Adiponectin Is a Link Among Inflammation, Insulin Resistance, and High‐Density Lipoprotein Cholesterol But Is Not Associated With Paraoxonase Activity in Premenopausal Women

Abstract

The aim of this study was to evaluate whether insulin sensitivity, inflammatory response, and plasma lipid profile are associated with circulating adiponectin levels in nondiabetic healthy women. The authors also assessed whether adiponectin has any effect on high‐density lipoprotein cholesterol–linked paraoxonase 1 (PON‐1) activity and on the susceptibility of low‐density lipoproteins to oxidation. Plasma adiponectin was measured in 91 nondiabetic premenopausal women, and the patients were then divided into quartiles. Circulating adiponectin was found to be associated with body mass index (r=.55, P<.001). After adjustment for body mass index, adiponectin showed an inverse correlation with the homeostasis model assessment of insulin resistance (HOMA‐IR) (r=−.41, P<.001) and a positive correlation with high‐density lipoprotein cholesterol (r=.43, P<.001). In linear regression analysis, HOMA‐IR, tumor necrosis factor α, and high‐density lipoprotein cholesterol levels were found to be independently associated with adiponectin. However, high‐density lipoprotein cholesterol–linked PON‐1 activity and the susceptibility of low‐density lipoproteins to in vitro oxidation did not seem to be related to plasma adiponectin concentrations.

Adiponectin, a recently identified novel peptide, is expressed specifically and abundantly in adipose tissue. Adiponectin is known to associate with both lipid and glucose metabolism ¹ , ² , ³ , ⁴ and to play a role as an anti‐inflammatory agent. Although produced exclusively in adipose tissue, circulating adiponectin levels are paradoxically lower in obese individuals. Plasma adiponectin concentrations have been found to be positively correlated with plasma high‐density lipoprotein cholesterol (HDL‐C) levels. Recently, human recombinant adiponectin has been reported to enhance the expression of ATP‐binding cassette transporter (ABCA1) and accelerate the synthesis of apolipoprotein A1 in human liver cell lines, suggesting that adiponectin might increase HDL‐C assembly in the liver. ⁵

Besides being a key player in the reverse cholesterol transport system, HDL‐C has the ability to protect low‐density lipoprotein cholesterol (LDL‐C) from oxidative modification. There is mounting evidence that high‐density lipoprotein–associated proteins such as paraoxonase 1 (PON‐1) could be implicated in this process. ⁶ , ⁷ , ⁸ PON‐1 (EC 3.1.8.1) is capable of hydrolyzing biologically active phospholipids present in oxidized LDL‐C and HDL‐C. PON‐1 activity has been shown to be diminished in pathologic conditions such as hypercholesterolemia, diabetes mellitus, ⁹ ischemic cardiac disease, ¹⁰ and renal failure ¹¹ and in aging. ¹²

Because of the ability of adiponectin to increase circulating HDL‐C concentrations, this adipokine might also have an effect on HDL‐C–linked PON‐1 activity, thereby causing an alteration on the susceptibility of LDL‐C to oxidation, which is believed to be an early event in the progress of atherosclerosis.

In this study, we aimed to investigate the relation of circulating adiponectin with insulin resistance, inflammatory markers, lipid profiles, and PON‐1 activity in premenopausal nondiabetic women. Attempts were also made to establish the involvement of adiponectin in the susceptibility of isolated LDL‐C to in vitro oxidation.

Materials and Methods

Patients

Ninety‐one healthy premenopausal women aged 19 to 49 years were included in this study after giving informed consent. All women were spontaneously menstrually active and nondiabetic. The prevalence of hypertension (systolic blood pressure ≥130 mm Hg, diastolic blood pressure ≥85 mm Hg) was 10%, and that of hyperlipidemia (triglyceride levels ≥200 mg/dL) was 5%; 49% of patients were obese (body mass index ≥30 kg/m²). None of the participants had any known cardiovascular disease, familial hyperlipidemia, chronic renal failure, or thyroid disorders.

Analytical Methods

Blood samples were drawn between 8 am and 9 am after at least 10 hours’ fasting. For PON‐1 activity determination, blood was collected without any additives. All blood samples were immediately centrifuged and stored at −80°C. Glucose, total cholesterol, LDL‐C, HDL‐C, and triglyceride levels were determined by an automated analyzer. Insulin was measured with an enzyme‐amplified chemiluminescence assay. Insulin resistance was evaluated by the homeostasis model assessment formula. This formula was calculated by multiplying fasting plasma insulin (μU/mL) with fasting plasma glucose (mmol/L) divided by 22.5. To analyze high‐sensitivity C‐reactive protein (hsCRP) levels, we used highly sensitive immunoassay (Cobas Integra 400 chemistry analyzer; Roche Diagnostics, GmbH, Mannheim, Germany). Only the patients and controls whose hsCRP values were <15 mg/L were included in this study, because values of hsCRP >15 mg/L likely indicate active inflammation or the presence of chronic inflammatory condition.

