OUP user menu

Adiponectin, risk of coronary heart disease and correlations with cardiovascular risk markers

Dietrich Rothenbacher, Hermann Brenner, Winfried März, Wolfgang Koenig
DOI: http://dx.doi.org/10.1093/eurheartj/ehi340 1640-1646 First published online: 2 June 2005


Aims We investigated the association of serum adiponectin concentrations with the risk of coronary heart disease (CHD) after careful adjustment for other established risk factors. In addition, we investigated the association between adiponectin levels and established sociodemographic and laboratory risk markers for CHD.

Methods and results Three hundred and twelve patients aged 40–68 with angiographically confirmed stable CHD and 476 age- and gender-matched controls were included in this case–control study. Adiponectin serum concentrations (adiponectin, R&D Systems, Wiesbaden, Germany), markers of inflammation and haemostasis, and an extensive lipid profile were determined. Adiponectin serum concentrations were lower in CHD patients when compared with age- and gender-matched controls, both in men (median 4.95 vs. 5.58 µmol/L, P=0.004) and in women (median 9.64 vs. 11.60 µmol/L, P=0.018). Adiponectin was strongly correlated with lipoproteins and apolipoproteins, in particular HDL-cholesterol (HDL-C), and to a lesser degree with markers of inflammation such as C-reactive protein, IL-6, or markers of coagulation or fibrinolysis. When compared with subjects with adiponectin serum concentrations in the lower quintile, the OR for CHD was 0.52 (95% CI 0.28–0.95) in the upper one after adjustment for covariates (P<0.007 for trend). After additional adjustment for HDL-C the association was strongly reduced, reflecting the close association between adiponectin and HDL-C.

Conclusion Adiponectin serum concentrations may have an important role in the development of CHD. The protective effect of high serum concentration may partly be mediated by effects on the metabolism of lipoproteins, especially on the metabolism of HDL.

  • Adiponectin
  • Coronary heart disease
  • Inflammation
  • Lipids
  • Case–control study

See page 1579 for the editorial comment on this article (doi:10.1093/eurheartj/ehi374)


Atherosclerosis is considered a chronic inflammatory process,1 which is present in the arterial wall2 but is also characterized by a low-grade systemic inflammatory response.3 Recently, it has been suggested that the adipose tissue may play an important role in mediating this chronic inflammatory process and, subsequently, cardiovascular disease risk4 and therefore may not only be considered as a storage site for fat.

Increasing evidence supports the notion that the adipocyte may have an active endocrine function; it produces several cytokines [among them interleukin-6 (Il-6) and tumour necrosis factor-α (TNF-α)] and adiponectin, a 30 kDa adipocyte complement-related protein.5 Adiponectin levels in serum are mainly determined by the size and amount of adipocytes. Highest serum concentrations are found in subjects with only few body fat. Adiponectin has insulin sensitizing and anti-atherogenic effects4,5 and lower serum levels have been reported in patients with coronary heart disease.6,7

A variety of inflammatory and other biochemical markers potentially related to atherogenesis have been identified,8,9 some of which may also originate from the adipose tissue. Currently, we are beginning to understand the potential pathophysiological role of adiponectin in atherosclerosis. A reciprocal association with several pro-inflammatory cytokines such as C-reactive protein,10 an important determinant of vascular disease, has been reported. Furthermore, hypoadiponectinaemia has been closely linked to endothelial dysfunction11 and may be involved in lipid meatobolism.12 Whether adiponectin is an independent risk factor in this signalling network or mainly associated with other inflammatory proteins such as Il-6 or C-reactive protein or other cardiovascular risk factors is yet unclear. The further elucidation of the pathogenetic role of adiponectin may allow to identify subjects at increased risk for coronary heart disease (CHD), and subsequently lead to more focused preventive measures and eventually to a specific treatment (e.g. application of recombinant adiponectin).

We investigated the association of adiponectin serum concentrations with a variety of established sociodemographic and laboratory risk markers for CHD and determined the correlation with various acute phase proteins, inflammation-associated cytokines, intracellular adhesion molecule-1 (ICAM-1), and various lipoproteins. Furthermore, we analysed data of a case–control study in patients with stable CHD in order to investigate the association of serum adiponectin concentrations with the risk of CHD, after careful adjustment for other established risk factors under special consideration of various pathogenic pathways.


