Metabolic flexibility differs by body composition in adults

Abstract

Background & Aims:

With the rise of global cardiometabolic diseases, it is important to investigate risk factors such as obesity. Metabolic flexibility, the ability to maintain metabolic homeostasis following an acute challenge, can reflect cardiometabolic health. We investigated the association between body composition and the metabolic flexibility following meal consumption in an adult population.

Methods:

In this study of 1,027 participants (mean age 44.0 y ± SD 4.2 y), we administered a mixed-macronutrient meal challenge. Fasting and two-hour postprandial plasma were assayed for lipids, glycemic, and inflammation biomarkers. We characterized metabolic flexibility through meal-induced biomarker responses (%Δ, the difference between postprandial and fasting, divided by fasting concentration). We then compared the responses by sex-specific tertiles of body mass index (BMI) and percent body fat.

Results:

With every unit (kg/m²) increase in BMI, %Δ (95% confidence interval) increased by 0.17% (0.09, 0.26%) for total cholesterol, 0.31% (0.07, 0.54%) for triglycerides, and 0.11% (0.01, 0.20%) for apoA-I, whereas insulin elevation was reduced (−6.30%; −8.41, −4.20%), and the reduction in leptin was attenuated (0.64%; 0.25, 1.05%). With each unit (percent) increase in body fat, we observed similar changes in the %Δ of total cholesterol and leptin but not in triglycerides, apoA-I, or insulin. Glucose response increased by 0.29% (0.06, 0.51%) as body fat increases by one unit.

Conclusion:

Metabolic flexibility, as assessed by biomarker responses to an acute physiological meal challenge, differed by body composition. These findings may help elucidate the pathways through which obesity contributes to cardiometabolic diseases.

Introduction

Globally, cardiometabolic disease (CMD) mortality is high, with cardiovascular diseases and type 2 diabetes accounting for 17.5 and 1.5 million annual deaths, respectively (1, 2). Low- and middle-income countries are experiencing increasing CMD burden (3). The reasons behind the emerging CMD burden in resource-limited settings is multifaceted, with obesity being a potential determinant (4). Body composition plays an important role in regulating and modifying metabolic processes (5). Circulating concentration of leptin, an adipose tissue-derived pro-inflammation hormone, is postulated to be a key mediator between fat mass and energy metabolism (5, 6).

A feature of the onset and progression of CMDs is cumulative disturbances in cardiometabolic pathways (7). Under normal circumstances, adaptive mechanisms can respond to stressors such as a high-fat and high-sugar diet (8, 9). This ability to maintain homeostasis, which we refer to as metabolic flexibility, has clinical implications (10). Persistent postprandial hyperglycemia and hyperlipidemia may exacerbate cardiometabolic perturbations through mechanisms such as inflammation and oxidative stress (9, 11). In healthy individuals, homeostasis is readily restored with the stimulation of anti-inflammatory and antioxidant processes (12, 13). Metabolic flexibility is therefore a robust measure of cardiometabolic health (14).

A common method to assess metabolic flexibility is a meal challenge followed by evaluation of changes in selected markers such as triglycerides and glucose (15–18). In order to maximize postprandial responses, many challenges involve meals containing large doses of fat, glucose, and calories (19–21). However, we postulate that responses to meals that contain physiological amounts of macronutrients may provide better understanding of the cumulative metabolic disturbances leading to abnormal cardiometabolic conditions. In this study, we assessed the effect of a mixed-macronutrient meal challenge within the physiological range. We quantify metabolic flexibility in this population and test the hypothesis that metabolic flexibility differs by body composition.

Material and methods

Study population

This study was conducted in a cohort of 1,027 free-living men and women in Guatemala (2015–17). Study participants were a subset of individuals who were enrolled, as children, in the (original cohort n=2,392) (22). From the 1,161 traceable, eligible, and consented participants, we excluded individuals who did not attend the clinical exam (n=16), who were not fasted at the time of the clinic visit (n=27), or who did not have postprandial samples (n=85), and pregnant or lactating women (n=6). A detailed participant flow chart was reported elsewhere (23).

