. Leksell’s gamma unit, can be achieved with proper LINAC arc determination or cobalt source plugging, at least for the 50% isodose volume.^43,44

Functional neurosurgery then grew as a result of the ability of neurosurgeons to make minute and precise lesions in the brain, which is unpredictable with radiation. The technique of choice for making lesions in the nuclei and pathways of the brain is, still today, the heat lesion.^45,46 Heat lesions can be graded and monitored because of their radiofrequency attributes. It is possible to glimpse the effects of the lesion by partially increasing the temperature, for example, to 43°C. The lesion is then completed by increasing the temperature beyond the level needed to denaturize protein (i.e., 54°C). Moreover, localization of the target by imaging can be aided by electrical stimulation of the tissue in question. This cannot be accomplished with radiosurgery.

The effects of radiation are achieved slowly, well after the final decision for the lesion’s location is made. Trials of lesion making for the treatment of tremor and other symptoms of Parkinson’s disease fell short of the results expected with radiofrequency lesions because of the unpredictability of the effects of radiation on the target.^47-53 Another important drawback of functional radiosurgery is its inability to confirm correct targeting with electrophysiology.^54,55 The effects of radiation are not finalized until months to years after its delivery. Radiation has acute, delayed, and late effects,⁵⁶ a period during which the patient may not experience the benefits of radiosurgery but is actually having to endure its undesirable side effects.^57-59 In contrast to radiofrequency lesions, in which the effects are immediate and final in all cell types inside the target volume, the effects of radiation are different in quality and time frame for each cell type. Formation of the lesion after a high dose of radiation takes place in several stages because of the ability of cells to program apoptosis⁶⁰ and make changes in their machinery to produce neurotransmitters²⁹ and hyaline material.^61,62 Such stages range from acute activation of the target area,⁶³ to production of neurotransmitters²⁹ and cytokines,⁶⁴ and ultimately to thrombosis and ischemia.⁶⁵

Prescribing to a Point

The dose prescribed for functional neurosurgery is by convention and tradition directed to the isocenter. This means that 100% of the dose (maximal dose) is prescribed to a target point (i.e., to the isocenter). The radiation prescription dose is the same as the maximal dose when prescribing to the maximum. The falloff distance, or the volume of tissue receiving at least 50% of the dose, is proportional to the diameter of the collimator because circular collimators are traditionally used for functional radiosurgery (Fig. 255-3). Application of this concept is nicely seen during targeting of the root entry zone for trigeminal neuralgia in LINAC radiosurgery, where 3-, 4-, and 5-mm collimators are available for use.

FIGURE 255-3 Trigeminal neuralgia target in the root entry zone (REZ). A, CISS magnetic resonance imaging sequence to demonstrate the anatomy of the trigeminal nerve (TN), gasserian ganglion (GG), and basilar artery (BA). Notice the dose distributions of the 3-mm collimator (B), 4-mm collimator (C), and 5-mm collimator (D). It is not advisable to use collimators larger than 5 mm for treatment of trigeminal neuralgia with linear accelerator radiosurgery. Note that the 50% isodose line represents the 45-Gy volume. The 100% dose is delivered to a single point inside the target.

Placement of the isocenter while planning radiosurgery for trigeminal neuralgia relies on the isodose line (IDL) to determine the distance from the isocenter to the brainstem. Although some LINAC radiosurgery treatments of trigeminal neuralgia have been delivered with the 5-mm collimator,^{21,32,49,66-68} the majority of data on trigeminal neuralgia published by gamma unit users has been amassed with the 4-mm collimator.^69-73 The dose distributions for treating trigeminal neuralgia very well exemplify the concept of prescribing to a point in LINAC radiosurgery (Fig. 255-3).

Please see the section “Functional Stereotactic Radiation Therapy for Change in Seizure Focus Firing” on Expert Consult.

Prescribing to a Volume

Considerations of Volumetric Dosimetry

The simplest dose distribution is achieved with a radiosurgery plan involving a single collimator. However, placement of functional lesions only in targets such as the trigeminal nerve and the thalamus, perfectly round metastases/primary tumors, or small round AVMs is amenable to a single isocenter with circular collimators in LINAC radiosurgery. All other lesions do not have a shape that can be covered with a single isocenter without distributing too much radiation to the surrounding brain tissue.

