Personalized Eating: The 'New' Diet?

John Watson


June 13, 2019

We have long assumed that the same "healthy" diet was healthy for everyone. But what if the food you are eating is all wrong for your body?

Welcome to personalized nutrition, an emerging field using a mix of science and technology that determines what foods are good or bad for you by designing an individualized diet. In just the past few years, the research has overturned conventional wisdom surrounding nutrition and may soon entirely collapse the monolithic dietary guidance upon which we've relied to date.

Although the definition of "personalized nutrition" is still evolving, it has been concisely described as "an approach that uses information on individual characteristics to develop targeted nutritional advice, products, or services."[1] Those "individual characteristics" can include physiologic features (eg, age, gender), fluctuating environmental factors (eg, sleep patterns, physical activity), more cutting-edge insights obtained by assessing a person's genetic or microbiome profile, or some combination thereof.

On its own, this information may not seem to have much disruptive power. Yet, using "big data" collected from thousands of people around the world to create algorithm-based prediction models has revealed surprising disparities in how individuals respond to different foods.

Eran Elinav, MD, PhD

We are only in the early stages of understanding how individuals respond to specific foods, with the best available data focusing on glycemic outcomes, yet there is considerable optimism that personalized nutrition will become an invaluable tool for helping patients decide what to include and exclude from their plates to maintain overall health and manage or prevent disease.

"We envision that in 5-10 years, one-size-fits-all diets would become obsolete and be replaced by diets that will be based on a combination of individual host and microbiome features, the medical condition to be treated, and personal preferences," said Eran Elinav, MD, PhD, a professor of immunology at the Weizmann Institute of Science in Israel, and a pioneer in the field.

State of Personalized Nutrition Science

So far, the most compelling data on the promise of personalized nutrition have come from studies incorporating the gut microbiome profile. These raise the tantalizing possibility that the information gleaned from a simple stool sample can help predict responses to specific foods.

Elinav and colleagues tested this theory in a watershed three-part study focusing on postprandial glycemic responses (PPGRs).[2] First, they enrolled 800 participants from Israel and tested glucose after every meal they consumed, confirming that PPGRs varied considerably among them, even when they ate identical foods. Next, they incorporated all the information they obtained from this cohort into an algorithm, along with other factors, such as age, weight, and microbiome profiles, and validated the algorithm's ability to accurately predict PPGRs in a separate, smaller cohort. Finally, they confirmed the algorithm's real-world value in a blinded, randomized study, in which microbiomes and clinical information were used to create customized diets for each participant based on that person's predicted PPGR to foods. The algorithm-designed diets resulted in significantly lower PPGRs. Thus, the study made a compelling case for proof of concept, validity, and applicability, all in one publication.

"Although the concept of interpersonal differences in response to food has been previously described, we believe that formulating individual host and microbiome features into dietary recommendations in a large cohort in an evidence-based manner was unprecedented and revolutionized the outlook on nutrition," Elinav said.

In an industry-sponsored study[3] published this year, researchers further validated this algorithm by testing its ability to predict PPGRs to food in 293 participants without diabetes. Because the microbiome is influenced by environmental factors beyond diet and can vary by region, reconfirming the earlier study's findings in a different population was crucial. These participants were almost entirely from the American Midwest.

Consistent with the previous findings, glycemic responses to identical food items varied significantly from person to person. For example, serum glucose levels after eating a bagel with cream cheese rose by a low of 6 mg/dL in some individuals (low responders) to a high of 94 mg/dL in others (high responders). And some findings were counterintuitive to commonly held nutritional guidance, as evidenced by blood glucose levels that rose higher after eating a banana than after eating a cookie.

Heidi Nelson, MD

This is particularly important, according to Heidi Nelson, MD, who leads the Center for Individualized Medicine Microbiome Program at the Mayo Clinic in Rochester, Minnesota, and is one of the authors of this study, because it replaces the relatively uninformative averages on which nutritional advice is based with precise variable-derived predictions.

"We want to know whether someone is a low or high responder to various foods, because you can substitute high-response foods with another food, or avoid those foods entirely. We've also shown that in some cases, combining foods lowers the glycemic response. Sometimes a few almonds added into a certain food mitigated the glycemic response," she said.

