Characterizing interfraction variations and their dosimetric effects in prostate cancer radiotherapy

To quantitatively characterize the interfraction variations and their dosimetric effects in radiotherapy for prostate cancer.

A total of 486 daily computed tomography (CT) sets acquired for 20 prostate cancer patients treated with daily CT-guided repositioning using a linear accelerator and CT-on-rail combination were analyzed. The prostate, rectum, and bladder, delineated on each daily CT data set, were compared with those from the planning CT scan. Several quantities, including Dice's coefficient and the maximal overlapping rate, were used to characterize the interfraction variations. The delivered dose was reconstructed by applying the original plan to the daily CT scan with consideration of proper repositioning.

The mean prostate Dice's coefficient and maximal overlapping rate after bony registration was 69.7%±13.8% (standard deviation) and 85.6%±7.8% (standard deviation), respectively. The daily delivered dose distributions were generally inferior to the planned dose distribution for target coverage and/or normal structure sparing. For example, for approximately 5% of the treatment fractions, the prostate volume receiving 100% of the prescription dose decreased dramatically (15-20%) compared with its planned value. The magnitudes of the interfraction variations and their dosimetric effects indicated that, statistically, current standard repositioning using prostate alignment might be adequate for two-thirds of the fractions, but for the rest of the fractions, better on-line correction strategies, such as on-line replanning, are needed.

Different adaptive correction schemes for prostatic interfraction changes can be used according to the anatomic changes, as quantified by the organ displacement and deformation parameters. On-line replanning is needed for approximately one-third of the treatment fractions.