Effects of organ motion on proton prostate treatments, as determined from analysis of daily CT imaging for patient positioning.

Abstract

We quantified interfractional movements of the prostate, seminal vesicles (SVs), and rectum during computed tomography (CT) image-guided proton therapy for prostate cancer and studied the range variation in opposed lateral proton beams. We analyzed 375 sets of daily CT images acquired throughout the proton therapy treatment of ten patients. We analyzed daily movements of the prostate, SVs, and rectum by simulating three image-matching strategies: bone matching, prostate center (PC) matching, and prostate-rectum boundary (PRB) matching. In the PC matching, translational movements of the prostate center were corrected after bone matching. In the PRB matching, we performed PC matching and correction along the anterior-posterior direction to match the boundary between the prostate and the rectum's anterior region. In each strategy, we evaluated systematic errors (Σ) and random errors (σ) by measuring the daily movements of certain points on each anatomic structure. The average positional deviations in millimeter of each point were determined by the Van Herk formula of 2.5Σ+0.7σ. Using these positional deviations, we created planning target volumes of the prostate and SVs and analyzed the daily variation in the water equivalent length (WEL) from the skin surface to the targetalong the lateral beam directions using the density converted from the daily CT number. Based on this analysis, we designed prostate cancer treatment planning and evaluated the dose volume histograms (DVHs) for these strategies. The SVs' daily movements showed large variations over the superior-inferior direction, as did the rectum's anterior region. The average positional deviations of the prostate in the anterior, posterior, superior, inferior, and lateral sides (mm) in bone matching, PC matching, and PRB matching were (8.9, 9.8, 7.5, 3.6, 1.6), (5.6, 6.1, 3.5, 4.5, 1.9), and (8.6, 3.2, 3.5, 4.5, 1.9) (mm), respectively. Moreover, the ones of the SV tip were similarly (22.5, 15.5, 11.0, 7.6, 6.0), (11.8, 8.4, 7.8, 5.2, 6.3), and (9.9, 7.5, 7.8, 5.2, 6.3). PRB matching showed the smallest positional deviations at all portions except for the anterior portion of the prostate and was able to markedly reduce the positional deviations at the posterior portion. The averaged WEL variations at the distal and proximal sides of planning target volumes were estimated 7-9mm and 4-6mm, respectively, and showed the increasing of a few millimeters in PC and PRB matching compared to bone matching. In the treatment planning simulation, the DVH values of the rectum in PRB matching were reduced compared to those obtained with other matching strategies. The positional deviations for the prostate on the posterior side and the SVs were smaller by PRB matching than the other strategies and effectively reduced the rectal dose. 3D dose calculations indicate that PRB matching with CT image guidance may do a better job relative to other positioning methods to effectively reduce the rectal complications. The WEL variation was quite large, and the appropriate margin (approx. 10mm) must be adapted to the proton range in an initial planning to maintain the coverage of target volumes throughout entire treatment.