Some fundamental concepts commonly discussed in dose prescription for LINAC radiosurgery are important to present at this point. With regard to brain metastases or small round gliomas, it is desirable to cover the gross contrast-enhancing lesion or gross target volume (GTV) along with some margin, usually 1 or 2 mm. Such a margin may encompass microdissemination of malignant cells surrounding the area defined by the contrast enhancement. This volume is called the clinical target volume (CTV). Besides the CTV, one should account for the uncertainties of the radiation delivery process, which at best approaches 2 mm with any technique (LINAC, Gamma Knife, CyberKnife). The final irradiated volume, defined as the CTV plus generally 2-mm margins, is called the planned target volume (PTV). The minimal radiation dose thought to be clinically safe and effective is prescribed to the PTV. The dose that adequately covers the target is designated the prescription radiation dose. Usually, in LINAC radiosurgery the prescription IDL is high (90%, 80%, or sometimes 70%), which means that the maximal dose is 10%, 20%, or 30% larger than the dose at the periphery of the lesion, depending on whether the prescription was to the 90%, 80%, or 70% IDL, respectively.

The different volumes just discussed are significant when treating malignant or benign recurrent tumors, but its real limits are beyond the contrast-enhancing margins used to determine a target volume. However, in some situations, addition of margins to the GTV is not biologically required. For instance, AVMs are not infiltrative lesions. Tighter plans are the best ones because the brain tissue affected by the AVM is normal and frequently eloquent.⁸⁴ Therefore, it does not make good sense to use additional margins. The progressive endothelial proliferation triggered by a single dose of radiation will promote cessation of blood flow through the AVM, even though a margin was not added to its boundaries. In the case of functional lesions, as discussed in the section “Prescribing to a Point,” the CTV and PTV concepts do not apply.

Radiation Dose Falloff

The dose falloff in LINAC radiosurgery is very steep. This accounts for the attractiveness of the method because it allows high radiation dose collimation inside the target with very fast radiation dose falloff in the normal brain tissue surrounding the target. Dose falloff varies according to collimator size and the type of planning used, such as multiple isocenter, dynamic arcs, or static beams (Fig. 255-4). This area of radiation falloff is called the penumbra. Consideration of the dosimetric consequences of the penumbra is important because the radiation dose may be still sufficiently high to cause toxicity in eloquent structures neighboring the lesion, such as the brainstem, motor area, and spinal cord.^86,87 Conversely, at the margin of a complex lesion, the falloff dose may still be effective in controlling tumor growth, although underdosed in relation to the remaining lesion volume.

FIGURE 255-4 A, Radiosurgery plan consisting of 16 isocenters for treatment of cavernous sinus meningioma. Notice the matching isodose line (IDL) in B, which has a single isocenter using a shaped-beam dynamic arc technique. Conformity of the prescribed radiation dose to the 50% (A) or (90%) IDL is satisfactory in both plans. C, Dose profiles of different circular collimator (cone) sizes. D, Insert showing the multileaf collimator.

In the context of skull base lesions abutting the optic apparatus or the brainstem or “donut”-shaped vertebral body lesions abutting the spinal cord, it is not wise to deliver the same radiation dose to these eloquent structures as prescribed to the lesion. Although it may be attractive to cover the lesion with additional safety margins, the risk of radiation-induced damage is not justifiable. In these situations, one possible approach is to slightly underdose the boundary of the lesion touching the eloquent structure so that the dose falloff occurs inside the lesion and not at the border of the lesion or outside the contrast-enhancing limits and additional three-dimensional margins. One modern technique available in LINAC radiosurgery to optimize dose falloff in the vicinity of eloquent structures is IMRS (Fig. 255-5).

FIGURE 255-5 A, Shaded representation of intensity modulation of the beam coming from the frontal region (yellow in the shaped beam in B) to treat a clival chordoma involving the brainstem (C). C, Dose distribution to avoid delivery of high isodose volumes to the pontine-mesencephalic region. Notice the cold spots (white arrow) and hot spots (black arrow) inside the intensity-modulated radiotherapy target volume.

Homogeneity

The addition of tridimensional margins to the GTV tailored to the constraints of the surrounding structures, as well as the different weighting and length of the arcs to achieve asymmetric dose falloff around a lesion, brings to the discussion the concept of homogeneity—a very controversial aspect of the radiation plan. LINAC users generally strive to implement the most homogeneous plan, whereas gamma unit users do not as a rule value homogeneity because of the limitations of the multiple-isocenter technique with regard to homogeneity.