Meal challenge procedure

A trained phlebotomist drew venous blood in EDTA-coated tubes from each participant upon confirming fasting status (≥8h). Participants were then given a freshly prepared liquid meal of 259g, consisting of 25g safflower oil, 52g sugar, 12g Incaparina powder (a soy and maize-based protein mixture developed by INCAP), in 170ml lactose-free skim milk. Each 100g provided 164.7kcal (31% from fat), 3.4g protein, 25.2g carbohydrate, and 5.7g fatty acids (including 3.0g monounsaturated and 0.9g polyunsaturated fatty acids), and 1.8mg cholesterol. Composition data were determined from three meal preparations collected randomly throughout the study period (Covance Laboratories Inc.; Madison, Wisconsin). Two hours after consumption of the liquid meal, the phlebotomist drew a second venous blood sample. The protocol for the meal challenge is depicted in Figure 1.

Figure 1.

Flowchart of the Study Procedure

Laboratory methods

Fasting and postprandial blood samples were kept on ice and centrifuged at 3000 RPM for 10min at 4°C within 2h of collection for isolation of plasma. Plasma samples were immediately aliquoted into cryovials and stored at −20°C. Aliquots of pre- and post-prandial plasma were thawed and assayed for glucose concentrations using enzymatic colorimetric methods (Cobas C111 analyzer, ROCHE, Indianapolis, IN, USA) at INCAP. Other frozen aliquots were transferred to a −80°C freezer for storage. Three shipments of frozen samples were transferred on dry ice to Atlanta, GA, US and stored at −80°C until assayed for lipids, insulin, and inflammation markers. The samples were thawed at 4°C in batches; Each batch contained approximately 40 matching pairs (fasting, post-prandial) of samples, selected at random in terms of data collection site, sex, and birth year.

All measurements other than glucose were performed on the AU480 automatic chemistry analyzer (Beckman Coulter Diagnostics, Fullerton CA, US). We assayed total cholesterol (TC) and triglycerides (TG) using enzymatic methods (Sekisui Diagnostics, Burlington, PA, US). We assayed high- and low-density lipoprotein cholesterol (HDLc, LDLc) using homogeneous method (Sekisui Diagnostics, Burlington, MA, US). Apolipoproteins (apoA-I and apoB) were assayed using immunoturbidimetric assay (Sekisui Diagnostics, Burlington, MA, US), and non-esterified fatty acids (NEFA) using calorimetric methods (Wako Chemicals Corporation, Richmond VA, US). Insulin and hsCRP were assayed using immunoturbidimetric method (Sekisui Diagnostics, Burlington, MA, US). Cytokines, including leptin, resistin, monocyte chemoattractant protein-1 (MCP-1), interleukin-10 (IL-10), and adiponectin, were determined in duplicates by ELISA (Boster Biologicals Technology, Pleasanton, CA, USA). Detailed methods were reported elsewhere (23).

Missingness of biomarkers included: 25.8% of IL-10, 12.8% of MCP-1, 7.3% of resistin, and 7.2% of adiponectin. Budgetary constraints limited our ability to conduct all the assays in all samples. However, because we sampled randomly for assays in each batch, the missingness pattern can be considered random (24).

Body weight and composition

Trained field workers collected anthropometric measurements according to standardized procedures (25). Body mass index (BMI) was calculated as weight divided by height squared (kg/m²). Normal weight was defined as BMI between 18.5 and 24.9kg/m², overweight as BMI between 25 and 29.9kg/m², and obesity as BMI ≥ 30kg/m² (26). Abdominal obesity was defined as waist circumference > 88cm in women and > 102cm in men (27). Body composition was assessed using the deuterium oxide dilution technique as previously described (28). Deuterium oxide (²H₂O) is used by the human body in the same way as water (H₂O), and is dispensed through various forms of body water (e.g., saliva, urine, sweat) within hours. Therefore, through measuring deuterium enrichment (as assessed by Fourier transform-infrared spectroscopy, Shimadzu 8400S) in dispensed body water, one can estimate total body water using preestablished mathematical models that quantify turnover of body water. In our study, before and 3h after administering a standardized deuterium oxide dose, saliva samples were collected for the estimation of total body water following the formula: TBW (kg) = Dose ²H₂O (mg) / enrichment ²H in saliva (mg/kg). Fat-free mass, calculated from total body water, under the assumption of a hydration constant of 0.732. Fat mass was calculated as the difference between body mass and fat-free mass. Percent body fat equals fat mass divided by total body mass (29).

