Energy Expenditure of Rhesus Monkeys Subjected to 11 Years of Dietary Restriction

Dietary restriction (DR) is currently the only paradigm that has consistently extended maximal life span and reduced the onset of age-related chronic diseases in all of the nonprimate species tested. Although it is controversial, some investigators have suggested that the underlying mechanisms may be mediated by adaptations in energy expenditure. We evaluated the extent to which DR alters energy metabolism in a unique cohort of rhesus monkeys submitted to DR for 11 yr. Total energy expenditure (doubly labeled water), resting energy expenditure (REE; indirect calorimetry), and nonbasal energy expenditure (calculated by difference) were measured in DR (n = 12) and control (n = 11) animals. Body composition was determined by dual energy x-ray absorptiometry. Both fat mass and fat-free mass were lower in the restricted animals (56 and 12%, respectively). DR induced a 17% lower total energy expenditure that was attributable to a 20% decrease in REE without changes in the nonbasal energy expenditure. Adjusted for fat-free mass, REE was 13% lower with DR (−250 kJ/d). Taken together with a reanalysis of previous DR experiments published in humans, rodents, and monkeys, these results suggest that DR may lower REE independent of the DR-induced changes in body composition. Whether this reduction in REE contributes to the life-extending properties of DR warrants further analysis, but it suggests that the long-standing debate regarding DR effects on metabolic rates may derive from the lack of consensus on how to adjust for body size and composition.

IN RODENTS, DIETARY RESTRICTION (DR) without malnutrition is the only intervention consistently shown to increase maximum life span and reduce or delay the onset of the diverse age-related diseases such as diabetes, hypertension, cancer, and bone demineralization (1). These results have been extended to other animal species (2), thereby suggesting that DR affects critical pathways in the mechanisms of aging.

In support of the free radical theory of aging, it has been suggested that energy expenditure (EE) and its accompanying consumption of oxygen may play a critical role in the life-extending properties of DR (1). This hypothesis proposes that a reduced oxygen consumption may lower the formation of reactive oxygen species, thus moderating oxidative damage and increasing life span. Indeed, to be valid from a thermodynamic point of view, such a hypothesis would imply that the DR-induced changes in EE are independent of the DR-induced changes in body size and composition.

Using whole body measurements, several attempts have been made in humans (3–8), rodents (9–18), and monkeys (19–22) to determine the role EE may play in the action of DR. As expected from the first thermodynamic law, most long-term studies—thereby excluding the short-term human investigations—report a decrease in EE at the whole body level. The discrepancies arise, however, when attempts are made to adjust metabolic rates for changes in body size and composition. In rats, adjusted total EE (TEE) and resting EE (REE) were reported unchanged (9, 11, 13–16) or decreased (10, 12, 17, 18) after different periods of DR ranging from 2–22 wk, with severities varying from 10–80%. In rhesus monkeys, Delany et al. (19) reported a reduction in adjusted TEE after 10 yr of weight clamping, whereas Lane et al. (20) did not observe differences after 5 yr of 30% DR. Ramsey et al. (22) reported a significant reduction in adjusted REE after 30 months of a 28% DR, but no difference in TEE. The REE difference, however, disappeared at later assessments (21). Consequently, it is still under debate whether or not DR can induce sustained changes in energy metabolism that are independent of the changes in body size and composition.

Our previous studies have used indirect respiration calorimetry chambers to measure EE. Although this method can provide detailed information about the components of TEE (REE, thermic effect of feeding, respiratory quotient, etc.), the method is not necessarily a reflection of EE in the home cage of the animal. To circumvent this potential problem, the EE measurements at the 11-yr assessments were expanded to include home cage doubly labeled water (DLW) measurements. The REE was assessed by indirect calorimetry, and the cost of physical activity was calculated by difference.

Materials and Methods

Experimental protocol

Rhesus monkeys were studied to determine whether moderate dietary restriction extends life span and slows the rate of age-associated diseases in primates. This longitudinal study was initiated in 1989 with a cohort of 30 male rhesus monkeys (8–14 yr of age) divided into a group of control monkeys allowed ad libitum access to food and a group of monkeys provided with approximately 30% less food than the controls. Details of the DR procedure were previously published (23). Briefly, the DR was achieved by randomly assigning animals to a treatment group after a 3- to 5-month period of baseline assessment during which food intake of the experimental diet was determined individually. Food intakes of the animals assigned to DR were progressively reduced from their baseline period averages by 10% per month for 3 months and then maintained at a 70% level relative to controls. The control animals continued to have free access to food during the 6- to 8-h daily feeding period.

