Strategies to reduce the systematic error due to tumor and rectum motion in radiotherapy of prostate cancer

Background and purpose: The goal of this work is to develop and evaluate strategies to reduce the uncertainty in the prostate position and rectum shape that arises in the preparation stage of the radiation treatment of prostate cancer. Patients and methods: Nineteen prostate cancer patients, who were treated with 3-dimensional conformal radiotherapy, received each a planning CT scan and 8-13 repeat CT scans during the treatment period. We quantified prostate motion relative to the pelvic bone by first matching the repeat CT scans on the planning CT scan using the bony anatomy. Subsequently, each contoured prostate, including seminal vesicles, was matched on the prostate in the planning CT scan to obtain the translations and rotations. The variation in prostate position was determined in terms of the systematic, random and group mean error. We tested the performance of two correction strategies to reduce the systematic error due to prostate motion. The first strategy, the pre-treatment strategy, used only the initial rectum volume in the planning CT scan to adjust the angle of the prostate with respect to the left-right (LR) axis and the shape and position of the rectum. The second strategy, the adaptive strategy, used the data of repeat CT scans to improve the estimate of the prostate position and rectum shape during the treatment. Results: The largest component of prostate motion was a rotation around the LR axis. The systematic error (1 SD) was 5.1 degrees and the random error was 3.6 degrees (1 SD). The average LR-axis rotation between the planning and the repeat CT scans correlated significantly with the rectum volume in the planning CT scan (r = 0.86, P < 0.0001). Correction of the rotational position on the basis of the planning rectum volume alone reduced the systematic error by 28%. A correction, based on the data of the planning CT scan and 4 repeat CT scans reduced the systematic error over the complete treatment period by a factor of 2. When the correction was carried out later in the treatment (based on the data of more scans) the overall reduction was less. For the rectum, the first strategy performed best at the upper anterior side, where a reduction of the anterior-posterior displacement of 30% could be achieved. The systematic error could be reduced by 43% for the whole rectum by using the data of 4 repeat CT scans and the planning CT scan. Conclusions: Both the pre-treatment as well as the adaptive correction strategy reduced the systematic error in the prostate position and rectum position and shape. A smaller systematic error makes it possible to safely reduce the margin around the clinical tumor volume, so that normal tissues can be spared or the prescription dose can be escalated. (c) 2005 Elsevier Ireland Ltd. All rights reserved.