It is intuitive to understand that a single collimator will distribute the radiation homogeneously throughout that area. To achieve conformity to an irregular external shape, multiple isocenters are used, which leads to an overlap of many single-isocenter dose distributions. The areas where the prescription dose of collimators intersects constitute “hot spots.” The number of hot spots in the PTV increases as the number of isocenters in the plan increases because of increased overlap. If one chooses to use intensity modulation, inhomogeneity generally increases. To allow steeper dose falloff in that particular border of the lesion, that area is considered cold and the beams are arranged in such a manner that many points inside the lesion become hot or cold to allow eloquent structures to be spared (see Fig. 255-5).

Inhomogeneity has been associated with increased complications after SRS.⁸⁷ The majority of LINAC users believe in the importance of homogeneity because of clinical evidence with regard to AVMs and the concept brought to the neurosurgery field by radiation oncologists.^16,19,88-90 Because the plans are designed to allow minimal difference between the radiation prescribed to cover the periphery of the lesion and maximal doses within the lesion, prescription volumes to a higher IDL are desirable. This paradigm can be achieved by treating irregular targets with shaped-beam techniques (discussed in the next subsection) but not with multiple-isocenter techniques.

Steeper falloff of a dose distribution is observed within the segment of the curve between the 70% and 90% IDL. Mathematical simulations of different IDL prescribing strategies for the same plan show that prescribing to the highest IDL has the advantage of sparing surrounding normal tissue.^40,87 However, the difference in sparing normal tissue with multiple-isocenter prescriptions at the 50% IDL versus LINAC shaped-beam prescriptions at the 90% IDL may be minimal in terms of clinical identification. Although hardly significant, less volume of normal tissue irradiated may account for fewer complications, even when treating large volumes.⁸⁴ For instance, if one prescribes 18 Gy to the 50% IDL, 20% of the total dose is 7.2 Gy. If the same prescription is to the 90% IDL, the dose drops to 4 Gy. With regard to treatment of an acoustic neuroma, for example, where the facial nerve is running in the anterior border of the neuroma, a dose of 3.2 Gy to the eloquent structure abutting the target is attractive. Although homogeneity is suggested to be advantageous mathematically, its translation into better clinical outcomes after radiosurgery remains to be proved. Unfortunately, a randomized clinical trial in which the IDL is the main end point variable, with balanced randomization of all others covariates playing an important prognostic role, is unlikely to be practical.

Volume-Shaping Techniques

The most challenging situation in making a radiosurgery plan is for an irregularly shaped lesion surrounding a critical structure. If the lesion is involving a critical structure, radiosurgery at doses effective in controlling the lesion may not be feasible. LINAC users take advantage of the radiobiology of fractionation and choose stereotactic radiation therapy.

Multiple-Isocenter Technique

Initially, the instrumentation available for delivering high doses of radiation to deep structures without subjecting normal brain tissue to such high doses used circular collimators. Careful plotting of many collimators of different size provided high conformality to very irregularly shaped lesions. The collimator sizes available in radiosurgery-dedicated LINACs generally range from 3 to 60 mm. As discussed earlier, a multiple-isocenter technique leads to inhomogeneous plans that increase the disparity between peripheral and maximal radiation doses inside the PTV. Proper arrangement of optimal isocenter and collimator size to achieve conformity in complex lesions is an art. Practitioners may take hours to be satisfied with a plan. Although modern software has automated simulation of the best arrangement, time and judgment by the medical physicist and medical team are still required to make the final decision. Furthermore, the ability of the delivery device to carry out the prescribed plan with multiple isocenters in a reasonable time frame is limited, which can lead to a compromise in conformality.

Shaped-Beam Techniques

Taking advantage of the ingenious idea of the collimator, the multileaf collimator was developed. This instrument consists of multiple leaves that move independent of each other and are capable of generating any format or size of collimator within the field limits of the LINAC. Basically, one piece of instrumentation allows infinite combinations of apertures of each leaf, thereby allowing modulation of the radiation dose according to the positioning of each beam and adaptation of the aperture to different shapes according to the beam’s eye view (BEV). Dynamic shaping is also evaluated by the BEV capability in the planning software.⁹¹ The BEV capability allows a three-dimensional perspective of the trajectory of the photon radiation directed by an arc or a beam. It is a practical approach to avoid undesirable delivery of radiation to eloquent structures. Initial evaluation of the radiosurgery plan is accomplished by direct visualization of the IDLs displayed on a given imaging modality (magnetic resonance imaging, computed tomography, positron emission tomography). The BEV helps refine evaluation of the entrance and exit trajectories of the photon beam. However, another important tool can be used for this purpose: the dose-volume histogram (DVH) (Fig. 255-6).

