Continuous localization technologies for radiotherapy delivery: Report of the American Society for Radiation Oncology Emerging Technology Committee.

Introduction

An active arena for technological advancement in radiation oncology treatment delivery has focused on the motion inherent in target structures and normal organs. With the advances over the last decade (and more so within the last few years), in intensity modulated radiation therapy (IMRT), stereotactic radiosurgery (SRS)/radiotherapy and stereotactic body radiotherapy (SBRT), and image-guided radiation therapy, it has become critical to position patients in the treatment positions precisely and reproducibly. To address these localization issues, devices have been developed that may be implanted in the organ or volume of interest and tracked during and between treatments

The goal of this report is to provide a review of nonionizing technologies designed to continuously localize the tumor, patient, or a surrogate. This report is concerned only with the completely independent (nonionizing) systems and not systems with integrated stereoscopic X-rays nor in house systems. Specifically, the 3 motion management systems that were Federal Drug Administration-approved and commercially available as of the closure date of this report (June 17, 2008) are discussed: the Calypso 4D (4-dimensional) Localization System as well as the RadioCameras (ZMed Inc/Varian Medical Systems, Ashland, MA) and the AlignRT system (VisionRTTM, London, UK).

Continuous localization systems are highly specialized technologies that require a significant quality assurance program in order to be effectively used to benefit patients. Given the scope of quality issues involved, the quality assurance program would need to address setup, testing, maintenance and interoperability of equipment, treatment planning, patient positioning, and process of care, as well as staffing, education, training, and appropriate supervision. While important, these aspects of target localization are outside the scope of this paper. Additional detail is available in the full Emerging Technology Committee report, which is available on the American Society for Radiation Oncology website (www.astro.org).

Description of the currently marketed devices

Calypso 4D localization system

The Calypso system uses an array of AC magnetic coils to generate a resonant response in implanted transponder beacons that is subsequently detected using a separate array of receiver coils. The beacons are inserted within the prostate gland under ultrasound guidance in a manner analogous to a needle biopsy, prior to computed tomographic (CT) simulation. Typically 3 are implanted, though the system can use as few as 2. Transponder coordinates on the CT scan relative to the treatment plan isocenter are determined and entered into the Calypso system. The location of the array is tracked in the treatment room using 3 ceiling-mounted infrared cameras. The system then monitors the intrafraction movement of the beacons relative to the calibrated isocenter during each treatment. Individual facilities may choose to relocalize the patient or interrupt treatment based on the observed intrafraction motion. This report focuses on the published literature as it applies to prostate localization as this represents the only currently Federal Drug Administration-approved indication.

RadioCameras

The RadioCameras system is specifically designed for intracranial radiosurgery with high-precision patient positioning. The system uses 2, two-dimensional charged couple devices that detect movement in a rigid array containing 4 infrared, passively reflective markers. The cameras are rigidly mounted in the ceiling of a treatment room and interfaced with a personal computer. The infrared markers fixed to the array are connected with the patient by rigid attachment of the array to a custom maxillary bite-block. The patient is CT-simulated with the bite-block in place and is registered in the treatment room in 3 dimensions by the passive infrared system. The camera system must be able to identify at least 3 of the 4 markers to determine the array's position and a software application will then display the displacement (with 0.1-mm precision) of the patient, along with the rotations (yaw, pitch, and roll with 0.1 degree precision) for each treatment. The system is used for frameless SRS/SBRT of the head and neck or central nervous system.

AlignRT

The AlignRT system uses a ceiling-mounted camera system to create a 3-dimensional topographic patient surface rendering that can be tracked for positioning during treatment. This is accomplished by registering the patient's unique 3D surface pattern using a system of special cameras and flashes. The 3D reference surface can be acquired during CT simulation scan or on the first day of treatment. Prior to each treatment the patient surface rendering is acquired and compared with the reference surface to make any daily changes in 6 degrees. This can also be done in a real-time continuous mode that allows for intrafraction monitoring. The data for this system have been used for breast cancer treatment.

Description of patients potentially benefiting from use of technology

It seems intuitive that the use of “continuous localization technologies” would be implemented hand-in-hand with the use of advanced radiotherapy planning and delivery techniques (eg, IMRT, SBRT, SRS, etc). The ability to conform the radiation dose distribution to the target(s) of interest while sparing surrounding normal tissues has the potential to reduce the margin for error with respect to localization uncertainty. The inability to localize appropriately may result in a geographic miss of the intended target tissues. The use of continuous localization technologies also has the potential to reduce normal tissue side effects as well as improve outcomes if able to correct for inaccurate targeting of advanced delivery techniques.

Tumors located in the pelvis, abdomen, and thorax are subject to motion during treatment caused by respiration, inherent bowel mobility and peristalsis, and cardiac motion. This motion is often accounted for by applying a margin to the target of interest to encompass the spatial variability of the target. However, treatment of this margin results in the delivery of unwanted dose to normal tissues when the target is not occupying a given position. The use of real-time tracking techniques may allow for a reduction in this margin and thus has the potential to reduce the morbidity associated with unnecessary dose to surrounding normal tissues. There is potential for all patients undergoing radiotherapy to the aforementioned body sites to benefit from real-time tracking techniques.

