A Multi-Scale Agent Based Model of Colon Carcinogenesis

Colorectal cancer (CRC) is a major cause of cancer mortality and there remain aspects of its
formation which are not understood. The colon contains an epithelium punctuated by flask
shaped invaginations called the crypts of Lieberkühn. These crypts are monoclonal in
nature while adenomas are thought to be polyclonal, suggesting that multiple crypts are
involved in carcinogenesis. It has been reported that fields of mutated tissue surround
adenomas but the causes and growth of these fields are not well understood. There are two
competing hypotheses regarding growth, the first being that mutated cells from one crypt
invade neighbouring crypts, and the second that mutated crypts replicate themselves more
often than wild-type crypts.
To investigate these processes two agent based models were developed. The first model
represents cells as agents and is similar to previous models in the field, but is novel in
including the geometry of the crypt mouth. This is necessary to model multiple interacting
crypts. This model is the first in the literature to be used to represent multiple crypts and is
used to investigate invasion of neighbour crypts by mutated cells. The second model
represents whole crypts as agents, which allows the entire colon to be simulated for multiple
decades of biological time, as far as we are aware this is the first such model.
The cell scale model predicts that crypt invasion does not occur, but that mutated cells can
invade the flat mucosa above neighbouring crypts. Analysis of in-vivo data is consistent with
this prediction. The crypt as agent model predicts fields of ~41,000 crypts, in agreement
with data in the literature, this corresponds to a field ~23mm in diameter. This project
models pre-cancerous fields for the first time over a variety of scales, making specific novel
predictions which are in agreement with in-vivo data where such data exist.
Two agent based models were created to study the development of precancerous fields, one
a model with cells as agents to study cell scale phenomena and the other with crypts as
agents to allow processes to be studied on larger spatial and temporal scales. These
models could potentially be used to refine clinic practice by predicting the required frequency
of post-intervention monitoring of patients or the necessity of further intervention.