FIGURE 255-6 Two dose-volume histograms (DVHs) from a residual clival chordoma radiosurgery plan. The prescription dose was 17 Gy to the 90% isodose line. The y-axis represents the volumetric percentage of the given structure, and the x-axis represents the percentage of the radiation dose delivered. A, DVH of the prescribed target volume (PTV). The dotted line shows that 98% of the PTV is covered by 90% of the total radiation dose (17 Gy in this example), whereas the dashed line shows that 90% of the PTV is covered by 95% of the total dose (17.94 Gy). GTV, gross tumor volume. B, DVH of the brainstem. The double vertical line shows that zero volume of brainstem is receiving 100% of the radiation dose. The dotted line shows that 50% of the brainstem volume is receiving 10% of the total radiation dose (1.89 Gy). C, Sagittal T1-weighted gadolinium-enhanced magnetic resonance image showing the structures to be preserved when treating this clival chordoma.

The DVH is the most objective informative tool for evaluating a radiosurgery plan. It offers a two-dimensional graphic representation of the full range of radiation doses delivered to volumetric percentages of a given structure. This structure can be the PTV or any other structure relevant to the planning. The DVH can be displayed in two formats: differential or cumulative. The radiation dose is represented on the x-axis and the volumetric percentage of the structure on the y-axis. The differential format shows histograms in which the height of each bin reflects the percentage of volume receiving a given dose within that range. The cumulative format displays the cumulative frequencies of the percent volume receiving an equal or greater dose of the radiation dose projected on the x-axis. The “ideal” DVH of a PTV shows a perfect horizontal line parallel to the x-axis denoting the entire PTV receiving 100% of the normalized radiation dose. The vertical line parallel to the y-axis reflects the precise and steep falloff denoting the absence of hot spots within the PTV. The ideal PTV curve for an eloquent structure to be spared would be the inverse-square curve described for the PTV. The vertical line would overlap on the y-axis, and therefore no volume of the eloquent structure would receive any radiation. The horizontal line would overlap on the x-axis, thus reinforcing the statement of the vertical portion of the square that no eloquent structure volume was receiving 10%, 20%, 30%, 90%, or 100% of the prescription dose. Obviously, the “ideal” DVH does not exist in clinical work (see Fig. 255-6).

Shaped-Beam Linear Accelerator Radiosurgery

Shaped-beam radiosurgery is possible with LINACs equipped with a multileaf collimator. Such a collimator automatically changes the shape of the radiation beam, depending on the information obtained by the BEV representation of the lesion. This is possible because of the availability of three-dimensional reconstruction of the target volume in modern software for radiosurgery planning. Two approaches for shaped-beam radiosurgery are possible today in LINAC radiosurgery.

Dynamic Arcs

Once the appropriate set of arcs is made, the aperture in the micro-multileaf collimator is defined in accordance with the shape and volume of the target.¹⁹

Selection of the arcs should be determined by the extension (starting and ending angles) of the gantry rotation, or the angle between the arcs and the trajectory of the photons delivered by the arc. An arc is defined as a “brush” of the LINAC radiation source governed by the angulations of the gantry in relation to a fixed couch position. Both the extension and the position of arcs within the stereotactic space are defined according to the constraints imposed by the proximity of the target to eloquent structures. Selection of the interval between arcs is also guided by these same constraints, but there is one additional physical limitation. Placing the arcs too close to each other or enlarging the aperture of the collimator will cause an increased overlap of the partial fields of the arcs. This overlap will lead to a severe increase in the intermediate volume (volume of normal tissue receiving a considerable dose of radiation),^19,40 thus defeating the purpose of radiosurgery. Moreover, as the target gets larger and consequently the overlap between partial fields of the arcs increases, the final lesion and normal tissue volume receiving a substantial dose of radiation resembles a circle. In this circumstance it is also clear that sparing of eloquent structures by avoidance of the use of some arcs is no longer effective.

