Volumetric modulated arc therapy for total body irradiation: A feasibility study using Pinnacle3 treatment planning system and Elekta Agility™ linac

Abstract A study was undertaken to explore the use of volumetric modulated arc therapy (VMAT) for total body irradiation (TBI). Five patient plans were created in Pinnacle3 using nine 6 MV photon dynamic arcs. A dose of 12 Gy in six fractions was prescribed. The planning target volume (PTV) was split into four subsections for the head, chest, abdomen, and pelvis. The head and chest beams were optimized together, followed by the abdomen and pelvis beams. The last stage of the planning process involved turning all beams on and performing a final optimization to achieve a clinically acceptable plan. Beam isocenters were shifted by 3 or 5 mm in the left–right, anterior–posterior, and superior–inferior directions to simulate the effect of setup errors on the dose distribution. Treatment plan verification consisted of ArcCheck measurements compared to calculated doses using a global 3%/3 mm gamma analysis. All five patient plans achieved the planning aim of delivering 12 Gy to at least 90% of the target. The mean dose in the PTV was 12.7 Gy. Mean lung dose was restricted to 8 Gy, and a dose reduction of up to 40% for organs such as the liver and kidneys proved feasible. The VMAT technique was found to be sensitive to patient setup errors particularly in the superior–inferior direction. The dose predicted by the planning system agreed with measured doses and had an average pass rate of 99.2% for all arcs. VMAT was found to be a viable treatment technique for total body irradiation.

from 12 to 15 Gy in 6-12 fractions over 4-6 days in myeloablative approaches, with the most common prescription being 12 Gy in six fractions. [1][2][3] Low-dose TBI (2)(3)(4)(5)(6)(7)(8) Gy in 1-4 fractions) can also be used as an effective form of conditioning for older patients who may not be able to tolerate myeloablation. 1,4,5 The most significant organ toxicity associated with TBI is lung toxicity. The lungs are typically shielded to <10 Gy to reduce the chance of radiation induced interstitial pneumonitis. 6 Other key organs at risk from TBI include the liver, kidneys, spleen, heart, and eyes. 7 TBI is traditionally treated using a conventional linac (linear accelerator) using static anterior-posterior/posterior-anterior (AP/PA) or parallel-opposed lateral beam arrangements at extended source-tosurface distance (SSD). Photon beam energies between 4 and 24 MV can be used with tissue compensators to boost regions of varying patient thickness or shielding blocks to limit dose to organs at risk (e.g., lungs, liver, and kidneys). The dose is prescribed to a single point at the midline of the patient with the aim of delivering a uniform dose of AE10%. 8 There are certain drawbacks associated with conventional techniques that have been well reported in the literature; 8,9 long treatment and setup times (which impact patient comfort and ability to maintain accurate positioning during treatment); a lack of accurate three-dimensional treatment planning data; and the requirement for large linac bunkers to accommodate extended SSDs; and specialized treatment equipment.
Other TBI treatment methods include positioning the patient supine and prone on a fixed couch at a large SSD underneath the linac and delivering modulated partial arcs. 10 This method offers greater dose uniformity; however, it is both time and labor-intensive for planning and treatment delivery. A translational couch technique maintains a static beam and gantry, while the patient travels under the linac on a specialized translational couch. 11 A homogeneous dose can be delivered by varying the speed of the couch movement. 12,13 Custom-made lung shields and beam spoilers may still be required for these techniques.
Recently, there has been a shift toward more advanced TBI treatments utilizing modulated arc techniques to target the hematopoietic tissues and reduce the dose to the surrounding healthy tissues. Helical tomotherapy (HT) can be used to treat TBI patients. [14][15][16] It has the advantage of the patient positioned supine, a homogeneous dose distribution without the need for junctions between beams, and with the ability to spare organs at risk. Total marrow irradiation (TMI) utilizes HT 14,17,18 or volumetric modulated arc therapy (VMAT) on conventional linacs [19][20][21] to treat the bone marrow itself, while reducing dose to the surrounding organs at risk and healthy tissue. Total marrow plus lymphoid irradiation (TMLI) targets the total marrow volume plus major lymph node chains, liver, spleen, and sanctuary sites such as the brain. In contrast to traditional TBI, this TMI technique has the potential to reduce both acute and chronic toxicities, reduce treatment time, increase patient comfort, and reduce the need for specialized equipment such as beam spoilers, shielding, and treatment frames. Dose escalation to the total marrow while limiting doses to normal organs to levels lower than in conventional TBI is currently being investigated. Wong et al. 22 reported dose escalation to 15 Gy combined with cyclophosphamide and etoposide therapy is associated with acceptable toxicities and encouraging outcomes in patients with advanced acute leukemia undergoing bone marrow transplantation. Further clinical trials are required to determine appropriate TMI and TMLI doses and whether dose escalation translates into improved control rates and survival.
The aim of this study was to investigate the feasibility of achieving clinically acceptable TBI plans with the Pinnacle 3 treatment planning system (TPS) and accurate delivery using an Elekta Agility TM linac. Previous studies have demonstrated the use of Varian Rapi-dArc in combination with the Eclipse TPS 23 with promising results in the first clinical cases. 24 A VMAT approach to TBI treatments has the potential to make TBI accessible to more clinical departments where equipment limitations or bunker size has restricted implementation. This work may also facilitate a move toward a TMI type treatment in future where the skeletal and hematopoietic tissues can be targeted and potentially receive an escalated dose regime as more clinical data become available. The planning target volume (PTV) was defined as the entire body, contracted to 5 mm below the skin. The patient datasets included in the planning study encompassed a range of patient shapes and sizes. Total PTV volume ranged from 39722 to 61900 cc and lung volume ranged from 2275 to 3534 cc. The planning aims were to deliver a uniform dose of 12 Gy to the PTV while limiting the mean lung dose to less than 8 Gy, and the mean kidney and liver doses to below 9 Gy. The PTV was extended 3 mm into the lungs in a pragmatic compromise between coverage of setup, geometric, and intrafraction motion uncertainties and sparing of the lungs. The 3 mm margin was chosen based on reported margins used in TMI. 14,18,20,21,25 A planning volume at risk (PRV) margin of 7 mm was applied to the kidneys in the superior-inferior direction to account for organ motion during treatment. 26 The Pinnacle 3 SmartEnterprise version 9.10 (Philips Healthcare, Andover, MA, USA) treatment planning system was used to optimize VMAT beams. The system consists of three application servers with dual 2.93 GHz Intel Xeon 5600 series processor and 96 GB of RAM.

