Most of the molecular characterisation we’ve done of the cancers has been at the gene level looking at DNA mutations through whole genome sequencing. Then we’re moving on to look at expression of those genes through RNA sequencing so it’s a bit of a different analysis. Trefoil factor 3 and the trefoil genes aren’t particularly mutated in this cancer, it’s just that the cell which is the precursor you get a metaplastic switch of the normal squamous oesophagus to Barrett’s, so you get a fundamental change in the differentiation status of the tissue lining and the glandular lining expresses trefoil factor 3. So it’s not mutated, it’s not a driver of the cancer.
In the sequencing work that we’ve done we’ve sequenced, as part of the International Cancer Genome Consortium, about 400 cases now, published on… the last paper was 129 cases that we characterised in detail. What have we learned? We’ve learned some of the things which we knew already which is that the most commonly mutated gene is p53 which was no surprise. What was a bit disappointing, I suppose, is that there aren’t other recurrently mutated genes like p53 in this cancer, it was hugely heterogeneous. So after p53 which is mutated in 70% of cases there’s a long, long tail of genes which are mutated in 10% or less. It’s a highly mutated cancer, actually a similar number of mutations to that seen in lung cancer and melanoma which have very high carcinogens, environmental exposures. So we think that the driver of this disease in terms of exposures is reflux but quite how that causes a mutation we don’t know, we don’t understand as well as we do for cigarette smoking and UV light.
It’s highly mutated, not many recurrent point mutations and small insertions and deletions but a lot of copy number changes, so a lot of amplifications, deletions, large scale structural rearrangements of the genome that are quite complicated. So actually the genome, in broad terms, is a bit similar to what we see in ovary cancer – p53 is usually mutated, a lot of copy number changes, a lot of structural rearrangement. So we’re working hard now to really characterise some of these structural rearrangements and try and understand how they occur. Some of them seem to occur through crises, chromothripsis type events; whole genome doubling seems to occur quite early on in at least 30% of cases.
What I’m also going to describe is the work we’ve done on looking at mutational signatures. So when we were doing the latest analysis on these 129 genomes what we wanted to do was come up with new targets for therapy. Because it’s so heterogeneous that’s actually very difficult to do except in a truly personalised way which isn’t very practical. A lot of the molecular targeted therapies that have been tried so far haven’t worked very well. So receptor tyrosine kinase inhibitors, for example, inhibiting B2 in fact is the only drug currently licensed and that’s in the metastatic setting for this cancer. So very disappointing results in the trials and because there aren’t many oncogenes recurrently mutated and a lot of the receptor tyrosine kinases are co-amplified, so if you just attack one bit of the pathway you’ve still got something else that will cause a problem, it was difficult to come up with new strategies.
But when we look at the mutational signature, so this is using the trinucleotide signatures described in the COMSIC database by Sanger colleagues Alexandrov and co., then we can have a different way of categorising the patients. What we see is six main signatures in this cancer. So in nearly all cancers you see an aging signature, which we see here; we see something called signature 17 which seems to be quite specific to this cancer, not only seen in this cancer but more commonly than in other cancers and we think that may be a signal from acid damage. When we characterise patients according to the predominant signature seen across their whole genome, then we get three main categories of patients. Actually this is very conserved for that patient so even if you take multiple biopsies from the tumour, which as I said it tends to be very heterogeneous even within a patient, the signatures are not heterogeneous, you get the same result across your three biopsies. Patients really do categorise into these three main types. When we use that kind of approach we can begin to come up with some more novel therapeutic strategies. So the biggest group have this dominant S17 signature which I mentioned and those cases also have the highest mutational burden, they have the highest expressed neoantigens, they have more CD8 cytotoxic T-cells infiltrating their tumour and those are all characteristics that in other tumour types have responded better to immunotherapy. There’s not much data yet on immunotherapy in this cancer type.
So that begins to give us some new ways in. This is hypothesis rather than proven in clinical trial but this is a bit of a different way of thinking about this tumour, maybe, that may give us some different approaches to try.