Standardized flattening filter free volumetric modulated arc therapy plans based on anteroposterior width for total body irradiation

Abstract In this work, the feasibility of using flattening filter free (FFF) beams in volumetric modulated arc therapy (VMAT) total body irradiation (TBI) treatment planning to decrease protracted beam‐on times for these treatments was investigated. In addition, a methodology was developed to generate standardized VMAT TBI treatment plans based on patient physical dimensions to eliminate plan optimization time. A planning study cohort of 47 TBI patients previously treated with optimized VMAT ARC 6 MV beams was retrospectively examined. These patients were sorted into six categories depending on height and anteroposterior (AP) width at the umbilicus. Using Varian Eclipse, clinical 40 cm × 10 cm open field arcs were substituted with 6 MV FFF. Mid‐plane lateral dose profiles in conjunction with relative arc output factors (RAOF) yielded how far a given multileaf collimator (MLC) leaf must move in order to achieve a mid‐plane 100% isodose for a specific control point. Linear interpolation gave the dynamic MLC aperture for the entire arc for each patient AP width category, which was subsequently applied through Python scripting. All FFF VMAT TBI plans were then evaluated by two radiation oncologists and deemed clinically acceptable. The FFF and clinical VMAT TBI plans had similar Body–5 mm D98% distributions, but overall the FFF plans had statistically significantly increased or broader Body–5 mm D2% and mean lung dose distributions. These differences are not considered clinically significant. Median beam‐on times for the FFF and clinical VMAT TBI plans were 11.07 and 18.06 min, respectively, and planning time for the FFF VMAT TBI plans was reduced by 34.1 min. In conclusion, use of FFF beams in VMAT TBI treatment planning resulted in dose homogeneity similar to our current VMAT TBI technique. Clinical dosimetric criteria were achieved for a majority of patients while planning and calculated beam‐on times were reduced, offering the possibility of improved patient experience.


| INTRODUCTION
Total body irradiation (TBI) has a prominent role in the treatment of a variety of blood disorders requiring hematopoietic stem cell transplantation (HSCT), including cancers such as leukemia and lymphoma. [1][2][3][4][5] The use of TBI in conjunction with chemotherapy for conditioning prior to HSCT has been shown to be clinically beneficial as TBI is able to target sanctuary sites such as the brain and testes, and aids in suppressing the immune system to improve the chances of successful stem cell engraftment. 1,4,5 TBI requires the delivery of a homogeneous dose of radiation to the entire body to target all malignant blood cells as well as the bone marrow where most blood cells are formed.
Recently, one focus of TBI technique development has been improving planning accuracy and decreasing toxicities associated with myeloablative therapy. This includes the use of volumetric modulated arc therapy (VMAT), which has shown promise in improving dose homogeneity and reducing the dose to organs at risk (OAR). Springer et al developed a 9-15 isocenter VMAT technique where the patient is longitudinally translated to capture the entire planning target volume (PTV) in eight segments. 6 Smooth junctions between segments were achieved using inverse planning, resulting in a homogeneous dose delivery. A similar technique was described by Tas et al, but only 3-5 isocenters were required. 7 They additionally completed three-dimensional dose-volume histogram (DVH) based quality assurance of their plans including their junction regions, thus presenting a robust and accurate method of TBI delivery. While VMAT TBI is resulting in higher quality plans, some problems persist. Moving to optimized plans requires an increase in treatment planning resources.
For the higher complexity VMAT TBI techniques, the contouring time can be up to six hours, and treatment planning time over a day. 6 In addition, highly modulated plans increase the already prolonged beam-on times for TBI treatments. Reported times including patient set-up for one fraction VMAT TBI are between 55 and 120 min for adult patients. 6,7 Multi-fraction treatments lead to considerable clinic resource utilization as well as the patient spending long periods on the treatment couch resulting in heightened discomfort.
In an effort to decrease the amount of time required for VMAT TBI while attaining similar dose homogeneity, we present a treatment planning study investigating the feasibility of a VMAT TBI technique utilizing flattening filter free (FFF) beams. FFF beams permit a higher dose rate delivery, so this modality substitution should result in decreased beam-on times and overall shorter treatments. In addition, the peaked dose profile for FFF beams provides the possibility of a body contour match at the umbilicus level in comparison to the standard flattened dose profile, with subsequent isodose lines that are flatter prior to multileaf collimator (MLC) modulation. To further improve time efficiency, a set of standardized plans with forward-planned MLC leaf positions are presented that are suitable for patients with certain physical dimensions, lessening the strain on planning resources. A discussion of potential lung toxicities with an increased dose rate then follows.

