This chapter summarizes session 4 of the workshop, a panel discussion involving Cecile Janssens, Douglas Wallace, Steven Zeisel, and Tim Morck, Spectrum Nutrition, LLC. The goal of the discussion, said moderator Patrick Stover, was to address the question: Where do we go from here, in terms of both where the field is moving and new expectations for nutrition?
Before introducing the panelists, Stover summarized the workshop as “an update to where the field of nutritional genomics is, based on the last meeting we had 10 years ago.” The field has evolved rapidly in his opinion, with respect to not only new knowledge about the role of genetics in nutrition but also expectations for nutrition in terms of chronic disease outcomes. In addition to the bar being raised for outcomes, that is, moving beyond functional indicators as discussed by Patsy Brannon in her opening presentation (see Chapter 1), he called attention to the fact that the bar has also been raised with respect to the type of evidence that is needed. He noted that a recent National Academies publication on chronic disease outcomes calls for grading evidence such that nutrition recommendations would have to have at least moderate evidence (NASEM, 2017a). Although he believes that observational data, which he said drive this field, can reach that level of moderate evidence, large sample sizes and a dose dependence will be required to demonstrate any effect of an intervention. Thus, he said, “we have a lot of challenges in front of us.”
To initiate the discussion, Stover asked the panelists how they viewed the role of precision medicine, that is, being able to classify individuals as responders versus nonresponders (i.e., to a drug), as a paradigm for nutrition. Zeisel did not answer the question immediately but stressed that now is a critical time to introduce nutrition into precision medicine, as major medical centers are beginning to establish data collection systems upon which precision medicine will be based. Yet, he said he was unaware of anyone who was thinking seriously about the nutrition information that would be collected, adding that this information is often viewed as being too difficult to collect. “But if we don’t insert ourselves into the field so that the data is collected,” he asserted, “nutrition will never be part of precision medicine. It’s a critical time.”
Regarding the appropriateness of the precision medicine paradigm for nutrition, in Zeisel’s opinion, the field of epidemiology “is being turned on its head by recognizing that people are different.” Before today’s nutrigenomics tools were available, he observed, one would have concluded, for example, that half of young women need a particular intervention and the other half do not. He stated that this type of result cannot be used to make a recommendation. With the new technology, however, what he described as “finer cuts” can be made. Now, he elaborated, one can examine the 50 percent that respond and the 50 percent that do not and determine why different people respond differently. Doing so will “get the noise out of nutrition data,” he said. Then, he suggested, it will be possible to develop targeted interventions that work more predictively on subpopulations with unique genetic characteristics or some unique combination of genetic and other characteristics.
Janssens cautioned, however, that a challenge to classifying people on the basis of their profiles is that “we very easily become unique.” As soon as everyone becomes unique and there is no other “me” from whom to learn, she argued, it is very difficult to say what the best treatment or diet is for an individual. In other words, she said, profiling can reach a point at which “there is no one else with that profile,” and the only way to test interventions is through trial and error. She was curious as to whether anyone had any ideas on how to address this challenge, saying, “That, for me, is the limiting factor both in precision medicine and in precision nutrition.”
In contrast, Zeisel expressed optimism and the view that many of the ways in which people are unique have little to do with the profiling factors used to characterize individuals as responders or nonresponders. He explained that it is possible to identify profiling factors that make significant contributions to determining whether an individual will be a responder or a nonresponder, although a decision must be made about what constitutes
significant: Is a 1 percent contribution sufficient, or should it be 10 percent? “We arbitrarily decide how close we need to be to make a recommendation,” he noted. He referred to José Ordovás’s comment earlier in the workshop about perfection being the enemy of good. In his opinion, if enough profiling factors can be identified that predict 92 percent of responders versus nonresponders, that is probably good enough, whereas if the prediction rate is only 32 percent, more work should probably be done to characterize response.
Janssens stated that she was not convinced the question is that simple. When a large amount of data is collected, she said, the data reflect many different profiles that distinguish “me” from others. In her opinion, it is very difficult to identify what accounts for why different people respond to different treatments.
