High serum total homocysteine levels accelerate hip bone loss in healthy premenopausal women and men

Bone 52 (2013) 56–62

Contents lists available at SciVerse ScienceDirect

Bone journal homepage: www.elsevier.com/locate/bone

Original Full Length Article

High serum total homocysteine levels accelerate hip bone loss in healthy premenopausal women and men Beom-Jun Kim a, Jung-Min Koh a,⁎, Seong Hee Ahn a, Seung Hun Lee a, Eun Hee Kim b, Sung Jin Bae b, Hong-Kyu Kim b, Jae Won Choe b, Keong-Hye Lim a, Kyung Ha Pyun a, Tae-Ho Kim c, Shin-Yoon Kim c, d, Ghi Su Kim a a

Division of Endocrinology and Metabolism, Asan Medical Center, University of Ulsan College of Medicine, 138-736, Seoul, Republic of Korea Health Promotion Center, Asan Medical Center, University of Ulsan College of Medicine, 138-736, Seoul, Republic of Korea c Skeletal Diseases Genome Research Center, Kyungpook National University Hospital, 700-412, Daegu, Republic of Korea d Department of Orthopedic Surgery, Kyungpook National University School of Medicine, 700-412, Daegu, Republic of Korea b

a r t i c l e

i n f o

Article history: Received 9 July 2012 Revised 10 September 2012 Accepted 18 September 2012 Available online 26 September 2012 Edited by: Toshio Matsumoto Keywords: Homocysteine Bone loss Risk factor Bone density

a b s t r a c t Introduction: Despite extensive evidence demonstrating the direct, detrimental role of homocysteine on bone metabolism, the effects of serum total homocysteine (tHcy) on bone loss are still equivocal. In the present study, we performed a longitudinal study on healthy participants of various ages of both sexes in order to investigate the association between serum tHcy concentrations and annualized changes in bone mineral density (BMD). Methods: A total of 460 Koreans ≥ 30 years of age received comprehensive, routine health examinations for an average period of 3 years. The BMD at proximal femur sites was measured with dual-energy X-ray absorptiometry using the same equipment at baseline and follow-up. Results: After adjusting for potential confounders, the rates of bone loss at the proximal femur sites were significantly accelerated in a dose–response fashion across increasing tHcy concentrations in premenopausal women and men, but not in postmenopausal women. Consistently, compared with subjects in the lowest tHcy quartile, premenopausal women in the third and/or highest tHcy quartile and men in the highest tHcy quartile showed significantly higher rates of bone loss at all proximal femur sites (p = 0.015–0.048) and at the total femur and femur neck (p = 0.008–0.013), respectively. In contrast, there were no differences in terms of bone loss among the tHcy quartiles for postmenopausal women. Conclusion: These data provide the first clinical evidence that increased tHcy levels could be an independent risk factor for the future deterioration of bone mass in premenopausal women and men. © 2012 Elsevier Inc. All rights reserved.

Introduction Many lines of evidence now indicate that elevation of homocysteine and its disulfide derivatives (tHcy) has direct detrimental effects on bone metabolism. Homocysteine may interfere with bone collagen cross-links, thereby increasing bone fragility [1]. In a murine experimental model, rats with elevated tHcy demonstrated severe trabecular bone loss, altered microarchitectural parameters, and decreased mechanical strength in the femoral neck [2]. In addition, our in vitro studies have reported that homocysteine induces apoptosis in bone marrow stromal cells via the reactive oxygen species (ROS)-mediated mitochondrial pathway and NF-kappa B activation [3], whereas it promotes bone resorption by stimulating p38 mitogen-activated ⁎ Corresponding author at: Division of Endocrinology and Metabolism, Asan Medical Center, University of Ulsan College of Medicine, 388-1 Poongnap2-Dong, Songpa-Gu, Seoul 138-736, Republic of Korea. Fax: +82 2 3010 6962. E-mail address: [email protected] (J.-M. Koh). 8756-3282/$ – see front matter © 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.bone.2012.09.022

