Breaking down silos for improved biomarker development: an interview with George Poste

Abstract

Interview with Dr George Poste by Claire Raison (Commissioning Editor)

George Poste, DVM, PhD, FRS, CBE, is the Chief Scientist at the Complex Adaptive Systems Initiative at Arizona State University (Tempe, AZ, USA) and serves as a member of the Editorial Advisory Board of Expert Review of Molecular Diagnostics. Dr Poste’s career has spanned 40 years, during which time he has been the head of research and development at SmithKline Beecham (now GSK), served as an advisor to the US Department of Defense on counter-bioterrorism and established the Biodesign Institute and the Complex Adaptive Systems Initiative at Arizona State University. Here, he discusses his work with the National Biomarker Development Alliance and his perspectives on the research and development challenges for progress in biomarkers and diagnostics.

•	What is the most promising area of your current research?

There are two. The first involves the formation of the National Biomarker Development Alliance (NBDA) [1]. This entity was established to address the multiplicity of issues responsible for the poor productivity record of biomarker research in the translation of early stage discovery into clinical trials and the enormous waste and cost when biomarkers fail in clinical trials. The objective of the NBDA, which was formed by my colleague, Dr Anna Barker (Arizona State University, Tempe, AZ, USA), previously director of the National Cancer Institute (Bethesda, MA, USA), is to build a coalition between academia and industry to address the problems responsible for the currently unacceptable situation of the poor reproducibility of too much published biomarker research and the urgent need for more robust validation methods.

Second, on a more fundamental laboratory research level, I am very excited by the diagnostic potential of exosomes. I am Vice Chairman of Caris Life Sciences (Phoenix, AZ, USA) [2], and Caris has made major R&D investments in exosomes over the past 5 years. There is a burgeoning literature on the role of exosomes as an important form of intercellular communication. Caris focuses on whether the quantity of exosomes in the circulation and their composition is altered in disease, with particular emphasis on the development of new exosome-based assays for cancer diagnosis and to monitor tumor responses to treatment. In functional terms, there is evidence that the uptake of exosomes by dendritic cells appears to be immunosuppressive, so they may contribute to the failure of the immune system to recognize cancer. Tumor cell-derived exosomes are proangiogenic and can promote the epithelial–mesenchymal transition, which is viewed as important as in the cancer invasion-metastasis cascade.

•	What are your predictions for the next 5 years in biomarker development and validation?

If I had a magic wand the first thing I’d do would be to have the biomarker field begin to realize how chaotic it has become because of a lack of standards. The promulgation of standards will be an essential dimension in long overdue reform. As diagnostic tests become more complex in transitioning from the historical unianalyte world to complex multianalyte panOmics assays, the validation of both clinical utility and regulatory complexity will increase proportionately. The debate has begun, at least in the USA, around the question of ‘who is responsible for oversight of this new generation of molecular diagnostic tests?’ Is it going to be the FDA or the CMS (Centers for Medicare and Medicaid Services)? Currently, the CMS has oversight of clinical laboratories established in an era dominated by unianalyte testing. With the rise of complex panOmics assays, there is appropriate debate about whether these tests are being launched prematurely and could potentially place patients at risk.

Equally important, if the molecular diagnostics industry is to prosper, the current myopic reimbursement policies are a major disincentive for commercial development of new panOmics assays. I continue to hold the view that I expressed many years ago that the availability of validated biomarkers is the intellectual foundation for precision medicine. However, unless we implement far more rigorous methods for biomarker validation, the evolution of precision medicine will lag and frustrate realization of its enormous promise. Without more enlightened reimbursement policies, which remain trapped in a pricing system established for unianalyte tests, this promise will not be realized.

My last point in looking at the 5-year horizon is that the current herd mentality that portrays genome sequencing as a panacea for understanding all aspects of disease pathogenesis will prove to be unrealistic. Sequencing genomes is undoubtedly important, but to believe that you can interpret disease pathology solely in the context of the genome is delusional. As with every other hype cycle created by a new technology, the balloon will burst and we will get back to a more balanced and pragmatic perspective. We will then recognize that we must adopt a systems-based approach in which the genome is interpreted as one part, albeit vital, but must be interpreted in the broader context of data on the epigenome, the universe of RNA-mediated gene regulatory networks and the largely unknown role of gene–gene interactions (epistasis) in the pathogenesis of complex late-onset multigenic diseases such as cancer, Type 2 diabetes and neurodegeneration. The role of whole-genome sequencing is currently overhyped; it’s necessary information, but it’s not sufficient.

We must also demonstrate that molecular profiling improves clinical outcomes. Currently, profiling and outcomes data are largely separated. Payors are asking for the evidence that molecular profiling does achieve better patient outcomes via improved treatment selection and clinical decisions. This is a legitimate challenge to the biomarker research community and for the commercialization of new molecular diagnostics.

•	In your recent writing, you mention the importance of cross-disciplinary integration of scientific and clinical expertise [3]. How might this be achieved?

I have a lot of ideas, but I think that my hopes could all too easily be dashed. Reductionist specialization is essential to dissect the complexities of biological systems. But at the same time, we need to avoid people ending up in fragmented silos of technical and intellectual specialization. This is now increasingly the norm. We’ve got to find ways of integrating across multiple scientific disciplines, to better link science and medicine and to recognize how the convergence of biomedicine with engineering and computing is increasingly relevant to molecular diagnostics. I appreciate that the words are easily stated, but practical translation will be more difficult.