Patients and controls

German speaking patients aged 40–68 who underwent coronary angiography at the University of Ulm Medical Center between April 1996 and November 1997 and who showed at least one coronary stenosis of >50% of the luminal diameter were included in the study. Main exclusion criteria for patients were first diagnosis of CHD >2 years ago, unstable angina pectoris, acute myocardial infarction within the past 4 weeks, infection within the past 3 weeks, malignant disease, and anti-coagulant therapy within previous 2 weeks. Out of 400 patients, 312 participated in this study (78%).

The control group consisted of subjects who were occasional blood donors at the local Red Cross Centre serving the University Hospitals of Ulm. None of the controls had a history of definite or suspected CHD, and they did not report infections or surgery within the previous 4 weeks. Response was 479 of 570 in controls (84%) of whom 477 could be included in this analysis (from two controls, no adiponectin measurements were available).

Frequency matching for age and gender was performed and a case–control ratio of 1:1.5 was intended. All subjects underwent standardized interviews conducted by trained interviewers. The primary objective of the study was to assess the effect of various infectious agents on CHD-risk13 and to investigate the role of other emerging risk factors. All subjects gave written informed consent and the study was approved by the Ethics Committee of the University of Ulm.

Laboratory methods

Venous blood was drawn in the morning under standardized conditions and a complete blood cell count was done (Coulter STKS chamber, Coulter Co., Krefeld, Germany). Within 30 min, the remaining blood was centrifuged at 3000g for 10 min, immediately aliquoted and frozen at −70°C until analysis. In cases, blood drawing was done before the angiographic procedure.

Adiponectin serum concentrations were determined by a commercially available ELISA (adiponectin, R&D Systems, Wiesbaden, Germany). In addition, the following markers of inflammation and haemostasis were determined by ELISA: IL-6 and TNF-α (Quantikine, R&D Systems, Wiesbaden, Germany), ICAM-1 (Diaclone, Besancon, France), plasminogen-activator-inhibitor-1 (PAI-1) activity (Immuno, Heidelberg, Germany), and von Willebrand factor (vWF) (Haemochrom, Essen, Germany). In addition, C-reactive protein determinations were done by an immunoradiometric assay (range 0.05–10 mg/L).14 Fibrinogen was measured by immunonephelometry (Dade Behring, Marburg, Germany). Serum amyloid A (SAA) was also determined by immunonephelometry (Dade Behring, Marburg, Germany) and finally, measurement of plasma viscosity (PV) was done in a Harkness–Coulter viscometer (Coulter Electronics, Luton, UK). Plasma levels of lipoprotein-associated phospholipase A2 (Lp-PLA2) were determined with a commercial Lp-PLA2-ELISA kit (PLACTM Test) was supplied by diaDexus Inc. (South San Francisco, CA, USA). Inter-assay coefficients of variation were 10.9% for adiponectin, 7.0% for IL-6, 17.9% for TNF-α, 14.2% for ICAM-1, 12.0% for C-reactive protein, 7.4% for SAA, 5.0% for fibrinogen, 11.0% for PAI-1, 15.8% for vWF, 2.0% for PV, and 9.6% for Lp-PLA2. HDL-cholesterol (HDL-C) concentrations were determined by routine enzymatic methods.

Statistical methods

Sociodemographic and medical characteristics of CHD cases and control subjects are presented in a descriptive fashion. Mean concentrations of adiponectin were calculated in cases and controls and compared using the Wilcoxon rank sum test. In addition, the distribution of adiponectin (in quintiles of controls) was compared among cases and controls and quantified by a Mantel–Haenszel χ2 test after adjustment for age and gender.

Mean concentrations of adiponectin were calculated for age, gender, and body mass index (BMI) after adjustment for case–control status; also for various levels of sociodemographic and other established CHD risk factors after adjustment for age, gender, BMI, and case–control status by a general linear regression method. Partial spearman correlation coefficients were calculated for adiponectin serum concentrations and lipoproteins, apolipoproteins, acute phase proteins, and other suggested laboratory CHD risk markers after adjustment for age, gender, BMI, and case–control status. A two sided P-value of 0.05 or less was considered to be statistically significant, if not indicated otherwise.