| METHODS AND MATERIALS
The PTV was split into subsections for the head, chest, abdomen, and pelvis. A combination of nine 6 MV photon beams were arranged along the patient's longitudinal axis using an isocenter at the middle of each sub-PTV. All isocenters had the same lateral and anterior-posterior coordinates to limit the couch moves required on treatment to only longitudinal shifts. The head, abdomen, and pelvis Due to limitations of our Pinnacle 3 system at optimizing and converting more than five beams concurrently over a large volume, the planning process for the nine VMAT arcs was broken into three stages. First, the head and chest beams were optimized and con-  Table 1. A uniform dose objective with a high weighting was set to keep dose uniformity in the PTV to acceptable levels.
Max EUD (equivalent uniform dose) objectives were set for the organs at risk. To maintain acceptable dose coverage to the ribs, the liver and lung objective ROIs were contracted by 5 mm (Liver_cont and Lungs_cont).
The beam arrangement and stages for TBI planning in Pinnacle. Nine VMAT arcs are arranged longitudinally along the patient and assigned to four isocenters. Each isocenter has a superior (red) and inferiorly (yellow) offset arc. The chest isocenter has an extra arc with no field size restrictions (orange). (a) Step 1: of the planning process is to optimize and convert the head and chest beams together. (b) Step 2: head and chest beams are set to "None" in IMRT parameters and the abdomen and pelvis beams are optimized together. (c) Step 3: All beams are turned on and a final optimization process performed to smooth out junction regions. During the optimization process, the air cavities within the patient were overridden to a density of 1 g/cm 3 , to restrict the optimizer from over increasing the photon fluence in low-density regions. The air cavity density override was switched off for the final dose computation. Initial optimization and conversion was performed with a dose grid of 5 9 5 9 5 mm. The final dose computation was performed with a dose grid of 3 9 3 9 3 mm using the collapsed cone convolution algorithm.
Treatment plan evaluation was performed using dose-volume histogram (DVH) analysis. The planning aims for the PTV were to deliver 12 Gy to at least 90% of the PTV (i.e., V 100% ≥ 90%) and 11.4 Gy to at least 95% of the PTV (i.e., V 95% ≥ 95%). For the PTV, the mean dose was recorded as well as the hottest dose to All measurements were analyzed in the SNC Patient software (version 6.6.2) using a global gamma analysis of 3%/3 mm on absolute dose.

