Introduction
In 1951 Lars Leksell coined the term stereotactic radiosurgery (SRS) [1]. A ceaseless innovator, his goal was to develop a method for ”the non-invasive destruction of intracranial . . . lesions that may be inaccessible or unsuitable for open surgery” [2]. The first procedures were done using an orthovoltage X-ray tube mounted on an early model of what is now known as the Leksell stereotactic frame, for the treatment of several patients with trigeminal neuralgia. After experimenting with particle beams and linear accelerators, Lek-sell and his colleagues ultimately designed the gamma knife, containing 179 cobalt sources in a hemispheric array (Figure 1). The first unit was operational in 1968 [3].
At the same time, work was continuing elsewhere with focussed heavy particle irradiation. Raymond Kjellberg spearheaded the use of proton beam treatments at the Harvard/Massachusetts General Hospital facility. A series of patients with arteriovenous malformations and pituitary tumors was amassed. Similar efforts were carried out in California (with helium ions) and Moscow [3]. Particle beams have the advantage of depositing their energy at a distinct point known as the Bragg peak, with minimal exit dose.
Figure 1 Schematic view of gamma knife.
In practice, the beams must be carefully shaped and spread in order to treat patients with intracranial lesions (Figure 2). The expense of building and maintaining a cyclotron has limited the use of heavy particle SRS to a few centers. (From Ref. 57 with permission of The McGraw-Hill Companies.)
Figure 2 Spreading and shaping of particle beam for SRS with a particle beam accelerator.
Through the 1970s SRS was used to treat intracranial targets that could be defined by plain X-rays or tomograms (e.g. acoustic neuromas arterio-venous malformations (AVMs), or the gasserian ganglion). The advent of computed tomography (CT) in the mid-70s opened up the possibility of direct targeting of tumors and other ”soft tissue” targets inside the skull. As the potential horizons of SRS broadened other investigators were able to adapt linear accelerators (linacs) for SRS (Figure 3). These devices were more available (and less expensive) than gamma knives or heavy particle accelerators [4-7]. As clinical experience increased, publications appeared, indications broadened, and vendors became increasingly interested. A debate emerged regarding the merits of the gamma knife versus linac based SRS. Clinical and physics studies seemed to have settled the issue that SRS could be delivered effectively and accurately with either method [8-10].
Figure 3 Diagram of linear accelerator, showing axes of rotation of couch and gantry.
Currently, SRS is a routine part of neurosurgical practice, and should be part of resident education. Below is an overview of the current indications and technical considerations for SRS.
Technical Considerations
The Treatment Team
SRS, however it is delivered and planned, has a final common pathway— the delivery of ionizing radiation to a well-defined, relatively small volume. Many factors go into the safe and effective use of this technique, not the least of which is a multidisciplinary team. Besides the neurosurgeon, the active involvement of a medical physicist is required. This is the only person with the expertise to translate the virtual computer based SRS plan into the physical reality of irradiation. Quality assurance (QA) of the SRS device must also be maintained by the physicist; otherwise the plans will not match the actual treatment.
In the U.S. only radiation oncologists are permitted to sign off on a plan of therapeutic irradiation. However, the oncologist has much more to offer than a clerical role. Many patients undergoing SRS may have received fractionated radiation therapy in the past, or may need it in the future. They may have types of cancers that neurosurgeons have little experience with. Furthermore, radiation oncologists view SRS as a form of focussed radiation therapy as opposed to a form of "minimally invasive neurosurgery.” Radiation oncologists can bring 100 years of radiation therapy experience to SRS planning and treatment.
Other staff members are of critical importance in the performance of SRS. A dedicated nurse, most often from the radiation oncology side, will ensure that equipment and any needed medication is prepared. He or she will sheperd the patient through the day, ensuring that discomfort is addressed as needed. Dosimetrists’, an invaluable help, may be actively involved in treatment planning in a busy department where the physicists must attend to other duties. The radiation therapists must be familiar with the treatment equipment and understand the QA needs of the device. Their professionalism is crucial in ensuring a smooth experience for the patient.
Imaging
SRS is based on imaging. The radiation must be aimed at a specific target seen on CT and/or MRI. Some commercially available devices (e.g. Radionics X-Knife®) require the use of CT scanning even if other images are used. This is due to the presumed greater stereotactic accuracy of CT scanning [11]; however, other authors have demonstrated that MRI can be an accurate tool for radiosurgical targeting and use this method exclusively [12]. A stereotactic phantom should be scanned and the accuracy of the CT and MRI units verified. MRI is needed to adequately target lesions at the skull base and in the posterior fossa. Coronal MRI is of particular value to image the cerebral convexity and the optic chiasm.