Animals are assessed through a battery of tests every 6 months according to protocols described in detail elsewhere (23, 24). The present study reports the investigation of energy metabolism at yr 11 of the study. The TEE was measured by the DLW method, the REE by indirect calorimetry, and the nonbasal EE (NB-EE) was calculated by difference. Lastly, body composition was assessed by dual x-ray absorptiometry (DXA). The indirect calorimetry measurements were performed 1–2 wk afterward to the DLW and DXA protocols.

Animals

The rhesus monkeys (Macaca mulatta) were born and raised at the Wisconsin Regional Primate Research Center. Twenty-four monkeys with a mean age of 20.3 ± 0.3 yr were studied. Of these 24 animals, 12 were in the DR group, and 12 were in the control group.

The animals are individually housed in cages (89 cm wide × 86 cm deep × 86 cm high) to allow measurement and control of food intake. The cages contain food hoppers and spigots providing continuous access to water. Room temperature is maintained at approximately 21 C with a 12-h light/dark cycle (lights on at 0600 h and off at 1800 h).

Diet

The monkeys are fed a defined, pelleted diet (no. 85387, Teklad Co., Madison, WI) containing 15% lactalbumin, 10% corn oil, and approximately 65% carbohydrate. Details about the diet were previously published (24). Food is provided in the morning and removed 6–8 h later. At this time, any spillage or food remaining is weighed, and the animals are given a piece of fresh fruit.

Because food intake can be variable in the control animals, the protocol of food intake was slightly modified in the calorimetry chambers. To ensure that all of the animals consumed their average daily food intake while in the chambers, a palatable sandwich (fat 49%, carbohydrate 38%, protein 13%) was provided for all measurements. The sandwiches were fed at two meals. The first meal was at 0900 h, and the energy content of this meal was equal for all animals. The afternoon meal was fed at 1500 h, and the energy content of this meal was adjusted to ensure that all animals consumed their normal daily energy intake.

Body composition and anthropometric data

Body composition was measured by DXA (Model DPX-L, GE/Lunar Corp., Madison, WI; Ref. 25) using the software Pediatric 4.0a (GE/Lunar Corp.). Animals were sedated with ketamine HCl (10 mg/kg im) plus xylazine (0.6 mg/kg im) for additional muscular relaxation and scanned in the supine position. Body weights were recorded, and the crown-to-rump lengths were measured. The body surface area of the animal was calculated by the DuBois formula.

DLW procedures

TEE was determined during a 5-d period by the two-point DLW methodology described by Schoeller et al. (26). After anesthesia (ketamine HCl, 15 mg/kg im), a baseline blood sample was collected, and a premixed 1.51 g/kg estimated total body water (TBW) dose of DLW was injected iv to the animals. The dose was composed of 0.3 g/kg estimated TBW of 94% H₂¹⁸O (Rotem Industries Ltd., Beer Sheva, Israel) and 0.17 g/kg estimated TBW of 99.9% ²H₂O (Cambridge Isotope Laboratories, Andover, MA) and was diluted with 3% NaCl to physiological osmolarity. The doses were calculated to ensure an in vivo enrichment of about 135 and 992 ‰ for ¹⁸O and deuterium, respectively [‰ (δ per mil) = (R_sample/R_standard − 1)·1000, with R being the ratio heavy to light isotope]. Isotopic equilibration in body water was determined through a blood sample collected at 2-h post dose. Blood samples were collected on d 1 and d 5 after dosing to complete DLW calculations. Immediately after collection, blood was centrifuged for 10 min at 3500 rpm for serum separation. Serum was stored at −20 C in cryogenically stable tubes until analysis by isotope ratio mass spectrometry.

Water from serum samples was extracted by centrifugation (4 C, 1 h at 10,000 rpm) on regenerated cellulose filters (YM-50, Centricon, Bedford, MA). Deuterium isotopic enrichments in hydrogen gas were measured after chromium reduction using a dual-inlet isotope ratio mass spectrometer (Delta Plus Mass Spectrometer, Finnigan MAT, San Jose, CA), as previously described (27). The ¹⁸O isotopic enrichment analyses were performed by the CO₂ equilibration method on a Delta-S isotope ratio mass spectrometer (Finnigan MAT) using a continuous flow inlet system developed in the laboratory (28). Analyses were performed in duplicate for deuterium and in triplicate for ¹⁸O and were repeated if sd exceeded 3 and 0.5‰, respectively.

