An empirical model to predict survival curve and relative biological effectiveness after helium and carbon ion irradiation based solely on the cell survival after photon irradiation

Abstract

Purpose:To develop an empirical model to predict radiosensitivity and relative biological effectiveness (RBE) after helium (He) and carbon (C) ion irradiation with or without DNA repair inhibitors. Methods: We characterized survival in eight human cancer cell lines exposed to 6 MV photons and to He-and C-ions with linear energy transfer (LET) values of 2.2-60.5 keV/µm to verify that the radiosensitivity parameters (D5%, D10%, D20%, D37%, D50% and SF2Gy) correlate linearly between photon and ion radiation with or without DNA-PKcs or ATR inhibitors. Then, we parameterized the LET response of the parameters governing these linear correlations up to LET values of 225 keV/μm using the data in the Particle Irradiation Data Ensemble (PIDE) v3.2 database, creating a model that predicts a cell's ion radiosensitivity, RBE and ion survival curve for a given LET on the basis of the cell's photon radiosensitivity. We then trained this model using the PIDE database as a training dataset, and validated it by predicting the radiosensitivity of the cell lines we exposed to He-and C-ions with LET ranging from 2.2-60.5 keV/μm. Results: Radiosensitivity to ions depended linearly with radiosensitivity of photons in the range of investigated LET values and the slopes and intercepts of these linear relationships within the PIDE database vary exponentially and linearly, respectively. Our model predicted ion radiosensitivity (e.g., D10%) within 5.1-21.3%, RBED10% within 5.0-17.1%, and ion mean inactivation dose within 6.7-25.1% for He-and C-ion LET ranging from 2.2-60.5 keV/µm. Conclusions: Radiosensitivity to He-and C-ions depend linearly with radiosensitivity to photons and can be used to predict ion radiosensitivity, RBE and cell survival curves for clinically relevant LET values from 2.2-60.5 keV/µm, with or without drug treatment. SUMMARYWe present a new empirical model capable of predicting clonogenic cell survival of cell lines exposed to helium and carbon ion beams. Our model is based on an observed linear correlation between radiosensitivity to ions and photons across a wide range of LET values. This linear correlation can be used to predict ion RBE, radiosensitivity, and the cell survival curve for a given LET all based on a cell's photon survival curve.

Cell lines of the same anatomic site and histologic type show large variability in intrinsic radiosensitivity and relative biological effectiveness to protons and carbon ions

“…Our finding that cells’ relative variability in intrinsic radiosensitivity between radiation qualities is the same, whether or not their DNA repair capacity has been modulated, might be understood in light of the observations by Suzuki et al (and by our group), that D 10% (D 50% ) for particles is linearly correlated with D 10% (D 50% ) for x‐rays 45,48 . This correlation implies that regardless of how these variations manifest themselves in a particular cell line's intrinsic photon radiosensitivity, they might simply map linearly to a cell's particle response.…”

“…26,27 Our finding that cells' relative variability in intrinsic radiosensitivity between radiation qualities is the same, whether or not their DNA repair capacity has been modulated, might be understood in light of the observations by Suzuki et al (and by our group), that D 10% (D 50% ) for particles is linearly correlated with D 10% (D 50% ) for x-rays. 45,48 This correlation implies that regardless of how these variations manifest themselves in a particular cell line's intrinsic photon radiosensitivity, they might simply map linearly to a cell's particle response. Accordingly, if a particular cell line is radiosensitive to photons (or made radiosensitive by DNA repair inhibition), this linear correlation would predict it to be relatively radiosensitive to particles, and by the same relative amount compared to the other cell lines in its group.…”

Robustness of carbon‐ion radiotherapy against DNA damage repair associated radiosensitivity variation based on a biophysical model

“…For the benchmarks against the carbon ion irradiation data by Furusawa et al 47 and Saager et al 48 the K iDSB and K cDSB were already obtained in previous publications. 15,16 In the case of the remaining predictions presented in this study, both the photon and carbon ion irradiation data were obtained from Suzuki et al, 49 Flint et al, 20 Cartwright et al, 50 and Lerch et al 51 Following the procedure detailed in a previous publication, 12 the K iDSB and K cDSB values are derived by fitting the model for sparsely ionizing radiation and Equation 1to the photon data using the curve_fit routine of the scipy library for Python. For cell lines that are reportedly subject to DDRi (e.g., DNA-PK deficient mutants), these two parameters are set to those derived for the wild type (e.g., DNA-PK proficient parent) of the cell line and Equation 2 is fitted to determine the RSF.…”

“…Figure 2a shows the predictions for survival data of M059K cells and their DNA-PK deficient mutant (M059J) after irradiation with photons and 12 C ions at 13.5, 27.9, and 60.5 keV/µm obtained by Flint et al 20 The cell line dependent parameters K iDSB and K cDSB as well as the RSF were derived from photon survival curves and are given in Table 1. The R 2 values (coefficient of 12 C ion beams at clinically relevant dose and linear energy transfer (LET) values from Furusawa et al, 47 with respective simulations by UNIVERSE and the modified Microdosimetric Kinetic Model (mMKM).…”

“…15 To our best knowledge, two other models have attempted to explicitly incorporate the effect of DDRi. However, one is purely empirical 20 and both are yet to be shown to provide accurate predictions in clinically relevant multienergy fields. 21 Here, after validating the applicability of the RSF for carbon ions in different (clinically relevant) scenarios, we utilize UNIVERSE predictions in a patient planning study.…”