Modelling cancer and bacterial evolution
Last week was a good week. A bunch of mathematical modelers specialized on evolution met in the town of Ploen, Germany, to discuss what the bacterial and cancer modeling communities could learn from each other. You can check Sandy Anderson's storify here. Jacob Scott also has a brief blog entry you can read here. This post just covers some of my highlights. Both the bacterial and cancer evolution communities have been working on mathematical models to study how their populations of interest (bacteria and cancer) evolve over space and time. There are a lot of commonalities between both fields. Both study complex evolving and heterogeneous populations where treatment and treatment evasion is important. But people working on bacterial evolution have been at it for much longer. Also, as a result of close and mature collaborations with experimentalists, and in many cases, because they established their own labs as well, their models are better parameterised and validated. Furthermore, bacteria are much faster than cancer cells so everything can be studied faster. As a result, much could be utilised from what bacterial researchers have learned about how to efficiently combine treatments to eradicate or control pathogenic bacterial populations. It might almost look like an exchange between bacterial and cancer modelers would be very one-sided. Where us, cancer modelers ,excel though is in spatial models that can recapitulate the spatial architecture of the microenvironment of the disease.
Other than bacterial vs cancer modelers, another gap that has to be bridged was the one between those studying evolution at the genetic level and those working at the phenotypic level. Those of you that have followed our work will probably be familiar with some of the main reasons I think that studying cancer phenotypes makes a lot of sense: main traits are common across different types of cancers and, fundamentally, selection acts on the phenotype, not the genotype. Andrea Sottoriva on the other hand, shows that working at the gene level presents many advantages: the definition of gene is much clearer, you can tag them, track them and study evolutionary history through them. He could have easily added that many of the newer treatments aim to target specific genes rather than phenotypic traits.
A number of talks (admittedly IMO-dominated) discussed the possibility that treatments could be used to steer tumour evolution and for it to lead to less lethal tumour populations. This might be a generalized version of the sucker's gambit suggested by Carlo Maley a while ago. But less about finding a population that is not heterogeneous and drug-able and more about making it controllable by maintaining heterogeneity. This is something that our group will be exploring in the context of bone metastases very soon.