Energy Balance Consortium Pilot Study

Principal Investigators: Kevin Hall (National Institutes of Health, NIDDK), Rudolph Leibel (Columbia University Medical College), Eric Ravussin (Pennington Biomedical Research Center), Marc Reitman (National Institutes of Health, NIDDK), Michael Rosenbaum (Columbia University Medical College), Steven Smith (Translational Research Institute for Metabolism and Diabetes)

Hall KD, Chen KY, Guo J, Lam YY, Leibel RL, Mayer LE, Reitman ML, Rosenbaum M, Smith SR, Walsh BT, Ravussin E.  Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men.  Am J Clin Nutr. 2016; 104: 324-33.  Read >>

Studies in free-living individuals are crucial to determine the effectiveness of dietary strategies to prevent or treat obesity and related metabolic conditions under real-world conditions. Such studies can provide tests of hypotheses about the cause of these disorders; however rigorous testing of these hypotheses requires greater control over the diets of study subjects than free-living studies permit. In practical terms, this means housing study participants in a facility where their diet is precisely defined, their food intake measured accurately, and their physical activity better controlled and monitored.

As a first step toward a rigorous test of the energy balance hypothesis of obesity – that foods influence fat accumulation only through the amount, but not type, of calories consumed – NuSI organized and funded a consortium of investigators to conduct a small preliminary or “pilot” study using participants confined to metabolic wards. The consortium researchers deemed it unlikely that they could reliably detect meaningful changes in body weight or fat content in an 8-week trial and instead chose to use energy expenditure as the primary outcome of the study.

To test the question of whether the amount or type of calories consumed is critical, participants were fed an equal number of calories from diets differing dramatically in fat and carbohydrate content. In addition to generating useful preliminary data that might address the hypothesis and help determine the number of subjects required for a full-scale study, the goals of this pilot study were operational and methodological: optimize the logistics required for a multi-site study, including, importantly, the assessment and fine-tuning of the precision and sensitivity of techniques for measuring energy expenditure.

A total of 17 overweight and obese male participants were housed in metabolic wards at four study sites. For 4 weeks they were fed a basal diet based on a standard American diet (50% carbohydrate, 35% fat, and 15% protein by calories) providing a number of calories intended to maintain them in energy balance as reflected by a stable body weight. Participants were then switched to a “ketogenic” diet of identical caloric content, but radically different macronutrient composition (5% carbohydrate, 80% fat, and 15% protein by calories). Energy expenditure was measured two ways: (i) by respirometry in which participants were housed in small room-sized metabolic chambers for two consecutive days each week throughout the study, and (ii) over the last 2 weeks of each diet period, by a technique known as doubly labeled water that does not require chamber confinement.

If, according to the energy balance hypothesis, fat accumulation depended only on the amount of energy consumed – if a calorie of carbohydrate is equally fattening as a calorie of fat – no difference would be expected in energy expenditure under the two diet conditions. However, because the very low-carbohydrate ketogenic diet would minimize secretion of insulin, a hormone that promotes the synthesis and storage of fat, an alternative hypothesis (known as the carbohydrate-insulin model or the hormonal/regulatory model of obesity) predicts that energy expenditure will increase during the ketogenic diet period as stored fat is mobilized from stored body fat and burned for fuel along with fat from the diet.

The results showed that energy expenditure measured in the metabolic chambers increased transiently during the first two weeks after the switch to the ketogenic diet by 60 to 100 kcal/day, depending on how expenditures were calculated relative to body weight or composition. As measured by doubly labeled water, energy expenditure increased during the last two weeks of the ketogenic diet period by 150 kcal/day, an observation that is difficult to reconcile with the energy balance, calorie-is-a-calorie perspective.

Interpretation of the results is confounded by limitations of this small pilot study:

  • Because the order in which diets were fed to participants were not randomized, no causal inference can be made about the role of diet composition as a cause of any change in energy expenditure. Strictly speaking, the results can be interpreted only in terms of before and after the diet switch and not necessarily because of the switch.
  • Although the study plan was to have subjects in energy balance during the end of the four-week basal diet period, this was not achieved. The subjects lost weight throughout the study. This unintentional weight loss complicates interpretation of the magnitude of the increase in energy expenditure because weight loss is typically associated with reduced expenditures, which could mitigate any increase caused by the diet switch.

