Effect of the Geometric Uncertainty of Jaw Positioning on the Use of Volumetric Modulated Arc Therapy in Stereotactic Radiosurgery.

Background and purpose Volumetric modulated arc therapy (VMAT) with a C-arm linear accelerator has been adopted for stereotactic radiosurgery (SRS) for treating brain tumors. Some treatment planning systems (TPSs) generate sequences that align the jaw with the edge of the multileaf collimator (MLC) radiation field during VMAT. However, the jaw exhibits greater geometric positioning uncertainty than the MLC. In this study, we investigated the effect of jaw positioning uncertainty on dose distribution in VMAT-SRS and evaluated the effectiveness of our proposed method. Methods The RayStation software (RaySearch Laboratories, Stockholm, Sweden) was used for the TPS and a TrueBeam STx linear accelerator (Varian Medical Systems, Palo Alto, USA) for the C-arm linear accelerator. A target simulating a brain tumor was placed in the phantom, for which three treatment plans implementing VMAT-SRS were created: one using the jaw tracking technique (JT plan), another with a fixed jaw (FJ plan), and a third with the jaw fixed 1 mm outward from the edge of the MLC (FJ1mm plan). The change in the dose-volume parameters relative to the original plan was evaluated when the jaw position at each control point in each plan was systematically changed by ±1 mm. Results The maximum changes in the absolute dose received by 99% of the gross tumor volume from the original plans were -7.0%, -5.6%, and -1.1% in the JT, FJ, and FJ1mm plans, respectively. The maximum changes in the absolute dose received by 99% of the planning target volume were -13.1%, -12.0%, and -2.2%, respectively. The ranges of change in the absolute volume of the normal brain receiving a dose greater than 12 Gy for the original plans were -1.3 to 0.6 cm, -0.7 to 0.4 cm, and -0.4 to 0.1 cm, respectively. Conclusion Even when the quality control levels recommended by the jaw positioning guidelines were met, target dose variations of >10% were observed depending on the existing VMAT-SRS. Our proposed method was the most robust, with a target dose variation of <3%.