SABR for Skull Base Malignancies: A Systematic Analysis of Set-Up and Positioning Accuracy.

Abstract

Here we provide an analysis of the set-up and positioning accuracy of SABR for skull base malignancies to evaluate the use of site- or axis-specific margins to reduce field size. Data were prospectively collected on 63 patients who received 304 fractions of SABR for recurrent/previously irradiated skull base tumors. Using our custom cushion-mask-bite-block immobilization system combined with ExacTrac x-ray and cone beam computed tomography (CBCT), set-up, residual, CBCT-positioning agreement, and intrafractional errors were measured. The resulting planning target volume (PTV) margins were estimated across 4 skull base subsites: anterior (group 1), central (group 2), posterolateral (group 3), and skull base-associated sites (eg, nasopharynx/retropharyngeal, cervical vertebrae 1-2, occiput) (group 4). On initial set-up, 66% of treatment courses required shifts of >2 mm or >2°, necessitating 4.9 mm PTV margins without image guidance. After correction, only 6 of 304 treatment sessions had residual errors >1 mm. CBCT-ExacTrac agreement was ≤1 mm in 89.1% of treatments and ≤1.5 mm in all but 1 session. Group 4 showed a higher rate of >1 mm or >1° CBCT-positioning differences compared with other groups (24.5% vs 7.8%; P = .0001), and the greatest variations occurred in the craniocaudal translational and the pitch rotational axes. Overall calculated PTV margins (based on intrafractional error) were 1.5 mm across subsites except for group 4, which required 2.0 mm margins. The use of 2.0 mm PTV margins for skull base SABR appears feasible using ExacTrac x-ray as the sole imaging modality for most subsites. However, PTVs were not uniformly equal, and the use of a site-specific nonuniform margin reduction to optimize critical-organ dose sparing may be feasible for select cases. These findings warrant clinical investigation.