5
Industry Perspectives

Speakers representing companies and professional organizations with an interest in biomarkers gave 5-minute presentations and participated in a panel discussion. These individuals represented a range of industries, including food, grocery manufacturers, nutritional supplements, pharmaceuticals, medical diagnostics, and devices.

The committee asked these speakers to address the following questions:

  • How do biomarkers impact patients, consumers, or clients in your industry stakeholder group?

  • How does the recommended biomarker evaluation framework relate to biomarker evaluation processes currently in use with your industry or stakeholder?

  • Will a prospective biomarker evaluation process add clarity to product development and consumer understanding?

  • To the degree it is possible to respond, if at all, what are your key concerns for implementation of the recommendations?

  • To the degree it is possible to respond, if at all, which recommendations of the report are the most useful or important?

Several themes emerged in the course of these presentations that were further explored during subsequent discussion. First, echoing its emphasis in the report, was the importance of context of biomarker use. As Allan Jaffe, professor of medicine at the Mayo Clinic, noted, these exchanges recapped discussions that occurred among committee members as they



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 51
5 Industry Perspectives Speakers representing companies and professional organizations with an interest in biomarkers gave 5-minute presentations and participated in a panel discussion. These individuals represented a range of industries, including food, grocery manufacturers, nutritional supplements, pharma- ceuticals, medical diagnostics, and devices. The committee asked these speakers to address the following questions: • How do biomarkers impact patients, consumers, or clients in your industry stakeholder group? • How does the recommended biomarker evaluation framework relate to biomarker evaluation processes currently in use with your industry or stakeholder? • Will a prospective biomarker evaluation process add clarity to product development and consumer understanding? • To the degree it is possible to respond, if at all, what are your key concerns for implementation of the recommendations? • To the degree it is possible to respond, if at all, which recommenda- tions of the report are the most useful or important? Several themes emerged in the course of these presentations that were further explored during subsequent discussion. First, echoing its empha - sis in the report, was the importance of context of biomarker use. As Allan Jaffe, professor of medicine at the Mayo Clinic, noted, these exchanges recapped discussions that occurred among committee members as they 1

OCR for page 51
2 BIOMARKER AND SURROGATE ENDPOINT EVALUATION determined whether to separate utilization from analytical validation and qualification in the recommended framework for biomarker evaluation. On a related point, some discussants raised the concern that case studies in the report might be interpreted as questioning the value of use- ful biomarkers that are not appropriate surrogate endpoints (for example, C-reactive protein [CRP]). Guy Johnson remarked that the predictability and shortcomings of biomarkers such as prostate-specific antigen (PSA), CRP, and low-density lipoprotein cholesterol (LDL-C) are well known by medical professionals, and that this information should also be available to consumers. However, he said, “in a quick read of the report, I’m wor- ried that its results could be misconstrued to undermine the value of some of these markers that are being used routinely.” Roberta Ness replied that the committee had shared this concern, and they included information and recommendations in the report emphasizing the importance of scientific literacy, numeracy, and com - munication with patients regarding biomarkers. “PSA is such a perfect example, because there is so much controversy about its use,” she said, noting that physicians do not consistently interpret the results of this assay. Maria Lopes-Virella, professor of bioengineering at the Medical University of South Carolina, added that while providing information about biomarkers is important, the public needs to be made aware that the same biomarker can be valid for use in one context and not in another. Dr. Jaffe noted that health care providers also need guidance on these issues. “Lots of physicians say, ‘That’s been in trial, [and] X showed better than Y; therefore, everybody gets X,’” he said. “That sort of thinking is a negative for the use of biomarkers, because it means when you use a bio- marker, you [reduce it] to its lowest common denominator and overutilize it.” Thus, it’s important to encourage more probabilistic thinking on the part of physicians as well as the public to combat the misguided notion that a biomarker has to be good for every use or no good at all, said Dr. Jaffe. Committee member Michelle A. Albert, assistant professor of medi- cine, Harvard Medical School, further noted that the phenomenon of “psychological toxicity” figured into the committee’s deliberations. In the case of very ill patients, she said, it is important to consider what effect it may have to “use the shrinkage of a tumor … to tell someone that they are improving when in fact they are not, and … what that does to them psychologically as an important endpoint.” That is an additional reason why the committee emphasized and separated context of use from bio- marker validation and qualification, she stated. Another concern raised by several discussants was the possibility that the standards established by the recommended biomarker evaluation framework will have a chilling effect on research in this area.

