Variants in adiponectin signaling pathway genes show little association with subclinical CVD in the diabetes heart study


  • Amanda J. Cox,

    1. Center for Human Genomics, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    2. Center for Diabetes Research, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    3. Department of Biochemistry, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    Search for more papers by this author
    • Disclosure: The authors declared no conflicts of interest.

  • John E. Lambird,

    1. Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    Search for more papers by this author
  • S. Sandy An,

    1. Center for Human Genomics, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    2. Center for Diabetes Research, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    3. Department of Biochemistry, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    Search for more papers by this author
  • Thomas C. Register,

    1. Center for Diabetes Research, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    2. Department of Pathology - Comparative Medicine, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    3. Department of Radiologic Sciences, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    Search for more papers by this author
  • Carl D. Langefeld,

    1. Department of Biostatistical Sciences, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    Search for more papers by this author
  • J.Jeffrey Carr,

    1. Department of Radiologic Sciences, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    Search for more papers by this author
  • Barry I. Freedman,

    1. Department of Internal Medicine-Nephrology, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    Search for more papers by this author
  • Donald W. Bowden

    Corresponding author
    1. Center for Diabetes Research, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    2. Department of Biochemistry, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    • Center for Human Genomics, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA
    Search for more papers by this author

  • Funding agencies: This study was supported in part by R01 HL67348, R01 HL09230, and R01 NS058700 to Dr Donald W Bowden.

Correspondence: Donald W Bowden (



Understanding the interplay between adiposity, inflammation, and cardiovascular complications in type 2 diabetes mellitus (T2DM) remains a challenge. Signaling from adipocytes is considered important in this context. Adiponectin is the most abundant adipocytokine and has been associated with various measures of cardiovascular disease (CVD). This study examines the relationships between genetic variants in the adiponectin (ADIPOQ) and adiponectin-related signaling pathway genes and measures of subclinical CVD (vascular calcified plaque and carotid intima-media thickness), plasma lipids, and inflammation in T2DM.

Design and Methods

Single-nucleotide polymorphisms (SNPs) in ADIPOQ (n = 45), SNPs tagging ADIPOR1 (n = 6), APIPOR2 (n = 8), APPL1 (n = 6) and known rare coding variants in KNG1 (n = 3) and LYZL1 (n = 3) were genotyped in 1220 European Americans from the family-based Diabetes Heart Study. Associations between SNPs and phenotypes of interest were assessed using a variance components analysis with adjustment for age, sex, T2DM-affected status, and body mass index.


There was minimal evidence of association between SNPs in the adiponectin signaling pathway genes and measures of calcified plaque; eight of the 71 SNPs showed evidence of association with subclinical CVD (P = 0.007-0.046) but not with other phenotypes examined. Nine additional SNPs were associated with at least one of the plasma lipid measures (P = 0.008-0.05).


Findings from this study do not support a significant role for variants in the adiponectin signaling pathway genes in contributing to risk for vascular calcification in T2DM. However, further understanding the interplay between adiposity, plasma lipids, and inflammation may prove important in the prediction and management of cardiovascular complications in T2DM.


Adipose tissue has gained acceptance as an active endocrine organ that may mediate the complex interplay between obesity, metabolic disease, and associated co-morbidities [1]. Of the many adipokines recognized as playing important roles in adipocyte signaling, adiponectin is the most abundant. Paradoxically, although adiponectin is produced primarily by adipocytes, plasma concentrations decrease with increases in fat mass [2]. Reduced plasma levels of adiponectin are correlated with reduced fasting insulin [2] and impaired glucose tolerance [3], suggesting that adiponectin signaling may be important in the regulation of glycemic control and changes in adiponectin signaling may be a potential risk factor for the development of type 2 diabetes mellitus (T2DM).

