Childhood Cancer 2016
Making sense of genetic associations with childhood leukaemia risk
Prof Mehmet Dorak - Liverpool Hope University, Liverpool, UK
I’m presenting actually one poster and one oral presentation I’m doing and in the oral presentation that is scheduled for today, it’s the last presentation in fact, I will be talking about how we can make sense, biological sense, of the genetic associations we are observing in these genome-wide association studies because they are usually just statistical correlations and then someone has to make biological sense of these results. That is what we have tried and I will present our take on the overall genome-wide association study results published so far as well as the meaning of a finding, a novel finding, we have observed ourselves but has not been published yet. So I will be presenting the initial results of our own study and biological meanings that we have been able to infer so far.
What is the area of your study?
The overall team is basically what causes childhood leukaemia and we are more on the genetic side however I have also worked in Environmental Health departments in my career so we never lose our sight on the environment. We try to combine both the environmental and genetic effects by primarily all we are doing is genetic association studies to obtain information from these genetic correlations with disease risk. So in our field there is a phrase that goes like, and that’s what we are doing, it goes like using genetic epidemiology as a probe for disease biology. So today’s presentation is basically a cumulative effect of all these kinds of efforts others have done and we are adding our own input.
How many people took part in the trial?
In the first part of the talk I’m going to review what is already published by others in genome-wide studies in childhood leukaemia. So it’s too many to count but mainly these studies are from populations of European descent, USA, UK and all of Europe included. So all together this is thousands of cases and thousands of healthy controls. We didn’t do a better meta-analysis or anything like that because all studies, more or less, have shown two genes are involved in this very consistently. So we are just basically looking at those two genes and the variance of those two genes and how they may be modifying the risk for childhood leukaemia by what mechanisms. We have mainly used dry laboratory, as it may be called, bioinformatics approaches using existing information from all genome surveys mainly in healthy people but there is data from leukaemia and other cancers as well. So we had a cumulative look at all of these and we have some conclusions that I am going to point out.
In the second half we are going to present our own study which was a pilot study, very small size, very modest size, for a genome-wide study. But it’s a pilot study and it is a novel approach – we did not use any controls, we only looked at 237 leukemic children, their genomes, in a genome-wide study and comparison groups were males versus females because it is very well known that males have a higher risk of childhood leukaemia but the reasons are totally unclear. So we wanted to get an insight, gain an insight, into this differential and looking at the genetic associations, how they differ for males against females. So we basically used females as a control group in looking at males, what genetic markers are enriched in that group which would then implicate those genes and those variants of those genes in the high risk for males.
What did you find?
First of all, this study was based on a US sample and samples were recruited in Baylor College of Medicine, so I have to give credit to my colleagues, my collaborators, there. The study was a PhD project for one of my students when I was working at Florida International University in my previous post. The study is now complete and what we have found is a genetic marker came up as the most significant difference between males and females which had not been implicated in any other disease before. But this isn’t necessarily pointing towards a false positive result because there’s no precedent. It points towards a methodological issue because all other studies so far have been analysed as a whole group, males and females together. But in this study we deliberately looked for markers that are present at a higher frequency in males and at a lower frequency in females so that we can see the difference. So if you put them together compared with healthy controls there will be no difference, that is why we believe this marker has never come up before. But it is a genetic variant in a highly cancer related gene, again that has not been implicated before. It is an inhibitor of a very well-known proto-oncogene, meaning a cancer causing gene, also highly relevant in childhood leukaemia. So the gene itself is RAS-SF2 and it binds to the proto-oncogene RAS which is very well known by everybody involved in cancer studies. This gene, this variant, increases the expression levels or protein levels of this RAS-binding protein, RAS-SF2; we believe this is how it modifies the risk. The statistical effect is that it is protective for males and increasing the risk for females and by increasing the levels of expression of this protein, which is a tumour suppressor gene because it’s inhibiting a proto-oncogene, that is how we believe it decreases the risk for males. But it is only statistical correlation, when I say decreases it will sound wrong to my colleagues because I’m implicating causality. Of course we don’t know that, it’s only statistical correlation. For that reason we have actually submitted a proposal to look at this at the experimental level and to confirm causality of this statistical observation.
What needs to happen next?
