MS01 - ONCO-04

Digital twins for clinical oncology and cancer research (Part 1)

Monday, July 14 at 10:20am

SMB2025 Follow

Note: this minisymposia has multiple sessions. The other session is: and Part-2.

Guillermo Lorenzo (University of A Coruna (Spain)), Chengyue Wu (The University of Texas MD Anderson Cancer Center, US), Ernesto A. B. F. Lima (The University of Texas at Austin, US)

Description:

The overall goal of this minisymposium is to present and discuss recent developments of digital twin technologies to (i) address the personalization of clinical management of cancers, and (ii) advance the research on biophysical mechanisms underlying these pathologies from the micro to the macroscale. A digital twin can be defined as a virtual representation of a physical object by means of a computational model that continuously assimilates object-specific data to enable-decision making about the physical object based on its current and future states. Digital twins have a widespread development in multiple areas of engineering and they have also been proven useful in several areas of medicine, such as surgical planning, cardiovascular disease interventions, and glucose monitoring in diabetic patients. Given the increasing success of computational models to predict the development of cancer and its response to treatments, these models could be employed to construct digital twins to support the optimal diagnosis, monitoring, and treatment of individual patients as well as to assist the research of this disease in vitro, in vivo, and in silico. The talks in this minisymposium will present recent efforts in building digital twins for the clinical management of cancers and the (pre)clinical research of these diseases, as well as ongoing work on practical mathematical models and computational methods for the development of these predictive technologies.

Thomas E. Yankeelov

The University of Texas at Austin

"A practical computational framework for systematically investigating alternative treatment strategies for cancer"

Over the last decade our team has developed a set of partial differential equations that capture key features of tumors growth and treatment response related to tumor cell migration, proliferation, treatment response, and tissue mechanical properties. These models can be calibrated with widely-available medical imaging data to accurately predict the spatio-temporal changes of solid tumors in response to both radiation and systemic therapies. We will briefly summarize those results for cancers of the brain and breast, and then show how these models power digital twins to identify alternative therapeutic regimens that are hypothesized to outperform the standard-of-care interventions. In the case of high-grade gliomas, we will show how our digital twins can be used to identify personalized treatment plans predicted to reduce tumor burden 24% more than the standard-of-care approach one month after treatment (n = 15), while maintaining toxicity in the organs-at-risk within acceptable clinical limits. In the case of breast cancer, we have used digital twins to personalize neoadjuvant chemotherapy schedules (n = 105) to yield therapeutic strategies that are hypothesized to increase the pathological complete response rate by at least 20%. Furthermore, we have used our digital twin formalism to virtually recapitulate the results of three key clinical trials that led to the current backbone for neoadjuvant therapy. While the results we will present will focus only on cancers of the brain and breast, we emphasize that since our digital twins are based on key underlying biology and physics features of cancer, they are applicable to any solid tumor for which the requisite imaging data is available.

Maximilian Strobl

Cleveland Clinic

"What pre-clinical experiments can teach us about digital twins for personalized cancer treatment scheduling"

Cancers are complex and evolving diseases. To tackle this complexity there has been growing interest in developing “digital twins” – personalized computational tumor models – to better inform when and how to treat to reduce toxicity and maximize tumor control. As this idea finds traction, the crucial question is how do we ensure efficacy and safety as we translate from bench to bedside? In this study, we test the digital twin approach to treatment scheduling in vitro, in the context of EGFR+ non-small cell lung cancer. Using fluorescent, time-lapse microscopy we characterize the evolutionary dynamics of co-cultures of Gefitinib-sensitive and paired resistant cell lines (PC9) across four different treatment schedules: i) continuous therapy, ii) intermittent therapy (on/off), iii) intermittent therapy (off/on), iv) continuous therapy at half the full dose. Our results demonstrate that both the dose and the frequency of treatment influence evolutionary dynamics. Intermittent therapy minimizes final resistant cell and total cell count after six treatment changes (18 days total), across four dose levels examined (2uM, 200nM, 100nM, 20nM Gefitinib). Moreover, the off/on intermittent schedule outperforms the on/off schedule, suggesting a role for spatial competition in suppressing resistant cells. Next, we test how well three commonly used mathematical models of sensitive-resistant dynamics can predict the observed dynamics: 1) A simple exponential model, 2) A logistic model which accounts for spatial competition, and 3) A 3-population model which includes an additional subpopulation of drug-tolerant cells in the “sensitive” population. While Models 1 and 2 can capture the dynamics under continuous treatment, the more complex Model 3 is required to predict the outcomes of intermittent treatment. Our work illustrates how in vitro experiments can support the development of digital twins, and how this process can uncover new insights into drug resistance evolution in cancer.

