From tissue regulation to cancer biomarkers: stochastic models of early tumorigenesis

The initiation of cancer is tightly linked to the failure of normal regulatory mechanisms that control cell populations in healthy tissue. These mechanisms are inherently stochastic, yet highly structured, giving rise to regulated variability in cell population levels rather than uncontrolled growth. In this talk, I will present stochastic modeling frameworks that describe how homeostasis is maintained in normal tissue, and how it is disrupted by early oncogenic events.

I will first focus on mathematical models of cell population dynamics in colonic crypts, which are the basic structural units of the colonic epithelium and the site where colorectal cancer originates. I develop multi-type stochastic branching process models with feedback to describe cell population dynamics within the crypt. These models explain how normal tissue regulation is maintained in healthy crypts, how crypt size changes following the first oncogenic mutations in colorectal tumorigenesis, and how intracellular population dynamics can trigger crypt fission, that is, the principal mechanism driving the expansion of premalignant lesions in the large intestine. Such stochastic, multi-scale models provide a quantitative framework for understanding how early cellular-level disruptions lead to tumor-level growth.

I will then present a second project focused on developing a mechanistic model for the dynamics of cell-free DNA (cfDNA), a cancer biomarker circulating in the bloodstream with potential for early cancer detection. Using simple mechanistic models informed by experimental data, I will show that the presence of early-stage cancer induces a systematic increase in circulating DNA that arises -counterintuitively- not from the tumor itself, but predominantly from healthy tissue. The stochastic analysis identifies saturation of cfDNA clearance in the liver as a critical mechanism amplifying cfDNA levels in early-stage cancer, providing a quantitative explanation for the observed increases in cfDNA in the absence of large tumor burden.