AACR Annual Meeting 2013
Genomic signatures and predictive biomarkers in breast cancer
Dr Charles Swanton – Cancer Research UK, London, UK
We’re interested in understanding how representative a single biopsy is of the entire genomic landscape of a tumour and we’re interested in how diversity within the tumour impacts upon survival outcomes and, ultimately, resistance to therapy and, in running experiments that we’ve been doing, trying to understand how tumours adapt and change over time and how therapy might influence that. So what we’re planning, going forward, is to try to set up longitudinal studies where we’re looking at tumours repeatedly during the disease course and understanding how they adapt, how diversity comes about in tumours and overall what the impact of that is on outcome.
How different is the metastasised tumour from the primary?
That’s certainly what the genomic studies have revealed over the last two or three years, that there are differences between the primary tumour and the metastatic site, there are differences within a primary tumour and then differences within individual metastases and between metastases. And these differences we strongly suspect are probably likely to contribute to some of the reasons for therapeutic failure in the clinical setting.
What does this mean for tumour resistance in therapy?
I mean, if you think about it in ecological terms, diversity is sort of the spice of life. Unfortunately the same laws probably apply in cancers as well, that the more heterogeneous or diverse a tumour, that there is certainly evidence going back through the literature in the last fifteen years or so that the more diverse a tumour, the worse the outcome. And I guess that makes sense, the more likely there is to be a resistant clone nestling in a sub-population of cells that when you apply a drug grows out through therapy and Darwinian selective terms. How therapy might adapt to the problems of heterogeneity, I think the first thing is that the obvious opportunity here is to think about new ways of structuring clinical trials; this is something we’re very interested in. Instead of calling it actionable mutation and actionable mutation based on whether it’s present in a tumour or not, I think some understanding of whether that actionable mutation is clonally dominant, i.e. is it an early founder event in the growth of the tumour, present in the trunk of the tumour’s evolutionary tree, if you like. By applying either multi-region sequencing or ultra-deep sequencing analysis we can start to resolve the patterns of mutations within an individual biopsy or between biopsies and actually identify whether those mutations are early founder events or later events in the tumour branches that are present in some sub-clones but not others or present in some regions of the tumour and not others. So the obvious study would be to set up a therapeutic study where you treat patients based on the identification of an actionable somatic mutation and then stratify progression free survival outcomes based on whether or not that mutation was clonally dominant or not. And the hypothesis would be that patients who have clonally dominant driver mutations that are actionable would do better with a targeted therapy than patients where that actionable event is present in some cells but not others.
Are there any therapies already developed against these so-called master regulators, as you put it? And I think the answer to that is almost certainly yes and if you take, for instance, EGFR activating mutations in non-small cell lung cancer, why is it that gefitinib is such an active drug or erlotinib is such an active drug in this context? And the hypothesis is, for us, that the EGFR activating event is an early founder event present in all sub-clones. And it may well be that the biomarkers that have really stood the test of time are those biomarkers that aren’t subject to tumour sampling bias present in some sub-clones but not others. But those are the biomarkers that are really validated and qualified for clinical use are the biomarkers that are essential early founder events present in all cells and therefore represent better drug targets.
Is there any cross tumour heterogeneity?
So we call that inter-tumour heterogeneity. So there are gross differences between tumours, genomic differences, and also within individual tumour types we’re seeing. Take ovarian cancer, there are very few mutations in ovarian cancer that are shared in more than 10% of patients. There’s gross sematic inter-tumour heterogeneity in many solid tumours with very few mutations being recurrent and present across large proportions of patients suffering from that disease subtype.
Do you think clinical trials should change?
We would like to see clinical trials be structured in this way, or some clinical trials be structured in this way, to address the relevance of heterogeneity longitudinally and in prospective studies. I think it’s important to say, though, that just by identifying what’s going on in the trunk of the tumour, we shouldn’t ignore what’s going on in the branches too. The diversity, as we mentioned a few minutes ago, is also likely to be the reason why the drugs stop working over time because there will be a sub-clone of cells, one or two cells, that will have a particular somatic event that enables them to overcome drug exposure. Despite the trunkal actionable mutation present in all cells there will be a minority population of cells that hold that actionable event but also harbour additional somatic changes, mutational changes, that allow them to become resistant to a drug over time. So, in other words, we’ve got to understand the diversity, I think, as well as understand the clonally dominant actionable events too.
Is it too complicated to have patients change therapy during treatment?
No, I don’t think it’s too complicated at all. I think that that is certainly one of the, for us, the holy grail of cancer therapy in a way, that we start to pre-empt what the tumour is about to do next and do it earlier than we would normally be able to do on a CT scan. So we react to tumours changing in the clinic. I’d like to envisage a future where we proactively manage cancers and block them going down particular evolutionary routes and adding a drug before the tumour has had the opportunity to adapt through that route and evolve down a particular pathway. That’s obviously a long way off, it may not be possible but that’s a future I’d be very optimistic about.
How would you like to see pharmaceutical companies help in this situation?
The problem with complexity, the level of complexity that we’re seeing, is it doesn’t fit terribly well in a standard pharma business model. This is the complexity we’re all witnessing, not just our labs but many others, in both haematological and solid tumours is pretty awe-inspiring, I would say, and is likely to be very costly to get to grips with. I don’t think there’s any doubt about that. How we get to grips with it is really another matter entirely and I think we’ve got to take one step at a time here. As I mentioned earlier, stratifying patients based on whether the targetable event is clonally dominant or not would be a good place to start. But in due course I’d like to see greater efforts towards really understanding what’s driving diversity in tumours and that’s something our lab is very interested in. We published a paper in Nature about a month ago in colorectal cancer models showing that a particular region of the genome is consistently lost in diverse colorectal cancers, so called aneuploid colorectal cancer. And that region of chromosome 18 that’s commonly lost in those diverse tumours encodes two or three genes that act to maintain genome stability and when they’re lost from the cancer genome you get this, we think, spray of diversity that creates evolutionary fitness and the power for tumours to adapt. So I think, for me, that’s where I see our lab’s future – really understanding how tumours drive diversity, how diversity comes about, what the genetic events are that cause diversity in tumours and ultimately the goal will be to target that and stop it happening or in advance of it happening or perhaps once it’s happened to try to limit any further diversity from happening by targeting the very processes in the cancer cell that generate the heterogeneity.
