Retrospective analysis of prostate cancer patients with implanted gold markers using off-line and adaptive therapy protocols

Purpose: To determine the efficacy of applying adaptive and off-line setup correction models to bony anatomy and gold fiducial markers implanted in the prostate, relative to daily alignment to skin tattoos and daily on-line corrections of the implanted gold markers. Methods and Materials: Ten prostate cancer patients with implanted gold fiducial markers were treated using a daily on-line setup correction protocol. The patients' positions were aligned to skin tattoos and two orthogonal diagnostic digital radiographs were obtained before treatment each day. These radiographs were compared with digitally reconstructed radiographs to obtain the translational setup errors of the bony anatomy and gold markers. The adaptive, no-action-level and shrinking-action-level off-line protocols were retrospectively applied to the bony anatomy to determine the change in the setup errors of the gold markers. The protocols were also applied to the gold markers directly to determine the residual setup errors. Results: The percentage of remaining fractions that the gold markers fell within the adaptive margins constructed with 1.5 sigma' (estimated random variation) after 5, 10, and 15 measurement fractions was 74%, 88%, and 93% for the prone patients and 55%, 77%, and 93% for the supine patients, respectively. Using 2 sigma', the percentage after 5, 10, and 15 measurements was 85%, 95%, and 97% for the prone patients and 68%, 87%, and 99% for the supine patients, respectively. The average initial three-dimensional (3D) setup error of the gold markers was 0.92 cm for the prone patients and 0.70 cm for the supine patients. Application of the no-action-level protocol to bony anatomy with N-m = 3 days resulted in significant benefit to 4 of 10 patients, but 3 were significantly worse. The residual average 3D setup error of the gold markers was 1.14 cm and 0.51 cm for the prone and supine patients, respectively. When applied directly to the gold markers with N-max = 3 days, 5 patients benefited and 3 were significantly worse. The residual 3D error of the gold markers was 1.14 cm and 0.76 cm for the prone and supine patients, respectively. Application of the shrinking-action-level protocol to bony anatomy with an initial action level of 1.0 cm and N-max = 5 days decreased the residual systematic offset of the gold markers in 2 of 10 patients. The residual average 3D setup error of the gold markers was 1.2 cm and 1.0 cm for the prone and supine patients, respectively. When applied directly to the gold markers with N-max = 5 days, the residual systematic offset of the gold markers decreased in 6 of 10 patients (0.84 cm and 0.67 cm for the prone and supine patients, respectively). In general, between 3 and 5 of the 10 patients showed significant decreases in setup errors with the application of these off-line protocols, and the remaining patients showed no significant improvement or showed significantly larger setup errors, as determined by the residual error of the gold markers. Conclusion: Changes in a prostate cancer patient's systematic and random setup characteristics during the course of therapy often violate the gaussian assumptions of adaptive and off-line correction models. Thus, off-line setup correction procedures, especially those directed at prostate localization using markers, will result in limited benefit to a minority of patients. The relative benefit of on-line localization is still potentially significant if the intrafraction motion is relatively small. (c) 2005 Elsevier Inc.