Despite these limitations, the results suggest that a well-controlled, full-scale study using appropriate methodologies is warranted to definitively address the question of whether fat accumulation is influenced by the macronutrient composition of the diet or only the amount of calories consumed.

Secondary analyses and ancillary studies

Rosenbaum M, Hall KD, Guo J, Ravussin E, Mayer LS, Reitman ML, Smith SR, Walsh BT, Leibel RL.  Glucose and lipid homeostasis and inflammation in humans following an isocaloric ketogenic diet.  Obesity. 2019; 27: 971-981.  Read >>

Switching overweight and obese participants in the EBC pilot study from the basal to the ketogenic diet was associated with changes in the blood concentrations of certain lipids and inflammatory markers and measures of glucose homeostasis.  Despite these effects of changing the diet, participants’ individual relative risk for metabolic disease was unchanged.

Hall KD, Guo J, Chen KY, Leibel RL, Reitman ML, Rosenbaum M, Smith SR, Ravussin E.  Methodologic considerations for measuring energy expenditure differences between diets varying in carbohydrate using the doubly labeled water method.  Am J Clin Nutr. 2019; 109: 1328–1334.  Read >>

As reported in their original 2016 paper, energy expenditure as measured using doubly labeled water (DLW) increased significantly after switching participants from the basal to the ketogenic diet. This reanalysis of the DLW data based on theoretical considerations regarding changes in diet and energy balance suggests that this effect of the diet switch was smaller than initially reported and could be practically eliminated by exclusion of data from two subjects deemed to be outliers in their responses to the dietary change.

Friedman MI, Appel S.  Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men: A secondary analysis of energy expenditure and physical activity. PLoS ONE. 2019; 14(12): e0222971.  Read >>

This secondary analysis of doubly labeled water (DLW) data from the EBC pilot study suggests that the effect on subject energy expenditure of switching from the basal to the ketogenic diet was underestimated because subjects were confined to metabolic chambers for 4 days during the 2-week DLW measurement period. Exclusion of DLW data is questioned given issues with the validity of the criteria for exclusion and post hoc selection of putative outliers.

Hall KD.  Mystery or method? Evaluating claims of increased energy expenditure during a ketogenic diet. PLoS ONE. 2019; 14(12): e0225944.  Read >>

In this comment on Friedman & Appel, Hall, using theoretical adjustments in respiratory quotient (RQ) values for energy balance and diet, confirms the effect of housing conditions on the calculated increase in energy expenditure measured using DLW after the switch to a ketogenic diet.  However, the increase in expenditures after the change in diet are again largely eliminated when data is excluded from two putative outliers identified using new criteria.

Friedman MI.  Effect of a ketogenic diet on energy expenditure: What conclusions can be drawn from a problematic pilot study?  PLoS ONE. 2019.  Read >>

Friedman’s response to Hall’s comment raises additional concerns about the new post hoc criteria for excluding putative outlier data from analysis and illustrates how exclusion of data can increase or decrease effect size depending on which outliers are selected.  In addition, Friedman emphasizes that while the EBC pilot study was helpful in the design and planning of future studies, it does not permit conclusions about causal effects of diet.

Ang QY, Alexander M, Newman JC, Tian Y, Cai J, Upadhyay V, Turnbaugh JA, Verdin E, Hall KD, Leibel RL, Ravussin E, Rosenbaum M, Patterson AD, Turnbaugh PJ.  Ketogenic diets alter the gut Microbiome resulting in decreased intestinal Th17 cells. Cell. 2020; 181: 1263-1275.  Read >>

An analysis of bacterial populations in stool samples collected from participants in the EBC pilot study, supplemented by controlled experiments in mice, suggest that ketogenic diets alter gut microbiota in a manner distinctly different from that seen with high-fat diets containing carbohydrate. Specifically, ketogenic diets reduce, rather than increase, a species of gut bacteria that is associated with inflammation and also reduces the level of an intestinal cell-type involved in the immune response.