Lifestyle‐Intervention‐Induced Reduction of Abdominal Fat Is Reflected by a Decreased Circulating Glycerol Level and an Increased HDL Diameter

Scope Abdominal obesity is one of the main modifiable risk factors of age‐related cardiometabolic disease. Cardiometabolic disease risk and its associated high abdominal fat mass, cholesterol, and glucose concentrations can be reduced by a healthier lifestyle. Hence, the aim is to understand the relation between lifestyle‐induced changes in body composition, and specifically abdominal fat, and accompanying changes in circulating metabolic biomarkers. Methods and results Data from the Growing Old Together (GOTO) study was used, which is a single arm lifestyle intervention in which 164 older adults (mean age 63 years, BMI 23–35 kg/m2) changed their lifestyle during 13 weeks by 12.5% caloric restriction plus 12.5% increase in energy expenditure. It is shown here that levels of circulating metabolic biomarkers, even after adjustment for body mass index, specifically associate with abdominal fat mass. The applied lifestyle intervention mainly reduces abdominal fat mass (−2.6%, SD = 3.0) and this reduction, when adjusted for general weight loss, is highly associated with decreased circulating glycerol concentrations and increased HDL diameter. Conclusion The lifestyle‐induced reduction of abdominal fat mass is particularly associated, independent of body mass index or general weight loss, with decreased circulating glycerol concentrations and increased HDL diameter.


Introduction
Abdominal obesity plays an important role in the development of cardiometabolic disease risk. [1,2] People with relatively high amounts of abdominal fat are characterized by increased insulin resistance [3] and a detrimental circulating metabolic biomarker profile encompassing high levels of glucose, cholesterol, and triglycerides, [4] all of which are known to be associated with type 2 diabetes and cardiovascular disease. [5][6][7][8][9][10][11][12] Reduction of cardiovascular risk can be achieved by lifestyle interventions aimed at increasing physical activity and/or reducing caloric intake. [13,14] Because abdominal fat is intimately linked to disease risk, it is imperative to understand the relation between the lifestyle-induced changes in body composition, specifically abdominal fat, and the accompanying changes in metabolic biomarkers.
To gain more insight than what would be achieved by only measuring the standard metabolic clinical chemistry parameters, such as cholesterol and glucose levels, state-of-the art 1 H-NMR metabolomics platforms have been used to investigate the relationship between metabolism and body composition. In young people (age between 25 and 30 years), a larger amount of abdominal fat has been associated with an unfavorable lipoprotein profile (i.e., high VLDL, IDL, LDL and small HDL particle concentrations, high IDL-and LDL-cholesterol, triglycerides, ApoB and ApoB to ApoA1 ratio and low large HDL particle concentration, HDLcholesterol and small HDL diameter). [5] In general, an unhealthy metabolic profile, as measured by the 1 H-NMR platform, can be improved by a lifestyle change [13][14][15][16] that is known to particularly reduce the amount of abdominal fat mass. [17][18][19] However, it remains unclear how and to what extent lifestyle-induced changes in body composition, specifically the reduction in abdominal fat, are reflected by circulating metabolic biomarkers.
We investigated the relation between lifestyle-induced changes in body composition and the altering blood metabolome, by exploring the data collected in the Growing Old Together study (GOTO); a 13-week-lifestyle intervention study in which older participants (N = 164, Age mean = 63 years old (age range 49-75 years), body mass index (BMI) mean = 27 (BMI range 23-35 kg m −2 ) at the moment of inclusion) increased physical activity by 12.5% and decreased energy intake by 12.5%. [13] Body composition parameters were measured with anthropometrics and a dual X-ray absorptiometry (DXA) scan, while metabolic biomarkers were measured in serum using 1 H-NMR metabolomics, both before and after the intervention. First, at baseline we crosssectionally correlated body composition measures with circulating metabolic biomarkers levels. Second, we determined the effect of the GOTO lifestyle intervention on 1 H-NMR metabolomic biomarkers. Third, we determined how body composition measures were affected by the lifestyle intervention. Finally, we investigated the associations between the change in multiple measures of body composition and the changes in metabolic biomarkers to determine which of these biomarkers reflected the alterations in body composition by a lifestyle change.