Evaluation/summary of results of existing studies

There have been several articles published on the use of the Calypso system both in phantom studies as well as in patients. These clinical studies have focused primarily on prostate cancer. In phantom measurements, Balter et al observed sub-millimeter localization and tracking capabilities of the Calypso system, with values that remained stable over prolonged periods of time. These results have been updated recently by Litzenberg et al. Willoughby et al, in reporting on the first human use of the system, evaluated the localization accuracy of the Calypso system relative to radiographic localization, and assessed its ability to track prostate motion in real time. Their findings indicated significant intrafraction prostate motion (greater than 10 mm) in 2 of 11 patients. The Calypso system demonstrated comparable (within 2 mm) isocenter localization accuracy compared with X-ray localization procedures. Kupelian et al reported on Calypso-based localization and continuous real-time monitoring of the prostate gland on a multi-institutional trial consisting of 41 patients treated at 5 institutions. They found differences between skin marks versus Calypso alignment to be greater than 5 mm in vector length in more than 75% of all fractions. They also observed that individual patients exhibited displacements of 5 mm or more, lasting at least 30 seconds, in 56% of all fractions. Using the criterion that 90% of patients receive 95% of the prescribed dose within the PTV, Litzenberg showed that margins required to accommodate intrafraction motion were approximately 2 mm in all directions, assuming that Calypso-based localization was performed for each fraction prior to the start of treatment. In the absence of Calypso-based localization these margins are approximately 10 mm, indicating that a substantial reduction in margins is possible when daily alignment is performed using the Calypso system. A recent study showed that when Calypso is used for daily localization, the setup uncertainty is reduced enough that planning treatment volumes can be reduced significantly without compromising dosimetric coverage.

The efficacy of Calypso localization among patients receiving androgen ablation therapy is not well defined. One small comparative study of 41 patients in which 14 received neoadjuvant and concurrent androgen suppression found that the implanted electromagnetic markers maintained a stable geometry within the prostate gland over time, both in patients treated with androgen deprivation and in patients treated with radiation therapy alone.

The impact of the electromagnetic detector array on the quality of radiation beams and portal images is of potential concern, as the array is placed several centimeters above the patients during treatment to detect signals from the transponders. Preliminary research demonstrated that the increase in skin dose attributable to the array was “within acceptable clinical limits” though this was not quantified. Additionally, researchers found that attenuation of the beams was less than 0.5% for radiation incident normal to the array; no comment was made regarding oblique or tangential beams. Finally, the researchers stated that “portal image quality due to presence of the array in the beam path was similar to that of patient support devices, such as nylon-strung tennis racquet table inserts.”

Researchers have also performed introductory studies analyzing possible application of the Calypso localization system to head and neck cancer patients. In one such study, a dental prosthesis was cast from a volunteer and some of the teeth were filled with dental amalgam. The prosthesis was placed under the detection array, adjacent to 3 transponders, and the resultant measurements were compared to those taken without the presence of the dental prosthesis. Despite the presence of the amalgam, the system could localize the transponders up to 20 cm from the array. Placement of the transponders within a mouthpiece does not increase backscatter dose to overlying oral mucosa.

The initial clinical study of this technology focused on stereotactic radiosurgery for central nervous system malignancies at the University of Florida. Sixty patients with benign and malignant tumors received a total of 1426 treatments using this frameless SRS system. The system proved to be robust with a misalignment vector error of 0.18 mm; the tolerance limit of 0.3 mm and 0.3 degrees was achieved in every case. This accuracy was determined to rival frame-based systems. Two follow-up studies have further evaluated this technology.13, 14 The most recent study followed 64 patients who received frameless stereotactic radiosurgery for intracranial metastatic disease. Some of the patients were treated up front and others for progressive disease after initial resection or whole brain radiotherapy.

The authors noted the advantage of the system being no pain or discomfort to the patient, ability to perform treatment planning and treatment delivery on different days, and reduction of resource utilization, personnel, cost, and complexity of the stereotactic procedure. The main disadvantage of the technique was noted to involve the potential for lengthy treatment times. Because of the use of conventional linac-based machines, treatment times were about 15 minutes per isocenter. Patients requiring more than 4 isocenters were found to be technically challenging and otherwise not good candidates because of patient fatigue during the lengthy treatment times.

A University of Wisconsin report utilized the system in ten patients with advanced head and neck cancers. These patients were prospectively enrolled to determine the potential impact of traditional daily setup variations with laser alignment and immobilization mask markings. The passive fiducial arrays were mounted to the maxillary bite-tray and acted as the “gold-standard.” They found an average composite mean vector error of 6.97 mm±3.63 mm with conventional methods. Based on these findings, the authors noted that the gross tumor volume and planning target volume were underdosed and critical normal structures like the parotid or globe were overdosed utilizing IMRT with conventional daily patient alignment.

