An in silico, three-dimensional model of tumour evolution suggests that cell motility is a key factor in the initial growth of a tumour mass. The model also reveals the dynamics of mutation spread. See Letter p.261
Evolutionary thinking is becoming an indispensable tool to understand cancer, and even to propose directions in the search for treatment strategies. In this issue, Waclaw et al.1 (page 261) use mathematical modelling based on evolutionary principles to provide an explanation for the observed architecture of tumours, and to argue that cell migration might be the key to tumour shape, spread and drug resistance. This study opens up the possibility of treatments that target genes related to cell motility and adhesion, rather than the conventional targets of genes governing cell division, death and differentiation.
Cancer is an unwanted evolutionary process whereby cells, driven by random mutations, escape the orchestrated behaviour of a functioning tissue and enter a phase of abnormal growth and, later, metastasis (tumour spread). We still lack understanding of many aspects of this complex process, and researchers in different fields are collaborating to solve this ultimate riddle. Evolutionary biologists approach the study of tumours in a similar manner to the study of viruses, bacteria or animals: they seek the mutations that give rise to the ever-changing variety of tumour cells, and they look at the forces of natural selection that allow certain mutants to proliferate, replace their competitors and give rise to new waves of evolutionary change.
Waclaw et al. combined methods from evolutionary biology and ecology with current knowledge of the molecular biology of cancer to design a versatile mathematical model in which tumour-cell populations undergo evolutionary change, guided by realistic parameters. The authors used this model to study the growth laws of 3D in silico tumours, focusing on the tumours' geometry and cellular composition. If the tumour cells are relatively immobile, then, as they proliferate and form a malignant mass, they crowd each other out and thus slow their own replication. Soon the tumour can grow only at its surface, which results in a relatively slow expansion (the mass grows as a cubic power of time).
But this slow growth cannot account for the relatively fast tumour expansion observed in many clinical studies. The resolution of this apparent paradox lies in cellular motility. By giving each cell in the model the ability to migrate, the researchers observed a much faster, exponential, growth, which also yielded a different, more realistic, tumour shape (Fig. 1). This result is consistent with the earlier proposition2 that migratory potential is a component of a cell's evolutionary fitness, in much the same way as is its replicative potential. However, it was previously thought that cell migration was mostly involved in the invasion of tissues by tumours or in metastasis. The direct, pivotal role of cell motility in tumour growth was under-appreciated and can now be considered a valid treatment target.
Another focus of the authors' study was tumour composition. In particular, they asked how quickly a particular mutation can propagate in a mass of cancer cells, thus changing the tumour's properties. Evolutionary processes and their outcomes are largely shaped by the environment in which they take place. For example, evolution in a well-mixed, homogeneous medium takes place at a different pace from evolution in an environment in which interactions are restricted by geometric constraints. And in the latter case, dimensionality is key. For example, it has been shown that inactivation of a tumour-suppressor gene (a two-hit evolutionary process in which the cells must first become less fit before becoming more fit) happens faster in 1D (a row of cells)3,4 than in 2D (a layer), and this is in turn faster than in a fully mixed system with no spatial constraints4,5,6,7. By contrast, in two-step processes in which the intermediate mutant confers a slight selective advantage, the relationship is the opposite, and a non-spatial, fully mixed environment promotes the fastest pace of evolution5. These phenomena seem less surprising if one notes how reminiscent they are of other fundamental laws of nature in which space dimensionality changes how things work, such as the different fundamental solutions of Poisson's equations in 1D and 2D.
Waclaw et al. then set out to understand why, given the high overall degree of tumour heterogeneity, some mutations are so prevalent among the cells of a given tumour. In the context of tumour progression, two broad classes of mutation have been identified8. Cells with driver mutations are characterized by having a growth advantage over other cells, and such mutations are thought to be responsible for cancer initiation and progression. Passenger mutations are genomic changes that do not really alter the cells' growth properties, and do not have a causal role in cancer origin or progression. Waclaw and colleagues show that, in the presence of even a small amount of selective advantage (that is, a driver mutation), the affected cells sweep rapidly through a 3D cell population. This explains the observed composition of large tumours, in which almost every cell contains the same driver mutations, and heterogeneity resulting from passenger mutations accumulates later, during tumour progression.
This idea is crucial in the context of cancer therapy. A mutation that confers resistance to a drug is usually a passenger mutation before treatment; such mutations are 'hiding' inside any tumour and are generated simply by chance as a result of the constant background mutation rate. But resistant mutants immediately gain a selective advantage once treatment is applied. Waclaw and colleagues' paper illustrates how rapidly resistant cells can accumulate, leading to regrowth and treatment failure. This happens even faster in the presence of mutations that increase cellular motility.
How far are we from being able to use evolution to our advantage? Understanding evolution's intricate ways brings us a step closer to being able to reverse malignant processes, and to channel the dynamics in the direction we want. And can we use the genes responsible for cell motility or cell adhesion as targets in future cancer treatments? Waclaw and colleagues' theoretical study suggests that this is a possibility, and it is to be hoped that others will take up the challenge.
Waclaw, B. et al. Nature 525, 261–264 (2015).
Thalhauser, C. J., Lowengrub, J. S., Stupack, D. & Komarova, N. L. Biol. Direct 5, 21 (2010).
Michor, F., Iwasa, Y., Rajagopalan, H., Lengauer, C. & Nowak, M. A. Cell Cycle 3, 356–360 (2004).
Komarova, N. L. Bull. Math. Biol. 68, 1573–1599 (2006).
Komarova, N. L., Shahriyari, L. & Wodarz, D. J. R. Soc. Interface 11, 20140014 (2014).
Durrett, R. & Moseley, S. Ann. Appl. Probab. 25, 104–115 (2015).
Komarova, N. L. Proc. Natl Acad. Sci. USA 111, 10789–10795 (2014).
Haber, D. A. & Settleman, J. Nature 446, 145–146 (2007).
Related links in Nature Research
About this article
Computational and Structural Biotechnology Journal (2019)
Hierarchical tissue organization as a general mechanism to limit the accumulation of somatic mutations
Nature Communications (2017)
PLOS Computational Biology (2017)