National Institutes of Health Perspectives
As noted in the report (IOM, 2010), the National Institutes of Health (NIH) has played an instrumental role in advancing biomarker discovery, development, and qualification. The agency has initiated a range of efforts, including workshops and consortia, aimed at improving stakeholder collaboration and expanding public access to information on promising biomarkers. The report recognized that the “expertise, leadership, and resources of the NIH enable much rigorous science, interagency and intersector collaboration, and the public availability of biomarker data that would otherwise not occur.” Furthermore, the report noted that the NIH “may also help play a role in prioritizing the development of biomarkers in underdeveloped areas, such as food and nutrition.”
The workshop included two presentations from representatives of the NIH: Michael Lauer, director of the National Heart, Lung, and Blood Institute (NHLBI) Division of Cardiovascular Sciences, and Paul Coates, director of the NIH Office of Dietary Supplements. Dr. Lauer’s talk offered a different perspective—one cautious about the utility of biomarkers—from that expressed in prior talks by representatives of the Food and Drug Administration’s Center for Food Safety and Applied Nutrition (FDA/CFSAN) (see Chapter 3). John A. Wagner said that the committee did discuss perspectives on the general utility of biomarkers during their initial meetings. In the end, he said, “the committee did come to the conclusion that biomarkers are really an important part of medical research and medical practice, and we need to grapple with … the complex issues that
surround biomarkers, and the interfaces they have to other measurements in medicine.”
The committee posed the following questions to the two speakers from the NIH, whose presentations are summarized below:
How are biomarkers currently evaluated at the NIH?
How does the NIH interact with the FDA regarding biomarker evaluation?
Are there priority areas for research with respect to particular biomarkers?
How does the recommended biomarker evaluation process differ from current processes at the NIH?
PRESENTATION BY MICHAEL LAUER, NHLBI
Dr. Lauer began by drawing an analogy between the use of biomarkers and an event in the history of baseball described in the book, Moneyball: The Art of Winning an Unfair Game (Lewis, 2003). To assemble a successful baseball team, Billy Beane, the general manager of the Oakland A’s, was the first to apply rigorous analytic methods to identify which kinds of data best predict how much an individual player could contribute to a team’s overall success, rather than rely on conventional metrics. Dr. Lauer noted that this strategy resembles the notion “that by using the right statistics—the right biomarkers, so to speak—like on-base percentage and pitch count, one can come up with winning baseball teams without having to spend that much money.”
Turning to the report, Dr. Lauer noted that it identifies five types of biomarkers: physiological measurements, blood tests and other chemical analyses, genetic data, metabolic data, and measurements obtained from images. He provided illustrative examples—in the form of cautionary tales—representing each of these biomarker classes.
As an example of a physiological biomarker, Dr. Lauer described the exploration of ventricular premature beats, a form of cardiac arrhythmia. “In the 1980s, the presumed belief was that people who have lots of premature beats were at increased risk for [cardiac sudden] death, and if you got rid of the premature beats, you would save lives,” he said. This hypothesis was tested in the Cardiac Arrhythmia Suppression Trial (CAST) (1989). So strong was the belief that suppressing premature beats was life-saving that some considered this trial to be unethical, he said. However, the results clearly associated treatment to suppress ventricular premature beats with increased death rates.
Decades later, the NHLBI funded another trial examining the potential of an implantable defibrillator to reduce sudden cardiac death rates,
said Dr. Lauer. This three-way trial randomized patients with heart failure to receive amiodarone, which at that time was believed to be the safest antiarrhythmia drug available, or a defibrillator (Bardy et al., 2005; Torpy et al., 2007). The result was that “amiodarone did nothing,” he said. “It didn’t make things worse, but it didn’t make things better, either. But there was a reduction in death rate with the defibrillator.” That finding has largely defined care for heart failure since then, he added.
Dr. Lauer’s second case involved biomarkers measured by blood tests, an area that receives significant research funding from the NIH. He noted that in 2006, two NHLBI-funded studies were published that reached opposite conclusions. One, which originated from the Framingham Heart Study, examined a panel of biomarkers1 in approximately 3,000 people and found that those who had more abnormal biomarker levels were found to have higher rates of major cardiovascular events (Wang et al., 2006). However, these findings were no more helpful than standard risk factors in identifying people likely to experience such events. The second study was a component of the Women’s Health Study, and measured C-reactive protein (CRP) along with a variety of other biomarkers in 15,000 women (Cook et al., 2006). In this trial, CRP was found to improve risk discrimination as well as many other standard risk factors, said Dr. Lauer. However, the Women’s Health Study data analysis differed from the Framingham study because it focused on a statistical approach known as reclassification, or the ability to reassign patients to different levels of risk. Cook and colleagues argued that this measure reflected the utility of CRP as a biomarker, according to Dr. Lauer.