Statistical analysis

We used Student’s t-test or chi-squared test to compare characteristics between men and women. For each biomarker, we computed the postprandial relative change (%Δ) as: $\frac{\mathit{\text{Postprandial}}\mathit{\text{concentration}}-\mathit{\text{Fasting}}\mathit{\text{concentration}}}{\mathit{\text{Fasting}}\mathit{\text{concentration}}}\times 100\%$. To describe the relationships among postprandial relative changes, we calculated Pearson correlation coefficients. To improve visual comparability across biomarkers, we standardized the %Δ of each marker. Z-scores were calculated as: $\frac{\mathit{\text{Observed}}(\%\Delta )-\mathit{\text{Mean}}(\%\Delta )}{\mathit{\text{Standard}}\mathit{\text{Deviation}}(\%\Delta )}$

To investigate the association of metabolic flexibility with body composition, we analyzed the %Δ data based on BMI and percent body fat first as continuous variables, and then as sex-specific tertiles. For each biomarker, %Δ was modeled as the dependent variable in least squares regression models. The independent variables were either BMI (kg/m²) or body fat (percent). In Model 1, we adjusted for age (year) at the time of data collection and sex. Model 2 was adjusted for age, sex, socio-economic status at the time of data collection, Guatemala City residency (to account for unknown confounders in the environment), parity in women (the number of livebirths, which is associated with BMI), current smoking status (potential impact on inflammation and body composition) and alcohol consumption (known impact on lipid metabolism), lipid-lowering medication, and village of birth (related to the INCAP trial in 1969–77) (22). All models were controlled for clustering at the household level (by maternal identification number) through generating cluster-robust estimates of the variance matrix.

We then created sex-specific tertiles of BMI and percent body fat (low=1; high=3). We calculated the mean %Δ Z-scores for each biomarker within tertiles. We visualized the %Δ Z-scores to investigate patterns in meal-induced responses by a priori determined cardiometabolic pathways: lipid responses, glycemic responses, pro- and anti-inflammatory responses.

We carried out all analyses in R version 4.0.2 (R Core Team, Foundation for Statistical Computing, Vienna, Austria). Statistical significance was set at p<0.05. All p-values were two-sided.

Research ethics

The study followed protocols approved by the Institutional Review Board at Emory University and the Ethics Review Committee of INCAP. All study participants provided written informed consent in Spanish.

Results

Selected characteristics and fasting biomarker profile

Among the 1,027 participants, 610 (59.4%) were women (Table 1). The mean (SD) age was 44.0 (4.2) years. More than 40% of women were obese, compared with 19.4% of men (p<0.001). In addition, 90.0% of women and 20.1% of men had abdominal obesity (p<0.001). Women had higher percent body fat than men (p<0.001). Men reported higher alcohol consumption and current smoking than women (both p<0.001).

Table 1:

Selected characteristics of the study population, by sex

Characteristicsa	Pooled (n = 1027)	Women (n = 610)	Men (n = 417)	p valuee
Age, years	44.0 (4.2)	44.1 (4.3)	43.9 (4.1)	0.26
Body weight and composition b
Body mass index, kg/m2	28.2 (5.0)	29.3 (5.8)	26.6 (4.7)	<0.001
Obesity, %	32.4	41.3	19.4	<0.001
Waist Circumference, cm	98.6 (12.1)	101.7 (13.6)	93.9 (11.3)	<0.001
Abdominal obesity, %	61.6	90.0	20.1	<0.001
Percent body fat, percent	36.8 (9.2)	42.3 (5.9)	28.6 (6.7)	<0.001
Additional characteristics
Socioeconomic status tertilesc				0.38
Lowest tertile, %	32.7	31.8	34.1	--
Middle tertile, %	32.9	34.6	30.5	--
Highest tertile, %	34.4	33.6	35.5	--
Guatemala City residencec, %	18.9	19.0	18.7	0.96
Parity (women), numberd	--	3 (2 – 4)	--	--
Alcohol consumptionc, %	17.7	4.6	36.9	<0.001
Smoking statusc, %	12.8	0.8	30.2	<0.001
Lipid-lowering medicationc, %	1.5	1.8	1.0	0.40

^aDefinitions: Data presented as mean (SD) or %; Overweight: BMI ≥ 25 kg/m² & BMI < 30 kg/m²; Obesity: BMI ≥ 30 kg/m². Abdominal obesity: waist circumference > 88 cm for women; > 102 cm for men

^bCorrelation between body weight and composition measurements: Pearson’s correlation coefficient (r) for BMI and waist circumference = 0.93 (p < 0.001); BMI and percent body fat r = 0.64 (p < 0.001); waist circumference and percent body fat r = 0.69 (p < 0.001).

