Adiponectin Is Associated with Favorable Lipoprotein Profile, Independent of BMI and Insulin Resistance, in Adolescents

Adiponectin levels in adolescents are inversely associated with an atherogenic panel of lipoprotein subclass particles, independent of body mass index and insulin resistance.

Abstract

Context:

Children with obesity and insulin resistance (IR) have decreased adiponectin and have increased cardiovascular risk. Adiponectin has antiatherogenic effects, but its mechanism is unclear.

Objectives:

Our objectives were 1) to compare lipoprotein subclass particles among obese and lean adolescents and delineate their relationships with IR and 2) to measure relationships between adiponectin and lipoproteins and their dependence on body mass index (BMI) and/or IR.

Design, Setting, Patients, and Main Outcome Measures:

This was a cross-sectional study of 57 obese and 38 lean pubertal adolescents, measuring lipoprotein subclass particles (nuclear magnetic resonance spectroscopy), lipids, adiponectin, and homeostasis model assessment of IR (HOMA-IR).

Results:

Obese had higher low-density lipoprotein (LDL) cholesterol (P = 0.018), higher small LDL particles (LDL-P) (P < 0.0005), smaller LDL-P size (P < 0.0005), smaller high-density lipoprotein particle (HDL-P) size (P < 0.0005), lower HDL cholesterol (HDL-C) (P < 0.0005), and higher small HDL-P (P = 0.009) compared with lean. HOMA-IR was higher in obese than lean (P < 0.0005) and positively associated with triglycerides, large very LDL-P, and small HDL-P and negatively with HDL-P size in obese. Adiponectin was lower in obese than lean (P < 0.0005) and was positively associated with LDL-P size, HDL-P size, and HDL-C and negatively with triglycerides, small LDL-P, large very LDL-P, and small HDL-P in obese. Using linear regression adjusting for demographics, Tanner stage, BMI, and HOMA-IR in all adolescents, adiponectin was positively associated with LDL-P size (P = 0.028), HDL-P size (P < 0.0005), and HDL-C (P = 0.042) and negatively with small LDL-P (P = 0.009) and small HDL-P (P = 0.004).

Conclusions:

Obese adolescents have lower adiponectin levels than lean, and a more atherogenic lipoprotein profile, associated with increased IR. Adiponectin was inversely associated with atherogenic lipoproteins in adolescents, even after adjusting for obesity and IR. This is the first such report in children, and suggests a relationship between adiponectin and lipoproteins in adolescents independent of BMI and IR.

Increased pediatric obesity has led to increased insulin resistance (IR) and type 2 diabetes mellitus (T2DM) in children (1). Both obesity and diabetes are risk factors for cardiovascular disease (CVD). Obesity and IR are also both components of the metabolic syndrome (2), a constellation of risk factors that additionally includes dyslipidemia and hypertension, and are predictive of CVD in adults. The development during childhood of obesity, dyslipidemia, and T2DM, all risk factors for CVD, may result in a generation presenting with CVD decades younger than previous generations, with dire repercussions on morbidity and mortality (3).

Adiponectin is an adipocytokine thought to have antiinflammatory and antiatherogenic effects (4). In adults, hypoadiponectinemia is an independent risk factor for T2DM (5), and low adiponectin levels are independently associated with coronary artery disease (6), including among type 2 diabetics (7). In children, adiponectin is decreased in metabolic syndrome and obesity (8) and inversely related to IR (9). The mechanism of adiponectin's seemingly protective effect is unknown, and it is unclear whether it is mediated through its relationship with obesity and/or IR.

We hypothesized that not only would obese adolescents have a more atherogenic lipoprotein profile, elevated homeostasis model assessment of IR (HOMA-IR), and reduced adiponectin levels compared with lean controls, but also that low adiponectin levels would be associated with an atherogenic lipoprotein profile independent of body mass index (BMI) and HOMA-IR. Thus, the objectives of this study were 1) to compare lipids and lipoprotein subclass particles among obese and lean adolescents and delineate their relationships with HOMA-IR and 2) to investigate the relationships between adiponectin and lipids/lipoprotein subclass particles in adolescents and whether these relationships are dependent on BMI and/or IR.