Dynamic arcs offer the advantage of faster delivery of radiation than possible with static beams or multiple-isocenter techniques. The disadvantage, however, is that the intermediate volume may be increased because of the increase in lesion volume. The shaping capability is lost, and the pattern of the final dose distribution approaches a round profile. The capability of avoiding photons crossing eloquent structures is also impaired. Therefore, large, irregularly shaped lesions amenable to fractionated shaped-beam radiation therapy are preferably planned with a static-beam technique with or without intensity modulation of the beam (IMRS/intensity-modulated radiotherapy [IMRT]).

Static Beams

The same principle of dynamic arcs applies with static beams, but delivery of radiation is not continuous throughout the entire movement of the arc. There are precalculated positions within each arc in which the beam is turned on. Planning is somewhat more time and effort consuming and delivery of radiation is also slightly longer, but this technique is preferable for lesions intermingled with multiple eloquent structures and for large complex-shaped lesions. One clinical example of the static-beam planning application is for a skull base chordoma. Although labeled as histologically “benign,” chordomas often recur and are known to be resistant to radiation. A clival location is the most common, which means that photons should not cross the optic apparatus, the brainstem, or the pituitary gland/stalk in their trajectory to the target. Basically, the radiation has to be as thoughtfully distributed as a surgeon’s manual dissection to reach the tumor. This is not accomplished with the use of bipolar, aspiration, or microneurosurgery instruments but rather with the use of an exquisite arrangement of the radiation source apertures and tridimensional beam architecture, including modulation of the photon beam, as well as understanding of the tolerance of structures to the radiation dose prescribed (see Fig. 255-5).

Advanced Methods of Conformality

Despite enormous development of the techniques of LINAC conformal radiation therapy, including multiple isocenters,^43,92 shaped beams,⁴⁰ and pencil beam approaches,³⁴ afforded by computerized imaging, three-dimensional treatment planning software, and fast-delivery LINACs, the need to further improve sparing of normal tissue and enhance the dose of radiation to the pathology remains important (Table 255-1),^93,94 as shown by the less than optimal results of radiosurgery when treating several pathologies.^95-98

A radiation dose response of tumors and AVMs has been demonstrated repeatedly.^61,99-103 Delivery of higher doses of radiation is associated with a higher probability of tumor and AVM control and cure, although accompanied by increased risk for complications involving normal tissue.^101,104,105 IMRT and IMRS are strategies to enhance the efficacy of delivery of radiation in that they potentially allow higher control and cure rates while simultaneously reducing morbidity. IMRT/IMRS is now routinely used for cranial and extracranial applications, including spine radiosurgery.^26,106-108

Beam Intensity Modulation

IMRT is relatively new, although the concept of deliberately designing a nonuniform radiation beam to achieve a superior dose distribution is not. Modulation of a heavy particle beam with a variable column of water or plastic that models the lesion in the beam pathway allowed placement of the Bragg peak inside tumors and vascular malformations since the early 1950s.^109-111 However, these techniques are not effective for modulating photon beans.

Several techniques have been proposed and used in an attempt to improve the distribution of the radiation dose when a photon beam is used.¹¹² Physical compensators have been used in limited fashion over the past 4 decades,^113-115 and physical wedges are still used routinely in most radiotherapy centers. Although these techniques can provide a more uniform target dose, they in no way provide the optimal solution that IMRT can provide with respect to avoidance of important normal structures that cannot otherwise be excluded from a radiation field.

Anders Brahme first described the modern IMRT concepts. In 1988 he proposed that the conventional trial-and-error paradigm for treatment planning be reversed and that one derive the optimum beam intensities from the desired dose distribution by using deterministic techniques.¹¹⁶ Since that time, several methods have been developed in both planning and delivery technology to allow the optimal intensity to be delivered. With the advent of micro-multileaf collimators, it is possible to perform IMRS. The success of IMRS hinges on the development and implementation of three key components: (1) inverse planning—or calculation of the optimal beam profiles given the physical or biologic constraints (or both) to a target and organs at risk; (2) leaf sequencing—or conversion of the beam profiles calculated by the inverse algorithm into a series of leaf positions and corresponding monitor units that are physically deliverable; and (3) delivery—or administration of IMRS with a tightly integrated accelerator and a multileaf collimator.

