When Calories Don’t Count
When nutrition professor Susan Roberts was designing a diet program, she tested the recommendations on herself. If she could maintain her normal weight on the meal plans, then she felt confident that they would work for her readers as well. The plans worked as intended, but only as long as she cooked at home. When she ate purchased entrees that supposedly contained the same number of calories, her weight started creeping up. To figure out why, she measured the energy in various low-calorie frozen dinners, which turned out to average 8 percent more calories than the labels claimed. Those averages are within the limits set by the FDA, which requires that grocery calorie labels be accurate only to within 20 percent.
This inaccuracy is a problem if weight loss is just math, as we often hear. The advice is all over magazines and the Internet: cut 100 calories a day to lose 10 pounds a year. The logic is that 3,500 calories is the amount of energy in one pound of fat, so 100 calories times 365 days equals 10.4 pounds. There are a bunch of problems with this reasoning, which relies on a naive view of weight regulation. To name a few, once people get below their defended range, their bodies will burn less energy due to metabolic suppression, enough to overwhelm the missing hundred calories. In addition, as people lose weight, fewer calories are required to support their lighter bodies, so the calorie deficit goes to zero over time. A dynamic model of how energy imbalance affects the body (without taking metabolic suppression into account) suggests that it would take more like three years, on average, to lose 10 pounds after cutting 100 calories—and the pounds will stay off only if the calorie reduction is also permanent. In any case, most people probably don’t care about the average, but about what will happen to them personally. Individual variations in physiology, based in part on each person’s population of gut bacteria, lead to huge differences in weight loss from the same calorie deficit. Finally, almost no one can count accurately enough to figure out how to cut 100 calories, let alone maintain that accuracy for years. The idea of counting calories appeals to our inner nerd because it feels rational and quantifiable, but getting accurate numbers turns out to be much harder than it looks.
Not Just Math
The information on grocery store labels looks precise—152 calories per six-ounce serving sounds like an exact measurement—but provides only an approximation of how much energy a particular person gets from eating that food. The reasons are found at every step of the process, from measuring calories to cooking food to digesting it. Considering that a 5 percent error in energy balance over time would be enough to cause noticeable weight gain, such uncertainty is an important consideration in deciding whether to manage food intake by counting calories. At the minimum, people would have to figure out their own energy needs by experimentation, based on the characteristics of their own bodies, which foods they habitually eat, and how they prepare those foods.
Obviously no one can deny the first law of thermodynamics: energy entering the body has to go somewhere. It can’t just disappear. But most “calories in/calories out” arguments about weight loss ignore vast individual differences in how bodies handle energy. They make as much sense as saying, “Look, a Humvee is basically a Prius. Just put gas in them, and they go.” The people buying the gas might be inclined to disagree. Two people’s response to the same energy deficit can be just as different.
Most people realize that food energy can fuel physical activity or be stored as fat, but the possibilities don’t end there. Food can also be excreted from the body because it has failed to be digested or absorbed. Energy can be used to grow or repair body tissues, including in pregnancy and breast-feeding. A lot of it goes to basal metabolism, the cost of keeping a body idling when it’s not doing work. That energy expenditure depends on body size and the proportion of muscle versus fat, along with an individual variation of about 250 calories per day among people of the same size. Contrary to popular belief, it is not affected by how many meals people eat or when they eat them. Finally, energy is wasted as heat during digestion, especially of protein, in what is called the thermal effect of food, which also differs across individuals. That reaction can be seen in the need to take off a sweater while eating a big meal. As weight is lost, all three components of energy expenditure—basal metabolism, the calories burned by a given amount of physical activity (because it costs more to move a larger body), and the thermal effect of food—decline until a new equilibrium is reached, at which point people stop losing weight unless they cut more calories.