Wallace suggested that there may be a continuum of variants, with different variants (i.e., what Zeisel termed profiling factors) having different impacts. Some act like single genes, he elaborated, which is the basis for newborn screening whereby a particular variant is identified, and a specific nutrient is given to compensate for that variant. In contrast, he continued, mitochondrial medicine, a field in which physiological processes are affected by broad groups of related genes, requires a more general approach. Depending on the pathway that is affected, he explained, certain kinds of defects may require a more glucose-rich diet, for example, while others may require more fatty acids in the diet. The more interesting challenge, in his opinion, is the percentage of variance that can actually be controlled with diet. He wondered whether there may be too much noise, or too much variation, to control a risk for a disease such as Alzheimer’s or diabetes.
Zeisel pointed out that even in medicine, efficacy is quite low. If 30 percent of people who receive a drug treatment benefit, he argued, “we’ve done well.” He cautioned against holding nutrition up to a more stringent standard. Given that the goal of precision medicine is to try to do better than this 30 percent, he asserted that if personalized nutrition guidelines could be developed for a reasonably sized subpopulation with the understanding that the guidelines could be wrong for any given individual, “you’d be doing really well.”
In that respect, Janssens said, the term “precision nutrition” is misleading. She suggested instead “stratified nutrition,” which she believes presents a more realistic picture of what can be expected.
Zeisel argued that adopting the same terminology used by the medical community will be necessary to ensure that nutrition is actually used by that community. If “precision medicine” is the term medical professionals recognize, he stated, then only by using the term “precision nutrition” will nutrition be recognized as “every bit as important as knowing what dose of a drug to give.”
Stover observed that the discussion to this point had focused on the classification of nutrition at a clinical level. He asked Morck about the impact of nutrigenomics on population as opposed to clinical nutrition.
“My bias is that the field of nutrition encompasses a pretty broad spectrum of health,” Morck replied. Nutrition is not just about food, he stated—it includes absorption, digestion, and all the other kinetic and dynamic processes that Patsy Brannon had laid out in her opening presentation. It also includes education (knowing what good nutrition looks like) and selection (actually making good food choices), he added. Even with all the genetic information and medical advice available, he said, “If we are not motivated to take that advice, it is worthless.” He emphasized the behavioral aspects of nutrigenomics that Brannon, Ahmed El-Sohemy, and Nathan Price had all touched on in their talks. For nutrigenomics to be successful, he asserted, the field needs to find a way to convince people that it is not just beneficial but may be imperative to change their diet should something be discovered about their genes that indicates risk. Without that motivation, he claimed, change will not happen. He pointed to smoking as an example of something that is known to be “bad,” but that people still do. Lifestyle factors need to be taken into account as well, he cautioned. With respect to precision nutrition, he imagined someone having a recommended precision diet based on genetic information, but then attending a family Thanksgiving dinner. He reminded the audience that people eat food in social settings, and suggested that precision nutrition could inadvertently impose a level of social isolation by not taking that fact into account.
For Morck, the personalization of nutrition is the personalization of one’s approach to incorporating nutrition into one’s lifestyle and as an important part of one’s future. He noted, for example, that the Mediterranean and Paleo diets provide guidelines, but posed the question of what will guide people to making better choices on an incrementally more frequent basis. Additionally, he stressed the importance of having validated biomarkers that are sensitive enough to show a metabolic or physiologic benefit when people make the effort to change their diet. At present, he asserted, the tools available are too crude to provide that type of reinforcement, and body weight, blood pressure, and cholesterol levels, for example, are not specific enough. He believes that metabolomic markers hold promise for providing patterns specific enough that people will be able to see changes within a few months and perceive the value of continuing the new dietary trends they began. He expressed the view that “the feedback is really critical . . . in making significant, long-term nutritional changes that are expected to produce real benefits.”
Zeisel suggested that there may be a biological component that helps explain why some individuals are willing to change their diet while others are not. Moreover, he suggested that this component may not be just genetic. For example, he observed, when the feces of a person with anorexia are transferred into another person, the other person starts behaving as though anorexic.
Wendy Johnson, session 2 moderator, agreed with Morck, saying, “We definitely have to have a behavioral piece . . . so that we can have the benefit of these new discoveries.” She referred to Zeisel’s earlier remark that nutrition needs to be included in precision medicine studies, or it will be left behind. She posited that the same is true of behavior—it needs to be included as well. She emphasized the importance of behavioral phenotyping and of understanding what causes people to adhere, whether that cause is genetic or something else.
