Age-Dependent Changes in Resting Energy Expenditure (REE): Insights from Detailed Body Composition Analysis in Normal and Overweight Healthy Caucasians

Age-related changes in organ and tissue masses may add to changes in the relationship between resting energy expenditure (REE) and fat free mass (FFM) in normal and overweight healthy Caucasians. Secondary analysis using cross-sectional data of 714 healthy normal and overweight Caucasian subjects (age 18–83 years) with comprehensive information on FFM, organ and tissue masses (as assessed by magnetic resonance imaging (MRI)), body density (as assessed by Air Displacement Plethysmography (ADP)) and hydration (as assessed by deuterium dilution (D2O)) and REE (as assessed by indirect calorimetry). High metabolic rate organs (HMR) summarized brain, heart, liver and kidney masses. Ratios of HMR organs and muscle mass (MM) in relation to FFM were considered. REE was calculated (REEc) using organ and tissue masses times their specific metabolic rates. REE, FFM, specific metabolic rates, the REE-FFM relationship, HOMA, CRP, and thyroid hormone levels change with age. The age-related decrease in FFM explained 59.7% of decreases in REE. Mean residuals of the REE-FFM association were positive in young adults but became negative in older subjects. When compared to young adults, proportions of MM to FFM decreased with age, whereas contributions of liver and heart did not differ between age groups. HOMA, TSH and inflammation (plasma CRP-levels) explained 4.2%, 2.0% and 1.4% of the variance in the REE-FFM residuals, but age and plasma T3-levels had no effects. HMR to FFM and MM to FFM ratios together added 11.8% on to the variance of REE-FFM residuals. Differences between REE and REEc increased with age, suggesting age-related changes in specific metabolic rates of organs and tissues. This bias was partly explained by plasmaT3-levels. Age-related changes in REE are explained by (i) decreases in fat free mass; (ii) a decrease in the contributions of organ and muscle masses to FFM; and (iii) decreases in specific organ and tissue metabolic rates. Age-dependent changes in the REE-FFMassociation are explained by composition of FFM, inflammation and thyroid hormones.


Introduction
Resting energy expenditure (REE) decreases from young to old age by 1% to 2% per decade [1]. This is partly explained by age-related decreases in fat free mass (FFM) [2]. FFM accounts for 50%-70% Table 1. Physical characteristics of the main study population (n = 714) and subgroup with detailed body composition (n = 369). Body composition was assessed by Air Displacement Plethysmography (ADP). ADP was performed by the BOD POD ® device (Cosmeds.r.l., Rome, Italy). Participants wore tight-fitting underwear and a swim cap. Two repeated measurements of body volume were performed, averaged and corrected for predicted body surface area and thoracic gas volume using BOD POD ® software (version 4.5.0). Percentage fat mass (FM ADP ) was calculated from body density using the equation by Siri et al. [21]. Fat free mass (FFM ADP ) was calculated as body weight minus FM ADP . Total body water was assessed by dilution method as previously described in detail [22] and used to calculate the hydration of FFM ADP .
In a subgroup of 369 subjects with BMI < 30 kg/m 2 muscle mass (MM), total, subcutaneous and visceral adipose tissue and organ masses were measured using whole body multislice MRI. Scans were obtained with a 1.5T scanner (Magnetom Vision Siemens, Erlangen, Germany) as previouslydescribed [23]. Areas and volumes of MM, adipose tissue and volumes of 5 internal organs (brain, heart, liver, spleen and kidney) were manually analyzed using the SliceOmatic software (version 4.3; Tomovision, Montreal, QC, Canada) as described earlier [8]. Intra-observer coefficient of variation (CV) was 1.8% for total SM, 1.8% for brain, 0.07% for liver, 1.7% for heart and 1.0% for kidney.
DXA whole body measurement was performed (QDR4500A, Hologic Inc., Bedford, MA, USA). Subjects laid in supine position during the 10 min scan. Manufactures software (version V8.26a:3, Hologic Inc., Bedford, MA, USA) was used for analysis of bone mineral content (BMC DXA ). Skeletal bone massDXA was calculated as BMC DXAˆ1 .85 and included in calculations of resting energy expenditure [24].
Statistical analysis was performed using SPSS statistical software (SPSS 22.0, Inc., Chicago, IL, USA). All data are given as median+interquartil range (IQR). Differences between women and men were tested by Mann-Whitney U-test and between age groups using the Kruskal-Wallis-test. To assess differences of age-related changes in organ masses and muscle mass in relation to FFM, a general linear regression model was used with FFM as a dependent variable. Linear regression models examined the association between residuals of the FFM to REE relationship, and differences between REE and REEc, HOMA, CRP and thyroid hormones. A paired t-test was used to analyze the significance of the difference (∆) between REE and REEc, and the differences were also plotted against age. A p-value < 0.05 was accepted as the limit of significance.