OCR for page 51
 INDUSTRY PERSPECTIVES “I think it doesn’t have a chilling effect at all,” Jennifer Van Eyk responded. “I think it’s very realistic, and I think people who are doing biomarker development—I can only speak from an academic [stand- point]—this is our reality. This is hard stuff to do. These are complicated diseases, and even when they are straightforward, the variation between patients, between individuals, between … population[s], is massive.” However, researchers believe that it will eventually be possible to identify biomarkers as surrogate endpoints for chronic diseases. “We all realize it’s a multistep pathway … [that] is well laid out [in the report].” “I think we simply formalized what had been in the minds of many and maybe even in the operation of many,” said Dr. Ness. The committee’s recommended framework for all biomarkers reflects their concern regard- ing biomarker usage that puts large numbers of people at risk, especially when the risk of harm outweighs the potential benefit, she added. Responding to this comment, Stephen Williams noted that biomarker qualification has typically been characterized by two competing philoso - phies. “One is to take the harm minimization approach, and to carefully characterize all the harm and to try and avoid it going forward,” he said. The other philosophy is “the tolerability of risk approach, which is to carefully characterize the harm and to carefully characterize the benefit of going forward with a surrogate versus going forward with the best available alternative and to choose the better option.” Both positions are expressed in different parts of the report, he observed, so he asked the committee, “is there an explicit embrace of tolerability of risk, or is there an explicit embrace of harm minimization? Or are people still ambiguous?” John R. Ball answered that the committee tried to “point out the real benefits of biomarkers in a lot of cases” but also to express caution learned from experience. This was also their intention in recommending that the FDA assemble expert panels to evaluate biomarkers. Such panels would include people with explicit conflicts of interest because their expertise is needed. “That would be the sort of failsafe mechanism that would take into account the balancing of the benefits and the potential harms,” he said. David DeMets noted the contention—that few biomarkers could serve as surrogate endpoints—was stated more than a decade earlier (Fleming and DeMets, 1996). Not surprisingly, he observed, “most biomarkers put forth as surrogates have failed. By reflecting that reality, the report shouldn’t chill biomarker research, he said, but focus efforts on more appropriate uses of biomarkers, such as the early screening of candidate drugs and the characterization of patient populations.” What really has a chilling effect, Dr. DeMets continued, is the inap - propriate use of a biomarker. For example, in the early days of the HIV/ AIDS epidemic, CD4 counts were used as a surrogate endpoint in drug

OCR for page 51
 BIOMARKER AND SURROGATE ENDPOINT EVALUATION research. “Over time [the scientific community] realized they weren’t such good surrogates,” he said, but there were no alternative biomarkers for this purpose. Members of the HIV/AIDS patient advocacy community also questioned the use of CD4 counts as surrogate endpoints. “They demanded that we shape up and demand the best data, the best evi- dence,” he said. “This is a population that had the most to lose, and yet they were the ones who understood what the risks were, because the [experimental AIDS] drugs were expensive, they were toxic, et cetera.” The distinction between foods and drugs, and in the ways each indus- try uses biomarkers, generated considerable discussion. Responding to Douglas Balentine, who remarked in his presentation (see below) that drugs cure and mitigate disease, while healthy foods reduce the risk of developing disease, Thomas Fleming noted that some drugs also serve the latter purpose; for example, interventions to prevent HIV transmis- sion, type 2 diabetes, and Alzheimer’s disease. Such drugs are rigorously scrutinized to ensure their efficacy, he continued, and if such claims are based on surrogate endpoints, those biomarkers must be validated. Since this is the case, he asked, “if foods want to have a health claim of reduc- ing the incidence of cardiovascular disease or cancer, why shouldn’t they undergo the same scrutiny as drugs would for making such claims?” Dr. Balentine answered that conducting such prospective cohort studies for foods would be difficult because people who develop health problems are treated with drugs not food. While he agreed that rigorous science is necessary to support health claims for foods, he contended that a different approach is required as compared with drugs. Dr. Krul noted that biomarkers currently used to make health claims for foods are relatively accurate and precise (for example, the 4 percent reduction of LDL-C claimed by Cheerios). The FDA wouldn’t approve a new drug that lowered cholesterol by such a modest amount, she observed, “but is that good enough for a health claim?” And if such a claim is clinically relevant and statistically significant, she continued, does that make a food a drug? Dr. Jaffe replied that the biological importance of a biomarker, such as LDL-C, should be considered during analytical validation and qualifica - tion. For example, he said, one could say that intravenous nitroglycerin following a mild acute infarction doubles the blood flow within the area of the infarction, but by increasing from 0.5 percent to 1.0 percent of nor- mal flow, that gain is biologically insignificant. Proper analytical valida - tion and qualification of biomarkers takes such information into account prior to considerations of context of use, he said. Dr. Jaffe also noted that presenters from the food and nutritional supplement industries had expressed concerns as to how the recom- mended framework for biomarker evaluation would be implemented,

