ICEP stands for the Integrated Cancer Epidemiology Programme and it’s a programme of work that combines genetics, something called epigenetics and metabolomics. What we’re looking at is the relationship of genetics, metabolomics and epigenetics to the causes of various cancers – why are some cancers rising in incidence, for example oesophageal cancer or renal cancer. We’re also interested in using genetics, epigenetics and metabolomics in what we call risk prediction, so can we identify people at greater risk of developing prostate cancer based on all those biomarkers and can we identify people who are more at risk of progression of disease based on those biomarkers.
So ICEP is a programme really with two main aims: one is what causes cancer and secondly can we predict cancer and can we predict its progression.
How does this integrate with big data?
What we do is we’re taking data from very, very large consortia of various different types of cancers. So, for example, we’re taking data from a consortium of approximately 30,000 prostate cancer cases and 40,000 prostate cancer control groups. We’re using that huge amount of data to really advance our understanding of what really causes cancer and also what potential mechanisms of cancer might be. So, for example, there’s a lot of work going on into trying to understand why obesity might cause different types of cancers; does obesity explain the rise in some cancers? We can use the access to these very, very large consortia not only to understand what causes cancer but what the mechanisms are that explain the relationship of different exposures to cancer such as why does obesity cause cancer – what are the metabolic pathways by which obesity, for example, might cause cancers.
What kind of collaborations are needed?
A major facet of our programme is collaborations with the International Agency for Research in Cancer; we collaborate with scientists there and the scientists at that agency have access to large consortia in Europe, Australia and the USA. Similarly we also have collaborations with various consortia in the UK and in the USA. Those collaborations are very, very important and the willingness of those collaborations to share data with us is very, very important. One of the things we’re doing within our collaboration to try and make sure that science advances more rapidly is we’re taking that big data, we’re putting it into various bioinformatics platforms that hold the data and that perform standard automated analyses and we’re putting that on open access software so that the whole scientific community can access those data and can access the analytical platforms that we have developed.
How can you employ modelling?
What we’re trying to do in terms of modelling, our main efforts there are to try and model what might improve the detection of different cancers. So, for example, we have done studies of DNA methylation; we have looked to see whether changes in DNA methylation that are associated with lung cancer can improve the detection of lung cancer and identify people who should be screened for lung cancer. By modelling that data, that DNA methylation data, we have shown that incorporating DNA methylation into a model of risk prediction for lung cancer could improve the effectiveness of screening for lung cancer. That’s the kind of modelling activity that we’re undertaking.
What other developments are likely to come from this project?
Some of the outputs so far, we’ve only been going for two years, but some of the outputs so far have been these potential new biomarkers for restratified screening in lung cancer. We have also contributed to the IARC efforts to identify which cancers are associated with obesity and IARC published that data last year in one of their monographs and that has become, then, the basis for a Cancer Research UK public information campaign. So we’re contributing robust evidence to public information campaigns. We are hoping as well that the work that we’re doing will identify eventually potential drug targets because we’re looking not just at metabolomics and epigenetics but we’re also looking at proteomics. By identifying proteins that are related to various different types of cancers and cancer progression we may be able to identify drug targets. So in the short term, and working with industry, we may be able to identify promising targets that industry would then take forward for further research. So those are the kinds of activities that we’re doing.
We’re also advising people who are thinking about undertaking randomised controlled trials about what sort of interventions they should be looking at. What sort of interventions should you prioritise if you’re trying to prevent, for example, the progression of prostate cancer? So we have been having talks with people who are running a big randomised controlled trial of aspirin as an adjunct to chemotherapy in various cancers and they want to know whether they should add in other low toxicity adjunctive treatments such as vitamin D. We have done a big study looking at the effect of vitamin D on various different types of cancers and that sort of data can then be used to decide whether or not you add in these kinds of low toxicity interventions into treatments for people for various different types of cancers and whether those kinds of low toxicity interventions should be entered into trials.
How can doctors and clinicians get involved?
Our main analytical platform is called MR Base; that contains data on nearly a million cases and controls and many, many, many cancers, many, many other disease outcomes and thousands of different exposures, different phenotypic traits. So you can access that data through the website. We have tutorials on how to use that data, how to analyse it and how to interpret it.
Anything else of interest that has come from this?
Another piece of software that we have developed that is open to scientists around the world is something that we have called MELODI which is essentially an acronym for Mining the Literature to Identify Mechanisms of Different Outcomes. So people can go to the MELODI website and they can search the worldwide literature for mechanisms linking different exposures to different outcomes. That’s a piece of software that our students are using, for example, when they’re trying to do systematic reviews of why they’ve seen different relationships. Biochemists within the university are very interested in using that software again to understand mechanisms when they identify particular interactions between metabolites and outcomes. So that’s another piece of software that people can use to try and understand mechanisms underpinning diseases.
