Relation of Serum Osteocalcin Level to Risk of Coronary Heart Disease in Chinese Adults

Relation of Serum Osteocalcin Level to Risk of Coronary Heart Disease in Chinese Adults Yifei Zhang, PhDa,†, Lu Qi, PhDc,†, Weiqiong Gu, PhDa, Qun Yan, MDa, Meng Dai, MDa, Juan Shi, MDa, Ying Zhai, MDa, Ying Chen, MDa, Jianmin Liu, MD, PhDa, Weiqing Wang, MD, PhDa, Guang Ning, MD, PhDa,b, and Jie Hong, PhDa,* Osteocalcin, a bone-derived polypeptide, was recently found to have hormonal function associated with metabolic disorders and atherosclerosis. Few studies have examined the association between circulating osteocalcin and coronary heart disease (CHD) risk. The aim of the present study was to investigate whether serum osteocalcin concentration was associated with CHD risk and metabolic profiles in Chinese adults. A total of 461 subjects (243 with CHD and 218 without CHD) who underwent coronary angiography were included. Serum osteocalcin, glucose, lipid profiles, and other biochemical markers were measured. Severity of coronary atherosclerosis was estimated by number of diseased vessels. Results showed that serum osteocalcin levels were significantly lower in the CHD group (12.2 ng/ml, 9.5 to 15.1) than in the non-CHD group (13.6 ng/ml, 10.7 to 18.0, p ⴝ 0.001) and were significantly decreased with the increasing of number of diseased vessels (p ⴝ 0.005). Serum osteocalcin concentration was inversely correlated with fasting and post load 2 hour plasma glucose and hemoglobin A1c (p ⴝ 0.044, 0.043, and 0.011, respectively), adjusting for CHD status. Odds ratios (95% confidence intervals) of CHD across increasing quartiles of serum osteocalcin were 0.68 (0.42 to 1.12), 0.59 (0.36 to 0.98), and 0.40 (0.23 to 0.69). The test for trend was significant (p ⴝ 0.0007). Adjusting for age, body mass index, and other conventional risk factors for CHD did not appreciably change the results. Spline regression analyses indicated a linear relation between serum osteocalcin level and CHD risk. In conclusion, our data indicate that serum osteocalcin level was associated with decreased risk of CHD and protective metabolic changes in Chinese adults. © 2010 Elsevier Inc. All rights reserved. (Am J Cardiol 2010;106:1461–1465) Bone and cardiovascular diseases have long been suspected to be linked to each other.1– 8 Osteocalcin is a bonederived 49-residue polypeptide.9 –11 In an animal study, osteocalcin knockout mice display an increased accumula-

a State Key Laboratory of Medical Genomics, Shanghai Clinical Center for Endocrine and Metabolic Diseases, Shanghai Institute of Endocrinology and Metabolism, Endocrine and Metabolic E-Institutes of Shanghai Universities (EISU), and Key Laboratory for Endocrinology and Metabolism of Chinese Health Ministry, Rui-jin Hospital, Shanghai Jiao-Tong University School of Medicine, and bLaboratory of Endocrinology and Metabolism, Institute of Health Sciences, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences/Shanghai Jiao-Tong University School of Medicine, Shanghai, China; and cDepartment of Nutrition, Harvard School of Public Health, Boston, Massachusetts. Manuscript received May 2, 2010; revised manuscript received and accepted July 12, 2010. This study was supported by Grant 2006 AA 02A409 from the 863 Project, Beijing/China, Grants 30971077 and 30890043 from the National Natural Science Foundation of China, Beijing/China, Grant Shdc12007309 from the Shanghai Shenkang Hospital Development Center, Shanghai/China, Grant 2008ZX09312/019 from the National Key New Drug Creation and Manufacturing Program, Beijing/China, and Grant 2008BAI52B03 from the National Key Technologies Research and Development Program, Beijing/China. *Corresponding author: Tel: 86-21-6437-0045, ext 665345; fax: 8621-6437-3514. E-mail address: [email protected] (J. Hong). †

Dr. Zhang and Dr. Qi contributed equally to this work.

