June 17, 2009

Metabolic Advantage of Ketogenic Diets Debunked? An Intriguing Study You Will Want to Read

I have been working my way through diet research lately, looking for those very few studies that are truly "scientific" in both design and execution. Because I'm interested in the impact of weight loss on metabolic rate, I did some searching for studies that measured REE--Resting Energy Expenditure, using techniques like the ones described in my post below about Dr. Liebel.

This led me to an intriguing study which compared a ketogenic low carb diet--which closely matched the Atkins diet as described in his 1998 book--to a nonketogenic lower carb diet very similar to Zone or South Beach.

You'll find it here:

Ketogenic low-carbohydrate diets have no metabolic advantage over nonketogenic low-carbohydrate diets Carol S Johnston, Sherrie L Tjonn, Pamela D Swan, Andrea White, Heather Hutchins and Barry Sears [emphasis mine]

Having recently reviewed some very poorly designed, high profile diet match up studies--including the NEJM Diet Bake-off study where the "Atkins" diet group turned out to be eating over 180 grams of carbohydrate a day, I did not expect much from this study. Especially not after noticing who was on the list of authors.

But after reading a few paragraphs, I realized I could not dismiss this study without giving it attention. The authors began by pointing out that other published studies of ketogenic diets had rarely found ketones in the urine of the study subjects who were supposedly eating these diets, a sure sign that they were not actually eating them. The design in this study ensured that the subjects were eating the diets they were studying because for the first six weeks the researchers provided their subjects with all their food.

Not only that, they tested them periodically for a ketone--ß-hydroxybutyrate, to ensure that the ketogenic dieters were in fact eating a ketogenic diet.

The two diets studied both were about 30% protein. The ketogenic diet was about 5% carbohydrate and 60% fat. This is exactly the ratio most of us who have used ketogenic diets successfully to lose weight have found most effective. The other diet (which probably is Zone given the authorship, but which also looks a lot like South Beach) was low fat 30% fat with 9% saturated fat and 40% carbohydrate. Since both diets were calorie controlled at 1500 calories a day, the non-ketogenic diet provided 150 grams of carbs.

Study participants ate nothing but prepared foods provided by the researchers for six weeks, which I felt was long enough for dieters to adapt to the ketogenic diet. After those six weeks were over, dieters were left to their own devices, but the ketogenic dieters were advised by dietitians to keep their carbohydrate intake below 40 grams a day and given counseling to help them achieve this.

Ten people pretty well matched were assigned to each group. (One dropped out of the ketogenic group after developing a heart arrhythmia). This is a small study, but given the expense of providing food and the costs of the detailed lab work that was done, this is understandable.

At the end of the six weeks the ketogenic dieters were in fact generating ketones. They were given a series of tests to track changes in lipids, CRP, liver function, kidney function, HOMA (insulin sensitivity), calcium levels, uric acid, and REE. Only weight and fat-free body mass were measured at ten weeks, so this should be treated as a six week study not a ten week study.

So what did they find that should interest us? A lot!

Where The Diets Matched

Dieters eating both diets lost statistically similar amounts of weight, fat mass, and BMI. Both diets caused a similar rise in HDL and a similar drop in the HOMA measure of insulin resistance. Both diets dropped triglycerides in a statistically similar manner.

Resting energy expenditure rose to a similar extent in both groups--a measure of increased metabolism- and dieters on both diets reported similar levels of hunger. Because the changes in REE did not correlate to the measured amount of ketones or changes in Fat-free mass, a major determinant of REE, the researchers attribute this finding to the very similar high protein content of both diets rather than to changes in carbohydrate intake.

Serum {gamma}-glutamyltransferase concentrations fell in both diet groups during the trial, a measure of liver health. C-reactive protein and 24-h urinary calcium concentrations were not significantly affected by either diet treatment. And though uric acid and creatinine clearance fluctuated, at the end of the six weeks they were lower than baseline in both groups, indicating both diets were not producing gout or damaging kidney function.

Where the Diets Differed

There were significant differences between the two diets in their impact on LDL. LDL rose significantly in some members the ketogenic diet group and it rose in direct proportion to the amount of ketones they were spilling.