Adiponectin is encoded for by a 15.8-kb gene (ADIPOQ) located on chromosome 3q27. This locus has been previously identified as a diabetes susceptibility locus [4, 5] and provides evidence that variants in ADIPOQ may contribute to risk for T2DM. Adiponectin is a 30-kDa protein known to self-assemble into three types of quaternary structures: low-molecular-weight (LMW), middle-molecular-weight (MMW), and high-molecular-weight (HMW) adiponectin. The HMW form of adiponectin appears to be the most functionally active [6, 7]. Interestingly, a number of low-frequency coding single-nucleotide polymorphisms (SNPs) have been identified within ADIPOQ that result in a loss of higher order structure formation [8, 9], demonstrating the potential for genetic variation to influence adiponectin function. Genome-wide association studies have also revealed other genetic loci, including LYZL1 [10] and KNG1 [11], to be associated with adiponectin concentrations; while the relationship between these genes and adiponectin concentrations remains unclear, they were also evaluated. Furthermore, genetic variants in the adiponectin receptors (ADIPOR1 and ADIPOR2) have also been associated with T2DM and related phenotypes [12, 13]. Considering other components of the adiponectin signaling pathway, such as the intracellular signaling intermediary APPL1 [14], when attempting to understand the genetic regulation of adiponectin signaling may also be informative.

In addition to contributing to risk for T2DM, adiponectin signaling may also be important in the development of T2DM-associated co-morbidities, including cardiovascular disease (CVD), the most common cause of death in T2DM [15]. Lower circulating adiponectin concentrations have been observed in CVD patients compared to controls in a number of cross-sectional studies [3, 16] and are associated with increased disease severity [17] and poorer prognosis [18] in individuals with established coronary artery disease (CAD). Also of note, a limited number of studies have reported associations between variants in ADIPOQ, ADIPOR1, and ADIPOR2 and CVD [19]. These studies included small numbers of SNPs (two to eight) and diverse phenotypes, leaving the exact contribution of variants in the adiponectin signaling pathway to CVD risk unclear. To date, the most comprehensive study examined the association between 11 ADIPOQ SNPs and measures of subclinical CVD in the Multi-Ethnic Study of Atherosclerosis (MESA) cohort [23]. There was no evidence of association between the ADIPOQ SNPs and the selected disease phenotypes in the subset of 712 Caucasian Americans; however, stronger relationships were observed for the African American and Hispanic subsets.

Given these limited and conflicting findings, this study examined the relationships between variants in ADIPOQ, and other genes involved in the adiponectin signaling pathway, with measures of subclinical CVD, and other known CVD risk factors, in a high-risk sample enriched for T2DM and CVD.

Methods and Procedures

Study design and sample

This investigation included 1220 self-described European American (EA) individuals (1020 T2DM-affected and 200 T2DM-unaffected individuals from 475 families) from the Diabetes Heart Study (DHS) cohort. Briefly, the DHS includes siblings concordant for T2DM but without advanced renal insufficiency. T2DM was clinically defined as diabetes developing after the age of 35 years and treated with insulin and/or oral agents in the absence of historical evidence of ketoacidosis. Ascertainment and recruitment have been described in detail previously [24, 25].

Study protocols were approved by the Institutional Review Board at Wake Forest School of Medicine, and all participants provided written informed consent prior to participation. Participant examinations were conducted in the General Clinical Research Center of the Wake Forest Baptist Medical Center and included interviews for medical history and health behaviors, anthropometric measures, resting blood pressure, electrocardiography, fasting blood sampling for laboratory analyses, and spot urine collection. Standard laboratory analyses included fasting glucose, glycated hemoglobin (HbA1C), total cholesterol, high-density lipoprotein cholesterol (HDL-C), and triglycerides. Low-density lipoprotein cholesterol (LDL-C) concentration was calculated using the Freidewald equation, and results were considered valid for subjects whose triglycerides were <796 mg/dL. High-sensitivity C-reactive protein (CRP) was measured using an ELISA kit from the American Laboratory Products Company (ALPCO, Windham, NH). Adiponectin concentrations were determined in a subset of participants (n = 450) using RIA kits and protocols from LINCO Research (St. Charles, MO). Intra- and inter-assay coefficients of variation in immunoassay were <7 and <10%, respectively.

