The successful development of a biomarker really begins when it becomes clear that the assay for its detection and measurement is accurate and reproducible within the intended context of use—in other words, it has evidence based potential for use in diagnosis, clinical decision making, or as a clinical tool (e.g., stratifying patients for trial). Being translatable also considers how the assay will ultimately perform in the real world of varying sample quality and its capacity to fit into the clinical diagnostic workflow. The decision to translate a biomarker from early discovery should be based on adherence to standards (guidelines, best practices) at every stage of R&D, especially the early and translatable discovery phases when assays are being developed. The NBDA believes that biomarker failure often begins with the decision to advance biomarkers that were not developed using robust standards and an end-to-end plan.
Each year scientists report the “discovery” of over 150,000 biomarkers. Unfortunately, these well-intentioned studies too often reflect early results from experiments that lacked a clear clinical question and context of use, employed poor quality samples, were statistically underpowered, and/or did not utilize robust data standards and analytics. In addition, nearly all classes of biomarkers utilize different technology platforms with highly specific performance characteristics. This latter issue is further complicated by changes in emerging technologies, such as whole genome sequencing, which introduce new variables that must be accounted for in the biomarker discovery process. In the aggregate, this myriad of factors produces a large number of biomarker “discoveries” that are often difficult or impossible to reproduce.
The NBDA process for moving a biomarker from early to translatable discovery requires first and foremost that best practices, guidelines, and standards be followed to reproduce the original finding. For a biomarker to ultimately be clinically useful, it should meet FDA standards as “fit-for-purpose” and also be scalable. In addition, if the biomarker is intended for use as a diagnostic on a large scale, it must make economic sense. Biomarkers that inform biology and are potentially useful as clinical tools versus companion or screening diagnostics should also meet criteria for standards use and reproducibility.
To be translatable, a biomarker assay should be supported by data that is transparent and “liquid,” i.e., capable of easily being moved among technology platforms. Data formats should be standardized to support rigorous archiving and analysis and demonstrate transparency relative to origin. In addition, appropriate metadata should accompany sample data, as part of an integrated effort to transform data into knowledge capable of ultimately producing a clinically useful (and potentially manufacturable) biomarker.
In summary, a translatable biomarker reproducibly measures a change in biology that addresses a specific clinical question. A decision to develop a biomarker beyond early discovery is expensive, so just as with drug candidates, the sooner it can be demonstrated that a biomarker is not sufficiently robust in the many areas needed for successful development the better. The NBDA’s intent is to assemble and/or develop the evidence needed to make these critical decisions at each stage of an end-to-end process.