Precision medicine is one of the hottest topics in healthcare, but it is not a new concept. The Human Genome Project promised myriad specific and sensitive tests, therapies, even cures, and billions of dollars have been invested to achieve those goals. “Omic” technologies that detect and harness biomarkers are key to realizing precision medicine’s promise. However, for every HER2 test linked to herceptin treatment, there have been countless biomarker tests and targeted therapies that sputtered and failed or showed merely equivocal results compared with the standard-of-care.

According to a report released last spring by the Institute of Medicine (IOM), Biomarker Tests for Molecularly Targeted Therapies: Key to Unlocking Precision Medicine, this problem stems from several issues, including widespread, incorrect assumptions around the regulation of biomarkers and the evidence needed to prove clinical utility. The authors also suggested that oversight of labs performing this type of testing should be beefed up, and that some type of “rapid learning system” is needed so that biomarker-related research findings inform clinical practice expediently. In addition, communication between all stakeholders needs to be improved.

“It’s a bit like the Wild West out there at the moment,” said IOM panelist Garret FitzGerald, MD, director of the University of Pennsylvania School of Medicine Institute for Translational Medicine and Therapeutics in Philadelphia. The solution, he said, is “not just a matter of having Bill talk to Jack. It’s an issue that demands co-investment by different types of players, so the challenge is that there are many people involved in deciding what happens, and many people who will likely benefit or not benefit.”

Confronting the Learning Curve

Experts are realizing the complex interplay between what biomarkers mean, how they should be measured, and whether or not assessing them brings health benefit. This is particularly evident in direct-to-consumer genomic tests, which have shifted awareness and interpretation of biomarkers into the hands of consumers who are ill-equipped to act upon the results or vulnerable to unverified marketing claims. But confusion lies not only with the public. A 2013 survey of clinical laboratorians and physicians showed that only 35% of the tests cleared by the Food and Drug Administration (FDA) since 1993 had been adopted by respondents’ laboratories (Clin Chem 2013;59:194–7). Furthermore, half of the respondents were unfamiliar with the clinical utility of nearly half of those tests.

“We’re still in a learning curve with genomic medicine,” explained Spencer Hey, PhD, a bioethicist at Harvard’s School of Medicine who has studied the sources of inefficiency in cancer biomarker development. “We’re figuring out that it’s a lot more complicated than we thought, and we have to be much more systematic in validating these potential biomarkers.”

One case that Hey and his colleagues described recently involves excision repair cross-complement group 1 protein (ERCC1), a candidate biomarker for predicting outcomes with platinum-based chemotherapy in non-small cell lung cancer. ERCC1-related research has been almost exclusively retrospective, and it has been riddled with problems. Hey found that key aspects of the published studies, like whether the evaluating pathologists were blinded or not, have not always been disclosed.

In addition, reagents used to test for the protein were later found to be inaccurate; tumor samples within individual studies were lumped together in the data analysis, regardless of site and primary or metastatic status; and the correlation between marker status and objective response or overall survival was unreported. As a result, after 12 years of research with samples from thousands of patients in dozens of studies across the globe, the question of whether ERCC1 can be used in patient management remains unanswered.

“What’s driving the whole enterprise right now are theoretical assumptions that justify diagnostic tests,” Hey explained. “But if we’re not careful about testing our assumptions and validating them, we can end up ... misclassifying patients and not getting better outcomes in trials.”

In Hey’s mind, biomarker research needs to take an approach similar to that used with the Human Genome Project, dividing avenues of research among teams that use standardized methods that have yielded compatible results, “rather than the proprietary, idiosyncratic approaches currently in play” he said. “We need to bring order to the chaos that we’re seeing. If we can just change course a little and use precision medicine to transform how we think about research, we might not get new therapies right away, but we might get a more efficient enterprise in the long term.”

Hey’s vision aligns with that of the IOM panel. Among the committee’s 10 recommendations, some are targeted to the private sector, but most demonstrate a top-down approach, calling on governmental agencies, including the Department of Health and Human Services (HHS), FDA, and others to lead the charge. Goal 1, for example, proposes that the secretary of HHS bring stakeholders together to develop common evidentiary standards of utility, a sort of playbook that will outline how to assess biomarkers for many complex diseases, among many complex populations, before anyone can claim that a gene or protein is clinically useful.

However, even small changes in bureaucracy can take considerable time. Victoria Pratt, PhD, an IOM panelist who directs pharmacogenomics and molecular genetics laboratories at the Indiana University School of Medicine in Bloomington, said that clinical laboratories independently might be able to make certain improvements in test labeling, but it would be difficult to achieve the level of data transparency described by the IOM report. “The challenge is finding a trusted source to rate the clinical validity,” she said. “A biomarker test may be FDA approved or cleared, but in general a clinical lab would not have information about how it impacts clinical outcomes for a specific patient population.”

A Laboratorian-Proposed Checklist

Another option available to clinical labs now may be to target inefficiency at the very earliest stages of biomarker development—before considering clinical validity and health outcomes—by asking a very simple and practical question: What is the true medical need for this test or treatment?

Members of the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) took this approach in developing a 14-point checklist meant to guide conversations between all the stakeholders in biomarker development (Clin Chim Acta 2016;460:211–19).

“It’s a set of common sense questions that you’d think people would be asking before they start their research, but usually these biomarkers come out from basic research, and there’s quite a disconnect between the stakeholders in the pipeline,” said EFLM member Rita Horvath, MD, PhD, DABCC, FACB, director of the SEALS department of clinical chemistry and endocrinology at the Prince of Wales Hospital in Sydney, Australia. “We thought as a profession that we should respond because we’re positioned amid researchers, clinicians, and industry to help move the test evaluation pathway forward.”

The EFLM checklist is clustered according to four domains: 1) identify the unmet clinical need for a biomarker; 2) verify the unmet need for the biomarker; 3) validate the intended use of the biomarker; and 4) assess the feasibility of using the biomarker. The authors offer two real-life examples, a point-of-care test for chlamydia and a triage test for pre-term labor, which walk users through the process.

“This tool can be used by all stakeholders regardless of setting because it really focuses on the intended use of the new biomarker, the current practice, and where a new biomarker fits in the clinical pathway,” said Phillip Monaghan, PhD, FRCPath, co-author of the checklist and consultant clinical scientist within the Christie Pathology Partnership in the United Kingdom. “The clinical pathway is the functional unit of patient care, and mapping the pathway to identify clinical management decisions places the unmet need into context.”

EFLM pilot-tested the tool with an international group of researchers, policymakers, clinicians, laboratorians, and payers at a recent meeting in the Netherlands, with positive results, according to Horvath. The authors now plan to put the checklist online, so that they can test it with a broader audience and gather feedback, which they expect will lead to future iterations of the tool.

“It may be taken for granted that the need for a biomarker has been identified at the outset, but often it hasn’t,” said Monaghan. Despite the strong push for new biomarkers, he said, “sometimes the perceived clinical need simply isn’t there.”

Brittany Moya del Pino is a freelance writer in Kailua, Hawaii. +Email: [email protected]