Target Site Analysis of Set-up and Positioning Accuracy for Head and Neck Stereotactic Ablative Radiation Therapy (SABR)

Abstract

Stereotactic ablative radiotherapy (SABR) is utilized to address recurrent and/or previously irradiated head and neck cancers (HNC) and requires high set-up accuracy. We previously demonstrated excellent results with our custom cushion/mask/bite-block (CMB) immobilization system. Here, we compare set-up and positioning accuracy across different head and neck tumor locations to further optimize PTV margins for SABR treatment planning and delivery. We analyzed 375 treatment sessions across 73 patients with recurrent HNC treated with SABR using our custom CMB immobilization system. Initial set-up was performed using ExacTrac infrared tracking, and interfractional error was determined by comparing initial X-ray to the treatment plan. Patients were then shifted using ExacTrac and the residual set-up error was measured with repeat X-ray verification. Next, cone-beam CT (CBCT) was obtained and compared to ExacTrac for 3-dimensional positioning agreement. Finally, intrafractional positioning errors were measured by taking pre-beam X-ray. The data was stratified based on treatment site: paranasal sinus, nasopharynx, skull base (SB), cervical and parotid nodes (Neck), and oropharynx, oral cavity, larynx (Mucosal). The median dose and volume were 45Gy in 5 fractions (range: 27-47.5Gy, 3-5 fractions) and 17.12cm3 (range: 1.47-69.20cm3). There were no significant differences in the mean age, treatment volume, or monitor units (MU) between groups. The initial translational set-up errors were -0.3 ± 1.1mm, 0.3 ± 1.5mm, and -0.2 ± 1.4mm in the AP, CC, and LR directions, respectively. Site-specific mean 3D interfractional vectors were 2.2 ± 1.2mm (SB; P<0.00001 vs others), 2.9 ± 1.4mm (Neck), and 2.5 ± 1.7mm (Mucosal). Overall, the mean ExacTrac-CBCT agreement was between -0.1 and 0.1mm in translational and rotation directions, and overall residual errors were -0.1 ± 0.2mm, 0.0 ± 0.2mm, and 0.0 ± 0.2mm in the AP, CC, and LR directions, respectively. Site-specific mean 3D residual vectors were 0.4 ± 0.3mm (SB), 0.7 ± 0.7mm (Neck), and 0.5 ± 0.3mm (Mucosal), with a residual 3D vector >1.0mm occurring in 5.4% of SB patients compared to 26.1% (Neck) and 26.7% (Mucosal)(P=0.013). Intrafractional errors showed a mean 3D vector of 0.7 ± 0.5mm (SB; P<0.05 vs others), 1.3 ± 0.9mm (Neck), and 1.4 ± 1.5mm (Mucosal). An intrafractional 3D vector >1.0mm occurred in 16.2% of SB patients compared to 52.2% (Neck) and 46.7% (Mucosal) (P<0.0001). The calculated PTV margin for each site was ≤1.3mm (Skull Base), ≤2.2mm (Neck), and ≤3.4mm (Mucosal). In a large comparison of set-up and positioning accuracy for head and neck SABR, PTV treatment margins can be optimized depending on the treatment site. The error magnitude is significantly smaller for skull base targets and a 2-mm PTV margin may be sufficient. For mucosal targets, a 4-mm PTV margin may be necessary to account for internal organ motion.