^cStatus at the time of data collection (2015–17).

^dMedian (Inter-quartile range) for parity, in this study defined as the number of livebirths for each woman

^eP-values based on Student’s t-test (continuous variables) or chi-squared test (categorical variables)

Fasting concentrations of most lipids differed between men and women (p<0.001 for TC, HDLc, LDLc, non-HDLc, apoA-I, apoB, and NEFA; p=0.05 for TG) (Table 2). Fasting insulin and glucose concentrations were higher in women than in men (p<0.001 and p=0.01, respectively). Among pro-inflammation biomarkers, hsCRP and leptin were both higher in women than in men (p<0.001). The anti-inflammatory cytokine, adiponectin, was higher in women than in men (p<0.001).

Table 2.

Fasting and postprandial changes in biomarker concentrations by sex

Biomarkers	Fasting concentrationa			Postprandial change (%Δ)b
Mean (SD)	Women (n = 610)	Men (n = 417)	p-valuec	Women (n = 610)	Men (n = 417)	p-valued
Lipids
TC (mmol/L)	4.7 (1.0)	4.4 (1.0)	<0.001	0.3 (6.3)	0.1 (7.6)	0.68
TG (mmol/L)	4.3 (2.3)	4.6 (3.0)	0.05	15.2 (18.9)	15.9 (18.2)	0.56
HDLc (mmol/L)	1.1 (0.2)	1.0 (0.2)	<0.001	1.3 (6.3)	0.8 (7.0)	0.27
LDLc (mmol/L)	3.0 (0.8)	2.7 (0.9)	<0.001	0.1 (7.7)	0.5 (11.4)	0.54
ApoA-I (g/L)	1.1 (0.3)	1.0 (0.3)	<0.001	0.5 (6.3)	0.9 (8.5)	0.37
ApoB (g/L)	0.9 (0.3)	0.8 (0.3)	<0.001	−0.5 (8.4)	0.2 (9.9)	0.23
NEFA (mEq/L)	1.0 (0.4)	0.8 (0.4)	<0.001	−52.5 (28.0)	−42.2 (31.6)	< 0.001
Glycemic markers
Insulin (pmol/L)	122.0 (73.9)	90.4 (72.5)	<0.001	264.2 (170.8)	196.0 (167.6)	< 0.001
Glucose (mmol/L)	5.7 (0.9)	5.5 (0.7)	0.01	19.7 (18.5)	7.3 (20.8)	< 0.001
Pro-inflammation markers
hsCRP (mg/L)	3.8 (3.7)	2.1 (2.9)	<0.001	0.6 (12.1)	1.5 (17.2)	0.35
Leptin (ng/mL)	17.9 (10.1)	4.1 (4.7)	<0.001	−10.4 (25.0)	−22.0 (29.7)	< 0.001
Resistin (ng/mL)	2.2 (2.5)	2.3 (3.8)	0.56	−4.0 (18.7)	−2.2 (23.5)	0.30
MCP-1 (pg/mL)	93.7 (98.6)	99.1 (70.2)	0.34	15.8 (111.2)	−6.3 (35.4)	0.01
Anti-inflammation markers
IL-10 (pg/mL)	70.0 (190.0)	93.8 (235.8)	0.14	15.8 (111.2)	15.3 (110.1)	0.95
Adiponectin (μg/mL)	12.5 (8.7)	9.1 (7.4)	<0.001	−1.3 (24.8)	4.4 (103.4)	0.38

Abbreviations: TC, total cholesterol; TG, triglycerides; HDLc, high density lipoprotein cholesterol; LDLc, low density lipoprotein cholesterol; apo, apolipoprotein; NEFA, non-esterified fatty acid; hsCRP, high sensitivity C-reactive protein; MCP-1, monocyte chemoattractant protein 1; IL-10, interleukin 10.