0002-9149/10/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2010.07.013

tion of visceral fat associated with glucose intolerance and insulin resistance early in life.12 In humans, serum osteocalcin was inversely associated with fasting glucose and insulin, homeostasis model assessment of insulin resistance (HOMA-IR), body mass index (BMI), and fat mass.13–16 A recent study in type 2 diabetic patients found that osteocalcin was negatively correlated with parameters of peripheral atherosclerosis, intima–media thickness, and ankle– brachial pulse-wave velocity.17 These data suggest that osteocalcin might be involved in the development of metabolic and cardiovascular diseases. However, direct evidence on the relation between serum osteocalcin level and risk of coronary heart disease (CHD) in humans is sparse. In this study we examined the association between serum osteocalcin level and CHD in Chinese adults. We also assessed correlations between serum osteocalcin and metabolic risk factors for CHD. Methods The present study included 461 consecutive patients (299 men and 162 women, age ranges 39 to 84 and 41 to 85 years, means 61.7 and 63.3, respectively) who were referred to the department of cardiology in Ruijin Hospital (Shanghai, China) because of symptoms in the chest such as chest pain, chest heaviness, periodic discomfort, and palpitations from January 2005 to December 2007. In all subjects, coronary angiography for diagnosis of CHD was performed. Those with medical illnesses such as acute infection, chronic hepatic and renal www.ajconline.org

1462

The American Journal of Cardiology (www.ajconline.org)

Table 1 Characteristics of participants according to coronary heart disease status Variable

Men/women Age (years) Current smoker Alcohol use Type 2 diabetes mellitus Hypertension Body mass index (kg/m2) Waist (cm) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Fasting plasma glucose (mmol/L) Plasma glucose 2 hours after load (mmol/L) Hemoglobin A1c (%) Triglyceride (mmol/L)/(mg/dl) Total cholesterol (mmol/L)/(mg/dl) High-density lipoprotein (mmol/L)/(mg/dl) Low-density lipoprotein (mmol/L)/(mg/dl) Fasting serum insulin (␮IU/ml) Serum insulin 2 hours after load (␮IU/ml) Homeostasis model assessment for insulin resistance (␮IU ⫻ mol/L2) High-sensitivity C-reactive protein (mg/L) Number of diseased vessels

CHD

p Value

Yes (n ⫽ 243)

No (n ⫽ 218)

189 (78%)/54 (22%) 62.7 ⫾ 8.6 120 (49.4%) 48 (19.8%) 84 (34.6%) 172 (70.8%) 25.1 ⫾ 3.4 90.3 ⫾ 9.0 133 ⫾ 19 79 ⫾ 10 5.78 ⫾ 1.82 9.12 ⫾ 3.85 6.3 (5.8–6.9) 1.92 ⫾ 1.06/170 ⫾ 94 4.53 ⫾ 1.08/175 ⫾ 42 1.17 ⫾ 0.34/45 ⫾ 13 2.69 ⫾ 0.96/104 ⫾ 37 9.2 (5.0–14.4) 56 (28–109) 2.3 (1.2–3.8)

110 (50%)/108 (50%) 61.8 ⫾ 9.2 52 (23.9%) 26 (11.9%) 57 (26.1%) 143 (65.6%) 25.5 ⫾ 3.1 90.7 ⫾ 8.1 130 ⫾ 18 79 ⫾ 11 5.55 ⫾ 1.53 8.71 ⫾ 3.59 6.0 (5.7–6.7) 1.96 ⫾ 1.63/173 ⫾ 144 4.60 ⫾ 1.03/178 ⫾ 40 1.21 ⫾ 0.32/47 ⫾ 12 2.69 ⫾ 0.80/104 ⫾ 31 8.2 (5.4–12.2) 46 (24–94) 1.9 (1.2–2.7)

⬍0.001 0.300 ⬍0.001 0.013 0.050 0.329 0.181 0.652 0.100 0.468 0.194 0.256 0.007 0.721 0.501 0.259 0.998 0.379 0.034 0.207

3.0 (1.2–7.3) 1.7 ⫾ 0.9

2.0 (0.8–4.6) 0

0.001 ⬍0.001

Data are presented as mean ⫾ SD for data normally distributed, number of patients (percentages), or median (interquartile range) for data not normally distributed. Statistical significances were determined using Student’s t test (for data normally distributed) or Mann-Whitney test (for data not normally distributed) and chi-square test (for data that were categorical variables).