CO₂ production was calculated according to the equation of Schoeller et al. (26):

$$rCO_{\mathrm{2}}\mathrm{(mol/d){=}N/2.078(1.007}\mathit{k}_{\mathit{o}}\mathrm{{-}1.041}\mathit{k}_{\mathit{d}}\mathrm{){-}0.0266{\cdot}N{\cdot}(1.007}\mathit{k}_{\mathit{o}}\mathrm{{-}1.041}\mathit{k}_{\mathit{d}}\mathrm{)}$$

where N represents the average isotope dilution space of deuterium and ¹⁸O calculated from Coward et al. (29) by the plateau method using the 2-h postdose sample and corrected for isotope exchange by the factors 1.041 and 1.007, respectively (30). The isotope dilution space ratio was 1.028 ± 0.009 (mean ± sd) in the control and 1.028 ± 0.018 in the restricted groups. k_o and k_d represent the isotope constant elimination rates calculated by linear regression of the natural logarithm of isotope enrichment as a function of elapsed time from d 1 samples. TEE was calculated by Wier’s equation (31) using a food quotient of 0.93 estimated from the animal’s diet. Day 1 samples were chosen in the calculation of k_o and k_d to avoid the potential artifacts (hypometabolism, hypoactivity, etc.) introduced in the TEE estimates by anesthesia. TEE estimated on d 1 and during the period d 1–5 differed by 16% in the control group (P = 0.08) and 9% in the restricted group (P = 0.7). Although statistical significance was not reached, the differences indicate that the precision of the measurements will increase by excluding d 1 from the elimination rate calculations.

Indirect calorimetry

Oxygen consumption and carbon dioxide production were measured using indirect calorimetry chambers that were previously described (22). The monkeys were individually placed inside transparent metabolic chambers with inside dimensions of approximately 75 cm wide × 75 cm deep × 80 cm high for a period of 72 h. The animals were able to see and hear each other the entire time. They were allowed to adapt to the chambers for 24 h before the start of the measurements and, on the basis of behavioral observations and food intake measurements, they did not appear to be adversely influenced by this environmental change. Moreover, chamber measurements were completed on all of the animals semiannually during the first 11 yr of the study, so all the animals had extensive prior experience with the calorimetry chambers.

The rates of O₂ consumption and CO₂ production were assessed by an indirect calorimeter consisting of an open-flow system using gas analyzers. Room air was drawn through the chambers at a controlled flow rate. Chamber exhaust air was then dried and continuously sampled at a rate of 100 ml/min for analysis of O₂ and CO₂ contents (S-3A O₂ analyzer and CD-3 CO₂ analyzer, AMETEK, Pittsburgh, PA). Outputs from the oxygen and CO₂ analyzers and the flow meter were recorded every 5 min using a Workbench Data Acquisition System (Strawberry Tree, Sunnyvale, CA) and a 486 PC. The system was calibrated by burning ethanol lamps and measuring the recovery of oxygen and carbon dioxide. Nighttime or sleeping metabolic rate was considered to represent the REE. EE were calculated from oxygen consumption and carbon dioxide production using the Weir equation (31).

Statistical analysis

Due to technical problems with the calorimetry chamber, REE measurements were not performed in one control animal. To ensure homogeneity in the statistical analysis, this monkey was excluded. The differences between the variables were tested by the mean of a factorial ANOVA using group (control vs. DR) as the main factor. REE and TEE were analyzed by an analysis of covariance (ANCOVA) with fat-free mass (FFM) as the covariate, because multiple regression demonstrated that FFM was the best predictor of both variables (R² = 0.50 and 0.45, respectively). REE and TEE values adjusted for the covariate were compared by a classical unpaired t test. The difference between energy intake and TEE was tested by a repeated measure ANOVA with group (control vs. DR) as the independent variable. ANOVAs were performed using StatView 5.01 (SAS Institute Inc., Cary, NC), and ANCOVA was performed by the PROC MIXED procedure of SAS 8.01 (SAS Institute Inc.). Values are expressed as means ± sd unless otherwise stated, and a P value less than 0.05 was considered significant.

