Purpose: To elucidate the role of LET in the appearance of abnormalities in MRI after undergoing proton radiotherapy. Results helped feed a normal tissue complication probability (NTCP) explicitly including LET as a predictor variable.
Methods: We calculated dose-averaged LET (LETd) in 26 meningioma patients that developed abnormalities in MRI studies following proton radiotherapy treatments. An analytical method (MicroCalc) based on the computation of the spectral fluence of primary protons, secondary protons and alpha particles was employed. MRI were registered to planning CT and LETd values across voxels belonging abnormalities were scored. We divided dose (D) and LETd ranges in bins and a priori probability of abnormality P(D,LETd) for each bin was obtained as the fraction of voxels belonging abnormalities and belonging normal areas. This probability is modeled as a function of dose and the linear-quadratic (LQ) parameters α(LETd)=α_0+α_1 LETd and β=β_0, which, in turn, depend on LETd in this model. Data from patients were used to estimate these dependencies under the model. Finally, NTCP was modeled assuming independence between voxels.
Results: For (D, LETd) pair-values with a priori probability higher than 50% of abnormality, LETd was found to increase roughly linearly as dose decreased, being above 8 keV/μm at doses below 1 Gy and about 3 keV/μm at a dose of 60 Gy. This points towards an enhancement of low dose effectiveness because of high LETd. LQ parameters were estimated as α_0= 0.011 ± 0.02 Gy-1, α_1= 0.033 ± 0.002 Gy-1 μm keV-1 and β_0= 0.0015 ± 0.0004 Gy-2.
Conclusion: We illustrated a practical method to infer the potential role that LET may play in the induction of normal tissue complications and on derived parameter estimates of an LET-dependent NTCP model. However, a larger patient cohort including patients without abnormalities are required to further validate these results.
Funding Support, Disclosures, and Conflict of Interest: A. Bertolet, R. Abolfath and A. Carabe-Fernandez were supported by Varian Medical Systems for this work