Coronary artery calcified plaque (CAC), carotid artery calcified plaque (CarCP), and infra-renal abdominal aortic calcified plaque (AACP) were measured using fast-gated helical CT scanners and calcium scores calculated as previously described and reported as an Agatston score [26, 27]. Carotid artery intima-media thickness (IMT) was measured by high-resolution B-mode ultrasonography with a 7.5-MHz transducer and a Biosound Esaote (AU5) ultrasound machine (Biosound Esaote, Indianapolis, IN) as previously described [28]. Not all measurements were available for all participants.

Genetic analysis

Total genomic DNA was purified from whole blood samples using the PUREGENE DNA isolation kit (Gentra, Minneapolis, MN). DNA concentration was quantified using standardized fluorometric readings on a Hoefer DyNA Quant 200 fluorometer (Hoefer Pharmacia Biotech, San Francisco, CA). Samples were diluted to a final concentration of 5 ng/μl.

SNPs in ADIPOQ (n = 45), as well as SNPs tagging ADIPOR1 (n = 6), ADIPOR2 (n = 8), APPL1 (n = 6), and known rare variants in KNG1 (n = 3), LYZL1 (n = 3), were selected for genotyping. Of these 71 SNPs, 22 were known coding variants selected from the literature [8, 21, 29] or rare variants identified from resequencing of the ADIPOQ gene or from the National Heart, Lung, and Blood Institute Exome Sequence Project database ( Existing literature further informed additional variant selection [11, 12, 14, 20, 22, 23]. In addition, common SNPs across the ADIPOQ gene were selected from the HapMap database (, and using the greedy pair-wise tagging algorithm implemented in the Tagger program of Haploview version 4.2 (Broad Institute, Cambridge, MA), 38 tag-SNPs were identified with an r2 threshold of 0.8 and MAF > 5% for inclusion in genotype design. Genotypes were determined using a MassARRAY SNP Genotyping System (Sequenom, San Diego, CA). Primers for PCR amplification and extension reactions were designed using the MassARRAY Assay Design Software (Sequenom, San Francisco, CA). Single-base extension reaction products were separated and scored using a matrix-assisted laser desorption ionization/time of flight mass spectrometer.

A total of 41 quality control samples were included as blind duplicates in the genotyping analysis to allow for evaluation of genotyping accuracy. The concordance rate for these blind duplicates was 100%. For all SNPs, the minimum acceptable call rate was 95%; the average call rate was 98.6 ± 0.7% (mean ± SD). Samples with genotyping efficiency rates <90% were excluded from further analysis; the average genotyping efficiency rate was 99.7 ± 0.01% (mean ± SD).

For ADIPOQ, ADIPOR1, ADIPOR2, and APPL1, genotype data were obtained from the HapMap database and were used to assess the ability of the genotyped SNPs to capture genotypic variation across each gene region defined by an average r2 value.

Statistical analysis

Relationships between adiponectin concentrations and CAC, CarCP, AACP, and IMT measures and CRP concentrations were examined using a Pearson correlation; variables were log-transformed prior to analysis to approximate conditional normality. For genotype data, allele and genotype frequencies were calculated for each SNP using a subset of unrelated individuals and tested for departures from Hardy–Weinberg equilibrium using a Chi-squared goodness-of-fit test. Association between each of the 71 SNPs and phenotypes of interest was examined using variance components methods implemented in Sequential Oligogenic Linkage Analysis Routines (SOLAR) version 4.2.7 (Texas Biomedical Research Institute, San Antonio, TX) to account for relatedness between subjects [32]. All associations were examined using additive, dominant, and recessive genetic models to determine the model of best fit. Again, continuous variables were transformed prior to analysis to approximate conditional normality and to reduce heterogeneity of residual phenotypic variance across SNP genotypes. Age, gender, diabetes-affected status, and BMI were included as covariates in all analyses. Phenotypes of interest included measures of subclinical CVD (vascular calcified plaque scores and IMT) as well as total cholesterol, HDL-C, LDL-C, and triglyceride concentrations, all known risk factors for CVD. Statistical significance was accepted at P < 0.05.