Blood concentrations of tumor necrosis factor α (TNF‐α), serum amyloid A (SAA) (Bio Source International, CA), interleukin‐8 (IL‐8) (Beckman Coulter Comp. Marseille, Cedex, France), and adiponectin (B‐Bridge International, CA) were determined by enzyme‐linked immunosorbent assay. Measurements of apolipoprotein AI and ceruloplasmin levels were made by nephelometric immunochemical assay (Dade Behring Inc, Newark, DE).

Paraoxonase activity was measured with phenyl acetate as the substrate ¹³ ; 500 μL of 1/100 prediluted serum was added to 2 mL of Tris–HCl buffer (9 mM, pH 8.00, 25°C) containing 0.9 mM CaCl₂ and 1.25 mM phenyl acetate. The rate of generation of phenol was determined at 270 nm (ε₂₇₀: 1306/M/cm).

LDL‐C was isolated by precipitation of 1 mL plasma (ethylenediaminetetraacetic acid [EDTA]) with 7 mL of 64 mM citrate buffer (pH: 5.05), containing 50,000 IU/L sodium heparinate. ¹⁴ Isolated LDL‐C was diluted in phosphate buffered saline (PBS) to 1 g of protein/L and dialyzed overnight against PBS at 4°C to remove the EDTA. LDL‐C protein concentration was determined with the Folin phenol reagent. LDL‐C (100 mg of protein/L) was incubated with 5 μM of CuSO₄ for 3 hours at 37°C. The formation of conjugated dienes was continuously monitored by measuring the increase in absorbance at 234 nm. The lag time required for the initiation of lipoprotein oxidation was calculated from the oxidation curve (Figure 1).

Figure 1.

 Oxidation curve of low‐density lipoprotein cholesterol isolated from a randomly selected participant (after incubation with 5 μM CuSO₄ for 3 hours at 37°C, the increase in absorbance at 234 nm was measured).

Statistical Analysis

Data are presented as mean ± SD. With continuous variables, group mean values were compared using the unpaired Student’s t‐test, as long as the variables were normally distributed. If the data distribution did not follow the normality assumption, Mann‐Whitney U test was utilized. Pearson’s correlation coefficients (r) were computed to explore the correlations between 2 variables. If variables were not normally distributed, they were log‐transformed prior to correlations. Further linear regression analysis was performed to evaluate the independent relationships by using a stepwise regression model.

Results

Clinical characteristics of participants divided into quartiles according to adiponectin levels are summarized in Table I. The mean of the lowest quartile of adiponectin was 5.51±0.98 (3.75–6.66) and that of the highest quartile was 16.24±3.07 (12.96–24.97).

Table I.

 Characteristics by Quartiles of Adiponectin in 91 Nondiabetic Healthy Women

Variable	Quartiles of Adiponectin				P for Trend
	Q1 3.75–6.66	Q2 6.74–8.91	Q3 9.10–12.75	Q4 12.96–24.97
Adiponectin (μg/mL)	5.51±0.98	7.94±0.72	11.04±0.97	16.24±3.07
Age (y)	34.62±7.26	33.59±5.86	34.00±7.03	30.78±9.44	0.12
BMI (kg/m2)	35.83±6.39	30.93±7.60	26.73±7.78	24.41±5.64	<0.001
SBP (mm Hg)	114.8±16.7	110.7±20.4	112.3±20.1	107.8±20.4	0.26
DBP (mm Hg)	79.6±8.7	74.4±11.4	74.2±10.7	71.8±12.0	0.02
Glucose (mg/dL)	85.29±8.21	85.91±7.76	83.13±9.59	81.41±11.15	0.18
Insulin (μU/mL)	14.85±7.05	9.99±3.62	6.88±3.33	6.12±3.23	<0.001
HOMA‐IR	3.11±1.46	2.24±0.88	1.45±0.78	1.29±0.78	<0.001
Triglycerides (mg/dL)	105.7±49.5	115.2±72.1	65.1±24.6	71.7±37.6	0.01
Cholesterol (mg/dL)	174.4±30.2	175.2±46.0	178.6±28.8	188.7±35.8	0.15
HDL‐C (mg/dL)	43.95±7.03	45.91±11.37	54.22±11.53	61.38±12.12	<0.001
Apolipoprotein A1 (mg/dL)	128.6±13.9	131.8±23.3	142.9±21.7	148.5±21.0	<0.001
LDL‐C (mg/dL)	105.7±24.3	102.3±35.2	102.6±25.0	105.4±30.3	0.97
PON‐1 (μmol/min/mL)	65.34±19.96	63.80±26.07	65.47±18.77	67.95±14.36	0.62
Lag time (min)a	98.50±31.97	92.00±24.70	80.27±38.25	81.00±22.73	0.17
hsCRP (mg/L)b	2.70 (0.50–13.00)	3.85 (0.30–10.00)	1.25 (0.30–6.40)	0.80 (0.20–1.80)	<0.001
TNF‐α (pg/mL)b	5.25 (0.00–61.80)	10.34 (0.00–82.71)	4.34 (0.00–193.62)	1.25 (0.00–134.35)	0.006
SAA (μg/mL)b	13.55 (0.30–46.60)	7.90 (1.80–50.70)	9.55 (3.70–83.30)	7.55 (1.50–21.50)	0.04
IL‐8 (pg/mL)b	27.88 (0.00–387.69)	27.88 (0.00–75.74)	21.53 (0.00–100.26)	8.68 (0.00–124.55)	0.06
Ceruloplasmin (mg/dL)	29.98±5.43	27.46±6.72	26.68±5.17	26.20±5.99	0.04

