Researchers from the Israeli Weizmann Institute have reported data that suggest general dietary recommendations may be of limited utility, because individuals vary enormously in their blood glucose response to any given food.

However, using a combination of individuals' responses to certain foods and the compositions of their microbiome, it was possible to predict individuals' glycemic responses and design diets that significantly reduced swings in blood sugar.

The study, which appeared in the Nov. 19, 2015, issue of Cell, "offers a distinct new look at nutrition, and how it affects our bodies, our microbiomes and, consequently, our health," co-corresponding author Eran Elinav told reporters at a press conference.

It also suggests both therapeutic possibilities for targeting the microbiome directly, and diagnostics that could improve clinical trials for drugs aimed at metabolic disorders and possibly, other types of disorders as well.

The paper "shows the power of this type of big data and the need to do this kind of work," Karim Dabbagh told BioWorld Today. Dabbagh is the chief scientific officer of Second Genome Inc., a company focused on therapeutics development based on an understanding of the microbiome.

The work now reported in Cell represents a possible starting point for that kind of work.

The study is "all correlation at this point," Dabbagh said. "But there are some very clear correlations" between the presence of certain bacteria in the microbiome, and appropriate glucose control.

One of those bacteria is Eubacterium rectale, and Dabbagh said one approach that might be taken from a therapeutics development standpoint would be to understand mechanistically what Eubacterium rectale is doing or producing that might have a positive effect on glucose control, and then mimic that effect "in the form of a drug, not a live bacterium."

But the findings could be used in myriad different therapeutic and diagnostic ways.

Dabbagh noted that in the short term, the complexity that is added by the new findings might seem like one more headache for drug development.

On one level, "the paper is a really elegant analysis of the obvious, which is that everything you do in life affects your health," he said.

With 800 patients and more than 100 parameters, it's also "a pretty heroic study."

The work demonstrates that in any trial of 800 patients, "there are a fair amount of patients that are not going to behave the way you expect them to" – and furthermore, that this is probably the case with every regulated process, "which is basically everything."

Having a way to predict those responses means that "you can now start thinking about much better drugs for the right individuals." This is also true during drug development, where a way to predict responses could lead to better ways to recruit and stratify patients.


Elinav said that the paper suggests that "a fundamental change from the way we've all viewed nutrition in the past few decades" is in order.

In the broadest sense, the paper suggests that the effect of a food is not an intrinsic quality of the food itself. Instead, the most informative approach is to find ways to measure, and ultimately predict, individuals' responses to foods.

"In our studies, these [responses] seem to be of much more potential" for understanding what a healthy diet would consist of for an individual, Elinav said.

Previous small studies had suggested that not every individual has the same blood sugar response to each food, and so Elinav, co-corresponding author Eran Segal and their teams decided to look at such responses both in a larger cohort – their study included 800 subjects, none of whom had diabetes but many of whom were prediabetic – and in greater depth.

They decided to focus on blood glucose levels, and more specifically on what's called the postprandial glucose response (PPGR) – the change in blood sugar levels that occurs within the first two hours after consuming a meal – for both scientific and practical reasons.

Multiple studies have shown that a PPGR characterized by spikes in blood sugar is a risk factor for diabetes, obesity, and other components of the so-called metabolic syndrome, a group of measures that collectively predicts high risk for cardiovascular disease.

Many things about the PPGR are also easy to measure.

"Within a short time we can obtain many different measurements of what we would refer to as healthy nutrition," Segal told reporters.

Measure they certainly did. The team subjected study participants to Big Brother-like surveillance for a week, measuring more than 100 physiological parameters, including blood glucose levels at five minute intervals, and detailed information about any and all food consumed, including both standardized meals and the regular diets of the participants (with ingredients weighed on investigator-provided scales).

They also collected blood samples and information about sleep schedules, exercise and various stressors.

Finally, they looked at the microbiome compositions of their study participants. In the larger initial group of 800 patients, that look took the form of a snapshot. In a smaller replication cohort of 26 individuals, the team repeatedly measured the microbiome to identify changes over time that correlated with the responses to food.

They found that whether a food led to blood sugar spikes was not an absolute quality, but one that depended on the eater, with the eater's microbiome playing a major role.

Though the project was conceptualized and conducted as basic research, the Weizmann Institute is in licensing discussions with several commercial entities who hope to develop diagnostic tests that could make individualized recommendations for what is a healthy diet.

Segal said that although his team cast a very wide net in their physiological measurements, "we believe that there is a much more minimal set of features that one can strive for. . . . Intuitively, we believe that the microbiome would hold much of the key to being able to make these predictions."

In addition to such predictions, the microbiome could also be modified to improve individuals' PPGR to certain foods, either through diet itself – which, the authors showed in their work, affected microbiome contribution – or through interventions such as probiotics.

"Unlike our genes, our microbiome has the potential to change," Segal noted.

The study was conducted in Israel, where Segal is a professor of mathematics and computer science, and Elinav is a professor of immunology at the Weizmann Institute of Science. In terms of their eating habits, their weights and their microbiomes, the subjects were similar to what has been reported for American and European populations.

Perhaps disappointingly, as the Eating Season that lasts roughly from Thanksgiving to New Year's begins next week, the study results do not imply that eating too much of the "right" foods is healthy.

"By no means do we overlook the importance of eating quantities that would not be excessive," Elinav told BioWorld Today at the press conference.

Still, the study showed that what might be sound advice for the population as a whole could be counterproductive in a significant fraction of people.

For example, eating bread with butter rather than without can smooth out blood sugar spikes in 80 to 90 percent of the population – but on the other 10 to 20 percent, it can exacerbate such spikes.

And at the press conference, the scientists described a woman whose blood sugar spiked in response to a food considered laudable by current dietary guidelines: tomatoes.