Body Composition
Body composition characteristics of the study population are presented in Table 1. Men were heavier and taller than women, and they had higher BMI, FFM and organ masses. Visceral adipose tissue volumes (VAT) were higher in men, whereas total adipose tissue was higher in women. In both sexes, there were age-related decreases in the individual components of FFM. By contrast, in men, heart mass and VAT increased with age. Relative amounts of tissue and organ masses as part of FFM showed an age-related decrease of muscle and spleen mass in women and men. In contrast, masses of heart, brain, liver, kidneys as well as bone mineral in relation to FFM, increased during aging, which could be explained by co-occurring FFM decline. Compared to men, women had a lower body density but higher FFM hydration. In both sexes, body density decreased with increasing age; in contrast, FFM hydration increased in women only.

REE
There were significant age and sex effects on REE, REE FFM and the difference between REE to REEc (∆REE-REEc), as calculated using the age-unspecific algorithm of Elia et al. (Table 2; see Methods). In contrast, no sex and age-differences in the ∆REE-REEc were observed using the age-specific algorithms of Wang et al. (Table 2; see Methods). Compared with women, men had higher levels of free triiodothyronineand lower levels of thyroid-stimulating hormone as well as free thyroxin. In women, only free thyroxin levels showed an age-related increase. This was also true in men, whereas thyroid-stimulating hormone and free triiodothyroninedecreased with age in men.
REE increased with FFM ( Figure 1A); however, REE related to FFM decreased with age. Comparing different age groups, residuals of the REE-FFM relationship turned from positive to negative residuals with increasing age ( Figure 1B). The ratio of muscle mass (as major low metabolic rate tissue) to high metabolic rate organ masses (HMR), and thus the relative proportion of MM to HMR, increased with FFM ( Figure 2A) and decreased nonlinearly with age ( Figure 2B).
The ratio of MM/FFM increased with increasing FFM, whereas the HMR/FFM ratio remained stable. The association between FFM, MM/FFM and HMR/FFM in young (18-39 years) middle-aged (40-59 years) and older (ě60 years) subjects showed no significant differences between the age groups. Comparing age groups, the relationship between FFM to MM/FFM became weaker (R 2 between 0.11 and 0.06) but FFM to HMR/FFM relationship remained constant (R 2 between 0.65 and 0.50). However, age affected the relationship of individual organ masses (i.e., heart, spleen and skeletal muscle) to FFM (Figure 3).
Differences between REE and REEc were plotted against age (Figure 4). Both REEc calculated according to either Elia or Wang et al. (see Methods) showed a significant bias of REE with age. This age bias was not explained by whole body density or FFM hydration. Plasma levels of free triiodothyronine ffected the relationship of individual organ masses (i.e., heart, spleen and skeletal muscle) to Figure 3). ifferences between REE and REEc were plotted against age (Figure 4). Both REEc calculated ing to either Elia or Wang et al. (see Methods) showed a significant bias of REE with age. This ias was not explained by whole body density or FFM hydration. Plasma levels of free othyronine explained 4.7% (Elia) and 2.5% (Wang et al.) of the variance in REE and REEc bias.   Table 3 shows cardiometabolic risk factors. Overall, there were no sex differences in biomar inflammation and insulin resistance. Plasma levels of C-reactive protein significantly increase age in women only. In contrast, an age-associated increase in HOMA index was only observed i

Cardiometabolic Risk
In multivariate regression analysis variance of REE-FFM, residuals were explained by H (4.2%), CRP (2.0%) and TSH (1.4%). However, MM/FFM and HMR/FFM ratios accounted for of REE-FFM residuals, whereas age added 1.9% to their variance.   Relationship between FFM and the ratio of muscle mass (MM) to high metabolic rate organs (HMR) per kg (A) and between the MM/HMR-ratio and age (B). HMR is the sum of masses of brain, heart, liver and kidneys.FFM was assessed by ADP, organ masses were assessed by whole body Magnetic Resonance Imaging (MRI) (for details, see Methods) (n = 369). Table 3 shows cardiometabolic risk factors. Overall, there were no sex differences in biomarkers of inflammation and insulin resistance. Plasma levels of C-reactive protein significantly increased with age in women only. In contrast, an age-associated increase in HOMA index was only observed in men.

Figure 2.
Relationship between FFM and the ratio of muscle mass (MM) to high metabolic rate organs (HMR) per kg (A) and between the MM/HMR-ratio and age (B). HMR is the sum of masses of brain, heart, liver and kidneys.FFM was assessed by ADP, organ masses were assessed by whole body Magnetic Resonance Imaging (MRI) (for details, see Methods) (n = 369).

Figure 1.
Age-dependent decrease in fat free mass (FFM)-resting energy expenditure (REE) relationship (A) and their residuals in different age groups (median; 95% CI) (B). Significant differences between age-groups are indicated by # as tested by Kruskal-Wallis-test (n = 714). FFM was assessed by Air Displacement Plethysmography (ADP) (for details, see Methods).