Image fusion, available with most commercial systems, permits the registration of CT and MRI scans. Thus, patients can undergo MRI scanning in advance, limiting the length of time needed for scanning on the day of the procedure, and making it easier to schedule the SRS. If a question remains regarding the spatial accuracy of stereotactic MRI, image fusion can combine the reliability of CT with the improved anatomical resolution of MRI [13]. Any other digital image set can be incorporated, such as positron emission tomography (PET) or functional MRI [14]. AVMs are often best seen on contrast-enhanced CT [15]. For some patients catheter angiography is still necessary for AVM visualization. In these cases digital angiography can be used, but software to correct the spatial distortion inherent in digital X-rays must be installed and verified for accuracy.
Treatment
Patients arrive the morning of treatment, and in most institutions will be discharged at the end of the day. Oral sedation is administered (for instance Valium 5 mg) and an intravenous angiocath inserted for contrast injection. The stereotactic frame is then applied.This placement must ensure that 1) the treatment volume is above the base ring of the frame; 2) as much of the cranium as possible is included within the bounds of the localizer, so that dosimetry will be accurate; 3) there will be no obstruction to securing the frame to the scanning and treatment couches, especially in the back of the head; 4) the patient can eat and drink after the scan; 5) there is no scalp pressure from any part of the apparatus. Symmetrical placement is ideal, but not essential.
Scanning is performed with CT and/or MRI. Attaching the frame to the scan table prevents patient movement and ensures that the image will be in an orthogonal plane. Contrast should be injected if this will help visualize the target. Scan slice thickness should be 3 mm and the field of view adequate to encompass the localizer rods or panels. After the scan is done the patient waits, preferably in a comfortable private area until treatment.
Treatment Planning
The stereotactic scan is transferred electronically to the SRS planning computer. Image fusion is done if necessary, and the visual fit between scans is confirmed in 3 planes. Each slice is checked. If the software requires, or allows for 3-dimensional contouring of anatomical structures, these virtual models are used only to begin the planning—in the end the plan must be based on the 2-dimensional CT and/or MR images themselves.
The two main technical goals of SRS planning are to achieve confor-mality and a steep dose gradient. Conformality means that the prescription dose of radiation will match the borders of the target as closely as possible, so that the target/volume ratio (TVR) is as close to 1:1 as possible. This number should be less than 2 and certainly not more than 3. Of course, if it is less than 1 the target will be undertreated. A steep dose gradient means that the falloff of radiation outside the target will be rapid. Thus, the volume of brain receiving lower (but still significant) amounts of radiation will be as small as possible. A steep dose gradient will also protect critical structures, e.g. the optic chiasm, allowing for SRS use to treat lesions as close as 3 mm from the chiasm. Since more radiation falloff is inevitable, this can also be aimed as needed in a safer direction. For example, in patients with acoustic neuromas, it is preferable to place additional doses in the temporal bone rather than the brainstem by orienting beams in a craniocaudal fashion [16].
If SRS is administered using circular collimators (as with the Gamma Knife® and ”conventional” linac systems), a collimator large enough to encompass the treatment volume should be chosen. As the Gamma Knife® has a maximum collimator size of 18 mm, often this will not be possible, in which case multiple treatment ”shots” will be employed to fashion an ideal plan.The maximum linac collimator size should be 40 mm, beyond which the volume treated will likely be too large for SRS. Rarely will a purely spherical treatment plan suffice, and even if a single linac collimator can be used for treatment, the beams must be angled, trimmed, or weighted in such a way as to conform to the target shape as much as possible [15]. Devices that use inverse planning such as the Peacock system or the CyberKnife® are not limited by collimator size or shape; their quality assurance may be more complicated.
When the appearance of the treatment plan is satisfactory, as confirmed in 3 planes and multiple slices of the 2-dimensional images, the prescription dose is chosen. Here the goals are to deliver a dose that is safe and effective. In general, complications of SRS are directly related to increasing dose, volume, and target location, with volume the factor best understood [17,18]. Other factors, such as prior or anticipated fractionated radiation therapy, should be considered as well.
