For tumor therapy with light ions and for experimental aspects in particle radiobiology the relative biological effectiveness (RBE) is an important quantity to describe the increased effectiveness of particle radiation. By establishing and analysing a database of ion and photon cell survival data, some remarkable properties of RBE-related quantities were observed. The database consists of 855 in vitro cell survival experiments after ion and photon irradiation. The experiments comprise curves obtained in different labs, using different ion species, different irradiation modalities, the whole range of accessible energies and linear energy transfers (LETs) and various cell types. Each survival curve has been parameterized using the linear-quadratic (LQ) model. The photon parameters, α and β, appear to be slightly anti-correlated, which might point toward an underlying biological mechanism. The RBE values derived from the survival curves support the known dependence of RBE on LET, on particle species and dose. A positive correlation of RBE with the ratio α/β of the photon LQ parameters is found at low doses, which unexpectedly changes to a negative correlation at high doses. Furthermore, we investigated the course of the β coefficient of the LQ model with increasing LET, finding typically a slight initial increase and a final falloff to zero. The observed fluctuations in RBE values of comparable experiments resemble overall RBE uncertainties, which is of relevance for treatment planning. The database can also be used for extensive testing of RBE models. We thus compare simulations with the local effect model to achieve this goal.

To present details of the recent version of the 'Local Effect Model' (LEM), that has been developed and implemented in treatment planning for the ion beam therapy pilot project performed at GSI Helmholtzzentrum für Schwerionenforschung in Darmstadt, Germany. The new version of the model is based on a detailed consideration of the spatial distribution of the initial damages, i.e., double-strand breaks (DSB). This spatial distribution of DSB is obtained from the radial dose profile of the ion track using Monte Carlo methods. These distributions are then analyzed with regard to the proximity of DSB. This version of the model also facilitates the calculation of full dose response curves up to arbitrary high doses, thus allowing to thoroughly check the approximations previously used to estimate the quadratic term (β-term) for the linear-quadratic description of dose response curves. The accuracy of the model predictions is demonstrated by good agreement of the relative biological effectiveness (RBE) as a function of the linear energy transfer (LET) with experimental data obtained for V79 cells after carbon irradiation. The β-values predicted by the full simulation tend to be larger as compared to the approximation in the intermediate LET range. The new version of the model allows a more mechanistic description of the biological effects of ion radiation. The full simulation is a prerequisite for tests of the validity of the approach at high doses, which are of particular interest for application in hypofractionation studies.

Solid tumours often present regions with severe oxygen deprivation (hypoxia), which are resistant to both chemotherapy and radiotherapy. Increased radiosensitivity as a function of the oxygen concentration is well described for X-rays. It has also been demonstrated that radioresistance in anoxia is reduced using high-LET radiation rather than conventional X-rays. However, the dependence of the oxygen enhancement ratio (OER) on radiation quality in the regions of intermediate oxygen concentrations, those normally found in tumours, had never been measured and biophysical models were based on extrapolations. Here we present a complete survival dataset of mammalian cells exposed to different ions in oxygen concentration ranging from normoxia (21%) to anoxia (0%). The data were used to generate a model of the dependence of the OER on oxygen concentration and particle energy. The model was implemented in the ion beam treatment planning system to prescribe uniform cell killing across volumes with heterogeneous radiosensitivity. The adaptive treatment plans have been validated in two different accelerator facilities, using a biological phantom where cells can be irradiated simultaneously at three different oxygen concentrations. We thus realized a hypoxia-adapted treatment plan, which will be used for painting by voxel of hypoxic tumours visualized by functional imaging.

Biological optimization for treatment planning in carbon ion therapy is currently based on the first version of the local effect model (LEM I). Further developments implemented in the latest version (LEM IV) allowed to predict more accurately the Relative Biological Effectiveness (RBE) in-vitro. The main goal of this study is to compare the LEM IV against LEM I under treatment-like conditions for idealized target geometries. Therefore, physical dose distributions resulting from the biological optimization with LEM I were used to recalculate the RBE-weighted dose distribution based on LEM IV. Input parameters representing the clinical endpoints late toxicity in the central nervous system and the tumor control for chordoma were chosen to investigate the impact of changes on the predicted isoeffective dose levels. The recalculated RBE-weighted dose distributions show an increase within the target region, and the mean RBE-weighted dose values are dependent on the geometry and decrease with increasing target dimension. The differences between predictions of LEM IV and LEM I are less than 10% for typical tumor volumes treated in the pilot project at GSI. Median RBE-weighted doses predicted by LEM IV in the target region are consistent with clinically observed dose-response behavior as demonstrated by comparison to the 5-year local control curve for skull base chordoma.

Background and Significance: The application of heavy ion beams in cancer therapy must account for the increasing relative biological effectiveness (RBE) with increasing penetration depth when determining dose prescriptions and organ at risk (OAR) constraints in treatment planning. Because RBE depends in a complex manner on factors such as the ion type, energy, cell and tissue radiosensitivity, physical dose, biological endpoint, and position within and outside treatment fields, biophysical models reflecting these dependencies are required for the personalization and optimization of treatment plans.Aim: To review and compare three mechanism‐inspired models which predict the complexities of particle RBE for various ion types, energies, linear energy transfer (LET) values and tissue radiation sensitivities.Methods: The review of models and mechanisms focuses on the Local Effect Model (LEM), the Microdosimetric‐Kinetic (MK) model, and the Repair‐Misrepair‐Fixation (RMF) model in combination with the Monte Carlo Damage Simulation (MCDS). These models relate the induction of potentially lethal double strand breaks (DSBs) to the subsequent interactions and biological processing of DSB into more lethal forms of damage. A key element to explain the increased biological effectiveness of high LET ions compared to MV x rays is the characterization of the number and local complexity (clustering) of the initial DSB produced within a cell. For high LET ions, the spatial density of DSB induction along an ion's trajectory is much greater than along the path of a low LET electron, such as the secondary electrons produced by the megavoltage (MV) x rays used in conventional radiation therapy. The main aspects of the three models are introduced and the conceptual similarities and differences are critiqued and highlighted. Model predictions are compared in terms of the RBE for DSB induction and for reproductive cell survival.Results and Conclusions: Comparisons of the RBE for DSB induction and for cell survival are presented for proton (¹H), helium (⁴He), and carbon (¹²C) ions for the therapeutically most relevant range of ion beam energies. The reviewed models embody mechanisms of action acting over the spatial scales underlying the biological processing of potentially lethal DSB into more lethal forms of damage. Differences among the number and types of input parameters, relevant biological targets, and the computational approaches among the LEM, MK and RMF models are summarized and critiqued. Potential experiments to test some of the seemingly contradictory aspects of the models are discussed.