Results

Body weight and composition

Body composition data are presented in Table 1. Body weight of the DR animals was 26% lower than the controls, due to a lower FFM (12%) and fat mass (56%). Expressed as a percentage of body weight, fat mass was 41% lower in the DR group.

Table 1.

Body composition and EE

Variables	Unit	Group		Difference (%)	F ratio	P
		Control	Restricted
No.		11	12
Body composition
Weight	kg	14.4 ± 2.3	10.6 ± 1.4	−26	23.6	<0.0001
Lean body mass	kg	9.8 ± 1.2	8.6 ± 1.1	−12	5.1	0.03
Fat mass	kg	4.8 ± 1.5	2.1 ± 0.8	−56	29.3	<0.0001
	%	32.3 ± 7.6	19.2 ± 6.2	−41	20.7	0.0002
Energy intake and expenditures
EI	MJ · d−1	2.84 ± 0.11	2.04 ± 0.08	−28	35.1	<0.0001
TEE	MJ · d−1	2.54 ± 0.30	2.11 ± 0.54	−17	5.6	0.03
	kJ · d−1 · kg weight−1	180 ± 33	201 ± 49	12	1.4	0.25
	kJ · d−1 · kg FFM−1	263 ± 33	243 ± 47	−8	1.5	0.24
	kJ · d−1 · kg weight−0.67	430 ± 56	435 ± 103	1	0.0	0.87
	kJ · d−1 · kg weight−0.75	348 ± 49	361 ± 86	4	0.2	0.66
	MJ · d−1 · m−2	4.42 ± 0.48	4.20 ± 0.98	−5	0.5	0.49
TEEFFMa	MJ · d−1	2.44 ± 0.44	2.29 ± 0.44	−6		0.42
REE	MJ · d−1	2.05 ± 0.30	1.65 ± 0.23	−20	13.1	0.002
	kJ · d−1 · kg FFM−1	211 ± 26	192 ± 21	−9	3.9	0.06
REEFFMa	MJ · d−1	1.97 ± 0.25	1.72 ± 0.25	−13		0.03
NB-EE	MJ · d−1	0.49 ± 0.18	0.46 ± 0.52	−6	0.03	0.86
	kJ · d−1 · kg−1	36.4 ± 18.3	43.6 ± 50.7	20	0.3	0.62

Values are means ± sd. F ratio and P value refer to the ANOVA. EI, Energy intake; REE_FFM, resting metabolic rate adjusted for FFM; TEE_FFM, total energy expenditure adjusted for FFM.

^a

P value, when alone, refers to the t test following the ANCOVA adjustment.

Energy intake and expenditure

As shown in Table 1, energy intake was reduced by 28% in the DR-restricted animals compared with the controls. This value is close to the study’s targeted goal of 30% energy restriction.

Repeated measures ANOVA showed a significant difference between TEE and dietary intake (5%; P = 0.0003). However, the presence of an interaction (P = 0.04) demonstrated that energy intake was higher than TEE in only the control group (11%), and not the DR animals (3%).

DLW-derived TEE estimates were reduced by 17% in the DR animals. This difference was no longer significant when TEE was either expressed per kilogram of body weight or FFM, as well as expressed for the metabolic body size or estimates of body surface area. When FFM was added as covariate, TEE was also not significantly different between groups. The NB-EE was similar between both groups, expressed either in megajoules per day or megajoules per day per kilogram of body weight. Thus, the differences in TEE were mainly attributable to differences in REE. In the DR animals, the REE was 20% lower than in the control group. Even after adjustment for FFM, REE was still 13% lower. The relationships of TEE and REE with FFM are presented in Fig. 1.

Figure 1.

Top, Relationship between TEE and FFM for the control (n = 11; Y = 1.33 + 0.13·X; R² = 0.27) and energy-restricted animals (n = 12; Y = −0.69 + 0.32·X; R² = 0.46). Bottom, Relationship between REE and FFM for the control (n = 11; Y = 0.69 + 0.14·X; R² = 0.33) and energy-restricted animals (n = 12; Y = 0.46 + 0.14·X; R² = 0.46).

