At PHG we have been co-leading one of the work packages of the EU-funded PROPHET project which has the overarching goal of providing concrete evidence and recommendations to decision-makers regarding implementation of personalised prevention initiatives across diverse health contexts within Europe. Details are available on the project page.
This February, eighteen months into the project, a milestone was achieved with numerous work packages submitting reports to the EU as well as the release of a paper outlining the project. Our report was the completion of an evaluation of the evidence for clinical utility of biomarkers for use in personalised prevention of common diseases – our contribution researched cardiovascular disease (CVD), while our work package co-lead, the Centro de Investigación Biomédica en Red (CIBER), undertook the same work looking at cancer and neurological disorders. Knowing what evidence for clinical utility exists for tests incorporating biomarkers can help determine where efforts should be focused, and where early wins can be achieved. Below are some reflections on the process and key learnings.
Demonstrating clinical utility
Demonstrating clinical utility of any test associated with a biomarker is not an easy task. Firstly, we recommend in our report that researchers develop a test definition for the biomarker they are studying, since for a biomarker to have clinical utility, it must be incorporated into a test. A test definition describes HOW an assay detects the biomarker, for a particular disease (WHAT), in a particular population (WHO), and for a particular purpose (WHY). This was not always described in some of the papers highlighting the biomarkers, making clinical utility assessment impossible for these particular biomarkers.
Numerous frameworks exist for assessment of clinical utility, for genetic biomarkers the most well-known is the ACCE framework. These frameworks bring together the elements important to assessing clinical utility for prevention. These elements include the ability of a test to identify the risk of a disease (known as analytical validity) as well as cost effectiveness. Other elements, amongst many, include:
- whether the test has been shown to work in the real world with real patients, how and if it can be incorporated into current healthcare workflows,
- that the test does not harm patients,
- that there are interventions available to help prevent or minimise disease as a result of using the test.
Evidence assessments are done by guidance developers (such as the UK’s National Institute for Health and Care Excellence, NICE), regulatory agencies (such as the UK’s Medicines and Healthcare products Regulatory Agency, MHRA), payers (e.g. individuals, laboratories, hospitals, departments of health), clinicians, patient interest groups, advisory committees (such as the UK’s National Screening Committee, NSC), amongst many others. The extent of these assessments will depend on the purpose of the assessment, the characteristics of the test, its intended use and the user group. Assessments by a panel of experts determine if something is suitable for use and can take months, sometimes years.
Searching for evidence of clinical utility
Together with CIBER we found over 1500 unique biomarkers that had been identified in scoping reviews as contributors across the three disease groups – cancer, cardiovascular diseases and neurodegenerative diseases. We developed a search strategy to find evidence of an assessment of clinical utility for tests incorporating novel biomarkers for the personalised prevention of these common non-communicable diseases. Using this strategy we identified published guidelines, health technology assessments (HTA) and cost-effective analyses (CEAs). A key principle is that in the development of these documents, authors and institutions will make some form of assessment of clinical utility of the tests being considered for clinical implementation.
We considered guidelines to be the highest level of evidence since clinical utility assessment is an integral part of the process of guideline development. In contrast, HTAs and CEAs provide assessment of potential clinical utility. By integrating several different databases, we were able to deliver the broadest possible evaluation of evidence. This strategy is disease agnostic and can be expanded for use with any test or disease. Another benefit is its flexibility, allowing users to start at different points depending on the data they are interested in.
Lessons from the results
The searches identified pieces of evidence that recommend the use of specific tests, but do not recommend the use of others, primarily due to a lack of evidence for clinical utility. This demonstrates there is a gap in the evidence needed for the implementation of tests for a comprehensive assessment of clinical utility. We outlined some of the challenges in terms of demonstrating clinical utility in our recent evaluation of polygenic score applications report. Much of the research on new tests focuses on demonstrating analytical validity, whereas the pathway to clinical utility requires additional elements to be demonstrated, such as clinical validity, as well as ethical, legal and social issues (ELSI). We have highlighted the need for a broader suite of evidence as part of the recommendations in our recent PROPHET report.
Another finding from this work was the value that integrated risk models have in prevention. Integrated risk models consider several different variables, such as age, gender and lifestyle, when calculating an individual’s risk of disease. Some of these risk prediction tools are already in use. However, when new variables are added to them new assessments and validation studies are needed to develop evidence that the specific variable is improving the risk calculation.
With more and more data becoming available further resources are needed to integrate data from multiple research areas (e.g. ‘omics), domains (e.g. social, environmental) and the integration of electronic health records into research. This would allow a more holistic view of the usefulness of a test or biomarker for personalised prevention. However evaluation and validation of these risk prediction models is not straightforward and more work is needed to determine how best to do this.
Moving forward
We are still involved with the PROPHET project and will continue to contribute to the development of its strategies and recommendations. Towards the end of the project in 2026, we will write a policy briefing summarising the key findings and recommendations on the implementation of personalised prevention programmes in Europe.
PROPHET is funded by the European Commission under the Horizon Europe research and innovation programme under Grant Agreement N°101057721 and the UK participant (Foundation for Genomics & Population Health) in Horizon Europe Project PROPHET is supported by UKRI grant number 10040946