In search of a one plan solution for VMAT post‐mastectomy chest wall irradiation

Abstract Purpose This study was designed to evaluate skin dose in both VMAT and tangent treatment deliveries for the purpose of identifying suitable bolus use protocols that should produce similar superficial doses. Methods Phantom measurements were used to investigate skin dose in chest wall radiotherapy with and without bolus for 3D and rotational treatment techniques. Optically stimulated luminescence dosimeters (OSLDs) with and without housing and EBT3 film were used. Superflab (3, 5, and 10 mm) and brass mesh were considered. Measured doses were compared with predictions by the Eclipse treatment planning system. Patient measurements were also performed and the bolusing effect of hospital gowns and blankets were highlighted. The effect of flash for VMAT plans was considered experimentally by using 2 mm couch shifts. Results For tangents, average skin doses without bolus were 0.64 (EBT3), 0.62 (bare OSLD), 0.77 (jacketed OSLD), and 0.68 (Eclipse) as a fraction of prescription. For VMAT, doses without bolus were 0.53 (EBT3), 0.53 (bare OSLD), 0.64 (jacketed OSLD), and 0.60 (Eclipse). For tangents, the average doses with different boluses as measured by EBT3 were 0.99 (brass mesh), 1.02 (3 mm), 1.03 (5 mm), and 1.07 (10 mm). For VMAT with bolus, average doses as measured by EBT3 were 0.83 (brass), 0.96 (3 mm), 1.03 (5 mm), and 1.04 (10 mm). Eclipse doses agreed with measurements to within 5% of measurements for all Superflab thicknesses and within 15% of measurements for no bolus. The presence of a hospital gown and blanket had a bolusing effect that increased the surface dose by approximately 10%. Conclusions Results of this work allow for consideration of different bolus thicknesses, materials, and usage schedules based on desired skin dose and choice of either tangents or an arc beam techniques.

generally excludes it as the first option, 1 VMAT planning techniques have gained some traction in recent years for treating difficult-toplan post-mastectomy and intact breast cases. [2][3][4][5] The primary advantage of VMAT over 3D techniques is sparing of heart and ipsilateral lung from high doses without sacrificing target coverage. This advantage comes at the price of higher dose to contralateral breast and lung and even higher low dose to ipsilateral organs at risk.
Differences in dosimetry between VMAT and tangential treatments are largely the result of the differences in incident angles of the beam. In tangential treatments, the beams are incident from two angles, medial and lateral. In VMAT planning, the planner may choose to limit more en face components of the beam by restricting the ranges of arc angles 6 or they may choose to use an arc extending from medial to lateral angles and let the optimization objectives drive the weighting of tangential versus en face components. 2 Regardless of the planning technique, there will be more en face weighting of the incident beam angles in VMAT compared with 3D conformal tangents.
The presence of the en face component raises questions regarding skin dose (including the role of bolus) and the importance of flash in VMAT planning. Historical treatments involving tangential fields with wedges facilitated an intuitive appreciation for how flash conferred reduced sensitivity to dose variation related to breathing motion, swelling, and setup uncertainty. Since forward planned IMRT effectively reproduces wedged tangents behavior (to a first approximation), similar understanding for field-in-field treatments exists. The variable gantry angle delivery and aperture modulation that characterizes VMAT treatment deliveries, however, makes prediction of the effects of flash more difficult. These changes in treatment characteristics also make it difficult to assume that skin dose either without or under a bolus would be the same for the two treatment techniques, potentially due to variability in path length through the skin and/or bolus as a function of gantry angle. To complicate matters further, a consensus definition of skin is difficult to establish. The ICRU states that the superficial layers of interest include the dermal lymphatics (to a depth of~1 mm) and the basal cell layer at about 70 microns. 7 Practical dosimetric quantities extracted from treatment planning systems are often on the order of 2 mm in thickness. 8 For our purposes, we define herein the dose reported from in vivo dosimeters as representative of skin dose and we evaluate their behavior under different irradiation conditions. Regardless of planning technique, to date, there has been a lack of consensus on whether the routine use of bolus in post-mastectomy radiation therapy is necessary or not. 9,10 The guidelines of the American Society of Clinical Oncology first published in 2001 11 and later updated in 2016 12 stated that "whether it is necessary to apply the bolus every day, less frequently, or at all is uncertain." As such, whether bolus is used routinely or not, its thickness and frequency are often decided by clinical experience and vary from center to center. 13 Regardless of whether bolus is used or not at a clinic, its thickness and frequency, when using a VMAT technique it might be desirable to match the skin dose to that which is consistent with clinical practice as established by the three-dimensional technique at that center.
To date, only a limited number of studies have investigated skin dose in VMAT treatments of chest wall. 14 Absent from the literature is a systematic study of the impact of different types and thicknesses of bolus on skin dose for both 3D conformal and VMAT treatment techniques. This study was designed to evaluate skin dose in both VMAT and tangent treatment deliveries for the purpose of identifying suitable bolus use protocols that should produce similar superficial doses. Skin dose in this setting is evaluated with three in vivo dosimetry measurements: Gafchromic film, optically stimulated luminescence dosimeters in their jackets, and OSLDs without their jackets. The second goal was to evaluate the effects of flash on dose variation caused by breathing motion in VMAT post-mastectomy radiation therapy. Here the goal was to determine whether the implementation of flash was necessary or not since the presence of more en face beams might make the distribution less susceptible to changes in outer body contour position due to breathing.

