Clearing up the carbohydrate confusion

"Interest in the ketogenic diet grows for weight loss and type 2 diabetes,” reported a news article (not a study) in the Journal of the American Medical Association  in January. A ketogenic diet—which is very low in carbohydrates and high in fat—may be getting attention, but the evidence to support it is far from solid.


Weight loss

“Low-carb and low-fat diets equally effective for losing weight: Study.”

That headline, which ran in Newsweek in February, was reporting on DIETFITS, a year-long study on 609 people who were assigned to eat healthy diets with as little fat or carbs as possible.1

It’s not just DIETFITS. The Pounds Lost trial tested diets that were higher or lower in carbs, fat, or protein on 811 people. After two years, weight loss was the same.2

“The Pounds Lost results make a very strong case that it doesn’t matter which you cut—fat, carbs, or protein,” says lead author Frank Sacks, professor of cardiovascular disease prevention at the Harvard T.H. Chan School of Public Health.

But those and similar studies didn’t end the debate over low-carb diets and weight loss.

“What is this keto diet you keep hearing about, and is it healthy?” asked Bon Appétit in August. “With celebrities like Halle Berry and Kourtney Kardashian getting on board, we have a feeling that interest won’t be waning any time soon.”

A ketogenic diet is not just low—it’s very low—in carbs: no more than 20 to 50 grams a day. (The average adult consumes about 245 grams of carbs a day.)

Without carbs to burn for fuel, the body breaks down fat—from food or the fat in your cells—into ketones (like acetone) that can be burned for fuel instead. When it does, you’re in ketosis.

A strict version of the diet has been used since the 1920s to treat children with drug-resistant epilepsy.

ketogenic foods
You can add low-carb veggies and berries to these staples of a ketogenic diet.

“It’s not an easy diet to follow even for a few months,” says Sacks. “Look at all the carbohydrate-containing foods you can’t eat.”

It’s not just carrots, oranges, beans, oatmeal, and other healthy carbs. It’s goodbye burgers, fries, sodas, shakes, pizza, burritos, sandwiches, breads, bagels, buns, pasta, rice, cereals, chips, muffins, cakes, cookies, candy, and just about all junk food.

And it’s hello fats—cream, butter, oils, nuts, avocado—along with fatty meats and cheese. You can eat protein, but not too much, because it boosts insulin levels.

“One of the things that’s driving the ketosis is the low levels of insulin,” says Kevin Hall, senior investigator at the National Institute of Diabetes and Digestive and Kidney Diseases.

“Keeping insulin low drives the liver to take up the fatty acids from fat cells and produce ketones.” That doesn’t happen if insulin goes up.

But those ketones have a downside.

“People don’t feel so well,” says Sacks. “They have side effects like fatigue, bad breath, bloating, and constipation.” Some diet books call it the “keto flu.”

They’re constipated because they can’t eat fiber-rich foods like whole grains, beans, or some fruits and vegetables.

“A fiber supplement is generally recommended because most people don’t get enough fiber to move their bowels efficiently,” says Judith Wylie-Rosett, who heads the division of health promotion and nutrition research at the Albert Einstein College of Medicine in New York.

You can eat some fruit (like berries) and some vegetables in some ketogenic diets. “They tend to focus on low-carb vegetables, like broccoli or cauliflower,” notes Wylie-Rosett.

“If you can’t have french fries, you may find broccoli much more attractive, because there’s only so much steak you can eat. And if you’re trying to eat more fat to get into ketosis, you need to put the fat on something. So oil and garlic on broccoli becomes extremely attractive.”

The long list of forbidden foods can help dieters, at least for a while.

“People often consume fewer calories on a low-carbohydrate diet because they have fewer foods to choose from,” says Wylie-Rosett.

But in studies that last at least a year, any difference in weight loss between low-carb and other diets shrinks or disappears. Of course, you may have heard otherwise.

“A meta-analysis of 13 randomized controlled trials suggested that people on ketogenic diets tend to lose more weight and keep more of it off than people on low-fat diets,” reported a news article in the Journal of the American Medical Association in January.3

What the reporter left out: The difference in weight loss after a year or more was only two pounds. And that was in people who were typically obese.4 “The differences appear to be of little clinical significance,” said the meta-analysis.

But if a new industry-funded study in people with type 2 diabetes is borne out by future research—and that’s a big if—it could be a game changer.


Type 2 diabetes

“Wilma got off insulin, reduced her A1c, lost 45 pounds, and reversed her diabetes using the clinically-proven Virta Treatment,” says the website for Virta Health Corporation, one of several high-tech companies that offer online programs to help people manage their type 2 diabetes remotely.

(Virta charges a one-time $500 initiation fee, plus $370 a month for one year and $199 a month subsequently, though some employers and health plans cover the cost, says Virta.)

Virta’s diet typically allows less than 30 grams of carbs a day, but the company doesn’t disclose the details.

carb foods
Goodbye carbs means goodbye soda, pizza, fries, and more.

“While there are many ways to achieve ketosis, Virta uses its own, proprietary, clinically-tested high fat, moderate protein, low carbohydrate protocol,” says the website.

