Room: Karl Dean Ballroom A1
Purpose: To evaluate the radiobiological advantage of tumor tracking in a MR-linac for three treatment sites: liver, pancreas and kidney.
Methods: A total of nine SBRT patients (three patients from each treatment site) were selected. Patient 4DCT data sets were used, representing 4D pseudo-CTs derived from 4DMRI. We applied two treatment planning approaches using the Monaco treatment planning system (Elekta research version 5.19.03) : 1) the conventional ITV method which uses a 6MV Elekta Agility beam and 2) a simulated tracking method which utilizes MLC tracking of the GTV (virtual couch shift) with a 7MV Elekta MR-linac beam model and a 1.5 T transverse magnetic field (4DMRL method). A 5mm isotropic PTV margin was added either to the ITV or the GTV with the criteria that 95% of the PTV volume received 100% of the prescription dose. To evaluate the potential radiobiological advantages of tumor tracking, the NTCP values (Normal Tissue Complication Probabilities) were calculated for each OAR (organ at risk) using the LKB (Layman Kutcher Burman) model.
Results: For each OAR, the differences in NTCP between the two methods (NTCPITV â€“ NTCP4DMRL) were calculated. The differences were null for 74% of the cases and positive for 24% of the cases. For all three disease sites, the largest NTCP improvement was shown for the kidney, the duodenum and the bowels (maximum improvement of 0.79, 0.35 and 0.69 respectively). OARs, such as the liver, showed little to no reduction in NTCP, in part due to the lack of seriality.
Conclusion: This study demonstrates the possible benefit of using a MRI-linac tracking system to reduce NTCPs. OARs with the most improvement in NTCP were the kidney, the bowels and the duodenum. This, in part, is due to the rapid changes in NTCP for small changes in dose for these OARs.
Funding Support, Disclosures, and Conflict of Interest: The senior author has received funding support from Elekta AB, Stockholm, Sweden
TH- Radiobiology(RBio)/Biology(Bio): RBio- LQ/TCP/NTCP/outcome modeling