Figure 2.
Relationship between FFM and the ratio of muscle mass (MM) to high metabolic rate organs (HMR) per kg (A) and between the MM/HMR-ratio and age (B). HMR is the sum of masses of brain, heart, liver and kidneys.FFM was assessed by ADP, organ masses were assessed by whole body Magnetic Resonance Imaging (MRI) (for details, see Methods) (n = 369). Table 3 shows cardiometabolic risk factors. Overall, there were no sex differences in biomarkers of inflammation and insulin resistance. Plasma levels of C-reactive protein significantly increased with age in women only. In contrast, an age-associated increase in HOMA index was only observed in men.

Cardiometabolic Risk
In multivariate regression analysis variance of REE-FFM, residuals were explained by HOMA (4.2%), CRP (2.0%) and TSH (1.4%). However, MM/FFM and HMR/FFM ratios accounted for 11.8% of REE-FFM residuals, whereas age added 1.9% to their variance.   . Age-dependency of differences between measured and calculated resting energy expenditure (REE-REEc). REEc was calculated from organ/tissue masses (as assessed by whole body MRI times organ-and tissue-mass-specific metabolic rates as published by Elia (circles) and Wang (triangles, i.e., using age-adjusted specific metabolic rates) (n = 217) (for details, see Methods).  . Age-dependency of differences between measured and calculated resting energy expenditure (REE-REEc). REEc was calculated from organ/tissue masses (as assessed by whole body MRI times organ-and tissue-mass-specific metabolic rates as published by Elia (circles) and Wang (triangles, i.e., using age-adjusted specific metabolic rates) (n = 217) (for details, see Methods).

Figure 4.
Age-dependency of differences between measured and calculated resting energy expenditure (REE-REEc). REEc was calculated from organ/tissue masses (as assessed by whole body MRI times organ-and tissue-mass-specific metabolic rates as published by Elia (circles) and Wang (triangles, i.e., using age-adjusted specific metabolic rates) (n = 217) (for details, see Methods). Table 3 shows cardiometabolic risk factors. Overall, there were no sex differences in biomarkers of inflammation and insulin resistance. Plasma levels of C-reactive protein significantly increased with age in women only. In contrast, an age-associated increase in HOMA index was only observed in men.  In multivariate regression analysis variance of REE-FFM, residuals were explained by HOMA (4.2%), CRP (2.0%) and TSH (1.4%). However, MM/FFM and HMR/FFM ratios accounted for 11.8% of REE-FFM residuals, whereas age added 1.9% to their variance.

Discussion
Both REE and the relationship between REE and FFM decreased with age [25]. The decrease in REE is explained by a reduction in FFM as well as by changes in the composition of FFM in normal and overweight subjects (Table 1, Figure 3). With increasing FFM, the proportion of HMR to FFM remained relatively constant, whereas the MM to FFM ratio decreased with age. Considering individual organ and tissue masses, heart mass and spleen masses per FFM increased with age (Figure 3). In a previous study of our group, Bosy-Westphal et al. [3] had already described that heart mass explained 58% of the variance in the difference between REE and REEc in elderly subjects. A higher heart mass with age was seen as a compensatory effect on chronic hypertensive load or hypertrophy of cardiocytes.
In addition to the age-related (i) decreases in FFM and (ii) changes in the proportion of organ/tissues mass to FFM, alterations of specific metabolic rates of individual organs and tissues add to age-related changes in REE. To address this issue, REE was calculated (REEc) based on organ and tissue masses times specific tissue metabolic rates. Elia [27] was first to publish metabolic rates of individual organs and tissues assuming constant values across lifespan. By contrast, Gallagher et al. [4] showed that specific metabolic rates based on data of younger subjects do not resemble the respective metabolic rates in elderly subjects. In a previous work, Wang et al. [13] calculated that specific metabolic rates of organs and tissues changed with age resulting in corrections of Elias constants. Anyhow, the present data revealed that the differences between REE and REEc increased with age. This was obvious for both the Elia and Wang predictions (Figure 4). The bias between REE and REEc could not be explained by the age-related changes in whole body density or FFM hydration. In our opinion, this supports the idea that the physical property of organ and tissue masses are not related to age-related changes in REE.
In the Baltimore Longitudinal Study of Aging, Ruggiero et al. [14] and Fabbri et al. [16] have already shown that higher resting metabolic rates in age were associated with multi-morbidity and mortality. We now add the finding that age-related changes in the REE-FFM-association (i.e., the REE-FFMresiduals) were related to inflammation. Thus, the variance in the FFM to REE relationship could reflect health status.
A limitation of this study was the assumption of age-independent constant organ and tissue densities that affects the estimate of organ and tissue masses from their volumes. In addition, we have used cross-sectional data only; thus, our data cannot give a future prospect on individual aging. The fact that data analysis included normal and overweight subjects (BMI < 30 kg/m 2 ) only could be seen as critical, but this cut off was chosen to cover current normal BMI ranges for all young as well as older adults.

Conclusions
In conclusion, age-related changes of REE relate to decreases in FFM as well as alterations in the REE-FFM relationship. In addition, proportional changes in FFM composition (i.e., the organ and tissue mass ratios to FFM) plus decreases in specific metabolic rates of organs and tissues add to the decrease in REE with age. The variance in the REE-FFM association is related to inflammation.