Evolving Our Thinking on Biomarker Assay Validation: Are we Ready for the Next Leap?

Keywords:

• assay
• biomarker
• validation

Over the last decade, there has been an explosion of interest in the discovery and development of biomarkers to facilitate drug development and improve patient care. The potential utility of biomarkers in drug development is broad in scope, with biomarkers being leveraged right from the earliest phases to demonstrate target engagement and proof of biology, thereby reducing the risk of investment in later stages. Indeed, biomarkers can accelerate overall drug development, defining appropriate patient subsets and providing early indicators of response. It is therefore not surprising that the US FDA Critical Path Initiative identified development of new biomarkers as the highest priority for focused scientific effort [1].

A critical element to successful deployment of a biomarker is, of course, the ability to reliably measure it. Accordingly, over the past few years, the topic of analytical validation of methods to measure biomarkers has been hotly debated, with scientists and regulators alike wrestling with how best to define expected performance criteria for biomarker assays. An interesting scientific excursion occurred during this time whereby, with the best of intentions and presumed desire for scientific rigor, it was suggested that biomarker assays be held to the standards of PK assays. In retrospect, this was an intriguing development, given that the seminal paper by Lee etal., entitled ‘Fit-for-purpose method development and validation for successful biomarker measurement’ [2], preceded these discussions by several years. Regrettably, as biomarker discussions had gained steam, the concept of fit-for-purpose (FFP) began to be interpreted by some as meaning lacking in rigor, which was quite opposite to its true meaning, which is application of appropriate scientific rigor.

With the aim of beginning to build common understanding, the AAPS Crystal City VI (CCVI) Workshop: ‘Bioanalytical Method Validation for Biomarkers’, was held in Baltimore, MD, USA, on 28–29 September 2015 to discuss the critical issues. At that meeting the simple refrain ‘Biomarker assays are not PK assays’ was first uttered and repeated – again and again [3,4]. That repetition served a purpose. It solidified the concept as foundational for the discussion, disrupted some preconceived patterns of thinking and catalyzed an evolutionary leap in the conversation on development and validation of biomarker assays. Critical takeaways from that meeting included: clarification that biomarker assays should be approached scientifically in their own right, with evaluations that focus on the measurement of endogenous analyte; highlighting the limitations of spike recovery experiments and the foundational nature of parallelism assessments; and building an understanding that some biomarker applications would require analytical criteria that would be even more stringent than those applied to PK assays [4]. It was also emphasized that in order to set appropriate performance criteria for a given biomarker assay, focus must be on the specific question being asked and the decision the biomarker data must support. Furthermore, in addition to evaluation of the analytical performance of the assay, an understanding of the pertinent biology and biological variability of the biomarker is necessary to determine whether the assay is suitable to meet the needs of the question. In short, biomarker assays should be assessed in context.

Another key outcome from CCVI was that it raised the volume on the discussion and thereafter virtually every scientific forum that brought together industry professionals and regulators continued the dialogue. Subsequently, the Public Workshop on ‘Scientific and Regulatory Considerations for the Analytical Validation of Assays Used in the Qualification of Biomarkers in Biological Matrices’ was held in Washington, DC, USA, on 14–15 June 2017 where these critical concepts were revisited in depth as the content of the draft white paper by the same name (final version in preparation) was discussed and debated [5]. Through that forum an even broader audience was reached and appreciation for the not-PK nature of biomarker assays was expanded. Since then, through various forums, consensus has continued to build, with scientific alignment on several key concepts occurring by the close of 2018. Importantly, the concept of Context of Use (COU), an elegant restatement of CCVI's “focus on the specific question being asked” has entrenched itself within the biomarker community. Common understanding has evolved beyond broad acknowledgment that biomarker assays are not PK assays to include specific conceptual leaps, such as:

• Recombinant/purified standard calibrator material ≠ endogenous analyte;

• Accurate recovery of spiked recombinant/purified material ≠ endogenous analyte accuracy;

• Recombinant/purified material stability ≠ endogenous analyte stability;

• Dilutional linearity ≠ parallelism.

It seems that we have succeeded in viewing biomarker assays through their own lens and successfully shed the ill-fitting mantle of PK assays.

Before we become too self-congratulatory or complacent, we must acknowledge that our work is far from over. The question that currently seems to get the most air time is how to provide clear guidance for biomarker assay validation. Despite the desire expressed by many to have concrete rules to follow, we must accept that the breadth of questions to be addressed by biomarkers is not amenable to such an approach. Recommendations can certainly be made, but they must provide adequate plasticity to enable broad applicability. So, are we ready for another leap in the evolution of our thinking? Can we embrace a level of ambiguity that has not generally been associated with bioanalysis, a field built on accuracy and precision? I not only believe we can, but that we must. To this end, I propose three simple recommendations: (i) be a scientist; (ii) embrace and own FFP and (iii) demand COU. No doubt, the vague nature of these recommendations will disappoint many. So, what do they really mean?

‘Be a scientist’ highlights the need for biomarker scientists to accept a new level of accountability in determining the right assessments to perform to ensure that their assays meet the needs of the biomarker. To do so requires that biomarker scientists embrace and ownFFP, appreciating that FFP does not mean lower quality or less rigor, but instead means ‘do good science’ and ‘get it right’. Furthermore, the purpose in FFP is in fact, COU. The biomarker scientist must assume responsibility for being the critical judge of what level of assay performance is required to meet a given COU. Therefore, to be a successful biomarker scientist, one must demand COU. A challenge faced in many organizations, however, is a disconnect between those requesting a biomarker assay and those who are expected to deliver it, whereby the proposed COU is not made evident to the latter. An approach frequently proposed to enable a path forward in such cases has been to develop the biomarker assay to the highest possible standard, so the assay will be capable of meeting all potential COUs. However, in my experience, proposed COUs evolve with the biomarker and clinical development program and are not amenable to a priori prediction. Even if such predictions were possible, the time and cost of developing every biomarker assay to meet all future conceivable COUs, many of which would never come to fruition, is not wise investment of resources. In a world of finite resources, the consequence of overinvestment is the lost opportunity to do other, genuinely impactful work. We should never be guilty of overinvesting because we were unwilling to take on, and fight for, scientific accountability for our own work. Demand COU.

Biomarker science is not for everyone. These broad stroke recommendations are necessary because flexibility is requisite. Biomarker assays serve unique and varied COUs, which cannot be addressed by one-size-fits-all guidance. When it comes to guidance for biomarker assays, we must continue to apply the not-PK-assay mindset. PK assays share the same COU, so a single guidance may reasonably be expected to ensure that they are fit for their purpose. Such is not the case for biomarker assays.

Those who remember why they became scientists – to engage in critical thinking, solve problems, create new knowledge and change the lives of patients – will relish the challenge. Biomarker science is not black and white, but countless shades of gray. We must therefore align on the fundamental principles of scientific rigor and accountability while accepting ambiguity. There can be no specific rule book. We must think critically, generate data, share our learnings and continue robust discourse that includes industry professionals, regulators, academics and clinicians and considers the best interests of the patients who are counting on biomarker scientists to do the right assay at the right time to generate the right data to improve their lives. It is time to take the next leap.

Acknowledgments

The author would like to thank C Stebbins and D Mehta for scientific discussions and review of the manuscript.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.