For a genetic biomarker example, Dr. Lauer described a recent genomewide association study of Alzheimer’s disease involving over 35,000 subjects, of whom 8,371 developed Alzheimer’s disease (Seshadri et al., 2010). He noted that the researchers confirmed two loci associated with the disease, and that they had identified and replicated two more such loci, but that compared with three well-known risk factors—age, gender, and apolipoprotein E (APOE) levels—these additional genetic markers had absolutely no impact on disease risk. “The authors concluded that, while not clinically useful, the value of these genetic markers may lie in the insights they could provide for research into the pathophysiologic mechanisms of Alzheimer’s disease,” said Dr. Lauer. He further noted that Harold Varmus, who at the time was about to become the director of the National Cancer Institute, was quoted as saying “genomics and related disciplines are more closely aligned with modern science
than with modern medicine. They produce knowledge that is broad in its scope, but only a few selected items of that new information are now widely used as guides to risk, diagnosis, or therapy” (Varmus, 2010).
As an example of metabolic data used as a biomarker, Dr. Lauer pointed out that it has been known for some time that elevated levels of hemoglobin A1c are associated with poor outcomes in patients with diabetes and more recently, in the general population. As a result, it was hypothesized that by aggressively reducing blood sugar, as reflected in reduced hemoglobin A1c levels, major cardiac events could be prevented. Dr. Lauer described the NHLBI-funded study, ACCORD (Action to Control Cardiovascular Risk in Diabetes), in which 10,251 patients were randomly assigned to intensive hemoglobin A1c reduction or to a more conservative hemoglobin A1c reduction strategy (Dluhy and McMahon, 2008). “The trialists were remarkably successful in getting the hemoglobin A1c down and keeping it down … but that did not translate into an improvement in outcome,” he said. In fact, the trial had to be stopped prematurely because patients who were randomized to intensive therapy demonstrated an increased mortality rate. “These data have been analyzed over and over and over again, and to date it is not at all clear why this happened,” he said.
The fifth and final biomarker Dr. Lauer described was an example of an imaging biomarker. In the 1990s, a series of observational studies found that patients who were clinically stable days after experiencing myocardial infarction (MI), and who went home with an open (known as “patent”) artery resulting either from spontaneous fibrinolysis or from balloon angioplasty, had better outcomes than stable MI patients who went home with occluded arteries. These results became widely accepted as the “open artery hypothesis,” said Dr. Lauer. There were some who questioned this theory, prompting the Occluded Artery Trial in which 2,166 patients were randomized to either receive a percutaneous coronary intervention (PCI)2 or not (Hochman et al., 2006). “The results were a bit disappointing and surprising, in that the patients who were randomized to receive a PCI did not have an improved outcome, and if anything, it was a bit worse,” he said. Within a year of this trial, the American College of Cardiology and American Heart Association changed their practice guidelines to indicate that PCI of a totally occluded artery after a completed infarct in an otherwise stable patient is not recommended (Antman et al., 2008).
Dr. Lauer then discussed NHLBI’s approach to biomarkers, which he said includes biomarker discovery and the use of biomarkers to
differentiate clinically relevant disease subtypes;
identify new molecular targets for intervention;
improve risk assessment, diagnosis, prognosis, and prediction of response to therapy; and
develop personalized preventive and therapeutic regimens.
One example of NHLBI’s current work in biomarkers is the Systems Approach to Biomarker Research in Cardiovascular Disease (SABRe in CVD) Initiative,3 which is a plan to combine peripheral biomarkers, genetic markers, and markers of gene expression to create sophisticated models of clinical disease using the Framingham Heart Study and other patient cohorts. Dr. Lauer noted that the NHLBI is also involved in the Clarification of Optimal Coagulation through Genetics (COAG) trial,4 which is a randomized clinical trial assessing the value of using genetic markers as a biomarker for patient response to anticoagulant therapy warfarin. Patients in the trial are being randomized to have their warfarin dosage determined by a standard clinical strategy versus a strategy informed by genomics; the primary outcome will be correct anticoagulation within a relatively short period of time, said Dr. Lauer. Another such study funded by the NHLBI, the Prospective Multicenter Imaging Study for Evaluation of Chest Pain (PROMISE) trial,5 will randomize 10,000 patients with suspected coronary disease to an anatomic test, coronary computed tomography (CT) angiography, versus functional tests, including exercise electrocardiogram on a treadmill, stress echocardiography, and stress nuclear imaging. Dr. Lauer noted that in this study, the biomarkers themselves are being rigorously tested through a randomized clinical trial.
