Skip to main content

Radiation Response Models for Personalised Radiation Oncology

Final Report Summary - RADRESPRO (Radiation Response Models for Personalised Radiation Oncology)

Marie Curie International Outgoing Fellowship – RadRESPro


Radiotherapy is one of the key tools in the treatment of cancer, delivered to over 1.5 million EU citizens each year. The RadResPRO project aimed to develop a multi-scale model framework to bridge the gap between our current macroscopic radiotherapy planning and our current understanding of the physical and biological responses on the microscale. This framework will allow the characterisation of survival on a cell-by-cell level and incorporation of factors such as radiation quality, intercellular signalling, and individual cellular genetic characteristics.

To deliver on this aim, this project had several key objectives:
1. Build a computational framework to model responses to radiation on the single-cell level incorporating micro-environmental variables and intercellular communication;
2. Combine this framework with modern Monte Carlo radiation transport codes, to account for differences in response to different types of radiation;
3. Validate models developed in objectives 1 and 2 against in vitro experimental measurements of responses to radiation, including advanced modalities such as highly modulated X-rays, protons and ions;
4. Adapt the model to clinical radiotherapy delivery scenarios and generate predictions for Tumour Control Probability (TCP);
5. Expand the model to enable the use of genetic factors as parameters to characterise cells, to reduce the dependence of model fitting on empirical parameters.

Together, these objectives will lay the foundation for future personalised radiotherapy approaches. This work is being carried out by Dr Stephen McMahon, working between Massachusetts General Hospital, Boston, USA (MGH) and Queen’s University Belfast, Northern Ireland (QUB).

- Description of work
In the outgoing phase of this fellowship, Dr McMahon was based at Massachusetts General Hospital. There, he developed computational models of cellular responses to ionising radiation, which predict the yields of mutations, chromosome aberrations and cell survival for a range of radiation exposure conditions, and expanded them to incorporate the impact of different types of ionising radiation.
This work has focused on several key areas – firstly, the development of a single-cell model which incorporates descriptions of DNA repair, cell cycling and cell death following exposure to ionising radiation. Importantly, this model focuses on characterising fundamental cellular responses, and as a result can be applied in a predictive manner to new cells or exposure conditions without the usage of additional cell-line specific fitting parameters. This model has been extensively validated across several hundred different cell lines and experimental conditions, showing good predictive power based on cellular genetic and phenotypic characteristics.

Dr McMahon has also expanded this model to describe the impact of different types of radiation (X-rays, protons and carbon ions) by incorporating physical models of energy deposition on the subcellular scale. This approach has demonstrated the ability to characterise the increased biological effectiveness of more densely ionising radiation, again without the need for cell-line specific parameters, being tested against over 500 different published reports in the literature across a wide range of cell lines and exposure conditions.

Further work has incorporated a temporal component of radiation response into this model, allowing for modelling of fractionated or protracted radiation exposures. This has also been used to model the effects of intercellular signalling stresses on cell survival, as protracted low-dose stresses.

This modelling was also supported by experimental work, of which Dr McMahon was a part, investigating the relative biological effectiveness of a range of ions at the NASA Space Radiation Laboratory at Brookhaven National Laboratory, carried out in collaboration with other investigators at Massachusetts General Hospital.

Finally, work has been carried out to integrate these models into a clinical planning workflow, showing that, if individual patient tumour phenotypes can be determined, individualised radiosensitivity predictions can also be generated based on their clinical planning data.

- Main results
This project has developed a mechanistic radiation response model which has been demonstrated to have widely applicable predictive power. Importantly, this is achieved without any cell-specific fitting parameters, instead predicting sensitivity from a model of fundamental biological responses. After initially validating the model against a panel of experimental mechanistic observations, an extended study was carried out across 100 different published cell line radiation response curves. This comparison showed that using a simple classification (based on species of origin and genes associated with cell cycle and DNA repair defects), the model could accurately explain between 65 and 80% of the variation in sensitivity between different cell lines. This is a significant novel result and provides considerable support for the model’s predictive power and potential application as a predictive measure of radiation sensitivity.

Further work has demonstrated the model’s ability to predict the Relative Biological Effectiveness of different types of radiation. This has shown that incorporating information on the sub-cellular distribution of radiation within the cell enables prediction of the impact of different radiation qualities, being tested against over 600 experimental studies. The predictive power in this model is comparable to purely empirical fitting models but does not require cell-specific parameters, and extends to both protons and heavier ions.

- Potential socio-economic and wider society impacts
This model has demonstrated the potential of mechanistic models to predict individual radiosensitivity on a patient-by-patient basis, and describe differential effects due to different types of exposures. This approach has significant translational benefit, potentially providing a foundation with which to personalise radiotherapy, either by tailoring individual doses or identifying those patients most likely to benefit from ion-based therapies. Ongoing work seeks to clinically validate these approaches as a first step towards personalised therapy and improved clinical outcomes.