New Study: Is a Calorie a Calorie?

A new study in JAMA led by Dr. Cara B. Ebbeling and colleagues purports to challenge the idea that all calories are equally fattening (1).  Let's have a look.  When thinking about the role of calorie intake in body fatness, there are basically three camps:

1.    Calories don’t matter at all, only diet composition matters.
2.    Calories are the only thing that matters, and diet composition is irrelevant.
3.    Calories matter, but diet composition may also play a role.

The first one is an odd position that is not very well populated.  The second one has a lot of adherents in the research world, and there’s enough evidence to make a good case for it.  It’s represented by the phrase ‘a calorie is a calorie’, i.e. all calories are equally fattening.  #1 and #2 are both extreme positions, and as such they get a lot of attention.  But the third group, although less vocal, may be closest to the truth. 
A Little Background on Food Intake, Diet Composition and Energy Expenditure

The energy content of the human body is determined by the amount of energy coming in, minus the energy exiting—simple on paper.  Since body fat is by far the main energy storage site in the body, it follows that the balance between energy intake and energy expenditure determines fat mass, and this has been experimentally confirmed many times.  But here’s the catch—energy expenditure is not always constant.  For example, we know that it increases as a person gains weight, and decreases when a person loses weight (2).  The decrease in energy expenditure with weight loss is caused by two things, 1) a smaller body requires fewer calories for maintenance, 2) a decrease in leptin, which decreases the metabolic rate by acting in the brain (3).

Some people have suggested that the type of food we eat, not just the amount, influences energy expenditure, and in particular that this is related to the diet's carbohydrate content.  In people who are not trying to lose weight (4, 5), or who are being overfed (6, 7), the carbohydrate:fat ratio in the diet has little or no detectable impact on energy expenditure, and if anything it favors carbohydrate, but could this be different during fat loss in people who start off overweight?  This idea has been called the ‘metabolic advantage’, most notably attributed to the low-carbohydrate diet.  The idea here is that you can lose fat eating the same number of calories if carbohydrate is kept low.

I’ve never really weighed in on this because it’s a topic of heated debate, and in any case it’s a fairly academic question.  Why is it academic?  Because previous weight loss studies have shown that if a metabolic advantage exists at all, it’s quite small, because the effect is undetectable in most studies (8, 9, 10).  People who are not associated with the low-carbohydrate community tend to conclude that there's no metabolic advantage when they review the literature (11), although I haven't reviewed it closely myself.  It’s clear that where fat loss is concerned, calorie intake is much more important than the amount of fat or carbohydrate in the diet.  What previous studies have suggested is that low-carbohydrate diets suppress appetite — often resulting in lower calorie intake (12, 13).  The reason for this remains a topic of speculation.  

That being said, I’m actually quite open to the idea that food quality in addition to quantity can influence body fatness, and I would encourage people to think outside the macronutrient box: there are probably many different dietary factors that can have such an effect.  Although this idea hasn’t received much support in the human literature so far, there’s quite a bit of evidence for it in the animal literature.  For example, when we want to produce obesity in rodents we typically use diets that are composed of refined ingredients, high in fat (40-60% of calories), contain some sugar (~10% of calories), and are highly palatable.  These diets are extremely fattening in susceptible strains, but their fattening ability is only partially dependent on increased calorie intake.  If you restrict an animal's energy intake so that it’s the same as rodents on a non-fattening unrefined diet (called ‘pair feeding’), they still gain most of the fat that they would have if you hadn’t restricted energy intake at all (14, 15)!  This suggests that these diets make their bodies ‘want’ to be fat, and they will accomplish this goal by increasing calorie intake and/or by decreasing energy expenditure.  This is related in large part to changes that occur in the brains of these animals (16). 

The Study

Enter the recent study by Dr. David Ludwig’s group (1).  This is a really fascinating, some might say groundbreaking, study.  Although we know that differences in dietary carbohydrate and fat content have little or no impact on energy expenditure during overfeeding, in weight-stable people, or during weight loss, we don’t know much about how these impact energy expenditure during weight maintenance after weight loss.  Weight maintenance is critical because it’s where most people’s weight loss efforts fail. 

Contrary to some of the claims I'm sure are being made about this study, it wasn’t actually designed to tell whether fat and carbohydrate per se influence energy expenditure.  What they compared were three distinct dietary patterns that differed in carbohydrate, fat, protein, and other aspects of diet composition, although carbohydrate content was certainly a major difference between groups. 