Renee Brady-Nicholls

H. Lee Moffitt Cancer Center & Research Institute

"Investigating Response Differences between African American and European American Prostate Cancer Patients Through an In Silico Study"

African American (AA) men have the highest incidence and mortality rates of prostate cancer (PCa) compared to any other racial group. The increased incidence as well as mortality are likely due to socioeconomic factors, environmental exposure, access to care, and biologic variations. Deciphering the specific drivers of increased incidence and mortality is difficult due to a scarcity in available data from AA patients. Mathematical modeling offers the opportunity to run in silico studies to investigate treatment responses in a larger cohort of virtual patients. Here, we investigate response differences between AA and European American (EA) prostate cancer patients receiving hormone therapy. We simulate prostate-specific antigen (PSA) dynamics, using a mathematical model of interactions between PCa stem cells and differentiated cells. We use propensity score matching to identify 15 EA patients that most closely matched the 10 AA patients. Bayesian calibration is used to determine plausible parameter sets that accurately describe longitudinal PSA dynamics on a per-patient basis. Model parameters are compared between AA and EA patients to determine potential drivers of resistance. Our findings show that the initial PSA levels, stem-cell self-renewal, and PSA production rates significantly differ between AA and EA patients. Using the plausible parameter sets drawn from the calibration, we simulate adaptive therapy as a potential strategy to maximally delay progression. Our findings show that both patient groups receive benefit from adaptive therapy when compared to continuous, with AA patients receiving a significantly greater advantage. Our study presents an important step in identifying race-specific, patient-specific treatment options that can be used to maximally delay time to progression.

Fatemeh Beigmohammadi

Université de Montréal

"Efficient methods for generating virtual patient cohorts using trajectory-matching ABC-MCMC"

Virtual patient cohorts (VPCs) are computer-generated representations of patients that mirror real-world populations. VPCs use mechanistic mathematical models to establish the effects of inter-patient variability on disease and treatment outcomes, thereby allowing for the comprehensive exploration of disease mechanisms and therapeutic strategies at low-cost and without burden to patients. Further, they may aid the development and validation of mechanistic mathematical models that aim to capture the underlying mechanisms of biological systems and their responses to drug treatments. Thus, the effective generation of virtual populations is critical. While several methods exist to create VPCs, there is a growing need for more computationally efficient techniques. Previous work combined Approximate Bayesian Computation (ABC) with Markov Chain Monte Carlo (MCMC) to generate heterogeneity in model parameters and predicted outcomes. Unfortunately, because the generated samples must simultaneously meet the ABC and MCMC acceptance criteria, this approach has a very high rejection rate. To reduce this computational burden, we developed a model-based technique we call Trajectory Matching ABC-MCMC (TM-ABC-MCMC). TM-ABC-MCMC modifies the acceptance criteria in ABC-MCMC to only require that model trajectories fall within observed bounds. In this talk, I will discuss the application of TM-ABC-MCMC to three existing mechanistic models of varying complexity and will show that it has a lower rejection rate compared to ABC-MCMC and maintains high fidelity with previous results. Thus, our approach significantly decreases computational costs, boosting the efficiency of virtual patient cohort generation.

#SMB2025 Follow

Annual Meeting for the Society for Mathematical Biology, 2025.