A simple technique for craniospinal radiotherapy in the supine position

Transcription

1 Radiotherapy and Oncology 78 (2006) Craniospinal radiotherapy A simple technique for craniospinal radiotherapy in the supine position William A. Parker a, *, Carolyn R. Freeman b a Department of Medical Physics, and b Department of Radiation Oncology, McGill University Health Centre, Montreal, Canada Abstract Purpose: Craniospinal irradiation poses technical difficulties that may be addressed with the use of the newer technologies that have become available over the past decade. The use of CT simulation allows improved target localisation and beam geometry definition while significantly reducing the treatment simulation time. We have developed a CT-based technique for whole CNS irradiation in the supine position that uses fixed field parameters, asymmetric jaws for field matching and drastically reduces simulation and treatment times. Methods: The patient is CT scanned and treated in the supine position. The clinical target volume and relevant critical structures are outlined on a planning CT scan. Half beam blocked lateral fields with a collimator rotation are used to match the beam divergence from the superior border of the spinal field at the C2 vertebral body. The shielding for the cranial fields is generated automatically, and the dose distribution is calculated using a 3D treatment planning system. Fixed field parameters are used for the planning and treatment. The position of the isocenter of the spine field is always a fixed longitudinal distance from the isocenter of the brain fields. If multiple posterior fields are required, the isocenter of the second spine field is always a fixed longitudinal distance from that of the first and the gap between the fields is determined using virtual simulation and feathered during treatment using the asymmetric jaws of the linear accelerator. All treatment portals are filmed daily during the first week of treatment, and after each junction change thereafter. Results and conclusion: The supine position provides numerous advantages. Patients are more comfortable, the treatment position is more reproducible, and access to the airway is possible, if necessary, for patient sedation. The use of CT simulation decreases the simulation time, allows for increased planning accuracy, and enables the use of multimodality image registration, and 3D treatment planning. The use of asymmetric jaws allows for junction feathering without changing the patient setup or using a couch angle. q 2005 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 78 (2006) Keywords: Craniospinal irradiation; Supine position; Medulloblastoma; Radiotherapy Craniospinal irradiation (CSI) is technically challenging because of the need to cover a complex clinical target volume (CTV) [5,6] that includes the whole brain and the whole length of the spinal axis and the covering meninges. In most radiotherapy departments, CSI is delivered to patients in the prone position using lateral opposed fields covering the whole brain and upper cervical spine matched to a direct posterior field that extends inferiorly to cover the caudal extent of the thecal sac [15]. Because many patients who require CSI are young children (less than 6 7 years of age) who may require anesthesia (with intubation), we have adapted our CSI technique to treat patients in the supine position. In contrast to other techniques described over recent years [2,4,11,12,16], the technique that we developed in our department uses fixed field parameters, requires only longitudinal couch motions, and is simple to plan and easy to incorporate into the workload of a busy radiotherapy department. A number of issues need to be considered concerning beam geometry and field matching. One relates to the junction between the brain and spinal fields. Careful positioning of the patient and optimal placement of the junction is important to avoid inclusion of the mandible in the exit of the spinal field, and immobilization is essential to ensure reproducibility of treatment from day-to-day. The lateral fields need to be carefully matched over the cervical spinal cord to avoid over- or under-dosage in this region and most centers will move (feather) the junction during treatment [14]. Many will use a couch rotation as well which further adds to the complexity of treatment. An extended SSD or second posterior field will be required if the length of the spinal field is excessive and the geometry at /$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi: /j.radonc