In this study, they started with overweight and obese volunteers who had lost an average of 14 percent of body weight using a low-calorie diet.  Then they placed them on three different diets, which were precisely controlled by the investigators:

1. A low-fat (LF) diet  “designed to reflect conventional recommendations to reduce dietary fat, emphasize whole grain products, and include a variety of vegetables and fruits”.  60-20-20 carbohydrate-fat-protein.  I think these people were eating a lot of breads, pastas, etc. 
2. A low glycemic index (LGI) diet diet designed to “achieve a moderate glycemic load by replacing some grain products and starchy vegetables with sources of healthful fat and low–glycemic index vegetables, legumes, and fruits”.  40-40-20 carbohydrate-fat-protein.  Low glycemic index carbohydrates are those that increase blood sugar less per unit carbohydrate eaten, such as beans, oatmeal, fruit, and sweet potatoes.
3. A very low carbohydrate (VLC) diet that was “modeled on the Atkins Diet and had a low glycemic load due to more severe restriction of carbohydrate”.  10-60-30 carbohydrate-fat-protein.  This is mostly meat, vegetables, eggs, nuts, and added fats.  They also took a fiber supplement as recommended by Atkins.

After four weeks, they measured energy expenditure, both at rest and over the course of 24 hours.  It’s worth mentioning that they used state of the art techniques to measure these variables.  What they found is that energy expenditure at rest, and over 24 hours, was highest in the VLC group, intermediate in the LGI, and lowest in the LF group—and the differences were substantial, up to 308 calories a day.  A notable limitation of this study is that each diet period only lasted 4 weeks—we don’t know what would have happened to these people after 6 months or a year.

Does this support the idea that there is a ‘metabolic advantage’ to low-carbohydrate diets?  Well, sort of.  It doesn’t change the previous findings that the carbohydrate:fat ratio has little or no impact on energy expenditure during overfeeding, in weight stable people, or during weight loss, but it does suggest that a VLC dietary pattern has a metabolic advantage over a LF diet specifically in the context of weight maintenance after weight loss.  It also suggests that a LGI diet has a smaller but still meaningful metabolic advantage in this setting, and that a LF diet are not very effective in this regard.  It also opens a whole new can of worms for the research world, investigating the effects of diet quality on energy expenditure. 

So should everyone eat a VLC diet for weight loss maintenance then?  Not so fast.  The VLC diet group experienced some troubling hormonal changes that seem to be pretty common with this kind of diet.  There were three main negative changes.  The first and perhaps most troubling was an increase in cortisol, a stress hormone that may contribute to serious health problems over the long term (17).  The second was a decrease in thyroid hormone, which is something that has been observed repeatedly with this kind of diet.  Many of the negative effects that some people develop on long-term VLC diets (constipation, lethargy, poor sleep, hair loss, irritability) could be related to low thyroid function.  This doesn't seem to be as much of an issue with moderate carbohydrate restriction.  The third was an increase in C-reactive protein (CRP), a marker of inflammation and heart attack risk.  Although the paper and media reports make a big deal out of the increase in CRP, it looks quite small to me—I’m not convinced it’s biologically significant. 

The low-fat diet came out looking pretty poor as well.  Energy expenditure was lowest on this diet, and estimated insulin sensitivity was also lowest (although the caveat here is that they didn't directly measure insulin sensitivity).  This is interesting because as they remarked in the paper, this is the dietary advice most people will receive from nutrition authorities.  This adds to the evidence that eating a bunch of whole grain breads and pastas is probably not a great strategy for weight loss maintenance.

Overall, the LGI diet came out looking the best overall.  It’s the least restrictive of the three, conferred a pretty good ‘metabolic advantage’, and was not associated with harmful hormonal changes.  I think this kind of diet would be a reasonable choice for weight maintenance following fat loss.  I’ve been skeptical of the glycemic index concept in the past, and I continue to be, because most controlled studies have shown that in isolation it has little or no impact on body weight or insulin sensitivity (18, 19).  The LGI diet in this study did indeed have a low glycemic index, but since it differed from the other diets in many other ways, we can’t know whether the glycemic index per se was a relevant factor. 

In any case, a diet that focuses on beans/lentils, meats, nuts, fruits, and vegetables at the expense of grains (particularly flours) may be a good choice for weight loss maintenance, and I think this squares well with research coming from other angles.  That being said, there are some major caveats to this study that must be kept in mind:

1. It was only four weeks long.  We don't know if this difference would have continued in the long run.
2. The study was not designed to measure long-term changes in fat mass, which is what we're ultimately interested in.
3. This was done under calorie-controlled conditions, not under free-living conditions as most people would apply these diets.  Under free-living conditions, the main determinant of long-term fat loss seems to be how well you stick to a diet, not which diet you choose (20, 21, 22).
4. There was a lot of individual variability in this study, confirming once again that everyone is different.  There is no reason to stick to a diet if it's not working for you, just because a study says it works well for the "average person".  Kudos to Drs. Ebbeling and Ludwig for reporting individual data.

Despite these important caveats, I think the results are fascinating and I look forward to more studies on this phenomenon.  If I have time, I may discuss the potential mechanisms for this effect in a future post.