ONCO-36

Personalized prediction and risk assessment of post-radiotherapy biochemical relapse of prostate cancer using mechanistic forecasts of prostate-specific antigen dynamics under uncertainty

SMB2025 SMB2025 Follow
Share this

Miguel AnxoVicente Pardal

Universidade da Coruña
"Personalized prediction and risk assessment of post-radiotherapy biochemical relapse of prostate cancer using mechanistic forecasts of prostate-specific antigen dynamics under uncertainty"
The analysis of prostate-specific antigen (PSA) dynamics after external beam radiotherapy is crucial for detecting prostate cancer recurrence. A significant increase in PSA post-radiotherapy often indicates biochemical relapse, although this evolution can be gradual and may take many years to manifest. Current clinical criteria for defining biochemical relapse rely on observation of population-based markers, using fixed thresholds to assess patient progression after a minimum value of PSA is reached. However, this approach does not account for individual tumor dynamics, which may delay recurrence detection and subsequent treatment. To overcome this limitation, we propose anticipating PSA increases using patient-specific forecasts obtained with a mechanistic model that describes post-radiotherapy tumor dynamics. This model utilizes longitudinal PSA measurements, which are routinely collected as part of standard-of-care management for prostate cancer before and after radiotherapy. By applying Bayesian calibration to the model using these data series, we can thus predict patient-specific PSA dynamics, accounting for the uncertainties in the model and data. Additionally, we can obtain the probabilistic distribution of key model-based biomarkers of biochemical relapse (e.g., surviving tumor cell proliferation rate, PSA nadir, and time to biochemical relapse), allowing for early identification of biochemically-relapsing patients. By leveraging our probabilistic formulation, we also introduce risk measures based on the distributions of these biomarkers to allow for a more accurate assessment of an individual’s risk of relapse (e.g., superquantiles). Finally, although validation in larger, more diverse cohorts is needed and extensions of the model could be implemented, this approach has the potential to improve clinical decision-making by personalizing the monitoring of prostate cancer patients after radiotherapy and anticipating disease progression to advanced stages.
Additional authors:


SMB2025
#SMB2025 Follow
Annual Meeting for the Society for Mathematical Biology, 2025.