I had the plenary lecture today and I talked about a few different topics. The first one was, to give you a little bit of background, on single cell genomics and translational applications, especially as they pertain to breast cancer tissues. I talked a bit about some of the new single cell sequencing methods we’ve developed that have spatial resolution so that you can pick out individual cells in a breast tissue section and look at pre-malignancies to understand if they are still in the ducts or if they have already started to invade the surrounding tissues. So I talked about our data on that project which reported this multiclonal invasion model of invasion which is a new way to think about invasion from an evolutionary perspective but also has important translational implications.

The second thing I talked about was our studies on chemoresistance in triple negative breast cancer and there we’ve used both single cell DNA sequencing methods as well as single cell RNA sequencing methods to track both genomic evolution and phenotypic evolution in response to neoadjuvant chemotherapy. What we found from that study which looked at around twenty patients, a subset of those, eight patients, were looked at with single cell methods in a lot of detail, was that genetic mutations were pre-existing in the population of tumour cells and when you treat the tumour with therapy you’re eliminating most of the tumour cells but you select for small populations that then expand afterwards and that’s true for genetic lesions such as copy number alterations but in terms of transcriptional reprogramming it looks like the therapy itself is inducing a reprogramming of the transcriptional profiles that cause the cells to be chemoresistant. We identified several signatures and pathways we think are interesting and could potentially be targeted to overcome chemoresistance which is a problem for triple negative patients because about 50% of those patients develop resistance within one to two years.
The final project was a new project that we’ve been working on called the Breast Cell Atlas. This has nothing to do with cancer directly but you can think of it as a human genome map of all the different cell types in the normal breast, including the epithelial components, so luminal and basal, but also the normal immune cells as well as things like fibroblasts and endothelial cells that are part of the vasculature. Trying to create a reference of what the normal transcriptional programmes and epigenetic programmes of those cell types are so that when we start to study cancer and we want to understand not just the tumour cells but the microenvironment we know what a normal reference should look like and so we can say that in these early lesions, maybe in DCIS or early breast cancer, we can see changes in fibroblasts but maybe not changes in other cell types. We can look at those pathways that are being co-opted and changed and they can have clinical significance. So it’s a team science project to really have this nice reference of the cell types in the human breast and it’s a bit more related to normal biology as well as normal variation across different women because we think there’s a lot of biological variables that are important, such as menopause, pregnancy, breast size, breast density, those different factors that can affect the cell types.

We developed a method called topographic single cell sequencing. The problem is that most single cell methods developed so far, genomic methods, require that you dissociate a tissue into a cell suspension and you lose all the spatial information; you don’t know where every individual cell came from in the initial tissue. So this is a method that is compatible with H&E sections from pathology so you can have important pathological features – morphologies, ducts, lobules, different structures – and then you can pick out individual cells from those structures and study their genomic profiles. We looked at synchronous DCIS IDC, so that means they both have areas that are in situ and still within the basement membrane of the ducts and it has areas adjacent that are invasive. We can study this transition as the tumour cells go from these premalignant lesions in the ducts into the invasive areas and start to expand to form the tumour mass. So by tracking copy number changes but also mutations we found that most of the action happens in the ducts prior to invasion. So this led to the idea that early pre-programming of the genome in terms of copy number aberrations and mutations are likely to cause those tumour cells to either remain indolent for the lifetime of the patient or cause those tumour cells to become invasive later on as they continue to progress and expand.

One analogy that people like to use that I did not come up with myself is a fruit smoothie. So most TCGA studies, tumour studies, that you see are going to be you take your tumour, you grind it up, you put it in a blender and then you get this mixture of lots of different DNA and RNA and you sequence everything and you try to figure out what’s in that mixture of DNA and RNA. Of course, a lot of times half of it is stroma, half of it is tumour, there’s a lot of diversity within the tumour cells, there’s a lot of diversity within the stroma; there might be immune cells or maybe blood coming in. Single cell sequencing is more like a fruit cup where you can tell what the berries are, the strawberries or the different components, and you can determine which mutations or transcriptional programmes are associated with all those different fruits that are essentially in the mixture. So it gives a very granular resolution of a tumour and it allows you to remove effects of cell types that you might not be interested in. So if you want to look at how is a fibroblast in a tumour different than a normal fibroblast you can do a cell type specific differential expression analysis which just hasn’t been possible before. With DNA you can look at individual cells and you can completely resolve the heterogeneity and even though you can get a list of mutations from a tumour this will tell you which combinations of those mutations are present in any given cell or subpopulation within the tumour.