The other difference was in self-reported vigor. To quote the study:
weekly fatigue-inertia scores, representing a mood of weariness, inertia, and low energy level, did not differ significantly by diet treatment or time; however, vigor-activity scores, representing a mood of vigorousness, ebullience, and high energy, were significantly higher in NLC dieters than in KLC dieters. These data suggest that, in the context of high-protein diets, small differences (as little as 50–60 g/d) in dietary carbohydrate may affect emotion, mood state, and, potentially, the desire to be physically active.
Though overall weight loss was statistically similar, the graphs presented show an interesting difference between the two diets. Through three weeks free fat mass and weigh loss are identical on the two diets. However, at week 4 the dieters on the low fat/moderate carb diet continue to lose while the ketogenic dieters look like they are beginning to experience the notorious Three Week Stall that is very often reported by people on ketogenic diets who post on diet forums.

At the end of ten weeks, the ketogenic diet group on average remains stalled, the higher carb dieters have continued to lose small amounts of body weight and free fat mass.

Implications for People With Diabetes

This is a small study, and given the financial interest of Dr. Sears, who is listed as an author, in the higher carb diet, one has to maintain some skepticism about the result.

It is also worth noting that this is a calorie controlled study, and that the claim made for the Atkins diet by those most enthusiastic about it is that it allows for weight loss at a much higher caloric intake than other diets. I experienced exactly that phenomenon early on in my first ketogenic diet, though after the "beginners luck" period I did find I had to lower calories to lose weight, as do many people who post on low carb diet boards.

Most obviously, the overweight participants in this study did not have diabetes or prediabetes, though their HOMA results at the beginning of the study were high enough to define them as having significant insulin resistance. Many people with diabetes cannot eat 150 grams of carbohydrate a day without experiencing dangerously high blood sugars.

But I have heard from so many readers of this blog and my Blood Sugar 101 web site who have used the "test, test, test" method to craft a diet that lowers their blood sugars who have ended up eatng a restricted carb diet that is closer to the non-ketogenic diet described here than an Atkins diet. They report achieving A1cs in the 5% range and significant weight loss.

So my thoughts after considering this study are these.

1. The study, like many others, finds no evidence for claims that a high protein diet of about 125 grams of protein a day damages kidneys or that it changes the calcium balance in a way that that damages bone.

2. The study suggests that a high protein intake of around 125 g per day revs up the resting metabolism at least in the first weeks of a high protein diet.

3. The study suggests that people who are concerned about their LDL rising on a low carb diet can continue to lose weight on a higher carb/lower fat diet as long as they monitor their blood sugars closely. From other studies we know that high blood sugar will adversely affect lipids, independent of fat intake, so there will be no benefit in lowering fat intake if we replace fat with carbohydrate to the extent that we lose blood sugar control.

4. People without diabetes who find themselves stalled on a Atkins-style diet might consider experimenting with a lower fat, slightly higher carb, calorie controlled diet and see if that helps break their stall.

5. This study does not address the question of whether ketogenic dieters can lose as much weight as non-ketogenic dieters do while eating a higher calorie intake. Other studies have suggested this may be true.

6. This diet does not address the big problem with all diets studied by researchers, which is what happens after six months when many dieters stall on all diets and compliance becomes a huge problem. Nor does it look at the changes in metabolism--measured carefully--that occur after 10% of starting weight has been lost. Given the expense of a large scale study that would look into these questions, it isn't likely we'll have solid answers any time soon. All large scale diet studies rely on cheap, superficial lab tests and diet and exercise questionnaires rather than careful measurements and controlled feeding because because of the huge expense of the latter approach.

UPDATE: As luck would have it, I found another study of people with Type 2 diabetes given 1500 calorie high protein/Low Fat/40% carb diets and tracked for 12 weeks.

This study tracked changes through 12 weeks and found the high protein made NO difference in the Resting Energy Expenditure when compared to the classic low fat/low protein diet.