^aMissingness (> 5.0%): 25.8% of IL-10, 12.8% of MCP-1, 7.3% of resistin, and 7.2% of adiponectin.

^bPostprandial change (%Δ) equals the difference between postprandial and fasting biomarker concentrations, divided by fasting concentrations, presented as percentages.

^cStudent’s t-test, comparing the fasting biomarker concentrations between men and women.

^dStudent’s t-test, comparing the postprandial changes (%Δ) between men and women.

Postprandial biomarker changes

TG, insulin, glucose, and IL-10 had the highest relative increase in both men and women (Table 2). NEFA and leptin had notable reductions in concentrations in both sexes. Postprandial biomarker responses differed between men and women for NEFA (larger reduction in women), insulin and glucose (larger increase in women), and leptin (larger reduction in men) (p<0.001 for all). The correlation matrix among biomarker responses is shown in Supplementary Figure S1: lipids (all except TG and NEFA) had positive correlations in postprandial responses, and the response for hsCRP was positively correlated with those of the lipids. Insulin and glucose responses were positively correlated, whereas glucose responses were inversely correlated with those of all lipids. There were weak correlations (−0.2 < Pearson’s r < 0.2) among most cytokines.

Postprandial biomarker changes and body composition

After adjusting for covariates, with each 1 kg/m² increase in BMI, %Δ (95% CI) increased by 0.17% (0.09, 0.26%; p<0.001) for TC, by 0.31% (0.07, 0.54 %; p<0.05) for TG, and by 0.11% (0.01, 0.20%; p<0.05) for apoA-I (Table 3). When BMI increases by 1 kg/m², insulin response was attenuated by 6.30 % (−8.41, −4.20%; p<0.001). Among inflammatory responses, postprandial leptin reduction was attenuated with each unit increase in BMI (0.64%; 0.24, 1.05%; p<0.01). The postprandial increase of anti-inflammatory marker IL-10 was attenuated by 1.66 % (−3.39, 0.07%; p<0.05) with each unit increase in BMI.

Table 3.

Association between postprandial biomarker changes and body composition^a

Postprandial %Δ	Body mass index (kg/m2)		Percent body fat (Percent)
	Model 1b β (95% CI)	Model 2c β (95% CI)	Model 1b β (95% CI)	Model 2c β (95% CI)
Lipid responses
TC	0.14 (0.06, 0.22) **	0.17 (0.09, 0.26) ***	0.08 (0.009, 0.14) *	0.10 (0.03, 0.16) **
TG	0.26 (0.04, 0.49) *	0.31 (0.07, 0.54) *	0.12 (−0.08, 0.33)	0.15 (−0.05, 0.36)
HDLc	0.06 (−0.02, 0.14)	0.07 (−0.01, 0.15)	0.01 (−0.05, 0.07)	0.03 (−0.03, 0.09)
LDLc	0.04 (−0.07, 0.14)	0.05 (−0.06, 0.17)	−0.02 (−0.12, 0.08)	0.0003 (−0.11, 0.11)
apoA-I	0.10 (0.01, 0.18) *	0.11 (0.01, 0.20) *	0.01 (−0.08, 0.10)	0.02 (−0.08, 0.11)
apoB	0.07 (−0.04, 0.19)	0.09 (−0.03, 0.21)	0.02 (−0.08, 0.11)	0.02 (−0.09, 0.12)
NEFA	0.10 (−0.25, 0.45)	0.11 (−0.22, 0.45)	0.08 (−0.19, 0.36)	0.00 (−0.17, 0.35)
Glycemic responses
Insulin	−5.59 (−7.61, −3.56) ***	−6.30 (−8.41, −4.20) ***	−0.68 (−2.44, 1.08)	−1.06 (−2.87, 0.75)
Glucose	0.15 (−0.17, 0.46)	0.10 (−0.22, 0.41)	0.34 (0.12, 0.56) ***	0.29 (0.06, 0.51) **
Pro-inflammatory responses
hsCRP	0.13 (−0.03, 0.29)	0.14 (−0.03, 0.30)	0.08 (−0.07, 0.23)	0.09 (−0.06, 0.25)
Leptin	0.60 (0.18, 1.02) **	0.64 (0.24, 1.05) **	0.60 (0.23, 0.97) ***	0.61 (0.24, 0.99) ***
Resistin	0.05 (−0.30, 0.34)	0.05 (−0.31, 0.40)	−0.003 (−0.28, 0.28)	0.02 (−0.26, 0.31)
MCP-1	0.24 (−0.26, 0.73)	0.07 (−0.44, 0.58)	−0.06 (−0.43, 0.31)	−0.14 (−0.52, 0.23)
Anti-inflammatory responses
IL-10	−1.68 (−3.44, 0.09) *	−1.66 (−3.39, 0.07) *	−1.22 (−2.70, 0.26) *	−1.22 (−2.64, 0.20) *
Adiponectin	0.26 (−0.13, 0.65)	0.26 (−0.16, 0.67)	0.24 (−0.01, 0.50)	0.24 (−0.02, 0.51)

