Patients, simulation, and contouring
This study was approved by our ethics committee, and written informed consent was obtained from all patients. Between April 2017 and April 2020, ten patients with angiosarcoma of the scalp were treated with non-coplanar VMAT at our institute. All patients were chosen for analysis in this study. These patients were aged 56–86 years (median: 76 years) and immobilized with thermoplastic head masks in the supine position. CT simulations (Revolution HD, GE Medical Systems, Milwaukee, WI) were performed with a 2.0-mm slice thickness and 500-mm field of view with dimensions of 512 × 512 pixels. All the CT images were transferred to a treatment planning system (TPS) (Eclipse version 15.5; Varian Medical Systems, Palo Alto, CA).
The target volume and OARs such as the brain, brain stem, optic nerve, chiasm, and lens, were delineated by experienced radiation oncologists in our institute. Contouring of clinical target volume (CTV) included the entire scalp bordered by the face anteriorly and the neck to the sides and posteriorly. The planning target volume (PTV) was derived from CTV plus a symmetrical 5-mm margin. Mean PTV size and standard deviation were 518.2 ± 116.6 cm3 (range: 362.8–678.6 cm3). To achieve high target coverage, this study employed the use of a 1-cm virtual bolus which was added as structure in an Eclipse and applied to the surface of the skin around the PTV. This bolus was used for the optimizations and subsequent final dose calculations.
Treatment planning
All treatment plans were performed using two photon beams of a linear accelerator, (TrueBeam Edge, Varian Medical Systems) equipped with high definition multileaf collimators (MLCs), the widths of which were 2.5 mm for the first 32 leaves from the central point and 5 mm for the rest. For the purpose of comparison, different treatment plans were designed for three techniques, including non-coplanar VMAT with flattening filter (FF) beam (VMAT-FF), HyperArc with FF beam (HyperArc-FF), and HyperArc with FFF beam (HyperArc-FFF), in all patients. The VMAT-FF is a clinical plan and the others are not. All plans were optimized with a photon optimizer (PO) ver.15.0 and calculated with the analytical anisotropic algorithm (AAA) for dose calculation with inhomogeneity corrections on the 2 mm grid size. Prescribed dose, which was determined as the mean dose for PTV, was 35 × 2 Gy (70 Gy).
VMAT plan
All VMAT-FF plans were generated using a 6 MV photon beam at a maximum dose rate of 600 MU per minute with non-coplanar arc fields. The number of isocenters, beam angle, field size and collimator angle were manually selected depending on the size and location of the target. Beam parameters, such as couch angle, collimator angle, and arc length, for each arc in VMAT-FF are summarized in Table 1. In the optimization process, the jaw tracking function was used. Moreover, normal tissue objective (NTO) and objectives for PTV were maintained at constant values to avoid bias. The optimization goals and constraints in the VMAT-FF plan are summarized in Table 2.
HyperArc plan
The HyperArc-FF and HyperArc-FFF plans were generated using a 6 MV photon beam at a maximum dose rates of 600 and 1400 MU per minute, respectively. The isocenter and field size were automatically determined based on the target structure in the HyperArc planning. In addition, four arc fields, three of which were non-coplanar, were automatically arranged as follow: one full or half coplanar arc without couch rotation and three half non-coplanar arcs with couch rotations of 315°, 45°, and 90°. The HyperArc plan used collimator angles of 5°, 345°, 15°, and 45° in the beam with couch rotations of 0°, 315°, 45°, and 90°, respectively (Table1). In the optimization process, the jaw tracking function was used. Moreover, the SRS NTO was used, which was designed to generate treatment plans that featured steep dose decay in space from target-specific dose levels to low asymptotic dose levels. The optimization goals and constraints in the HyperArc plan are summarized in Table 2.
Dosimetric comparisons
The treatment plans were evaluated according to the standard dose volume histograms (DVH) for VMAT-FF, HyperArc-FF, and HyperArc-FFF. For the evaluation of target dose, the homogeneity index (HI) was defined as follows: HI = (D2% - D98%)/D50%, where D50% is the median absorbed dose and D2% and D98% represent the doses received by 2% and 98% of the PTV. The conformity index (CI) was defined as follows: CI = (TVPV×TVPV)/(TV×PV), where TVPV, TV, and PV represent the volume of the target covered by the prescription dose, target volume, and prescription isodose volume, respectively. For normal brain tissues excluding the PTV, the volumes that received a specific dose in a range of 10 to 60 Gy (V10Gy− V60Gy) and the mean dose were compared. In addition, the doses receiving 0.1 cc of the volume for surrounding critical organs, such as the brain stem, chiasm, optic nerve, and lens, were evaluated. Total MUs and beam-on time (BOT) were compared for the three plans.
To evaluate the complexity of the MLC patterns, the modulation complexity score for VMAT (MCSV) for each plan was calculated using our in-house software (MATLAB R2016a; MathWorks, Natick, MA), and the overall MCSV was defined as the mean of the MCSV for each treatment beam. The MCSV was calculated based on the leaf sequence variability (LSV) parameter and aperture area variability (AAV) as described by Masi et al. [15]:
where MUcpi, i+1 indicates the number of MUs delivered between 2 successive control points (namely, CPi and CP(i+1)). The value of the MCSV decreases with an increase in modulation complexity.
Dosimetric verification
In all treatment plans, dosimetric verification was performed by using the electronic portal imaging device (EPID, aS1200 flat panel detector, Varian Medical Systems) mounted on the linear accelerator. The square pixels of the EPID had a side length of 0.34 mm, which yielded a total area of approximately 40 × 40 cm2 (1190 × 1190 pixels). All EPID images were obtained in the integrated acquisition mode without any obstructions at the source-to-imager distance of 150 cm and 170 cm for FF and FFF beams, respectively. The acquired images were automatically retrieved to a commercial software (PerFRACTION version 2.0.4, Sun Nuclear Corporation, Melbourne, FL) and compared against the baseline images, which were generated from the DICOM files of treatment plan from the TPS. The quantitative evaluation of the dosimetric accuracy was performed using gamma method. Analysis criteria of 3%/2 mm and 2%/2 mm in gamma was used above a 10% maximum signal threshold.
Statistical analysis
The paired Wilcoxon signed-rank test was performed on the data not following normal distribution (SPSS, version 24; IBM, Armonk, NY) for the statistical measurement of the differences between the following: VMAT-FF vs. HyperArc-FF, VMAT-FF vs. HyperArc-FFF, and HyperArc-FF vs. HyperArc-FFF. A p value below 0.05 was considered to indicate statistical significance.