The demographic and clinical characteristics of the DHS cohort are presented in Table 1. As anticipated in a sample enriched for T2DM, a predominance of known CVD risk factors, including high body mass, hypertension, and abnormal lipid concentrations, was evident. Vascular calcified plaque scores for the coronary artery, carotid artery, and abdominal aorta reflect a substantial burden of subclinical CVD in these diabetes-affected individuals and their siblings. Adiponectin concentrations were poorly correlated with CAC, IMT, total cholesterol, and LDL-C (Table 2). Modest inverse relationships were observed between adiponectin and triglyceride and adiponectin and CRP concentrations, and a strong positive relationship was observed between adiponectin and HDL-C concentrations (Table 2).

Table 1. Demographic characteristics of the DHS cohort
 Mean ± SD or %
  1. Either coronary artery bypass grafting, coronary angiography or carotid endarterectomy.
Age (years)61.5 ± 9.4
Gender (% female)53.3%
Height (cm)168.6 ± 9.7
Weight (kg)90.6 ± 20.3
BMI (kg/m2)31.8 ± 6.5
Diabetes affected (%)83.6%
Diabetes duration (years)10.4 ± 7.2
Length of follow-up (years)7.3 ± 2.3
Smoking current or past (%)59.1%
Systolic BP (mmHg)139 ± 20
Diastolic BP (mmHg)73 ± 10
Hypertension (%)86.2%
Lipid lowering medication (%)49.6%
Anti-hypertensive medication (%)62.5%
Laboratory measures 
Total cholesterol (mg/ dL)186.8 ± 42.1
HDL-C (mg/dL)43.1 ± 12.5
LDL-C (mg/dL)105.4 ± 32.6
Triglycerides (mg/dL)200.1 ± 131.2
Glucose (mg/dL)138.2 ± 54.9
HbA1C (%)7.3 ± 1.7
CRP (mg/L)6.0 ± 9.9
Adiponectin (mg/L)28.5 ± 30.1
Self-reported history of 
Cardiovascular intervention*18.6%
Vascular imaging 
CAC1672 ± 3179
CarCP312.5 ± 670.9
AACP11158 ± 15939
IMT0.693 ± 0.143
Table 2. Pearson correlation coefficients for relationships between serum adiponectin concentrations and CVD risk phenotypes
TraitnPearson correlation coefficientP value
Total cholesterol433−0.0350.464

Using available HapMap data, SNPs selected for genotyping in ADIPOQ, ADIPOR1, ADIPOR2, and APPL1-captured genetic variation across each of the selected gene regions with average r2 values of 0.804, 0.662, 0.509, and 0.576, respectively. These coverage estimates are potentially underestimated as many more SNPs, particularly in ADIPOQ, were genotyped for which HapMap data were not available. Genotype frequencies for all SNPs were consistent with Hardy–Weinberg proportions (P = 0.12-1.00), with the exception of rs3818551 in LYZL1 (P < 0.001). Of the 71 SNPs, 12 were monomorphic in this sample (supplementary material).

Association results for SNPs showing nominal or trending evidence of association with one or more of the primary phenotypes of interest are summarized in Table 3. Association results for all genotyped SNPs are included in the supplementary material. Three SNPs in ADIPOQ (rs6810075, rs17300539, rs16861209; P = 0.014-0.049) and three SNPs in ADIPOR2 (rs16928751, rs9805042, rs767870; P = 0.023-0.035) were nominally associated with adiponectin concentrations directly. The impact of these SNPs on adiponectin concentrations was moderate, with concentrations differing between major and minor allele homozygotes on average by 25-35%; however, the number of minor allele homozygotes was limited (ranging from n = 2-39; Table 3). These particular SNPs did not show any evidence of association with any of the clinical phenotypes examined.

