The effect of designing a rotational planning target volume on sparing pharyngeal constrictor muscles in patients with oropharyngeal cancer

Abstract Background Planning target volume (PTV) has been used to account for variations in tissue, patient and beam position. In oropharyngeal cancers, an isotropic expanded PTV has been used. Aim The aim of this study was to design a new margin formula that would cover the space occupied by an oropharyngeal clinical target volume (CTV) with ±5‐degree rotation around the spine in order to reduce the pharyngeal constrictors overlap with PTV compared to an isotropic expanded PTV. Methods We retrospectively evaluated 20 volumetric‐modulated arc therapy (VMAT) plans. In order to perform an off‐axis rotation, a hypothetical point was placed through the center of the cervical spinal canal and the image was then rotated around the longitudinal axis ±5 degrees. This created a new set of CTVs that were combined to form the new rotational PTV. The overlap between the pharyngeal constrictor muscles (PCMs) and both PTVs was then evaluated. Results The new rotational PTV causes reduction in the superior PCM overlap in the base of tongue (BOT) lesions compared to tonsillar lesion, 57.8% vs 25.8%, P = 0.01, as well as middle PCM overlap, 73% vs 49%, P = 0.04. Average percent change for PTV volume and overlap with the superior, middle, and inferior PCMs are as followed: −19%, −37%, −59.4%, and −45.2. The smallest isotropic expansion that covers the new rotational PTV was between 3 and 5mm with the average tumor center shift of 0.49 cm. Conclusion This new rotational PTV causes significant reduction of the overlap volume between PCMs and PTVs in order to spare the PCMs compared to isotropic expanded PTV.

anatomic motion, delineation errors, and setup errors. 2 This expanded safety volume is known as planning target volume (PTV).
Historically, a uniform margin around the clinical target volume (CTV) has been used to define the PTV. Target displacement can be decomposed into translations and rotations. The rotational component is of great importance especially when the target has a nonspherical shape or the rotation is off axis, as a small rotation can cause deviation of the dose distribution. 3,4 As shown in a study by Peng et al, in large targets with irregular shapes, target coverage can decrease significantly when rotational error of 5 degrees or more is present. 5 Both translations and rotations should be considered to form the swept space of the target in designing precise PTVs, but rotations are typically neglected. In addition, estimating the swept volume of an object with both rotation and translation is not simply solved. 6,7 Based on a study done by Hong et al considering 20 institutes, the average recommended PTV expansion from CTV was 4.11 mm with a standard deviation of 3.19 mm. 8 . Another study done by Djordjevic et al showed considerable local residual setup error even with daily imaging in head and neck cancers and required PTV margins ranging 4.5 to 9.3 mm for each subregion. This study suggested designing a variable margin related to the tumor site to account for minor cervical deformations. 9 Most planning systems allow margins to be specified along the "three Cartesian dimensions only," a limitation attested in the ICRU 83 report. 10 Based on the American Cancer Society (ACR) reports, the 5-yr relative survival rates for cancers of the oral cavity and pharynx is 65% and can be as high as 84% in patients with early stage disease. 7 This emphasizes the importance of reducing tissue morbidity by delivering a more focused radiation treatment. Dysphagia is a potentially a devastating toxicity of radiation therapy (RT) in this population as 59% of head and neck patients report persistent dysphagia at an average of 33-month follow-up. 11 This can lead to limited oral intake and possibility of developing aspirations which can lead to life-threatening aspiration pneumonia as well as feeding tube dependence. 60% of patients are feeding tube dependent during their treatment course. 12,13 The cricopharyngeus muscle, inlet of esophagus, superior, middle and inferior constrictors play an important role in the swallowing process. 14 There are several studies that have evaluated the correlation between dose to pharyngeal constrictors and dysphagia rates. In one study, for a median dose of 50 Gy to superior and middle pharyngeal constrictors, the probability of developing grade 3 and 4 dysphagia is approximately 20% and if the dose is reduced to 22 Gy, the probability of developing dysphagia is as low as 2%. 14 Another study evaluated the effect of reducing PTV margin on radiation induced toxicity. A 2 mm reduction in the in the PTV margin resulted in a significant reduction of acute dysphagia defined as feeding tube dependence by 50% and late dysphagia from 22% to 11%. 15 This emphasizes the importance of margin construction based on real anatomic motion rather than uniform expansion of the tumor. Cervical spinal rotation can happen during treatment.
In a study done by Kapanen et al, improvement in the formation of thermoplastic masks by making them tighter and improved image matching to vertebrae reduced residual random errors especially by reducing the rotation of vertebrae and head. 16 The positional uncertainty contributed by rotational displacements are significant. Rotations greater than 3 degrees have been observed. 17 In a study done by Nakata et al, the setup and rotational shifts in head and neck cancer patients undergoing IMRT with the use of an immobilization device and an IGRT system were evaluated. This study showed variability in random translational errors for different regions in the anatomy of head and neck cancer patients due to rotational shifts happening inter and intrafraction. It also estimated rotational shift using stereoscopic projections and provided estimates of mean and standard deviation. 18 The intrafractional systematic (Σ) and random (σ) rotational displacements of the spine in the upper neck found from repeat stereoscopic projections 18 would yield a needed coverage range of ±5 degree along one axis, using the Van Herk formula of 2.5 Σ + 0.7 σ for drawing a range of displacements to encompass likely target shift. 19 Most previous studies of rotational effects have considered the geometric center of the gross tumor volume as their axis of rotation.
We evaluated an off-axis rotation around the spine, as this is the more likely anatomic axis of rotation. 20 In this study, the hypothesis was that forming a margin that would cover the space occupied by an oropharyngeal CTV with ±5-degree rotation around the spine would spare more of the pharyngeal constrictors than would current practice using an isotropic expanded PTV. We have retrospectively evaluated 20 patients with head and neck cancer who completed VMAT in our institute for oropharyngeal carcinomas. PTVs were specified on the physician request form to be 0.3 cm and formed isotropically. A new PTV was designed considering ±5 degree rotation along the spinal axis for each patient. The overall objective of this study was to evaluate the difference in the overlap between PTV and the pharyngeal constrictor muscles when using the original PTV based on isotropic expansion compared to a new PTV that tracked rotational deviations.