protein kinase (MAPK) activity and the generation of intracellular ROS in osteoclasts [4]. The first clinically relevant evidence of these in vitro and animal findings was observed in patients with homocystinuria due to cystathionine beta-synthase deficiency, an inborn error of metabolism that is characterized by very high plasma concentrations of tHcy and, among several clinical manifestations, premature osteoporosis and fractures [5]. Based on these findings, numerous epidemiological studies have been performed to assess the role of serum tHcy as a risk factor for osteoporosis-related phenotypes. All prospective trials that enrolled >1000 patients found a significant positive relationship between tHcy and osteoporotic fractures [6–9]. In contrast, studies on the association between tHcy and bone mineral density (BMD) have yielded inconsistent results [10]. However, because almost all of these studies were cross-sectional in design [11,12], the role of tHcy in bone mass could not be appropriately investigated. Furthermore, the only two studies of longitudinal bone loss that have been previously reported were based on data from elderly patients > 75 years of age on average

B.-J. Kim et al. / Bone 52 (2013) 56–62

and had contradictory findings [13,14]. Thus, the effects of tHcy on bone density are still equivocal. It is well known that the biologic differences between men and women, as well as menopause status, are important factors that influence bone metabolism, implying that the effects of various risk factors on bone health could be different depending on the sex and menopause status of the individual patient. Actually, several epidemiological studies have reported different relationships of serum tHcy levels with osteoporosis-related phenotypes in men and women [7,9,12]. In the present study, we performed a longitudinal study on healthy pre- and postmenopausal women and men in order to investigate the association between serum tHcy concentrations and annualized changes in BMD. Materials and methods Study participants This was a 3-year longitudinal health promotion center-based study. The study population consisted of subjects ≥ 30 years of age who had undergone comprehensive routine health examinations at the Asan Medical Center (AMC, Seoul, Republic of Korea) in 2007 and had returned for follow-up examinations in 2010; BMD and serum tHcy concentrations were measured during these examinations. The examinations consisted of extensive screening tests for the early detection of malignancy, diabetes, osteoporosis, and other age-related diseases. Initially, 525 subjects were identified. Of these, based on the 2007 medical records, subjects with serum liver enzyme (aspartate aminotransferase [AST] and alanine aminotransferase [ALT]) activities above 2 times the upper normal limits, increased serum creatinine (>1.4 mg/dL [> 123.8 μmol/L]), and/or abnormal thyroid function (serum thyrotropin b 0.4 mU/L or > 5.0 mU/L) were excluded from this study. In addition, subjects were excluded if they had taken drugs, such as bisphosphonates and glucocorticoids, during the study period or 12 months prior to baseline, due to their potential impacts on bone metabolism. Finally, patients that suffered from diseases such as hyperparathyroidism or rheumatoid arthritis, which might affect bone metabolism, were also excluded from the study. Some subjects met ≥ 2 exclusion criteria. The remaining 460 subjects (150 premenopausal women, 138 postmenopausal women, and 172 men) were eligible to participate in this study. Postmenopausal status was defined as the cessation of menses for ≥ 1 year, which was confirmed by a serum follicle-stimulating hormone concentration of > 30 IU/L. Lifestyle factors and anthropometric measurements All participants were interviewed and examined by physicians at the health promotion center. Information on medication usage and history of previous medical or surgical procedures were obtained for each subject. Smoking (never, past, or current) and drinking habits (no or yes) were categorized. Dairy product consumption and physical exercise were also categorized according to frequency (b or ≥ 3 times/week). Height (cm) and weight (kg) were measured using standardized protocols while the subject was dressed in light clothing and without shoes. Body mass index (BMI; kg/m 2) was calculated from the height and weight. Biochemical measurements Morning blood samples were obtained 12 h after fasting and subsequently analyzed at the certified laboratory at AMC. Serum tHcy concentrations were measured by a competitive immunoassay, using direct chemiluminescent technology (ADVIA Centaur kit; Bayer, Morristown, NJ, USA), which had intra- and interassay coefficients of variations (CVs) of 3.4–4.3% and 2.5–2.6%, respectively, and