OCR for page 51
 INDUSTRY PERSPECTIVES and in particular, how their industries could afford to conduct research on health claims. “I wish I knew all the answers about how to operation - alize all of this in terms of … surveillance or appropriate studies,” he said. However, he added, “one of the nice things about combining drugs, devices, nutraceuticals, and food under a similar umbrella is there is a huge amount of experience … at the industry level that can be tapped to help develop the appropriate paradigms.” One such paradigm, Dr. Jaffe added, is likely to involve data sharing and collaboration on biomarker identification, as is already occurring among pharmaceutical and medical device companies. PRESENTATION By DOugLAS BALENTINE, uNILEVER Dr. Balentine said his remarks would try to illustrate the challenges and benefits of using biomarkers to support health claims in the food industry. “The industry needs clear guidance as our goal is to make truthful and not misleading claims and messages on products,” he said. “As we develop products to help consumers choose healthier diets, clear science-based guidelines are essential.” The food industry also needs to have a clear understanding of which biomarkers or surrogate endpoints the FDA will accept, based on rigorous scientific research, in support of health claims. Foods and drugs are viewed differently, Dr. Balentine emphasized: foods can promote and maintain overall health and wellness while drugs can cure or mitigate disease. Both purposes demand scientific rigor, he said, but the type of studies required to demonstrate each type of claim would be different, as befits the context of use for a drug versus a food. However, studies of biomarkers that the National Institutes of Health (NIH) or pharmaceutical industry conduct can benefit the food industry, which can apply their results to help determine health claims for foods. Drugs are often a specific compound that is used to produce specific effects, whereas foods consist of many compounds which may produce many different effects. The complexity of food makes the health effects asso- ciated with dietary interventions difficult to study, said Dr. Balentine. This problem is further complicated by the fact that people in control groups for food studies are often exposed to the components being tested (for example, vitamin C, carotenoids) through other aspects of their daily diet. Thus, he said, there cannot be a true placebo-controlled trial of most foods. The food industry uses biomarkers to support health claims, which involve reduced risk for a chronic disease endpoint, and also to support structure–function claims, which involve maintainance of a healthy pro- cess. In either case, Dr. Balentine emphasized that “biomarkers need to be

OCR for page 51
 BIOMARKER AND SURROGATE ENDPOINT EVALUATION strong and they need to be clear, and they need to be scientifically sup - ported.” The biomarker evaluation framework should help to establish that clarity, he added, provided foods, medical devices, and supplements are each evaluated in context and with appropriate scientific rigor. PRESENTATION By MELISSA MuSIkER, gROCERy MANuFACTuRERS ASSOCIATION The Grocery Manufacturers Association (GMA) represents the world’s leading food, beverage, and consumer products companies, by promot- ing sound policy, championing initiatives that increase productivity and growth, and helping to ensure the safety and security of consumer pack - aged goods through scientific excellence, said Melissa Musiker, senior manager of science policy, nutrition, and health, at the Grocery Manufac - turers Association. She noted that the report’s publication had sparked discussion in the popular press as to the validity and appropriateness of all types of voluntary nutrition claims used on food product labels. The GMA was concerned with this interpretation of the findings of the report because biomarkers are used in a different context when applied to food and vol - untary nutrition label claims as compared with drugs. “Under the current regulatory framework, a variety of voluntary claims are permitted for use in food labeling, provided that they are truthful and not misleading,” she explained. “There are also three other types of voluntary claims related to nutri - tion and health that are permitted for use on food labels but without prior approval,” Ms. Musiker continued. “The first is a dietary guidance statement, which typically refers to the benefits of the broad class of foods and reduced risk of disease or health conditions. The second is a structure–function claim … that encompasses both expressed and implied claims regarding the benefits of food or food components in promoting and maintaining normal structures and functions of the body. The third is a nutrient content claim, a descriptive statement that characterizes the level of nutrients in the food.” All three types of claims imply that consuming the food bearing them leads to the maintenance of health and promotion of normal physiology, Ms. Musiker said. “Any food or component of a food, characterized by a voluntary label claim, should be of importance in human nutrition by virtue of its presence or absence at the levels the claim describes.” In fact, she added, “all information provided as a mandatory nutrition-labeling element within the nutrition facts panel is assumed to meet the same standard.” Ultimately, she concluded, “all nutrients on the label should