The Trouble with Labels
The problems with calorie counting begin before the first bite. Food labels can be highly inaccurate, as filmmaker Casey Neistat discovered when he analyzed some of his favorite meals. A tofu sandwich advertised as healthy turned out to have almost twice its labeled calories when it was tested in the laboratory. Food producers have an incentive to claim lower calorie counts to get health-conscious consumers to buy more of their products. When Susan Roberts and her research group went to chain restaurants and purchased items labeled as low in calories (under five hundred), the portions averaged 18 percent more calories than advertised, up to twice as many. The free side dishes often had as many calories as the entree they accompanied, but because each item was listed separately, a customer would have to look carefully to notice that fact. The take-home message seems clear: a person could gain a lot of weight relying on calorie labels.
Getting calorie counts right is also hard for meals cooked at home. To determine how many calories a food contains, the most accurate method uses a bomb calorimeter. That’s pretty much what it sounds like, a sealed chamber in which electricity burns the food to release its energy as heat, measured by a temperature increase in a surrounding water bath. The idea that a pound of fat contains 3,500 calories comes from bomb calorimetry measurements. To label every food in the grocery store, though, that procedure would be time-consuming and expensive.
Instead the label numbers come from a century-old system developed by Wilbur Atwater. He fed people diets with differing proportions of carbohydrates, fats, and proteins and used a bomb calorimeter to measure the calories in the food they ate and the waste they excreted. Then he calculated how much energy his subjects were getting from carbohydrates (four calories per gram), fats (nine), protein (four), and alcohol (seven). Aside from a few tweaks, for instance to account for high-fiber foods, scientists still use his formula today.
Not every gram of carbohydrate is created equal, however. Nor protein, nor fat. At every step of extracting the energy from foods, some is lost. How much? Well, it depends. Chemical reactions in the gut break food down into sugars, fatty acids, and amino acids that can be absorbed into the blood. The efficiency of these reactions depends on many factors, which may be specific to the food or the individual or both.
Break It Down Now
To digest food, enzymes in the gut must contact it. The smaller the pieces that end up in the stomach, the more thoroughly that food will be broken down. Chewing helps with that, and so does chopping. A pureed carrot yields more calories than a whole one, for instance. These differences can be large. Almonds have a calculated value of 170 calories, but people extract only 130 of those calories, probably because some of the fat is not available for digestion. Similarly, people who eat peanut butter absorb more fat calories than those who eat whole roasted peanuts. The Atwater numbers in general are close to the values of processed foods, but can be substantially off for whole foods.
Cooking releases extra energy from food by breaking down cell walls and other structures that lock energy away from digestion. That may explain why cooking is used in every human culture that’s ever been studied. Richard Wrangham at Harvard believes that when our ancestors figured out how to cook, they got so much extra energy that they were able to support large brains for the first time in evolutionary history. For comparison, researchers calculate that if we ate the same diet of raw plants as other primates, people would need to eat for 9.3 hours per day just to get enough calories. People who eat nothing but raw foods are indeed a lot thinner than vegetarians who cook. Half the women who eat only raw food have such low body fat that they don’t menstruate, suggesting that their bodies aren’t getting enough energy.
The effect of cooking is largest for starch. A raw potato, for example, has half as many usable calories as a cooked whole potato, and a handful more calories can be extracted from mashed potatoes. Cooking wheat makes a third more calories available. Raw meat also has fewer available calories than cooked meat. To estimate the difference, researchers measured oxygen consumption in Burmese pythons. Cooking reduced their cost of digestion by 12.7 percent, grinding the meat reduced it by 12.4 percent, and the combination by 23.4 percent. In people, cooking egg whites, compared with eating them raw, increases the amount of digestible protein by 50 percent. Similarly, eating peanuts roasted instead of raw allows mice to harvest more of their fat content.
Another reason we don’t get all the energy out of our food is because digestion itself uses energy. Fat is the easiest macronutrient to digest, with losses of only 3 to 5 percent, meaning that when people eat a hundred calories of fat, they get about ninety-six usable calories. Carbohydrate digestion costs 5 to 15 percent of the food’s energy, so that a hundred calories of food yields about ninety calories. Proteins are the hardest to digest, costing 20 to 35 percent of their energy, so a person might get as few as sixty-five calories from eating a hundred calories of protein. People who believe they’re not cutting calories on their low-carb diet might just be mistaken, given how hard it is to digest meat.
