ESTRO 37 Abstract book

ESTRO 37

S531

Material and Methods All patients treated for prostate cancer with elective lymph node irradiation between August 2016 and September 2017 at our institution were selected for the present study (n=58). Patients were treated on Elekta linacs using 6MV VMAT. Dose plans were made in RayStation. Treatment was prescribed as 39 fractions totaling a dose of 78 Gy to the prostate, 56 Gy to elective lymph nodes, and 56 or 78 Gy to the seminal vesicles. Patients had daily pre-treatment CBCT imaging, using the Elekta XVI R4.5 system, with dual registration on bony anatomy and implanted seeds in the prostate. If deviations between the two registrations in x, y, or z, were larger than our CTV to PTV margin (5 mm LR, 8 mm CC, 7 mm AP), a flag was raised. Three flags triggered an investigation by a physicist to determine if ART was required. The number of treatment adaptations prescribed as a result of differential motion of the prostate and elective lymph node targets was determined by chart examination. From the log of the daily CBCT match, the lateral, longitudinal and vertical differences between the match on bones and prostate seeds were retrieved and analyzed. Manual evaluation of the displacement over the entire treatment course for each patient was used to determine which patients should have received ART to ensure proper target coverage. Based on the extracted data, we propose a model to predict the necessity of ART if the mean +/- one standard deviation between bone and prostate seed match is larger than our CTV to PTV margin minus 2 mm, calculated after 8 treated fractions. Results Of the 58 patients, 18 should have received ART due to either systematic or large random shifts between the prostate and bony anatomy (example provided in Figure 1). Our current practice of manually flagging deviations between bone and seed match resulted in 14/18 patients receiving ART, prescribed at a median time of 14 fractions (full range: 5-30). The proposed model predicted the need for ART in 17/18 patients after only 8 fractions. True and false positive and negative rates of the manual and predicted screening for ART are reported in Table 1.

Conclusion By calculating the mean difference between CBCT match on bone and implanted seeds on fraction 8, we were able to predict the need for ART on 17/18 patients early in the treatment course. In contrast, our current manual protocols for ART only flagged 14/18 patients, and did so almost a week later in median. The proposed model is easy to use, and suggests ART earlier in the treatment course than our current clinical practice. We are currently collecting data to validate the model in an independent patient cohort.

Poster: Physics track: CT Imaging for treatment preparation