Effects of Energy-Restricted Diets Containing Increased Protein on Weight Loss, Resting Energy Expenditure, and the Thermic Effect of Feeding in Type 2 Diabetes. N.D. Luscomb et al. Diabetes Care. April 2002 vol. 25 no. 4 652-657

The researchers supplied some, but not all the foods, which were the usual low fat frankfood.
These included preweighed portions of beef and chicken suitable for six meals per week, shortbread biscuits [i.e. cookies], Canola Lite margarine, and Sunola oil-plus (MeadowLea Foods, Mascot, Australia), Kraft Free cheese (3% fat; Kraft Foods, Melbourne, Australia), skim milk powder, and diet yogurt for the HP diet and sultanas [raisins] and rice for the LP diet
As the subjects were eating more freely in this study, one has no way of knowing what they were eally eating. I know what I would have been eating after a month of diet marge and rubber low fat cheese. . .


ItsTheWooo said...

This study makes the same error a lot of diet studies do. It puts dieters on calorie levels which, truthfully, are far too low to test the REAL difference macronutrients have.

If someone is restricted to 1500 cals, assuming they are getting the bare minimum of nutrients (adequate protein, adequate fats, adequate vitamins/minerals) we aren't going to see much of a difference between 40% carb or 5% carb. 1500 calories is just too friggen low to really test what food does to us, because the body is running on empty and dipping into fat stores anyway.

I think there is some kind of scale between calories and macronutrients... and restricting calories can cover up a lot of the ugliness that the wrong balance of macronutrients can cause, because of how caloric deficit forces blood sugar/insulin low and makes us dip into adipose. Which is pretty much the whole point of restricting carbo and eating more pro/fat in the first place.

To really get a good feel of how important it is to eat a certain way (macronutrient wise) the diet should be higher in calorie than 1500, especially when dealing with healthy overweight subjects (high metabolism). When cals are that low I wouldn't expect much of a difference between 5 and 40% carb anyway assuming protein/fats are adequate.

Trinkwasser said...

Hadn't seen that correlation between ketones and LDL before, that's interesting.

Now it's common for LDL to increase on high fat diets BUT often HDL increases proportionally more. Since this and trigs (and the ratio) are more closely correlated to cardiovascular risk one wonders why it was not measured or reported (I suppose you could back-calculate from the other numbers)

Would be interesting to see the same study redone on Type 2 diabetics where I suspect IR would significantly alter the outcomes especially if they actually reported HDL levels.

Don't those carbophiles purely love to find reasons to eat more carbs?


Stephan Guyenet said...

Hi Jenny,

There was a similar paper published today in the AJCN that lasted 12 months:


I haven't looked at it carefully to see if they checked for ketones.

It was also calorie-restricted. There was a trend toward greater weight loss in the ketogenic diet but it wasn't significant. They also saw increases in LDL and HDL in the ketogenic group.

I agree with ItsTheWoo that weight loss on calorie-restricted diets doesn't necessarily reflect what you see in calorie-replete groups. The body is able to waste calories by heat and that ability varies between people, and is partially controlled by diet in my opinion.

When calories are restricted, a healthy body seems to use them with equal efficiency from fat or carbs.

Jenny said...


That new study is timely. I wish they had included the ketone status of the LC dieters, because so many of the long term studies suggest people aren't really eating LC diets by six months in.

That matches what I see on the diet support groups too.

The 50% drop out rate in this new study is very typical too.

I would love to see someone analyze the physiological dynamics of the long term stalls that so many low carb dieters, even ones who are devoted, run into by six months.

I found some interesting data on fasting and the resulting IR which intrigues me and I think we may be too swift to dismiss the way that ketogenic diets heighten insulin resistance. I don't think it is the same KIND of insulin resistance as is found in Type 2 diabetes, but I think it would be good to examine low carb induced IR in more physiological detail.

Gretchen said...

Saturated fat increases IR, and most people on LC diets eat a lot of saturated fat.

Jenny said...


The extreme rise in IR after fasting in normal people has nothing to do with saturated fat. It appears to be an adaptation to promote the preferential use of carbohydrate to refill glycogen.

Trinkwasser said...

Dunno about that, saturated fat *in the presence of excess carbs* probably does increase IR but strangely it seems to have reduced mine, looking both at trigs/HDL ratio (it put the HDL up at the expense of LDL) and general BG stability over time it seems to work better *for me* than the much lauded "Heart Healthy" Omega 6 oils (I'm not the only one)