If you’re starving and need to extract every possible calorie from food, make sure it’s low in fiber and as processed as possible. In long-term studies, weight gain is best predicted by eating French fries, potato chips, and sugared soda. The time and energy saved on food preparation may be lost later on to illness, though. On the other hand, to get the maximum possible health benefits and feelings of fullness from eating, the evidence suggests concentrating on whole foods that are cooked at home. In the same long-term studies, avoiding weight gain is predicted by eating whole grains, vegetables, fruits, and nuts.
Your Bacterial Passengers
The last factor that contributes to individual variability in how many calories a person extracts from food is the gut bacteria population. The human gut contains tens of trillions of bacteria; many more of our cells are bacterial than human. We support these hitchhikers because, like other animals, we need gut bacteria to digest our own food. These resident bugs come from three main groups, called Bacteroidetes, Firmicutes, and Actinobacteria, which break down parts of our food that would otherwise be wasted. The Bacteroidetes even synthesize vitamin K, a necessary nutrient that we can’t make for ourselves. These bacteria don’t just nibble around the edges of our diets, either. Their contribution is large. Without them, we would have a much harder time meeting our nutritional needs. Gut bacteria are also important for the normal development of the immune system, as they influence the initial population of T cells, the white blood cells that respond to cellular damage or infections.
Researchers can produce “germ-free mice” without any gut bacteria by birthing them through Cesarean section (newborns get a lot of their initial bacteria colony from the mother’s vagina) and raising them in an isolation chamber. Germ-free mice eat 29 percent more food but have 42 percent less body fat than similar mice that were raised normally. When fed a tasty high-fat, high-carbohydrate diet, the germ-free mice eat as much as normal mice but gain much less weight. Eating more and weighing less might be a dieter’s dream, but it would have been a serious disadvantage when food was hard to get, throughout most of human history. Bacteria release a lot more energy from food than our own digestive enzymes can extract. If germ-free mice receive transplanted gut bacteria from a normal mouse (yes, that’s a polite way of saying a poop transplant), they start eating 27 percent less food, the same as a normal mouse, while increasing their body fat by 60 percent in two weeks.
Although not having any bacteria at all is an extreme experimental manipulation, the gut microbiome is a vibrant living community that can be transformed through food choice, illness, or antibiotic use. For example, researchers put people on a diet of meat, eggs, and cheese, giving them 70 percent of their calories from fat and the rest from protein, with almost no fiber. Within a day, they had more Bacteroidetes, which are tolerant of bile, a secretion involved in fat digestion. At the same time, the Firmicutes, which digest plant starches, went down. A whole-grain, vegetarian diet had the opposite effects. These changes are similar to differences between meat-eating and plant-eating species. When the participants resumed their regular eating habits, their gut bacteria returned to normal in two days.
Bacteria Influence Weight
Gut bacteria have wide-ranging effects on energy pathways throughout the body, and the relative abundance of different species varies with body weight. Mice that are obese because they cannot make leptin have 50 percent fewer Bacteroidetes and 50 percent more Firmicutes compared with their thin counterparts. A similar pattern is seen in mice that became obese on a high-fat/high-sugar diet, as well as in obese people. When germ-free mice are transplanted with gut bacteria from an obese mouse or an obese person, they gain about twice as much weight as when the bacteria come from a lean mouse. These differences are reversed when animals lose weight on either low-fat or low-carbohydrate diets, at least while the weight loss lasts.
The relative prevalence of gut bacteria can also change over generations, across the population. The species Helicobacter pylori was the most common resident of the stomach since ancient times, making up more than half of all stomach bacteria, but it has steadily declined in people in developed countries around the world over the past century. In 1900, more than 80 percent of people carried this species, but by 2000 only 6 percent of children in the United States, Germany, and Sweden had it. The antibiotics often used to treat childhood ear infections wipe out this species in 20 to 50 percent of people who receive them, which explains some of the decline. Because H. pylori produces a lot of acid, its loss is a mixed blessing, increasing esophageal reflux and associated disorders but decreasing stomach cancers and ulcers. The disappearance of this species also changes the production of hormones involved in regulating appetite, reducing leptin, at least in the stomach, and increasing ghrelin in the blood, which may help to set the brain’s defended range in early life. Successful eradication of H. pylori by antibiotic treatment in adults is often associated with weight gain, relative to cases in which the same antibiotics fail to kill their intended target.
