Prostate Motion and Set-up Error Progression During the Treatment Session Studied by Electronic Portal Imaging in Cine Mode

Abstract

Image-guided prostate localization and couch correction before each radiotherapy treatment fraction has become a routine clinical practice in many radiotherapy departments in order to account for interfraction prostate motion. However, the prostate may also move during each treatment fraction. Determination of appropriate set-up margins requires knowledge about the set-up error remaining after prostate localization and its progression throughout the treatment session as a result of prostate motion. These issues were addressed in the present study by cine mode portal imaging of prostate cancer patients with implanted fiducial markers. Twenty patients with localized prostate cancer received a 5-field conformal treatment of 78 Gy in 39 fractions with daily marker-based prostate localization using an on-board imager system. Cine mode portal imaging of the entire treatment was acquired at three early treatment fractions followed by a weekly treatment fraction, in total 10 fractions per patient. The set-up error at treatment start and the prostate motion during the treatment session were evaluated from the fiducial marker positions on the portal images. At the onset of the treatment, the standard deviation (SD) of the set-up error was 1.0 mm in the lateral direction and 1.6 mm in the craniocaudal (CC) direction. The SD did not depend on the duration of the image-guided set-up procedure (median time 2.5 minutes, time span 1.25–25 minutes). After the first two treatment fields (approximately 2 minutes) the SD of the CC set-up error stabilized at 2.0 mm. Set-up errors momentarily exceeding 8 mm were observed. While the distribution of prostate deviations from the initial position at treatment start broadened throughout the three first treatment fields, it stabilized during the remaining two treatment fields. The data showed that the intrafraction prostate motion at the first fractions was not a reliable predictor for the prostate mobility during the rest of the treatment course. The distribution of set-up errors residual after the image-guided prostate localization broadened only slightly during treatment delivery. The findings suggest that the additional population based average set-up margins to account for prostate movement during treatment are small. However, in individual patients large intrafraction prostate motion may occur.