OCR for page 51
 INDUSTRY PERSPECTIVES be relevant to the promotion of good health and by extension have an impact on a biomarker.” “The GMA agrees with the committee that a consistent level of scien- tific rigor should be utilized when evaluating biomarkers and surrogate endpoints in their ability to serve as predictive models from morbidity and mortality outcomes,” said Ms. Musiker. However, since health claims are preventive in nature, the type of study needed to assess such a claim should differ from that used to support a disease treatment, she argued. “The scientific research needed to assess the validity of a biomarker for the purpose of voluntary nutrition claims presents unique challenges for foods,” she said. “A large-scale, double-blind, placebo-controlled, crossover, randomized clinical trial would be very difficult to design and successfully implement when studying foods and diets,” she noted. Echoing a point made by Dr. Balentine, she added, “when examining nutrients within a food as opposed to a nutrient consumed in isolation, it is very challenging to isolate the impact of just one nutrient or food, and then to generalize this finding across populations … because all people eat food as a part of a daily diet and a dietary pattern, which can be so highly variable.” All foods are composed of a number of components (for example, fats, carbohydrates, fiber, protein, water, micronutrients, and phytochemicals) known collectively as the food matrix, Ms. Musiker said. These complex and highly variable food matrices can modify nutrient bioavailability, such as when a nutrient is consumed in combination with other nutrients. Due to these modifying effects of food matrices and overall dietary pat - terns, the committee’s recommendations, as she interpreted them, would require a unique study to determine the validity of a biomarker relative to a nutrition claim for any given food or combination of foods. “It’s really not practical for novel research to be required each time a company would like to put a nutrition claim on a food,” she said. “There is a well-established body of science that provides the fundamental basis for the vast majority of voluntary nutrition claims, linking nutrients and foods in the diet to the maintenance of normal physiology and good health.” She noted that most of this science is articulated in the Dietary Guidelines for Americans or the Dietary Reference Intake report (HHS, 2005; IOM, 2006). “The GMA proposes that a system be developed to provide guidance to food companies when developing and applying voluntary nutrition claims,” Ms. Musiker said. This system should be designed with the com - mittee’s recommendations in mind, but should also take into account the challenges presented by studying foods or nutrients consumed as com - ponents of a meal, within a diet, and as part of a person’s lifestyle. “The GMA believes the types of scientific studies needed … for the evaluation

OCR for page 51
 BIOMARKER AND SURROGATE ENDPOINT EVALUATION of the appropriateness of dietary guidance statements, structure–function claims, and even nutrient content claims, may be inherently different from those needed to develop health or drug claims,” she added. Therefore, the GMA “encourages those evaluating nutrition label claims to remember that there is a distinction between health claims related to the prevention and mitigation of abnormal biomarker status and other types of voluntary claims related to the maintenance of normal status.” PRESENTATION By ANDREW SHAO, COuNCIL FOR RESPONSIBLE NuTRITION “The lack of validated biomarkers for exposure to nutrition interven- tions and surrogate endpoints for chronic disease limits the amount of research that can be conducted, especially for prospective randomized trials, due to cost and other logistical issues,” Andrew Shao said. “This, in turn, limits the ability to derive answers to important questions relating to the ability of diet, food, and food components to promote health and reduce the risk of chronic disease.” The Council for Responsible Nutrition (CRN) has commented on the need for additional biomarkers as surrogate endpoints to both the FDA and the NIH, he reported. Having a framework for biomarker evaluation is a step in the right direction, Dr. Shao said. “We anticipate that a formal biomarker evaluation process will add clarity to product development, [and that] as companies that choose to invest in research [we] will have a better understanding a priori that the research will have broader acceptability and applicabil- ity to public health recommendations, such as health claims,” he said. Regarding consumer understanding of biomarkers and their evaluation, he asserted, “in the end, consumers are not so much interested in the bio - marker as they are in the clinical endpoint that it represents.” The CRN’s primary concern regarding the committee’s recommen - dations is that human or financial resources may prove inadequate to implement them, Dr. Shao said. He also expressed concern that some recommendations might be misinterpreted; for example, he interprets Recommendation 3 (see Box 2-1) to mean “when it comes to relying on a biomarker as a surrogate for a clinical endpoint, the product application, whether a food, drug, or device, is irrelevant, and that there should be a single standard,” he said. He added that he agreed with this position, “whether we’re talking about a statin or we’re talking about dietary fiber.” However, he asserted, “text in the body of the report associated with the recommendation is not consistent with that interpretation.” For example, he quoted from the report, “the FDA’s regulation claims and

