Low Adiponectin Levels Are Associated With Atherogenic Dyslipidemia and Lipid-Rich Plaque in Nondiabetic Coronary Arteries

Study subjects were enrolled in an IVUS substudy of the Diabetes Genome Project (DGP). Briefly, the DGP is a single-center prospective gene and biomarker banking registry designed to collect extensive clinical and anatomic information on patients undergoing coronary angiography. All patients consented to provide fasting blood samples. Patients aged <18 years with hemoglobin <9 g/dl and undergoing sedation within the past 12 h were not eligible for enrollment. Patients enrolled in the IVUS substudy underwent IVUS at the discretion of the operator for the following indications: 1) assessment of an indeterminate lesion; 2) investigation of a culprit lesion for optimal device sizing; 3) poststent assessment; or 4) clarification of coronary lesion extent. This study was approved by the Saint Luke's Hospital Institutional Review Board.

Vessels were imaged using IVUS-VH (Volcano, Rancho Cordova, CA) (8–12). In brief, IVUS-VH used the mathematical technique of autoregressive modeling to classify IVUS-radiofrequency data into one of four color-coded plaque components: fibrous (green), fibrofatty (light green), dense calcium (white), and necrotic core (red). IVUS was performed using a 20-MHz catheter (Eagle Eye Gold; Volcano). Retrograde imaging was performed using a continuous motorized pullback system at a speed of 0.5 mm/s (R-100 and TrakBack II; Volcano). Images were interpreted offline in our IVUS core laboratory. Software was used to reconstruct IVUS B-mode images from the radiofrequency data (pcVH; Volcano) and calculate geometric and composition data for each frame. Analysis of the medial-adventitial and luminal borders was performed in each analyzable cross-sectional frame for the entire pullback length.

Intra- and interuser validation studies have been performed between our IVUS core laboratory technicians and an IVUS technician from an outside core laboratory, with intraobserver correlation coefficients for the lumen cross-sectional area (CSA) and external elastic membrane (EEM) cross-sectional area (EEM_CSA), ranging from 0.98 to 0.99 and interobserver correlation coefficients ranging from 0.97 to 0.99. The mean percent relative differences among our IVUS technicians were 2.3 ± 2.5% for EEM_CSA and 4.4 ± 5.0% for lumen_CSA.

Qualitative analysis was performed according to the American College of Cardiology consensus on IVUS (13). Plaque volume was derived using Simpson's rule within the defined segment as EEM volume − lumen volume. Normalized plaque volume was defined as [plaque volume/(original pullback length)] × [median pullback length of all vessels in the cohort]. The region of interest for this analysis consisted of all frames within the entire IVUS pullback.

We further characterized plaque using a modified classification system developed by Virmani et al. (14) and used by others (15,16). Plaque was classified as follows: 1) early atherosclerotic plaque: IVUS-derived adaptive intimal thickening (containing <5% fibrofatty, <5% calcified, and <5% necrotic core); 2) intermediate atherosclerotic plaque: IVUS-derived pathological intimal thickening (plaque burden >40% and containing >5% fibrofatty, <5% calcified, and <5% necrotic core); 3) late-stage atherosclerotic plaque: IVUS-derived fibrocalcific (plaque burden >40% and containing >5% dense calcium and <5% necrotic core), IVUS-derived fibroatheroma (plaque burden >40% with >10% confluent necrotic core and evidence of an overlying fibrous cap), or IVUS-derived thin cap fibroatheroma (plaque burden >40% with >10% confluent necrotic core and no evidence of an overlying fibrous cap). Representative images for this classification scheme are shown in Fig. 1.

Diabetes was defined by American Diabetes Association (17) and World Health Organization (18) criteria, reported history, or treatment (pharmacologic or diet).Acute coronary syndromes were defined as either unstable angina, non–ST-segment elevation myocardial infarction, or ST-segment elevation myocardial infarction as defined by American College of Cardiology/American Heart Association guidelines. Insulin resistance was defined according to the homeostasis model assessment formula: (fasting insulin [microunits per milliliter] × fasting glucose [millimoles per liter]/22.5).