Abbreviations: BMI, body mass index; DBP, diastolic blood pressure; HDL‐C, high‐density lipoprotein cholesterol; HOMA‐IR, homeostasis model assessment of insulin resistance; hsCRP, high‐sensitivity C‐reactive protein; IL‐8, interleukin‐8; LDL‐C, low‐density lipoprotein cholesterol; PON, paraoxonase; SAA, serum amyloid A; SBP, systolic blood pressure; TNF‐α, tumor necrosis factor α. Values are mean ± SD. P values demonstrate statistical significance between lowest and highest quartiles of adiponectin; ^an=9 or 10 in each quartile; ^bvalues shown are median (range). Values in boldface are statistically significant.

There was a significant difference in body mass index, homeostasis model assessment of insulin resistance (HOMA‐IR), diastolic blood pressure (DBP), insulin, triglycerides, HDL‐C, apolipoprotein A1, hsCRP, TNF‐α, ceruloplasmin, and SAA levels between the lowest and highest quartiles of adiponectin (Table I). Plasma adiponectin levels were inversely correlated with body mass index (r=−.55, P<.001), HOMA‐IR (r=−.54, P<.001), triglyceride levels (r=−.36, P<.001), hsCRP levels (r=−.52, P<.001), and DBP values (r=−.23, P<.05) and positively correlated with HDL‐C levels (r=.56, P<.001) (Table II). The application of body mass index–adjusted partial correlation coefficients yielded a significant inverse correlation of adiponectin with HOMA‐IR (r=−.41, P<.001) and IL‐8 levels (r=−.40, P<.02) and a positive correlation with HDL‐C levels (r=.43, P<.001). In linear regression analysis using a stepwise regression model, HOMA‐IR, TNF‐α values, and HDL‐C levels were found to be independent factors associated with plasma adiponectin levels (Table III).

Table II.

 Simple and Partial Pearson’s Correlation Coefficients Between Adiponectin and Other Variables Measured

	Simple r	Partial r
BMI	−0.55 a	Adjusted
Insulin	−0.56 a	−0.41 a
HOMA‐IR	−0.54 a	−0.40 a
Triglycerides	−0.36 a	NS
HDL‐C	0.56 a	0.43 a
DBP	−0.23 b	NS
Log hsCRP	−0.52 a	NS
Log TNF‐α	−0.21	NS
Log SAA	−0.18	NS
Log IL‐8	−0.24	−0.40 b
Ceruloplasmin	−0.16	NS

Abbreviations: BMI, body mass index; DBP, diastolic blood pressure; HDL‐C, high‐density lipoprotein cholesterol; HOMA‐IR, homeostasis model assessment of insulin resistance; hsCRP, high‐sensitivity C‐reactive protein; IL‐interleukin‐8; SAA, serum amyloid A; TNF‐α, tumor necrosis factor α. ^a P<.001 or less; ^b P<.05. Values in boldface are statistically significant.

Table III.

 Linear Regression Analysis of Variables Associated With Adiponectin in All Patients

	Estimate	P Value
HOMA‐IR	−3.061	.003
Log TNF‐α	−2.746	.008
HDL‐C	2.294	.026
BMI	−1.383	.172

Abbreviations: BMI, body mass index; HOMA‐IR, homeostasis model assessment of insulin resistance; HDL‐C, high‐density lipoprotein cholesterol; TNF‐α, tumor necrosis factor α. The percentage of variance of adiponectin by these parameters was 58.8 (P<.001). Values in boldface are statistically significant.

