Practical considerations of linear accelerator‐based frameless extracranial radiosurgery for treatment of occipital neuralgia for nonsurgical candidates

Abstract Occipital neuralgia generally responds to medical or invasive procedures. Repeated invasive procedures generate increasing complications and are often contraindicated. Stereotactic radiosurgery (SRS) has not been reported as a treatment option largely due to the extracranial nature of the target as opposed to the similar, more established trigeminal neuralgia. A dedicated phantom study was conducted to determine the optimum imaging studies, fusion matrices, and treatment planning parameters to target the C2 dorsal root ganglion which forms the occipital nerve. The conditions created from the phantom were applied to a patient with medically and surgically refractory occipital neuralgia. A dose of 80 Gy in one fraction was prescribed to the C2 occipital dorsal root ganglion. The phantom study resulted in a treatment achieved with an average translational magnitude of correction of 1.35 mm with an acceptable tolerance of 0.5 mm and an average rotational magnitude of correction of 0.4° with an acceptable tolerance of 1.0°. For the patient, the spinal cord was 12.0 mm at its closest distance to the isocenter and received a maximum dose of 3.36 Gy, a dose to 0.35 cc of 1.84 Gy, and a dose to 1.2 cc of 0.79 Gy. The brain maximum dose was 2.20 Gy. Treatment time was 59 min for 18, 323 MUs. Imaging was performed prior to each arc delivery resulting in 21 imaging sessions. The average deviation magnitude requiring a positional or rotational correction was 0.96 ± 0.25 mm, 0.8 ± 0.41°, whereas the average deviation magnitude deemed within tolerance was 0.41 ± 0.12 mm, 0.57 ± 0.28°. Dedicated quality assurance of the treatment planning and delivery is necessary for safe and accurate SRS to the cervical spine dorsal root ganglion. With additional prospective study, linear accelerator‐based frameless radiosurgery can provide an accurate, noninvasive alternative for treating occipital neuralgia where an invasive procedure is contraindicated.


| INTRODUCTION
Occipital neuralgia is a neurological condition characterized by paroxysms of intense pain transmitted by the greater occipital nerves in the back of the head and neck often accompanied by a dull ache. [1][2][3][4] The condition is differentially diagnosed from other headache types by using patient descriptions of the pain, noting the location of tenderness associated with pain episodes, and achieving prompt relief of pain following an anesthetic block of the greater occipital nerve. 5,6 The incidence of occipital neuralgia in the general population remains unknown but is thought to be less than that of trigeminal neuralgia and glossopharyngeal neuralgia which have an incidence of 20/ 100,000 per year and 0.7/100,000 per year, respectively. 7 Medical management for patients diagnosed with occipital neuralgia generally includes analgesics or anti-inflammatories which proves effective for most patients. Numerous treatment options may be warranted for patients with continued disabling and intractable pain despite temporary treatments or when invasive therapies such as surgical incision, radiofrequency ablation, injected neurotoxin facilitated nerve blocking, implanted nerve stimulator, or surgically decompressing the nerve fail to provide relief. 6,[8][9][10] If the condition continues to be refractory, nerve sparing procedures are utilized in preference to neurodestructive surgeries. [11][12][13][14] Neurodestructive procedures, such as neurectomies, are highly invasive and carry some degree of risk of permanent complication. 15,16 For those patients whose pain recurs following an invasive procedure, a secondary invasive procedure is often contraindicated due to compounding risk. 6 Although occipital neuralgia has been reported to be a complication resulting from frame installation for frame-based radiosurgery, the use of SRS for the treatment of occipital neuralgia has not been reported to date. 17 The success of radiosurgery in the management of trigeminal and glossopharyngeal neuralgia (both cranial-based functional diseases) has been well established. [18][19][20][21][22][23][24][25][26] SRS for trigeminal neuralgia and glossopharyngeal neuralgia involves single-fraction high doses to the isocenter placed along the course of the nerve after exiting the central nervous system. In comparison, radiosurgery for occipital neuralgia has not been well explored due to limitations in the ability of traditional SRS treatment modalities in delivering extracranial applications. 27,28 With the recent advances in intrafractional image guidance, we report here the treatment of occipital neuralgia using linear accelerator-based frameless radiosurgery.

