Room: AAPM ePoster Library
Purpose: Despite actual clinical consensus of RBE constant and equal to 1.1 in proton radiotherapy, there is growing evidence of variable RBE within a patient geometry. To date, no contrasted study compares RBE predicted by LET-based models to other models based on microdosimetry, such as the Microdosimetric Kinetic Model (MKM). Here we assess whether the use of microdosimetry provides different results for RBE calculations in a prostate case.
Methods: We have developed an analytical algorithm to compute the spectral fluence of proton beams in a voxelized geometry and, indirectly, calculate dose and LETd. In parallel, models for the microdosimetric deposition of monoenergetic protons are obtained from simulations with Geant4-DNA, in order to obtain dose-mean lineal energy (yD) out of the spectral fluence. By means of an ESAPI script, we compute LETd and yD for prostate cases as phenomenological models for RBE use LETd while MKM employs yD as the indicator for quality beam. MKM also introduces a dependency on the ‘domain size’, characteristic of each tissue/cell line, which we have estimated as 960 nm for prostate tumor using data published by Butterworth et al, 2012. MKM, together with Carabe, Wedenberg and McNamara phenomenological models are implemented in our ESAPI script. Comparisons among them are performed for different prostate cases, considering an a/ß ratio of 1.877 Gy for tumor and 3 Gy for normal tissue.
Results: The three considered phenomenological models tend to overestimate RBE by 15% ± 10% with respect to MKM. However, this tendency tends to be reduced and eventually inverted as smaller domains are considered.
Conclusion: The mechanistic MKM exhibits differences with respect to phenomenological models in protontherapy, lowering the RBE suggested by those. However, as we deal with a first estimation of the domain size, further biological investigation to confirm these results is required before clinical implementation.
Funding Support, Disclosures, and Conflict of Interest: This project is supported by Varian Medical Systems, Palo Alto, California, USA.