Zooming in on cancer genes
More than a century ago, the German biologist Theodor Boveri put forward the idea that something inside cells was responsible for causing cancer. Today, we know that the disease is fundamentally driven by faults in the genes that normally act to keep our cells and bodies healthy, only dividing when they should, and dying if they’re damaged.
Kat Arney went to The Institute of Cancer Research in Surrey to meet some of the UK’s leading experts on cancer genetics, to find out where we’ve come from in terms of understanding the faulty genes that drive the disease - whether inherited or picked up during our lifetime - and how this might shape treatment in the future.
First she spoke with cancer genetics expert Professor Nazneen Rahman, to find out how studying genetics is useful in cancer.
Nazneen - There are different ways in which we use genetics in cancer. The first fundamental difference is whether we’re actually testing the cancer itself or whether we’re testing the whole person. If you test the cancer itself, there are all sorts of genetic changes that have made that cell turn into a cancer cell. By doing that kind of genetic testing, we can work out what type cancer it is and what kind of treatment it might have. So that’s one side in which genetics is incredibly useful.
We can also look at the DNA of the whole person to see whether there were some potential reason why they developed cancer in the first place, and also whether or not there might be information there relevant to other members of the family due to inherited predisposition to cancer. So that’s the other way in which genetic information is incredibly helpful in cancer.
Kat - This genetic connection raises the question of whether cancer runs in families. But as I found out from Professor Richard Houlston who’s spent his career tracking down genetic variations that increase cancer risk, just having several cancer cases in a family doesn’t automatically mean that there’s an inherited gene fault at work.
Richard - As many people know, cancer is actually a relatively common disease. So effectively, if everyone lives to a ripe old age then you're actually going to get lots of cancers in the family because essentially – in western countries, it’s that and coronary heart disease and strokethat are the major causes of death. And the next point is that determining cancers are highly environmentally or lifestyle induced. Example, if you had a large number of elderly people who’d actually all developed lung cancer and they're all smokers, you get a conglomeration of cancers within the family simply because of that. What to be alert is usually the fact there might be an inherited disposition in families – these are cancer diagnosed at a young age, and second, when you have a more than one cancer diagnosed in young age of the same type of cancer. So an example is early onset of colorectal cancer in the family under the age of 45, so that’s quite unusual. So you’ve got some essentially designed consolations of specific cancers or site specific cancers, but generally, diagnosed at early age.
Kat - When did people first start noticing this, that it could be hereditary, and then start trying to track down what was going on?
Richard - Warthin in 1913 reported a family, G, that actually had a large number of people who were affected with bowel cancer and also, uterine cancer that appeared to be transmitted in the family. That was effectively assigned to the back shelf of the book case. And then Henry Lynch repick that up and revisited the family in the early ‘70s. He tracked the offspring and all the people in the family G. I think he actually looked at over 120 descendants of the original people in the family and showed that they also had a very high incidence of both bowel and uterine cancer. That then led to the definition of the Lynch syndrome which we now know from molecular testing to be due to mutations in the mismatch repair genes. So effectively, in those cases, mutation in one of the mismatch repair genes causes an increased risk of primarily bowel and also uterine carcinoma.
Kat - The gene faults that tracks strongly in families that significantly increase the risk of specific types of cancer, is that where their story ends or do we know that they are involved in some cancers more commonly in population?
Richard - We know that mismatch repair gene defects actually had around about 13 per cent of all colorectal cancers. They're not actually inherited. They're just actually acquired mutations. They underlie a specific form of bowel cancer which tends to be associated with better prognosis, and in fact, responds differently to drug medication.
Kat - In people in families that have this strong risk, they have inherited a lot of these genes that normally keep cancer under control. But then in cancers that spontaneously arise in people without this inherited risk, it’s something gone wrong in those cells as the cancer develops.
Richard - Yes, that’s correct. Effectively, the point about the mismatch repair aptly illustrates the fact that it’s the same genes, but effectively, they're somatically or acquired mutations in those, whereas in fact, other people, it’s inherited.
Kat - Given that there is more and more genetic data coming out, we have more advanced gene sequencing and gene analysis techniques, what would you like to see in the future in terms of using this information?