2 218 Whole CNS irradiation this junction will differ from that over the cervical spinal cord. Target volume coverage is not a simple matter, especially with respect to the area of the cribriform plate. Efforts to spare the lenses to avoid cataract formation may result in under-dosage in this region and lead to treatment failure [1]. Special attention is needed as well to ensure adequate coverage of the caudal extent of the thecal sac which is not well seen on simulator radiographs, and in a significant proportion of cases extends below the classical inferior limit for CSI of S2 [3]. The use of CT simulation and multi-modality image registration with MRI greatly simplifies and improves target volume definition [7 10]. Methods Patient position and simulation Patients are simulated in the supine position using a CTsimulator. The patient s head is positioned on a custom headrest that is placed flush with a Styrofoam board that lies underneath the whole length of the patient in order to allow for maximum neck extension and avoid inclusion of the mandible in the exit of the posterior field used to treat the spine. An anterior CT scout view is used to check that the spine is straight and a lateral CT scout view to verify the position of the mandible. The patient s head is immobilized using a custom thermoplastic mask. Anterior and lateral radio-opaque fiducial markers are placed on the mask at the level of the C2 vertebral body. Two scans of the patient are acquired using standard oncology CT-scanning protocols. The brain scan, which includes the entire head, down to the level of the shoulders, is taken with 3 mm slice thickness and separation. The scan of the spine, which includes the whole body contour from the mandible to the inferior limit of the S4 vertebral body, is taken with 10 mm slice thickness and separation. Both scans include the fiducial markers on the mask. Two scans are necessary because many treatment planning systems have a limitation on the number of CT-slices that can be used but in addition a finer slice thickness is required to resolve some anatomical structures in the brain. A reference isocenter, marked at the time of the CT scan at the level of the fiducial markers, is the same on both scans and typically becomes the treatment isocenter for the brain volume. Virtual simulation and treatment field definition The surface of the brain and the whole length of the spinal axis are carefully contoured on every CT slice and defines the CTV. CT-MRI image registration using postoperative diagnostic MRI studies is performed in order to define the target volume for the boost treatment which is typically given after the completion of CSI. The reference isocenter for the brain fields is defined at mid-plane and mid-line at the level of the C2 vertebral body. Fig. 1. Schematic diagram of the CSI technique as applied to a supine patient.

3 W.A. Parker, C.R. Freeman / Radiotherapy and Oncology 78 (2006) Fixed field parameters are used as shown diagrammatically in Fig. 1. The lateral fields used to treat the brain are fixed in length at 20 cm (Y 2 Z20 cm), with the length of the field asymmetrically half-blocked (Y 1 Z0 cm) at the isocenter in order to provide a non-divergent junction with the posterior spine field obviating the need for any couch rotations. The treatment isocenter is typically defined at the reference isocenter location but may be defined elsewhere if the patient anatomy demands it. For example, the maximum half-blocked field length of 20 cm might not cover the entire brain and meninges with an adequate margin for some adult patients, in particular if the patient s neck is not fully extended. The principal factors to be considered with respect to with the location of the isocenter however, are that the CTV is adequately covered, that the lateral beams do not enter the patient through the shoulders, that there is enough neck inferior to the isocenter to allow for junction feathering, and that the junction is positioned such as to minimize the exit dose through the thyroid, and to avoid having the spine field exit through the mandible. A collimator rotation of 118 is used to match the inferior limit of the brain fields with the divergent superior border of the posterior spine field. The spine field isocenter is always located at a point 20 cm distal to the brain field isocenter, and the length of the superior portion of the posterior field is initially fixed at 20 cm, hence the collimator rotation of 118 for the lateral brain fields. If a second spinal field is required, the isocenter of the inferior spinal field is always 30 cm distal to the isocenter of the superior field as shown in Fig. 1. The length of the superior portion of the inferior spine field is adjusted with asymmetric jaws in order to match the inferior limit of the superior spine field at the depth of the posterior surface of the vertebral body at that level. The length of the inferior portion of the inferior spine field is adjusted with asymmetric jaws to cover the caudal extent of the thecal sac as determined on diagnostic MRI or CT-based digitally reconstructed radiographs (DRR). Since all of the fields used to treat the patients are at a common source-axis distance (SAD), only longitudinal motion of the couch is ever required. The technique is illustrated for a typical patient on a sagittal CT reconstruction in Fig. 2. The width of the brain fields is adjusted using the asymmetric jaws. Shielding for the cranial fields is generated automatically with an auto-blocking function available with the virtual simulation software. The shielding is designed such that a 10 mm margin exists between the CTV and the blocks. This allows for patient setup error, patient motion, and ensures that the CTV is covered by at least the 95% isodose line when the plan is normalized to 100% at the center of the whole brain field. Lead alloy blocks are favored over multi-leaf collimation (MLC) for shielding in order to provide better field definition and avoid the leakage/transmission dose through the MLC. The width of the spine field(s) is adjusted using asymmetric jaws to cover the neural foramina and the field may be shaped appropriately by using MLC or mounted blocks. Treatment planning and dose calculation Following virtual simulation, the data are transferred to a 3D treatment planning system and the dose distributions are calculated. The lateral brain fields do not usually require any beam modification. If necessary, the posterior spine field(s) may be treated using simple intensity modulation to provide dose compensation. Each posterior field thus consists of several sub-field segments that simply boost cold regions of the distribution and are automatically delivered using the dynamic MLC mode of the linear accelerator in a step-andshoot fashion. Treatment Patients are treated with 6 MV photons on one of several linear accelerators available at our facility with asymmetric jaws and 120-leaf MLCs with dynamic beam delivery capabilities. The entire treatment sequence is transferred automatically to a record and verify system. Since Fig. 2. Sagittal multi-planar reconstruction of a patient treated in the supine position with the CSI technique.