Furthermore, we used unconditional logistic regression to assess the association of adiponectin serum concentrations (in quintiles) with CHD, while simultaneously controlling for age and gender and additionally controlling for BMI, duration of school education, cigarette smoking (pack-years), alcohol consumption, history of hypertension, and history of diabetes mellitus (which are all established risk factors for CHD). The linearity assumption was assessed for continuous variables by a goodness-of-fit test (significance level α=0.1). To further elucidate the pathogenic link between adiponectin and CHD risk, we also included the basic model several laboratory markers, which represent defined changes in the pathogenic processes of atherogenesis and showed an association with adiponectin (all variables considered with P<0.1 in bivariate analysis). These were triglyceride-values, HDL-C values, apo-lipoproteins (ApoA1, ApoA2, ApoB, ApoC2, ApoC3, and ApoE), markers of inflammation (C-reactive protein, IL-6 and TNF-α, and leukocyte count), and PAI-1 and D-dimers. All statistical procedures were carried out with the SAS® statistical software package (release 8.2, Cary, NC USA: SAS Institute Inc., 1999).


In total, 788 subjects were enroled in the study (312 patients with stable CHD and 476 age- and gender-matched controls) which had adiponectin concentrations measured. Table 1 gives main characteristics of the study population. CHD patients had more often a lower school education compared with control subjects and established cardiovascular risk factors were more unfavourably distributed in patients when compared with controls.

View this table:
Table 1

Characteristics of patients with CHD and controls

CHD patientsControls
n 312476
Age (µ±SD)57.7±7.455.8±7.2
Gender (male)267 (85.6)357 (75.0)
Family status married268 (85.9)399 (83.8)
School education <10 years216 (69.2)277 (58.2)
Daily alcohol consumption92 (29.5)134 (28.3)
Smoking status
 Current smoker30 (9.6)66 (13.9)
 Ex-smoker207 (66.3)201 (42.2)
 Never smoker75 (24.0)209 (43.9)
Smoked pack-years (µ±SD)20.3±23.710.9±17.1
BMI (kg/m2), (µ±SD)27.3±3.626.3±3.2
History of high blood pressure180 (57.7)98 (20.6)
History of diabetes mellitus42 (13.5)13 (2.7)
History of hyperlipidaemia210 (67.3)100 (21.0)
HDL-C (mg/dL), (µ±SD)42.4±10.351.7±13.2

Values in parentheses indicate n (%).

Table 2 shows the distribution of serum adiponectin concentrations in CHD patients and controls. In general, adiponectin concentrations were lower in CHD patients when compared with age- and gender-matched controls, both in men (median 4.95 vs. 5.58 µmol/L, P=0.004) and in women (median 9.64 vs. 11.60 µmol/L, P=0.018). This was also evident when looking at the adiponectin distribution in quintiles (P<0.001 after adjustment for age and gender).

View this table:
Table 2

Distribution of adiponectin serum concentrations in CHD patients and control

CHD patientsControlsP-valueCHD patientsControlsP-valueCHD patientsControlsP-value
n 26735745119312476
Adiponectin (µmol/L)
Quintiles (µmol/L) (%)
 First (0 to ≤4.12)37.126.315.62.534.020.4
 Second (4.12 to ≤5.67)24.023.317.
 Third (5.67 to ≤7.65)
 Fourth (7.65 to ≤11)14.220.517.818.514.720.0
 Fifth (>11)<0.001b

aWilcoxon rank sum test.

bAfter adjustment for age and gender.

We then calculated mean serum concentrations of adiponectin according to the levels of various cardiovascular risk factors (Table 3). The group of age 50–59 had the highest adiponectin levels. There was a clear association with gender and BMI: females had much higher levels when compared with males (P<0.0001), and subjects in the upper quartile of the BMI distribution had the lowest adiponectin values (P<0.0001).

View this table:
Table 3

Mean serum concentrations of adiponectin according to risk factors of CHD and other sociodemographic variables

nAdiponectin (µmol/L)P-value
Age (years)a
BMIa (kg/m2)
School educationb
 <10 years4967.22
 ≥10 years2927.200.93
Alcohol consumptionb (within past 12 months)
Smoking statusb
 Never smoker2846.96
 Current smoker967.430.43
Physical activityb, c
History of high blood pressureb
History of diabetesb
History of hyperlipidAemiab

aAfter adjustment for case–control status.

bAfter adjustment for age, gender, BMI, and case–control status.

cPhysical activity defined as no leisure time physical activity in summer and winter and no or very moderate physical strain at work.

The level of school education, alcohol consumption, smoking status, and an indicator of physicial activity showed no clear association with adiponectin serum concentrations after adjustment for age, gender, BMI, and case–control status. Subjects with a self reported history of hyperlipidaemia showed only decreased adiponectin serum concentrations after adjustment (P=0.008).