2.B.2 | FFF delivery technique
The clinical VMAT TBI technique was used as a template for the development of the FFF delivery technique. Patient setup, orientation, and immobilization were kept the same for the FFF VMAT TBI technique.
Two major changes to the clinical technique were made: the beam energy was changed from 6 MV to 6 MV FFF, and the MLC shapes were pregenerated using custom code to achieve the desired dose distribution within the body, resulting in a set of standard treatment plans. The process of standard plan generation follows.

Relative arc output factors (RAOF)
FFF plan construction began with a 310°-60°40 cm × 10 cm open field arc with meterset weights based off of the clinical VMAT TBI technique, summarized in Table 2. The beam energy was 6 MV FFF, the nominal dose rate 1400 MU/min, and dose was calculated in Eclipse using the analytical anisotropic algorithm (AAA) version 11.0.31 with grid size 0.5 cm.
The goal is to produce a dose distribution that is flat in both the lateral and cranial-caudal directions at a mid-plane depth, with 100% of the dose for each patient orientation at mid-plane, allowing for full even coverage when combined with both orientations. To assess the homogeneity of the dose distribution, discrete lateral dose profiles in the patient's AP mid-plane were collected for both the supine and prone orientations at 9-10 cranial-caudal locations for three patients from each category, and then averaged patient-wise. Profiles were discretized using reference points laterally spaced by 5 cm. Lateral dose profiles were longitudinally spaced by approximately 15 cm with the exception of the cranial region where profiles were typically closer spaced due to large variation in body contour (ranging between 10 and 16 cm) and the simulated legs where less profiles were required. Spacing was kept as consistent as possible between patients. One of the cranial-caudal locations was chosen to be the lungs to provide shielding. An example of the cranial-caudal spacing is shown in Fig. 2, and a lateral dose profile in Fig. 3.
After collecting the open field dose values, a method was developed to change the field size to achieve a flat 100% isodose line at mid-plane. The factor required to change the mid-plane profile isodose values to 100% was calculated for each point as follows: Open Field Isodose % is the isodose value at a particular point on the profile described as a percentage of the prescription dose, and RAOF is the relative arc output factor. To deduce the necessary

Plan evaluation
All fffTBI were reviewed by two radiation oncologists. Conservative dosimetric criteria (as a percentage of the prescription dose) that were used for plan evaluation include: Body-5 mm D98% ≥95%, Body-5 mm D2% ≤115%, and MLD ≤105%. The criterion that takes precedence is target coverage (D98%). An ALARA perspective was considered for the hotspot (D2%) and MLD objectives as they were more difficult to meet in most cases for both the fffTBI and clini-calTBI.

2.C | FFF plan robustness to set-up errors
Testing the performance of the fffTBI with systematic set-up errors was completed using Eclipse. Three patients with variable heights and widths from each category were selected for examination, for a total of 18 patients. Each arc isocenter was manually shifted 2 cm in the lateral, cranial-caudal, and anteroposterior directions, and in both the positive and negative directions for these axes. The shifts were completed for both the supine and prone plans. Dose was recalculated for each separate shift using the same monitor units (MU) as the original fffTBI, and then a plan sum of the supine and prone plans corresponding to the same shift were completed to assess maximum error. The DVH parameters of the D98% and D2% of the Body-5 mm structure and MLD were collected, and the difference between the shifted FFF plan parameter and the original FFF plan parameter was taken.

3.A | Dosimetry
An example of CT data from one patient with the fffTBI and clini-calTBI in dose color wash in the coronal and sagittal planes is shown in Fig. 7. Qualitatively, the dose distributions vary between the two plans, particularly when examining hotspot location and lower dose regions. To quantitatively compare the plans with respect to dose homogeneity and lung dose, the D98% and D2% of the Body-5 mm structure and MLD were collected for both plans and all 47 patients.
Uniformity is also reported via the homogeneity index. The resulting distributions are summarized in Fig. 8 and Table 3  Patients with increased AP width did not meet the dose homogeneity goal as consistently, illustrated by the increased dissimilarity between the fffTBI and clinicalTBI Body-5 mm D2% distributions  1) is then applied to each point. The required field size is then calculated at each point using the linear fit from Fig. 4 (Equation 2). In order to keep the standard plans more generalized, the field size is symmetrized over the patient's lateral mid-line (x = 0 cm), favoring the larger field size changes. Finally, the MLC leaf position coded into the plan DICOM file is calculated. These positions correspond to those in the BEV shown next to the bar graph. MLC, multileaf collimator; RAOF, relative arc output factor.
Body-5 mm D2% and MLD (Table 3). Across all of the categories, the differences between fffTBI and clinicalTBI medians for the Body-5 mm D2% and MLD were also significant ( Table 3). In general, for all patient categories, the Body-5 mm D2% and MLD distributions for fffTBI were broader and shifted higher than those for clinicalTBI.

