A new mathematical model for the computation of optimal cancer surveillance schedules
To this end, a promising tool is represented by liquid biopsies. This term generally refers to non-invasive tests aiming to detect and analyze specific biomarkers in bodily fluids. Clinical trials are currently underway for several cancer types to assess the full potential and limitations of liquid biopsies for disease surveillance and tracking over time.
For both screening and relapse monitoring purposes, however, a more general problem arises: even if a test with high sensitivity and specificity is designed and developed, its efficacy in detecting a disease might depend on the times at which the test is taken. We have developed a stochastic model of cancer evolution and ctDNA shedding to determine the optimal testing frequency. We provide a new way to optimize the testing frequency by minimizing the expected tumor detection size for a fixed number of liquid biopsies. With small adjustments, this setup adapts to both cancer screening and cancer relapse detection.
We aim to analyze the NLST dataset to validate our framework. In particular, we will exploit the dataset to:
- infer key parameters of our model, especially for lung cancer;
- compare the model predictions with clinical observations;
- analyze the differences in the results yielded by our framework for different cancer types.
Johannes Reiter - Stanford University
Stefano Avanzini - Stanford University