Richard - Well, I think it’s fair to say that really, we’ve entered the world of the so called thousand dollar genome but million dollar interpretation problem. So I think the first point is that we actually have to interpret the data we have more effectively – that’s one issue. Second, the other thing is, that you record linkage. It’s going to be critical to understanding the real – from an individual point of view, what do these actually mean? I think it’s fascinating that Genomics England, they tried to actually put record linkage together with whole genome sequencing. That theoretically might produce a faster rate of data that might be useful from our point of view of actually understanding the true functional effects of many of the variations. It’s going to be a very challenging matter to understand the nuances of genetic variation in terms of its effect on the whole person.
Kat - So the challenge is not just getting the information about the genes, but it’s actually tracking that to medical records, and what it means.
Richard - Yeah, I think it is. I mean, I think it puts everything in context and dare I say the word ‘big data’ always creeps, but it is also a big data challenge and that’s going to be a very significant problem in the future.
Kat - It’s certainly a massive challenge, but it’s one that scientists are working hard at. When it comes to the genetic changes involved in cancer, there are relatively rare ‘strong’ hereditary gene faults that significantly increase the risk of one or a few types of cancer in a family. Then there’s a much wider spectrum of genetic variations across the whole population that have a less severe impact on cancer risk.
On top of this, our DNA is under assault every day - from things in the world around us, such as tobacco smoke, and even the fundamental processes of life at work within our cells. And it’s this damage, combined with our underlying genetic makeup that can result in cancer. Thanks to advances in DNA-reading technology known as sequencing. Scientists can now delve into the detailed genetic makeup of an individual patient’s cancer, and even use that information to shape treatment.
One example already in the clinic is the breast cancer drug Herceptin, designed to target an overactive protein molecule resulting from a genetic mistake in some types of breast cancer. But it doesn’t work for every patient because it turns out that all breast cancers are not the same.
To find out how the genetic revolution is changing our understanding of cancer, I spoke to Dr Anguraj Sadanandam, who’s combining lab research with computer modelling to target tumours more effectively.
Anguraj - Conventionally, people were thinking that a breast cancer or colon, or pancreatic cancer is a one disease that has been tried and therapies have been given so far and it was not successful because different people respond differently to these drugs. But with genomics improving so much we can untangle that and try to identify the subdisease within each cancer type and try to match some of it to therapies.
Kat - So basically, all breast cancers are not the same, even though they might all start in the same part of the body.
Anguraj - Yes. So though they have started in the same part of the body, the question is that, are they starting from the same type of the cell? That is one of the interesting question. If they're not starting from the same type of cell then they will have a different prognosis and therapeutic responses.
Kat - So by finding the key genes, the key gene faults that are driving these different groups of cancer cells, how then do you target those with treatments?
Anguraj - These key genes later, they translate into a protein and the proteins are the functional units of the cells. These proteins help particular structure and that structure defines how a cell would move and how the cell would change their shape, and all those things. So, we can try to by computation modelling make certain chemical molecules which would go and bind at that particular structure. So that way, the cells cannot get the signals to move or change shape, so that you can control cancer cells.
Kat - So these are very, very targeted therapies, targeting the faulty products of these genes in the cancer cells, wipe ‘em out.
Anguraj - Yes. So that’s the basic idea so that these cells don’t behave that way.
Kat - What do you say as the really big questions that still needs to be answered when it comes to using genetics to understand tumours and how to treat them? What are still the big mysteries?
Anguraj - So one of the major complications, now we are thinking that these patients, one patient’s tumour is different from another patient. But the much more complex in these within a patient, there are a lot of subtypes and subgroups existing because as I mentioned, if they're starting from the same cell within the tumour, they can become into multiple groups. If you take one patient’s tumour and cut them into let’s say, 20 or 30 pieces, there are possibilities that if they're not going to be in 20 different ways but at least 4 ways or 5 ways, or 10 ways, that is a more complicated thing. Still, we are in a really early stage to understand how these changes are happening. So, if we go and start treating them thinking that it’s one tumour, it’s not going to work.
Kat - I guess it’s like looking at large families and saying, “Well, you’ve all got a lot of things in common” but when we really drill down to it, each member of that family is unique and individual.
Anguraj - So between individuals in a family that may be a significant change to make them behave differently and also, do things differently, similarly, the cancer cells can have very tiny changes but those tiny changes can make them interact differently. At the same time, there are not only cancer cells within a tumour but there are also other cells – immune cells. All those interact and make them even much more complicated and they behave differently then that’s major challenge to untangle right now.