57

a lower limit of detection of 0.5 μmol/L. Serum calcium and phosphorus concentrations were measured by the cresolphthalein complexone and the phosphomolybdate ultraviolet methods, respectively, using the Toshiba 200FR system (Toshiba Medical System Co., Tokyo, Japan). Fasting total cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, uric acid, AST, and ALT were measured using the enzymatic colorimetric method (Toshiba). Serum alkaline phosphatase (ALP), gamma glutamyltransferase (GGT), and total bilirubin concentrations were measured using the Bowers and McComb method (Toshiba), the L-γ-glutamyl-p-nitroanilide method (Toshiba), and the vanadate oxidation method (Toshiba), respectively. Serum glucose and creatinine levels were measured by the glucose oxidase method (Toshiba) and the Jaffe method (Toshiba), respectively. The serum erythrocyte sedimentation rate (ESR) was measured using the quantitative capillary photometry method (TEST-1; ALIFAX S.P.A., Polverara, Italy). The glomerular filtration rate (GFR) was estimated using the Cockcroft–Gault formula [15]. The intra- and interassay CVs for these analyses were consistently b3.5%. BMD measurements The areal BMDs (g/cm2) of the total femur, femur neck, and trochanter were measured at baseline and on the follow-up examinations by dual-energy X-ray absorptiometry (DXA) using Lunar equipment (Prodigy; Madison, WI, USA). Repeated measurements were performed using the same instruments used for the initial measurements. The in vivo precisions of the machine were 1.08%, 1.02%, and 1.06% for the femur neck, total femur, and trochanter, respectively. These values were obtained by scanning 17 volunteers who were not enrolled in the study. Each volunteer underwent 5 scans on the same day, getting on and off the table between examinations. A quality control laboratory, certified DXA technicians, and standardized procedures for scanning were implemented in order to ensure reliable DXA measurements. The rate of change in BMD was expressed as the annualized percentage of the difference between the follow-up BMD and the initial BMD, divided by the initial BMD reflecting the examination intervals. Statistical analysis Continuous and categorical variables are reported as the means ± standard deviations (SDs) and percentages, respectively, unless otherwise specified. The baseline characteristics of the 3 groups were compared using one-way analysis of variance (ANOVA) for continuous variables and the Chi-square test for categorical variables. Age and GFR-adjusted least-square means (95% CI) of the tHcy levels in preand postmenopausal women and men were estimated and compared using analysis of covariance (ANCOVA). Tests to determine the level of interaction between variables were performed using likelihood ratio tests by comparing 2 nested models, one with the main effects only and the other with both the main effects and interaction terms. To examine the relationship between annualized BMD changes at the total femur, femur neck, and trochanter and covariates, linear univariate regression analyses were performed. Next, in order to determine the independent effects of the serum tHcy level on annualized BMD changes at various proximal femur sites, we used a multiple regression model with the annualized BMD change as a dependent variable and the serum tHcy level as an independent variable. In these analyses, the serum tHcy concentration was logarithmically transformed because the distribution was positively skewed. Confounding independent variables were selected on the basis of being clinically applicable and/or their statistical significance according to linear univariate regression models (inclusion criterion: p b 0.2). In addition, the same variables were forced into the multiple regression models of the 3 proximal femur sites in each study group, regardless of the significance (p b or >0.2) at the other sites in order to obtain comparable analyses. Consequently, the base adjustment model included age, BMI, baseline BMD,

58

B.-J. Kim et al. / Bone 52 (2013) 56–62

GFR, smoking and drinking habits, exercise habits, and dairy consumption. In addition to the factors included in the base model, the multivariable adjustment model for premenopausal women included waist circumference, systolic BP, and corrected calcium, fasting glucose, triglycerides, and GGT levels, while the multivariable adjustment model for postmenopausal women included waist circumference, diastolic BP, and corrected calcium, fasting glucose, total cholesterol, triglycerides, HDL cholesterol, ALP, and GGT levels. The multivariable adjustment model for men included waist circumference, systolic BP, diastolic BP, and fasting glucose, uric acid, total cholesterol, triglycerides, GGT, and total bilirubin levels as well as the other factors that were included in the base model. In the multiple regression analyses, we used the “enter method”, meaning that all variables were entered simultaneously into the model. To further test our hypothesis that higher tHcy levels may be associated with accelerated bone loss, we categorized our subjects into 4 groups according to baseline serum tHcy concentration, and then performed the analyses. The multivariateadjusted least-square mean (95% CI) annualized BMD changes, in terms of the serum tHcy quartiles, were estimated using ANCOVA after adjusting for confounders. The trends of the annualized BMD changes in each proximal femur site across increasing tHcy quartiles were checked by examining the p values for each trend using multiple linear regression analysis, with the annualized BMD changes as the dependent variable and the ordinal tHcy quartiles as the independent variable. ANCOVA was used to compare annualized BMD changes according to serum tHcy quartiles after adjusting for confounding variables. All statistical analyses were performed using SPSS statistical software (SPSS Inc., Chicago, IL, USA), and p b 0.05 was considered statistically significant.