Figure 1. Comparisons of logarithmic transformation (lg) of serum osteocalcin (A) between the CHD and non-CHD groups and (B) according to number of stenotic coronary arteries.

dysfunctions (including serum alanine aminotransferase ⬎120 IU/L, aspartate aminotransaminase ⬎80 IU/L, and serum creatinine ⬎2.0 mg/dl) or nutritional derangements, malignancies, and other severe medical illnesses were excluded (n ⫽ 63). All patients were free of drugs known to influence bone and calcium metabolism. All patients were Chinese living in the Shanghai region and gave informed consent. This study was approved by the institutional review board of Ruijin Hospital and complied with the Declaration of Helsinki.

Coronary angiography was performed in multiple projections with the Judkins technique. Coronary stenosis with lumen narrowing ⬎50% was considered significant. CHD was diagnosed as the presence of ⱖ1 vessel with significant stenosis in a given subject. Extent of coronary atherosclerosis was defined as number of diseased vessels with significant stenosis in a given subject. All patients were examined in the morning after an overnight fast of 10 to 12 hours. Date of birth, smoking history,

Preventive Cardiology/Osteocalcin and Coronary Atherosclerosis Table 2 Correlations between serum osteocalcin and risk factors for coronary heart disease Variable

Age (years) Body mass index (kg/m2) Waist (cm) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Fasting plasma glucose (mmol/L) Plasma glucose 2 hours after load (mmol/L) Hemoglobin A1c (%) Triglyceride (mmol/L)/(mg/dl) Total cholesterol (mmol/L)/ (mg/dl) High-density lipoprotein (mmol/L)/(mg/dl) Low-density lipoprotein (mmol/L)/(mg/dl) Fasting serum insulin (␮IU/ml) Post load 2 hour serum insulin (␮IU/ml) Homeostasis model assessment for insulin resistance (␮IU ⫻ mol/L2) High-sensitivity C-reactive protein (mg/L)

Crude

Control for CHD

r

p Value

r

p Value

⫺0.123 0.025 ⫺0.012 0.003 0.006 ⫺0.059 ⫺0.068

0.008 0.586 0.817 0.948 0.900 0.213 0.161

⫺0.175 0.126 0.028 0.054 0.040 ⫺0.148 ⫺0.149

0.017 0.088 0.701 0.464 0.588 0.044 0.043

⫺0.090 ⫺0.026 0.008

0.078 0.587 0.862

⫺0.186 ⫺0.028 0.068

0.011 0.707 0.356

0.031

0.523

0.022

0.771

⫺0.010

0.843

0.032

0.661

0.020 ⫺0.020

0.713 0.712

0.029 0.041

0.698 0.578

⫺0.002

0.977

⫺0.077

0.297

0.012

0.808

⫺0.135

0.067

Correlations were determined using Spearman correlation coefficients.

