NEW YORK (GenomeWeb) – Despite concerns over the scientific validity and usefulness of genetic tests that purport to personalize diet, exercise, and weight loss strategies, such products continue to proliferate.
Cecile Janssens, research professor of translational epidemiology at Emory University's Rollins School of Public Health, has been researching consumer genomics firms, the evidence underlying the tests they sell, and the terms of service that people agree to when buying these tests.
Over the past decade, while the public's interest in genomics, particularly precision medicine, has grown, Janssens research indicates that there is still a lack of evidence supporting the use of DNA information to guide dietary decisions. The entrepreneurs behind these companies often lack clinical genetics expertise and the terms of service that customers agree to in purchasing the tests make it difficult to understand what they're buying and all that the firms can do with their data.
Janssens recently discussed the field of nutrigenomics with GenomeWeb. Below is an edited transcript of the interview.
You and your colleagues did research 10 years ago evaluating genes being tested by consumer genetics companies and found the clinical validity of these genes lacking. Although most of these companies are no longer around, could you elaborate on what you found with regard to the evidence base underlying nutrigenomic testing back then?
Ten years ago, we wrote a paper about the genetic tests offered by companies who wanted to give you insight into your risk of various diseases, and related to that give recommendations on how to change your diet and lifestyle. At that point, the research was really premature and we wondered whether there was scientific evidence and whether these genes were associated with the risk of disease. We found that these companies were testing for genes for which we didn't find any statistically significant association as published in meta-analyses.
We also found that companies were testing for genes that were more strongly associated with diseases other than the ones they claimed to predict. So, the paper showed that really there was no evidence for the tests these companies were offering.
These days it's a bit different because the companies that are selling these diets and exercise recommendations don't claim to predict disease. They don't tell you that you have an increased risk for heart disease, therefore you adopt a healthier lifestyle. They just focus on genes that impact how our body processes or metabolizes certain nutrients. For example, they can look at genes related to how well our bodies process caffeine — some process it fast and others slow. What some companies are now doing is saying, for example, that if you're a fast metabolizer of caffeine, you fine, but if you're a slow metabolizer, you need to reduce the number of cups of coffee you drink a day.
These companies need to establish a gene-nutrient interaction, which is more difficult to establish than just a genetic association. They need to show that if they find variation in the response of those genes, how much you need to adjust your diet to obtain the effect they want you to have. They also have to specify what will be the benefit if you change your diet. In that respect, there needs to be at least two kinds of evidence: the evidence of the association but then also the evidence that changing your diet delivers the benefits you want it to have.
There has been a recent resurgence in nutrigenomic testing. Why do you think, 10 years since your research, there seems to be new companies marketing this type of testing?
There are several trends ongoing. First of all, the genome is hot. A lot of people are interested in the genome not only for discovering risk factors for diseases, but also in making changes to DNA as we see with CRISPR. We are also increasingly interested in precision medicine, which means we feel that general health recommendations don't work for everyone — which we've know for a long time — and that people might benefit more if we tailor recommendations to their genetic makeup. The interest in precision medicine, I think, is an important factor behind the renewed interest in nutrigenomics.
A third reason is genotyping is really cheap these days. It's something you can easily offer for a low price and that attracts consumers. Ten years ago, when these companies were offering these tests, it was still around $400, sometimes $1,000. Now you see the prices are already less than $100.
Have you taken a look at the scientific basis of some of the more recent nutrigenomic offerings that promise to personalize diets and weight loss regimens? Has the science advanced within these newer offerings?
There isn't much evidence out there around the gene/diet/outcome association. We know how your body responds to certain nutrients depending on your genetic makeup, but the studies out there on whether changing your diet helps are usually very small, and show very minimal results. Also a recent, large multicenter study showed some effect but rather minimal. So, there is no robust evidence out there that shows that tailoring your diet to your genetic makeup makes any sense.
While nutrigenomics companies are launching these tests, have expert groups issued any guidelines or recommendations around this type of testing?
There was a 2014 position paper by the Academy of Nutrition and Dietetics where they say that nutritional genomics holds promise but that there is insufficient evidence at this moment that tailoring diet to a person's genotype has any benefits for the person. They are currently revising this position paper but I would be surprised if their position changes radically because the evidence hasn't changed in the meantime.
Among geneticists, it my understanding, most don't take this extremely seriously and feel there is no need yet to jump in with a position paper.
Is there any kind of guidance at least on whether this type of information should be sold directly to consumers or provided through a doctor?
No. I mostly see blogs and journalists who pick it up summarizing the little bit of evidence that's out there and then concluding that it's not ready for prime time, and that people shouldn't do it.