Table 4 shows the correlation between adiponectin serum concentrations and lipid variables, markers of coagulation, fibrinolysis and inflammation, and Lp-PLA2. There were statistically significant positive correlations of adiponectin concentrations seen with HDL, ApoA1, and ApoA2. Furthermore, negative correlations were seen with PAI-1, C-reactive protein, TNF-α, and leukocyte count. There were no statistically significant correlations with fibrinogen, d-dimers, vWF, ICAM-1, IL-6, PV, albumin, and Lp-PLA2.

View this table:
Table 4

Levels of lipoproteins, markers of coagulation, fibrinolysis and inflammation, and their correlation with adiponectin

Meana(25th to 75th percentile)Correlation with adiponectin
HDL (mmol/L)48.0(39.0–55.0)0.31<0.0001
ApoA1 (mg/dL)138.8(122.0–154.0)0.28<0.0001
ApoA2 (mg/dL)47.7(42.0–52.5)0.14<0.0001
ApoB (mg/dL)104.6(87.7–120.0)−0.030.35
ApoC2 (mg/dL)4.21(2.88–4.96)−0.070.04
ApoC3 (mg/dL)15.1(12.1–17.1)−0.060.12
ApoE (mg/dL)9.17(7.30–10.4)−0.040.32
Lp(a) (mg/dL)d10.3(4.33–32.0)0.040.21
PAI-1 activity (U/mL)d8.78(5.14–16.0)−0.17<0.0001
Fibrinogen (g/L)2.67(2.34–2.92)−0.040.21
D-dimer (µg/L)d4.00(0.50–21.5)0.050.15
vWF (% activity)140.9(105.0–168.0)0.030.42
ICAM-1 [ng/mL]500.5(396.0–574.0)−0.040.29
C-reactive protein (mg/L)d1.31(0.58–2.98)−0.120.0009
IL-6 (pg/mL)d1.82(1.11–2.62)−0.050.15
SAA (mg/L)d2.95(1.80–4.40)0.020.60
TNF-α (pg/mL)d2.08(1.50–2.97)−0.110.002
Leukocytes (103/µL)6.21(5.00–7.20)−0.120.0005
PA (mPa s)1.20(1.16–1.24)−0.020.66
Albumin (g/L)39.6(37.3–42.3)−0.050.17
Lp-PLA2 (ng/mL)277.9(197.9–336.3)−0.030.40

aArithmetric, if skewed (and therefore log-transformed) geometric.

bSpearman rank correlation coefficient, adjusted for age, gender, BMI, and case–control status.

cAn α-value <0.0024 indicates statistical significance (Bonferroni correction).


Table 5 shows a clear inverse association of the OR for CHD associated with adiponectin serum concentrations in the multivariable analysis. Compared to subjects with adiponectin serum concentrations in the lowest quintile, the OR decreased with increasing quintiles from 0.69 (95% CI 0.45–1.05) to 0.49 (95% CI 0.32–0.76), to 0.44 (95% CI 0.23–0.69), and finally to 0.42 (95% CI 0.25–0.72) after adjustment for age and gender (P<0.0001 for trend). ORs were slightly attenuated after additional adjustment for BMI, duration of school education, smoked pack-years (cigarettes), alcohol consumption, history of hypertension, and history of diabetes. However, the initial pattern remained and the OR for CHD in the top quintile was finally 0.52 (95% CI 0.28–0.95, P=0.007 for trend). If adiponectin was included as a continuous variable in the latter model, an increase of 1 U (µmol/L) was associated with an OR for CHD of 0.93 (95% CI 0.89–0.97) in the partially adjusted model and with an OR for CHD of 0.95 (95% CI 0.91–0.99) in the multivariable-adjusted model.

View this table:
Table 5

ORs for the presence of primary CHD associated with adiponectin serum concentrations

Adiponectin quintiles (µmol/L)P-value for trend
First (0 to ≤4.12)Second (4.12 to ≤5.67)Third (5.67 to ≤7.65)Fourth (7.65 to ≤11)Fifth (>11)
Partly adjusteda OR (95% CI)1 (reference)0.69 (0.45–1.05)0.49 (0.32–0.76)0.44 (0.28–0.69)0.42 (0.25–0.72)<0.0001
Multivariable-adjustedb OR (95% CI)1 (reference)0.72 (0.45–1.17)0.56 (0.34–0.92)0.55 (0.33–0.93)0.52 (0.28–0.95)0.007

aAdjusted for age and gender.

bAdjusted for age and gender, BMI, duration of school education, smoked pack-years (cigarettes), alcohol consumption, history of hypertension, and history of diabetes.