4 220 Whole CNS irradiation the location of the spine field isocenter is always 20 cm distal to the brain field isocenter and both are treated at the same source-axis distance (SAD), the couch need only be moved longitudinally for treatment of the spine field and controlled by using the digital readout of the accelerator. The same applies if a second spinal field is required; in this circumstance the couch is moved longitudinally to a distance 30 cm inferior to the isocenter of the superior spinal field. The longitudinal shift distance is always double checked with the use of a ruler after each couch motion. Since the patient is treated in the supine position, the posterior fields are difficult to visualize on the patient. The setup is therefore based on anterior setup fields with the same isocenters and field dimensions as the treatment fields. Gaps between the posterior fields can be assessed on the anterior setup fields. These gaps can be accurately determined from reconstructed sagittal CT images obtained from the treatment planning system, and verified by a quick hand calculation (see Fig. 3). The gaps are verified on a daily basis on the patient with the use of a ruler. The junctions between the fields are feathered during the course of treatment by using the asymmetric jaws. The jaw defining the inferior limit of the brain fields is opened by 1 cm every 9 Gy, while the superior limit of the spinal field abutting the brain fields is decreased by 1 cm. The small change in field size presents a negligible difference in the divergence of the beams, and the collimator angle of the brain fields remains unchanged at 118. If a second spine field is required, the inferior limit of the superior field is decreased by 1 cm, and the superior limit of the inferior field is increased by 1 cm. Portal imaging All fields are imaged on a daily basis during the first week of treatment, and once a week (following each junction change) thereafter. The portal images are compared to simulation DRRs. Quality assurance Although the integrity of the cranial spinal junction depends on the patient position, the accuracy of treatment is critically dependent on several mechanical tolerances of the treatment unit, namely field size, collimator rotation, collimator setting, and longitudinal couch movement. Before each patient begins a course of treatment, a film (within a phantom) is taken at the junction between Fig. 3. Sagittal multi-planar reconstruction of a patient treated in the supine position with the CSI technique. The anterior fields are shown and the gap can be measured directly from the image.

5 W.A. Parker, C.R. Freeman / Radiotherapy and Oncology 78 (2006) Fig. 4. The film in the figure is taken in the sagittal plane. The film verifies that couch motions, collimator rotations and field sizes are precisely defined, and that the chosen parameters provide a geometrically correct junction. the lateral brain and posterior spine fields to confirm the geometric and mechanical accuracy. The film is placed perpendicular to the axis of the lateral brain fields (longitudinally along the axis of the spine field) at the level of the cranial fields isocenter (Fig. 4). If multiple spine fields are used, a film may also be used to verify gap calculations and beam matching depths of the spine fields. Quality control procedures for reproducibility of collimator settings at our clinic have shown the positional accuracy of the jaws to be within G1 mm. In any event, the patient is protected from any gross over or under dosing through the feathering of the junctions throughout the treatment. Discussion Target volume definition, sparing of critical normal structures, dose homogeneity, and junctioning have all been problematic in CSI and are especially critical in pediatric cases. These technical difficulties may be addressed utilizing the newer technologies now becoming widely available for radiotherapy treatment planning and delivery. In our experience the use of CT simulation with multi-modality image registration in preparation for CSI results in better definition of the target volume for the brain fields such that coverage of the region of the cribriform plate can be assured and the amount of bone and soft tissue included in the field over the middle cranial fossa minimised [7]. At the McGill University Health Centre, we have used CT simulation for craniospinal irradiation since In 1998 we modified our technique to treat patients whenever possible in the supine position. The amount of time required at the time of CT simulation has been drastically reduced from up to 2 h (with conventional simulation) to approximately 30 min. The supine technique is much better tolerated and more stable than the prone technique and as a consequence results in a reduced daily treatment time of the order of 15 min. The anatomic information available from CT simulation ensures adequate target coverage while maximizing normal tissue sparing as well as providing a platform for 3D dose calculation. The use of fixed field parameters greatly facilitates treatment planning and significantly reduces the amount of time that the patient is on the CT simulator table because the planning itself is performed after the patient has left the department. This is not a minor point for young patients who have recently undergone major surgery and are often traumatized and uncomfortable. The problem of junctioning non-coplanar fields over the cervical spinal cord is not trivial and many different solutions have been proposed. Our approach that uses collimator rotation and half beam blocking with asymmetric jaws for the brain fields and fixed field geometry is a practical solution to the problem that effectively eliminates hot or cold spots at the junction. Although the use of feathering, even for traditional techniques, has been questioned by others [13], it is our practice to feather the junction during treatment and this is accomplished by simply moving the appropriate jaws independently. When the length of the spinal field exceeds 40 cm, a second posterior field will be required. The beam junction between the two spinal fields is precisely determined using virtual simulation using sagittal CT reconstructions to determine the ideal junction position and to determine the skin gap (on the setup fields). As others have shown [2,9,11], the use of fixed field parameters and verification using anterior setup marks effectively addresses concerns with respect to verification of field placement. Because the spine in patients treated supine is usually almost parallel to the treatment couch a missing tissue compensator is less often required, but if dose compensation is needed the use of a simple, few segment, step-and-shoot IMRT technique greatly reduces the treatment planning time as compared to a static compensator. Automatic delivery by the accelerator makes the process seamless at treatment time and avoids errors. Although our technique is based on CT simulation, the same concepts can be applied using conventional simulation. Patients can be simulated using a fixed beam geometry that will reduce time taken for simulation; adjustments can be performed later using the asymmetric jaws. Of course, the use of conventional simulation precludes the use of CT data and CT-MRI image registration for volume delineation or 3D dose calculation. * Corresponding author. William A. Parker, Department of Medical Physics, McGill University Health Centre, Montreal General Hospital, 1650 Cedar Avenue, Room L5-129, Montreal, Que., Canada H3G 1A4. address: william@medphys.mcgill.ca Received 1 October 2004; received in revised form 17 October 2005; accepted 10 November 2005; available online 5 December 2005