3.B | Plan parameters
The number of MUs planned per arc for the supine and prone orientations were summed to give the total MUs per fraction for fffTBI and clinicalTBI. Beam-on time was deduced based on the gantry speed at all control points over all arcs. The average dose rate at mid-plane was the prescribed dose divided by the beam-on time.
The instantaneous dose rate at mid-plane directly under isocenter is 250-300 and~100-130 cGy/min for fffTBI and clinicalTBI respectively, and is dependent on the patient AP width and MLC

3.C | Plan performance with set-up errors
Systematic set-up errors were modeled in Eclipse by shifting arc isocenters of the fffTBI for 18/47 patients. Differences between the Body-5 mm D98% and D2%, and MLD for the shifted and original plans are summarized in Table 5  Six sets of supine and prone standardized MLC modulations were planned, and patients were assigned a plan based on AP width and height.
The change from using a standard linac beam with the flattening filter inserted to a FFF beam caused variations in the treatment plan dose distribution, as expected with a change in dose profile (Fig. 7).
Even with the modified dose distribution, the same dosimetric criteria are generally met using either treatment plan. The Body-5 mm D98% had similar distributions for the fffTBI compared to the  MV FFF VMAT plans, which is an area we wish to investigate in the future with respect to 10 MV FFF VMAT. Furthermore, the dose profile of 10 MV FFF is more peaked than 6 MV FFF, meaning that plans using 10 MV FFF would likely need even more MLC modulation to achieve the required dose homogeneity. Highly modulated plans are impacted more by set-up errors, and would not take full advantage of the increased dose rate. It is therefore unlikely that moving to a higher energy would be advantageous for the patient cohort in this work. However, 10 MV FFF plans may provide an overall benefit for patients with AP widths greater than what was examined here.
It is expected that patient set-up would be more critical when using FFF beams due to the accompanying increase in dose profile variation. In the case of TBI treatments, this is an especially important consideration as they often occur at an extended SSD without the use of image guidance. The dosimetric impact of having a large set-up error of 2 cm was at most 3.0% for the lateral and cranialcaudal directions. An SSD mismatch was more substantial, with changes between the shifted and original FFF plans being up to 5.0%. This indicates that care must be taken when moving the patient to the correct SSD. Similar trends were seen when examining the plan robustness to set-up uncertainties for our current clinical technique. 8 Overall, the fffTBI shows unanticipated robustness to set-up errors, which is likely due to the MLCs decreasing the FFF dose profile peak and essentially flattening the dose profile.
The ultimate goal of using FFF beams in TBI VMAT treatment planning was to reduce treatment time to increase patient comfort.
This objective would theoretically be achieved, with an average decrease in beam-on time of 39.2% when moving to FFF treatment plans ( Table 4)  we currently treat at much higher dose rates than those often used in published data (instantaneous dose rates of~100 cGy/min versus ≤15 cGy/min respectively), and an internal retrospective analysis has found no instances of radiation pneumonitis. Clearly the pathogenesis of lung toxicity during HSCT is multifactorial and the role of radiation dose rate remains unknown.
A definitive relationship that has been acknowledged is that of pulmonary complications and total lung dose. 19,21 Lung shielding is often seen for both single-dose and fractionated schemes to reduce lung dose and toxicity. Della Volpe et al have suggested MLD limits of approximately 9 Gy for a 10 Gy in three fractions and 5.5 cGy/ min dose rate regimen to reduce the rate of lethal pulmonary complications to less than 5%, 22 with this being the restriction many centers place on lung dose for a 10 Gy total body prescription currently. This is also in agreement with the lower incidences of lung toxicity for reduced-intensity conditioning HSCT 23 which generally uses radiation prescription doses that are less than those seen in conventional TBI, though the influence of reduced-intensity chemotherapy on these toxicities must be recognized.
Though no patients have been treated with the fffTBI yet and there are no data on patient outcomes using this technique, we believe that high dose rate treatments for low-dose TBI would be safe to perform due to the reasons outlined above. The TBCC's regimen results in lung doses that are far below those suggested to reduce radiation-induced lung toxicity risk to an acceptable level, and modifying the treatment plans to FFF beams did not alter the MLD to a level of concern even though it was significantly higher for fffTBI (P = 0.043). Note:: D98% and D2% both correspond to the Body-5 mm structure, and mean lung dose (MLD) corresponds to the mean dose to the lung structure.

| CONCLUSION S
FREDERICK ET AL.