alcohol consumption, and medical history were recorded. Height and weight (light clothes and without shoes), waist and hip circumferences, and blood pressure while sitting (measured on a patient’s nondominant arm supported at heart level) were determined by an experienced physician. All serum and plasma samples were collected in the morning after an overnight fasting of 10 to 12 hours and without smoking. Samples were frozen immediately and stored at ⫺80°C until assayed. Biochemical measurements of serum lipids and insulin were performed in a central laboratory (Shanghai Institute of Endocrinology and Metabolism, Shanghai, China). All patients were required to refrain from alcohol, cigarettes, and heavy physical exercise for ⱖ1 week before obtaining blood samples for biochemical measurement and performing the 75-g oral glucose tolerance test. Glucose was measured immediately using an enzymatic method (CX-7 Biochemical Autoanalyzer, Beckman Coulter, Inc, Brea, California). Serum insulin was measured using a double-antibody radioimmunoassay (DSL, Webster, Texas). Serum total cholesterol and triglycerides were measured by enzymatic methods (Beckman Coulter, Inc., Fullerton, California). High-density lipoprotein cholesterol and low-density lipoprotein cholesterol were determined by immunoinhibition methods (High-Density Lipoprotein Cholesterol and Low-Density Lipoprotein Cholesterol Direct, Wake Pure Chemical Industries, Ltd. GmbH, Neuss, Germany). Serum high-sensitivity C-reactive protein was measured using an enzyme-linked immunosorbent assay kit (BioCheck, Inc., Foster City, California). Serum osteocalcin levels were measured by electrochemiluminescence (Elecsys N-MID Osteocalcin Calset; Roche Diagnos-

1463

tics, Indianapolis, Indiana; interassay coefficient of variation ⬍5%). HOMA-IR was calculated according to the following formula: (Fasting Serum Insulin [international microunits per milliliter] ⫻ Fasting Plasma Glucose [millimoles per liter])/ 22.5. Statistical analysis was performed using the SPSS 13.0 for Windows (SPSS, Inc., Chicago, Illinois). Logarithmic transformation was performed for serum insulin, HOMAIR, high-sensitivity C-reactive protein, hemoglobin A1c (HbA1c), and serum osteocalcin to achieve normal distribution. Student’s t test (for data that were normally distributed) or Mann-Whitney test (not normally distributed) and chi-square test (for data that were categorical variables) were used to compare CHD and non-CHD samples. Oneway analysis of variance was used to compare serum osteocalcin levels in groups according to number of diseased vessels (n ⫽ 0, 1, 2, ⱖ3). Correlations between serum osteocalcin and risk factors for CHD were determined using Spearman correlation coefficients. Odds ratios of CHD were calculated using a logistic regression models, adjusting for covariates including age, gender, BMI, smoking, alcohol, and family history of CHD and biochemical risk factors low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, and diabetes. Serum osteocalcin level was analyzed in quartiles with the lowest quartile (quartile 1) as the reference. We also used restricted cubic spline regressions18 to model associations between serum osteocalcin concentration as a continuous variable and risk of CHD. Two-sided p values ⬍0.05 were considered statistically significant. Results Clinical characteristics of the CHD and non-CHD groups are listed in Table 1. Of participants included in this study, 243 (53%) were found to have CHD and 218 (47%) were found not to have CHD. Smoking history, alcohol use, type 2 diabetes percentage, HbA1c, post load 2h serum insulin, and high-sensitivity C-reactive protein levels were significantly higher in the CHD group than in the non-CHD group (different at p ⱕ0.05). Serum osteocalcin levels in the CHD group were significantly lower (12.2 ng/ml, 9.5 to 15.1) than in the non-CHD group (13.6 ng/ml, 10.7 to 18.0, p ⫽ 0.001; Figure 1). When all subjects were further divided into 4 groups according to number of diseased vessels (n ⫽ 0, 1, 2, ⱖ3), serum osteocalcin levels were 13.6 ng/ml (10.8 to 17.8), 12.6 ng/ml (9.5 to 15.3), 11.3 ng/ml (9.2 to 14.9), and 12.2 ng/ml (9.5 to 14.8) in groups with 0 and 1 diseased vessel and 2 and ⱖ3 diseased vessels, respectively (p for trend ⫽ 0.005). Level of osteocalcin in the group with 0 diseased vessel was significantly higher than those in groups with 1 diseased vessel and 2 and ⱖ3 diseased vessels (p ⫽ 0.039, 0.005, and 0.010, respectively). Serum osteocalcin levels were not significantly different among groups with ⱖ1 diseased vessel (Figure 1). We analyzed correlations between serum osteocalcin levels and risk factors for CHD, including metabolic measurements for glucose, insulin sensitivity, blood pressures, and inflammation (Table 2). Significant negative correlations were observed between serum osteocalcin levels and plasma glucose levels after fasting and after a 2-hour load and HbA1c (different at p ⬍0.05), adjusting for CHD status. Serum osteocalcin levels were not