Subjects and Methods

This observational, cross-sectional study involved obese and lean (BMI ≥ 95th and ≤85th percentile for age and sex, respectively) adolescents. Inclusion criteria were 1) ages 12–17 yr and 2) Tanner stage higher than 1. Exclusion criteria were 1) oral glucose tolerance test showing impaired fasting glucose (fasting glucose ≥ 100 mg/dl) or impaired glucose tolerance (2-h glucose = 140–199 mg/dl), 2) major chronic illness, 3) pregnancy, 4) genetic syndrome known to affect glucose tolerance, 5) familial hypercholesterolemia, 6) treatment with medications known to affect insulin sensitivity (systemic steroids in the last 1 month) or lipid profiles (statins or high-dose vitamin A), and 7) treatment with high doses of inhaled steroids (>1000 μg/d). Both groups were recruited from four primary care clinics affiliated with The Children's Hospital of Philadelphia (CHOP) serving a largely African-American population from inner-city Philadelphia. Written informed consent and age-appropriate assent were obtained from all subjects before participation, and the study was approved by the CHOP Institutional Review Board.

Study visits took place from October 2007 through November 2009 at the Clinical and Translational Research Center of CHOP and the Hospital of the University of Pennsylvania. Tanner staging for pubertal assessment was based on breast development in girls and testicular size in boys. Weight was measured with the subject wearing a light gown without shoes by use of a Scaletronix digital scale (Scaletronix, White Plains, NY), calibrated daily. Height was measured using a wall-mounted stadiometer (Holtain Inc., Crymych, UK). BMI percentiles were assessed using age- and gender-specific BMI reference data (10).

After a 12-h overnight fast, a blood sample was obtained for glucose, insulin, adiponectin, lipid panel, and lipoprotein subclass particle analysis. Obese subjects then underwent a 2-h oral glucose tolerance test. Fasting insulin and glucose were used to calculate HOMA-IR: [fasting insulin (μIU/ml) × fasting glycemia (mmol/liter)]/22.5. Adiponectin was measured by ELISA, using a kit from ALPCO Diagnostics (Salem, NH). Triglycerides, total cholesterol, and high-density lipoprotein cholesterol (HDL-C) were assayed on a Hitachi 912 using Roche reagents. Low-density lipoprotein cholesterol (LDL-C) was calculated using the Freidewald equation [LDL-C = total cholesterol − HDL-C − (triglycerides)/5)]. Insulin was measured by ELISA, using a kit from ALPCO Diagnostics. Lipoprotein subclass analysis was performed by LipoScience, Inc. (Raleigh, NC), using nuclear magnetic resonance (NMR) spectroscopy. In this manuscript, lipid cholesterol levels are designated by C following the lipoprotein, and lipoprotein particle numbers measured by NMR are designated by P.

Statistical analyses were performed using SPSS for Windows release 15 software (SPSS Inc., Chicago, IL). A P value ≤ 0.05 was considered statistically significant. Kolgomorov-Smirnov tests were used to examine the distribution of variables. Logarithmic transformations were applied as needed. Lipids and lipoprotein subclass particles were compared between obese vs. lean groups using two-sample t tests or Wilcoxon rank sum tests depending on normality of distribution.

Wilcoxon, Kruskal-Wallis, t tests, and correlation coefficients were used to explore the impact of possible confounders (race, gender, age, Tanner stage, BMI, and HOMA-IR) on lipoprotein outcomes. Pearson or Spearman correlation coefficients were used to examine the linear or rank order relationship between HOMA-IR or adiponectin, and the lipids and lipoprotein subclass particle levels, in obese and lean groups.