In 2017, you published a commentary, entitled "Uninformed consent in nutrigenomic research," where you discuss how consent was garnered in studies where APOE testing was offered. Can you elaborate on why those consent policies were problematic?
This commentary was about two European studies, one large and the other smaller, where researchers were investigating whether people respond to diets tailored to their genetic makeup. One of the genes being tested in both studies was APOE, which is related to cholesterol regulation, and so both studies recommended people reduce their saturated fat intake depending on their genotype. But maybe what [investigators conducting] both studies did not realize was that APOE is also very strongly associated with the risk for late-onset Alzheimer's disease.
So, the people who needed to reduce their saturated fat intake were also the ones with this increased risk for Alzheimer's disease. We argued in the paper, that this is very irresponsible, but it also indicates that clinical geneticists are not involved in these studies because APOE is such a famous gene that any clinical geneticist would have picked up that this [information] cannot be returned to people without informing them that this gene is a major risk factor for Alzheimer's. To avoid this risk, this is why it's so important to have clinical geneticists at least look at the designs of these studies or the products that companies want to offer.
You see on the internet that the people behind nutrigenomic tests, or DNA tests for sports [performance], are entrepreneurs. They may have a medical expert or geneticist on board, but the geneticist is usually not a clinical geneticist. It's usually a molecular geneticist or a population geneticist who may not know the big genes out there for clinical genetics. To prevent these types of mistakes that are really irresponsible these companies better team up with clinical geneticists or genetic counselors.
One of the important observations in this commentary was that even the term "nutrigenomics" is problematic in that a consumer might assume that the term suggests this is an interdisciplinary field that involves clinical genetics. But that's not really the case.
The term "nutrigenomics" may sound like a sub-discipline of genomics, but it's more like a sub-discipline of the nutrition sciences, and it's often conducted by people who don't have too much understanding of genetics. The minimum you can do when you start researching this topic or start offering commercial DNA tests is to make sure that you have the appropriate knowledge in house. But because most companies may have a molecular geneticist on the team, it's unlikely that consumers will understand that this might not be sufficient.
I have studied the terms of service of several consumer testing companies and my first impression, even as a scientist, is that they are very difficult to read. They basically tell you [the companies] are not responsible for anything they offer, and that you're taking the test at your own risk, and that there are a lot of things they can do with your DNA that they don't explain in all the details.
For example, you have this new platform, called Helix, which is an app store for genetic applications. Helix has its terms of service, and each of the companies that are offering apps through Helix also have their terms of services. When you order a personalized diet test on this platform, you order that with the company offering the app but Helix is doing the [exome] sequencing. So, when you buy the app, the [app company] gets only part of your DNA information, but you consent that Helix can sequence your genes. I am not sure how clear that is to customers.
Also, the terms of service between Helix and the app companies are not always identical, so there may be differences between the responsibilities that Helix takes and the responsibilities that the app companies take. We need a far more careful review of that and if that is in the consumers' interest.
For the health-related apps within the Helix marketplace, genetic counseling is provided to customers if they want it. Should pre-test and post-test counseling be included as a mandatory part of all these apps?
It depends on what the company is sharing with you. If a nutrition company, for example, is testing for the APOE gene, among others, to tailor your diet, then yes, people need to be informed that APOE is also an important risk factor for Alzheimer's. That may be a reason that people don't want to hear their nutrition advice. But most of the genes these companies are testing, they may have a tiny effect on many diseases, but the effects are so very small that it's unlikely they have any impact on consumers whether they know them or don't know them. But for the genes we know from clinical genetics that really have an impact on people's lives, because your risk of disease becomes so high, yes, people need to be counseled before and after testing. You don't want to be surprised by this knowledge without considering whether you wanted to know.
The FDA has stepped in to regulate health-related DTC genetic testing, but hasn't said too much about nutrigenomic testing. Is this an area FDA should regulate?
I don't know if the FDA has to get involved at this moment. There's not much to regulate, because most of these tests don't target diseases. If the tests focus on nutrition and exercise, they don't need FDA approval.
Lastly, you may have seen that 23andMe recently announced it was launching a study involving 100,000 customers to investigate the genetic variants associated with weight loss. Do you think a GWAS of this size could help add more scientific rigor to the nutrigenomics space?
I saw the 23andMe study. I don't think that makes any difference. If this is an ordinary trial, then I don't expect anything major. Losing weight depends on so many factors, adherence to your diet being the major one, that genes unlikely are of any relevance. They may find some statistically significant results, but these are unlikely useful for recommending diets.