Some types of gut bacteria are influenced by their host person’s genes, while others are not. The prevalence of bacteria from the Bacteroidetes family responds to diet and does not depend on the person’s genetic heritage. In contrast, a twin study found the highest heritability in a family called Christensenellaceae, which is not easily modified by diet. When transplanted to germ-free mice, a member of this family caused weight loss and reduced body fat. In another study, germ-free mice were given gut bacteria from human twin pairs with one lean and one obese twin. When the mice were housed separately, they became lean or obese, just like the twin whose bacteria they had received. When the pair was housed together, the lean twin’s bacteria became established in both mice, and the obese mouse lost weight, but only on a diet low in fat and high in vegetables. If the diet was high in fat and low in vegetables, the obese mouse stayed obese.
The bacteria types common in obesity lead to weight gain because they help us get more energy from food. They are better at breaking down complex carbohydrates like cellulose, xylan, and pectin. For people who eat a lot of fruits and vegetables, that improvement in digestion may provide an extra 140 to 180 calories per day of usable energy. Gut bacteria influence the absorption of nutrients in a variety of ways, for instance by slowing the movement of food through the gut to allow more complete extraction of nutrients, and by increasing the production of an enzyme that moves glucose from the small intestine into the blood. They also suppress an enzyme called lipoprotein lipase that normally limits the ability of fat cells to take up fatty acids and triglycerides from the blood, resulting in more fat storage. This mechanism seems to be particularly important for obesity, as germ-free mice gain only 10 percent of their weight instead of 60 percent if a regulator of this pathway is blocked after they receive the gut bacteria transplant. Gut bacteria also reduce the use of fat for energy in the liver and muscles.
Weight-loss surgery may take advantage of these pathways to change the defended range of the brain’s energy-balance system. Gut bacteria modify signals of hunger and fullness, so they might act on the brain’s energy-regulation system. Weight-loss surgery causes changes in gut bacteria in both mice and people, similar to those seen during diet-induced weight loss, and also reduces diabetes symptoms. Both these effects of bariatric surgery occur before weight loss and are independent of diet. When bacteria from mice that have had surgery are transplanted into germ-free mice, they lose weight, even though they are very thin already.
In one person, a gut bacteria transplant may have caused obesity. Such transplants are an effective way to cure infections by Clostridium difficile, which can take over when the other gut bacteria have been killed by antibiotics, causing almost nonstop diarrhea. After multiple attempts to eradicate her C. difficile infection with antibiotics failed, a thirty-two-year-old woman agreed to try a fecal transplant, which cured her infection. At that time, she weighed 136 pounds and had never been obese. Her daughter, the donor, weighed 140 pounds before the transplant, but her weight shot up to 170 pounds soon afterward. The mother also gained weight, despite repeated attempts to control it with diet and exercise, reaching 177 pounds by three years after the transplant. Given that both of them gained weight at the same time, their shared gut bacteria seems to be the most likely explanation.
Gut Bacteria for Weight Loss?
Although researchers understand some of the details of how gut bacteria influence weight, they have not yet figured out how to apply that knowledge to make people lose weight. One possibility is a bacterial species called Akkermansia muciniphila, which becomes much more prevalent after weight-loss surgery and prevents mice from gaining weight on a high-fat diet. A clinical trial of this species is under way in people, but it will be a while before microbiome therapy for obesity is ready for action.
Researchers are also testing a drug that mimics the effects of weight-loss surgery on gut bacteria, at least in mice. After surgery, human patients show a substantial increase in bile acids circulating in the blood. In addition to their main job of helping to digest fats, bile acids released during meals activate a receptor called FXR, which regulates fat and sugar metabolism. Both directly and through FXR, bile acids also influence the abundance and composition of gut microbes. In mice that are genetically modified to lack this receptor, weight-loss surgery causes only a transient decrease in weight, which is fully regained within five weeks. (This study provides more evidence that the effect of surgery is not due to limiting meal volume, as these animals were able to gain weight despite their surgically reduced stomach size.) Gut bacteria also react differently to surgery in mice lacking FXR. They do not show the usual reduced abundance of Bacteroidetes or the increased abundance of a bacterial group called Roseburia, which is low in people with diabetes.