Discussion

It has been hypothesized that a reduced metabolic rate may be one of the mechanisms by which DR increases life span (1, 32). Using a nonhuman primate model of aging, we observed that after 11 yr of DR the unadjusted TEE is reduced by 17% partly due to a reduction in REE (250 kJ/d) adjusted for FFM. The mass-adjusted TEE difference is not significant because of the variability observed in individual responses to DR. This is shown by the fact that, at similar body weights, some DR monkeys present TEE equivalent to the control animals. When focusing on REE, thereby excluding the variable cost of physical activity and other confounding factors, the variability decreases. This may likely explain why the decrease in TEE adjusted for FFM does not reach significance, whereas REE adjusted for FFM was significantly lower in the DR group.

The reduction in REE adjusted for FFM has not been consistently observed in this colony. At 30 months of DR in these animals, a significant decrease in mass-adjusted REE was reported, but no significant difference (P > 0.05) in mass-adjusted TEE was observed in DR compared with control animals (22). This difference in REE, however, was no longer significant at later assessments, and we have never observed a difference in 24-h TEE (1, 22, 24). The reason for the reemergence of the decrease in FFM-adjusted REE is not entirely clear. It is possible that DR induced a sustained decrease in REE, but the magnitude of the difference was small and thus differences could not be consistently detected. Although this is possible, REE comparisons between the control and DR animals have often shown no hint of a consistent reduction in the DR monkeys (24). It is also possible that age-related changes in EE magnified or stimulated the observed changes in REE. To pursue this result, we used a repeated measure ANCOVA to test differences between REE adjusted for FFM across yr 9, 10, and 11. The statistical model showed no effect of DR (F = 2.6; P = 0.1). However, there was a significant group-by-year interaction (F = 4.3; P = 0.02), suggesting that the lack of difference between control and DR animals was dependent on the year studied. Indeed, when yr 9 was removed from the model, the REE adjusted for FFM was significantly lower in the DR animals compared with the controls (−220 kJ/d; F = 5.4; P = 0.03), and the above interaction disappeared (F = 0.1; P = 0.8). This clearly demonstrates that there is a resurgence of the FFM-adjusted REE difference at yr 10, even if at this time the mechanisms underlying this phenomenon remain unclear.

Three studies are currently using nonhuman primate models to study long-term DR. The universities of Wisconsin and Maryland initiated DR in adult monkeys, whereas the National Institute on Aging (NIA) started DR in juvenile age (1–2 yr) and young-adult age (3–5 yr). Two of these studies have reported a significant decrease in TEE (19, 22), but only one (19) has reported a decrease in mass-adjusted TEE with long-term DR. It should be mentioned that the latter study applied a weight-clamping procedure that cannot be completely compared to a DR. Nevertheless, human and rodent studies support these discrepancies. Several explanations have been suggested to elucidate the discrepancies of TEE responses to DR such as the inclusion age, the initial body composition, and the severity and duration of the DR (1). No clear explanations have emerged. Certainly, numerous inappropriate methods of metabolic rate adjustments have confounded interspecies and study comparisons. In published DR studies, metabolic rate is expressed per kilogram of body weight, estimated FFM, metabolic body size, body surface area, or adjusted by ANCOVA. With these different adjustments, no clear pattern of responses emerged.

Proper metabolic rate adjustment is central to the understanding of the mechanisms by which DR may affect EE and promote its life-extending properties. According to the first thermodynamic law, it is expected that most DR studies in humans, monkeys, and rodents should demonstrate a decrease of TEE. To some degree, animals adapt to sustained DR by progressively reducing body weight until EE matches energy intake. Consequently, it seems reasonable to argue that if EE plays a role in the action of DR, the observed changes in raw metabolic rates should be independent of the changes in body size and composition. Yet, the heterogeneity of the published data cannot answer this question.

To give new insights regarding the effects of DR on energy metabolism, we orientate the following discussion by focusing on the limitations, advantages, and results of the different methods of metabolic rate adjustments applied during DR studies rather than doing a typical comparison of the DR effects regardless of the adjustment methods. Lastly, we attempt to perform a common adjustment of metabolic rates of published results to determine whether across studies and species DR-induced changes in metabolic rates are independent of the DR-induced changes in body size and composition.