Participants get video chats with a physician, a personal health coach, and a Starter Kit with lancets so they can take blood samples to test their blood sugar and blood ketone levels at home.

The catch: Virta’s treatment was “clinically proven” in the company’s own non-randomized study.5

Instead of randomly assigning people to either Virta or a control treatment, the researchers—mostly Virta employees who had been offered stock options—compared 262 people with type 2 diabetes who volunteered to try Virta’s program to 87 others who were not interested, so they just kept seeing their own doctors and eating their usual diet.

Not surprisingly, nothing changed in the control group after a year. But among those who got the Virta treatment, the one-year results were impressive:

Hemoglobin A1c (a long-term measure of blood sugar) dropped from an average of 7.6 to 6.3 percent. (An A1c of 6.5 or higher is diabetes. Prediabetes is 5.7 to 6.4.)

Weight dropped an average of 30 pounds (down from an average of 256 pounds).

Diabetes medications dropped. At the outset, 57 percent of the Virta group took drugs other than metformin. After a year, it was only 30 percent.

“The results are very promising,” says Hall. But questions remain:

Selection bias? “Because the trial wasn’t randomized, you can only draw limited conclusions from the data,” says Hall. “You have to worry about selection bias, because the folks who got the Virta treatment were highly motivated to stick to the program.”

So you can’t chalk up their greater success only to the treatment.

What’s more, adds Hall, “we don’t know whether these results will translate to a wider group of people who are less motivated.”

Performance bias? When researchers give one group more attention than others, that can create a bias.

“These motivated, self-selected folks underwent intensive coaching and monitoring as part of this remote delivery system,” says Hall. The control group got no intervention.

“What’s unclear is whether an equally intensive intervention with a different diet could have led to similar effects,” notes Hall.

Weight loss or diet? Also unclear is how much weight loss helped the participants manage their diabetes.

“It would be interesting to see another group that lost the same weight on a different diet, so you could distinguish the effects of the weight loss versus the type of diet,” says Hall.

“If you went on a ketogenic diet and didn’t lose weight, it’s not clear how it would affect insulin sensitivity and beta-cell function.” (Insulin is secreted by the beta cells in the pancreas.)

Of course, the Virta group did lose weight—an average of 30 pounds after a year. That beats the usual 12 or so pounds in weight loss studies that last a year. The ketone tests could explain why.

“The tests allowed people to monitor whether they went off the diet,” says Hall. “People can’t tell precisely how many calories they eat each day, but they can tell whether they ate too many carbs or too much protein to knock them out of ketosis. That’s a real advantage.”

Safety? What with the cream, butter, meat, and cheese in the ketogenic diet, it’s no surprise that LDL (“bad”) cholesterol rose in the Virta group (from an average of 103 to 113 mg/dL).6

Odds are, LDL didn’t exceed what used to be called the “near optimal” range (100 to 129), because half of the participants were taking statins, and most lost weight over the year.

However, triglyceride (fat) levels after meals—which weren’t measured—could have risen.

“The high-fat, low-carb diet improves fasting triglyceride levels,” says Hall.

“But it’s likely to increase post-meal triglyceride levels, which predict cardiovascular risk more than fasting triglycerides. So we just don’t know what the net effect on risk is.”

virta ad
The Virta Treatment, which offers intensive online coaching and lets people check their ketones and blood sugar daily, has promise. But it has only been “clinically proven” in an industryfunded non-randomized study run largely by Virta staff.

The long haul

The catch with any diet is whether you can stay on it.

That’s why Hall is encouraged by the Virta results.

“The frequent interaction with participants and the ability to track their adherence are wonderful factors that may help people stick to the diet over time,” he says.

Not many studies have kept people on ketogenic diets for more than six months. And most studies that test lower-carb diets—ketogenic or not—haven’t had much long-term success.

“People often get off to a good start with a low-carbohydrate diet,” says Wylie-Rosett. “The question is, will they stick with it.”

Her study randomly assigned 105 people with type 2 diabetes to either a low-fat or low-carb (though not low enough to be ketogenic) diet. The low-carb group lost more weight after the first three months, but by one year there was no difference in weight or A1c.7 Other studies find similar results.8

“People who are big advocates for low-carb diets make the results sound impressive,” says Wylie-Rosett. “But they tend to have a study without a control group or a study of only completers. If you ignore the dropouts, the low-carb diets tend to look better.”

However, she adds, “the American Diabetes Association now recognizes that people can lower their blood sugar with a variety of approaches, so there’s no one diet for people with diabetes.”9

And patients may be more likely to stick to a diet if they choose it.

“The person with diabetes makes the decision and is guided by the healthcare team,” says Wylie-Rosett. “The patient is in the driver’s seat.”

However, she adds, people with type 2 diabetes shouldn’t cut way back on carbs without a doctor’s supervision.

“If you dramatically reduce carbs on your own and don’t adjust medications that raise blood insulin, you could have a real crisis” if your blood sugar drops too low.