Hierarchical multiple linear regression models were used in the sample as a whole to examine the interrelationships between lipoprotein outcomes and adiponectin, obesity (BMI as a continuous variable), and IR (HOMA-IR). For each lipid and lipoprotein subclass outcome of interest, demographics [gender, age, race (African-American vs. others)] were entered in the first step, a dummy-coded Tanner stage (2 and 3 vs. 4 vs. 5) was entered in a second step, BMI was entered in a third, then HOMA-IR, and then finally adiponectin was entered in the final step. At each step, change in coefficient of determination of the model (R²) was observed. This change in coefficient of determination of the model (R²) indicated the amount of variance in the outcome that could be explained by the addition of the new predictor variable, over and above that already explained by preceding predictors.

Results

Table 1 shows the distribution of demographic variables between the 57 obese and 38 lean adolescent participants, with no significant differences in age, sex, or race between the two groups. Both groups were largely African-American. Puberty stage was significantly different between the two groups, with more obese subjects in the Tanner 5 group, but more than 90% of both groups were Tanner 4 or 5.

Table 1.

Comparison of obese and lean groups

	Obese, n = 57a	Lean, n = 38a	P valueb
Age (yr)	14.5 ± 1.3	14.7 ± 1.3	0.45
Gender [n (%) male]	22 (38.6)	20 (52.6)	0.21
Race [n (%) African-American]	44 (77.2)	31 (81.6)	0.34
Tanner stage [n (%)]			0.035
2	1 (1.8)	0 (0.0)
3	2 (3.5)	3 (7.9)
4	14 (24.6)	18 (47.4)
5	40 (70.2)	17 (44.7)
BMI (kg/m2)	34.3 ± 5.5	19.9 ± 1.9	<0.0005
BMI z-score	2.23 ± 0.34	−0.01 ± 0.67	<0.0005
LDL-C (mmol/liter)	2.45 ± 0.77	2.10 ± 0.52	0.018
Triglycerides (mmol/liter)	0.85 ± 0.41	0.67 ± 0.26	0.057
Total cholesterol (mmol/liter)	3.97 ± 0.83	3.82 ± 0.65	0.35
Small LDL-P (nmol/liter)	525.6 ± 266.9	247.1 ± 168.2	<0.0005
Large VLDL-P and chylomicrons (nmol/liter)	1.3 ± 2.1	0.2 ± 0.5	0.061
LDL-P size (nm)	21.19 ± 0.68	21.82 ± 0.72	<0.0005
HDL-P size (nm)	9.02 ± 0.42	9.56 ± 0.34	<0.0005
HDL-C (mmol/literc)	1.13 ± 0.25	1.42 ± 0.30	<0.0005
Small HDL-P (μmol/liter)	16.7 ± 4.1	14.6 ± 3.4	0.009
HOMA-IRd	4.48 ± 2.85	1.73 ± 0.93	<0.0005
Adiponectin (ng/ml)	4149.8 ± 1695.2	6562.3 ± 2356.5	<0.0005

^aThe value of n may vary across variables due to missing values.

^bAnalysis by Fisher's exact test for gender, race, and tanner stage; analysis for continuous variables by t test except for triglycerides and large VLDL-P (used Mann-Whitney test because not normally distributed).

^cTo convert from mg/dl of cholesterol to mmol/liter of cholesterol, multiply by 0.02586. To convert from mg/dl of triglycerides to mmol/liter of triglycerides, multiply by 0.01129.

^dHomeostasis Model Assessment–Insulin Resistance Index; HOMA-IR = [fasting insulin (uIU/mL) x fasting glycemia (mmol/liter)] / 22.5.

The obese group had higher concentrations of LDL-C and small LDL particles (LDL-P), and smaller LDL-P size compared with lean subjects (Table 1), with a trend toward higher large very LDL particles (VLDL-P) and triglycerides as well. Obese adolescents also had lower HDL-C, higher small HDL particles (HDL-P), and smaller HDL-P size than lean subjects.