Stover pointed out that, in addition to genetics, several speakers had discussed epigenetics (Ordovás), the microbiome (Price), and other measurable features. He asked whether it made sense to continue the focus on nutrigenetics in this era of data integration.
Zeisel remarked that the nutrigenetics toolset is 5 years ahead of other toolsets, which in his opinion is the only reason it has become the priority. The microbiome is much further behind, he added, given that it attracted little attention until about 10 years ago, and the problem with epigenetics is that target tissues cannot be collected, only blood samples. “I think it is a mistake,” he argued, “to try to push too hard when you don’t have the toolset yet.”
Wallace wondered how metabolomic biomarkers in particular will be found in the future, given the inherent variability of humans. In his own research, he works with inbred mice and therefore is able to remove most of the variance so that significant chemical signatures can be identified. With that knowledge, he asserted, one should then be able to study that same chemical signature in humans and identify which groups of individuals show variation and which do not. “We have to go back and define . . . what the real markers of relevance are,” he said, “then go back to the [human] population and stratify the population.”
Zeisel added that in his opinion, the “challenge test” will be an important tool for nutrition metabolomics. He suggested that it may be the only way to see metabolic variation because unless a system is challenged, the body’s homeostatic mechanisms are capable of managing any variation in metabolism. A challenge test pushes the system and reveals weaknesses, he added.
Wallace agreed that in mitochondrial medicine, when a nutritionist believes it is appropriate, administering a challenge test such as the glucose challenge can help identify which part of the metabolic pathway is most impaired. With that knowledge, he explained, dietary recommendations can be tailored to meet that need. “That’s a really important tool for us,” he said.
Stover commented on how it has been estimated that nutrition-related chronic diseases cost the U.S. economy about $1 trillion annually. Thus, he asserted, one of the goals in nutrition is to lower rates of chronic disease, with much of the discussion at the workshop being relevant to those efforts. Yet, he observed, chronic disease is present mainly in low-income communities that are least likely to benefit from nutrigenomics and have bigger problems to deal with. He asked, “How do we deal with the equity issue as we begin to think about advancing the science toward reducing chronic disease when the target population that is driving the chronic disease may be the least receptive to what we have to offer?”
Wallace commented on the importance and difficulty of studying different ethnic groups. He mentioned that he and his team have been conducting fairly large studies on macular degeneration and glaucoma, which he characterized as having very different risk states in African Americans, Asian Americans, Native Americans, and European Americans. Thus, he stated, it is necessary to stratify by ethnicity, then substratify within each group, so that variable(s) that are contributing to the differences in disease can be identified and the percentage of variance due to those variables determined. Additionally, he and his research team have been very interested in studying Asian Americans in California, but have spent almost all of their budget just assembling a population that would be representative of both first- and second-generation Asian Americans. “I think this is important,” he said, “but there are also constraints, and I don’t know how we are going to manage that.”
Another point to be made, Wallace continued, is that there is an inherent assumption that a person of lower socioeconomic status has a clinical problem because of his or her socioeconomic status. He views that assumption as a sociologist’s perspective, and as a geneticist, he disagrees. He cited as an example the assumption that African Americans have a higher rate of preterm birth because they do not receive good health care. But a study in Chicago stratified African Americans by socioeconomic status and showed that preterm birth was the same regardless (Collins et al., 2007). Wallace interpreted those results to mean that in fact, there may be much more going on genetically than has been assumed in the past.
Zeisel mentioned studies conducted in Great Britain showing that socioeconomic status in early life modifies epigenetic markers and expressed concern that corporate databases are being built from people who can afford to buy genetic tests, so that the data will be heavily biased toward rich people. He suggested that because health care systems have proposed collecting data on all patients, perhaps the collection and analysis of those data could be funded in a way that would allow for the data to be correlated not only with outcomes but also with socioeconomic status. He mentioned the National Institutes of Health as a possible source of such funding.
For Janssens, the question of equality is so important that she feels uncomfortable discussing nutrigenomics and puzzling over such detailed tweaking of diet when there is an enormous nutrition problem in society that will likely require completely different types of solutions. “I’m sure that when I walk out of the building here later, and I see the people on the street,” she said, “that I’ll question myself, ‘What have I been doing this entire day? Why have I not tried to solve a bigger problem for them, instead of trying to find a little benefit in nutrigenomics?’”