Ethics This study was approved by the Institutional Review Board of AMC, and written informed consent was obtained from all study subjects. Results The baseline characteristics of the 460 study subjects are shown in Table 1. The mean ages of the pre- and postmenopausal women and men were 45.6 ± 3.5 years (range: 30–54 years), 55.6 ± 6.2 years (range: 44–80 years), and 55.0 ± 8.4 years (range: 35–80 years), respectively. BMI and the percentage of current smokers and drinkers were significantly higher in men than in women, whereas the percentage of subjects who consumed dairy products was higher in women than in men. Women demonstrated significantly lower BMD values at the total femur and trochanter than men, and postmenopausal women also demonstrated lower BMD values at these sites than premenopausal women. The BMD of the femur neck was higher in men than postmenopausal women, but it was not significantly different between men and premenopausal women. The serum tHcy level was significantly higher in men (12.3 μmol/L; 95% CI: 11.7–13.0 μmol/L) than in women, and, among women, it was higher in postmenopausal women (9.8 μmol/L; 95% CI: 9.4–10.1 μmol/L) than in premenopausal women (8.7 μmol/L; 95% CI: 8.4–9.1 μmol/L; Fig. 1). After adjusting for age and GFR, the wellknown main determinants of tHcy levels [16,17], men (12.4 μmol/L; 95% CI: 11.9–12.9 μmol/L) still had higher tHcy levels than women. However, the level of statistical significance between pre- (8.8 μmol/L;

Table 1 Baseline characteristics of the study population. Variables

Premenopausal (n = 150)

Postmenopausal (n = 138)

Men (n = 172)

p value

Follow-up interval (month) Age (year) Body mass index (kg/m2) Current smokers (%) Drinkers (%) Dairy product consumption (≥3 times/week) (%) Physical exercise (≥3 times/week) (%) Total femur BMD (g/cm2) Femur neck BMD (g/cm2) Trochanter BMD (g/cm2) GFR (mL/min) Waist circumference (cm) Systolic BP (mm Hg) Diastolic BP (mm Hg) Corrected calcium (mg/dL) a Phosphorus (mg/dL) Uric acid (mg/dL) ESR (mm/h) Fasting glucose (mg/dL) Total cholesterol (mg/dL) HDL cholesterol (mg/dL) Triglycerides (mg/dL) AST (IU/L) ALT (IU/L) ALP (IU/L) GGT (IU/L) Total bilirubin (mg/dL)

35.6 ± 7.1 45.6 ± 3.5d 22.7 ± 3.2d 6.0 11.3 43.3 46.0 0.989 ± 0.116d 0.935 ± 0.111d 0.771 ± 0.105d 91.6 ± 17.0d 73.5 ± 11.3 111.1 ± 14.7d 70.2 ± 10.3 8.97 ± 0.23d 3.90 ± 0.64d 4.12 ± 0.80d 19.7 ± 12.9 92.3 ± 12.6d 184.5 ± 27.5d 63.9 ± 15.1 98.9 ± 58.1 21.6 ± 10.2d 17.4 ± 11.7d 48.9 ± 12.7d 16.0 ± 14.3 0.99 ± 0.32

36.9 ± 7.6 55.6 ± 6.2c 23.6 ± 3.1c 4.3 13.8 40.6 47.1 0.930 ± 0.109c 0.878 ± 0.110c 0.731 ± 0.094c 81.1 ± 15.8c 75.6 ± 15.3 118.1 ± 15.6c 72.9 ± 9.2 9.18 ± 0.24c 4.27 ± 0.53c 4.47 ± 0.86c 23.1 ± 12.9 97.5 ± 17.4c 199.7 ± 34.8c 61.1 ± 13.9 111.7 ± 54.1 25.0 ± 11.9c 21.7 ± 15.3c 67.1 ± 17.7c 18.8 ± 21.0 0.95 ± 0.28