2.A | Phantom feasibility study
Quality assurance for geometric accuracy, precision, and dosimetric accuracy was established during commissioning of the functional SRS program. The initial commissioning included evaluation of cone positioning reproducibility, providing accuracy within 0.5 mm, and intercomparison of PDD, off-axis factors, and total scatter factors with another institution. Principles of small-field dosimetry were applied by measuring output with film, diode, and a small volume ionization chamber. [29][30][31][32][33][34] Relative output factors were obtained from the ratio of the dose at isocenter at depth d max for the conical collimator (4 mm in our case) relative via the "daisy chain" method described by Dieterich et al. to the dose measured for a 100 by 100 mm 2 square field size at a depth of d max both at 1000 mm source to isocenter distance providing a ratio of 0.6668. 35,36 Total scatter factors compared against four other institutions results in agreement within 1%. The SRS single-beam phantom from the Imaging and Radiation Oncology Core -Houston (IROC) was used to verify the output of the SRS cone program. The measurements and commissioning process received peer review by a medical physicist expert in radiosurgery as part of Novalis Certification.
A phantom feasibility study was performed by CT simulating, planning, and treating an SRS anthropomorphic head phantom (CIRS Computerized Imaging Reference System Inc. VA, USA). This phantom included simulated bony anatomy that is visible to both CT and x ray and included a neck which was necessary in consideration of the target localization under investigation.  Fig. 1 can be interpreted as the hemisphere of that spherical space corresponding to the superior hemisphere surrounding the patient's head. For additional orientation, consider the anterior pole to colloquially correspond to the patient's nose. It is not intended to be a "planning solution" for all patients whose anatomy could vary significantly. Patient-specific collision avoidance verification tests were performed for the phantom and patient included in this study.
Entry dose through the spinal cord was avoided by limiting or avoiding the use of beams entering through the contralateral side. The anterior and posterior poles were labeled as avoidance zones to prevent beam overlap. Brain entry dose was noted to avoid beams that would result in the beam first traveling through the brain to reach the target. Two elliptical shapes are represented on the left and right side of Fig. 1 that reflect two additional zones of high collision risk and represent the corners of the treatment couch used for this study which has rounded square corners and is not semicircular. Because of this design, there was an increased risk of collision as the conical collimator would start to dip below the horizon defined by the treatment couch with the contralateral side representing a larger risk area due to the lateral shift required to align the patient's target side to the radiation isocenter. The green zone represented a lowered collision risk. Due to this supine patient setup, radiation beams passed through the treatment couch prior to the patient. For this reason, the treatment couch and resultant attenuation was taken into account in the treatment planning system.

2.B | Case selection
For this case study, a 53-year-old male patient presented with complaints of severe cervical neck pain diagnosed as occipital neuralgia.
The patient had undergone both radiofrequency ablation with 1 month of relief and cervical spine decompression and fusion of C4-7 with no improvement. Various local injections provided only short lasting relief. The pain was reported to be debilitating with the maximum score of V on the Barrow Neurological Institute pain intensity scale. 37 As the occipital neuralgia condition was refractory to radiofrequency ablation, surgery, and medical management, the option of radiosurgery was presented.  Figure 3 shows an anatomical illustration of the intended target along with axial, sagittal, and coronal views of the target. The CT myelogram was preferable for target definition, in this case, to MRI for two reasons. First, artifacts from the titanium hardware were present on MRI. Secondly, like-modality image registration was preferable to cross-modality image registration in an effort to decrease uncertainty introduced in the image registration step. 38 Figure 4 shows all available image modalities.

2.C | Anatomy, targeting, and immobilization
X-ray image guidance was performed throughout the treatment delivery, and the results of the ExacTrac-based coregistration F I G . 1. A projection treatment map for a left-sided occipital neuralgia typical target is presented as a planning guide and result of a phantom feasibility test performed prior to attempting patient treatment. It is noted that a right-sided target would need a guide that is mirror imaged and symmetric about the sagittal axis.

