University of Texas at Austin


Using Biomechanistic Modeling, Researchers Predict Prostate Cancer Growth

By Aira Balasubramanian

Published April 3, 2024

Examples of personalized predictions of prostate cancer growth using a biomechanistic model.

Prostate cancer is an exceedingly common disease, affecting 1 in 8 men. When diagnosed early, many can live for years without symptoms. Rather than subject early-disease stage patients to the debilitating side-effects of radiation and surgery, clinicians typically monitor these patients through a process called Active Surveillance (AS), using magnetic resonance imaging (MRI) data to track tumor progression, which guides decisions regarding when to begin treatment. Current medical practice uses population level statistics in order to predict and understand how prostate tumors tend to grow. However, this population-based lens may impact detection of patient-specific tumor progression, leading to delayed implementation of life-saving treatment. 

New research led by Guillermo Lorenzo, postdoctoral researcher at the Oden Institute of Computational Engineering and Sciences, is catalyzing a revolution in personalized prostate cancer tumor forecasting. Published in Cancer Research Communications, his team’s  paper explores how computational models (similarly used by Oden Institute researchers to predict how gliomas and breast cancers will respond to treatment) can be used to predict how tumors develop prior to the treatment stage of this disease. According to Thomas Yankeelov, Oden Institute core faculty and Director of the Center for Computational Oncology, this work “underscores the next generation of tools for guiding the treatment of prostate cancer.”

Lorenzo’s team developed a biomechanistic model that uses MRI data from patients’ routine screenings’ during active surveillance to provide personalized tumor forecasting, a coveted capability not yet available in clinical practice. By analyzing cancer cell mobility, which increases tumor size, and the rate of cancer cell division, which increases tumor density, their model is able to computationally and spatiotemporally predict how these interplaying mechanisms will influence future tumor growth within the 3D geometry of a patient’s prostate. 

The research team’s technology also allows for the classification and forecasting of  dynamic changes in biomarkers that correlate with high-risk tumors. While many prostate cancers begin as low-risk, low-symptom conditions that do not require intervention, they can develop into high-risk cases requiring immediate surgery or radiotherapy. These changes may be missed in the intermittent MRI screenings used in standard-of-care active surveillance. Protocols for detecting increasing cancer risk levels do not account for the vast heterogeneity of prostate cancer growth dynamics across patients.

“Our computational technology for tumor forecasting is a decisive step towards defining a predictive, personalized strategy for the clinical management of prostate cancer that combats the treatment excesses and deficiencies from the current observational, population-based protocols," said Lorenzo. 

Our computational technology for tumor forecasting is a decisive step towards defining a predictive, personalized strategy for the clinical management of prostate cancer.

— Guillermo Lorenzo

By classifying biomarker dynamics that are indicative of quickly-progressing or more aggressive tumors, Lorenzo’s research allows for early identification of higher risk tumors, and helps clinicians determine the optimal timing of each MRI scan and ultimately, treatment administration. Based on this model, “risk can be determined and treatment planned at an earlier stage than ever before. This is the theme of our work, and we believe its potential to save many lives is significant,” said Thomas J.R Hughes, Oden Institute core faculty and lead of the Computational Mechanics Group

Results from the pilot study show real promise in forecasting individual patients' tumor burden, coupled with the ability to classify prostate cancer risk based on biomarker availability. While many advancements in medical technology are difficult to scale and provide to patients and physicians, Lorenzo’s work is intrinsically accessible.

“Our methods rely on data that is regularly collected from prostate cancer patients in active surveillance after their tumors are diagnosed,” he shared. “In the future, our computational technology is expected to be usable by virtually any medical center with this monitoring approach.” This has the potential to shift the trajectory of prostate cancer's treatment and management at a fundamental level, and provides a glimpse into the future of personalized medicine. 

In the future, our computational technology is expected to be usable by virtually any medical center.

— Guillermo Lorenzo

The research team behind Lorenzo’s publication is interdisciplinary, and consists of experts across universities and fields. Michael Liss, Jon S. Heiselman, Michael I. Miga, Hector Gomez, Alessandro Reali, Thomas E. Yankeelov (Oden Institute core faculty member, Director of Center for Computational Oncology), and Thomas J. R. Hughes (Oden Institute core faculty member, Director of the Computational Mechanics Group) served as collaborators and mentors throughout the process of development and publication. 

This work has been possible thanks to a Marie Skłodowska-Curie Global Postdoctoral Fellowship that Lorenzo was awarded from the European Commission (GA No. 838786), which supported his research work at UT Austin and the University of Pavia.


Guillermo Lorenzo