Gene variations and cancer risk – more results, more answers and more questions

Scientists have found around eighty new gene variations linked to breast, prostate and ovarian cancers

A thousand scientists from one hundred international research groups working over four years. Thirteen papers spread across five journals. DNA analysis of two hundred thousand people. And eighty new genetic variations, or SNPs (pronounced “snips”) linked to three different types of cancer, doubling the current total known about so far.

These are impressive, big figures from an equally impressive, big piece of science, which Cancer Research UK helped to fund (here’s the press release). But what does it all mean?

To find out, we spoke to Professor Doug Easton from the University of Cambridge, one of the leaders of the project.

Cancer Research UK: What exactly are SNPs?

Prof Easton: SNP stands for “single nucleotide polymorphism”, and it’s a single ‘letter’ difference in the DNA between individuals. Your DNA is made up of around 3 billion of these ‘letters’ – there are four possible letters you can have: A, C, T and G – so a SNP is just a single place in your genome where you might have one particular letter, and someone else has a different one.

To explain a bit more about SNPs and what they do, have a look at this short animation:

Cancer Research UK: Are these what we would think of as mutations, such as faults in particular genes? Or are they just variations between us?

Prof Easton: It’s a bit of both. We normally think of mutations as quite rare changes in a particular gene that affect its function, so it makes a faulty protein or is broken in some other way that is harmful to a person.

Most SNPs, of which there are many millions, don’t usually occur in genes (although some of them do), and can be quite common. They may be in the control regions that help to switch a gene on or off, but most of them are just markers – they act as ‘flags’ to tell us that something else that is important in cancer is nearby.

Cancer Research UK: How do you find them, and how do you link them to cancer risk?

Prof Easton: This is the work of a huge consortium of researchers, known as COGS (the Collaborative Oncological Gene-environment Study). What we did was to use a little chip to look at many hundreds of thousands of these SNPs at the same time in a very large number of people. Some of these we already had an inkling they might be involved in cancer from previous studies – the usual suspects, if you like.

We looked at 100,000 people with cancer, focusing on ovarian, breast and prostate cancer patients, and the same number of people who didn’t have cancer. We looked at the frequency of particular SNPs (for example, an A in a particular place as opposed to a C) in the cancer patients and the healthy controls. And if we saw a particular letter come up time and time again in people with a specific type of cancer, but not in the people without, then we can infer that that SNP is probably linked to that cancer in some way.

Altogether we found eighty new SNPs associated with breast, prostate or ovarian cancer, more than doubling the number that we knew about before. We expected to find quite a few, but weren’t sure exactly how many would turn up. In fact, we had a sweepstake amongst all the members of the COGS consortium on how many SNPs we would find, ranging from around five or six to more than a hundred.

There were more than a thousand people involved in this study, spread over more than a hundred research groups. It was quite tricky keeping track of everything as there were a lot of different stages – designing the study, getting the samples together, getting the SNP testing done, having it quality-controlled and analysed and then finally writing up the results for publication.

The whole process took around four years. It’s not what people might think of as traditional science, but I think it’s been really worth it.

Cancer Research UK: How did this whole project start?

Prof Easton: Initially it kicked off to follow up on the results of previous studies that we’d published, which were the first step. We had two major projects to do this – one was a Cancer Research UK grant to look at breast cancer, and the other was a big EU grant to look at the genes involved in breast, ovarian and prostate cancers.

So we decided to combine these efforts together, and at the same time the technology to look at huge numbers of SNPs became available, and we realised there could be an economy of scale – if you do a big number of samples it becomes cheaper. And then more and more people got involved and it kind of snowballed.

Cancer Research UK: Now we have a collection of SNPs that we know are linked to these cancers. How much do we know about what they actually do, or what genes they affect?

Prof Easton: At the moment we only know about very few of them in that kind of detail. We’ve looked at a couple of the regions of the genome in depth and pinned down a few genes that we think are involved in causing cancer when they’re faulty.

But a lot of that work still has to be done. In quite a lot of cases we have a fairly good idea of what the gene involved might be, because there’s an obvious candidate nearby. But for most cases we simply don’t know.