OCR for page 51
 INDUSTRY PERSPECTIVES the scientific standards for evaluating such claims are governed by dif - ferent regulatory frameworks as compared to drugs. Legislation may be required to revise the science-based standards and regulatory processes for these nondrug products.” That passage, he explained, caused him to ask whether the committee was suggesting that foods and supplements be regulated like drugs. Calling Recommendations 1 and 2 (see Box 2-1) “the most useful and relevant recommendations,” Dr. Shao said that they “represent what we believe the IOM committee was charged with accomplishing and set a solid foundation for a scientific framework that can be applied to the literature and future research efforts.” On the other hand, he described Recommendations 3 through 5 as “both unnecessary and confusing.” “I certainly won’t speak for the FDA, but it seems to me that recom - mendations dealing with food policy and regulation, including that of dietary supplements, seem to go beyond what the committee’s mandate was,” Dr. Shao said. “For example, Recommendation 4 really seems to relate to general issues regarding health claims … but it says nothing about biomarkers.” By contrast, “We feel Ancillary Recommendation 6a [which encour- ages the collection and sharing of biomarker data across the Department of Health and Human Services] is very important for the implementation of the first two recommendations.” A number of different federal agencies conduct research on and evaluate biomarkers, Dr. Shao pointed out. “To fully leverage all of these resources, these efforts cannot be siloed, but instead should be consistently shared across the various agencies.” PRESENTATION By STEPHEN WILLIAMS, SOMALOgIC, INC. “Pharma is a very diverse range of interests,” Dr. Williams pointed out, and while he did not seek to represent them all, he expected that his opinion, which reflects his many years of involvement in the industry, was “reasonably representative of mainstream opinion.” Dr. Williams said that in regard to biomarkers, the pharmaceutical industry wants the same thing as the FDA, as reflected in their charge to the committee: “to establish a framework for the level of evidence and the nature of evidence for biomarker utilization.” However, such a frame- work should also enable drug companies and their regulators to reach a decision as to what level of evidence is needed for a particular biomarker application. He added that the pharmaceutical industry wants “consis - tency and transparency in biomarker evaluation, rather than something … unpredictable and opaque,” as reflected in the report. By focusing on two questions, “how good is good enough?” and “what type of evidence?,” he briefly reviewed how well he thought the report and its recommenda -

OCR for page 51
0 BIOMARKER AND SURROGATE ENDPOINT EVALUATION tions had answered these questions, and in what ways he felt they had fallen short. The committee’s recommendation that the FDA establish expert com - mittees for biomarker evaluation across its regulatory areas “would move forward the ability to decide how good was good enough,” Dr. Williams said. He similarly applauded the three-part framework, its consistent application across all biomarker uses, the continual reevaluation of bio - markers, and the comprehensive evaluation of risks associated with the use of biomarkers as surrogate endpoints. An especially important feature of the report is its advocacy for biomarker consortia, he added. Pharma- ceutical companies had realized for some time that more evidence might be needed to evaluate a biomarker or a surrogate endpoint than any one firm could generate on its own, and that even if it were feasible, doing so would not often be cost-effective. Turning to his concerns about the report and its recommendations, Dr. Williams reiterated a concern he had expressed earlier in the workshop, the notion that one must define a specific intended use for a biomarker in order to determine both the benefits of success and the consequences of error in its usage. “I don’t think a committee’s going to be sitting there wondering about what level of evidence there is for a biomarker when they don’t have a purpose in mind [for it],” he said. Context “comes first because without that you cannot define an acceptable performance stan - dard for analytic validation or qualification,” he continued. Dr. Williams also argued that the case studies in the report did not actually assess the recommended evaluation framework. For example, the report said that “tumor shrinkage is not acceptable as a surrogate end - point … partly because the analytic validation is not very good.” How- ever, within individual clinical trials, Dr. Williams noted that tumor size can be gauged accurately, “so the analytic validation can be controlled.” In addition, he suggested that tumor shrinkage could have been used as an example to demonstrate evidence thresholds. “I would accept completely that if I’m developing a drug … [that is] 14th in class, and the cancer [it treats] is already well served by lots of other drugs, then I’m not going to ask you to tolerate the risks of using tumor shrinkage,” he said. “But if I’m developing a drug for a cancer where there is absolutely no therapy today, it is completely fatal, the drug is known to be safe, and if I delay that drug on the market by another year to … [use patient] survival [as the endpoint] … another 10,000 people will die … [then] the value of using tumor shrinkage in that case is 10,000 lives. The consequence of error, to say I’m wrong and the surrogate endpoint didn’t work, is zero. It is the cost of the drug company expense on the drug, but actually, if those people were going to die anyway, letting the drug through until you find