Then the other thing is just to be able to really study rare subpopulations that haven’t been amenable to genomic methods before, not just circulating tumour cells but also things like cancer stem cells or other rare subpopulations.

This is consistent with what we call punctuated evolution which we’ve been studying for some time in copy number where we think these early initial bursts in genome instability can give rise to lots of different clones but then is turned off later on and those clones that are generated and are viable and have a fitness advantage they will continue to expand and will eventually escape the ducts. So this data in pre-malignancy is DCIS, where there really isn’t a lot of genomic data so far, is consistent with the evolutionary models we’ve inferred from the more invasive cancers which is good news. But, again, this study was about ten patients and it did focus on synchronous DCIS and IDC and a future direction and even better study would be to have pure DCIS samples and then recurrent samples 5-10 years later where the patient comes back and has IDC. Then you can look at patients that progress to IDC and patients that remain pure DCIS for their whole lifetimes which is better than the synchronous samples. The problem, though, is that all this material is FFPE and that’s been very difficult for single cell genomics so we’ve been doing a lot of effort and work into developing single cell DNA sequencing methods that are compatible with FFPE tissues so that we can do studies like this.

So what’s next?

The question is how far along are these technologies, are they translational at this point in time or are they still only in the expert labs. So for single cell RNA sequencing it’s been heavily commercialised because it’s being used not just in cancer research but in neurobiology, in development, immunology, pretty much every field. So a number of companies, 10x Genomics, Bio-Rad, a number of companies developed methods to do very high throughput single cell RNA sequencing that are based on papers that came out in 2015 that used droplet-like technologies to encapsulate beads and cells and then be able to profile 10,000 cells in one experiment. So that’s brought the costs down a lot and has also marketised the methods. Now they are widely available and it’s becoming a standard molecular biology tool.

Single cell DNA sequencing has been a little bit harder to develop the methods, it requires multiple steps in chemistry and, unlike RNA where you have lots of molecules that you can amplify, DNA you’re usually dealing with a diploid genome which is two molecules; you don’t have the poly(A) tail which is used for a lot of the chemistries. But single cell DNA sequencing has come a long way, we reported the first method in 2011 and it has been commercialised also very recently, just in the last few months or last year. So it’s very interesting to see as these technologies go from expert labs like our lab into the clinic and into more research labs around the world what types of applications they’ll be used for. So I think it’s a pretty exciting time; we’re going past the initial technology development phases, which have really taken almost ten years or so, into much more stable technologies and now we can think about applying them on large numbers, hundreds of patients and also in a more clinical framework where you have a system that can easily be run by a technician or someone else. So there are some important translational applications, both in looking at circulating tumour cells to do longitudinal monitoring of therapeutic response, also to look at scarce clinical samples such as fine needle aspirations in the breast or core biopsy samples, to do immunoprofiling, microenvironment profiling. And also to guide targeted therapies because you can get very detailed maps of the clonal diversity in the tumour but also the lineage of the tumour cells and you can make decisions about do you want to target a driver mutation that was acquired early and was inherited by all the tumour cells so that you can target everything, or is there a subpopulation that you think is going to be more metastatic and should you only target that population.

So there are a lot of exciting clinical applications and us and other groups are starting to work on projects that use these technologies. Circulating tumour cells are probably the one that’s going to be the most translational first, just because there are good techniques to do enumeration and there are also pretty good techniques for picking out the CTCs. So now we have the techniques to actually do the single cell, genome or transcriptome profiling.

Is there anything you would like to add?

No, I don’t think so. I think it’s an exciting time and these technologies have matured now to the point where we can start looking at large cohorts of patients, using them to analyse samples from clinical trials. It really gives us a lot of information.