1464

The American Journal of Cardiology (www.ajconline.org)

Table 3 Associations between serum osteocalcin (in quartiles) and risk of coronary heart disease OR (95% CI)

Model 1: crude, no adjustment p Values Model 2: adjusting for age, gender, body mass index p Values Model 3: adjusting for age, gender, body mass index, smoking, alcohol, and family history of coronary heart disease p Values Model 4: adjusting for age, gender, body mass index, smoking, alcohol, and family history of coronary heart disease plus biochemical risk factors lowdensity lipoprotein, high-density lipoprotein, triglyceride, and diabetes p Values

Q1 (n ⫽ 151)

Q2 (n ⫽ 117)

Q3 (n ⫽ 105)

Q4 (n ⫽ 88)

1.0

1.0

0.68 (0.42–1.12) 0.65 0.777 (0.463–1.302) 0.338 0.79 (0.46–1.34)

0.59 (0.36–0.98) 0.7 0.603 (0.356–1.023) 0.061 0.60 (0.35–1.04)

0.40 (0.23–0.69) 0.01 0.476 (0.270–0.838) 0.01 0.54 (0.30–0.81)

1.0

0.38 0.738 (0.361–1.506)

0.07 0.495 (0.241–1.018)

0.03 0.497 (0.215–1.006)

0.403

0.056

0.052

1.0

p Value for Trend

0.0007 0.006 0.018

0.02

Odds ratios of coronary heart disease were calculated using logistic regression models. Serum osteocalcin level was analyzed in quartiles with the lowest quartile (quartile 1) as the reference. CI ⫽ confidence interval; OR ⫽ odds ratio; Q1 ⫽ quartile 1; Q2 ⫽ quartile 2; Q3 ⫽ quartile 3; Q4 ⫽ quartile 4.

Figure 2. Restricted cubic spline regressions modeling associations between serum osteocalcin and risk of CHD. CI ⫽ confidence interval.

significantly correlated with insulin levels after fasting and after a 2-hour load, lipid concentrations, and other risk factors. Table 3 presents associations between serum osteocalcin level (in quartiles) and CHD risk. In the crude analysis (model 1), the odds ratios (95% confidence intervals) of CHD across increasing quartiles of serum osteocalcin were 0.68 (0.42 to 1.12), 0.59 (0.36 to 0.98), and 0.40 (0.23 to 0.69). The test for trend was significant (p ⫽ 0.0007). We further performed a series of analyses in model 2 adjusting for age, gender, and BMI; model 3 adjusting for age, gender, BMI, smoking, alcohol, and family history of CHD; and model 4 adjusting for the previous factors and for biochemical risk factors low-density lipoprotein cholesterol,

high-density lipoprotein cholesterol, triglycerides, and diabetes. Adjustment for the covariates did not appreciably change the associations (adjusted p values for trend were 0.006, 0.018, and 0.02, respectively, in models 2, 3, and 4). We further used restricted cubic spline regressions to model associations between CHD risk and serum osteocalcin continuously (Figure 2). Regression splines showed a linear relation between serum osteocalcin and decreased risk of CHD. Discussion The present study for the first time evaluated associations of serum osteocalcin level with direct parameters of atheroscle-