2.A | Phantom
A replica of a left-sided chest wall CT set of a patient was 3D printed using PLA and a Lulzbot Taz5 MOARstruder, 100% infill ( Fig. 1). The scan of the phantom showed a physical density of 1.1 g/cm 3 and Hounsfield unit of 160. The phantom contains an insert to hold a PTW-60019 microDiamond (PTW_Freiburg) detector. The microDiamond detector was chosen due to its shallow effective point of measurement (1 mm) and angular independence. 15 This detector was only used for relative measurements and assurance that the phantom was set up reproducibly each time a measurement was repeated. No absolute dose readings were acquired using this detector.

2.B | Planning
A CT image of the 3D-printed phantom was taken with 2.5 mm slice spacing. Target and organ at risk volumes were drawn to permit creation of VMAT plans. A list of plans that were created by one experienced planner (Eclipse 13.6, AAA, 2 mm calculation grid) is given below. An identifier for each plan is provided using the following format: Technique:#Bolus:(±)#Cropping. Technique refers to VMAT or  Table 1. Although the prescription is to a point in tangents and a volume in VMAT, both techniques follow the PTV_EVAL coverage goals listed in Table 1. The VMAT plans included three arcs, the stop and start angles are shorter by 10°medially (310°for VMAT) and wider by 40°laterally (165°for VMAT) compared with tangential fields. Collimator angles are 20°, 340°, and 355°. The x jaw setting was roughly 16 cm for all three arcs and jaw tracking was used.

2.C | Phantom doses
All measurements were performed on a TrueBeam TM linear accelerator. Cone beam CT images were acquired to ensure accuracy of phantom setup. Skin doses were evaluated with Gafchromic film (Ashland Advanced Materials, Bridgewater, USA) and nanoDot™ OSLDs (Landauer Inc., Glenwood, USA). OSLD measurements were performed with the dosimeter in its housing and with the dosimeter extracted from its housing (to minimize the effective depth of measurement compared to film 16,17 ). Data are clearly labeled with respect to the OSLD configuration in the subsequent sections.
To quantify representative skin doses, for each plan and bolus combination, we placed a strip of GafChromic EBT3 film of size 9 × 4 cm 2 on the phantom [ Fig. 1(a)] such that the longer side extended in the medial to lateral direction. The strip of film was then taken off and replaced with three OSLDs as indicated in Figure 1 and the plan was delivered twice more, once with three OSLDs in their housing and once with three OSLDs taken out of their housing. Care was taken to keep the room dark, not touch the OSLD, and place them back in their housing as soon as the treatment was delivered.
Scanned EBT3 images were measured using Film QA Pro software (Ashland Advanced Materials, Bridgewater, USA). The methodology described in Ref. [18] was used to analyze the film readings. OSLDs were read with an InLight microStar reader (Landauer Inc., Glenwood, USA). The effects of supralinearity (2.5-3%) were corrected for OSLD readings of doses larger than 250 cGy. 19  We adopted the approach of single optimization and renormalization as described here in the interest of clinical efficiency.

2.D | Patient data
We performed in vivo readings on 10 QOD clinical VMAT: 10 mmBolus:0mmCrop VMAT patients with OSLDs in their housing on bolus (10 mm) and no bolus days. The OSLDs were placed on patient skin in a manner similar to Fig. 1(a). Similar to the phantom study on no bolus days, the patients received the same optimized plan but with the monitor units adjusted down by 5%. Since the majority of the patients at our institute use a gown and a hospital blanket, we repeated our phantom study in the presence of a hospital gown and knitted blanket. This was done to establish if the presence of these materials acts as a bolus and affects the baseline doses established by the phantom.

| RESULTS
All skin doses are reported as a fraction of prescription dose, that is, skin dose of 0.6 means detector reading for one fraction is equal to 60% of 266.7 cGy which is the prescribed daily dose.

3.B | VMAT
In contrast to the field in field tangent plans, the doses across the films in the VMAT plans were effectively uniform as shown in Fig. 3 (a). Furthermore, they are uniformly lower than the tangent plans.
This is likely attributable to the greater degree of en face beam delivery in VMAT with its concomitant skin sparing.
We found that moving the    The results of our in vivo study and the subsequent investigation into the bolusing effects of hospital gown and blankets, emphasize the importance of understanding the bolusing behavior of these materials.

3.C | Patient data
In the absence of bolus, we measured approximately 10% (of the prescription) increase in dose when a gown and blanket were both used.  presented here, but the same is necessarily true of any model of postmastectomy radiation therapy skin dose.

| CONCLUSIONS
Clinical experience with skin dose in a post-mastectomy radiation therapy setting has historically been derived from treatment plans that make use of tangential beam arrangements. The data in this study indicate that VMAT treatments can produce significantly different results depending on the bolus material and thickness. With the materials in this study, it was not possible to identify a single bolus policy (i.e., same material, thickness, and frequency) that pro-

ACKNOWLEDGMENTS
The authors thank Jim Clancey and Jim Allan for their help with phantom design and printing. They thank Scott Dube for thoughtful discussions on the topic of chest wall bolus.