Kat - What do you want to find out next? What are you looking at now?
Anguraj - So the major challenge right now is to find the key sort of genes which we call biomarkers because these biomarkers would define these subgroups. That’s more for us a challenge and in bowel cancer for example, we have defined those genes. Second thing is which type of technology that would take that to the clinic. The question now of how to handle a lot of data, given a patient coming to the clinic in a single day or two is very, very challenging. New methods and technologies are coming which don’t have to deal with thousands and thousands of genes, rather tens and hundreds of genes so that we can give the answer to the patient within a day or two.
Kat - And finally, if you imagine say, five years into the future, what would you like to see cancer testing and cancer treatment look like based on the kind of work that you and your colleagues are doing?
Anguraj - So when a patient enters into the clinic, the blood and the tumour material from that patient should be taken to ask the question that what kind of subgroup they belong to. If that is the case, with existing drugs available, what would be the therapy that they can go for. So for example, patient A entering in, “Okay, you have this particular genetic type so you are qualified for drug X, and patient B, you'd go for drug Y” and that’s the way I'm thinking five years from now. If not completely achieved in a subset of patients, we could probably find a way to go in that direction.
Kat - Advances in gene sequencing have the potential to target cancer in a more precise and personalised way than ever before, but this new wealth of data brings its own problems, as Nazneen Rahman explains.
Nazneen - So we’re at an extremely exciting time in genetics at the moment, we’re able to generate vast amount of information about the genetics for an individual. But our problem really is about curating that information, understanding its meaning. There are lots of different ways in which we’re not able to use that information the best way possible. And so, one of the things that we’re currently doing together with a lot of different people across the UK and indeed the world is an initiative called the Transforming Genetic Medicine Initiative (TGMI). Really, what that’s trying to do is harness all of this information and use it to bring knowledge and hopefully, wisdom so that we can make sure that we’re using the information appropriately to bring the maximum benefits. But also, to make sure that the information is not used inappropriately as well. I think that’s the key challenge for us in genetic medicine currently.
Kat - How do you go about doing that? How are you bringing all this information together?
Nazneen - At many levels, it’s really quite prosaic and a lot of these is sort of logistically trying to pull together lots of different information, and then sort of curating it. So, there are 20,000 genes. At the moment, it’s not entirely clear how many of those genes can be associated with human diseases and how many are not. Historically, there's a lot of unclear information that was unclear before because we are having more and more knowledge but you have to go back now and then look at the previous things and think, “Now, I've got this knowledge. How do we have to re-interpret previous information?” At the very highest level, we would really like to go through all of the genes and say, “Okay, these genes can be associated with human disease and these ones are not” and be able to separate these out. And then for the ones that are associated with human disease, how are they? What diseases? What types of mutations? And then once we sorted that out, how does that then translate into how those then are going to be tested and what are the best ways of doing that. So at the heart, there are really quite simple questions which until now, we haven't had enough data to be able to answer in a sort of comprehensive and careful way.
Kat - It feels like there's a lot of philosophical changes that need to happen. Instead of saying, “You have breast cancer” to saying, “You have this particular genetic subtype of breast cancer”. And then there needs to be a sort of philosophical change in the way we develop drugs, the way we test drugs, the way we license drugs. How can we make that happen?
Nazneen - Well, I think recognising the problem is a key step, debating it. I think with the scale of technological change that allowed us to go from taking 10 years to do one genome to being able to do 10 genomes in one day is such a massive change that it makes the whole system require a sort of ground up, just let’s think about it again and how we’re going to make it work in this modern world. So I think the key thing is really recognising the issues and debating them. There are philosophical debates to be had. At least about the sort of predictive information as well: do we want them as individuals in society? Do we want to have this window into the future? When do we want it? How do we want it? How are we going to act upon it? These are not things that we’ve had to think about very much. So, I think that’s the first thing that has to happen. But then we have to then overlay that with really some quite pragmatic solutions and I think we might have to almost do it in a sort of like experimental way: try things, get the information iteratively then improve it. I don’t think we can solve it in isolation and create the perfect system ahead of time. Not at least because everything is still changing very fast. So, I would like to see some ways in which we can develop the system and improve it as we’re sort of working through it. Otherwise, we’ll always be chasing our tail, and we may never get there.
Kat - Nazneen Rahman and before her Anguraj Sadanandam and Richard Houlston, all from The Institute of Cancer Research.