The challenges of biomarkers

An Ask the Expert feature with Definiens AG

The implementation of biomarkers for patient stratification to improve oncology-targeted therapies is still an extremely challenging endeavour with limited success.

“The implementation of biomarkers for patient stratification to improve diagnostics, prognostics and companion assays for oncology targeted therapies is still an extremely challenging endeavour with limited success”

Although we are now beginning to benefit from the genetic, genomic, and proteomic discoveries over the last decade, the implementation of biomarkers for patient stratification to improve diagnostics, prognostics, and companion assays for oncology targeted therapies is still an extremely challenging endeavour with limited success. This is due to many factors including the lack of implementation of true systems-based approaches in clinical development to achieve multiplexed, multi-parametric, phenotypic profiling of protein biomarkers and corresponding tissue microanatomy.

Accurate prediction of biological events, whether it be disease prognosis, toxicity, or therapeutic response, will become increasingly similar to how we predict other multivariate phenomenon; for example, the weather. 100 years ago farmers had to rely largely on qualitative, non-integrated methods such as a weathervane, cloud patterns, and an almanac for identifying weather trends. Today, when one wishes to see the weather forecast, a visit online to any one of a myriad of weather related websites provides a variety of images and quantitative information at our fingertips. Multivariate modeling integrates several factors (images, temperature, barometric pressure, wind speed, historical data, etc.) yielding accurate weather forecast information, which we are then able to make decisions with.

There is no doubt that our efforts in ‘forecasting’ biological events could benefit from a similar approach. However, at the present time, we are often looking for ‘silver bullets’ that will facilitate medical decisions, drug prioritisation, or clinical trial enrichment. In the press, we often read about a single gene or protein that is responsible for cancer. This kind of over simplification of the problem further inhibits our abilities to make strides in diagnosis, prognosis, and drug development.

Compounding the problem is the fact that the current sequential clinical development paradigm does not facilitate the development of multivariate assays that can aid in a better understanding of how multiple signal transduction pathways can affect the progression or therapeutic response of an individual’s unique cancer. This is because in order to develop these kinds of assays for a specific patient set, pathologists need access to patient tissue samples from the specific clinical trials in which they are conducting. Phase III clinical trials are where there are usually adequate patient numbers for multivariate assay development. However, by the time clinical trials are in phase III, it is too late to develop a training model, validate the model, and then run an additional prospective study to demonstrate feasibility under FDA guidance. As such, bio-pharmaceutical companies are forced into a dynamic that often only considers the target in question (i.e., Her2/neu, EGFR, etc.) or related molecules within the pathway that the therapeutic agent seeks to perturbate. The result is that even though we have examples of personalised medicine such as Herceptin, response rates are still lower than desired even among patients who have been classified as potential responders via a companion assay (i.e., immunohistochemistry). Although a particular patient may have an over active angiogenesis-related biomarker which shows up in the immunohistochemical test, there is currently little opportunity in the clinical development process to stratify patients by utilising the simultaneous measurement of biomarkers that belong to multiple pathways. These other pathways may be compensating when the original drug target is inhibited, leading to the low response rates we are faced with today.

A few solutions might be for the FDA to facilitate a clinical development process that is more interactive than sequential which could facilitate true systems-based, multivariate approaches. Additionally, further guidance for in vitro diagnostic multivariate index assays (IVDMIAs) is needed. Bio-pharmaceutical drug developers, cancer research centers, technology providers, and clinical service labs should work together more closely, synergising efforts where possible. As described in the article in ‘Imaging is Key to Success in Translational Research and Drug Development,’ image intelligence solutions that can be implemented across the entire biomedical continuum will also be needed. Further, we should accept that complex biological problems will require systems-based approaches to understand them as opposed to approaches based on an oversimplification of the underlying biology. 

Peter Duncan joined Definiens AG in December of 2008 as Global Director, Marketing and Business Development, Life Sciences. In his capacity, Duncan manages external collaborations with leading cancer centers, industry partners and bio-pharmaceutical companies. Prior to Definiens, he served as Vice President, Business Development at Aureon Laboratories and was responsible for several collaborations around applying morphometric imaging techniques to multiplexed immunofluorescence assays applied to formalin-fixed, paraffin embedded tumor tissue. Peter has over 15 years of executive sales, marketing and business development experience spanning the analytical chemistry, biotechnology, and diagnostic industries. He holds a BS degree in Biochemistry from the University of Vermont.