Table 3. Association results for AdipoQ signaling pathway genes and CVD risk traits in the DHS sample
      Mean trait values ± SD (n)Association results
GeneSNPLocationMAFAlleles (1/2)Trait1/11/22/2Modelβ ± SEP value
  1. Mean trait values ± standard deviation (SD) are reported. Association results are expressed as beta-value ± standard error (SE) and reported for the model of best fit (A = additive, D = dominant, R = recessive) adjusted for age, sex, T2DM-affected status and BMI. MAF = minor allele frequency; allele 1= major allele; allele 2 = minor allele.
ADIPOQrs6810075Upstream0.317T/CAdiponectin30.2 ± 29.7 (210)28.2 ± 31.0 (202)20.8 ± 27.1 (39)A−0.171 ± 0.0730.019
ADIPOQrs16861194Upstream0.082A/GCholesterol187 ± 42 (971)184 ± 39 (182)258 ± 91 (4)R0.273 ± 0.1110.014
ADIPOQrs17300539Upstream0.080G/AAdiponectin26.7 ± 28.7 (386)39.1 ± 36.0 (59)40.0 ± 33.9 (6)A0.280 ± 0.1140.014
ADIPOQrs16861205Intronic0.084G/ACholesterol187 ± 42 (963)185 ± 39 (179)258 ± 91 (4)R0.274 ± 0.1110.014
Triglycerides198 ± 132 (963)205 ± 126 (179)341 ± 230 (4)A0.087 ± 0.0450.050
ADIPOQrs16861209Intronic0.091C/AAdiponectin26.6 ± 28.7 (443)38.6 ± 35.5 (6)35.2 ± 33.5 (2)D0.245 ± 0.1240.049
ADIPOQrs822391Intronic0.191T/CCRP0.558 ± 0.90 (639)0.665 ± 1.12 (291)0.750 ± 1.20 (45)A0.177 ± 0.0700.011
LDL-C105.5 ± 32.0 (639)103.8 ± 33.5 (291)115.7 ± 33.2 (45)R12.5 ± 4.80.009
ADIPOQrs822394Intronic0.175C/ACRP0.556 ± 0.89 (666)0.693 ± 1.20 (279)0.652 ± 0.81 (38)A0.202 ± 0.0720.005
Cholesterol187 ± 42 (797)185 ± 42 (316)196 ± 42 (46)R0.067 ± 0.0340.048
LDL-C105.6 ± 32.5 (747)103.6 ± 32.6 (299)114.3 ± 32.0 (38)R12.1 ± 5.10.019
ADIPOQrs822396Intronic0.174A/GCholesterol187 ± 42 (796)185 ± 41 (318)198 ± 39 (45)R0.086 ± 0.0340.012
LDL-C105.6 ± 32.5 (745)103.6 ± 32.8 (299)109.0 ± 29.2 (37)R13.8 ± 5.20.008
ADIPOQrs2036373Intronic0.064T/GAACP26246 ± 34680 (737)9664 ± 12890 (101)11218 ± 16011 (7)R61.7 ± 23.10.008
ADIPOQrs3821799Intronic0.465C/TIMT0.691 ± 0.15 (309)0.698 ± 0.14 (543)0.686 ± 0.14 (224)R−0.014 ± 0.0060.014
CRP0.650 ± 1.10 (279)0.619 ± 1.10 (496)0.482 ± 0.74 (206)R−0.251 ± 0.0980.011