Ordovás asserted that “most of the time, the people who are missing from this discussion are the policy makers.” He mentioned how the policy of subsidizing meat, for example, is known to have affected the nutritional health of certain neighborhoods. Additionally, he observed, it is known from human breastfeeding data that the quality of milk varies among neighborhoods depending on socioeconomic status. Given these circumstances, the component that is missing from the nutrigenomics discussion, in his opinion, is public policy making.
Janssens agreed with Ordovás, but only partly. “I think also,” she said, “to find a solution for the people who need it, we have to first ask ourselves whether the solution for them is in nutrigenomics.” In her opinion, this is not likely.
Zeisel disagreed. He characterized data studies as noisy, generating a great deal of variability. Results of one study will come out and lead to a new policy, for example, to recommend eating more cocoa; then another study will lead to a different policy because it was based on a different population. These discrepancies, Zeisel argued, turn people away from nutrition and from using it in policy at all. In his opinion, imprecise, noisy data lead to bad policy. Thus, he said, “I think everything we can do to refine our ability to explain the noise and to understand why some people respond and others do not is useful.” “Sure,” he added, “the big problems are going to be solved by policy, but [policy makers] won’t come up with the right answers unless our research techniques are refined.”
Johnson wondered whether a cost-effectiveness or cost-utility analysis of some sort could be carried out and the results used to talk with policy
makers about whether genetic testing is a good investment in populations that cannot otherwise afford it.
An audience member described how a genetics lab in Texas used DNA to perform facial reconstruction and how law enforcement officials, in turn, used the facial reconstruction to arrest a suspect for a crime. Historically, she pointed out, communities such as the black community have experienced what she called “medical malfeasance.” “So there’s a lot of distrust in wanting to give genetic information,” she said, even if it is for science. She asked the panelists how this distrust would impact nutrigenomics.
In Zeisel’s opinion, the use of DNA in facial reconstruction is the same as asking a witness to describe a suspect. The only difference, in his view, is that instead of being obtained from an observer, information is being sought from DNA based on what is known about genetic differences among, for example, different ancestries. Regarding medical malfeasance, he mentioned the Tuskegee study, in which black people were infected with syphilis to study its effects, greatly harming credibility. Today, by contrast, there is an expectation and a requirement that researchers consider a community’s interests. He cited the example of his own research team, which, especially with studies pertaining to genetic testing, solicits input from a community advisory board. He described the members of the board as members of the local community who can provide perspectives that the researcher may not be able to perceive. He also emphasized the importance of good communication and the sharing of benefits with respect to the knowledge generated by a study. He stated that participants should be told about the uses of the data, and they must give permission for the data to be used for other research purposes.
“But in the end, you are taking a risk when you give your genetic data,” Zeisel continued. For example, he noted, concern has been raised that if an investigator were to be subpoenaed by the U.S. court system, he or she would have to release study data even if the research subjects had been promised that their data would be confidential. He stated that for this reason, some investigators have been storing their genetic data on servers outside of the United States. He emphasized the importance of investigators informing individuals of the risks of their participation in a study and what is going to be done to mitigate those risks. Then, he said, individuals can decide whether they want to participate.
An audience member asked about the potential impact of the consumption of genetically modified foods on a person’s long-term genetics. Zeisel replied that the genetics used to modify a plant has little to do with nutrigenomics, except for the fact that better data on people’s nutritional needs could be used to try to design better foods. He added that traditional breeding is a form of genetic manipulation. In the past, farmers chose plants with the largest fruits, for example, and bred those plants, whereas with the new techniques available today, they are making plants artificially instead of breeding. But he said that in terms of the genes being inserted into those plants, other than the effect of any food on one’s genetics, he was unaware of any inserted gene that could also enter the human genome upon consumption.
While the discussion was on the topic of agriculture, Naomi Fukagawa pointed out that much of the genetic variation in humans also exists in plants, animals, and the food that is produced from those plants and animals. In her opinion, agricultural production goes hand in hand with efforts in medicine aimed at wellness and the prevention of disease. She noted that even Hippocrates, who is credited with the precept “first do no harm,” also stated, “food is thy medicine.”
This page intentionally left blank.