| RESULTS
An example dose distribution for patient 3 is shown in Fig. 3   techniques. 14,24 Although no strict limit is set on D 2 cc or the dose homogeneity, regions of hotspots are evaluated based on their location in the patient. Hotspots may be acceptable if they are away from critical structures and mostly in external tissues. Due to the normalization of the PTV dose (V 100% ≥ 90%), the average dose is shifted higher than the prescribed dose for the VMAT plans (mean dose in the PTV over the five patients was 12.7 Gy). In this study, the PTV was trimmed to 5 mm below the surface of the patient as certain problems arise when attempting achieve the planned absorbed dose in this region. Within the buildup region and with the coarse voxel resolution used during optimization (5 9 5 9 5 mm), the dose increases rapidly and has large uncertainty. It is difficult to achieve the planned dose in this region due to the lack of electronic equilibrium; the inverse optimizer must therefore increase the dose in this region by strongly increasing the photon fluence which can lead to reduced homogeneity in the PTV. Hot spots can also be created by small setup errors during treatment. For the above reasons, it is also difficult to assess the skin dose, but in practice the combination of multiple arcs, oblique beam incidence, and beam exit from all angles significantly reduces the normal photon beam skin-sparing effect. Although not included in this study, skin dose could be boosted by using well-documented techniques such as bolus, a beam spoiler, or a virtual bolus in the planning system. 28 The practice of delivering TBI at low dose rates stems from radiobiological considerations, namely the sparing of damage as a result of cellular recovery. 29,30 Pneumonitis is one of the major toxicity concerns with TBI, and was originally related to dose rate; however, since the introduction of fractionated regimes, several publications have shown the dose rate to have little effect. 31 It is important to note the irradiation of the lower limbs was not included in this paper. It is intended the legs be treated in the feet-first direction using AP/PA beams with conventional static fields. The ideal method would be to treat the lower legs with a series of VMAT arcs also to smooth doses in the junction regions. This has proved a challenge due to the difficulties in junctioning two VMAT arcs that have been planned on two CTs with different treatment orientations (i.e., head-first and feet-first orientation). Although Springer et al. 24 have described a method for irradiating the legs with VMAT by summating the resulting head-first and feet-first plans in Eclipse and performing a final optimization of the junction region. Junctioned AP/PA beams for the legs are deemed acceptable owing to the absence of any organs at risk, and are the simplest approach to achieving the prescribed dose to the target.
Moving to a VMAT TBI technique will be both labor and resource intensive. CT simulation is estimated to take between 1 and 1.5 hr. The contouring of required PTVs and organs at risk is expected to take approximately 2 hr using the auto-contouring software MIM Maestro TM (MIM software). Total time for optimization of the nine VMAT arc is approximately 21 hr (or 3-4 working days), but requires minimal user input with automated scripting in Pinnacle 3 . This does not include the time required for planning of the legs, plan checking, and export to the record and verify system. However, total time on the treatment machine will be decreased as appointments for simulation and imaging checks on the shielding blocks will no longer be required. It is estimated the time per treatment fraction will be reduced also from 1.5 to 2 hr for the 2D extended SSD technique to 1-1.5 hr per fraction for VMAT. The total time for physics quality assurance is 6-8 hr for dose calculation and 2-3 hr of machine measurements.

| CONCLUSION
This study has demonstrated the feasibility of achieving clinically acceptable VMAT TBI plans with the Pinnacle 3 treatment planning system and accurate delivery using an Elekta Agility TM linac. The advantages of this technique include improved patient comfort and positioning reproducibility during treatment, accurate 3D dose information, and the ability to selectively spare organs at risk.

CONFLI CT OF INTEREST
The authors declare no conflict of interest.