ADIPOQrs3774261Intronic0.401G/AIMT0.696 ± 0.15 (404)0.697 ± 0.14 (483)0.676 ± 0.13 (182)R−0.019 ± 0.0060.002
ADIPOQrs17366743Exon30.029T/CCAC1721 ± 3250 (1049)901 ± 1446 (63)16 (1)D−0.674 ± 0.3060.028
ADIPOQrs67739573′-UTR0.400G/AIMT0.697 ± 0.15 (402)0.699 ± 0.14 (502)0.676 ± 0.13 (182)R−0.019 ± 0.0060.002
ADIPOQrs76393523′-UTR0.286C/TLDL-C105.6 ± 33.6 (570)104.2 ± 30.7 (420)110.6 ± 35.4 (97)R8.4 ± 3.50.015
ADIPOQrs64441753′-UTR0.285G/ACholesterol186 ± 41 (605)187 ± 44 (450)193 ± 39 (103)R0.053 ± 0.0230.020
LDL-C104.5 ± 32.7 (569)105.1 ± 32.7 (420)111.7 ± 30.9 (97)R8.4 ± 3.50.015
ADIPOQrs13085499Downstream0.484A/GCRP0.645 ± 1.07 (254)0.631 ± 1.04 (505)0.474 ± 0.47 (222)R−0.262 ± 0.0950.006
ADIPOR1rs10494839Intronic0.325T/CCAC1776 ± 2532 (504)1786 ± 3014 (488)1532 ± 3465 (124)D0.323 ± 0.1390.020
ADIPOR1rs12733285Intronic0.269C/TAACP11247 ± 15491 (439)11853 ± 16876 (338)7172 ± 13620 (68)R−18.24 ± 7.890.021
ADIPOR1rs1342387Intronic0.419C/TTriglycerides212 ± 139 (389)197 ± 135 (554)185 ± 101 (212)A−0.050 ± 0.0240.034
ADIPOR1rs75395423′-UTR0.283C/GCholesterol186 ± 42 (588)186 ± 41 (470)196 ± 45 (95)R0.052 ± 0.0240.030
ADIPOR1rs10920531Downstream0.351C/ACRP0.676 ± 1.12 (389)0.545 ± 0.92 (475)0.559 ± 0.72 (119)D−0.247 ± 0.0830.003
ADIPOR2rs767870Intronic0.165A/GAdiponectin27.3 ± 29.7 (313)30.7 ± 31.0 (120)36.9 ± 30.6 (16)A0.196 ± 0.0880.025
ADIPOR2rs16928751Exon 50.142G/AAdiponectin27.2 ± 29.3 (338)32.1 ± 32.4 (99)37.0 ± 30.0 (13)A0.216 ± 0.0950.023
ADIPOR2rs9805042Exon 60.145C/TAdiponectin27.4 ± 29.4 (333)31.2 ± 32.1 (104)35.4 ± 29.7 (14)A0.201 ± 0.0930.032
APPL1rs11544593Exon220.178A/GIMT0.690 ± 0.15 (736)0.697 ± 0.14 (296)0.729 ± 0.13 (43)A0.009 ± 0.0040.037
APPL1rs4640525Intronic0.423G/CTriglycerides199 ± 137 (387)197 ± 131 (560)209 ± 120 (210)D0.087 ± 0.0420.050
APPL1rs19133023′-UTR0.365A/GAACP11520 ± 16004 (351)11597 ± 16166 (368)9063 ± 15095 (122)R−15.3 ± 6.110.013
KNG1rs5030025Intronic0.223A/CAACP11696 ± 16032 (505)10296 ± 15953 (295)10757 ± 14835 (44)A−7.41 ± 3.710.046
LYZL1rs41289053Exon 50.055G/ACRP0.569 ± 0.96 (881)0.833 ± 1.21 (678)1.28 ± 0.59 (5)A0.380 ± 0.1240.002

