• Louise Leong

    Posted in category Opinion by Louise Leong on 07/02/2014

    Generating a NICE assessment for stratified medicines

Nowadays, diagnostics provide much more than simple assays conducted in test tubes and examining glass dishes under microscopes. They are used in large scale population screening, such as for cervical cancer, as well as for predicting whether a specific medicine or treatment will work on a patient.

 

​As the number of approved medicines linked to companion diagnostic tests increases, we need to ensure that these are appropriately evaluated using Health Technology Assessment (HTA) so that patients can benefit.

We must recognise that the objectives of HTA and stratified medicine are aligned – ultimately increasing patient access to life-saving and life-enhancing treatments. With this in mind, last week ABPI and BIVDA jointly hosted the third pharma-diagnostics stratified medicine networking event, which focused on the principles of NICE assessment of stratified medicines, issues and future trends.

The event featured thought provoking perspectives from Adrian Newland (Chair, NICE Diagnostics Advisory Committee), Greg Rossi (AstraZeneca), Berwyn Clarke (Nalia Systems) and Adrian Towse (Office of Health Economics). A few common themes emerged in the panel and audience discussion:

  • The HTA process for diagnostics may not provide the most appropriate framework for companion diagnostics. Currently, diagnostics are assessed via a different process to medicines. In cases where a medicine and a companion diagnostic are linked and launched simultaneously, a synchronised HTA process should be undertaken, by the same committee, to ensure appropriate assessment of their combined value.
  • The value of companion diagnostics with clinical utility data should be recognised and such evidence generation incentivised.  It should, however, also be recognised that clinical validation studies for tests are often much smaller than the randomised controlled trials required for medicines.  Ideally, medicines and companion diagnostics would be prospectively co-developed. However, more often than not, due to the current state of art in science and technology, biomarker information will emerge during late phase clinical studies. Hence there should be facility for proportionate retrospective analysis when biomarkers are identified post-hoc. The future will also bring a further level of complexity with more than one stratifying biomarker or multiplex platforms.
  • There is not a level playing field between commercial and in-house tests, when industry has invested upfront in clinical evaluation studies with the expectation of capturing a return on investment. Commercial laboratories are required to use a regulatory approved IVD test where one exists. However, laboratories within health institutions are permitted to develop assays in-house which are not subject to the same regulatory standards. Further, we have heard that there is no known evidence to suggest that lab-developed tests are more cost-effective than commercial ones.
  • If health systems are to reimburse stratified medicines at levels that reflect their value, flexible pricing will be required. This takes into account that evidence for its value in treatment of different groups of patients will change over time.

There are many stakeholders involved in making stratified medicine a reality. As we go forward, it is of paramount importance to ensure full engagement in these issues.

Moving forward there will be more public sector support for the generation of robust evidence on IVD tests – e.g. NIHR Diagnostic Evidence Co-operatives, the Technology Strategy Board’s Stratified Medicine Innovation Platform, the UK Genetic Testing Network activity on gene dossiers and the Royal College of Pathologists work on the National Laboratory Medicine Catalogue.

 
Dr Louise Leong
ABPI Head of Research and Development.
 
Print this page icon Print this page
 
Please choose the subject of your enquiry from the list below:
 
Code to enter for submitting the form
Enter the above code here:
Can't read? Try different words.