Objectives: To study the impact of pelvic shape on prostate motion and the implications for radiotherapy target design and treatment outcomes.
Methods: A total of 3741 measurements (daily shift moves in the 3 principal directions) on 29 consecutive prostate cancer patients were reviewed. All patients had 3 prostatic fiducials placed and were tracked using kilovoltage on-board imaging. Pelvic shape was categorized into android (n = 21; 2580 measurements) and gynecoid (n = 8; 1161 measurements) (defined geometrically by postoutlet to preoutlet ratio). Multivariate analyses of means/standard deviations in each principal direction were performed using major demographic, disease, anatomic, and treatment factors as covariates. Toxicity rates were compared using Fisher exact test.
Results: On simple t test comparisons, no mean/standard deviation reached significance, although there was a nonsignificant (0.38 vs. 0.31 cm, P = 0.083) larger mean antero-posterior (AP) movement in the gynecoid group. On multivariate analyses, gynecoid shape (P = 0.032) significantly predicted for mean AP movement, and gynecoid shape (P = 0.045) significantly predicted for standard deviation of AP movement. Pelvic AP and RL dimensions also correlated with mean and standard movement along the respective axes. There were no differences between rates of acute (GI or GU) toxicity (P = 0.456) between the android and gynecoid groups.
Conclusions: Treatment strategies that do not employ daily motion tracking may require wider planning target volume margins in gynecoid patients. Tracking the prostate daily, as done in our case using fiducials/on-board imaging, can counter differences in pelvic shape to produce similar treatment outcomes.