Short-Term Paleo Diet Improves Glucose Control in Obese Type 2 Diabetes (the Masharani Study)

UCSF is here

UCSF is here

A three-week Paleolithic-style diet improved blood sugars and lipids in obese type 2 diabetics, according to researchers at the University of California—San Francisco. This is the Lynda Frassetto study I’ve been waiting over a year for. The first named author is U. Masharani, so I’ll refer to this work in the future as the Masharani study. Sorry, Lynda.

To understand the impact of this study, you need to know about a blood test called fructosamine, which reflects blood sugar levels over the preceding 2–3 weeks. You may already be familiar with a blood test called hemoglobin A1c: it tells us about blood sugars over the preceding three months. Blood glucose binds to proteins in our blood in a process called glycation. The higher the blood glucose, the more bonding. Glucose bound to hemoglobin molecules is measured in HgbA1c. Glucose bound to plasma proteins (predominantly albumin) is measured as fructosamine. It probably has nothing to do with fructose. Fructosamine is a generic name for plasma ketoamines.

If you’re doing a diabetic diet study over over 2–3 weeks, as in the report at hand, changes in glucose control will mostly be detected in fructosamine rather than HgbA1c levels.

How Was the Research Done?

Twenty-five obese diabetics in the San Francisco Bay area were randomly assigned to either a paleo-style diet or one based on American Diabetes Association (ADA) guidelines. They followed the diets for three weeks, with various measurements taken before and after intervention.

Participants were aged 50-69; you have to guess the sex breakdown. Average body mass index was 34. Over half (63%) were White/European American; there were three each of Asian, African American, and Hispanic ethnicity. They had normal blood pressures and diabetes was well controlled, with hemoglobin A1c’s around 7% and fructosamine levels close to normal. Four subjects were on no diabetes medications; 14 were taking metformin alone, five were on metformin and a sulfonylurea, one was on long-acting insulin and a sulfonylurea. No drug dosages were changed during the study.

Both intervention diets were designed for weight maintenance, i.e., avoidance of weight loss or gain. If participants lost weight, they were instructed to eat more. All food was prepared and provided for the participants. Three meals and three snacks were provided for daily consumption.

Fourteen subjects completed the paleo diet intervention. They ate lean meats, fruits, vegetables, tree nuts, poultry, eggs, canola oil, mayonnaise, and honey. No added salt. No cereal grains, dairy, legumes, or potatoes. Calorie percentages from protein, fat, and carbohydrate were 18%, 27%, and 58%, respectively. Compared to the ADA diet, the paleo diet was significantly lower in saturated fat, calcium, and sodium (under half as much), while higher in potassium (twice as much). These dieters eased into the full paleo diet over the first week, allowing bodies to adjust to higher fiber and potassium consumption. The paleo diet had about 40 grams of fiber, over twice as much as the ADA diet.

[I wonder why they chose canola over other oils.]

Ten subjects completed the ADA diet, which included moderate salt, low-fat dairy, whole grains, rice, bread, legumes, and pasta. Calorie percentages from protein, fat, and carbohydrate were 20%, 29%, and 54%, respectively (very similar to the paleo diet). I don’t have any additional description for you. I assume it included meat, poultry, eggs, and fruit.

Diet compliance was confirmed via urine measurements of sodium, potassium, pH, and calcium.

What Did the Researchers Find?

Both groups on average lost about 2 kg (4-5 lb).

Compared to their baseline values, the paleo group saw reductions in total cholesterol, HDL cholesterol, LDL cholesterol, HgbA1c (down 0.3% absolute reduction), and fructosamine. Fructosamine fell from 294 to 260 micromole/L. [The normal non-diabetic range for fructosamine is 190-270 micromole/L.]

Compared to their baseline values, the ADA diet group saw reductions in HDL cholesterol and HgbA1c (down 0.2% absolute reduction) but no change in fructosamine, total cholesterol, and LDL cholesterol.

Comparing the groups to each other, the difference in fructosamine change was right on the cusp of statistical significance at p = 0.06.

Within each group, insulin resistance trended down, but didn’t reach statistical significance. However, when they looked at the folks who were the most insulin resistant, only the paleo dieters improved their resistance. By the way, insulin resistance was measure via euglycemic hyperinsulinemic clamp instead of the short-cut HOMA-IR method.

Blood pressures didn’t change.

The authors don’t mention hypoglycemia at all, nor alcohol consumption.

They note that some of the paleo dieters complained about the volume of food they had to eat.


I found what I think are a couple misprints. Table 1 has incorrect numbers for the amount of sodium and potassium in the ADA diet. See the text for correct values. Table 2 give fructosamine values in mg/dl; they should be micromoles/L.

Final Thoughts

This particular version of the paleo diet indeed seems to have potential to help control diabetes in obese type 2’s, perhaps even better than an ADA diet, and despite the high carb content. Obviously, it’s a very small study and I’d like to see it tested in a larger population for several months, and in type 1 diabetics. But it will be years, if ever, before we see those research results. Diabetics alive today have to decide what they’ll eat tomorrow.

I wish the researchers had explained why they chose their paleo diet macronutrient breakdown: calorie percentages from protein, fat, and carbohydrate were 18%, 27%, and 58%, respectively. Perhaps they were trying to match the ratios of the ADA diet. But from what I’ve read, the average ancestral paleo diet carbohydrate energy percentage is 30-35%, not close to 60%. My experience is that reducing carb calorie consumption to 30% or less helps even more with glucose control. Reducing carbs that low in this study would have necessitated diabetes drug adjustments and increased the risk of hypoglycemia.

The authors wonder if the high fiber content of the paleo diet drove the lowered glucose levels.

High HDL cholesterol is thought to be protective against coronary artery disease and other types of atherosclerosis. Both diet groups here saw reductions in HDL. That’s something to keep an eye on.

The ADA diet group saw a drop in HgbA1c but not fructosamine. I can’t explain how HgbA1c goes down over three weeks without a change in fructosamine level.

You have to wonder if the paleo diet results would have been more impressive if the test subjects at baseline had been sicker, with poorly controlled blood pressures and HgbA1c’s of 9% or higher. And it sounds like some of these folks would have lost weight if not forced to eat more. The paleo diet is more satiating than some.

The article was well-written and a pleasure to read, in contrast to some I’ve suffered through recently.

Steve Parker, M.D.

Reference: Masharani, U., et al. Metabolic and physiologic effects from consuming a hunter-gatherer (Paleolithic)-type diet in type 2 diabetes. European Journal of Clinical Nutrition, advance online April 1, 2015. doi: 10.1038/ejcn.2015.39

3 responses to “Short-Term Paleo Diet Improves Glucose Control in Obese Type 2 Diabetes (the Masharani Study)

  1. Yes, it really does work. My own practitioner has the book, “The Whole 30” in her office. Things are starting to move in a healthy direction.