Inverse-Planning Techniques

Inverse planning is an optimization process whereby one specifies a desired dose distribution and searches for the beam intensity distribution that will satisfy the request. This is generally accomplished with an objective function that is subsequently minimized through a mathematical operation. In theory and practice, there are a number of functions, both physically and biologically based, that can be used as the objective function. The physical method called the dynamically penalized maximum likelihood (DPL) algorithm has been integrated into commercial treatment planning systems for inverse planning of intracranial targets. A key strength of the DPL approach is the ability to compute a number of inverse plans simultaneously. This allows, for example, varying levels of emphasis to be placed on the target and organs at risk, with the neurosurgeon selecting the appropriate plan for the individual tumor site and patient on the basis of the DVH and dose distribution information (Fig. 255-7).

FIGURE 255-7 A, Dose-volume histogram (DVH) of a giant arteriovenous malformation (AVM) showing that at least 90% of the AVM volume is receiving the prescribed dose, with the minimal dose to the lesion being 5.18 Gy and the maximal dose to the lesion being 7.02 Gy per fraction. PTV, prescribed tumor volume. B and C, Axial views of a T2-weighted magnetic resonance image at different levels of the AVM (volume, 142 cm³) treated with hypofractionated stereotactic radiotherapy (HSRT) consisting of five fractions of 6 Gy. HSRT was chosen instead of radiosurgery because of the large AVM volume, which precludes safe delivery of an effective single dose of radiation. Notice the areas in red indicated by the arrow, which represent hot spots (B). D, Static intensity-modulated radiotherapy (IMRT) beam. E, Notice the cold area (arrow) in the direction of the optic chiasm in the shaded IMRT shaped field. The intensity modulation profile shown in E is derived from the beam’s eye view of this static IMRT beam (yellow in the beam in D). F, DVH of the optic chiasm showing minimal dose to the structure selected to be avoided in this inverse plan.

Please see the sections “Leaf Sequencing” and “Intensity-Modulated Delivery” on Expert Consult.

Temporal Modulation

The development of fast-delivery LINACs capable of multiple energy modes is changing the concept of intensity modulation to temporal modulation by blocking the beam. The use of compensators, wedges, dynamic asymmetric jaws, multileaf collimators, and scanning beams is targeted toward that process through the use of blocking techniques.

Another and a more advanced method of achieving modulation across the radiation field is temporal modulation. The pioneers of the early concepts of MLC IMRT, which is also a type of temporal modulation, were Proimos and Takahashi.^120,121 Dynamic treatments were planned and delivered by these groups with the first prototypes of MLC devices.¹²² The latest advancement in beam-modulating techniques is volumetric radiotherapy technology, which can result in comparable or improved dose conformity while significantly shortening treatment times. RapidArc is an example of volumetric arc therapy that delivers a radiation dose with a single or several gantry rotations of the LINAC. It is made possible by a treatment-planning algorithm that simultaneously changes three parameters during treatment: speed of rotation of the gantry, shape of the treatment aperture with a multileaf collimator, and the radiation dose rate. This allows delivery of treatment several times faster than with other dynamic treatments. The short treatment time allows improved care of patients (Fig. 255-8).

FIGURE 255-8 Representation of a modern linear accelerator capable of temporal intensity modulation by using several energy levels, multiple delivery speed, beam shaping, and modulation.

Frameless Radiosurgery and the Spine

The overall accuracy achievable with frame-based cranial radiosurgery is typically recognized to be 2 to 3 mm in actual patient treatments.³³ This kind of accuracy has been obtained with frame-based immobilization in which a coordinate system is also defined for treatment planning. However, the use of frame-based immobilization and localization has shortcomings. Frame placement is an invasive procedure and a significantly more complex process for extracranial treatments that inhibits conventional fractionation.¹²³ Frameless, image-guided radiosurgery overcomes these limitations (Fig. 255-9). However, because the patient is not immobilized with a rigid frame, patient movement is more probable. Therefore, when the goal is to achieve high accuracy, it is important to characterize the patient’s movement during image-guided radiosurgery because even small movements of the patient or organ (or both) can significantly compromise radiation delivery and therefore the clinical outcome.

FIGURE 255-9 Modern radiosurgery-dedicated linear accelerator with a capability of image guidance based on infrared triangulation (A), oblique x-rays for fusion of images and checking precision (B), cone computed tomography during treatment (C), and a robotic table with 6 degrees of freedom (D).