6 222 Whole CNS irradiation References [1] Carrie C, Hoffstetter S, Gomez F, et al. Impact of targeting deviations on outcome in medulloblastoma: study of the French society of pediatric oncology (SFOP). Int J Radiat Oncol Biol Phys 1999;45: [2] Chang EL, Wong P-F, Forster KM, et al. Verification techniques and dose distribution for computed tomographic planned supine craniospinal radiation therapy. Med Dosim 2003;28: [3] Dunbar SF, Barnes PD, Tarbell NJ. Radiologic determination of the caudal border of the spinal field in cranial spinal irradiation. Int J Radiat Oncol Biol Phys 1993;26: [4] Hawkins RB. A simple method of radiation treatment of craniospinal fields with the patient supine. Int J Radiat Oncol Biol Phys 2001;49: [5] International Commission On Radiation Units Report 50, Bethesda, MD; [6] International Commission on Radiation Units Report 62 (Supplement to Report 50), Bethesda, MD; [7] Liu M, Carrie C, Parker W, et al. Comparison of computergenerated and customized blocks for whole brain fields used in the treatment of medulloblastoma. Med Pediatr Oncol 2002;38: [8] Mah K, Danjoux C, Manship S, et al. Computed tomographic simulation of craniospinal fields in pediatric patients: improved treatment accuracy and patient comfort. Int J Radiat Oncol Biol Phys 1998;41: [9] Michalski JM, Klein EE, Gerber R. Method to plan, administer, and verify supine craniospinal irradiation. J Appl Clin Med Phys 2002;3: [10] Parker W, Tsien C, Gingras C, et al. 3D treatment planning for pediatric whole CNS irradiation using CT simulation and intensity modulated radiation therapy. Int J Radiat Oncol Biol Phys 1999;45:2286. [11] Rades D, Holtzhauer R, Baumann R, et al. Craniospinal axis irradiation in children. Treatment in supine position including field verification as a prerequisite for anesthesia without intubation. Strahlenther Onkol 1999;175: [12] Thomadsen B, Mehta M, Howard S, et al. Craniospinal treatment with the patient supine. Med Dosim 2003;28:35 8. [13] Tinkler SD, Lucraft HH. Are moving junctions in craniospinal irradiation for medulloblastoma really necessary? Br J Radiol 1995;68: [14] Urie M, FitzGerald TJ, Followill D, et al. Current calibration, treatment, and treatment planning techniques among institutions participating in the Children s Oncology Group. Int J Radiat Oncol Biol Phys 2003;55: [15] Van Dyk J, Jenkin RD, Leung PM, et al. Medulloblastoma: treatment technique and radiation dosimetry. Int J Radiat Oncol Biol Phys 1977;2: [16] Verellen D, Van Den Heuvel F, De Neve W, et al. Dynamic radiotherapy: interactive movement of patient couch for treatment of craniospinal axis. Int J Radiat Oncol Biol Phys 1996;35:771 7.