Adjustment of metabolic rates per kilogram of tissue during DR

Older literature using animal models often expressed EE per kilogram of body weight or FFM (33). This approach, however, is a misuse of a ratio because the relationship between metabolic rate and body weight/FFM has an intercept different from zero (34). This approach, however, has been used in several DR studies. In rodents, McCarter et al. (15) using 4.5 months of 40% DR reported no changes in the 24-h TEE expressed per kilogram of FFM. Gonzales-Pacheco et al. (12) reported a decrease in FFM-adjusted 24-h TEE and REE in DR animals after the same duration and severity of DR; however, no differences were noted after adjustment for weight. In the same way, Dulloo et al. (10) observed a drop in the 24-h TEE expressed either per kilogram of weight or FFM after 22 wk of 50% DR. Lastly, after 24 months of 40% DR, McCarter et al. (16) found a drop of the 24-h TEE expressed either per kilogram of weight or FFM, and REE was responsible for this decrease. We did not find any evidence that differences in the sample sizes (from 6–10 animals per group), the age at the time DR was initiated, or initial body composition explained these discrepancies. We found only one experiment in humans using this type of adjustment during a DR protocol of 4–7 wk of semistarvation (3.3MJ/d) followed until 10% body weight loss was reached (6), and TEE per kilogram FFM decreased. However, this magnitude of DR is much greater than could safely be used in long-term DR and aging studies. In monkeys, the NIA (20) did not observe differences in TEE expressed per kilogram of weight or FFM after 5 yr of 30% DR, whereas the University of Maryland weight-clamped monkeys showed a sustained TEE depression with both normalizations after 10 yr (19). The differences are difficult to explain, but it is interesting to note that only Lane et al. (20) used a private laboratory to perform isotopic analysis of the DLW method. No effect of DR was noted when this method of adjustment was applied to our measures of either TEE or REE.

Adjustment of metabolic rates for the metabolic body size during DR

Adjustment for metabolic size is the second classically used method (33). It is achieved by dividing the EE by the body weight expressed to the three fourths power and is based on Kleiber’s observation (33) that across species, the relationship between the logarithm of metabolic rate against the logarithm of body mass is a linear function with a slope of three fourths. Kleiber (33) also demonstrated that “the experimental line representing proportionality between metabolic rate and body weight stayed within the same confidence intervals of the theoretical line expressing metabolic rate to the three fourths power of body weight through a ratio of weight of 3.2.” This means that comparison of metabolic rates expressed to the three fourths power of body weight is meaningless unless the weight of the leanest animal under study is at least three times smaller than the largest one, which is not the case in DR studies.

Regardless of the above considerations, this way of adjustment has been frequently used to study the effects of DR. In rodents, TEE·kg^−0.75 is either unchanged (14–16) or decreased (9, 10, 12) after long-term DR. Similarly, REE is either decreased during DR (12, 17), shows only a transient drop (13), or is decreased during severe but not moderate DR (9). In humans, metabolic rates are generally not corrected in such a manner, and we found no DR studies applying such an adjustment. TEE expressed per metabolic body size in monkeys was shown to either decrease (19) or remain stable (20). The results of the present experiment do not support differences in TEE to the three fourths power of body weight after long-term DR.

Adjustment of metabolic rates for the body surface area during DR

The third adjustment method is based on the surface law, in which metabolic rates are proportional to the body surface area (33). This adjustment is hampered by the difficulty of accurately determining a body surface area. In practice, such adjustment is achieved by dividing metabolic rate by the body weight to the two thirds power. This is based on the fact that large and small animals of similar shape have a surface area in proportion to the square of their linear dimensions (i.e. to two thirds power of their volume). However, the body surface area is proportional to the two thirds power of body weight if, and only if, they have the same density. Thus, this method can give biased results when comparing animals of the same body surface area having different body composition, which is obviously the case during DR studies. Nevertheless, this way of adjustment has been used frequently to test the effect of DR on EE.

In rodents, four studies of similar DR duration and severity reported either a sustained decrease (10, 18) or no change (15, 16) in TEE·kg^−0.67. The same discrepancy is observed in the monkey models, with the University of Maryland study (19) reporting a decrease in TEE·kg^−0.67, whereas no changes were reported in the NIA study (20). Because monkeys and humans are primates of similar shapes, we determined the body surface area of the monkeys in our study by the DuBois formula. This equation has the advantage of being dimensionally correct and therefore valid for different sizes of body. Dimensionally correct means that the body surface is expressed in proportion to the square of the body’s linear dimension or to the two thirds power of their volumes. When applied to our data, no difference between the DR and control groups was observed.