The NBDA was designed specifically to try and break down these organizational silos and to catalyze the required holistic, transdisciplinary approach. It requires sophisticated logistics and substantial funding to assemble the needed skills if we really want to make precision medicine a reality. It requires that we approach problems in an end-to-end, system-based approach: from early concept to final demonstration of clinical utility. I hope we’ll be successful but I don’t minimize the challenges. The incentive structure for research grants in academia is still set up to largely support individual investigators rather than the multidisciplinary team-based research and clinical validation efforts now required. I consider that there is an urgent need for major reform in public funding agency policies. It’s a harsh reality, but everyone follows the money. If a funding agency said that too much current biomarker research is being wasted because efforts are uncoordinated and fragmented, that requirements for specimens of the right quality are not being met and that the requisite scale needed to launch multidisciplinarian teams is being ignored and then refused to fund proposals that propagate these deficits, they could change many of the current problems in biomarker discovery overnight.

•	What is the most important change that needs to be made in molecular diagnostics research and development?

We need clinically well-annotated specimens and more attention to be paid to the impact of preanalytical variables on biomarker expression. This lies at the core of the growing concern about the reproducibility of many published findings in the life sciences. Everyone is doing things differently. People think that they’re doing things in the same way, but they’re not. There has to be standardization. There is a need for national/international leadership to set up large-scale biobanks to create specimens of requisite quality with accompanying detailed clinical annotation. There must be clear protocols for how specimens are handled. Scale matters. The statistical validation of the clinical utility of new biomarkers panels may require analysis of several thousands of samples. The logistics of research on this scale means you can’t typically do it within a single institution. You need to build cross-institutional networks. Once again, leadership must come from the public funding agencies. The NBDA was set up as a public–private partnership to start addressing the problems highlighted here.

•	Which area of molecular diagnostics do you think deserves more funding or attention?

Rather than a lack of money, I think the more serious underlying problem is a lack of coherent organization and prioritization. This has allowed the silo-based approach to fragment the field and the lack of standards has created a chaotic situation. As emphasized, the solution lies in systematic end-to-end approaches. This is much more difficult to achieve. It requires cross-institutional collaborations and a profound shift in public funding policies to support such efforts. You need a way to fund these large collaborative networks, without which we will continue to propagate the wasteful and unproductive efforts. We need a better balance in funding policies between individual investigator-centric research versus team-based research. This is anathema to many academics because the pursuit of tenure favors isolationism and the race to publish irrespective of the quality of the work or its potential value for translation in to improve clinical care.

•	What are the big data challenges in diagnostics and how might these be overcome?

Big data marches under the colloquialism of the ‘V5’ challenge: volume, variety, velocity, veracity and value. Regarding volume, the evolution of panOmics profiling is already yielding half a terabyte of information per patient. If you extrapolate this scale to proposals for whole genome sequencing of a million individuals, your data storage and analytical costs are going to be astronomical.

The most important ‘V’ of all is value. The only meaningful definition of value is ‘does a new diagnostic test alter clinical care to improve patient outcomes?’ If patient care becomes more efficient it’s also better for society, since it’s likely that better outcomes will lower cost. This is an area of vast neglect. Everyone gets excited for quite understandable reasons thinking about the potential for precision medicine but not enough reflection is being given to thinking about the downstream infrastructure and personnel requirements needed to manage vast data streams on this scale.

Even groups like the NBDA are only just beginning to come to terms with this challenge and major public funding agencies have not addressed this at all. We’re potentially facing what I call ‘digital Malthusianism’, which will generate an increasingly stark divide between institutions that have begun to plan to manage data on this scale and those who haven’t assessed this looming challenge and will lack the resources to participate. The latter will atrophy and be rendered irrelevant.

Healthcare and the life sciences are not the first to be handling big data. There are precedents that we can import from other disciplines, such as financial services, national security or social media. Those domains are already dealing routinely with multipetabyte data streams. Surely we can import some of their learnings into life sciences and healthcare. However, until we are prepared to address the current lack of standardized formats for data and noninteroperable databases, we will continue to generate fragmented data destined to be trapped and unused in ‘data tombs’.

•	What do you consider to be your greatest achievement in your career so far?

I think it varies by period. In my academic career, it was research on the identification of clonal heterogeneity in tumors. Heterogeneity is now acknowledged as the core obstacle in effective cancer treatment. Much of the foundational research recognizing this complexity was done in the 1980s, even if no longer cited in the more contemporary literature. The second phase was as head of research and development at SmithKline Beecham, now GSK. I consider that being first of the large pharmaceutical companies to get into genomics on a major scale in the 1990s catalyzed a sea change in the industry.

Later achievements were in my role as a member of the US Defense Science Board in bioterrorism countermeasures and then moving to Arizona to establish the Biodesign Institute at Arizona State University [4]. This involved building a state-of-the-art facility and recruitment of several members of the National Academy and a Nobel Prize winner. Now I am taking on a new challenge with my colleague Ann Barker to build the NBDA as part of the Complex Adaptive Systems Initiative at ASU [5]. If we can look back 5 years from now and say the launch of NBDA was a catalyst in changing the way that research is done on biomarkers, I would view that as a compelling accomplishment.

Finally, at a personal level, my election to the Royal Society in the UK was a very special moment. It is something that, at the beginning of my career, I could never have expected.

Disclaimer

The opinions expressed in this interview are those of the interviewee and do not necessarily reflect the views of Expert Reviews Ltd.

Financial & competing interests disclosure

The interviewee is Vice Chairman of Caris Life Sciences. The interviewee has no other 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 apart from those disclosed.