Table 6 then shows the relationship between adiponectin serum concentrations and risk of CHD after additional adjustment for other established cardiovascular risk markers (lipid variables, markers of coagulation, fibrinolysis, and inflammation). Most notably, after additional inclusion of HDL-C to the basic model, the associations of adiponectin serum concentrations with CHD was strongly reduced and no more statistically significant (OR in top quintile 1.02, 95 % CI 0.53–1.98; P=0.5 for trend), indicating that HDL-C may represent an important intermediate step in the pathway between adiponectin serum concentrations and risk of CHD. A similar attenuation, although not so strong, was evident if ApoA1 and ApoA2 were included. The inclusion of ApoC2, ApoC3, and ApoE or ApoB into the basic model did not change the OR in a relevant manner.

View this table:
Table 6

ORs for primary CHD associated with adiponectin serum concentrations after additional adjustment for various lipoproteins, markers of coagulation, fibrinolysis, and inflammation

Basic modelaAdiponectin quintiles (µmol/L)P-value for trend
First (0 to ≤4.12)Second (4.12 to ≤5.67)Third (5.67 to ≤7.65)Fourth (7.65 to ≤11)Fifth (>11)
HDL cholesterol OR (95% CI)1 (reference)0.90 (0.55–1.48)0.69 (0.41–1.16)0.82 (0.47–1.42)1.02 (0.53–1.97)0.6
ApoA1, ApoA2 OR (95% CI)1 (reference)0.92 (0.56–1.50)0.67 (0.41–1.15)0.75 (0.43–1.29)0.90 (0.47–1.73)0.3
ApoB OR (95% CI)1 (reference)0.72 (0.45–1.17)0.56 (0.34–0.92)0.56 (0.33–0.94)0.52 (0.28–0.96)0.07
ApoC2, ApoC3, ApoE OR (95% CI)1 (reference)0.72 (0.45–1.17)0.56 (0.34–0.92)0.56 (0.33–0.94)0.53 (0.29–0.98)0.07
CRP, IL-6, TNF-α, and leukocytes OR (95% CI)1 (reference)0.95 (0.56–1.61)0.64 (0.37–1.11)0.80 (0.45–1.41)0.70 (0.35–1.40)0.07
PAI-1, D-dimer OR (95% CI)1 (reference)0.78 (0.48–1.28)0.58 (0.35–0.96)0.63 (0.37–1.07)0.64 (0.34–1.24)<0.05

aBasic model adjusted for age, gender, BMI, duration of school education, smoked pack-years (cigarettes), alcohol consumption, history of hypertension, and history of diabetes.

The inclusion of C-reactive protein, IL-6, TNF-α, and leukocyte count into the basic model resulted also in an increase in the ORs. However, the OR in the top quintile of the adiponectin distribution was still 0.70 (95 % CI 0.35–1.40) and the P-value for trend was of borderline significance (P=0.07). If PAI-1 and d-dimers were included in the basic model, the ORs associated with higher adiponectin levels were slightly attenuated; however, the trend was still statistically significant (P>0.05).


In this large case–control study including patients with angiographically defined stable CHD, we observed a strong inverse association between serum levels of adiponectin and risk of CHD. The association showed a clear dose–response relationship and persisted after adjustment for conventional CHD risk factors including a history of diabetes and BMI. Adiponectin serum concentrations were strongly correlated with various lipoproteins, among them the strongest correlation was seen with HDL-C, and to a lesser degree with markers of inflammation such as C-reactive protein, IL-6 and markers of coagulation or fibrinolysis, but no correlation was seen with ICAM-1 or Lp-PLA2. Most notably, the relationship between adiponectin and CHD risk essentially vanished after adjustment for HDL. It was reduced to a lesser degree after adjustment for other markers such as C-reactive protein and Il-6, indicating that there is a very close association of adiponectin and HDL metabolism.

Our data are in line with other reports describing an associaton between low levels of adiponectin serum concentration and risk of CHD. Kumada et al.6 estimated a two-fold risk for ischaemic heart disease associated with low adiponectin levels. Furthermore, a reduced risk of subsequent acute myocardial infarction associated with higher levels of adiponectin in serum at baseline was described in a nested case–control study,7 notably, this association was also reduced after adjustment for covariates, but persisted after additional adjustment for lipids, glycemic status, and C-reactive protein.