Abbreviations: BMI, body mass index; CI, confidence interval; TC, total cholesterol; TG, triglycerides; HDLc, high density lipoprotein cholesterol; LDLc, low density lipoprotein cholesterol; apo, apolipoprotein; NEFA, non-esterified fatty acid; hsCRP, high sensitivity C-reactive protein; MCP-1, monocyte chemoattractant protein 1; IL-10, interleukin 10.

^aBased on least squares regression models: dependent variables are individual biomarker changes postprandial (%Δ), independent variables are BMI or percent body fat, respectively. All models are controlled for clustering at the household level (by mother’s ID) by generating cluster-robust estimates of the variance matrix. F-statistic in least squares regression: * P-value < 0.05, ** P-value < 0.01, *** P-value < 0.001.

^bModel 1 is adjusted for age at the time of data collection (year) and sex (female/male).

^cModel 2 is adjusted for age at the time of data collection (year), sex (female/male), socio-economic status at the time of data collection (tertiles), Guatemala City residency (reside in Guatemala City/Other), current alcohol consumption (yes/no), current smoking status (yes/no), lipid-lowering medication (yes/no), village of birth (one in four villages), and parity (women only).

After adjusting for all covariates, with each 1 percent increase in body fat, only %Δ TC increased in postprandial response (0.10%; 95%CI: 0.03, 0.16 %; p<0.01) among all lipids (Table 3). Insulin response was not significantly affected by body fat percent differences, but glucose response increased by 0.29% (0.06, 0.51%; p<0.01). Among all pro- and anti-inflammatory markers: with 1 percent increase in body fat, postprandial leptin reduction was attenuated by 0.61% (0.24, 0.99%; p<0.001), and IL-10 response was reduced by 1.22% (−2.64, 0.20; p<0.05) (Table 3).

Figure 2 (color version: Supplementary Figure S2) shows the mean %Δ for each biomarker by sex-specific BMI tertiles. Among women, lipid responses were generally higher in the 3^rd than the 1^st tertile. Glucose response was similar across all tertiles in women, whereas insulin response was attenuated in the higher tertile. Leptin reduction was attenuated (the mean %Δ value was higher) in the 3^rd versus 1^st tertile in women. Among men, the responses for TG, HDLc, LDLc were lower for 3^rd versus 1^st tertile. Glucose response increased in the 3^rd versus 1^st tertile among men. Similar to women, insulin response and leptin reduction were both attenuated in the 3^rd versus 1^st BMI tertile in men.

Figure 2. Postprandial biomarker changes by body mass index tertile for women (A) and men (B).

Abbreviations: TC, total cholesterol; TG, triglycerides; HDLc, high density lipoprotein cholesterol; LDLc, low density lipoprotein cholesterol; apo, apolipoprotein; NEFA, non-esterified fatty acid; hsCRP, high sensitivity C-reactive protein; MCP-1, monocyte chemoattractant protein 1; IL-10, interleukin 10. Sample sizes: BMI tertiles 1 (low) to 3 (high) for women: 204, 203, 203. BMI tertiles 1 to 3 for men: 139, 139, 139. Sample size varied by biomarker. This figure presented the mean z scores of postprandial changes (%Δ z-score) for each biomarker, categorized by BMI tertiles for men and women separately. %Δ equals the difference between postprandial and fasting biomarker concentrations, divided by fasting concentration. Standardized %Δ z-scores were calculated as postprandial response (%Δ) minus the mean and divided by the standard deviation for each biomarker.

