METHODS AND MATERIALS
Ten pediatric patients (age range, 3 – 17 y.o.) treated with the VMAT-TBI technique on the Varian TrueBeam from November 2019 to August 2020 were included in this IRB approved study. Patients were treated with either non-myeloablative (2 Gy in one fraction) or myeloablative (12 Gy in six fractions) regimens. Corresponding 2D plans replicating our current clinical setup with lung blocks and chest-wall boost were created for all ten patients for dosimetric comparison between the 2D and VMAT techniques.
For VMAT-TBI, patients were simulated head-first supine (HFS) position in a CIVCO long vac-lok bag positioned on the in-house developed rotational platform. Patients’ necks were extended resting on the AccuForm cushion (CIVCO) neck support, arms tightly clasped straight alongside the body, and the CIVCO knee fix and/or feet fix were placed under patient’s knees for comfort and leg position reproducibility. The in-house-developed auto-planning script within Eclipse v15.6 Application Programming Interface (API) was used for treatment planning16. The VMAT plans were auto-generated with three isocenters (head, chest/abdomen, pevis/legs) in HFS position. Additional AP/PA plans with 1 to 2 isocenters in the feet-first supine (FFS) position were used if patient height exceeded 115 cm (due to the longitudinal table limit in HFS position). Upper-body VMAT plans using 6 MV or 10 MV were optimized with all isocenters included in one plan with at least 2 cm overlap between the fields. The AP/PA Upper Leg plan was used a baseline dose in the optimization to homogenize the dose distribution near the matchline area. The auto-feathering optimization option was turned on to create smooth dose gradients in the fields’ overlapping areas and to prevent extreme dose heterogeneity in the event of larger setup variations. ​ The VMAT plans were optimized and normalized to achieve at least 90% coverage of the whole body PTV with the prescription dose. The PTV volume was created by cropping 3 mm from skin and critical normal tissues.
For each VMAT-TBI plan, a simulated 2D plan was developed in Eclipse based on the VMAT-TBI CT scan. The 2D plan fields were setup for AP/PA technique at an extended SSD (~608 cm), collimators at 45° and 135°, field sizes set to 40 x 40 and 15 MV energy. Umbilicus was used as the isocenter. Screen-to-skin distance, separation and off-axis measurements that are typically acquired during clinical setup were measured based on the CT scan. The compensator layers were re-created on the CT scan to homogenize dose distribution based on the following midline points: head, chin, neck, suprasternal notch (SSN), xiphoid, umbilicus, hip/pubis, thigh, knee, calf, and ankle. Lung blocks were generated using the lung contour contracted by 1 cm (lungs-1cm) and 1 cm below clavicle with the constant thickness of 2.5 cm of Cerrobend for every patient. The block transmission measurements in the middle of 10 cm thick lung slab sandwiched in-between two 4-cm thick solid water slabs was 50% for 15 MV beam. Electron chest wall (CW) boost fields were created for all patients prescribed to 50% of TBI photon prescription to depth of maximum dose. The accuracy of the dose calculation was verified using the ion chamber measurement in the phantom composed of the lung slab sandwiched between the solid water slabs.
The clinically relevant metrics in this study, including plan global Dmax, PTV D90%, PTV V110%, lungs and lungs-1cm Dmean, were analyzed and compared between the VMAT-TBI and simulated 2D plans. For gonadal sparing comparison, the VMAT-TBI plans were compared to 2D plans for a girl (Patient 4) and a boy (Patient 3) assuming 5 cm lead shield for testes/ovaries in front of the patient with 5 mm margin. All dosimetric comparisons between the VMAT and 2D plans were made with the dose expressed as a percentage of the prescription dose and the volume expressed as a percentage of the PTV volume. Paired t-test was used to compare the dosimetric indices between the VMAT and 2D TBI plans with p-values less than 0.05 considered to be statistically significant.