Saumil Shah, Lisa-Marie Philipp, Susanne Sebens, Arne Traulsen, Michael RaatzRead the preprint
Metastasis is a hallmark of cancer. During spread and establishment tumor cells face different and fluctuating selection pressures (e.g. attack by the immune system in the circulation or resource competition at a tumor site). Tumor cells adapt to these changing pressures by resorting to reprogramming processes that allow them to plastically change their phenotype. Moreover, plasticity may also enable drug tolerance and temporary resistance in tumors.
In our recent preprint, we focus on carcinomas, cancer of epithelial cells, that are known to show epithelial-mesenchymal plasticity. This plasticity is driven by the transition between more proliferative epithelial and more invasive mesenchymal phenotypes (see also this related MathOnco blogpost).
In the context of this plasticity, We wanted to quantify the following:
We use the competitive Lotka-Volterra equations to describe the temporal dynamics of phenotypes along the continuous proliferative-invasive trait axis. We add transition terms between the phenotypes to represent plasticity.
Additionally, we included the differential effects of various interventions on distinct phenotypes. For example, chemotherapy targets cell proliferation; thus, chemotherapy affects epithelial-like cells more. This has fascinating implications for treatment and relapse dynamics (which we have outlined in detail here).
We find that phenotypic plasticity can generate and maintain intra-tumor heterogeneity. Complementing the recently emphasized importance of the microenvironment, we show that also microenvironment-independent phenotypic plasticity can produce experimentally observed patterns of phenotype change between primary and secondary sites. This suggests that invoking complex and distinct tumor microenvironments at different sites of cancers may not be necessary to observe phenotype transitions as microenvironment-independent plasticity could be sufficient to drive the onset of metastasis.
By adding treatment types inspired from chemo- and immunotherapy (growth-dependent and growth-independent), we find that distinct treatment types modify the tumor phenotype distribution differently during treatment. Here, a dynamic phenotype response reduces treatment efficacy gradually as the least affected phenotype acts as a refuge during treatment.
Finally, by in-silico use of an additional, transition-modulating drug inspired by proteins like TGF-β, we 'trap' the tumor in a phenotype most affected by the main treatment. Therefore, controlling and modulating the tumor phenotype distribution may not only maintain the efficiency of the main treatment but actually improve it. Additionally, both metastasis formation and treatment evasion could be influenced by tracking and controlling the tumor phenotype turning plasticity as one of the tumor cells' strengths into a weakness.
© 2023 - The Mathematical Oncology Blog