Various physiological mechanisms of adiponectin have been identified in vitro which could play a role in atherogenesis.6 Systemically measurable markers of low-grade inflammation are important predictors of CHD risk4 and are increased in patients with stable CHD. As the adipose tissue itself at least, in part, generates these markers (TNF-α, IL-6 or C-reactive protein as well as of PAI-1), a direct link between adiponectin and these pro-inflammatory cytokines seems conceivable.4 In addition, TNF-α and IL-6 are also negatively related to insulin sensitivity,15,16 whereas adiponectin has a positive effect on it. In addition, TNF-α has also been suggested as a strong inhibitor of adiponectin promoter activity17 and may explain partly the inverse association between the amount of accumulated visceral fat, increased TNF-α secretion, and decreased adiponectin levels.

Although the physiological role of adiponectin has not yet been fully elucidated, it may well be involved in the regulation of many of the inflammatory processes or in the lipid metabolisms, which are contributing to atherosclerosis. However, we found only moderate correlations between adiponectin serum concentrations and these inflammatory factors, of which correlations with TNF-α and C-reactive protein were the strongest.

Among the many risk factors involved in atherogenesis disorders of lipoprotein metabolism are considered to play the most important role.18 In particular, high LDL and low HDL concentrations are main determinants of CHD risk. We indeed found the strongest correlations of adiponectin serum concentrations with HDL levels. The correlation to ApoA1, which is essential to HDL-formation was of similar strength. Similar findings were also reported by others.19 Most notably, plasma levels of adiponectin and its relationship to serum HDL and triglyceride concentrations were independent from BMI12 and this supports the suggestion that adiponectin levels play a role in the regulation of lipid metabolism. Furthermore, in our study, the association between adiponectin and CHD was abrogated by including HDL in the model adjusted for conventional risk factors. The most important anti-atherogenic function of HDL considered is its participation in the reverse cholesterol transport, which delivers excess cholesterol from systemic vasculature to the liver for disposal as bile salt.20 In addition, HDL also has antioxidant and anti-inflammatory properties.21

The biochemical mechanisms linking adiponectin and HDL metabolism have not been clarified so far. Adiponectin also seems to have a key role in the metabolic syndrome,22 an accumulation of multiple risk factors. It may therefore represent the link between obesity (or even more important visceral fat accumulation), insulin resistance, and diabetes. Low concentrations of adiponectin may lead to insulin resistance.23 Insulin resistance, in turn, may lower the concentration of HDL through different mechanisms, directly and indirectly. First, insulin may directly stimulate the transcriptional activity of ApoA1, the major apolipoprotein of HDL.24 Secondly, insulin may decrease the production of VLDL25 and enhances the expression of lipoprotein lipase.26 Insulin resistance thus may raise the concentration of triglyceride-rich lipoproteins in the circulation, which may alter the formation and remodelling of HDL particles. Altogether, this raises the possibility that low adiponectin concentrations may cause low HDL-C and that the pro-atherogenic effects of low adiponectin may be mediated by its effects on HDL metabolism. As we found a closer association of adiponectin with HDL than with inflammatory markers, the lipoprotein effects of adiponectin may be more important than the anti-inflammatory links with TNF-α, IL-6, or C-reactive protein,27 or the suggested effects on ICAM-1,28 an adhesion molecule that regulates the attachment and transmigration of leukocytes across the vascular endothelium.

The present study has several limitations which should be addressed. CHD was defined invasively by coronary angiography in cases, but for ethical reasons, no coronary angiogram could be obtained in controls. Although we excluded controls with a history or characteristic symptoms of CHD, the presence of asymptomatic CHD cannot be definitely ruled out; however, the prevalence of asymptomatic CHD cases seems to be low in a middle-aged population.29 Furthermore, the choice of blood donors as controls can be considered as suboptimal, as they might be healthier than the target population the cases were drawn from. We tried to minimize this potential bias by carrying out multivariable adjustments for a variety of covariates. As always in case–control studies in which exposure and outcome are collected at one point in time, it is difficult to assess the time-sequence of the described associations and therefore, it is highly desirable to replicate our results in prospective studies. The limited sample size may also be a reason why the relationship of adiponectin with markers of inflammation did not reach statistical significance.

Despite these limitations, the current study provides evidence that adiponectin serum concentrations may have an important role in the development of CHD, and it raises the possibility that the protective effect of high serum concentrations may partly be mediated by effects on the lipid metabolism, especially on the HDL levels.


View Abstract