Commercialization Before Validation Raises Concerns

Although these forays into personalized nutrition have garnered significant attention, concern remains that expectations for its actual application may be unrealistic at present. But commercial entities aren't waiting for more data; they have already begun marketing these tools to the general public.

DayTwo, Inc., is using the technology to provide personalized nutrition recommendations that predict a person's glycemic response to various foods. It's as simple as buying a kit and mailing in a stool sample. After testing the stool, DayTwo claims to provide access to dietary advice that can help modify outcomes ranging from weight loss and gastrointestinal symptoms (eg, bloating, constipation) to mental and immunologic health. Other companies offer broadly comparable services, but their prediction tools haven't been vetted in the peer-reviewed literature to the same level.[2]

Several personalized nutrition companies—for example, Habit, GenoPalate, Nutrogenomix, and Profile—focus on DNA testing. Though they differ in their offerings, the basic pitch is the same. Consumers purchase a kit, either through the mail or at a brick-and-mortar location, which they use to provide blood samples and/or cheek swabs. The kits are sent to the companies, which then analyze the person's genotype and phenotype and compare them against established genetic markers. Consumers are informed what foods are right for them based on their predicted ability to metabolize them, which nutrients they should be seeking and avoiding, and what is likely to result in the most weight loss. Advice comes in printed reports or on digital apps, right down to personalized recipes.

These kits range in price from $150 to $350, although add-on features (memberships, extended coaching sessions with dietitians, and foods and supplements offered by the companies) are also available. These companies are doing more than selling a one-time curiosity product; they are establishing relationships with consumers to provide long-term health advice. The model is some combination of genetic information providers, such as 23andMe or, and popular weight-loss institutions, such as Jenny Craig or Weight Watchers. It's easy to envision why this would be intriguing to both consumers and investors. However, their websites are often vague about what they can provide or the science underlying their services.

Alexander Khoruts, MD

"My discomfort is that these proprietary methods have gone into full commercialization before any real validation has occurred," said Alexander Khoruts, MD, a scientific advisory board member for the American Gastroenterological Association Center for Gut Microbiome Research & Education and medical director of the Microbiota Therapeutics Program at the University of Minnesota in Minneapolis.

Other researchers are similarly dismayed by the claims some companies are making in the absence of large-scale validation studies.

"Personalized nutrition services should be scrutinized," said Elinav. "Constituents other than glucose, such as fat, protein, vitamins, and minerals, should be taken into account in dietary planning. Food allergies and other individual considerations shouldn't be overlooked. Blindly following automated reports is never advisable."

They are also largely in agreement about the key limitations of personalized nutrition research to date. Khoruts identified one of the lingering issues to be the lack of standardization among testing kits. Elinav agreed with this characterization, noting that whereas standardization has been achieved in his own model, it is a bigger concern with more inadequately studied direct-to-consumer kits.

How Personalized Nutrition Can Realize Its Potential

Nonetheless, Khoruts acknowledges that because of these direct-to-consumer services, physicians are likely to encounter more patients asking about them.

"Gastroenterologists are seeing patients who bring in these reports to the office, because the companies are marketing directly to consumers. They have pretty websites that promise a lot," he said.

The trouble is that the reports are difficult to explain to patients, even for an expert such as Khoruts. For those incorporating microbiome analyses, patients are often told whether their profiles are within ranges considered normal, when in fact sufficient analyses have not yet been provided to establish these ranges convincingly. This means these reports offer little actionable advice for physicians to work with, although Khoruts does see their value in simply generating productive interactions with patients.

"If patients are engaged and are doing some of this research on their own to seek answers, it is also an opportunity for a physician to engage in conversation with the patient."

Elinav added that reports generated by a personalized nutrition company should never replace common sense. "They should be integrated into the armamentarium of physicians and medical personnel as an auxiliary evidence-based means to enhance precision in dietary planning," he said.

Regardless of these current limitations, the potential to use this research to inform diagnosis and treatment is considerable, given the microbiome's hypothesized role in so many elements of human health, such as cancer progression, infection susceptibility, inflammatory bowel disease, neurologic disorders, and metabolic conditions, said Elinav.

Nelson added that perhaps the most important thing to remember is that this is a young field of science that will take time to sort out.

To fulfill its promise, more well-designed studies of the type conducted by Elinav, Nelson, and others are needed, and perhaps a little less of the hype coming from the commercial side of this field.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.