Adiponectin was measured using a Quantikine Human Adiponectin/Acrp30 immunoassay (R&D Systems, Minneapolis, MN). A description of this assay is available at http://www.rndsystems.com/pdf/drp300.pdf. Comprehensive lipoprotein profiles for each patient were obtained using the Vertical Auto Profile-II (19,20) (Atherotech, Birmingham, AL). In brief, the Vertical Auto Profile-II separates lipoprotein classes [HDL, LDL, VLDL, and lipoprotein(a), and intermediate-density lipoprotein] and subclasses (LDL₁–LDL₄, with LDL₁ being the most buoyant and LDL₄ most dense) by a single vertical spin density gradient ultracentrifugation using a Beckman Optima-XL 100K ultracentrifuge at 416,000g for 36 min. Separated lipoprotein classes and subclasses are then continuously drained from the bottom of the centrifuge tube into the Vertical Auto Profile-II continuous flow analyzer where they react sequentially with a cholesterol-specific enzymatic reagent producing a cholesterol concentration-dependent lipoprotein absorbance curve monitored by a spectrophotometer. The digital form of the absorbance curve is further deconvoluted using software to provide cholesterol concentrations of lipoprotein classes and subclasses. This procedure is highly reproducible and has been validated using the Lipid Research Clinics Beta Quantification reference method.

Continuous variables were compared using Student's t test or Wilcoxon's rank-sum test. Categorical variables were compared using a χ² or Fisher's exact test. All demographic summaries are presented as median (interquartile range [IQR]) for continuous variables and number (percent) for ordinal variables. Statistical significance was defined as P ≤ 0.05.The primary analysis was to describe an association between adiponectin levels and atherosclerotic burden and composition. Additional analyses were performed to identify associations between adiponectin and various lipoproteins. Adiponectin values were non-normally distributed according to the Shapiro-Wilk test (P < 0.0001); therefore, a log transformation was used to adhere to normal distribution assumptions (P = 0.9). Pearson product-moment correlation coefficients were calculated to test the correlation between log adiponectin and various markers of interest. Spearman's ρ and Kendall's τ, both nonparametric correlation tests, gave similar results. When adiponectin values were compared across population subgroups, Wilcoxon's rank-sum test was performed, and results are reported as median (IQR). Analyses were conducted using SAS (version 9.1; SAS Institute, Cary, NC).

Characteristics of the study population are presented in Table 1. On admission, 138 (75%) patients were receiving lipid-lowering drugs, the majority of which were statins (128 [93%]). Antihypertensive medications consisted of β-blockers (124 [67%]) and renin-angiotensin system (RAS) inhibitors (99 [54%]). Diabetic patients were more likely to be receiving a RAS inhibitor compared with nondiabetic patients (65 vs. 47%, P = 0.02). Among the 66 diabetic patients, 36 (55%) were treated with oral agents only, 14 (21%) with insulin only, 8 (12%) with oral and insulin therapy, 5 (8%) with no therapy, and 3 (5%) with diet only. Oral diabetic agents consisted of biguanides (n = 27), sulfonylureas (n = 19), thiazolidinediones (n = 6), or a combination (n = 7).

The median (IQR) adiponectin level for the study group was 6.1 (2.5–9.3) μg/ml. Adiponectin levels were significantly lower in men than in women: 5.5 (3.4–8.4) vs. 8.7 (4.5–15.0) μg/ml (P < 0.001). Adiponectin correlated with age (r = 0.3, P < 0.0001), BMI (r = −0.17, P = 0.02), homeostasis model assessment (r = −0.17, P = 0.02), and fasting insulin (r = −0.20, P = 0.008). Adiponectin did not correlate with diabetes, A1C, fasting glucose, or albuminuria (data not shown).

Adiponectin levels correlated with atherogenic lipoproteins. Adiponectin correlated with triglycerides (r = −0.27, P = 0.0002) and HDL cholesterol (r = 0.4, P < 0.001). Although there was no significant correlation between adiponectin and total or LDL cholesterol, there were significant associations with the small dense LDL₃ (r = −0.27, P = 0.003) and LDL₄ (r = −0.25, P = 0.0005) subfractions and the larger, more buoyant LDL₂ (r = 0.18, P = 0.015).