In recent years, LINAC-delivered single-fraction radiosurgery has become a treatment option for some patients with spinal and spine-related tumors. The technique is limited, however, by the tolerance of normal cord and the ability to deliver focused radiation accurately to the lesion. Highly conformal delivery methods such as multiple circularly collimated beams, dynamic conformal arcs, and IMRS, as outlined earlier, have been used to achieve the prescribed tumor dose while maintaining the dose to the cord at or below tolerance.^26,124,125 Such delivery techniques in combination with accurate target localization and immobilization are critical for successful treatment of lesions near the cord.

Image guidance with either radiopaque fiducials implanted in the vertebrae or direct imaging of vertebral anatomy has been used to localize spinal anatomy and associated tumors.^36,126,127 Although the extracranial stereotactic radiosurgery frame performs both immobilization and localization, image-guidance techniques require ancillary immobilization. This is generally accomplished by using moldable cushions with the patient lying in the supine position to reduce target motion as a result of respiration (Fig. 255-10). Yin and coworkers observed less than 1 mm of respiratory-induced motion in vertebral bodies during fluoroscopic studies of patients lying in the supine position.¹²⁷ However, spinal anatomy may move more than 2 mm during the delivery of radiosurgery.¹²⁸ Agazaryan and associates observed vertebral anatomy movements that vary as much as 3 mm and could occur in as little as 5 minutes. These results suggest a need for intrafraction patient monitoring and correctional shifts, even for patients whose overall treatment times are expected to be relatively short.¹²⁸

FIGURE 255-10 Patient immobilization for stereotactic radiosurgery on a spinal lesion. A beanbag (A) is used to mold to the patient’s shape in the supine position to allow repeated applications. Plastic wrap (B) is used for additional immobilization of the patient to allow minimal movement and thus minimal adjustments at each beam or arc delivery. A vacuum device (C) is used to complete adjustment of the plastic wrap and beanbag. A robotic table (D) with 6 degrees of freedom is used for correction of patient position before beam or arc delivery.

Spine radiosurgery for primary benign, primary malignant, and metastatic tumors has the potential to improve local control by escalation of the effective radiation dose while minimizing the risk for spinal cord injury. An essential requirement for spine radiosurgery is the ability to accurately determine the spinal cord dose associated with a planned treatment so that an overdose to the cord can be avoided. Additionally, our understanding of the tolerance of the spinal cord to radiosurgery continues to evolve, and an accurate determination of the spinal cord dose associated with the treatment is required. Besides small systematic errors associated with positioning uncertainty during spine radiosurgery,¹²⁹ random errors associated with patient movement create uncertainty regarding the actual radiation dose that the spinal cord receives. Published series of spine radiosurgery have explored and quantified intrafraction patient movement during spine radiosurgery, although additional data are needed.^25,129,130 These data indicate that significant movement can occur during treatment. Separately, some groups have examined the uncertainty of the spinal cord dose associated with simulated patient positioning errors and demonstrated the need for an accurate understanding of uncertainty in setup and movement.^131,132 Several cases of myelitis have been reported after spine radiosurgery.^133-135 In the published accounts, the spinal cord doses appear to have been within limits widely accepted as being safe for almost all cases. However, only the planned cord dose could be reported, which ignores the possibility of intrafraction patient movement. It would be greatly beneficial to our understanding of spinal cord tolerance if a study were performed in which online measurements of patient movement were taken into account in providing an estimate of the actual dose delivered to the spinal cord. The measured patient translations and rotations can be simulated in the planning system after the treatment to determine the actual dose delivered to the spinal cord. Later, these data could be analyzed in terms of complications. Such monitoring capabilities do exist with modern radiosurgery systems (see Fig. 255-9).

Conclusion

SRS has evolved beyond the boundaries of the imagination of its creator Dr. Lars Leksell. The evolution imposed by the versatility of LINAC radiosurgery has stimulated the interest of medical physicists and radiation oncologists who use stereotactic techniques in their day-to-day work. The frame, a neurosurgeons’ tool, is becoming progressively obsolete with advances in the initially frameless stereotactic technique, now simply called image-guided radiosurgery or radiation therapy. Image guidance left the boundaries of neurosurgery to become the generic term precision radiation therapy, which is being applied to pathologies beyond the scope of neurosurgery. Within neurosurgery, the term stereotactic radiosurgery persists and should be applied to cranial and spine surgery. Applications of interest to neurosurgeons include all malignances, primary and metastatic, and all benign tumors and vascular malformations involving the central and peripheral nervous system. Additionally, functional applications encompassing pain control, epilepsy, movement disorders, and selected psychiatric diseases persist, but on a much smaller scale.