| MATERIALS AND METHODS
This study was approved by the IRB of our institute. All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2008. Since this was an IRB approved retrospective study reviewing treatment plans of patients who received radiation in our department, individual consent forms were not obtained. We retrospectively evaluated 20 VMAT plans of patients who received EBRT in our department for oropharyngeal cancer. The Eclipse treatment planning system, version 13.7 by Varian Medical systems, was used to generate the VMAT plans. Patients were immobilized using the Klarity S type head and shoulder mask. The treatment-planning CT scan was acquired using 3-mm slice scan with intravenous contrast.
Since this is a retrospective study, all contours were done by the primary treatment team. The following contoured OARS were available: Brain stem, spinal cord, bilateral parotid, esophagus, trachea, brachial ARBAB ET AL. Eclipse. Figure 1 illustrates the technique used. Figure 1 In order to evaluate the effect of the rotational PTV on the mean PCM dose, one patient was randomly selected and the VMAT plan was generated using three dose levels, 70 Gy to GTV and positive nodes, 63 Gy to high-risk volume and 54 Gy to bilateral elective cervical nodes.
All PTV levels were designed using the same rotational technique. Based on the Nutting et al study, 23 the mean dose to the volume of superior and middle PCM outside the high-dose volume was set as a mandatory constraint and the expected mean dose was <50 Gy. 2 sets of plans were generated using the exact same dose constraints and optimization techniques. In the first plan, the standard isotropic expanded PTV was used and in the second plan, the rotational PTV was used instead. In order to calculate physician-rated swallowing dysfunction in 6 months, the NTCP model discussed in Christianen et al paper was used. 24 This model uses superior pharyngeal constrictor and supraglottic larynx mean dose for NTCP calculation.
JMP version 13 was used to analyze the data. Descriptive statistics and t-test were used and P < 0.05 was considered as statistically significant.

| RESULTS
With regard to the baseline tumor characteristics, 25% of patients had base of tongue lesion and 75% had tonsillar lesions. Sixty-five percent of cases were human papilloma virus (HPV) positive. T2 was the most common T staging (35%) followed by T4 (30%), T1 (20%), and T3 (15%). Forty-five percent of patients had N2 disease, 30% had N1, 15% had N0, and 10% had N3 disease. None of the patients had metastatic disease. All patients received bilateral elective cervical neck radiation. Table 1

| DISCUSSION
Per ICRU, PTV expansions have largely been built from recorded translational shifts only, neglecting rotational components. 10 When rotations have been considered in PTV construction using sampling methods, the translational and rotational components have been treated as statistically independent by assuming a given and fixed rotation center. 25 In this study, the hypothesis was to design the PTV based on ±5-degree rotation instead of isotropic expansion of CTV. The rotation center was set in the center of the cervical spinal canal, which is the anatomic location for rotations in the head and neck region. 26 Standard practice is to shift isocenter relative to the spine because its center is well detected on orthogonal KV projections, using the spinous process and borders of the vertebral body.