In this study, we measured plasma concentrations of antioxidant proteins associated with HDL‐C such as PON‐1. No difference was detected in PON‐1 activity between the lowest and highest adiponectin quartiles.

To estimate the susceptibility of LDL‐C to oxidation, which shows the capacity of HDL‐C to protect LDL‐C, we calculated the lag time: the time required for the initiation of isolated LDL‐C oxidation by in vitro CuSO₄. Lag time values did not change with increasing adiponectin levels (Figure 2).

Figure 2.

 Lag time values according to adiponectin quartiles.

Discussion

Adiponectin is an adipocyte‐derived protein that has been closely and inversely associated with obesity. Low levels of adiponectin have been thought profoundly responsible for obesity‐related disorders including insulin resistance and atherosclerosis. The present study demonstrated that circulating adiponectin levels are lower in women with higher weight. The adiponectin concentration associates inversely and independently with insulin resistance as assessed by HOMA‐IR. In addition, a positive correlation exists between adiponectin and HDL‐C levels. These results are consistent with a previous report that serum adiponectin levels were negatively correlated with HOMA‐IR and positively correlated with HDL‐C values independent of age, sex, and body mass index. ¹⁵

The pathologic processes that are triggered by increased body weight and lead to impaired glucose metabolism are closely related to inflammatory conditions. A number of studies have investigated the association between adiponectin levels and proinflammatory markers in various populations. C‐reactive protein is a sensitive marker of inflammation and is considered to be an independent predictor of future risk for cardiovascular diseases. In previous studies, circulating adiponectin levels have been found to be negatively correlated with C‐reactive protein levels in nondiabetic, healthy individuals. ¹⁶ , ¹⁷ This association has not been detected in patients with coronary artery disease. ¹⁸ In our study group, which consisted of premenopausal women with body mass indices ranging from 16 to 50 kg/m², adiponectin concentrations seemed to be negatively correlated with hsCRP, but this relation no longer existed after the adjustment of body mass indices.

When the association of the proinflammatory cytokine TNF‐α with adiponectin levels was investigated, significantly low TNF‐α concentrations were observed in persons with high adiponectin levels. In a linear regression analysis, TNF‐α emerged as a significant predictor of adiponectin. It has been shown that the anti‐inflammatory potency of adiponectin is—at least in part—through the modulation of proinflammatory effects of TNF‐α. ¹⁹ The suppressive effect of TNF‐α on adiponectin gene expression or vice versa was also demonstrated by in vitro studies. ²⁰ , ²¹ Likewise, patients with the highest levels of adiponectin mRNA expression secreted the lowest levels of TNF‐α from adipose tissues. ²² Our findings support the evidence indicating a counteraction between adiponectin and TNF‐α. Furthermore, other markers of inflammation (SAA, ceruloplasmin, and IL‐8) seem to be at relatively low levels in those with high adiponectin levels. Altogether, it is possible to consider adiponectin as a potent anti‐inflammatory agent.

In linear regression analysis, adiponectin also emerged as a predictor of insulin resistance and HDL‐C levels. In our study, a strong link among inflammation, insulin resistance, and dyslipidemia evidenced with low HDL‐C appeared to be adiponectin. Among inflammatory parameters, TNF‐α was prominently involved in this association. Increased TNF‐α production due to a proinflammatory state suppresses adiponectin expression in adipose tissue, thus preventing the role of adiponectin in insulin sensitivity and lipid profile.

Low plasma HDL‐C levels are closely associated with the incidence of cardiovascular disease. As shown in the present study, those with high plasma adiponectin levels have high plasma HDL‐C values; this is in good agreement with the antiatherosclerotic properties of adiponectin. Since a recent study presented adiponectin as a dependent variable of serum PON‐1, ²³ we wanted to assess whether adiponectin exerts its antiatherogenic properties by affecting HDL‐C–linked antioxidant proteins such as PON‐1. However, we did not detect an association between PON‐1 activity and adiponectin levels. Despite low levels of HDL‐C, these patients sustained normal activity of PON‐1, possibly because of HDL‐C reorganization that furnished HDL‐C still functional.

In conclusion, our study showed that diminished adiponectin in plasma was associated closely with the development of cardiovascular risk factors such as insulin resistance and atherogenic lipid profile, while no evidence suggested the profound effect of adiponectin on HDL‐C–linked PON‐1 activity.