Prediction of normal tissue complication probability for rat spinal cord tolerance following ion irradiations

Objective. Currently, treatment planning in cancer hadrontherapy relies on dose-volume criteria and physical quantities constraints. However, incorporating biologically related models of tumor control probability and of normal tissue complication probability (NTCP) would help further minimizing adverse tissue reactions, and would allow achieving a more patient-specific strategy. The aim of this work was therefore the development of a mechanistic approach to predict NTCP for late tissue reactions following ion irradiation. Approach. A dataset on the tolerance of the rat spinal cord was considered, providing NTCP (for paresis of at least grade II) experimental data following irradiation by photons, protons, helium and carbon ions, under different fractionation schemes. The photon data were fit by a mechanistic NTCP model with four parameters, called Critical Element Model; this allowed fixing the two parameters that only depend on the tissue features. Afterwards, the two parameters depending on radiation quality were predicted by applying the BIophysical ANalysis of Cell death and chromosome Aberrations biophysical model, for each ion type and dose-averaged linear energy transfer value. Main results. The predicted NTCP curves for ion irradiation were tested against the ion experimental data, by Chi-Square and p-value calculations. The model passed a significance test at 1% for all the datasets, and 5% for 13 out of 16 datasets, thus showing a good predictive power. The Relative biological effectiveness (RBE) was also calculated and compared with the data for the endpoint of NTCP equal to 50 % , and a considerable discrepancy with the commonly calculated RBE for cell survival was shown. Significance. This study highlights the importance of considering the endpoint of interest when computing the RBE, through the application of a NTCP model, and it represents a first step towards the development of an approach to improve treatment plan optimization in therapy. To this aim, the approach needs to be extended to other endpoints and to be applied to patients’ data.