Study Design
The single arm GOTO lifestyle intervention study has previously been described by van de Rest et al. [13] The Medical Ethical Committee of the Leiden University Medical Center approved the study (P11.187) and all participants signed a written informed consent. All experiments were performed in accordance with relevant and approved guidelines and regulations. This trial was registered at the Dutch Trial Register (http://www.trialregister.nl) as NTR3499.
In short, the single arm lifestyle intervention comprised 13 weeks of 25% lowered energy balance by 12.5% reduction in energy intake and 12.5% increase in physical activity under supervision of a dietician and a physiotherapist. Participants were recruited between February and October 2012 within the Leiden Longevity Study, [20] consisting of a member of a long-lived family and their partner, and as a couple from the same household they participated in the GOTO study. In case one of the two was not eligible to participate, single members of a long-lived family or just volunteers were included to obtain the required sample size. Participants (N = 164) were between 46 and 75 years (mean age 63 years), had a BMI between 23 and 35 kg m −2 (mean BMI = 27 kg m −2 ), no diabetes (fasting glucose <7.0 mmol L −1 ), or any disease or condition that seriously affects body weight and/or body composition including active types of cancer (Table S1, Supporting Information).
The participants provided a report of their pharmacist about their current medication use, from which the use of lipid lowering medication (fibrates, niacin, bile acid sequestrants, 3hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors) and hypertension medication (diuretics, beta-blockers, calcium channel blockers, agents acting on the renin-angiotensin system) were deduced.
In the present paper the analyses were performed on the subgroup of 132 participants for whom data on anthropometrics, DXA measures, and NMR metabolomics (Nightingale Health) were available at baseline as well as at the endpoint of the study ( Table 1).

Body Composition Measurements
Data were available for seven anthropometric measures based on weight, height, waist circumference, and hip circumference. Weight was measured to the nearest 0.1 kg using a digital personal scale (Seca Clara 803 scale, Seca Deutschland, Hamburg, Germany) with the person dressed in light clothing and without shoes. Height, waist circumference (midpoint between the lowest rib and the top of the iliac crest), and hip circumference (largest circumference of buttocks) were measured to the nearest 0.1 cm with a non-elastic tape in standing position without shoes. BMI was calculated using the Quetelet index: weight(kg)/(height(cm)) 2 . Waist hip ratio is the ratio of waist circumference (cm) over hip circumference (cm), and waist height ratio is the ratio of waist circumference (cm) over height (cm).
Eleven body composition features were measured using whole-body DXA (Discovery A, Hologic Inc., Bedford, MA,USA): whole body lean mass in kilogram, whole body fat in kilogram and percentage of whole body weight, trunk fat in kilogram and percentage of trunk weight, android fat in kilogram and percentage of android weight, genoid fat in kilogram and percentage of genoid weight, leg fat in kilogram and percentage of leg weight. In addition, six ratios were calculated: trunk fat over whole body fat ratio, android fat over whole body fat ratio, gynoid fat over whole body fat ratio, leg fat over whole body fat ratio, android fat over gynoid fat ratio, whole body fat over whole body lean mass ratio ( Figure S1, Supporting Information).
A detailed description of the DXA measurement and an indication of the trunk, android, and genoid body regions can be found in the Supporting Information.

Metabolic Biomarker Profiling
Blood collection took place between 8 and 9 a.m. after at least 10 h of fasting. Metabolic biomarkers were quantified from serum samples of 164 individuals using high-throughput www.advancedsciencenews.com www.mnf-journal.com  Figure S1, Supporting Information, for recognition of body regions in DXA images. SD, standard deviation.
1 H-NMR metabolomics (Nightingale Health Ltd, Helsinki, Finland). Details of the experimentation and applications of the NMR metabolomics platform have been described previously. [21] This method provides simultaneous quantification of routine lipids, lipoprotein subclass profiling with lipid concentrations within 14 subclasses, fatty acid composition, and various lowmolecular metabolites including amino acids, ketone bodies, and gluconeogenesis-related metabolites in molar concentration units. The 14 lipoprotein subclass sizes were defined as follows: extremely large VLDL with particle diameters from 75 nm upward and a possible contribution of chylomicrons, five VLDL subclasses (average particle diameters of 64.0, 53.6, 44.5, 36.8, and 31.3 nm), IDL (28.6 nm), three LDL subclasses (25.5, 23.0, and 18.7 nm), and four HDL subclasses (14.3, 12.1, 10.9, and 8.7 nm). The mean size for VLDL, LDL, and HDL particles was calculated by weighting the corresponding subclass diameters with their particle concentrations. Due to the high correlation among the metabolic biomarkers, only the 65 biomarkers were analyzed that were previously ex-plored for cardiovascular risk by Würtz et al. [22] to enhance interpretability. The selection of these biomarkers was based on previous studies using this platform and the current list comprised the total lipid concentrations, fatty acid composition, and lowmolecular-weight metabolites, including amino acids, glycolysisrelated metabolites, ketone bodies, and metabolites involved in fluid balance and immunity (Table S2, Supporting Information).