A drug called fexaramine activates the FXR receptor, mimicking the effects of bile acids. When mice swallow this drug, it acts in the intestine to reduce weight gain in response to a high-fat diet, without decreasing the amount that animals eat. Instead the drug reduces inflammation and production of glucose by the liver. It also increases insulin sensitivity and basal metabolic rate, as the mice burn enough extra sugar and fat to raise their body temperature by 2.7 degrees Fahrenheit. These effects must result from activation of the FXR receptor, because they don’t occur in drug-treated mice that lack the receptor.
A critical difference between animal research and clinical therapy is that doctors can’t transplant bacteria into germ-free people. Instead researchers have to contend with the species that are already present in the gut, an assortment that is different for each individual. Changing the microbiome is less like giving a drug and more like trying to restore a damaged habitat, with a lot of potential interactions between species that are difficult to predict. For example, when scientists removed all the starfish from a stretch of shoreline, the diversity of species was cut in half as mussels took over much more territory than before, successfully crowding out their competitors. Different species of bacteria also compete with one another, so if one type is too prevalent, it might be able to elbow out any replacement that doctors put in. The number of possible outcomes is head-spinningly complicated, so it will probably take a long time to work out the details.
Another issue with applying this research in the clinic is defining what a desirable population of gut bacteria would look like. As a first guess, we might be tempted to replicate the ancestral microbiome by copying the gut bacteria of modern hunter-gatherers. There are certainly some major differences. Hunter-gatherers have a much more diverse set of species than ours, with estimates ranging from 50 percent more to twice as many. But there are also some problems with the idea of copying their microbiome. Gut microbes vary across different populations of hunter-gatherers and change seasonally as food sources shift, suggesting that there was probably no single ancestral state. More important, the gut bacteria of these populations are adapted to their own lifestyles and may not play well with ours. Bacteria that specialize in digesting fiber would have trouble finding anything to eat in the guts of a typical suburbanite.
In Sickness and in Health
The gut microbiome also contributes to disease. Because gut bacteria help calibrate the immune system early in life—particularly its ability to distinguish self (“that’s me, so leave it alone”) from no self (“foreign invaders must die”)—changing the microbiome in adulthood might lead the immune system to attack the intruders, potentially causing autoimmune diseases.
Diabetes is associated with shifts in the gut microbiome, regardless of body weight. Unlike normal mice, germ-free mice do not develop insulin resistance on a high-fat, high-carbohydrate diet. After receiving a bacterial transplant from normal mice, however, they show a strong increase in blood sugar and insulin resistance. This result makes sense because gut bacteria can influence the uptake of glucose into the blood. One group of gut bacteria may help to cause diabetes. These are the bacteria that produce a signal called lipopolysaccharide, which triggers the production of many inflammatory molecules, especially in animals fed a high-fat diet. The resulting inflammation in fat cells and elsewhere seems to produce both type 2 diabetes and obesity. Similarly, when gut bacteria from lean people were transplanted into obese people with metabolic syndrome, their insulin sensitivity increased within six weeks. At six months, though, the gut bacteria had returned to their normal state, in line with the idea that a bacterial ecosystem can push back against changes.
Researchers isolated a single species of bacteria that produces lipopolysaccharide from the gut of a very obese person. Initially, when he weighed 385 pounds, this bacterial species made up 35 percent of his microbiome, but six months later, after he had lost 110 pounds, it was undetectable. When germ-free mice were transplanted with this species alone, they became obese and diabetic, but only if they ate a high-fat diet. In another experiment, feeding mice a high-fat diet for a month increased the population of bacteria that produce lipopolysaccharide, raising its blood levels by two- to threefold. The mice gained weight and started showing symptoms of diabetes. The same things happened when lipopolysaccharide was infused into mice eating a normal diet. In contrast, blocking the receptor that receives signals from lipopolysaccharide protected mice from the effects of a high-fat diet. These experiments suggest that it might eventually be possible to prevent the development of diabetes even in people who are unable to exercise or eat well, though such a drug remains a long way off.