Adjustment of metabolic rates during DR by ANCOVA

The last method of adjustment, considered the more statistically appropriate, is the ANCOVA. This approach was adopted to overcome the weakness of the ratio method that aimed to control the influence of the variable in the denominator and thus adjust data. However, as mentioned above, the validity of the ratio adjustment implies that there is a linear relationship between the numerator and the denominator and that the intercept is zero. This last condition is the necessary and sufficient condition for the effect of the denominator to be removed when using the ratio. Obviously, the intercept is rarely zero in physiology, and especially between metabolic rates and most determinants. This appears clearly from the data of the present study (Fig. 1), and this is why regression based approaches (ANCOVA) have been proposed as alternatives (34). From a practical point of view, the ANCOVA is used to test the main and interaction effects of categorical variables (control vs. DR) on a continuous dependent variable (i.e. TEE or REE), controlling for the effects of selected other continuous variables, which covary with the dependent (i.e. FFM or body weight; Ref. 35). Several assumptions and requirements, which are beyond the present discussion, are needed to perform the ANCOVA. Basically, this results in individual values of metabolic rates adjusted (i.e. weighed) for a group mean of FFM, providing that FFM has been shown to be the most significant determinant of TEE by regression analysis [i.e. TEE _{FFM adjusted} = TEE _individual + slope (FFM _mean − FFM _individual)]. This method has never been applied in rodents during DR protocols. In humans, Weyer et al. (8) showed that after 2 yr of DR during the Biosphere experiment, TEE adjusted for age, sex, FFM, and fat mass was lower compared with a measure performed 6 months after the end of the protocol. The same depression of TEE and REE adjusted for FFM was also observed after a stabilized weight loss of 10% (6). In monkeys, Ramsey et al. (22) after 30 months of DR observed no changes in TEE adjusted for FFM, whereas REE was reduced. As previously reported, at later assessments, this REE difference disappeared (21). DeLany et al. (19) after 10 yr of weight clamping noted a reduction in TEE adjusted either for FFM or body weight. In the present experiment, TEE adjusted for body weight was not significantly different between groups. However, REE adjusted for FFM was reduced in the DR group compared with the control one, as reported at 30 months in the University of Wisconsin study (22).

Inter- and intraspecies effects of DR after common adjustment of the literature data

From the above review, two points emerged. First, the responses of EE to DR are clearly nonhomogeneous either within or between any metabolic rate adjustment methods. Second, the ANCOVA appears to be the most statistically appropriate method of adjustment of metabolic rate. However, the recent use of this method does not allow for comparison of sufficient data to test the effect of DR on TEE.

To analyze the published data in a uniform manner, we performed a review of the studies that investigated the TEE adaptation to DR in humans, monkeys, and rats. Any human study that included overweight or obese subjects was excluded. Similarly, we excluded studies in which initial and final body weights or initial and final TEE were not mentioned or not deducible with good confidence. Lastly, the DR studies in which weight loss was not stabilized were excluded to avoid the confounding effects of continued weight loss. In rodents, we chose a cut-off value of 30 d as a minimal duration of DR (3% of the animal life span). Only four studies in humans (3, 6–8), three in monkeys (19, 20, 22), and seven in rats (9, 10, 12, 13, 15, 16, 18) met our criteria.¹ A repeated measures ANCOVA was used to test the hypothesis that the DR-induced changes in TEE are independent of the DR-induced changes in body weight. To allow interspecies comparison, we represented the logarithm of TEE as a function of the logarithm of body weights (Fig. 2). For this interspecies comparison, we indeed observed that DR induces a decrease in TEE adjusted for body weight. The mean adjusted decrease in TEE associated with DR was −232 ± 144 kJ/d [mean ± sd, 95% confidence interval (CI), −171, −293 kJ/d] with a P value below 0.0001. Because changes in TEE are explained by modification in its components, we applied the same statistical approach based on resting metabolic rate and FFM. Unfortunately, this reduces the number of experimental groups to eight in rodents (12, 13, 16), two in monkeys (22 and the present study), and three in humans (Refs. 6–8 ; Fig. 3). Across species, the repeated measures ANCOVA showed that DR-induced reduction in TEE is mainly explained by a significant −196 ± 71 kJ/d (95% CI, −153, −239 kJ/d; P < 0.0001) reduction in REE, independent of the changes in FFM. These results are of the same magnitude to what we observed in the present study.

Figure 2.

