The working hypothesis in cancer research is that
cells within a tumor have descended from a single
initiated aberrant cell. In contrast to this clonality
hypothesis, data have emerged which support the polyclonality
of intestinal tumors. I will discuss statistical and
biological aspects of polyclonality from a recent study of mouse
aggregation chimeras. Three important issues are: (1) Random collision:
Is it possible that observed polyclonality can be explained by
the chance proximity of initiated clones? A hypothesis
test is derived using some classical results from stochastic
geometry. (2) What is the spatial extent of interaction among
initiated clones? Techniques from Bayesian image analysis are
used to combine image data on the chimeric patch structure
of the intestinal wall with tumor count data. (3) What
fraction of tumors are polyclonal? The lower bound
proposed by Novelli and colleagues is shown to be flawed
owing to a misinterpretation of conditional probability.
This is a joint project with A. Thliveris, R. Halberg,
L. Clipson, R. Sullivan and W.F. Dove from the McArdle
Laboratory, and S. Stanhope from Statistics. The
central publication is
http://www.pnas.org/cgi/content/short/102/19/6960