In concluding, Dr. Lauer noted that the report, which he called a “very thoughtful document,” nonetheless contains some concerning statements. For example, the report states that biomarkers “can enable faster and more efficient clinical trials for lifesaving and health-promoting interventions.” Dr. Lauer suggested perhaps the opposite of this statement is true, and reiterated hypothetical questions previously posed about hormone replacement therapy (HRT): what might have happened if, decades ago, a definitive trial was conducted to directly examine the effects of post-menopausal HRT on clinical outcomes instead of relying on a biomarker that was eventually disproven? How many women would not have been prescribed this therapy and how many lives could have been saved if a more expeditious course focusing on clinical outcomes had been taken (Califf, 2006)?
On the other hand, he applauded the report’s statement that “we need to gather available evidence to predict effective interventions and clinical endpoints of interest,” which, he said, “points out why we cannot avoid doing clinical endpoint randomized trials.”
Dr. Lauer noted that years of research such as the studies he described, have led to a better understanding of the failings of biomarkers. However, this increase in knowledge means that it is now more difficult to make a general judgment about the value of all biomarkers.
PRESENTATION BY PAUL COATES, NIH OFFICE OF DIETARY SUPPLEMENTS
Created by an act of Congress in 1994, the NIH Office of Dietary Supplements (ODS) collaborates with other NIH institutes and centers on areas of common interest, including biomarkers, said Paul Coates. He used the example of vitamin D, a putative biomarker for various chronic conditions, to illustrate the challenges his office faces with respect to biomarker evaluation, and how they address these challenges through collaboration with a range of federal agencies.
Vitamin D status—measured by the level of 25-hydroxy vitamin D in blood—is most closely associated with rickets in children and osteoporosis in the elderly, said Dr. Coates. “Over the years, the very best data have been developed against those endpoints in those two rather specific populations,” he observed, and the possibility that vitamin D status affects other chronic conditions is “part of a very active, ongoing collaborative set of studies that are funded by the NIH, with lots of input from others.”
The ODS works with the Centers for Disease Control and Prevention (CDC) to enhance availability of data on vitamin D status in a nationally representative sample of the United States through the National Health and Nutrition Examination Survey (NHANES). Over the years that these measurements have been taken, several methodological challenges have emerged. For example, a proposed change in the method for measuring 25-hydroxy vitamin D revealed “major differences” in values measured by different methods, said Dr. Coates. To find a rational approach to deal with these variable measurements, the ODS, the National Center for Health Statistics (which runs the NHANES vitamin D study), and the National Center for Environmental Health (which measures 25-hydroxy-D levels in the NHANES samples), convened a roundtable discussion involving laboratory scientists, vitamin D researchers, statisticians, population scientists, and other experts, said Dr. Coates. The results of these discussions were published in the Journal of Nutrition (Yetley et al., 2010). Dr. Coates noted that this situation demonstrates the importance of the
analytic validation of biomarkers, and suggested that not enough attention is paid to this issue.
Dr. Coates then addressed the question of what vitamin D status actually indicates. For the purpose of argument, Dr. Coates suggested that the measured level of 25-hydroxy vitamin D in an average human derives from several sources, including sunlight, supplements, and foods. Whether vitamin D status indicates anything except level of risk for rickets in children and for falls and fractures in the elderly remains unknown. “Given that,” he said, “there are important reasons for people to want to consider taking vitamin D: the notion that their risk for type 2 diabetes, type 1 diabetes, cardiovascular disease, cancer, and a great many other chronic disease endpoints may go down,” so it is important that studies address the link between vitamin D and a variety of chronic diseases. He also noted that the Institute of Medicine is currently completing a review of dietary reference intakes for vitamin D and calcium. This report will be based on a systematic evaluation of the existing literature relating vitamin D status to chronic disease endpoints, other than falls and fractures and rickets. “In some respects this parallels the committee’s analysis of biomarkers in chronic disease management” said Dr. Coates. “We’ll see what that thoughtful approach to the available science means in terms of making recommendations for dietary intake.”
Dr. Coates noted that the ODS convened a working group on vitamin D comprising representatives from the NIH institutes, the FDA, the United States Department of Agriculture (USDA), the CDC, and corresponding agencies in the Canadian federal government. Because the ODS is a small office with limited resources, its effectiveness results from collaborating closely with other agencies. He noted that this is an approach the ODS is also taking to study a range of nutrient biomarkers with potential relationships to chronic disease, including omega-3 fatty acids, folate, and vitamin B12.