Background and purpose: The purposes of this multicentric study are (a) the evaluation of four different commercially available treatment planning systems (TPSs) and (b) to verify whether the dosimetric results are comparable, also when considering the inter-observer variabilities and the different scanning protocols used. This work is to be considered a first step to test the value of multicentric studies based on dosimetric evaluation of the quality of the implants.
Patients and methods: Four different TPSs were used and the following tests were performed:Comparison of the parameters and mathematical algorithms used; comparison of the dose distributions generated by three different geometries of sources based on 32 dose-points on each source geometry. An octagonal geometric phantom was used to compare volume algorithms and dose-volume histogram (DVH) calculations (V150(Gy), V100(Gy), V50(Gy) and V25(Gy)). Comparison of the post-plan source distribution performed on a prostate-phantom implanted with (125)I seeds. A CT scan of the phantom was obtained at each participating center. Both the geometrical coordinates (with respect to the most caudal one), and the spread of the geometrical distribution, were calculated. The volumes included within different isodoses were also collected. Comparison of the post-plan source distribution performed on an actual patient. Post-plan V100% and D90(Gy) derived from seed distributions obtained by different operators were calculated, using the same target delineation.
Results: All the considered TPSs satisfied the AAPM dosimetric parameter recommendations. Point-dose examinations revealed differences smaller than 5%, except for one of the systems. Although the volume algorithm was not the same for all systems, no statistically significant difference was found in the volume measurements. The DVHs also presented differences smaller than 5%, except for one TPS. The distances between the seeds, based on the same CT images, showed a mean SD of 0.13 mm. The mean maximum difference of the position of each seed was 0.36 mm. The most significant errors were made in the cranio-caudal direction (mean maximal difference: 0.44 mm); here the size of the step between slices played an important role. The algorithm of source positioning of the different TPSs may also help explain this difference. The compiled DVHs showed differences smaller than 5%. Post-plans derived from different seed distributions showed a mild dependence upon operators. We obtained a mean value of 97.8 and 152.7 with a percentage of SD of 0.43 and 1.7, respectively, for V100% and D90(Gy).
Conclusions: Three-dimensional (3D) geometric reconstructions of seed distributions are slightly dependent upon the operators and the scanning protocols have little effect on the dosimetric evaluation. Some relevant discrepancies were found between one of the TPSs and the other three if few sources were used; increasing the number of seeds those differences became less pronounced. Multicentric studies on the quality of prostate implants based on post-implant dosimetry are feasible, provided an accurate step-wise evaluation of the procedure be performed.