A concept to personalize radiation oncology: Predicting cell-specific survival prior to treatment


A concept to personalize radiation oncology: Predicting cell-specific survival prior to treatment

Oesten, H.; von Neubeck, C.; Löck, S.; Enghardt, W.; Krause, M.; Mcmahon, S.; Grassberger, C.; Paganetti, H.; Lühr, A.

Material and Methods
Based on a previously developed mechanistic radiation response model of DNA repair and cell survival (CS) prediction for normal tissue cells, we simulated measured radiobiological parameters (α and β) of 19 in vitro cancer cell lines (skin, lung, brain). The radiation model incorporated four cell-specific parameters: number of chromosomes, p53 mutation status, cell-cycle distribution and the effective genome size (GS). Only the first three input parameters were experimentally available; the latter was obtained by minimizing the difference between the simulated and measured α and β values. A parametrization of the GS as a function of the cells’ chromosome number and nucleus volume was proposed. The use of these input parameters was validated by comparing the simulated outcome of time-dependent γH2AX data over 24h with independent experimental datasets.
Results
Overall good agreement between simulated and measured in vitro cancer CS curves was achieved (Fig. 1). The measured β values increased quadratically with the obtained GS (R2=0.81) irrespective of other cell-specific parameters (Fig. 2b). The measured α values increased linearly with GS manifesting different slopes distinguishable into the cells’ p53 mutation status (Fig. 2a). Measured α and β values were predictable based on GS with a one-sigma uncertainty: σ=0.04Gy-1 for α and σ=0.01Gy-2 for β. The GS correlated (R2=0.70) with the number of chromosomes for all but four cell lines. The detailed cell-specific cell cycle distribution had a negligible impact on α and β. Measured time-dependent γH2AX data were consistent with the repair kinetics simulations (R2=0.95).
Conclusion
A mechanistic model for radiation response of normal human cells was successfully modified to simulate measured in vitro CS of 19 cancer cell lines. Independent of cancer entity, the radiobiological value β was predictable only with known GS while the prediction of α additionally required at least knowledge of the p53 mutation status. An observed correlation of GS with the number of chromosomes and nucleus size, both clinically accessible from a biopsy prior to treatment, may facilitate individualized radiotherapy based on cell-specific survival prediction.

Permalink: https://www.hzdr.de/publications/Publ-27968
Publ.-Id: 27968