Magnitude and Agreement of Cone Beam Computed Tomography Shifts for Patients With Intact and Postoperative Prostate Cancer

Abstract

To assess the magnitude and variability of pre-treatment alignment shifts between therapists for prostate cancer patients treated with IGRT/IMRT and positioned using CBCT. Eighteen patients treated for intact (IP, n = 8) and postoperative (PP, n = 10) prostate cancer during 2010 were selected. Patients with prostate markers were excluded. For each patient, daily CBCT images were reviewed and two representative images were selected, one in which the bladder resembled the planning CT (B0) and one in which the bladder was different from the planning CT (B1). Therapists viewed each of the 36 images in scanning acquisition position and aligned the cone beam to the planning CT as if they were preparing to treat the patient. Shifts in the AP, SI and LR dimensions were recorded. One-way ANOVA was used to compare the magnitude of shifts in the SI, AP, and LR directions. Repeated measures ANOVA was used to assess the variability between therapists in positioning the same images. Two way ANOVA was used to compare ranges between IP and PP patients and between B0 and B1. Eight therapists participated in the study (mean years experience with CBCT: 3.6, range: 2-6). Eighteen patients, each with two images, were included for a total of 288 shifts. Average magnitude of shifts was greater in the AP (3.6 mm) and LR (3.4 mm) direction than in the SI direction (1.2 mm, p<0.001). There was no difference between the average magnitudes in the AP vs. LR direction (p = 0.66). There was significant variation among therapists for shifts in the AP direction (p<0.001), but not in the SI (p = 0.29) and LR (p = 0.10) directions. There were no significant differences in the average AP, SI, or LR shifts for IP vs. PP (p = 0.23, 0.58, 0.09) or for B0 vs. B1 (p = 0.50, 0.43,0.56). The mean range to encompass the shifts for all 8 therapists in the AP, SI, and LR directions were 7.7 (min: 3, max: 19), 4.7 (min: 1, max: 11), and 2.7 mm (min: 1, max: 5). The mean range was smaller for the B0 image in the AP (6.7 vs. 8.6 mm, p = 0.09), SI (4.2 vs. 5.1 mm, p = 0.27) and LR (2.3 vs. 3.1 mm, p = 0.04) directions, but this was only statistically significant in the LR direction. The mean range was smaller for IP patients in the AP (7.5 vs. 7.8 mm, p = 0.81), SI (4.5 vs. 4.9 mm, p = 0.63), and LR directions (2.5 vs. 2.9 mm, p = 0.36), but this was not statistically significant. Significant variation exists in the AP direction between different therapists making CBCT shifts of the same images. Greater variation in the AP direction may be due to deformation of the rectal wall by varying amounts of stool or bowel gas, which can create uncertainty in the match. Careful evaluation of bladder filling to resemble sim CT may decrease shifts. Larger PTV margins may be needed to account for variation between therapists when using CBCT for IGRT.