Caloric restriction and weight loss are associated with decreases in resting energy expenditure (REE). This makes sense when we understand that less body mass means there is less tissue that needs to be kept alive. However, there is often a decrease in REE beyond those decreases accounted for by changed fat-free mass (FFM) and fat mass (FM), which has become known as adaptive thermogenesis (AT). This is essentially the body’s way of fighting against weight loss in response to perceived starvation.

AT was first described in the Minnesota Starvation Experiment, where men ate half of their daily energy requirements for six months and ultimately lost 24% of their initial bodyweight. Their REE was reduced by 40%, of which only two-thirds could be attributed to weight loss. However, the Minnesota Starvation Experiment was conducted in the 1940s and 50s, leaving open room for error through the use of now outdated technology. Additionally, the diet consisted primarily of potatoes, cabbage, turnips, and porridge, making it very low-fat (27 g/d) and low-protein (55 g/d).

Although AT is established, what causes it remains elusive. Most authors have assumed that AT is an autoregulatory response explained by the reduced activity of the sympathetic nervous system (SNS) and low concentrations of thyroid hormones and leptin. Now, researchers from Christian-Albrechts University, Germany may have an answer. Their study was essentially two experiments.

The first experiment involved 32 healthy young men recruited from the Christian-Albrechts University campus. For six weeks they resided at the research institute and were subjected to 1-week of overfeeding (OF; +50% energy requirements), 3-weeks of caloric restriction (CR; -50%), and 2-weeks of refeeding (RF; +50%). All foods and drinks were provided and water intake was unrestricted.

The second experiment was conducted 1.5 years later in eight of the original 32 participants. This experiment was much shorter, involving only 1 week each of OF and CR, and was intended to provide more detailed information on the short-term adaptations to dieting, as well as the reproducibility of the first experiment.

Body composition

Body composition was determined using the 4-compartment method, which is the most accurate method researchers have to measure body fat. Whole-body MRI was used to assess the volumes of adipose tissue, skeletal muscle mass, and internal organs, all of which were transformed into masses based on established organ and tissue densities. This was combined with data from air-displacement plethysmography (BodPod) and total body water dilution techniques to give a complete 4-comparment model picture of body composition.

On average, the participants gained 2.3% of bodyweight with OF, lost 7.5% with CR, and regained 4.5% during RF. Something I find fascinating is how machine-like these changes in bodyweight were. All three phases either added or subtracted 50% of energy requirements, and in all three phases the weight change was 2.3-2.5 % bodyweight per week. Despite day-to-day fluctuations in weight of 1.5-2 %, which would easily mask any noticeable change, ultimately the first law of thermodynamics holds true – calories matter.

Although bodyweight was predictable, changes in body composition were clearly influenced by the feeding conditions of the participants. Average percentage changes in FM were +5.8% (OF), -17.8% (CR), and +10% (RF), while corresponding changes in FFM were +1.5, -5.2, +3.6 %. It appears that OF and CR resulted in similar changes, but RF favored the repletion of lost FFM rather than FM. The changes in FFM were accounted for primarily by changes in the masses of skeletal muscle and the liver.

Energy Expenditure

Over the course of 10 weeks before the study began, habitual food intake (with the use of dietitian-guided dietary records), REE (with the use of indirect calorimetry), and physical activity (with the use of 24-hour heart rate monitoring and accelerometry pedometers) were assessed 3 times to calculate individual energy needs.

During the study, indirect calorimetry was again used to determine actual REE. This was compared against a predicted REE that was based on the standard energy expenditure of body organs and tissues, which were measured via the whole-body MRI scan. The difference between the two values represents adaptive thermogenesis.

During CR, there were statistically significant declines in 24-hour energy expenditure, REE, the respiratory quotient (RQ), heart rate, blood pressure, and the energetic cost of walking, all of which returned to baseline values after RF. A similar reduction and rebound to baseline during CR and RF, respectively, was observed for SNS activity, TSH and free T3, leptin, insulin, adiponectin, and testosterone. However, OF resulted in an increase in the RQ only, indicating an increased reliance on glucose to supply energy. These data suggest that metabolic adaptations are more pronounced with weight loss than weight gain, favoring a conservation of energy with the former.

The changes in REE, RQ, and AT became significant by day 3 of CR, suggesting there is a lag-period before adaptation occurs. However, there was no significant difference between reductions in the first week and the remaining weeks, suggesting that AT is primarily a short-term response. Over the entire CR period, AT amounted to 72 kcal on average. Greater reductions in weight, FFM, fluid balance, heart rate, and C-peptide (insulin secretion) all predicted the extent of AT, but that was it. There were no correlations between the extent of AT and changes in body fat, organ masses, T3, insulin, leptin, or SNS activity.

Finally, changes in body composition and energy expenditure were all statistically significantly reproduced in experiment 2 that was conducted 1.5 years later, again supporting the notion that we are somewhat machine-like.

So what does this all mean?

The primary purpose of this study was to investigate adaptive thermogenesis (AT) and its regulation in response to caloric restriction. The results suggest that assumptions regarding the importance of SNS activity, thyroid function, and leptin are not entirely accurate. Rather, a reduced heart rate, kidney function, and body temperature together with increased hepatic gluconeogenesis are sufficient to explain AT.

The calculation of REE according to the observed decreases in function-related changes in the specific metabolic rates of the heart (decrease in heart rate: 213%; 2384 kcal/kg) and kidneys (decrease in kidney function: 239%; 2269 kcal/kg) together with the apparent increases in liver-specific metabolic rates (206 kcal/kg as a result of increased gluconeogenesis as calculated from urinary urea excretion) add up to 40 kcal/d, which leaves 32 kcal/d or 44% of “true” AT unexplained. Taking into account the (nonsignificant) decline in body temperature (20.38C; Table 2) and a temperature coefficient (Q10) of 2 (= 238 kcal/d) would explain AT.

Therefore, it appears that CR-associated changes in leptin, T3, and SNS activity reflect adaptations to weight loss, but they are not related to AT. Importantly, these results do not argue against the known thermic effects of T3, leptin, or the catecholamines, they merely suggest that these effects do not regulate adaptations to caloric restriction. This is supported by research showing that inhibiting the creation of T3 does not lower REE, and by research showing that leptin is only thermogenic in underweight individuals.

The other important piece of information this study shows is that AT has a lag period of about three days, is reduced primarily within the first week, and returns to baseline within 2 weeks of refeeding. The implications for anyone choosing to lose weight are simply that adaptations do not continue to build upon themselves and are completely reversible relatively quickly. This certainly goes against the popular notion of “damaged” metabolisms. Interestingly, the lag-period also suggests that any metabolic adaptations will not occur if someone cycles between low and high calorie days.

This study is not without limitations unfortunately. It must of course be mentioned that protein intake was well below optimal. Average intake during the CR phase was only 49 grams per day, which corresponds to 0.63 g/kg bodyweight. Also, the participants were all sedentary and normal weight or overweight, and the dieting duration was only 3 weeks. Therefore, we cannot say whether different outcomes would have occurred with an exercising or obese population consuming more protein over a longer time period.