OCR for page 51
1 INDUSTRY PERSPECTIVES out that it didn’t translate into a benefit has no consequence at all,” he asserted. “So in that case, I might choose to accept tumor shrinkage.” Regarding these points, Dr. Fleming later expressed concern that Dr. Williams was “advocating for a lower bar that could be used to justify [using tumor shrinkage] as a surrogate.” Dr. Fleming noted the potential for false negative risks in such cases as renal cell cancer, which fit the scenario Dr. Williams described. There were no effective therapies for that cancer when sorafenib came along several years ago, said Dr. Fleming. If tumor shrinkage had been the measure by which that drug was judged, its benefit would have been missed because it doesn’t shrink tumors; neither do some other drugs that have had major effects in colorectal and lung cancer, he added. He also noted the potential for false positive effects, which might arise when side effects significantly increase risk relative to benefit. Moreover, he added, the law states that approval of a drug requires evidence of efficacy, so standards for approval are not simply at the discretion of the FDA. There are additional downsides to using an imperfect surrogate endpoint in such a situation, Dr. Fleming continued. A patient taking a therapy that doesn’t work but was erroneously approved on the basis of a surrogate endpoint might otherwise have joined clinical trials for altruistic reasons, as well as for the potential benefit of the experimental therapy, he said. Finally, he argued, limited health care resources should be focused on therapies likely to provide more benefit than risk. “So,” Dr. Fleming asked, “how can we defend a low bar for a surrogate that is very unlikely to reliably predict benefits simply because there are no other effective therapies?” Dr. Williams agreed that there should be a “level playing field for benefit and risk.” He said that he “wasn’t advocating a blind acceptance of a surrogate just because there were no other therapies” but rather that its evaluation include an assessment of the consequences of all possible errors associated with its use, as compared to all possible associated bene - fits. Such an analysis might have excluded tumor shrinkage as a surrogate endpoint, he said, but the case study in the report should have reflected that process in its entirety. “What I was looking for was an example … that demonstrated to the audience how one might change the balance of the weight of evidence depending on the … harm of failure or value of success,” he explained. Dr. Fleming responded, “I believe the reason you didn’t see that is when we have 35 years of experience using this measure and now have extensive data regarding the relationship between the effects on tumor shrinkage and the effects on clinical endpoints, we’re finding that we’re not seeing a reliable assessment of efficacy using tumor shrinkage.”

OCR for page 51
2 BIOMARKER AND SURROGATE ENDPOINT EVALUATION But many readers of the report don’t have the benefit of these years of experience, Dr. Williams pointed out, so they might have benefited from learning how the threshold for use was defined for tumor shrinkage as a surrogate endpoint, and how the biomarker doesn’t meet this threshold. For example, the report might have stated how many instances when tri - als using tumor shrinkage as a surrogate endpoint had failed to translate into improvements in survival postmarket. Then, he added, the report could have explained why this statistic is not tolerable or under what circumstances it might be acceptable. “I was looking for some kind of con- nection between the failure rate of the surrogate and the value of using it versus the harm of not using it,” Dr. Williams said. Dr. Fleming maintained that it isn’t appropriate to lower the bar for acceptability of surrogate endpoints when nothing else works, because “something unreliable could lead us to missing effective therapies or declaring that something should be used when it is actually unfavorable in benefit-to-risk.” In a similar vein, Dr. Williams said he felt the report underplayed the value of surrogate endpoints, particularly as compared with their risks. “The report really nicely goes through all the different kinds of risks that one might come across when using a surrogate endpoint,” he said. However, if you don’t recognize the outcome of a short trial in terms of lives saved, you wouldn’t tolerate the risks inherent in using surrogate endpoints. Focusing on risk and downplaying benefit leads to “a kind of precautionary view of surrogate endpoints,” much as Michael Lauer had expressed, Dr. Williams noted. “I’m surprised that someone from cardio- vascular health can say that they are concerned about whether surrogates are useful, [given] the … millions of lives saved by using blood pressure or LDL [as biomarkers],” he said. Another benefit of surrogate endpoints is the advantage they offer to researchers seeking new therapies, Dr. Williams said. He noted that a recent review found that diseases for which a surrogate endpoint or efficient clinical endpoint1 is available have an average of 100 therapies on the market, as compared with an average of less than one therapy available for diseases that lack a surrogate endpoint or efficient clinical endpoint (Lathia et al., 2009). That’s another reason to take a balanced view of tolerable risk in using surrogate endpoints, he suggested. Dr. Williams was also concerned that the evaluation framework did not include an analysis of the cost-effectiveness of surrogate endpoints. He argued that cost-effectiveness is a standard of the British National 1 An efficient clinical endpoint is defined in Lathia et al. (2009) as those that enable proof of concept studies in 6 weeks or less and pivotal trials in 6 months of dosing or less.