Overall, there was limited evidence of association between SNPs in the adiponectin signaling pathway genes and measures of vascular calcified plaque. Only rs17366743 in ADIPOQ (β ± SE: −0.674 ± 0.0306; P = 0.028) and rs1044839 in ADIPOR1 (0.323 ± 0.139; P = 0.020) were nominally associated with CAC. In addition, only rs2036373 in ADIPOQ (61.7 ± 23.1; P = 0.008), rs12733285 in ADIPOR1 (−18.24 ± 7.89; P = 0.021), rs1913302 in APPL1 (−15.3 ± 6.11; P = 0.013), and rs5030025 in KNG1 (−7.41 ± 3.71; P = 0.046) were associated with AACP. There was no evidence of association with CarCP. None of the SNPs associated with CAC or AACP were associated with any of the other phenotypes examined (or with adiponectin, as noted above).

Other results included evidence of significant association between three SNPs in ADIPOQ (rs3774261, rs3821799, and rs6773957) and one SNP in APPL1 (rs11544593) with IMT (P = 0.002-0.037). Furthermore, four SNPs in ADIOPQ (rs3821799, rs13085499, rs822391, and rs822394; P = 0.005-0.011), one SNP in ADIPOR1 (rs10920531; P = 0.003), and one SNP in LYZL1 (rs41289053; P = 0.002) were found to be associated with CRP concentrations. Notably, rs3821799 was associated with both IMT and CRP.

Finally, a smaller number of SNPs (seven SNPs in ADIPOQ, two SNPs in ADIPOR1, and one SNP in APPL1) were found to be associated with plasma lipid measures (Table 3). Of note, rs16861205 in ADIPOQ was associated with total cholesterol and triglycerides (P = 0.014 and P = 0.05, respectively). Three other SNPs in ADIPOQ (rs822394, rs822396, and rs6444175) were also found to be associated with total cholesterol and LDL-C (P = 0.008-0.048) with rs822394 also associated with CRP.


Although adiponectin has received much attention over the last decade in the context of risk for T2DM, the potential role of adiponectin signaling in the development of diabetic complications is still unclear. This study examined the association between 71 SNPs (including 22 known common and low-frequency coding variants) in the adiponectin signaling pathway genes ADIPOQ, ADIPOR1, ADIPOR2, APPL1, and KNG1 and measures of subclinical CVD in European American families enriched for T2DM. Although there were numerous nominal associations, when taking into consideration the number of SNPs tested (i.e. 71), overall, there was limited evidence to support a major contribution of variants in these adiponectin signaling pathway genes to risk for subclinical CVD in the form of vascular calcified plaque or carotid wall thickness in European Americans in the DHS. Patterns of association with other CVD risk phenotypes, including total cholesterol, LDL-C, and CRP were sporadic and at best modest.

One important observation from the current study was the limited association between variants in ADIPOQ and adiponectin concentrations. While a small number of nominal associations were observed between this set of SNPs and adiponectin concentrations, several SNPs previously reported to be associated with adiponectin concentrations in European-derived samples, including rs17300539, rs266729, and rs2241766 [33, 34] showed no evidence of association in the current study. These findings may be interpreted as indicating that adiponectin concentrations are unlikely to be influenced to a significant degree by these, mostly common, variants in European Americans with T2DM. Alternatively, given that adiponectin concentrations were available in only a proportion of our sample (n = 450), it is possible that the current study was not sufficiently powered to detect SNPs making small contributions to the variance in adiponectin concentrations at a level considered to be statistically significant. In addition, a number of variants selected specifically for inclusion in the current study based on prior evidence of association with adiponectin concentrations in other population groups (African Americans, Hispanic Americans, and Asians) [8, 9, 35] were either found to be monomorphic in this sample (see supplementary material) or were not associated with adiponectin concentrations. The lack of association with adiponectin concentrations is an important observation in the context of our study as we anticipated that if adiponectin concentrations influenced subclinical CVD risk, then polymorphisms associated with adiponectin would also be likely to show association with CVD phenotypes. Associations with vascular calcification were generally not seen in the current study. Furthermore, the generally poor correlation between adiponectin concentrations and vascular calcified plaque suggests that these traits are regulated independently of each other and that variants in adiponectin signaling are unlikely to be an important determinant of CVD risk in T2DM.