Statistical Analysis
For the following metabolic biomarkers, serum levels were below the detection level for at least one measurement: lipid concentration in Extremely Large VLDL (2.3%), Very Large VLDL (3.0%), Large VLDL (1.5%), and Large HDL (2.3%), and these values were considered as missing (Table S2, Supporting Information). All metabolic biomarkers were LN-transformed and consecutively Zscaled (resulting in a mean of 0 and a standard deviation of 1). To be able to compare the effects of body composition parameters, all measurement levels were Z-scaled.
To confirm association between metabolic biomarker levels in plasma and body composition parameters the baseline data were investigated. Because the participants were couples that lived together in the same household, there may be a random effect influencing the metabolic biomarker levels by their lifestyle. To be able to model the household of the participants as a random effect, a mixed effects model at baseline was used. Partial correlation of metabolic biomarkers and body composition parameters at baseline was determined using a linear mixed model adjusted for age, gender, status (longevity family member or partner), lipid lowering medication, hypertension medication (fixed effects), and household (random effect) with the body composition parameters as outcome. A random effect for household was included to account for the potentially increased similarity among household members (85% belong to a couple sharing a household, i.e., 56 couples in the study), as they generally share diet and other lifestyle factors.
To determine the partial correlation of the change in the metabolic biomarker levels and the change in the body composition parameters after the intervention, a linear mixed model was used with the metabolic biomarker levels as outcome and body composition as determinant adjusted for age, gender, status (longevity family member or partner), lipid lowering medication, hypertension medication (fixed effects), household, and individual (random effects). For additional analyses, weight was added to the model to determine general weight loss-independent effects.
All statistical analyses were performed with STATA/SE 13.1 and heatmaps were generated using the heatmap.2 function of the gplots package in R. Since 65 metabolic biomarkers and 22 body composition phenotypes were tested p < 3.5 × 10 −5 (0.05/(65 × 22)) was considered as significant after adjustment for multiple testing.

Study Population
The current investigation of the relation between changing body composition and circulating metabolic biomarkers was performed in a representative subgroup of 132 participants of the GOTO study of whom body composition measures and 1 H-NMR circulating metabolic biomarkers were available before and after the intervention ( Table 1). The mean age of the study participants was 63 years (range 46-75 years), they had a mean BMI of 27 kg m −2 (SD 2.4) and 18 participants (11%) were obese (BMI>30 kg m −2 ).

Abdominal Fat Associates with Circulating Metabolic Biomarkers and Body Composition at Baseline
In order to compare our findings with those from a previous study in younger individuals, [5] we first investigated the association between body composition parameters and circulating metabolic biomarkers of the GOTO study at baseline ( Figure S2, Supporting Information). We confirmed that body composition features, especially large amount of abdominal fat measures, mainly associated with smaller HDL diameter, higher VLDL particle concentrations, and higher circulating levels of triglyceride and glycoprotein acetyls ( Figure S2, Supporting Informa-tion). In contrast, in the GOTO study that consisted of older adults, a larger amount of abdominal fat mass was additionally associated with higher circulating concentrations of glycerol and 3-hydroxybutyrate. Furthermore, we observed stronger associations with the DXA fat measures than with the anthropometrics parameters of body composition.
We subsequently investigated whether the associations between body composition and circulating metabolic biomarker levels would still hold after adjustment for BMI. Figure 1 shows a heatmap of the partial correlation (adjusted for BMI) between metabolic biomarkers and body composition parameters at baseline. The hierarchical clustering on basis of the partial correlations between body composition measures and metabolic biomarkers, clusters body composition parameters roughly into five clusters (Figure 1  circulating metabolic biomarkers. Since the lower body fat measures themselves do not show any association with metabolic biomarkers, the latter association seems to be driven by the whole body fat measure. Hence, after adjustment for BMI, we observed that a higher percentage of fat in the trunk or android body regions (abdominal fat) associated with a lower concentration of lipids in (extra) large HDL particles and smaller HDL diameter, a higher concentration of lipids in (extra) large VLDL particles, and higher circulating levels of leucin, isoleucine, serum triglycerides, glycoprotein acetyls, 3-hydroxybutyrate, and glycerol.