Practically speaking, balancing safety and efficacy means that treatment doses will lie somewhere between 10 Gy and 20 Gy for patients with tumors; occasionally a higher dose can be aimed at a small metastasis. Functional radiosurgical treatments will require a higher dose (such as maximum of 70 Gy for patients with trigeminal neuralgia); the very small target volume permits this to be done safely [19]. The difference between prescription and maximum doses must be kept in mind. The use of multiple collimators with overlapping treatment spheres may result in a maximum dose much higher than the prescription dose, depending on the SRS system being used. It must be ascertained that any major areas of overlap remain within the target volume. Coordination with the medical physicist is crucial to ensure that the translation from relative isodose shells, rendered in percentages of the maximum dose, and the actual, absolute dose in Gy is accurate. For patients with malignant disease in whom other radiation is needed, input from the radiation oncologist should be especially welcome.
Treatment
When it is time to turn the virtual treatment plan on the computer screen into a physical reality of delivered radiation, the strictest QA measures must be followed. The isocentric accuracy of any SRS system must be absolutely ensured—otherwise the high dose of radiation will go somewhere other than planned. Double and triple checking of stereotactic settings by multiple personnel is desirable. The physicist is responsible for verifying that the dose output of the treatment device is predicted.
The patient should lie as comfortably as possible for the treatment to avoid attempted head movement. Head fixation with the stereotactic frame (or with the mask in case of the CyberKnife®) is mandatory. After completion of the treatment and removal of the head ring, the patient may be discharged. Clinical and imaging followup will depend on the patient’s status and the lesion treated. In general, patients with malignant disease will be seen sooner and imaged more often than those with benign tumors, AVMs, or with ”functional” disorders.
INDICATIONS FOR SRS
SRS has evolved from an esoteric treatment to an option that should be offered to many patients with intracranial lesions. The older or more medically fragile the patient, the smaller the target. The more hazardous the surgical option, the more SRS should be offered as an alternative to surgery, or as the primary treatment. Radiobiology suggests that there is a particular advantage for single-session irradiation of benign tissues [20]. Thus, certain patients with AVMs, acoustic neuromas, and small cavernous sinus meningiomas are ideal candidates for primary SRS. Ample clinical evidence exists that SRS is an effective treatment for such lesions, with a morbidity rate less than surgery. The question is where to draw the line—e.g., should all patients with acoustic neuromas be offered SRS, or should this be reserved only for those considered elderly or medically infirm? Thus, the debate is not whether SRS is an effective treatment, but rather which patients are candidates for treatment.
At this time, patient choice is considered as valid an indication as any medical reason for SRS. However, it is up to the neurosurgeon to understand when SRS is a reasonable choice before presenting it to the patient as an option. For this reason some training in SRS and an understanding of the underlying physics and radiobiology are essential; guidelines for this have been published as a consensus statement by leading neurosurgeons, radiation oncologists, and physicists [21].
Increasing experience and documentation of its efficacy has fueled the expansion of indications for SRS. In truth the increase in SRS centers and practitioners may have played a role as well. There are more patients with malignant intracranial tumors than with benign lesions. There is evidence that SRS increases survival with maintained quality of life for patients with high grade gliomas, but this is an incremental improvement at best [22]. For a patient in good clinical shape with a ”recurrent” glioma that is small enough to be targeted effectively and safely, SRS is a reasonable option. On the other hand, patients with metastatic tumors, especially single ones, may be better served by SRS compared to surgery. Tumors that are resistant to fractionated RT (e.g. melanoma) may be controlled with SRS [23].
The role of SRS in functional neurosurgery is being explored. For patients with trigeminal neuralgia, a condition where the target can be defined with great precision on MRI, SRS seems to be an effective option. For alleviation of pain or movement disorders, more uncertainty exists. Some encouraging reports have been published [24,25]. However, direct physiological target confirmation is lacking in SRS. For targets whose exact location may not be predictable with anatomic imaging alone (e.g. the Vim nucleus of the thalamus)—or more precisely, when effective lesion placement depends on physiological feedback—the best technique and indication for SRS remains to be determined.
The indications and technique of SRS with different tools and for different indications are discussed in detail in the topics that follow.
CONCLUSION
SRS is an accepted and even ubiquitous part of contemporary neurosurgical practice. For the appropriate patients and in the proper hands, it provides minimally invasive, safe, and effective treatment. Neurosurgeons performing SRS should have the necessary skills and understanding of the basic principles underlying SRS, and be involved in every step of the procedure. With judicious use, SRS will remain an excellent primary and adjuvant treatment modality for many of our patients in neurosurgery.
