Subgroup Contributed Talks

Renee Brady-Nicholls

Department of Integrated Mathematical Oncology, H. Lee Moffitt Cancer Center Tampa, USA

"Forecasting Individual Responses to Intermittent Androgen Deprivation Therapy Using PSA Dynamics"

Prostate cancer (PCa) is the second most common cancer in men and the fifth leading cause of death worldwide. Intermittent androgen deprivation therapy (IADT), whereby treatment is cycled on and off, is an attractive therapy option that can reduce cumulative dose and limit toxicities, when compared to continuous therapy. We simulate prostate-specific antigen (PSA) dynamics, with enrichment of PCa stem-like cell (PCaSC) during treatment as a plausible mechanism of resistance evolution. Simulated PCaSC proliferation patterns correlate with longitudinal serum PSA measurements in 70 biochemically recurrent PCa patients. Learning dynamics from each treatment cycle in a leave-one-out study, model simulations predict patient-specific evolution of resistance with an overall accuracy of 89% (sensitivity=73%, specificity=91%). Previous studies have shown a benefit of concurrent therapies with ADT in both low- and high-volume metastatic hormone-sensitive PCa. Model simulations based on response dynamics from the first IADT cycle identify patients who may benefit from concurrent docetaxel, demonstrating the feasibility and potential value of adaptive clinical trials guided by patient-specific mathematical models of intratumoral evolutionary dynamics.

Balázs G. Madas

Centre for Energy Research, Hungary

"Low dose hyper-radiosensitivity as a consequence of bystander signalling aiming to reduce the mutation load of the cell population"

Radiation therapy is frequently applied in cancer treatment. The fraction of surviving cells is a crucial quantity predicting both tumour control and normal tissue toxicity. The gold standard in vitro measurement for cell survival is the colony forming assay (CFA), which is based on the ability of a single cell to grow into a colony. The relationship between surviving fraction (SF) and absorbed dose (D) can be described by a linear-quadratic function: SF=exp(-aD-bD^2), where a and b depend on radiation type other cell properties. However, many cell lines show hyper-radiosensitivity (HRS) resulting in a local minimum in the surviving curve below 1 Gy. It is an interesting phenomenon and it may also have clinical implications related to the potential benefits of hyperfractionation. It is generally thought that HRS is a mechanism that protects against carcinogenesis following low dose ionizing radiation. Recently, it has been shown that minimizing the mutation load at the cell population level may result in local minimum in survival-dose relationships if both spontaneous and radiation induced mutations are considered. The model proposed, however, does not account for how cells receive the information required to reduce the mutation load. The aim of the present study was to test whether bystander signalling through the medium provides enough information for the cells to reduce the mutation load of the population with their individual survival or death decision. For this purpose, a mathematical model of bystander signalling was elaborated. Cells are located randomly in a disc shaped dish. Extracellular signals are produced by cells with mutagenic damages. The signal production is proportional to the number of mutagenic damages, while the signal strength decreases by the distance from the signal producing cell. A cell goes into apoptosis if the difference between the number of its mutagenic damages and the number of spontanenous mutations per cell division is high compared to the signal strength at its location. The threshold signal strength for this decision equals to the mean signal strength if all cells have one mutagenic damage. The number of mutagenic damages follows Poisson distribution, and its mean is proportional to absorbed dose. Simulations were performed to reproduce the deviation from the linear quadratic survival curve in case of 45 experimental datasets. During this fitting process, the spontaneous mutation rate and the mutagenic damage induction rate were varied. The model describes most experimental data very well, while the spontanenous mutation rate is close to the reasonable number of 1 mutation per cell division, and the mutagenic damage induction rate is in the order of magnitude of 1 per Gy. These results confirm that low dose hyper-radiosensitivity can be a consequence of bystander signalling aiming to reduce the mutation load of the cell population.

Guillermo Lorenzo

The University of Texas at Austin, USA

"Image-based mechanistic modeling of prostate cancer for personalized forecasting of tumor growth"

The current clinical management of prostate cancer (PCa) enables its detection at early organ-confined stages by combining regular screening and patient classification in risk groups. Although these newly diagnosed tumors do not usually pose a threat to the patient, most PCa cases are prescribed a radical treatment immediately after diagnosis (e.g., surgery or radiotherapy). However, the limited individualization of the clinical management beyond risk-group definition has led to significant overtreatment and undertreatment rates, which might adversely impact the patients' lives and life expectancy, respectively. Thus, PCa is a paradigmatic disease in which an individualized predictive technology could make a crucial difference in clinical practice, thereby separating less aggressive tumors that could be safely monitored from lethal tumors that require immediate treatment. To address this critical need, we propose to use routine clinical and imaging data to construct and parametrize personalized mathematical models of PCa growth including the key mechanisms involved in this pathology. We can then run computer simulations with these models to forecast the growth of a patient's tumor, which may assist physicians in clinical-decision making. Our models have been able to qualitatively reproduce tumor growth over the local anatomy of a patient's prostate extracted from imaging data. We have also been able to capture the dynamics of the Prostate Specific Antigen (PSA), which is a ubiquitous biomarker in PCa clinical management. We have also studied the inhibitive effect of growth-induced mechanical stress on PCa and how the compression exerted by concomitant benign prostatic hyperplasia dramatically impedes tumor growth. Our ongoing efforts aim at leveraging longitudinal clinically-available quantitative magnetic resonance data to initialize and parameterize our PCa growth models. We believe that our imaging-based models could constitute a promising computational technology to assist physicians to provide a personalized clinical management of PCa.