Preventive Cardiology/Osteocalcin and Coronary Atherosclerosis

rosis and metabolic phenotype in Chinese adult patients who underwent coronary angiography. We found a significant association between serum osteocalcin level and a decreased risk of CHD. We also found that number of stenotic coronary arteries was associated with decreasing serum osteocalcin levels. Our present findings are in accordance with a recent study by Kanazawa et al17 who found that serum osteocalcin level was negatively associated with atherosclerotic parameters intima–media thickness and ankle– brachial pulsewave velocity, independent of other cardiovascular risk factors in diabetic men. Several recent studies have demonstrated that human vascular calcification might have similar regulatory mechanisms as bone modeling and remodeling.19 –24 In 1 study,20 osteocalcin was detected in large calcified areas as large calcium deposits and smaller calcification foci in human carotid arteries from endarterectomy samples, in addition to osteopontin and osteonectin. Dhore et al22 found that several bone matrix regulatory proteins including osteocalcin, matrix Gla protein, bone sialoprotein, bone morphogenetic protein, osteopontin, and osteonectin were expressed in atherosclerotic arteries. Osteocalcin was present at all stages of human atherosclerosis and was a continuous inhibitor of calcification in the atherosclerotic vessel wall. These observations suggested a tight regulation of the expression of osteocalcin during human atherogenesis and were in line with a previous population-based study17 and our present study. Potential mechanisms underlying the association between serum osteocalcin and CHD risk remain unclear. In our study, negative correlations were found between serum osteocalcin and plasma glucose levels after fasting and after a 2-hour load and HbA1c independent of CHD status. Associations of serum osteocalcin with these glucose metabolic phenotypes agreed with other studies.13–17 Moreover, these findings could also be supported by an animal study of Lee et al,12 which indicated that osteocalcin regulated glucose homeostasis by favoring ␤-cell proliferation and insulin secretion and increasing insulin sensitivity at least partly through upregulating adiponectin. However, in the present study, adjustment for metabolic markers did not alter associations between serum osteocalcin level and CHD. These observations suggest osteocalcin may affect CHD risk through other independent pathways. We did not observe significant correlations between serum osteocalcin and other markers such as serum insulin and HOMA-IR, although these markers have been related to osteocalcin in some studies.13–16 Reasons for the discrepancy between studies may be partly due to the ethnic diversity and difference in study designs. Several limitations of our study need attention. First, the study subjects might be heterogenous in treatments for diabetes and cardiovascular disease, and we could not eliminate all influences of drugs in the present work. However, adjustment of medication use did not change the major findings. Second, the present study was cross sectional in nature and the sample was relatively small. Therefore, we cannot claim causality and would be cautious in interpretations of the findings. 1. Lieben L, Callewaert F, Bouillon R. Bone and metabolism: a complex crosstalk. Horm Res 2009;71(suppl 1):134 –138. 2. Strotmeyer ES, Cauley JA. Diabetes mellitus, bone mineral density, and fracture risk. Curr Opin Endocrinol Diabetes Obes 2007;14:429 – 435.

