Room: Karl Dean Ballroom A2
Purpose: The relative biological effectiveness (RBE) of protons increases with linear energy transfer (LET). In current practice, however, treatment planning for intensity-modulated proton therapy (IMPT) is based on a constant RBE of 1.1. IMPT optimization based on RBE-weighted dose rather than physical dose has been suggested but is not applied due to uncertainties in RBE. We present a robust optimization approach to incorporate uncertainty in RBE into RBE-based IMPT planning.
Methods: We consider a standard RBE model in which RBE increases linearly with LET: RBE=(câ‚?+câ‚‚LET). The model contains the constant RBE model corresponding to current practice (câ‚?=1.1,câ‚‚=0) as well as LET-dependent variable RBE models (câ‚?<1.1,câ‚‚>0). Uncertainty in the parameters câ‚? and câ‚‚ is incorporated into treatment planning via a stochastic programming approach, i.e. we evaluate the objective function for RBE-weighted dose for a range of realistic parameter values (câ‚?,câ‚‚) and optimize the average objective value with respect to the pencil beam intensities. The method was evaluated for intracranial lesions (atypical meningiomas, ependymomas, chordomas).
Results: IMPT planning based on a constant RBE of 1.1 may lead to LET hotspots (possibly corresponding to elevated RBE-weighted doses) in critical structures within or near the target volume. Contrary, IMPT planning based on a variable RBE model may lead to reduced physical doses in parts of the target volume, based on the assumption that RBE is higher than 1.1 in regions of high LET (possibly corresponding to underdosage of the target if RBE is overestimated). Accounting for uncertainty in RBE via robust optimization can avoid the disadvantages of both methods and generate IMPT plans that are acceptable under a wide range of RBE model parameters.
Conclusion: Robust optimization is a suitable method to account for RBE uncertainty, which may make RBE-based IMPT planning practical. The method is associated with virtually no increase in computation time.