Computer Model Predicts Therapy Outcomes in Prostate Cancer Patients With Bone Metastases

Prostate cancer typically metastasizes to bone tissue, resulting in lesions in the part of the bone where the cycle of bone formation and degradation occurs. To better understand this process and to help predict the efficacy of potential therapies, researchers at the Moffitt Cancer Center in Tampa, Florida have created a computational model to simulate the bone metastasis process and to predict the outcomes of specific prostate cancer therapies. The study was published in Cancer Research.

The model can be tailored to quickly assess putative therapies for patients at risk or with prostate cancer bone metastases, according to the authors.

Studying the bone metastasis process from a single metastatic prostate cancer cell is particularly difficult. The utility of this model, according to the authors, is to reproduce the process based on the knowledge that has been accumulated on the normal bone and bone metastasis processes. The goal is to understand the optimal timing for giving patients anti–bone metastasis therapy to provide maximum benefit-an insight that has thus far remained elusive.

Arturo Araujo, PhD, a postdoctoral fellow at the Moffitt Cancer Center, and colleagues created a computer model that is detailed enough to simulate the distinct phases of both bone formation and degradation, as well as the role of the bone microenvironment in the process, and also gives researchers the ability to zero in on a single metastatic cell within the bone. Model parameters include speed of movement of different cell types and volume of bone resorbed per day.

The model also allows simulations of interactions of different cell types and different microenvironment factors, such as the receptor activator of nuclear factor kappa-B ligand (RANKL) and transforming growth factor (TFG)-beta proteins.

A single metastatic prostate cell in the model was able to simulate the bone metastasis process 7 out of 25 times, according to the study. The model has been able to generate a bone lesion from a single metastatic cell 18 out of 25 times, suggesting that not every metastatic cancer cell that escapes the primary tumor and ends up in the bone tissue will create a bone lesion. The simulations were partly validated in an in vivo mouse model where it was observed that the tumor growth rate predicted from the mathematical model was similar to that seen in mice.

“The advantage of using a computational model is that, just as experimental data is being fed into the model, the model also informs the experiments,” Araujo told Cancer Network. “This allows us to rigorously test our understanding of the biology and to be able to predict how the interactions between different cell types and the microenvironment drive overall tumor behavior.”

Applying anti-RANKL and bisphosphonate therapies to the model, it showed that anti-RANKL therapy was more effective, completely eradicating bone metastases. When a reduced dose of anti-RANKL therapy was used, the results were similar to those seen in the clinic, suggesting that tailored dosing of anti-RANKL therapy may improve patient survival.

The authors are now improving on their model by integrating further algorithms based on biological data, including additional cellular components that contribute to cancer intra-tumor heterogeneity. “Integrating patient-derived information in regards to mutations, it will also be possible to personalize computational models in a bid to provide better treatment options for each patient,” said Araujo.

This study was funded by the US Department of Defense and the National Cancer Institute.