Figure 3 (color version: Supplementary Figure S3) shows the mean %Δ for each biomarker by sex-specific body fat tertiles. Among women, lipid responses were increased (except NEFA) in higher tertile of body fat. Insulin response was attenuated in the higher body fat tertile among women. Both hsCRP and leptin had increased responses (attenuated reduction in the case of leptin) among women as body fat tertile increased from low to high. Among men, both glucose and insulin response increased from low to high body fat tertile. HsCRP and leptin showed higher mean %Δ value as body fat increased among men. No clear pattern was observed in inflammatory responses by BMI or body fat tertile.

Figure 3. Postprandial biomarker changes by percent body fat tertile for women (A) and men (B).

Abbreviations: TC, total cholesterol; TG, triglycerides; HDLc, high density lipoprotein cholesterol; LDLc, low density lipoprotein cholesterol; apo, apolipoprotein; NEFA, non-esterified fatty acid; hsCRP, high sensitivity C-reactive protein; MCP-1, monocyte chemoattractant protein 1; IL-10, interleukin 10. Sample sizes: Body fat tertiles 1 (low) to 3 (high) for women: 199, 198, 198. Body fat tertiles 1 to 3 for men: 136, 135, 135. Sample size varied by biomarker. This figure presented the mean z scores of postprandial changes (%Δ z-score) for each biomarker, categorized by body fat tertiles for men and women separately. %Δ equals the difference between postprandial and fasting biomarker concentrations, divided by fasting concentration. Standardized %Δ z-scores were calculated as postprandial response (%Δ) minus the mean and divided by the standard deviation for each biomarker.

Discussion

Summary of findings

Through a standardized, liquid meal challenge containing macronutrient composition within the physiological range, we assessed metabolic flexibility in a large free-living cohort of adults in Guatemala. We confirmed our hypothesis that metabolic flexibility, as assessed by postprandial biomarker responses, differed by body composition. We also observed differences in the meal-induced biomarker responses between men and women.

Postprandial lipid responses and body composition

We observed significant associations of postprandial lipid responses with body composition. The meal-induced elevation in total cholesterol was greater in individuals with higher BMI and in those with greater body fat. This may reflect delayed clearance of intestinal chylomicrons transporting newly absorbed triglycerides and cholesterol (30). We also observed increased overall postprandial lipid responses, particularly in women, with increasing BMI. More pronounced postprandial increase in TG was observed in individuals with higher BMI – a finding consistent with previous research (31). Of interest is the lack of notable difference in the TG response among individuals with different percent body fat. Since postprandial TG levels reflect the rate of appearance of intestinal chylomicrons into the bloodstream, it is possible that BMI and percent body fat have differential association with the rate of dietary fat absorption.

Postprandial glycemic responses and body composition

The relationship between obesity and type 2 diabetes is well studied, with insulin resistance being a key linkage therein (32). In our study, we observed that both men and women showed approximately two-fold increase in insulin concentration 2h after the meal challenge. Postprandial insulin increase was significantly attenuated in individuals with higher BMI, but there was no significant difference in response among individuals with varying degrees of percent body fat. It is plausible that those with high BMI but normal percent body fat actually have higher fat-free mass, and utilizes insulin more efficiently (33). Contrary to the pattern of insulin response, postprandial glucose elevation was associated with percent body fat, but not with higher BMI. We postulate that individuals with high percent body mass may experience insulin resistance, which in this study manifested as elevated glucose concentration for longer duration following the meal challenge (34).