OCR for page 51
 INDUSTRY PERSPECTIVES Health Service, the largest single-payer health care organization in the world, and is also a factor in decision making for many federal agencies, including the Environmental Protection Agency and the Federal Aviation Administration. Finally, Dr. Williams noted that pharmaceutical companies have been involved in the testing and validation of biomarkers for at least a decade, and that some people within the industry have run dozens or even hun - dreds of biomarker validation trials. “If there is to be a committee in the future that makes recommendations on whether … [a biomarker or a sur- rogate endpoint] is acceptable or not, the conventional wisdom would say, don’t include industry experts, because they are conflicted,” Dr. Williams said. He would prefer to see people from industry contribute their consid - erable expertise to biomarker evaluation committees. Responding to this point in discussion, Dr. Ball noted that the committee recommended that such expert panels comprise people representing a range of perspectives, because those are the people with the greatest expertise on the topic; he also noted that the FDA typically includes experts with these types of conflicts of interest on such panels. PRESENTATION By JAMES MAyNE, PFIzER, INC. At the outset of his remarks, James Mayne, senior director at Pfizer congratulated the committee on its work: “This is clearly a very timely, very powerful, and very scholarly work that sets a framework and a road- map for the future of biomarker development and qualification.” Speak - ing on behalf of Pfizer and himself, Dr. Mayne noted, “we are very, very appreciative that the committee didn’t step back from the ambitious scope that it undertook at the outset of this work.” He believed the committee has created a unified decision-making system that encompasses all types of biomarkers, products, and decisions regarding their usage. Recalling a presentation he delivered to the committee during its deliberations, Dr. Mayne noted that he had advocated a two-step frame - work consisting of analytical validation for biomarkers, followed by “fit- for-purpose” qualification. When he first reviewed the report, he said, he questioned why the qualification and utilization were not integrated. However, “as I read through the report and saw the rationale, it does make a tremendous amount of sense; … [the framework] can be used across many different product types and many different market types.” The report clearly defines what the framework is not intended to shape, Dr. Mayne emphasized. “It is not intended for biomarkers used in the dis- covery space,” he said. “It is not intended for the ‘tools of the trade,’ as we sometimes call them in the drug industry, by which we make decisions on which chemical moieties to advance, which diseases to pursue, and where

OCR for page 51
 BIOMARKER AND SURROGATE ENDPOINT EVALUATION the best opportunities may lie. Very wisely … the committee made it clear that those areas are omitted so as to not constrain innovation.” At the conclusion of his remarks, Dr. Mayne offered what he called “a criticism and a challenge.” He said that the report failed to specify the “actual elements of the decision framework,” or the criteria by which regulatory decisions should be made. “That was not provided, at least not in the detail I was looking for,” he said. “That’s fine; you have to start somewhere.” PRESENTATION By JACk zAkOWSkI, BECkMAN COuLTER, INC. Although he noted that his comments did not necessarily represent the position of his company, nor of the diagnostic industry in general, Jack Zakowski, director of scientific affairs and professional relations at Beckman Coulter, Inc., began his remarks by voicing general agreement with the report’s evaluation framework and in particular, the interde - pendence of its steps, as depicted in Figure 2-1. However, he thought the arrows in the figure should be double-headed, indicating bidirectionality, to better reflect that interdependency. Dr. Zakowski then sought to answer the session questions, as stated above. He said in vitro devices (IVD) that assay biomarkers aid in the overall assessment of patient status by diagnosing and monitoring chronic disease, guiding therapy, and predicting outcomes. The committee’s framework resembles current practices of biomarker evaluation within his industry, but he emphasized that IVD assays measure biomarker concentrations, and are not biomarkers in and of themselves. “I think too often we have not drawn that distinction,” he said. In considering whether a prospective biomarker evaluation process would add clarity to product development or consumer understanding with respect to IVD biomarker assays, Dr. Zakowski noted that such clar- ity is achieved through the following routes, most of which were also noted by Dr. Williams: • Clear definitions of terminology, • Common conceptual framework (as shown in Figure 2-1), • Clear common goals, • A priori assessment criteria and analysis tools, and • Transparency and predictability of process. Regarding terminology, he asked, “when we use words as supposedly simple as sensitivity and specificity, are we talking about clinical or diag- nostic sensitivity and specificity, ROC [receiver operating characteristic] curves, or what? Or do we mean analytical sensitivity, which could be a

