Development and Verification of Multi-Ion Particle Treatments

Abstract

In hadron therapy, there are two widely used particles: protons (P) and carbon ions (CI), while helium ions (HI) are under consideration. Each ion species has unique benefits and tradeoffs; P undergo extensive scattering which widens the dose penumbra and can lead to increased lateral dose spreading, while HI and CI need sophisticated biophysical models to estimate the relative biological effectiveness (RBE) in tissue. Moreover, all heavy ions undergo fragmentation in tissue, producing a tail dose after the Bragg peak. It has been theorized that by combining two (or more) particle beams, novel treatments with biological and physical features unattainable using single ion modalities can be created. To that end, we developed and verified a multi-ion particle treatment modality, yielding constant RBE and potentially reduced dose to normal tissue. FRoG (Fast dose Recalculation on GPU), the GPU dose engine, has been coupled with an external optimization algorithm to establish the PRECISE (PaRticle thErapy using single and Combined Ion optimization StratEgies) treatment planning system. Besides SFUD and IMPT, PRECISE supports multi-ion optimizations, as well as the usage of different biophysical dose models such as the local effect model version 4 (LEM4) or modified microdosimetric kinetic model (mMKM). With PRECISE, combined ion-beam with constant RBE (CICR) plans using CI and P (CICRCI-P) as well as CI and HI (CICRCI-HI) have been optimized using LEM4. Extensive validation studies performed CICR plan optimizations with a 3 GyRBE target dose in both homogeneous and heterogeneous settings using a water and anthropomorphic head-phantom, respectively. The plans were irradiated at our facility and dosimetrically and biologically verified. Furthermore, in-silico patient treatment plans were optimized with P, HI, CI as well as CICRCI-P and CICRCI-HI. Dose-averaged linear energy transfer (LETd) and RBE with in the target were analyzed and for one case dose to normal tissue surrounding the tumor was investigated. Measured physical dose differences were ∼3 %, while RBE prediction was within 1 %. A biological robustness study for one patient with glioma yielded a similar biophysical stability to P, with CICRCI-P dose in tissue surrounding the target comparable to that of HI. Median LETd values in the targets of up to ∼30 keVμm-1 and ∼50 keVμm-1 were seen with CICRCI-P and CICRCI-HI patient plans. A smaller LETd, RBE, Dphys variability in the target was observed for all CICR plans, compared to SFUD CI treatments. Multi-field CICRCI-P plans achieved higher homogeneity in RBE, LETd and Dphys distributions. In this work, we showed that by combining ions in single and multiple fields, more biologically robust and more conformal treatment plans can be delivered. We also performed the first biological and dosimetric verification of multi-ion treatments in a homogeneous and heterogeneous setting.