Introducing Proton Track-End Objectives as a Tool to Mitigate the Elevated Relative Biological Effectiveness in Critical Structures

Abstract

The increased linear energy transfer (LET) near the end of the proton range often reach into normal tissues. This could potentially increase the risk of adverse effects due to the elevated relative biological effectiveness (RBE) with increased LET. By introducing proton track-end objectives in the optimization, the increased RBE in critical structures may be mitigated by lowering the LET. Intensity modulated proton therapy (IMPT) plans were generated for a head and heck (H&N) and an intracranial patient case. 2 Gy(RBE) per fraction (assuming the generic RBE of 1.1) were prescribed to the target using physical dose objectives. 35 fractions were planned for the H&N case and 30 for the intracranial case. Subsequently, both plans were re-optimized using proton track-end objectives in addition to the physical dose objectives. The track-end objectives penalized track-ends falling in the volume of choice and were used for the brainstem, chiasm, parotid glands and spinal cord. The resulting dose distributions were compared with the original plans using both RBE=1.1 and the variable RBE model by Wedenberg et al. (2013), where the RBE is a function of dose, dose-averaged LET (LETd) and α/β. An α/β of 10 Gy was assumed for the targets and 3 Gy for all normal tissues. Additionally, the distributions of proton track-ends and LETd were analyzed. The original plans satisfied the clinical goals for the target and the critical structures assuming RBE=1.1. The redistribution of the proton track-ends in the re-optimized plans allowed for LETd and RBE reductions in the critical structures, with only minimal degradation of the physical dose distribution. The median LETd in the H&N plan was reduced from 1.8 to 1.7 keV/μm for the brainstem, from 3.5 to 2.4 keV/μm for the parotid glands and from 2.2 to 2.1 keV/μm for the spinal cord when introducing the track-end objectives. The corresponding reduction for the intracranial plan was 4.7 to 4.1 keV/μm for the brainstem and 6.9 to 6.1 keV/μm for the chiasm. This allowed for a reduction from 29.3 to 27.6 Gy(RBE) in the mean dose to the parotid glands and 64.2 to 62.1 Gy(RBE) in D2% for the chiasm using the Wedenberg model. Although the track-end objectives allowed for this RBE-weighted dose reduction, it still predicted substantially higher values than the generic RBE of 1.1, which predicted a mean dose of 25.2 Gy(RBE) to the parotid glands and a D2% of 53.9 Gy(RBE) for the chiasm. Variable RBE models often predicts RBE values substantially higher than 1.1 in critical structures due to high LETd and low α/β values. By including proton track-end objectives in the optimization of IMPT plans, LETd reductions in critical structures may be accomplished without compromising the physical dose. This could mitigate potential adverse effects near the end of the proton range independently of tissue or patient-specific RBE.