Clinical Implementation of Robust Multi-Isocentric VMAT for Craniospinal Irradiation

Abstract

To implement a robust multi-isocentric VMAT technique for craniospinal irradiation (CSI), evaluate the dosimetric effect of simulated set-up errors on clinical treatment plans, and quantify its delivery accuracy. Brain and spinal clinical target volumes (CTV) were defined and expanded with variable margins to spinal and combined planning target volumes (PTV). In patients still growing, the vertebral column was contoured and planned to achieve a homogeneous dose. CSI VMAT plans were produced using two or three isocenters, one in the brain and one or two in the spine, depending on the combined PTV length. One partial VMAT arc was used to treat the brain, avoiding highly attenuating couch regions, and three partial arcs were used at each isocenter in the spine, avoiding the arms and shoulders. Overlap of beams from different isocenters was restricted to the neck and, if necessary, mid spine. Additional robust objectives were used to optimize the dose accounting for isocenter position errors of up to 5 mm superoinferiorly. The final clinical plan was evaluated for robustness by simulating systematic shifts of the spine isocenter(s) by ±3 mm and ±5 mm superoinferiorly. Verification of plan delivery was performed using a 3D diode array prior to treatment. CSI VMAT was implemented in May 2019, and 9 patients started treatment prior to February 2020. Patients were diagnosed with medulloblastoma (6/9), pineoblastoma (1/9), choroid plexus carcinoma (1/9), and high grade neuroepithelial tumor (1/9). Median (range) age at treatment was 15 (3 – 32) years and combined PTV length was 61.8 cm (46.8 – 75.4 cm). CSI prescription dose was 36 Gy (23.4 – 39.6 Gy) in 1.8 Gy (1.67 – 1.8 Gy) fractions. Combined PTV D0.1cc increased from a median relative dose of 105.0 % (103.2 - 107.7 %) to 113.7 % (107.8 - 120.2 %) when the superior spine isocenter was moved 3 mm closer to the brain, and to 119.5 % (111.3 - 128.2 %) when moved by 5 mm. Spine PTV D99% reduced from 96.2 % (95.4 - 97.5%) to 90.8 % (85.9 - 93.3 %) when the superior spine isocenter was moved 3 mm further away from the brain, and to 87.4 % (78.8 - 90.1 %) when moved by 5 mm. For the 5/9 patients planned with three isocenters, the combined PTV D0.1cc increased from 104.7 % (103.2 - 107.7 %) to 107.1 % (106.3 - 107.8 %) when the inferior spine isocenter was moved 3 mm superiorly, and to 109.2 % (108.5 - 111.7 %) when moved 5 mm. Spine PTV D99% reduced from 96.8 % (95.9 - 97.5 %) to 95.2 % (94.8 - 95.4 %) when the inferior spine isocenter was moved 3 mm inferiorly, and to 92.7 % (92.6 - 93.7 %) when moved 5 mm. Plan verification results passed local tolerances (greater than 95 % of pixels achieving a gamma value < 1 for global 3 % and 3 mm criteria with 5% low dose threshold) in all cases. Multi-isocentric CSI VMAT has been implemented clinically with acceptable robustness to systematic errors in superoinferior isocenter position and accurate treatment delivery.