2.D | Treatment and IGRT equipment
The treatment was successfully performed with a NovalisTX (Varian Medical Systems, Palo Alto, CA, USA) equipped with ExacTrac stereoscopic x-ray image guidance (BrainLab AG, Feldkirchen, Germany). The treatment couch was an IGRT couch top (BrainLab AG, Feldkirchen, Germany). We previously determined the congruency between mechanical (including gantry and couch rotational isocenters), IGRT, in-room laser, and treatment beam isocenters specific for this radiation treatment device and imaging platform. 39 This study quantified an average magnitude of distance for the laser-defined alignment isocenter to be 0.58 mm. The effect of gantry sag was 0.4 mm in magnitude. The IGRT isocenter was within 0.5 mm of the radiation-defined isocenter. Couch walkout had a maximum discordance of 0.72 mm with the isocenter. These values were based on a statistical analysis of 149 individual isocenter congruency tests performed for this machine and imaging combination. The magnitude of difference between the radiation-defined and ExacTrac IGRT-defined isocenter was significantly less than the couch walkout. Therefore, imaging was performed for each treatment couch angle prior to treatment arc delivery.

3.B | Plan characteristics
The dose grid resolution in the treatment planning system was set to 0.5 mm. For small objects, such as the contoured target, the grid size was automatically adjusted such that at least 10 voxels for each F I G . 2. Image guidance for intrafractional positional corrections during the treatment of occipital neuralgia using a frameless approach requires an inverse approach as compared to traditional cranial-based SRS. Image registration for IGRT during treatment should be performed using anatomy in the vicinity of the target (i.e., on the C1/ C2 level). Other bony anatomy such as the skull, mandible, and the lower C-spine should be understood as capable of moving independent of the target and, therefore, should not be used as registration references when aligning the patient to the isocenter.

3.C | Image guidance
The results of intrafractional IGRT are graphically represented in

3.D | Patient outcome
At most recent follow-up, the patient reported 4 months of pain  End-to-end testing using this modality was completed during the commissioning of both the linear accelerator and beam model in the Post-treatment delivery analysis, using the results of image guidance determined for each delivered arc, showed minimal deviations from treatment planning projections. The exception to this was the maximum spinal cord dose which was projected to have received a maximum dose of 3.59 AE 0.08 Gy versus the planning calculation dose of 3.36 Gy. This deviation was noted to be well within the acceptable limits asserted by the radiation oncologist.
After treatment of trigeminal neuralgia, SRS is thought to cause axonal degeneration and necrosis, with pain relief after radiosurgery occurring from 2 to 6 weeks after the procedure. 18 A review of three separate case series with validated pain endpoints found that the 1 year pain-free outcome was 69% of patients, dropping to 52% at 3 year. 19 We would propose that larger series of occipital neuralgia patients would likely mirror these results if in fact occipital neuralgia has the same underlying mechanism as trigeminal neuralgia.
We are prospectively following our occipital neuralgia patients and anticipate reporting their outcomes once the sample size is statistically more valid.
Careful image registration for the actual patient was paramount for successful target definition and localization. Registration for anatomy in the cervical spine can be difficult due to the mobilization of the neck and a lack of an ability to share immobilization between diagnostic and therapeutic applications. 65,66 In addition, limitations of the rigid registration available for use with respect to the target location were noted with the unavailability of a nonrigid registration platform for this study. Such registration must be carried out with a focus on the level of C2 and with careful evaluation of the registration success for both the target-sided dorsal root ganglion and the spinal cord and canal.

| CONCLUSION S
We report the first application of SRS for the treatment of occipital neuralgia. We performed quality assurance testing on the linear accelerator isocentricity, the fusion of planning datasets, and image guidance to ensure accurate delivery. Initial short-term follow-up is encouraging. Additional prospective study is needed before SRS can be considered an appropriate clinical option for occipital neuralgia.

ACKNOWLEDG MENTS
We acknowledge Norton Healthcare for their continued support as well as the Associates in Medical Physics, LLC.