Review of studies on the effects of DR (>30 d) on the TEE. Data on monkeys (four studies, including the present one), rodents (seven studies), and humans (four studies excluding overweight and obese subjects) are included. These are presented as the logarithm of TEE vs. the logarithm of body weight to allow across-species comparisons. Also represented are data from the recent review of Nagy et al. (36 ) on the free-living metabolic rate relationship with body weights across species, which can be viewed as an update of the initial work of Kleiber (33 ). Because this review only deals with animals, the relationship was weighted by adding data on humans taken at three different age ranges (1–6, 30–39, and >75 yr) from the review of Black et al. (37 ). From a repeated measures ANCOVA, no differences were noted between the slopes (F = 0.40; P = 0.53) and the intercepts (F = 0.74; P = 0.38) of the control diet (n = 24; Y = 0.28 + 0.78·X; R² = 0.97) and the DR groups (n = 24; Y = 0.14 + 0.81·X; R² = 0.96). Adjusted TEE estimates for body weight were significantly different between control diet and DR groups [mean difference, −232 ± 144 kJ/d (mean ± sd); 95% CI, −171, −293 kJ/d; P < 0.0001]. Note, however, the slope difference with the free-living animals, suggesting that the cost of living in wildlife is higher than the cost of living in a laboratory setting.

Figure 3.

Similar to Fig. 2, the logarithm of the REE is plotted vs. the logarithm of FFM from data on monkeys (two studies, including the present one), rodents (three studies), and humans (three studies excluding overweight and obese subjects). By ANCOVA, no differences were noted between the slopes (F = 0.14; P = 0.72) and the intercepts (F = 0.66; P = 0.43) of the control diet groups (n = 13; Y = 0.44 + 0.73·X; R² = 0.98) and the energy-restricted groups (n = 13; Y = 0.32 + 0.74·X; R² = 0.99). Adjusted REE estimates for FFM were significantly lower after energy restriction (mean difference = −196 ± 71 kJ/d; 95% CI, −153, −239 kJ/d; P < 0.0001).

Lastly, it has been suggested that the duration and severity of DR may explain the discrepancies of TEE responses between studies (1). We tested this hypothesis by plotting the TEE adjusted for body weight and the REE adjusted for FFM against either the duration or the severity of DR. We did not observe an effect of time/severity with the data sets used for this regression analysis (Fig. 4).

Figure 4.

Comparison across studies and species of the relationship between the differences in adjusted metabolic rates (energy-restricted vs. control diet) and the logarithm of the duration [expressed in percentage of maximal life span of humans (38 ), monkeys (24 ), and rodents (38 )] and severity of energy restriction. REE_FFM, REE adjusted for FFM; TEE_weight, TEE adjusted for body weight. See Fig. 2 legend for details on the reviewed studies.

Conclusion

The present experiment suggests that 1) rhesus monkeys subjected to 11 yr of DR presented a reduced TEE attributable to a 250 kJ/d drop in REE independent of the DR-induced reduction in FFM, and 2) the discrepancies in the literature regarding the role TEE may play in the action of DR may be partly attributable to different and often inappropriate methods of adjusting metabolic rate for body size. By performing a statistically suitable and uniform adjustment of metabolic rates on the previously published DR experiments in humans, monkeys, and rodents, we observed that the variability in TEE response to DR often hides significant changes in REE that are independent of FFM (−196 ± 71 kJ/d). Inexplicably however, the reduction in adjusted REE within the Wisconsin DR colony has not been detected throughout study. This demonstrates a need for further longitudinal testing to evaluate whether this represents an interaction with aging. Taken together, these results suggest that long-term DR can reduce REE independent of the changes in the lean tissue mass. Whether or not this decrease in adjusted REE plays a role in the life extension action of DR warrants further experiments.

Acknowledgements

This study is supported by National Institutes of Health Grants PSI RR00167 and PO1 AG11915. This publication is the 42-008 of the Wisconsin Regional Primate Research Center and 03-01 of the Veterans Administration Hospital, Geriatric Research, Education and Clinical Center of Madison.

1

A summary table of the data is available upon request to the authors.

Abbreviations:

 
• ANCOVA,

  Analysis of covariance;

 
• CI,

  confidence interval;

 
• DLW,

  doubly labeled water;

 
• DR,

  dietary restriction;

 
• DXA,

  dual x-ray absorptiometry;

 
• EE,

  energy expenditure(s);

 
• FFM,

  fat-free mass;

 
• NB-EE,

  nonbasal EE;

 
• REE,

  resting EE;

 
• TBW,

  total body weight;

 
• TEE,

  total EE.