Previous investigation of the role of ADIPOQ SNPs in CVD risk has been limited and produced conflicting results. The ADIPOQ SNP rs2241766 was reported to be associated with angiography-defined CAD in a study of 418 individuals with T2DM [19], but these findings contradicted an earlier study examining associations with incident CVD in a case-control cohort of 1251 males [20]. Similarly, the SNP rs1501299 was reported to be associated with CAD in a case-control sample of 376 individuals with T2DM [36]; subsequent studies failed to replicate this finding [19, 20]. In the current study, the association for rs2241766 was not replicated with measures of vascular calcified plaque. While rs1501299 was not genotyped, available HapMap data suggest that this SNP is in complete linkage disequilibrium with rs2241767 and rs3821799 (supplementary material); however, we did not find either rs2241767 or rs3821799 to be significantly associated with measures of vascular calcification. Our findings indicated a general lack of association between ADIPOQ SNPs and vascular calcified plaque. In another relatively large sample, 11 ADIPOQ SNPs were genotyped in 712 European Americans with no known CVD and also showed no evidence of association with CAC or carotid IMT [23]. These findings, in conjunction with our own, suggest that ADIPOQ SNPs are unlikely to exert a large effect on risk for CVD development, including in high-risk individuals with T2DM.

The role of SNPs in ADIPOR1 and ADIPOR2 in contributing to CVD risk has received even less attention. One case/control study involving 944 individuals with T2DM [22] reported association between three SNPs in ADIPOR1 (rs7539542, rs10920531, and rs4950894) and angiography-defined CAD. A small case/control study involving 68 individuals reported an association between the SNP rs767870 in ADIPOR2 and CAD [21]. These four SNPs were all genotyped in the DHS sample and were not found to be associated with any of the subclinical CVD phenotypes examined. However, rs767870 did show a modest association with adiponectin concentrations. Overall, findings from this study do not support a clear role for adiponectin receptor variants in mediating risk for subclinical CVD in T2DM.

It should be acknowledged that, in addition to varying sample sizes, the discrepant findings between these earlier investigations and those reported here could be accounted for by the different phenotypes examined. The current study used subclinical measures of CVD and other known risk factors for CVD as the main phenotypes of interest and not angiography-defined CAD or verified CVD events. It could be argued that adiponectin variants may only be important in advanced CVD and hence were not found to be associated with the subclinical disease measures used here. However, the burden of vascular calcified plaque in this sample was considerable, and given the documented relationships between vascular calcification and stenosis [37] and the proposed risk threshold of coronary calcium >400 [38], we would consider advanced CVD likely in a large proportion of the DHS cohort. While we did observe a small number of modest associations between some of the adiponectin signaling variants and the CVD phenotypes, these were generally not seen across the multiple related traits examined and would not survive correction for the multiple comparisons undertaken. Accordingly, these findings suggest that contributions of adiponectin signaling variants to risk for CVD development in T2DM are likely to be small and not clinically meaningful. It is unlikely that the lack of association observed was a false-negative result as the study has >80% power to detect differences as small as 15% for SNPs with a MAF of 0.2.

In contrast to the subclinical CVD measures, several ADIPOQ SNPs were associated with more than one of the related serum lipid traits, or with both a lipid trait and CRP concentrations. Although these associations were modest, they suggest a potential interplay between adiposity (reflected by adiponectin), lipid concentrations, and inflammation; however, this finding needs to be further replicated in additional cohorts. A positive correlation between adiponectin and HDL-C has previously been reported [39], as has an inverse relationship between adiponectin and CRP concentrations [40], supporting a propensity for chronic inflammation in states of obesity. Although adiponectin signaling, via AMP-activated protein kinase, has been suggested to a play a role in lipid and cholesterol metabolism [39], evidence from this study is not sufficient to support, or otherwise, a role for variants in the adiponectin signaling pathway genes in influencing these relationships, particularly in the absence of any evidence of direct effects of these SNPs on adiponectin function.

Compared to a number of previous studies investigating ADIPOQ variants where sample sizes and/or tagging of the gene region were limited, the current study represents the most comprehensive examination of variants in the adiponectin signaling pathway genes that we are aware of to date. As such, findings from the current study could be considered to provide a more realistic insight into the role of adiponectin signaling variants in risk for subclinical CVD. The findings presented herein do not reveal compelling relationships between adiponectin signaling pathway variants and measures of subclinical CVD in T2DM. While there was modest evidence to suggest a possible interplay between adiponectin, plasma lipids, and inflammation, this observation requires further investigation.


The authors thank the other investigators, the staff, and the participants of the DHS study for their valuable contributions.