The next one is, first of all, as the rules of the game, we have to first replicate this finding. This is a very novel first time observation in a pilot study so we are well aware of that, we don’t want to make a big meal out of this. But we are very confident that it will be replicated in a second study but rather than spending a lot of money in a brand new study we are contacting other people who have made similar studies, even existing genome-wide association datasets can be used, it will be very economical to do so. So we are contacting other researchers in the field whether they can revisit their data and look at male specificity of this association which they may have missed inadvertently. So this is the first thing.
Once it is replicated then we will have a more solid ground to go further ahead and the next step will be, of course, biological experiments in the labs to confirm that what we are inferring from other existing data indeed happens and it makes a difference in childhood leukaemia. So this will require clinical samples and myself and my colleagues in Baylor College of Medicine have already made plans and currently, of course, are seeking funding to do these experiments. So we’ll be looking at healthy cells from healthy children and also leukemic cells from leukemic children to see, depending on the presence of this genetic marker or not we see the expected consequences of this genetic variation. Then that would add further credibility to currently what is our hypothesis.
What would be the ultimate aim of all this?
The ultimate idea is to take all of these to the clinic for the benefit of these patients but because I am primarily an epidemiologist I am also equally interested in prevention. So first of all this looks like a very strong marker, much stronger than previously reported markers, in terms of what’s called in our field effect size. So its impact is greater than any other marker known except that this is going to be male versus female or male-specific and female-specific risk modification. So a preventive value may well be very high but again from an epidemiological or prevention point of view. Whatever intervention will be applied it will have to be sex-specific because a one size fits all type of approach would actually harm half of the population because, as I have already said, this marker confers protection for males, risk for females. If you only go to enhance this effect then you will be harming the females in the population. Therefore it is a very, very new approach and we will bear this in mind in our future studies. So the ultimate aim is to use this as a marker for prediction of future leukaemia so this could be like if there’s another leukaemia case in the family of course the family will be very nervous. If this marker becomes a biomarker, and there’s a big difference, a lot of other steps in between, then it would be a nice sex-specific marker for childhood leukaemia. But this applies to any other marker that has been identified except that this may be a little bit stronger and it will be sex specific. But ultimately, of course, the treatment, there is a big reality out there that we are still seeing a lot of childhood leukaemia, it is still a big burden, both psychological, social and financial to the governments in healthcare.
So the ultimate aim is to find ways, either pharmacological or dietary or other ways, to make the modifications that this variant is making, maybe using more natural ways in people who are missing this genotype, this marker, and then ultimately have the same benefit. In fact I’m going to be presenting this afternoon that for some of the previously observed associations they affect the expression levels of certain genes and using an existing database accessible over the internet we will be presenting this afternoon that some of these genes are influenced the same way as the risk markers for leukaemia by some already known carcinogen exposures. I will be giving an example of valproic acid, for example, which is an anti-epileptic agent. It is already known that it can cause leukaemia as an adverse effect and it happens that it increases the expression levels of a particular gene called R85-B and most of the already established childhood leukaemia risk markers also act and increase the risk by elevating the expression levels of the same gene. So this is the starting point, there may already be actual remedies out there that will cause the same changes that the genetic variants, either protective or risk, are causing in our genome so we can use them accordingly for prevention is the idea. But pharmacological interventions, of course, are always out there.
What is your opinion on the balance between the environment and genetics in causing childhood leukaemia?
As a geneticist I actually talk more about the environment in my classes, in my lectures and in my research also I always keep an eye on the environment. First of all I don’t believe that phrase nature versus nurture kind of things, there is no such thing, they always act hand in hand together. So they are inseparable. There is no way we can give a number for their individual contributions but, again, as an intelligent guess I will go along with my colleagues in the field that the environment actually has a higher say in our risk. The genetic markers mainly act as modifiers or intermediates of environmental effects, so they make us less or more susceptible to the environmental effects.
But in childhood leukaemia it’s a slightly different story. The markers that have already been identified, and very strong ones, seem to be primarily genetic but this is because they come out of purely genetic studies. But there are similar environmental effects, as I have just mentioned, on the same genes but overall in all cancers experts believe, and I am one of them, that the environment has a higher say than genetics.