Most of the data for the AlignRT system have been provided during talks or supplemental abstracts and have primarily been related to the setup of breast cancer patients.16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29 The Massachusetts General Hospital used the system to assess its utility in patient setup for APBI (accelerated partial breast irradiation). The accuracy of the system (in 9 patients) was compared with traditional laser and portal image patient setup. Mean 3-dimensional displacements were 7.3±4.4 mm and 7.6±4.2 mm for laser and portal image setup, respectively, as compared with 1.0±1.2 mm for AlignRT. Breathing motion datum at isocenter was 1.9±1.1 mm. As a comparison, the system was used to evaluate the surface motion of the abdomen and 5.7±1.3 mm of displacement was noted. Other sites explored in abstracts noted increased accuracy in positioning the head for use in stereotactic radiosurgery guidance.21, 25

Good correlation has been noted between surface and bony anatomy. The practical utility of the system may be limited by skin-to-tumor positional correlation, which was investigated in one abstract for the case of APBI, where the registration of lumpectomy-site clip-based imaging (“gold standard”) was compared with the skin alignment assessed by AlignRT. Both agreed within 1 mm, suggesting that the surface of the breast may be a reasonable surrogate for the treatment volume.

Identification, analysis, and evaluation of consequences of nonuse

The inability to localize appropriately when using advanced radiation delivery techniques may result in a geographic miss of the intended target tissues, resulting in uncertainties with respect to tumor control. Additionally, a geographic miss of the target(s) generally results in the unintended delivery of high dose to healthy tissues and has the potential to result in unexpected additional morbidity.

The use of real-time tracking techniques has the potential to limit morbidity by decreasing the dose to normal structures through the reduction of target margins utilized for spatial uncertainty. Not employing tracking techniques, whether ionizing or nonionizing, makes it imperative that these margins be applied in order to maintain treatment outcomes gained to date.

Previous localization and tracking techniques typically utilize ionizing radiation. A notable exception is the use of ultrasound, although to date this usage is limited to a few tumor sites. None of the technologies evaluated in this study involve the use of ionizing radiation.

Future prediction based on technology development

Given the growing popularity of dose escalation, hypofractionation, and respiratory gating, and the potential improvement of clinical outcomes from each, it can be postulated that the use of real-time tracking techniques will increase in the radiation oncology community. Through further research and clinical trials, the Calypso system is being expanded to use in body sites outside the pelvis, which may lead to its use on a more routine basis for a larger population of patients.

The use of technologies that allow the registration of patient topography for use with respiratory gating will also likely increase. Current gating systems suffer from the uncertainty of correlation of external markers with internal structure movement. It may be that the increase in the number of registration points (the body surface) will decrease these uncertainties.

There are 2 main areas of clinical outcomes improvement that may be expected as a result of more accurate real-time localization using these technologies. The first involves target localization, in which the treatment fields are centered on a per-fraction basis on the center of mass of the target volume itself, as opposed to stable but unrelated bony or other anatomic landmarks. Use of unrelated landmarks requires the use of wider margins around the target volume, as described in this report, and these margins must be particularly large in the case of very mobile targets, including tumors located in the thorax and abdomen. These same tumors generally have poor outcomes overall, and this is in part due to the inability to escalate dose to large volumes of the surrounding normal tissues. Therefore the use of real-time continuous localization techniques may allow for clinically relevant reduction in margins, which will then allow for reduced normal tissue dose-volumes and subsequently dose escalation to the target volume that may lead to improved tumor control. These are the next phase of clinical trials which need to be conducted to verify outcomes following the implementation of continuous localization technologies.

The second important area of potential improvement in clinical outcomes involves dose to normal tissue. Radiation toxicity to normal tissue is directly related to the volume of the normal tissue that receives any given percentage of the prescribed dose. This outcome is important for all treatment sites, even in the case of tumors that do not exhibit a lot of inherent motion. For example, in prostate or breast cancer, in which the tumor motion is generally less extreme than for tumors in the thorax, and in which dose escalation is either already feasible (prostate) or not indicated (breast), it is still important to minimize the dose to critical normal tissues that surround the target volumes in order to reduce acute and especially long-term toxicity. Many patients with prostate and breast cancer will enjoy normal life spans after treatment; therefore the avoidance of late toxicities to the bowel, bladder, lung, and heart should contribute to quality of life and reduce the cost of post-treatment care. In addition, the feasibility of hypofractionation depends upon the ability to very accurately localize the target volume with minimal margins, as treatment of large volumes of the surrounding normal tissues would result in a higher likelihood of developing late toxicities due to the large dose per fraction used in these regimens. Hypofractionation, therefore, is highly dependent upon technologies that allow precise and real-time target localization in order to reduce normal tissue dose volumes. While the demonstration of reduced late toxicity may take many years to demonstrate in clinical trials, these outcomes should also be examined. In the interim, patients may well benefit from the use of continuous localization techniques as a component of image guided radiotherapy, and their use should be considered one method for achieving greater accuracy and precision.