There are some interesting things coming out about the clustering of SNPs in particular places throughout the genome. There are certainly regions where you get multiple hits. For example, we find a lot of SNPs for different types of cancer in a particular region on chromosome 8 – a kind of genomic ‘blackspot’, if you like.

One of the things that came out quite clearly in the COGS papers is that there are a lot more of these blackspots than we previously thought, so you get regions where there’s a hit for all three cancer types. They’re usually not the same SNP, but we presume that they are all affecting the same gene but in different ways.

Cancer Research UK: How close are we to being able to use this information to benefit people?

Prof Easton:  In the very near future, we hope that our knowledge will be helpful in terms of risk prediction – studies are ongoing to figure this out, as we don’t quite know how best to use it. It’s important to point out that you have to take all the SNPs as a set – individually, having a single “bad” SNP is not going to affect your risk very much, but if you have many of them, then it starts to add up.

One obvious example of how this could be used is in women with faults in the BRCA1 or 2 genes. We know that overall they have a greatly increased risk of breast cancer compared to the general population, but there’s quite a big range.

And because the SNPS that influence breast cancer risk in the general population also influence risk in these women, you could imagine a scenario where you can test a range of SNPs and give a woman a better idea of her overall risk and help her make choices about how to play it. If it comes out as, say, 30 per cent, she might just want to have regular screening. But if it’s 80 per cent, then she might want to have her breasts or ovaries removed as a preventive measure.

Looking further into the future over the coming years, this kind of information could change how screening is delivered. At the moment in the UK we have a national “one size fits all” breast screening programme, but it has its limitations. It would make sense to screen people at higher risk from a younger age, and screen people at lower risk less.

As another example, for prostate cancer the SNPs we know about are much more predictive about risk, and we don’t currently have a good screening technique. So knowing more about a man’s individual risk would be very helpful in terms of working out who might benefit from screening or further investigation.

One further idea is chemoprevention – using certain drugs to prevent cancer in those at high risk. This is still at an early stage, and the drugs also come with side effects. But using genetic information to figure out those people for whom the benefits of chemoprevention would outweigh the risks could be useful.

We also know that there are different subtypes of cancer – particularly breast and ovarian cancer – and it looks like different combinations of SNPs can predict the specific subtype of cancer a person has, which might affect how they are treated. Understanding that in more detail is an important future aim.

And finally, the SNP studies we’ve done so far can’t tell us how someone will respond to treatment, but they can tell us a lot about the underlying biology of the disease, which could point to new targets for the development of future treatments.

Cancer Research UK: The cost of whole genome sequencing – looking at the entire genome, not just specific letters –  is falling rapidly. Are the two approaches complementary or will whole genome analysis eventually overtake SNPs?

Prof Easton: In principle, whole genome studies could eventually supercede SNP analysis. But at the moment it’s much more difficult to do a whole genome study, and the cost difference is huge – it’s roughly a hundred times more expensive to do a person’s whole genome than look at all their SNPs using one of our chips, say £3,000 compared to £30.

But the cost of whole genome sequencing is coming down all the time, so it will change. The data handling is much more complex for whole genome analysis, so I think for the next few years they will be complementary approaches. But as the whole genome studies get bigger, they’ll become big enough to be really useful.

The advantage of a whole genome study over SNP analysis is that it can tell you about every variant, not just the ones we’re specifically looking at. At the moment we only find variations in the specific SNPs we’re looking at, and it’s good for looking at common gene variations that confer relatively small increases in cancer risk. But it’s not so good at spotting rare variants that may have a bigger influence. So hopefully more of that kind of information will come out of the whole genome studies in the future.

For now, we still have a huge data set that hasn’t really been mined yet, so there will be a lot of papers to come out over the next few years. We know that there are a few surprises in there, and much more to be discovered on the functional side, exploring what these genes actually do, as well as looking at the different types of cancer. So these are exciting times.

Interview by Kat Arney

Further information

• iCOGs Focus Issue in Nature Genetics