Session Item

Conventional lung SBRT requires large treatment margins to cover tumor motion resulting from respiration. This may avoid underdosage but increases toxicity risks. To maximize healthy tissue sparing, we previously developed MRI-guided MLC tumor tracking for the 1.5 T Unity MR-linac (Elekta AB, Stockholm, SE) in combination with IMRT. Recently, we also piloted VMAT deliveries on Unity to further maximize plan conformality and delivery efficiency. In this study, we demonstrate the feasibility of a first experimental setup on an MR-linac that combines VMAT with MLC-tracking for a range of lung SBRT indications. 

All experiments were performed on a 1.5 T Unity MR-linac in research mode.

A Quasar MRI4D phantom (ModusQA, London, CA) was used to generate: no motion (static reference), Lujan motion (cos⁴, peak-to-peak amplitude A = 20 mm, f = 0.25 Hz), and subject-derived real respiratory motion (average A = 11 mm, average f = 0.33 Hz) with an average baseline drift of 0.6 mm/min. The phantom contained a film insert with a 3 cm spherical target (GTV) that could be positioned centrally or 10 cm off-center (peripheral) in a water-filled body oval. Target positions were continuously estimated from 2D cine-MR (4 Hz). A linear regression prediction filter compensated for system latency. Predicted positions were used continuously to realign the MLC with the target position.

We created three VMAT treatment plans with 3 mm GTV-to-PTV margins following the clinical planning template for lung SBRT: a central plan (8x7.5 Gy) and two peripheral plans (3x18 Gy and 1x34 Gy). Gafchromic EBT3 or EBTXD films were used to measure the delivered dose. A 1%/1mm local Gamma-analysis quantified dose differences between the static reference and tracking cases. Additionally, the dose area histogram (DAH) was determined for the target.

The VMAT plans had a conformity index (prescribed isodose volume/ PTV) of 1.4-1.5 and an MU-weighted mean-field area of 13-16 cm². Treatment delivery times were: 6.7 min, 13.1 min, and 24.2 min, for the 8x7.5 Gy, 3x18 Gy, 1x34 Gy lung SBRT plans respectively. The plans required an RMS leaf speed of 0.5-0.7 cm/s. Tracking required a maximal additional 2.4 cm/s leaf speed. Each plan was delivered in respectively 2, 4, and 6 arcs.

The local gamma analysis for the central delivery shows that MLC-tracking improved the gamma pass-rate from 67.5% to 98.3% for Lujan motion and to 94.2% for the real respiratory trace. For peripheral deliveries with real respiratory motion, the 3x18 Gy delivery had a 97.3% pass-rate and the 1x34 Gy delivery had a 96.8% pass-rate (Fig.1).

The DAH (Fig.2) shows that the target dose agrees well between static and tracking deliveries with real respiratory motion. The figure also shows that the minimum dose in the target is well above the prescribed dose. 

We provided a first experimental demonstration of the technical feasibility of VMAT combined with MR-guided MLC-tracking for central and peripheral lung SBRT.