Postprandial inflammatory responses and body composition

Postprandial hyperlipidemia and hyperglycemia have cumulative impact on inflammatory cytokines (16). In our data, higher BMI and higher percent body fat were both associated with the pro-inflammatory response following meal consumption – there was greater postprandial increase in the pro-inflammatory marker leptin and more pronounced reduction in the anti-inflammatory marker IL-10. Among the cytokines measured, adipokines have previously been reported to be important in predicting future cardiometabolic risks due to their effects on insulin sensitivity and inflammation (35). In our study, leptin, was significantly lower 2h after the meal challenge (36). This observation was plausible based on its catabolic functions, but unexpected in terms of its pro-inflammatory effect (37). In our study, postprandial reduction in leptin was attenuated as BMI and percent body fat increase. This may be attributable to the positive correlation between leptin and adiposity, and to the common presence of leptin resistance in obesity (38). There was no significant meal-associated change in any other cytokines.

Sex-specific differences in postprandial biomarker responses

As previously reported, there were significant differences in cardiometabolic characteristics between men and women in our cohort (24). These differences translated into differential fasting concentrations of markers with the exception of TG, resistin, MCP-1, and IL-10. However, meal-induced responses (%Δ) were different between men and women for only NEFA, insulin, glucose, and leptin. Meal-induced increases in both insulin and glucose were greater in women than in men. Postprandial reduction in NEFA was more pronounced in women, whereas the reduction in leptin was greater for men. Meal consumption was associated with a reduction in MCP-1 among men but an increase among women. Previous research have reported that women had higher postprandial glucose uptake rate, better insulin secretion, and lower oxidative stress than men (39, 40). The sex-specific differences observed in our study may be due to the significantly higher prevalence of central obesity in women than men, in addition to other biological factors that contribute to sexual dimorphism (41). For instance, in a previous study, Bianchi and colleagues reported a correlation between low testosterone level and insulin resistance, and noted the different roles played by androgen and estrogen in terms of body fat distribution and associated cardiometabolic health outcomes (42).

Strengths and Limitations

To our knowledge, this is the first study in a low- or middle-income country setting to test meal-induced stress response in a cohort of adults. Very few studies in similar settings have a comprehensive panel of cardiometabolic biomarkers, let alone testing a metabolic stress model. Our study had over 80% power to detect a medium effect size, except for IL-10 and MCP-1. Another strength is that the meal challenge provided macronutrients within a physiological range consistent with normal human diets, which enabled us to observe biologically plausible responses. This may also explain the modest responses of several biomarkers. In addition, we conducted the analyses and interpreted the data based on both individual biomarker responses and overall patterns.

There were a few limitations to this study. First, we only obtained samples at fasting and 2h postprandial, which do not describe the full response trajectory of the biomarkers. The lack of response in some biomarkers may be due to a true resilience to stress, or to the timing of measurement. Second, the use of BMI and percent body fat as measurements for body composition have limitations. In addition, there may be other confounding factors that are associated with both body composition and cardiometabolic health status. We recognize this possibility, and adjusted for contextual covariates in the regression models as proxy factors for such unmeasured confounders.

Conclusion

Obesity is a key risk factor for CMDs. We observed that the metabolic flexibility differed by body composition. We also observed differences in the biomarker responses between two body composition measurements: BMI and percent body fat. Sex-specific differences were observed in the fasting biomarker concentrations and postprandial responses, and were more pronounced in the former than the latter. Through this research, we characterized the impact of obesity (through the lens of body composition) on cardiometabolic responses in a cohort of free-living adults in Guatemala. We also reported that observable responses in biomarkers can be triggered using a relatively modest macronutrient meal challenge. Building upon these results, we recommend that future studies with larger sample sizes compare the structural relationship of biomarker responses across body composition strata, which will further elucidate the nutritional and metabolic mechanisms for the onset and progression of cardiometabolic conditions. We also caution against simplified interpretation of the study findings, due to the complexity of obesity as a phenomenon, the unreliability of body composition measurements as proxy for obesity status, and the non-linear association between obesity and CMD outcomes.

Abbreviations:

ApoA-I

apolipoprotein A-I

ApoB

apolipoprotein B

BMI

body mass index

CI

confidence intervals

HDLc

high-density lipoprotein cholesterol

HsCRP

high-sensitivity C-reactive protein

IL-10

interleukin-10

INCAP

Institute of Nutrition of Central America and Panama

LDLc

low-density lipoprotein cholesterol

MCP-1

monocyte chemoattractant protein-1

NEFA

non-esterified fatty acids

Non-HDLc

subtracting HDLc from TC concentration

TC

total cholesterol

TG

triglycerides