Lifestyle Change Reduces CVD-Associated Metabolic Biomarkers and Abdominal Fat
Second, we investigated which of the 65 circulating metabolic biomarkers changed in response to the intervention. In total 46 metabolic biomarkers changed significantly due to the intervention and the most prominent effects were observed for LDL and VLDL subclass concentrations, and levels of apoB, monounsaturated fatty acids, triglycerides, and cholesterol (Table S3, Supporting Information). The metabolic biomarkers responding to the lifestyle intervention combining less caloric intake and more physical activity, clearly responded in the direction of lower risk for cardiovascular disease (CVD) as reported by Würtz et al. [6] We also investigated which DXA measure of body composition changed in response to the lifestyle intervention. Both men and women reduced their whole body fat with 1.5% (IQR = −0.5 to −2.6%) (Figure 2). As expected, android fat and trunk fat reduced most in both women (−2.4% (IQR = −0.5 to −4.7%) and −2.1% (IQR = −0.6 to −3.4%), respectively) and men (−2.9% (IQR = −0.9 to −5.0%) and −2.3% (IQR = −1.0 to −3.6%), respectively). Likewise, waist circumference, waist/hip ratio, and fat/lean ratio decreased similarly in men and women (Table S4, Supporting Information).

Partial Correlations between a Change in Metabolic Biomarkers and Change in Abdominal Fat
To determine whether the change in circulating metabolic biomarker levels can be explained by the reduction in abdominal fat, we investigated whether the change in abdominal fat (Δ abdominal fat) correlates with the change in the levels of 46 metabolic biomarkers (Δ metabolic biomarker) that was altered by the lifestyle intervention (Table S3, Supporting Information), while adjusting for general weight loss. The reduction of android and trunk fat, both measured in grams and the percentage of fat in android and trunk, was most strongly associated, independent of general weight loss, with a decrease in circulating glycerol levels ( Figure S3, Supporting Information). If the android fat mass decreased with 1 SD, glycerol levels decreased with 0.35 SD (p-value = 1.45 × 10 −11 : Table 2). Decreasing ratios of abdominal fat over whole body fat, or android fat over genoid fat were most strongly associated with an increasing HDL diameter. If the trunk fat over whole body mass ratio decreased with 1 SD, the HDL diameter increased with 0.36 SD (p-value = 2.56 × 10 −9 : Table 3).
In addition, although to a lower extent, the levels of lipids in VLDL particles, serum triglycerides, glycoprotein acetyls, www.advancedsciencenews.com www.mnf-journal.com apolipoprotein B, total fatty acids, monounsaturated fatty acids, and leucine decreased when abdominal fat reduced.