1465

3. Ravn P, Cizza G, Bjarnason NH, Thompson D, Daley M, Wasnich RD, McClung M, Hosking D, Yates AJ, Christiansen C. Low body mass index is an important risk factor for low bone mass and increased bone loss in early postmenopausal women. Early Postmenopausal Intervention Cohort (EPIC) study group. J Bone Miner Res 1999;14:1622–1627. 4. Zhao LJ, Liu YJ, Liu PY, Hamilton J, Recker RR, Deng HW. Relationship of obesity with osteoporosis. J Clin Endocrinol Metab 2007; 92:1640 –1646. 5. McFarlane SI, Muniyappa R, Shin JJ, Bathtiyar G, Sowers JR. Osteoporosis and cardiovascular disease: brittle bones and boned arteries, is there a link? Endocrine 2004;23:1–10. 6. von der Recke P, Hansen MA, Hassager C. The association between low bone mass at the menopause and cardiovascular mortality. Am J Med 1999;106:273–278. 7. Tanko LB, Christiansen C, Cox DA, Geiger MJ, McNabb MA, Cummings SR. Relationship between osteoporosis and cardiovascular disease in postmenopausal women. J Bone Miner Res 2005;20:1912–1920. 8. Srivatsa SS, Harrity PJ, Maercklein PB, Kleppe L, Veinot J, Edwards WD, Johnson CM, Fitzpatick LA. Increased cellular expression of matrix proteins that regulate mineralization is associated with calcification of native human and porcine xenograft bioprosthetic heart valves. J Clin Invest 1997;99:996 –1009. 9. Hauschka PV, Lian JB, Cole DE, Gundberg CM. Osteocalcin and matrix protein: vitamin K-dependent proteins in bone. Physiol Rev 1989;69:990 –1047. 10. Lee AJ, Hodges S, Eastell R. Measurement of osteocalcin. Ann Clin Biochem 2000;37:432– 446. 11. Gerdhem P, Ivaska KK, Alatalo SL, Halleen JM, Hellman J, Isaksson A, Pettersson K, Vaananen HK, Akesson K, Obrant KJ. Biochemical markers of bone metabolism and prediction of fracture in elderly women. J Bone Miner Res 2004;19:386 –393. 12. Lee NK, Sowa H, Hinoi E, Ferron M, Ahn JD, Confavreux C, Dacquin R, Mee PJ, McKee MD, Jung DY, Zhang Z, Kim JK, Mauvais-Jarvis F, Ducy P, Karsenty G. Endocrine regulation of energy metabolism by the skeleton. Cell 2007;130:456 – 469. 13. Pittas AG, Harris SS, Eliades M, Stark P, Dawson-Hughes B. Association between serum osteocalcin and markers of metabolic phenotype. J Clin Endocrinol Metab 2009;94:827– 832. 14. Im JA, Yu BP, Jeon JY, Kim SH. Relationship between osteocalcin and glucose metabolism in postmenopausal women. Clin Chim Acta 2008;396:66 – 69. 15. Kindblom JM, Ohlsson C, Ljunggren O, Karlsson MK, Tivesten A, Smith U, Mellström D. Plasma osteocalcin is inversely related to fat mass and plasma glucose in elderly Swedish men. J Bone Miner Res 2009;24:785–791. 16. Saleem U, Mosley TH, Jr., Kullo IJ. Serum osteocalcin is associated with measures of insulin resistance, adipokine levels, and the presence of metabolic syndrome. Arterioscler Thromb Vasc Biol 2010;30:1474–1478. 17. Kanazawa I, Yamaguchi T, Yamamoto M, Yamauchi M, Kurioka S, Yano S, Sugimoto T. Serum osteocalcin level is associated with glucose metabolism and atherosclerosis parameters in type 2 diabetes mellitus. J Clin Endocrinol Metab 2009;94:45– 49. 18. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989;8:551–561. 19. Shanahan CM, Cary NR, Metcalfe JC, Weissberg PL. High expression of genes for calcification-regulating protein in human atherosclerotic plaques. J Clin Invest 1994;93:2393–2402. 20. Bini A, Mann KG, Kudryk BJ, Schoen FJ. Noncollagenous bone matrix proteins, calcification, and thrombosis in carotid artery atherosclerosis. Arterioscler Thromb Vasc Biol 1999;19:1852–1861. 21. Watson KE, Bostrom K, Ravindranath R, Lam T, Norton B, Demer LL. TGF-␤1 and 25-hydroxycholesterol stimulate osteoblast-like vascular cells to calcify. J Clin Invest 1994;93:2106 –2113. 22. Dhore CR, Cleutjens JP, Lutgens E, Cleutjens KB, Geusens PP, Kitslaar PJ, Tordoir JH, Spronk HM, Vermeer C, Daemen MJ. Differential expression of bone matrix regulatory proteins in human atherosclerotic plaques. Arterioscler Thromb Vasc Biol 2001;21:1998 –2003. 23. Pennisi P, Signorelli SS, Riccobene S, Celotta G, Di Pino L, La Malfa T, Fiore CE. Low bone density and abnormal bone turnover in patients with atherosclerosis of peripheral vessels. Osteoporos Int 2004;15:389 –395. 24. O’Brien KD, Kuusisto J, Reichenbach DD, Ferguson M, Giachelli C, Alpers CE, Otto CM. Osteopontin is expressed in human aortic valvular lesions. Circulation 1995;92:2163–2168.