2275 - On Table Adaptation in MR Guided Prostate Cancer Treatments. Can RTTs Replace the Physician?

D. Epstein, V. Greenberg, Y. Birshtein, T. Shacham, Q. Tamimi, and D. Levin; Assuta Medical Centers, Tel Aviv, Israel

Purpose/Objective(s): Use of MR guided Radiotherapy (MRgRT) for low and intermediate risk prostate cancer (PC) is becoming more widespread. A major advantage of MRgRT is the ability to adapt the plan every treatment fraction based on the patient anatomy of the day. Currently, we require physician presence at the machine to contour the CTV and possibly the organs at risk. To streamline the workflow and minimize disruptions to physician time, this study aims to determine if RTTs can contour the CTV as accurately as physicians, alleviating the need to call the physician to the machine. Materials/

Methods: Seven RTTs and 5 physicians took part in this study. We evaluated 30 PC patients (150 treatment fractions) treated on the MRI-guided real-time on-table adaptive radiotherapy. For each fraction, RTTs contoured the CTV (CTV_RTT), and then, as per our standard of care, the physician on call contoured the daily CTV, blinded to the RTT contours. Treatment was delivered adapting the plan to the physician contoured CTV (CTV_Treat). Offline, a single radiation oncologist, experienced in MR prostate contouring, contoured the CTV for each treatment fraction, blinded to both CTV_RTT and CTV_Treat. This CTV was designated CTV_GT – the ground truth to which CTV_RTT and CTV_Treat were compared for volume and target coverage metrics. The accuracy of online CTV_RTT and CTV_Treat contouring was evaluated by calculating Dice similarity coefficient (DSC) for both CTVs compared to CTV_GT. For each treatment fraction we generated PTV_GT using a 3 mm uniform expansion of the CTV_GT. We then recalculated all 150 treatment plans optimized to 95% dose coverage to 95% of PTV_GT volume. We checked the 95% isodose coverage for physician and RTT PTVs according to these new plans.
Results: CTVs evaluation Using the Mann-Whitney Rank-Sum test we compared volume differences between both CTVs and CTV_GT. The results were not statistically significant (p>0.5) for either CTV. The DSC for both CTVs were in good agreement with CTV_GT: the average DSCs were 0.92±0.02 and 0.91±0.02 for CTV_Treat and CTV_RTT respectively. Target coverage evaluation The ground truth generated plan covered CTV_Treat and CTV_RTT to 95% of the prescribed dose in 96.7% and 98% cases respectively. For the PTV coverage by the 95% isodose the results showed that 46% of PTV_Treat contours and 23.3% of PTV_RTT contours met that criterion.
Conclusion: While CTV volumes and coverage were in good agreement between physicians, RTTs and ground truth, for PTV 95% coverage the results were markedly worse for both physician and RTT contours. This result probably indicates that the differences are in the margin expansion to PTV. This could be due to inter-physician and/or inter-RTT variability, which is likely to be more pronounced at the boundaries of the prostate, which are less clearly defined by a 0.35 T magnet. At this point our results indicate that we cannot forego physician presence at treatment for CTV contouring in PC patients treated adaptively. Further investigation is needed to overcome the poor PTV dose coverage.