Understanding metastasis in the light of evolution

Metastasis is the most common cause of cancer-related mortality.

Understanding the mechanisms of this process will allow us to bring greater insight to bear on the design of treatment strategies. This could be the early prediction of the metastatic potential of primary tumours, or the development of specific drug targets with the highest potential of dissemination.

However, our understanding of the metastatic process is currently poor. Limited by the complexity resulting from the many distinct subpopulations of cancer cells, or clones, that can have different metastatic potentials, analysis have proved to be difficult. While existing DNA sequencing technologies provide a way to explore the genome of these cells, these technologies analyse unknown mixtures of millions of different cells together. This limits our ability to use these data for studying different tumour clones and their metastatic potential.

Escaping the mix

Now, single-cell DNA sequencing technologies offer an exciting way of overcoming this lack of specificity.

These technologies enable us to analyse thousands of individual cancer cells in parallel – this is the key because it allows the direct investigation of cells belonging to different clones. The data generated by these new technologies are profoundly different than those previously generated and require the development of new computational methods to be accurately analysed.

And it is just this – the use of novel single-cell DNA sequencing and development of associated mathematical models – that my research group is working on as part of a recently CRUK-funded project. We’ll do this work  as part of the non-small cell lung cancer national study TRACERx and the PEACE autopsy programme, providing a unique platform to integrate several orthogonal and clinical data into our study.

Our aim is to obtain a much better understanding of metastasis, specifically around three questions: which specific cancer cells disseminate to seed metastases, where do they disseminate, and when does their dissemination start?

Which?

Identifying which specific cancer cells disseminate and seed metastases is challenging. Each primary tumour is composed of many different clones and, in principle, metastatic cancer cells might originate from any of these clones.

An added layer of complexity here is that after the successful seeding of metastases, metastatic cancer cells keep evolving by further accumulating many other genomic mutations and structural alterations that differentiate these cells more and more from their originating clones in the primary tumour. For this reason, this first question cannot be easily answered by simply comparing the cancer cells identified in primary tumours with those identified in matched metastases.

However, we can address it in the light of evolution. During cell divisions, each cancer cell passes on all the genomic mutations and alterations acquired so far to their daughter cells, leaving unique markers that can be used to track back the evolutionary history of each cell to its ancestors. This is, of course, similar to phylogenetic studies used to investigate the history of human evolution and the tree of life of all species on our planet.

We are developing computational methods that reconstruct the evolutionary history of metastatic cancer cells using the novel single-cell DNA sequencing data. The hope is to identify hallmarks that uniquely distinguish metastatic vs non-metastatic tumour clones, which can be incorporated into models to predict the metastatic potential of different tumour clones.

Where?

Metastatic cancer cells do not simply migrate from the primary tumour to one metastatic site but generally undergo much more complex dissemination patterns. And this is true at the clonal level as well – cancer cells from the same tumour clone can seed multiple metastases in the same or different anatomical sites.

Also, metastatic cancer cells located in one metastasis can further disseminate and be responsible of seeding other metastases (a phenomenon often referred to as “met-to-met seeding”), or they can return back and “re-seed” the primary tumour (if it is still in situ) or other previously-seeded metastases. Reconstructing these migrations patterns is challenging because we do not observe this process while it is ongoing, but we rely on data obtained after this complex migration process has already occurred.

However, once again, we can address this question in the light of evolution. If we know the parental relationships between different tumour clones, we can estimate the metastatic migrations that have occurred by comparing the anatomical location of parental clones with that of their descendants. I like the analogy with human migration here – if my parents are in Italy and I am currently in the UK, we can reasonably guess that I must have migrated from Italy to UK.

In this study, we are aiming to develop algorithms to reconstruct metastatic migration patterns by utilising the evolutionary histories of different tumour clones. The hope is to identify hallmarks of tumour clones that tend to disseminate to certain anatomical sites, providing the possibility to predict where certain clones could seed metastases before they actually occur.

When?

Metastases can be seeded by disseminating cancer cells years before they are clinically detectable, or they can be seeded suddenly and grow at a very high pace.

Disentangling these two possibilities is challenging because metastatic cancer cells can seed micro metastases composed of small niches of cancer cells in distant anatomical sites that can remain undetectable using patient clinical imaging.

However, once again, we can address this question in the light of evolution. Cancer cells acquire certain types of genomic mutations at a constant rate through time and during cell divisions. Therefore, counting the number of these mutations in metastatic cancer cells can be used to obtain a “molecular clock” which can be used to estimate the timing of metastatic seeding.

We are developing algorithms to count the number of these “clock” mutations in individual cancer cells and translate these numbers into estimates for the timing of different steps – the which and where – of the metastatic cascade. The hope is to identify metastatic hallmarks to predict whether metastasis has already started at some point during disease progression (for example, at the time of surgery), and thus inform the use of additional, systematic treatments to target potential metastatic cells that have already disseminated.

Incorporating many disciplines

Thanks to the support of CRUK, we are generating a unique dataset of 192,000 cancer cells from primary tumours and matched metastases obtained from patients with non-small cell lung cancer enrolled in the TRACERx study and in the PEACE autopsy programme.

Generating this dataset and answering these three important questions is only achievable by integrating the multi-disciplinary expertise in our group. Computer scientists, evolutionary biologists, cancer genomicists, clinical oncologists and more need to come together to make progress. Not only that, we must also incorporate this study into the multi-disciplinary and collaborative network of the TRACERx and PEACE studies.

For this study, it’s clear that formally integrating the expertise from different disciplines is of fundamental importance. As our project develops, I hope that not only can we shed light on the metastatic process, but also show that integration of a multidisciplinary approach will be important if we are to enhance our understanding of cancer and to identify opportunities for translating this into benefits for patients.