OCR for page 51
 INDUSTRY PERSPECTIVES lower limit of detection? Or freedom from interferences?” Such clarity is needed in the report, he observed. Dr. Zakowski listed three key concerns for implementing the com- mittee’s recommendations: the potential conflation of assays with bio - markers (as previously stated); the need to emphasize that assays are both independent of and interdependent on the biomarker’s utility; and the tendency toward “on-off” clinical thinking that ignores the fact that biomarkers measure the likelihood of disease and not disease itself. He described several examples of tests that have been improved by the selec - tion of a better biomarker, such as the transition from glycated hemoglo - bin to hemoglobin A1c as an indicator for diabetes risk. He also noted that some assays have evolved to better measure biomarkers, as occurred with the advent of ultrasensitive thyroid-stimulating hormone (TSH) tests. Concerns that biomarker assays encourage “all or nothing” thinking are another way of saying “context matters,” Dr. Zakowski observed. The upper limit of normal “cholesterol … is 200,” he said. “If your cholesterol is 199, you are not going to get prescribed a statin. If your cholesterol is 201, you are going to get prescribed a statin,” he said, because “medicine is practiced like that too often.” Thus, as biomarker assays improve, care must be taken to assess their results in the context of everything known about each patient, he warned. “Don’t allow a specific number to drive a patient diagnosis, or treatment, or therapy.” In concluding his talk, Dr. Zakowski observed the following aspects of the report as being the most useful and important: • Provide clarity and common understanding of definitions, process, and criteria. • Provide transparent scientific basis for evaluation. • Recognize the interdependence of analytical validation, qualifica- tion, and utilization. PRESENTATION By RICHARD kuNTz, MEDTRONIC Richard Kuntz, senior vice president and chief scientific, clinical, and regulatory officer at Medtronic, Inc., noted that his remarks reflected a quick review of the report from his personal perspective and not as a representa- tive of the medical device industry. He began by describing the approval process for medical devices, which commences with a premarket or preap- proval phase characterized by “a lot of complicated bench measurement,” he said. He noted efforts underway to reduce the amount of testing and timing that occur during this phase in order to expedite the device approval process through such means as computational bioengineering modeling.

OCR for page 51
 BIOMARKER AND SURROGATE ENDPOINT EVALUATION Device failure—especially for implantable preventative devices such as cardioverter-defibrillators, deep brain stimulators, or insulin infusion pumps—does not manifest as a clinical endpoint, making surrogate end - points for the failure of these devices desirable, Dr. Kuntz said. However, these are difficult to develop. He said that his own efforts to design a model to measure the narrowing of a coronary artery—which he char- acterized as a “very simple process … and mechanism”—was a difficult endeavor with important caveats and limitations (Mauri et al., 2005). “We started out many, many years ago trying to describe the dynam - ics of opening and closing an artery by a stent or any other device by using metrics of diameter measurements, which we called ‘acute gain’ and ‘late loss,’ ” Dr. Kuntz said. “Our interest was to measure this late loss, which we could do with some degree of accuracy through quantitative angiography as a surrogate … for the need for revascularization, which was the clinical interest,” he continued. To do this, he and his colleagues examined the results of several studies, including both drug-eluting stents and bare metal stents for which the endpoint of interest was target lesion revascularization (TLR) rate or need for revascularization, he said. The researchers found a loose correlation between late loss and TLR, after which they conducted a series of studies in an attempt to describe this relationship in a sufficiently reliable way so that TLR could serve as a surrogate endpoint for late loss, said Dr. Kuntz. They also found that perturbing this system with treatments resulted in correlated responses by TLR and late loss. After more than 15 publications on their methodol - ogy over the course of nearly a decade, he and his colleagues felt they had a established a simple relationship between late loss and TLR and a relatively simple mechanical model explaining the relationship, which would be useful in studying late loss (Mauri et al., 2005). However, Dr. Kuntz noted, as a surrogate for coronary stents, their model was ultimately limited. “I think we demonstrated through a lot of work that it was a good surrogate for binary angiographic restenosis and probably TLR,” he said. However, in addition to revascularization— which is the main interest of reducing restenosis with such devices as drug-eluting stents—it is also important to measure stent thrombosis of other revascularizations outside the artery, he said. Their model did not account for such effects, “so therefore, it really did not turn out to be a fantastic way” to test new devices. “We really do have few device surrogates at this point, and they cer- tainly are not as simple as the one we studied,” Dr. Kuntz summarized. “I think the biggest interest we have is trying to allow the timing of postap- proval to occur during the product life cycle in time that can keep pace with technology, and focus more on the postmarket analysis,” he said. Dr. Kuntz expressed his agreement with Recommendations 5 and 6. “There has to be

OCR for page 51
 INDUSTRY PERSPECTIVES more rigor in the postmarket, especially for devices,” he stressed. “We have to be able to look at product release, from efficacy studies to the real world, in a variety of dimensions … [involving] new operators and new patients, with more efficient systems in the postmarket.” He added that the develop- ment of patient registries, along with advances in computational models and observational statistics, should encourage such efforts. The concept of surrogate endpoints could be extended to several device design elements such as computational bioengineering model- ing, Dr. Kuntz suggested. “Product performance is another interesting endpoint in and of itself,” he added, which could be measured when a product that has no surrogate fails. Such an event could also be recorded in a patient registry, he said. In general, Dr. Kuntz reflected that more rigorous postapproval stud- ies need to be conducted, using better observational statistical methods, in order to balance pre- and postapproval data collection and to keep pace with rapid technological development.

OCR for page 51