Predictive Biomarker Research

A predictive biomarker is a test which provides information for medical teams and patients about how likely a particular treatment is to benefit that patient and or the risk of side effects. Using predictive biomarkers is an important part of the aim to deliver “personalised medicine”.  That is, that we need to move away from using the same treatment for everyone, towards choosing treatments more likely to work. There are in fact fewer predictive biomarkers actually being used in practice in 2017, than hoped. It has indeed proved difficult to develop such tests. These are some questions and answers about biomarker research.

Q. Why would treatments for advanced melanoma differ for different patients?

A. Melanoma is a cancer of pigment cells and all melanomas share some characteristics but they also differ: the genetic changes which define cancers occur in the cancer cells at random and cancers are all therefore a little different. Some changes are of no consequence but others mean that a particular drug may or may not to what it is intended to do. Similarly, only a proportion of patients develop most drug side effects and it may be that lifestyles, inherited difference between patients or other health issues play a role.  Personalised medicine aims to be able to predict side effects and whether or not a drug will work so that patients and their medical teams can choose the best option.

 

Q. Why has it been so difficult to find good predictive biomarkers?

A. There are many reasons. First it requires very large studies of many patients willing to give samples of their blood or their tumours: maybe thousands of patients being treated in many different clinics. Second, the patients need to be followed up for some time, and information about their response to the treatment carefully collected at each visit. Third, the tests performed must be chosen correctly based upon science, the samples must be collected and processed in exactly the same way in all the clinics, the chosen tests must perform very well in the laboratory and the analysis performed correctly. In fact the way to do the analyses optimally is still being developed using computer tests which are both statistical or a specialist statistical approach called “machine learning”. Finally, the scientific approaches to the analysis of samples is changing all the time: that is that options are increasing all the time and the treatment options are changing all the time so large scale studies set up to recruit sufficient patients and test them in a particular way must be “future proofed” so that the results stand the test of time.

Q. What will improve the identification of good biomarkers?

A. We think that the best way is to work together across the UK and internationally to perform large studies, using the best tests and getting the best analysis teams together. Indeed this is the view of the UK Medical Research Council which funds MRC Stratified Medicine Consortia. Some of the GenoMEL/MELGEN research groups have submitted an application in June 2017 to the MRC for a Melanoma Stratified Medicine consortium. It is very important to involve patients in discussions about research such as this and Professor Newton-Bishop presented the issues to the UK Melanoma Patient Conference in June 2017. We hope that some of the attendees at this meeting might advise the consortium if funded about how the research should be performed and how to develop a biomarker of meaningful value to patients. A link to this talk is provided here.