Cancer genomics and pancreatic cancer

Share :
Published: 11 Nov 2014
Views: 4247
Prof Andrew Biankin - University of Glasgow, Glasgow, UK

Prof Biankin talks to ecancertv at NCRI 2014 about the advancement of cancer genomics via cross-institutional collaboration, with specific reference to pancreatic cancer and the difficulty in finding biomarkers.

We’re really going to be talking about our International Cancer Genome Consortium project and the data that are emerging from that. It’s a team effort; initially we started in Australia as the Australian Pancreatic Cancer Genome Initiative with Professor Sean Grimmond who has also now moved to Glasgow. Now essentially we’re really working on advancing cancer genomics through to the clinic with our team members and colleagues in Glasgow and also through the UK, really working with the NCRI networks etc. to really try and take these data into the clinic. At the moment we’re just getting some early signals but when you start to look closely at these pancreatic cancer genomes, you actually see targets to existing therapies but they often occur at low frequency or low prevalence. So how do we design particular clinical trials to really get to those? In the case of pancreatic cancer really we haven’t made any progress in fifty years. We do have some therapies that will sort of work and that’s probably because they work in subgroups of patients but we don’t know which groups they are. So we’re really working on, number one, identifying what particular biomarkers might be good for selecting patients for existing therapies, so better using the therapies we currently have, particularly those that are particularly toxic, and secondly seeing if we can repurpose or rescue therapeutics that haven’t shown efficacy in all comers types of clinical trials but developing biomarkers to better target those patients. And, of course, identifying novel therapeutic targets through our extensive cancer genome sequencing.

What has your research found so far and what types of pancreatic cancer will you be looking at?

I’ll be talking about really the map or the catalogue of genetic mutations we’re starting to see in pancreatic cancer, we’re up to about 450 pancreatic cancers at the moment, looking at what driving mutations might be driving those. But my particular focus is really looking at what are actionable targets or what we call actionable phenotypes where we see good targets for existing therapeutics where we can start to target those particular subtypes more appropriately and some other responses that we might be seeing already. When we designed the study initially we did it prospectively so that way we could track every individual patient, what treatment they received, how they responded. We couldn’t intervene like you do in a clinical trial and it wasn’t really a controlled environment, but having that degree of detail, clinical data, was quite important so that we can start looking back and looking for signals.

We also made mouse models from many of these samples, patient-derived xenografts and some cell lines and we could perform therapeutic testing in those to really complement some of the signals we were starting to see in the patient data. This starts to drive a biomarker hypothesis where we’re trying to push things into the clinic.

What are the genes that you have been looking at?

We’ve been looking, in a broad sense, for some of the therapies that are already existing, so platinum based therapies. I know there’s a lot of activity with regard to looking at what subtypes might be susceptible to these particular therapies and also other therapies that target signalling mechanisms such as PARP inhibitors. At the moment we’re in-depth analysing many of those genomes and it’s hard to get the right numbers but we certainly see exceptional responders, patients that respond dramatically to a particular therapeutic type, and then try to really unpick what might be the responsible mutation or what might be the particular phenotype that makes that cancer susceptible to that particular therapeutic.

How have patients responded to treatment?

It’s complex because they often receive combinations of therapies etc. but we have some hypothesised biomarkers that we think might be relevant. But it’s early days yet and these are just really hypothesis generating to identify something that we might test in a clinical trial. The real challenge is how do we do those clinical trials and how do we do that in a current environment that really isn’t geared for that? What we really want to do in the future is every single pancreatic cancer patient at diagnosis gets a good biopsy that we can sequence, we can identify a particular opportunity for a therapeutic and give the right drug to the right patients in a clinical trial setting. So because at the moment the drugs that we give mostly don’t work and we don’t get a lot of patients enrolled into clinical trials because the opportunities just aren’t there. What we want to do is create an environment where every single patient with pancreatic cancer has the opportunity for novel therapy because what we’re doing at the moment really isn’t working. The way we can do that is to take a good biopsy of every patient at the time of diagnosis. At this point in time we’re very pragmatic, we use a selection of gene approach where we use a panel which is diagnostically tractable. We also integrate that with a research environment where we have a dynamic platform that allows for stratified therapeutic development. So we take the sample from the patient, not only do we sequence that to try and offer them a clinical trial, a targeted clinical trial if a particular therapeutic exists for their particular subtype, and if not then some of those therapies, some of the newer therapies like immunotherapy where, because we’ve got the sample, we’re already set to go back and see if only a subgroup responded why did that particular subgroup respond? And then from that sample we also try, at least in some, is to create new model systems, organoids, initially re-programmed cells, patient derived xenografts etc., so then we can start to study and understand why they’re inherently resistant, why particular therapies are effective and then we have a model system to work with.

We also want to try and track these patients so that when they recur before they can be re-enrolled into this particular study then, in that circumstance, we can take those samples and do exactly the same thing, look for a target, grow them and start to understand clonal evolution, start to understand potential novel targets that are emerging, start to identify biomarkers of therapeutic responsiveness and resistance that can help us in the clinic but also understand biology more in-depth.

Are we looking to simply avoid unnecessary treatment or find new therapies too?

Well, it’s not really over-treatment, it’s treatment that doesn’t particularly work. So, at the moment, there are probably two or three key therapies that we use in pancreatic cancer which is a very different situation to what we had five or six years ago where we only really had one - gemcitabine. Now we have gemcitabine that can be combined with either capecitabine or erlotinib and if you look at these studies they make this much difference, it’s really, basically, only three weeks or four weeks difference. But there are subgroups of patients that respond so what’s happening in this case is that you’re giving people treatment that doesn’t work. That’s OK if the cancer takes a long time to progress but usually pancreatic cancer progresses very rapidly and you only really have one good shot and only about half of patients get any treatment at all. Only half of those, or less, get a second line therapy and only about 10% of those get a third line therapy. So you’ve really got to try and get it right in the first instance. We have some newer therapies, a combination called FOLFIRINOX and also a combination called gemcitabine plus Abraxane, or nab-paclitaxel, which are options for fitter patients. But there is some associated toxicity and, again, similarly who do you decide which treatment to give? Do you roll a dice, do you take a guess? We need to understand which therapy we want to give first in this circumstance where the stakes are very high. So we want to introduce stratified or personalised or precision medicine for pancreatic cancer because we’re not really making much progress but the evidence we’re seeing at the moment is that the degree of diversity of the disease and the fact that we’re seeing very small signals in clinical trials may be because we’re not really targeting the right patients.

Can anything be done to catch this earlier?

Early detection is also a great way to move forward. One of the challenges is really accessing the pancreas with these precursor lesions, understanding these precursor lesions, and that’s where our challenge is at the moment. If you have a colonic polyp, for example, you can take a biopsy of that relatively straightforwardly. The pancreas is deep in the retroperitoneum, hard to access, hard to image, and the challenge is really getting appropriate samples. There are certainly lots of people working on it using techniques like endoscopic ultrasound etc. to try and identify pancreatic cancer early. We certainly do have a lead time, some of the more recent studies that have shown that we have something like twenty years between the cell first initiating on a pathway of carcinogenesis until the tumour mass appears.