World Cancer Leaders’ Summit 2015
Using tumour genomics in cancer prevention
Dr Paul Brennan - IARC, Lyon, France
IARC has got a massive investment in prevention research and your team in genetics are obviously intimately involved in it. Tell us what the relevance is of genome and prevention.
The main focus of the agency with cancer research is primary prevention, i.e. identifying new causes, and secondary prevention – early detection. If you look at the majority of genome work that has been launched in the last ten years and there have been some major international efforts with lots of funding behind them, things like the International Cancer Genome Consortium, the Cancer Genome Atlas in the US, all of the genome-wide studies that have been launched and the main rationale for those studies is biology, it’s to understand how cancer develops and also really to try and identify new treatments. That’s really the main push behind these studies. We think in the agency that there is an area of how genomics can help inform primary prevention and secondary prevention that has been overlooked and we think that there are great opportunities here.
Give me some examples.
So primary prevention, identifying new causes. There are lots of cancers for which we still really don’t know why there are massive international differences. You can think of colorectal cancer which is one of the most common cancers, we have no idea why it’s so common in a place like the Czech Republic or in a place like Japan and it’s a lot less common even in parts of Europe or in Africa. So one of the ways that we think we can identify new causes or understand the aetiology of some of these cancers is looking at the tumours, looking at the genomics of the tumours, looking at signatures that might link to particular exposures. Just to give you one example, we recently helped to finish a large genomic sequencing study of renal cancers. So this was about a hundred cases spread throughout Europe where we had full genome sequencing on all of these cases. There were about a dozen cases from Romania, now nearly all of these cases had a particular mutation sequence that you only find with exposure to aristolochic acid. It was incredibly bizarre, none of the other cases from any of the other countries had this finding, it was completely unexpected. The one interesting thing about the region of Romania and the other Balkan countries is that they have had an exposure to aristolochic acid that tends to occur along the Danube and causes quite a rare renal condition but this was never suspected as causing renal cancer in this region. So now we’ve gone back, we’ve looked in fresh tumours, fresh renal tumours, we’ve identified aristolochic acid adducts so we’re pretty sure that this is implicated in causing renal cancer in that region. How it causes it or why people are exposed to it we’re still a little bit unsure, whether it’s through contamination of wheat products, which some people suspect, or whether it’s through herbal remedies which have become very common recently. But it does appear to be a new cause of renal cancer in the region. So that’s an example of how looking at the tumour you can identify particular sequences that can link back to particular exposures.
Secondary prevention, early detection, we think that there are some really quite novel techniques coming through that can identify individuals with preclinical cancer symptoms. For example, it’s now apparent that from a blood sample you can detect minute pieces of the tumour in the blood. Now this has been used in the clinical setting to identify patients who might relapse, it’s also been called a liquid biopsy so for tumours that are particularly difficult to get biopsies for one could try and use a blood sample. What we’re interested in at the agency is whether we can use this technique to pick up pieces of the tumour or to identify the tumour in the blood sample before the symptoms occur. So we’ve been looking in very early stage tumours at the moment, so for example we’ve been looking in stage 1 or 2 lung cancers, we’ve been looking in pancreatic cancers and we’ve been looking at these blood samples and we have been able to identify circulating tumour DNA in the blood in a reasonable proportion of these cases. What we need to do now is to identify whether we can identify the circulating tumour DNA before the actual cancer occurs so that means going to a cohort study like the EPIC cohort where we have blood samples before the disease occurs and doing that type of study. Technically it’s incredibly challenging but the pay-off for this could be absolutely enormous if it works.
mRNA is another great example and there have been really quite exciting data for microRNA signatures, so using panels of microRNAs that appear to be different in people in the period before they’re diagnosed with a cancer. There are incredibly exciting results that have been identified for lung cancer in a few small trials, two of them in Italy, and in IARC we’ve been looking in some of our very early stage lung cancer cases and we have replicated the results that others have found. It really does appear to be a genuine finding. We have a study underway at the moment in the EPIC cohort, so looking at microRNA panels, and we know that in some other cohorts as well they’re trying to follow up on these results. This is a great potential to be able to identify high risk individuals who might then be able to be prioritised for screening.
And then 3 and 4?
What else can we do? We can look at Mendelian randomization is another wonderful potential to be able to identify causes of cancer. So it’s using genes to identify causal exposures for lifestyle exposures. So to give an example, we don’t know, for example, if the relationship with obesity and some cancers is causal or not. What we can do is we can look at genes that are associated with obesity and see if those genes are associated with a cancer because this idea of Mendelian randomization is that the genes you inherit that might make you prone to be a little bit heavier or not, these are unlikely to be related to other risk factors, like smoking for example. To give you a practical example, we’ve recently completed an analysis of genes associated with obesity and lung cancer. So we’ve done this on tens of thousands of individuals for which we now have quite complete genome information on lung cancer. Previously observational studies had shown that lung cancer patients tend to be a little bit lower weight; when we run this genetic analysis we don’t find that association. So there appears to be causal mechanism between being underweight and lung cancer. If there was that association it was probably due to confounding by smoking.
One interesting thing from the analysis is that for some of the histological subtypes of lung cancer there appears to be quite a strong association with being overweight. So it could be the case that being overweight could increase the risk of some subtypes of lung cancer. That would appear to be a causal relationship.
Adenocarcinoma, which subtypes?
The strongest was with small cell, there was also an association with squamous cell and for adenocarcinoma there was nothing.
Any other issues that the agency are tackling on genome and prevention?
The last area that could be important is identifying high risk individuals using genetic susceptibility scores because now we’re getting more and more information on the genes that make an individual susceptible to developing a particular cancer. So one way that that information could be particularly useful is identifying individuals who are at genetically high risk so that they might then be prioritised for screening programmes. There has been an awful lot of effort, resources, co-ordination that has been put into some of the large genome sequencing programmes, genetic susceptibility programmes, with respect to trying to identify biological pathways, treatment targets etc. I think we need the same level of international co-ordination and effort to target these programmes for primary and secondary prevention.