Vessels analyzed consisted of the left anterior descending artery (92 [50%]), the right coronary artery (50 [27%]), the left circumflex artery (35 [19%]), and the left main artery, branch vessel, or ramus intermediate (8 [4%[). Median (IQR) IVUS pullback length was 57.5 (38.2–76.9) mm. Normalized plaque volume was 346.3 (272.4–475.6) mm³. By composition, normalized plaque volumes were 96.3 (58.8–148.6) mm³ for fibrous, 28.4 (16.6–50.5) mm³ for fibrofatty, 11.2 (4.2–23.7) mm³ for dense calcium, and 18.1 (10.7–32.7) mm³ for necrotic core.

Adiponectin was associated with normalized plaque (r = −0.16, P = 0.025) and fibrofatty (Fig. 2A) volume and percent fibrofatty composition (r = −0.19, P = 0.009). There was also an association with normalized fibrous volume (r = −0.18, P = 0.013) but no association with normalized volume of calcium (r = 0.008, P = 0.9) or necrotic core (r = 0.13, P = 0.13). There were also quantitative differences in fibrofatty content by adiponectin quartiles (Table 2 and Fig. 2B). Normalized volume values for other plaque constituents are shown in Table 2.

Adiponectin levels were similar in diabetic and nondiabetic patients (median 5.70 [IQR 3.3–9.1] vs. 6.2 [3.6–10.3] μg/ml, respectively, P = 0.49). In nondiabetic patients, adiponectin correlated with plaque volume (r = −0.22, P = 0.01), fibrofatty volume (r = −0.23, P = 0.01), and percent fibrofatty composition (r = −0.23, P = 0.009); however, there were no significant associations in diabetic patients. With increasing adiponectin quartiles, normalized fibrofatty volume decreased in nondiabetic but not diabetic patients (Table 2).

Among all patients, adiponectin did not correlate with IVUS-derived adaptive intimal thickening, IVUS-derived thin cap fibroatheroma, or IVUS-derived fibroatheroma; however, there was a significant correlation with IVUS-derived pathological intimal thickening (r = −0.18, P = 0.01) and with IVUS-derived fibrocalcific (r = 0.17, P = 0.02). Increasing adiponectin quartiles were also associated with significantly lower median (IQR) percent contributions of IVUS-derived pathological intimal thickening (Q1 16.7 [9.5–33.9]; Q2 12.3 [3.6–34.0]; Q3 15.9 [4.3–27.6]; and Q4 6.4 [1.5–20.8]; P = 0.03). In nondiabetic patients, adiponectin also correlated with IVUS-derived pathological intimal thickening (r = −0.20, P = 0.03), whereas the association was not significant in diabetic patients (r = −0.16, P = 0.20). With increasing adiponectin quartiles, the percent contributions of IVUS-derived pathological intimal thickening also declined in the nondiabetic cohort (Q1 20.4 [11.7–36.3]; Q2 11.5 [3.6–24.8]; Q3 10.8 [1.5–26.8]; and Q4 8.0-[3.1, 21.8]; P = 0.04).

The novel findings of this study include a 1) correlation between adiponectin and quantitative IVUS measures of coronary atherosclerosis, 2) correlation between adiponectin and plaque lipid content in IVUS-VH entire pullback analysis; and 3) higher frequency of IVUS-derived pathological intimal thickening in patients with lower adiponectin levels. These findings were seen mostly in nondiabetic patients. Our findings are consistent with those of others (5) regarding the association between adiponectin and atherogenic dyslipidemia including elevated triglycerides, low HDL, and small dense LDL cholesterol .