Gut bacteria are also modified by pregnancy, when increased fatness and reduced insulin sensitivity help the mother’s body to provide the energy needed for fetal growth and breast-feeding. A study of ninety-one women in Finland found that pregnancy has a huge effect on gut bacteria. Proteobacteria and Actinobacteria became more abundant in the majority of women, with no overall change in the abundance of Firmicutes or Bacteroidetes. Third-trimester bacteria contained fewer species than first-trimester bacteria, which are more similar to no pregnant gut bacteria. By the third trimester and continuing to one month after birth, the mothers’ bacterial colonies were more individually distinct than they had been early in pregnancy. Their babies also had a wide diversity of bacteria at six months of age but reached adult levels of similarity across individuals by four years of age. At this age, their gut bacteria were similar to the mother’s as well. When germ-free mice were transplanted with bacteria pooled from five pregnant women, the third-trimester sample caused them to gain one third more weight than the first-trimester sample, even though they ate the same amount of food. The third-trimester sample also reduced glucose tolerance in the mice. These changes increase the risk of diabetes, so they may contribute to the development of gestational diabetes in some women.
The number of different bacterial species in the gut may also matter. People with a less diverse set of gut bacteria are more likely to be obese and to have metabolic syndrome. In fact, researchers can predict whether people are fat or thin more effectively from their bacterial genome (90 percent accuracy) than from the same people’s own genes (58 percent accuracy). Diversity can be reduced by antibiotic treatment, which not only kills bacteria but also favors certain species over others.
We can get a hint of which species are more likely to survive by recalling that industrial agriculture has used low doses of antibiotics in feed or water to promote weight gain in meat animals for decades. The strategy works in cows, sheep, pigs, chickens, and turkeys. The younger the animals are when treated, the more weight they gain. When researchers give mice low-dose antibiotics from birth, they end up with a higher proportion of Firmicutes and fewer Bacteroidetes (the pattern associated with obesity) and have more fat tissue than their littermates without antibiotics. When mice receive antibiotics for four weeks in infancy, their gut bacteria population gradually goes back to normal, but then in adolescence they develop obesity and metabolic problems, including accumulation of fat in the abdomen and the liver. Starting antibiotics earlier in life, before birth, leads to more weight gain. Putting antibiotic-treated mice onto a high-fat diet produces the strongest effects of all, especially in males, which are more susceptible to diet-induced obesity.
A similar relationship may occur in people. For example, among Canadian babies who did not receive antibiotics in their first year of life, 18.2 percent were overweight or obese at age twelve. In contrast, 32.4 percent of children exposed to early antibiotics became overweight or obese. After adjustment for potential confounding factors like birth weight and the mother’s weight, the relationship remained statistically significant in boys but not girls. Boys were more than five times as likely to become overweight at twelve if they had received antibiotics and almost three times as likely to have a lot of visceral fat. Similarly, among more than twenty-seven thousand Danish children born to normal-weight mothers, antibiotics in the first six months of life predicted a 54 percent higher chance of being overweight at seven years. However, the opposite result was found for children of overweight mothers, who were half as likely to be overweight at seven if they had received antibiotics, suggesting that the effect of these drugs may vary depending on the child’s initial gut bacteria population.
For now, we can take a couple of lessons from this research. Parents should minimize antibiotic use in children, especially in the first year of life, because changes in gut bacteria at that age can have lasting consequences. The average child in the United States receives ten to twenty courses of antibiotics before age eighteen, increasing the risks of asthma, allergies, and inflammatory bowel disease, in addition to obesity and diabetes. Someday we may have drugs or probiotics based on this research, but they’re not coming anytime soon. In the meantime, we should all think twice about blaming people for their weight. The composition of our gut bacteria profoundly affects our bodies, yet no one knows how to use that knowledge to change them.
What are your experiences with weight gain and antibiotics?