Discussion
We investigated the relation between lifestyle interventioninduced changes in body composition, specifically abdominal fat, and the accompanying molecular changes in older adults independent of general weight loss. In young and old individuals abdominal fat mainly associates with a smaller HDL diameter, higher VLDL particle concentrations, and higher circulating levels of triglycerides and glycoprotein acetyls. In older adults abdominal fat is additionally associated with higher circulating levels of glycerol and 3-hydroxybutyrate. Furthermore, we showed that BMI, as a measure of overall adiposity and known to correlate with abdominal fat mass, does not explain all associations between circulating metabolic biomarker levels and abdominal fat. Independent of BMI, HDL diameter, concentration of lipids in VLDL particles, and glycerol levels associate with abdominal fat mass. Specifically, the more abdominal fat, the smaller the HDL diameter, and the lower the concentration of lipids is in HDL particles, and the higher the concentration of lipids in VLDL particles. In addition, if there was more abdominal fat, the circulating levels of glycerol, 3-hydroxybutyrate, leucine, and glycoprotein acetyls were higher. This metabolic biomarker profile associated with abdominal fat indicates a high risk for cardiovascular disease. [6] We next showed that the intervention beneficially affected especially abdominal fat as well as the majority of the tested metabolic biomarkers (46), of which 26 are known to associate with cardiovascular disease. [6] Next, we show that the lifestyle-induced decrease of circulating glycerol levels and increase in HDL diameter can be explained by the loss of abdominal fat. Hence, the lifestyle-induced reduction of abdominal fat in older adults is reflected by decreased circulating glycerol levels and larger HDL diameter.
In older people, measures of BMI or body weight are not able to discriminate with high cardio-metabolic disease risk, that is, low muscle mass and high abdominal fat mass, from those with low risk, that is, high muscle mass and low abdominal fat mass. [23][24][25] Of the people with an average BMI, around 50% has a percentage of body fat that is too high for their age and gender [26] and 30% is metabolically unhealthy. [27] It is known that high body fat and not so much high body weight is associated with an increased risk for cardio-metabolic disease. [28] The relatively simple DXA measures for (abdominal) fat mass would then better be able to identify people with high cardiometabolic risk. Because a lifestyle change that reduces caloric intake and increases physical activity may not be beneficial for each older person, it would be crucial to monitor the metabolic effects of lifestyle interventions aimed at reducing the cardio-metabolic disease risk. We found that the healthy reduction in abdominal fat during a lifestyle change was, independent of general weight loss, reflected in by lower circulating glycerol concentrations and larger HDL diameter. Hence, these 1 H-NMR measures may be further explored to monitor the beneficial effects of a lifestyle change in older people. Circulating glycerol levels and HDL diameter may be valuable tools to monitor cardiometabolic health in older people performing a lifestyle change.
The association of circulating metabolic biomarkers with abdominal fat has been frequently observed in previous studies. [5,29,30] We now showed in older adults, that after adjusting for BMI, glycerol particularly associated with abdominal fat measures. Glycerol is produced by white adipose tissue to dispose of excess glucose [31] leading, via hepatic gluconeogenesis, to an increase in circulating glucose levels. A high level of circulating glycerol is a known biomarker for an increasing risk for hyperglycemia and type 2 diabetes. [11] Increased HDL diameter also reflects reduced abdominal fat independent of BMI, which is in concordance with previous observations. [32] Small, dense HDL subfractions promote cholesterol efflux from foam cell macrophages in the artery wall, [33] which would reduce atherosclerotic lesions. We hypothesize that when there is a large amount of abdominal fat, high levels of cholesterol require large cholesterol efflux to clear the foam cells. Hence, when abdominal fat is reduced, for example by a lifestyle intervention, the cholesterol efflux is lowered and the number of small HDL particles is reduced, resulting in higher overall HDL diameter. This suggests, in combination with our findings that older people during a lifestyle intervention mainly lose abdominal fat and decrease their cardiometabolic disease risk, that cardiometabolic disease risk is influenced by abdominal fat, independent of BMI and general weight loss, via circulating glycerol levels and HDL diameter.
The design of the GOTO study has some limitations. The change in lifestyle was for example not controlled, but guided to be feasible for participants. Because we endeavored a reduction of 12.5% of caloric intake and 12.5% more physical activity, for each of the participants a personalized diet and exercise strategy was proposed. The heterogeneity in the body composition and current lifestyle of the participants at baseline in combination with the personalized intervention strategy, caused large heterogeneity in the response to the lifestyle intervention. Hence, whether the changes in metabolic biomarker levels and abdominal fat were mainly due to the change in dietary pattern, physical exercise, or the combination of both is still to be examined. Another limitation is that the sample size of the GOTO study does not allow for gender stratified analyses. However, the majority of the female GOTO participants is postmenopausal and sex difference in body composition may therefore be limited. Since body composition and the accompanying cardiometabolic disease risk is a serious issue among older people, the older age of the GOTO study participants is advantageous.
In conclusion, the reduction of abdominal fat in older people due to a lifestyle change, is specifically reflected by decreased circulating glycerol concentration and larger HDL particle diameter, independent of general weight loss. Hence, to monitor the beneficial effects of a lifestyle change at older age circulating glycerol concentration and HDL diameter may be valuable tools.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author. agreement number 259679. This study was financially supported by the Netherlands Consortium for Healthy Ageing (grant 050-060-810), in the framework of the Netherlands Genomics Initiative, Netherlands Organization for Scientific Research (NWO); by BBMRI-NL, a Research Infrastructure financed by the Dutch government (NWO 184.021.007, 184.033.111) and by the Netherlands CardioVascular Research Initiative (CVON201-03). J.D. was financially supported by the Alexander von Humboldt Foundation. The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. The authors thank all staff members and Bachelor's and Master's students who contributed to the preparation, design, and performance of this intervention trial and/or assisted on the project. Last but not least, the authors would like to express their gratitude to all participants who did their very best to adhere to the intervention guidelines and underwent all measurements.