The mean HOMA-IR was higher in obese than lean subjects (Table 1) but varied considerably even within the obese group. Correlation analyses showed that within the obese group, HOMA-IR was positively associated with triglycerides (r = 0.39; P = 0.006), large VLDL-P (r = 0.41; P = 0.002), and small HDL-P (r = 0.28; P = 0.032) and negatively associated with HDL-P size (r = −0.28; P = 0.035). In contrast, in the lean group, there were no associations of HOMA-IR with any lipoprotein (P values > 0.1).

Mean adiponectin levels were lower in obese than lean subjects (Table 1). In obese subjects, adiponectin was negatively associated with triglycerides (r = −0.33; P = 0.021), small LDL-P (r = −0.35; P = 0.008), large VLDL-P (r = −0.27; P = 0.042), and small HDL-P (r = −0.31; P = 0.018) and positively associated with LDL-P size (r = 0.30; P = 0.026), HDL-P size (r = 0.44; P = 0.001), and HDL-C (r = 0.38; P = 0.008). In the lean group, adiponectin was negatively associated with small LDL-P (r = −0.39; P = 0.017) and small HDL-P (r = −0.43; P = 0.007) and positively associated with HDL-P size (r = 0.53; P = 0.001).

Using multiple linear regression in both groups combined (Table 2), there were no effects of demographics as a whole or Tanner stage on any of the outcomes. BMI had a significant effect on all lipid/lipoprotein outcomes after controlling for demographics and Tanner stage. HOMA-IR had a significant effect on log of large VLDL-P and small HDL-P after controlling for demographics, Tanner stage, and BMI. Adiponectin had significant effects on lipid/lipoprotein outcomes and was positively associated with LDL-P size, HDL-P size, and HDL-C and negatively associated with small LDL-P and small HDL-P after controlling for demographics, Tanner stage, BMI, and HOMA-IR (Table 2).

Table 2.

Linear regression model: all adolescents

Lipid/lipoprotein	Demographica		+ Tanner stageb		+ BMI		+ HOMA-IR		+Adiponectin
	ΔR2	P	ΔR2	P	ΔR2	P	ΔR2	P	ΔR2	P
LDL-C	0.029	0.496	0.007	0.769	0.062	0.024	0.003	0.646	0.002	0.663
LN triglyceride	0.015	0.745	0.039	0.211	0.092	0.005	0.028	0.116	0.004	0.538
Small LDL-P	0.020	0.627	0.004	0.822	0.305	<0.0005	0.017	0.139	0.052	0.009
LN large VLDL-P	0.018	0.667	0.008	0.689	0.092	0.004	0.043	0.042	0.001	0.714
LDL-P size	0.053	0.189	0.011	0.594	0.202	<0.0005	0.005	0.472	0.041	0.028
HDL-P size	0.041	0.301	0.036	0.196	0.345	<0.0005	0.024	0.060	0.084	<0.0005
HDL-C	0.034	0.426	0.030	0.292	0.247	<0.0005	0.000	0.824	0.037	0.042
Small HDL-P	0.034	0.379	0.016	0.480	0.066	0.014	0.083	0.004	0.077	0.004

^aAge, gender, race (African-American vs. all other racial groups).

^bBecause of only one subject who was Tanner 2, Tanner 2 and 3 were combined for regression analysis; therefore, comparison was Tanner 2 and 3 vs. 4 vs. 5.

Discussion

In this study, we demonstrate a significant inverse relationship between adiponectin and an atherogenic pattern of lipids and lipoprotein subclass particles in adolescents, independent of obesity and IR. This is the first such report in a pediatric population. Not surprisingly, our obese adolescents have lower adiponectin levels than lean adolescents and a more atherogenic lipoprotein profile (traditionally measured lipids as well as lipoprotein subclass particles measured by NMR). In the obese group, this atherogenic lipoprotein profile was associated with increased IR as well. However, adiponectin was significantly inversely associated with an atherogenic lipoprotein pattern, even after adjusting for the effects of BMI and HOMA-IR. The results of this study are important because the mechanisms of adiponectin's antiatherogenic and antiinflammatory effects are not known, and our results suggest that they are not mediated purely through obesity and IR.

