Think about the taste of sourdough. That distinctive tang is the work of microbes used in baking the bread—a “starter culture” of wild yeast strains and bacteria that fill the loafs with sour acids. Unlike industrially-made white loaves, which are baked using yeasts that date back just 150 years, the microbes in sourdough cultures have been used since ancient times. That’s why the food journalist Michael Pollan once described sourdough as “the proper way to make bread.”

The acids produced by those microbes have another purported benefit. According to The Guardian, they “slow down the rate at which glucose is released into the blood-stream.” In other words, it has a low glycemic index, making it, as the Globe and Mail advises, “a good choice for anyone managing their blood glucose levels,” such as diabetics.

But according to a team of Israeli scientists led by Eran Segal and Eran Elinav at the Weizmann Institute of Science, this common claim is wrong—or, at least, not universally right. In a small but thorough study, they put volunteers on week-long stints of eating either white bread or sourdough. All the while, they performed a wide battery of blood tests, and they analyzed the community of microbes that live in their guts. “To our great shock and surprise,” says Segal, “we found no significant differences between the two breads on any of the parameters that we collected.”

What’s more, their results showed that people vary substantially in how the two breads affect their blood sugar: Predictably, some people’s blood sugar spiked more dramatically after eating white than sourdough, but unexpectedly, others’ did the reverse. If these results can be confirmed in larger groups of people, it suggests that the common belief that sourdough has a lower glycemic index than white bread is only true for some people; in others, it’s the opposite. “We’re realizing that our nutrition needs to be personalized,” says Segal.

He and Elinav first came to that conclusion in 2015. By comprehensively monitoring the blood sugar, diets, and other traits of 800 volunteers, they showed that people differ in how their blood sugar spikes after eating the same foods. As I reported at the time, the Israeli team built an algorithm to predict those spikes, and used those predictions to develop personalized diets for keeping blood sugar in check. Their menus sometimes included unconventional items like chocolate and ice-cream, and were so counter-intuitive that they baffled both the participants and dietitians involved in the study. But they seemed to work when assessed in a clinical trial.

On the back of that study, the duo decided to focus on bread—a food that’s eaten by billions of people every day and makes up 10 percent of the calories in the diet of an average adult. In a study led by students Tal Korem and David Zeevi, the Israeli team picked two extremes from the bread world. They hired a local baker to prepare artisanal sourdough from whole-grain flour. They also bought mass-manufactured loaves of white bread, made from refined flour and loaded with preservatives.

The team recruited 20 volunteers and asked half to spend a week eating the white bread and another eating the sourdough. The other volunteers did the same in the reverse order. Before and after each bread-filled week, the team took a census of the bacteria in each volunteer’s gut, as well as measured 20 variables including blood pressure, weight, blood sugar, cholesterol, triglycerides, and various hormones. They found that the bread the participants ate had no significant effect on any of these factors. Even the microbiome, which can shift quickly and extensively after a change in diet, was barely affected by the choice of breads.

Susan Roberts, a professor of nutrition at Tufts University, is unimpressed. “Of course you wouldn’t expect to see significant effects in one week, with a small amount of bread, in 20 people,” she says. “That doesn’t mean there isn’t an effect. It just means this study was underpowered.” Susan Jebb, a professor of diet and population health at the University of Oxford, concurs. “It’s a weak study design, and I wouldn’t want to draw any conclusion from this analysis,” she says.

But Segal argues that their study packs extra statistical punch despite including just 20 volunteers. That’s because it’s a crossover trial—each participant ate both breads in turn, and so could be compared against themselves. If the sourdough bread changed any of the clinical variables by just 5 to 10 percent compared to the white bread, the team would have spotted that difference—and they didn’t. They also showed that overall, eating bread did change a variety of factors including cholesterol, iron, and calcium levels. It wasn’t that eating bread did nothing; it was that eating sourdough wasn’t radically different to eating white.

Why? “One possibility is that the two breads induce exactly the same effect,” says Segal. “The more intriguing possibility is that each bread induces different effects in different people.” And that’s what his team found in the new study. The team could tell who was whom based on their microbiome, and they used their data to develop an algorithm that could look at a person’s gut bacteria and accurately predict how they’d respond to different breads.

Again, Roberts is skeptical. “This is wildly over-extrapolating,” she says. “It’s more likely that people liked the two breads more or less, and so were eating more or less before the testing days. That would influence their glycemic response. There might have been a personalized response to the breads, but this study doesn’t tell us whether it was a methodological error or some real individual fingerprint.”

Segal says that his team gave each volunteer a fixed amount of bread to eat during breakfast, calibrated so that each serving had the same amount of carbohydrates. They also logged all their meals on a smartphone app. Based on these recordings, it seems like the volunteers were sticking to their instructions, and weren’t just shunning one bread or another based on their preferences.

That being said, people have a tendency to lie when logging their own meals, and such self-reports are notoriously inaccurate. A single study like this also cannot refute a large body of epidemiological evidence showing that people who eat more whole-grains tend to have lower risks of cancer, heart disease, type 2 diabetes, and other conditions. If you assume that these effects are causal, rather than just correlational, it may be that you need to eat a lot of whole grains, over long periods of time, to experience any benefits. Alternatively, large population-wide studies might mask the fact that only some people would benefit from these foods.

That’s what Elinav and Segal want to know. They’re acting as scientific consultants for a company that was set up on the back of their research, and that analyzes the microbiome to offer personalized dietary advice for controlling blood-sugar levels. They’re also setting up a year-long experiment to see if their recommendations can improve people’s health in the long-term.

This, they say, is the future of nutrition. After all, advising people to eat more sourdough because it officially has a lower glycemic index than white bread would have been misleading for half the volunteers in their experiment. Instead of offering universal dietary recommendations, or categorizing foods as “healthy” or “unhealthy,” the science of nutrition might need a more personal touch.