What are the key attributes of a biomarker, which has achieved widespread adoption? How do you spot a potential winner in the sea of biomarker research? We speak to Mr. Chris Peters, Surgeon and clinical senior lecturer at Imperial College, and his team, to find out.
Summary
Data suggests that less than 2% of biomarkers proposed in the scientific literature achieve clinical adoption with the rest stalling or failing, normally after only one paper.
Successful biomarkers have high quality clinical utility studies and transition to high throughput systems at an early stage, making them more practical to adopt in the real world.
Healthcare stakeholders can utilise a biomarker toolkit (>120 attributes of successful biomarkers) to direct clinical development plans, with the aim of producing a biomarker that will make a difference to patients.
Mr. Chris Peters and his team discusses how to improve biomarker research, with the aim making an impact on patients
Centivis: Why is biomarker research and development suboptimal today?
Chris Peters: We live in an era that we like to think serves personalised treatments, but in reality, the vast majority of medicine that's delivered, isn't particularly personalised yet.
In last couple of decades, our ability to do molecular omics, whether it's genomics, proteomics, or any kind of omics has exponentially increased with amazing advances in technology.
Trawling for biomarker discovery has become easy with data and artificial intelligence. However, what has not increased is our ability to turn these discoveries into clinical tools.
The success rate of biomarkers is small. When we looked into breast cancer, we found that less than 2% of biomarkers have been successfully adopted, meaning that over 98% of them have stalled.
There is a gap between research and development and how biomarkers are brought to the market to benefit patients. Academics are often experts in their platform of choice. But they are less well-placed to think about how a biomarker brings value to different stakeholders, how it integrates for health care pathway or what the potential barriers to adoption are. Therefore, many biomarkers get discovered but much fewer get brought to market to benefit patients.
Just 1.13% of published Breast cancer recurrence biomarkers have reached the patient and most have only ever appeared in one paper.
Centivis: What makes a successful biomarker?
Chris Peters: We conducted some research looking at a huge volume of published guidelines for Biomarker research and then tested the desirable features in over 400 papers in breast and colorectal cancer. We found that the biggest difference in successful biomarkers is carrying out a clinical utility studies earlier and in higher volume.
Successful biomarkers have robust analytical validity. But it is not the performance of the biomarker that is the distinguishing factor. Because actually, if you look at published biomarkers, their performance is all very good because otherwise they wouldn't have been published.
Successful biomarkers usually begin doing clinical utility studies quite early on, usually year 5 after first publication. These are studies that don’t just consider how well the Biomarker works but think about how they will impact on a clinical service- what diagnosing a condition earlier means for the patient for example. It is these studies that actually show outcomes will be improved. The lack of clinical utility studies in biomarker research could partially be due to academic grant funding cycles-- it is often easier to get funding for biomarker discovery and apply a platform to a population, rather than doing a large-scale clinical utility study of a biomarker in a clinical setting. This leads to a cycle of partial validation, discovery, partial validation and so on.
One other attribute which correlated with biomarker success is early transition onto high throughput clinical quality equipment. So, moving away from hand pipetting and doing manual PCR- essentially doing kind of case sample assessment- and transitioning to a high throughput semi-automated system that is far more likely to be scalable in the real world.
In short, successful biomarkers have robust analytical validity and good clinical validity. They also have good clinical utility studies, which should be done suitably early. Early transition to a high throughput system also makes a difference.
Centivis: What can we do to solve the problem?
Chris Peters: My team, Melody NI, Katerina-Vanessa Savva and Alice Baggaley and I and working to develop a solution dubbed the “biomarker toolkit”. It contains over 120 attributes of biomarker research, which we have demonstrated are beneficial in creating a robust biomarker. These attributes are broadly divided into the rationale of the biomarker, analytical validity, clinical validity and clinical utility. It is effectively a scoring system for a biomarker that will allow you to determine how likely that biomarker is to be successful.
If you are a diagnostic manufacturer, it could help you pick a winner to invest in, amongst a sea of publications. Or you might be able to use it to guide your clinical development plans of the academics doing the research.
If you are an academic, we really hope the biomarker toolkit can focus your mind on how you should be doing biomarker research, so that your hard work eventually benefits patients.
If you are a Payer, this could provide structure to assess a biomarker and the value of adopting it.
For funders, the biomarker toolkit could not only provide a way to assess the quality of grant applications, but also shape products and applications.
Also, the biomarker toolkit has potential for rescuing the 98% of stalled biomarkers. Many will have immense promise and could potentially change outcomes for the patients, but they are languishing in one or two research articles and never to go further.
We would like to explore the idea of applying the toolkit to these biomarkers. It may identify the stalled biomarkers which really are worthy of further research. When assessing stalled and successful biomarkers with the toolkit, we saw that some stalled ones actually do not look dissimilar to the successful counterparts. The question is whether those biomarkers with a bit of more push and development could potentially be beneficial.
Key elements of the biomarker toolkit
Centivis: Is the biomarker toolkit specific to any disease area? How did you develop it?
Chris Peters: We developed that list of over 120 attributes by doing an exhaustive literature search with biomarker guidelines (e.g., REMARK criteria). We evaluated papers, editorials and opinion pieces, and collated the attributes that were reflected in the literature. Then we grouped them together and combined similar ones.
We then did a series of stakeholder interviews and and surveys with industry academics, clinicians, patients, to refine that list. We removed a few attributes that were judged to be less important and added some from patients that were not covered by it. This is how we ended up with our final list.
The attributes we came up with are mostly generic, good laboratory practice and good scientific principles and solid clinical utility assessments and value assessments. We have tested the toolkit so far in prognostic biomarkers breast cancer, colorectal cancer and we are currently testing it in diagnostic biomarkers in colon cancer. We also tested it in a biomarker for hepatitis. The toolkit may be applied to biomarkers across disease areas, and we do not expect big differences in how it performs.
Centivis: How are access and reimbursement challenges represented in this toolkit?
Chris Peters: We are in the process of building biomarker toolkit 2.0. In this version, we want to encourage users to think about the use of the biomarker and why healthcare systems would adopt it. For example, where a biomarker for earlier diagnosis of the colorectal cancer is going to lead to improved outcomes, decreased spending on chemotherapy, decreased loss of life etc. These are attributes that socialised healthcare systems, and organisations like NICE may consider.
It is not just about discovering a biomarker of convenience. We want to prompt researchers to think to themselves, why is this biomarker going to be valuable (or not)?
Centivis: Can industry play a role in helping to develop biomarkers in a more optimal manner?
Chris Peters: We are planning for a biomarker matchmaking event between industry and academia, hopefully in November this year. This is being held under the banner of a new group we have created called BREACH- Biomedical Research Exchange for Academics, Clinicians, and Health tech. This is a group set up by Imperial College. NIHR London IVD Cooperative, the Diagnostics Growth hub, and MedCity We have got support from London universities, Oxford, Cambridge, Liverpool, Bradford, Belfast and Edinburgh. We will be asking academics from those centres to pitch their research and explain why they think their research question is important, why the biomarker works and explain next steps planned.
We will also be asking participating companies to present their biomarker technology, explain how their platform works, what samples their platform can be applied to. We hope to find a match between academics and industry, so they can pursue next steps together. We are really keen to hear from new companies are keen to engage with academics and either wish to sponsor this event or just present their platforms at it.
Interested in finding out more about the biomarker toolkit, or participating in the biomarker matchmaking event? Contact Chris Peters at Christopher.peters@imperial.ac.uk