Minimum Bladder Reference Contour To Guide Patient Setup For Prostate Radiotherapy

Abstract

To evaluate if a model—based approach for bladder matching at the time of treatment reduces imaging and treatment time for prostate SBRT patients treated with a full bladder protocol. Prospectively collected data from prostate SBRT patients consecutively treated between 9/2019 and 1/2020 was used to select patients who experienced at least one treatment delay due to perceived underfilling of their bladder on CBCT. At our institution, a full bladder protocol (1 cup of water 45 min prior) checked by CBCT is employed for all treatments. When the pre-treatment bladder volume is smaller than the simulated volume, patients are asked to either wait longer and/or drink more water. A minimum bladder reference contour [min_bladder] was calculated for each patient by deforming the bladder contour at simulation [sim_bladder] with decreasing margins (1, 1.5, 2, 2.5, 3, 3.5, 4.5, and 5 cm) and identifying the smallest contour that met the bladder planning constraints (Maximum < 105%, V20Gy < 50% and V 36Gy < 10%). To evaluate the accuracy and efficacy of a model-based approach where min_bladder was used as the threshold for treatment initiation, the bladder on initial rejected CBCT was contoured, dose calculated and compared to min_bladder dose. Furthermore, treatment setup time reductions were estimated retrospectively as if the model had been used in the clinic. 64 patients were included in this study, of which 23 (36%) had delays due to perceived bladder underfilling (35/115 fractions). The average ± standard deviation sim_bladder and min_bladder size were 321 ± 117 and 134 ± 53 cm3, respectively. Comparing the min_bladder to the actual rejected setup bladder volume [rej_bladder] on CBCT predicted that the bladder would have been large enough for 22 fractions (63%) while for 13 fractions (37%) additional filling would have been necessary. When reviewing the doses to the rej_bladder, we found a model accuracy of 82%. For 3 fractions, the model predicted the volume was too small, but the rej_bladder met the dose constraints. For one patient, the rej_bladder shape did not conform well to the model and even though the rej_bladder was larger than the min_bladder, it did not meet the bladder constraints. The average total treatment times as recorded in Aria for fractions without delays versus those with delays was 28.1 ± 18 vs 49.6 ± 15 min (p<0.0001) (Student's t-test). Average estimated time that could be saved using the minimum bladder reference was 25± 11 min per fraction and number of CBCTs reduced from 2 to 1 per fraction. We found 36% of our prostate SBRT patients were delayed at least once during their treatment due to underfilling of their bladder. We demonstrate that employing a patient-specific model-based approach using a minimum bladder reference created at the time of planning could prevent treatment delays in the majority of cases and reduce the number of CBCTs used to setup patients for prostate SBRT.