Adiponectin is a unique adipokine, downregulated in the presence of increasing central adiposity and associated with insulin resistance, inflammation, risk for metabolic syndrome, type 2 diabetes (3), decreased LDL particle size, and small dense HDL. Increased levels have also been associated with reduced risk of myocardial infarction even after adjustment for traditional cardiovascular risk factors (4). However, the association between serum adiponectin and coronary events remains controversial (21–24). Prior studies have shown a significant inverse correlation between coronary lumen narrowing as assessed by angiography and plasma adiponectin levels (25,26). To date, no studies have established a relationship between adiponectin and atherosclerotic burden and plaque composition.

Recent studies indicate that elevated adiponectin levels are associated with adverse outcomes in patients with established coronary atherosclerosis (24,27–29). This work suggests a complex relationship between adiponectin and atherosclerosis. Although our study does not resolve this issue, it does suggest that low levels of adiponectin are associated with relatively early plaque development and risk of plaque maturation. Consistent with the findings of others that high adiponectin levels are associated with adverse events in patients with advanced atherosclerosis, we found a positive correlation between higher adiponectin levels and the advanced (IVUS-derived fibrocalcific) plaque phenotype. However, this apparent paradox can only be resolved by a longitudinal study with serial adiponectin and IVUS measurements and collection of vital statistics.

Classification of lesion morphology (14) has been used in prior histopathology and IVUS studies (15,16). Adaptive and pathological intimal thickening are thought to represent early atherosclerosis. Adaptive intimal thickening is characterized by the presence of smooth muscle cells and the absence of foam cells, thrombus, lipid, and necrotic core. Whereas thrombus and necrotic core are also absent, increased lipid accumulation is implicit in the transformation from adaptive to pathological intimal thickening (14,16). Although extensively studied, lipid accumulation in the vessel wall is incompletely understood. It is believed that transport of lipoproteins across the endothelial cell monolayer is an initial step in atherogenesis and is also probably enhanced in the presence of oxidized LDL. A biological association between adiponectin and atherosclerosis seems plausible. In our study, low adiponectin levels were associated with a higher prevalence of small dense LDL particles. LDL may be indirectly associated with plaque lipid accumulation via oxidized small dense LDL particles. Small dense LDL is associated with endothelial cell injury and increased permeability (30). Further, adiponectin may play a regulatory role in foam cell maturation. At physiologic concentrations of adiponectin, the expression of class A macrophage scavenger receptor is suppressed. Adiponectin also dose dependently decreases class A macrophage scavenger receptor ligand binding and uptake activities, suggesting a preventive role in foam cell maturation and subsequent atherosclerosis progression (31).

Despite the novelty of our findings, there are several important limitations . This was an association study and does not establish a causal relationship between adiponectin and lipid accumulation in human atherosclerosis. Our correlation coefficients between adiponectin and lipid volume are modest. Adiponectin accounted for ∼5–6% of the variability in plaque lipid accumulation; thus, other biomarkers may be principally involved in lipid accumulation. It is also conceivable that the association between adiponectin and lipid in plaque was weakened by our use of a nonsensitive adiponectin assay that did not quantify high molecular weight multimers (32). Emerging data suggest that the high molecular weight isomer is bioactive (33). Although we found associations between low adiponectin levels and measures of plaque burden and increased lipid content, the clinical implications of these findings are currently unknown. There are no IVUS-VH data demonstrating that lipid-rich plaque is a requisite intermediate step in the progression to advanced plaque phenotypes (thin cap fibroatheroma and fibroatheroma). Another confounding issue for adiponectin levels is the imbalance in the use of concomitant medications (6). RAS blockers and peroxisome proliferator–activated receptor-γ agonists have been shown to increase adiponectin levels. RAS blockers were used in 54% of patients and were more common in diabetic patients. Furthermore ∼10% of diabetic patients received thiazolidinediones.

In a cohort of patients with coronary artery disease, lower adiponectin levels are associated with small dense LDL cholesterol, increased plaque volume as measured by lipid-rich atheroma, and a higher prevalence of IVUS-derived pathological intimal thickening in nondiabetic patients, suggesting an antiatherogenic role for adiponectin.

This work was supported by a grant from the American Diabetes Association Amaranth Diabetes Fund.

We thank Jose Aceituno and Joseph Murphy for publication assistance.