1JAMA 319: 667, 2018.
2N. Engl. J. Med. 360: 859, 2009.
3JAMA 319: 215, 2018.
4Br. J. Nutr. 110: 1178, 2013.
5Diabetes Ther. 9: 583, 2018.
6Cardiovasc. Diabetol. 17: 56, 2018.
7Diabetes Care 32: 1147, 2009.
8Am. J. Clin. Nutr. 108: 1, 2018.
9Diabetes Care 41(Suppl 1): S38, 2018.


Got prediabetes?

Want to avoid moving from prediabetes to type 2 diabetes? Few studies have looked at whether ketogenic diets can help.

But one thing is clear: “Weight loss is the strongest predictor of whether you convert to diabetes,” says Judith Wylie-Rosett of the Albert Einstein College of Medicine.

omada ad

That’s based on the Diabetes Prevention Program (DPP), a randomized clinical trial on 3,234 people with what we now call prediabetes. After three years, diet plus exercise lowered the risk of diabetes by 58 percent. Metformin (a drug that lowers blood sugar) cut the risk by 31 percent.1

“The fact that lifestyle was twice as effective as medication was a shock even to the investigators,” says Wylie-Rosett.

Thanks to the DPP, Medicare pays for older people with prediabetes to participate in approved DPP-like programs.

And some insurance plans cover online digital coaching programs like Omadahealth.com. (Note: Omada’s trials are non-randomized and are funded and run by the company.)

Though the DPP ended in 2002, Wylie-Rosett and other researchers are still tracking the participants.

“The improvement in A1c is largely related to weight loss,” she notes. “The composition of the diet—carbs, fat, protein—doesn’t seem to make any difference.”

Want to try a DPP program? The Centers for Disease Control and Prevention lists groups that meet DPP standards, though Medicare hasn’t approved them all (see nccd.cdc.gov/DDT_DPRP/Registry.aspx).

1N. Engl. J. Med. 346: 393, 2002.


Can a ketogenic diet make new cancer drugs work better?

“Cancer just loves refined sugar,” warned an article on “The 20 Most Carcinogenic Foods” on msn.com in June. “Why? Because this foodstuff helps cancer cells multiply.”

Whoa. It’s a huge leap to go from saying that blood sugar fuels cancer cells to calling sugar a carcinogen.

But a recent study in mice does suggest that very-low-carb diets may make some cancer-fighting drugs more effective.1

mouse
A ketogenic diet helps a new class of cancer drugs work, at least in mice. But the diet alone made some cancers worse.

“A series of cancer drugs are being developed to inhibit an enzyme called PI3-kinase,” explains Benjamin Hopkins, a postdoctoral associate at the Weill Cornell Medical College.

Mutations in the PI3K pathway are found in more than a third of all tumors in the breast, uterus, prostate, colon, and other organs.2 So drugs that turn off the enzyme hold promise.

“People were excited about the PI3K inhibitors because PI3K is a fundamentally important pathway in that tumors need nutrients to grow,” notes Hopkins.

PI3-kinase is activated by insulin. “Researchers thought that by turning off insulin signaling, you’re telling tumor cells that they don’t have the resources to continue to divide,” says Hopkins.

But trials testing the drugs have been disappointing. So far, the FDA has approved only one, Aliqopa, and only for treating relapsed follicular lymphoma.

“For almost a decade now, clinical trials on PI3K inhibitors have never quite had the impact on cancer that clinicians thought they should,” notes Hopkins.

So he and his colleagues, including lab director Lewis Cantley and author and oncologist Siddhartha Mukherjee, tried to figure out why.

“We and others observed that when patients received these compounds, their blood sugar spiked,” says Hopkins.

That’s because when the drugs blocked insulin, which allows blood sugar to enter cells, the cells ran short on fuel. That prompted the liver to break down its carbohydrate (glycogen) stores, sending a burst of sugar into the bloodstream. But the rise in blood sugar didn’t last.

“If your glucose goes high, insulin is released by your pancreas to lower it,” says Hopkins. “That’s a sign that your body is reactivating the very insulin-signaling pathway that the drugs are targeting. It’s undercutting the drugs before they start working.”

But that didn’t happen when Hopkins put mice on a ketogenic diet.

“If you’re eating very, very low levels of carbohydrates, your body no longer has glycogen stores to release when blood sugar is too low,” he explains. “There’s no sugar to release so you don’t undercut the drugs.”

Of course, it’s too early to know if the drugs would work in people who eat a very-low-carb diet. Trials are under way.

But these promising results don’t show that a very-low-carb diet alone can fight cancer.

“There are tumor types where the ketogenic diet alone does seem to help, there are tumors that don’t seem to care, and there are even some tumors that grow faster on a ketogenic diet,” says Hopkins.

“So the study definitely does not show that a ketogenic diet is a good idea for all cancer patients. And it absolutely does not show that sugar causes cancer.” However, “it doesn’t argue in favor of sugar, either.”

1Nature 2018. doi:10.1038/s41586-018-0343-4.
2JAMA Oncol. 2: 1565, 2016.


Photos: Danny Moloshok/Reuters (top), ricka_kinamoto/stock.adobe.com (foods in ketogenic diet), Nitr/stock.adobe.com (fast food collage), Vasiliy/stock.adobe.com (mouse).