Quantifying the Intra-fraction Treatment Variation in the Shift Off Bone in Fiducial-Based IG-SBRT for Pancreatic Adenocarcinoma

Abstract

Given the proximity of the radio-sensitive bowel, stereotactic body radiotherapy (SBRT) for pancreatic cancer requires precise immobilization and target localization, which can allow for steeper dose gradients and tighter margins. Currently, however, wide variation exists in the selection of planning target volume (PTV) margins that should be used in this setting. As intra-fraction variation represents a primary component of an appropriate PTV margin, we present our experience with intra-fraction shifts required during pancreatic SBRT, which has not previously been characterized in the literature. We analyzed 30 consecutively patients treated with SBRT for pancreatic cancer at our institution using volumetric modulated arc therapy (VMAT), active breathing control (ABC), and image-guided (IG) cone-beam computed tomography (CBCT). Patients were initially aligned to bony landmarks and then shifted to fiducials that had been endoscopically-placed. Halfway through treatment, an intra-fraction CBCT was obtained. Using the pre-treatment and intra-fraction CBCT, with the spine as a reference point, we calculated the intra-fraction shift in each axis along with a three-dimensional (3D) vector shift. The mean, maximum, and standard deviation of the intra-fraction shift in each direction, including the 3D vector, were calculated for individual patients across the five fractions. Population averages and standard deviations of each of these metrics were also calculated. Additionally, a linear regression model was developed to understand if the pre-treatment shift from bone to fiducial was related to the intra-fraction shift. On average, the mean intra-fraction shifts in the L-R, A-P, and S-I directions were 1.91± 0.82 mm, 1.62 ± 0.70 mm, and 2.33 ± 1.45 mm, respectively, and the mean intra-fraction shift as a 3D vector was 4.06 ± 1.51 mm. On average, the maximum intra-fraction shifts in the L-R, A-P, and S-I directions were 4.41 ± 2.37 mm, 3.51 ± 1.68 mm and 5.22 ± 3.79 mm, and the maximum intra-fraction shift as a 3D vector was 7.46 ± 3.70 mm. The linear regression model identified a statistically significant relationship between the mean pre-treatment shift from bone to fiducial and the mean intra-fraction shift in the A-P (p = 0.003) and S-I (p = 0.004) directions, along with for the 3D vector (p = 0.019). We report a single-institutional experience with intra-fractions shifts for patients undergoing SBRT for pancreatic cancer. These values can better inform appropriate PTV margins. As an example, to ensure a 95% likelihood that the target is encompassed within the PTV throughout treatment, radial margins of roughly 3-3.5mm and an SI margin of roughly 5mm should be considered. However, more data across institutions and using different motion management strategies is required for more generalizable results.