Adiponectin is an important adipocytokine, whose functions are not fully understood. Its antiinflammatory effects include inducing the production of antiinflammatory mediators IL-10 and IL-1RA in human inflammatory cells (11). Adiponectin can also cause vasodilation by stimulating nitric oxide production in endothelial cells (12), felt to contribute to its antiatherogenic effects. However, contradictory results have been reported in patients with end-stage renal failure (13) or heart failure (14), and further research is needed to clarify these findings.

Obesity and IR lead to a specific pattern of metabolic dyslipidemia, a highly atherogenic lipid pattern consisting of high triglycerides, low HDL-C, and increased small subspecies of LDL particles. Lipoprotein subclass particle analysis by NMR spectroscopy in patients with obesity and IR can provide important supplemental information about CVD risk, which may not be evident from traditional lipid analysis (15). Small LDL (16), large VLDL, and small HDL (17) particles are associated with increased CVD risk in adults. Furthermore, in a study of adult healthy men (mostly nonobese), Kazumi et al. (18) showed that adiponectin levels were positively associated with HDL-C and LDL particle size, and negatively associated with triglycerides. Percent body fat, but not HOMA-IR, was found to be a determinant of adiponectin in that study. Recent data by Weiss et al. (19) showed an inverse association between adiponectin and an atherogenic lipoprotein pattern in adults, independent of obesity and IR. These studies were all performed in adults and were not targeted at obese individuals.

In children, smaller HDL (20), small LDL, and larger VLDL (21) particles are positively associated with obesity and IR (22). However, the association of lipoprotein subclass particles with adiponectin has not been explored previously in the pediatric population. Instead of being causative, associations between obesity and IR may be unrelated to adiponectin and its effects or may serve to exaggerate an existing phenomenon. The existence of these associations at such a young age suggests that there could be genetic or other early determinants involved. Future studies will be needed to elucidate these issues.

The importance of our findings stems from the fact that obese adolescents have hormonal and environmental influences distinct from adults, putting them at a unique risk for increased IR, T2DM, and metabolic dyslipidemia. During puberty, adolescents experience a period of decreased insulin sensitivity (23) thought to be secondary to increased GH secretion. In addition, adolescent girls have increased inactivity (24), a CVD risk factor, which can also lead to decreased HDL-C. Increased soft drink consumption during adolescence (25) can increase caloric intake and triglyceride levels as well. Although still only in their second decade of life, obese adolescents already have a more atherogenic pattern of lipoproteins than lean children, implying a greater duration of disease, with possibly a higher likelihood of developing atherosclerosis. It is important to establish the extent of this risk early and to identify potential targets for future intervention and treatment.

The current study has some limitations. Our population was largely African-American, limiting the generalizability of our results outside of this population. In addition, there was a statistically significant difference in Tanner stage detected between the lean and obese groups (Table 1). However, by looking at Table 1, it is clear that the majority of subjects are Tanner 4 or 5 and that the statistical difference between the two groups lies in the relative number that is Tanner 4 or Tanner 5. This difference, however, is accounted for in our analysis by adjusting for Tanner stage. Another limitation is that we cannot rule out the possibility of type II error, that there actually may be associations between adiponectin and other lipoprotein subclass particles that we were unable to detect, because of sample size. Finally, any cross-sectional study can only show association and not causation, so future longitudinal studies will be needed.

In conclusion, lipoprotein subclass particle analysis demonstrated an atherogenic pattern of lipids and lipoprotein subclass particles in obese vs. lean adolescents. The pattern of atherogenic lipids correlated positively with IR in obese and negatively with adiponectin levels. In the whole group, the associations between adiponectin and small LDL-P, LDL-P size, HDL-P size, HDL-C, and small HDL-P were independent of BMI and HOMA-IR. Thus, variation in adiponectin production in adolescents may be a causal determinant of an atherogenic lipoprotein profile.