36.4 ± 8.3 55.0 ± 8.4c 24.7 ± 2.3c,d 30.8 57.6 27.3 48.3 1.022 ± 0.126c,d 0.953 ± 0.116d 0.858 ± 0.115c,d 88.4 ± 18.5d 86.8 ± 6.4c,d 119.6 ± 16.9c 74.0 ± 10.6c 9.10 ± 0.25c,d 3.67 ± 0.69c,d 5.83 ± 1.17c,d 14.9 ± 10.7c,d 106.5 ± 23.6c,d 189.1 ± 31.6c,d 53.2 ± 12.7c,d 143.5 ± 77.6c,d 26.8 ± 11.2c 27.1 ± 16.3c,d 59.5 ± 16.2c,d 42.1 ± 48.6c,d 1.12 ± 0.32c,d

0.352 b0.001 b0.001 b0.001 b0.001 0.006 0.921 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 0.004 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001 b0.001

b

Values are presented as the mean ± SD unless otherwise specified. BMD, bone mineral density; GFR, glomerular filtration rate; BP, blood pressure; ESR, erythrocyte sedimentation rate; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, gamma glutamyltransferase. SI conversion factors: to convert mg/dL to mmol/L for calcium, multiply values by 0.2495; to convert mg/dL to mmol/L for phosphorus, multiply values by 0.3229; to convert mg/dL to μmol/L for uric acid, multiply values by 59.48; to convert mg/dL to mmol/L for glucose, multiply values by 0.0555; to convert mg/dL to mmol/L for cholesterol, multiply values by 0.0259; to convert mg/dL to mmol/L for triglycerides, multiply values by 0.0113; to convert mg/dL to μmol/L for bilirubin, multiply values by 17.1. a Corrected calcium concentration (mg/dL) = total calcium concentration (mg/dL) + 0.8 × [4.0 (g/dL) − serum albumin concentration (g/dL)]. b These p values for comparing the baseline characteristics of the 3 groups were generated using analysis of variance (ANOVA) for continuous variables and the χ2-test for categorical variables. c p b 0.05 vs. premenopausal women by post-hoc analysis using the ANOVA test. d p b 0.05 vs. postmenopausal women by post-hoc analysis using the ANOVA test. Bold means that values are statistically significant.

B.-J. Kim et al. / Bone 52 (2013) 56–62

59

Fig. 1. Levels of serum total homocysteine before and after adjustment for age and GFR in pre- and postmenopausal women and men. Values before and after adjusting for age and GFR were generated and compared using analysis of variance (ANOVA) and analysis of covariance (ANCOVA), respectively. Values are presented as mean ± 95% CI. GFR, glomerular filtration rate.

95% CI: 8.2–9.4 μmol/L) and postmenopausal women (9.6 μmol/L; 95% CI: 9.1–10.1 μmol/L) was marginal. When we tested whether any association between tHcy levels and the rate of BMD changes might be modified by the sex and menopause status, there was strong evidence supporting interactions between the tHcy level (expressed as a continuous variable) and the sex and menopause status (coded as 0/1/2 for pre- and postmenopausal women and men, respectively, and expressed as a categorical variable) for the prediction of the rate of BMD changes at all proximal femur sites (p for tests of interaction = 0.006 to b0.001). Therefore,

we separately performed further analyses in pre- and postmenopausal women and men. Multivariate analyses were used to examine the independent effects of serum tHcy levels on annualized BMD changes at various proximal femur sites after considering other possible covariates in each study group (Table 2). A high serum tHcy concentration was independently associated with accelerated bone loss at all proximal femur sites before and after adopting various adjustment models in premenopausal women, whereas it was inversely associated with annualized BMD change at the femur neck in postmenopausal women

Table 2 Multiple regression analysis for determining the independent effects of serum total homocysteine on annualized BMD changes at various proximal femur sites in pre- and postmenopausal women and men. Adjustment

Annualized BMD changes in the total femur

Annualized BMD changes in the femur neck

Annualized BMD changes in the trochanter

S.E

p value

R2

β

S.E

p value

R2

β

S.E

p value

R2

A) Premenopausal women Unadjusted −1.041 Age and GFR −0.864 Base −0.846 Multivariablea −1.002

0.368 0.356 0.368 0.389

0.005 0.017 0.023 0.011

0.051 0.170 0.191 0.230

−0.876 −0.713 −0.742 −0.841

0.370 0.362 0.373 0.393

0.019 0.051 0.049 0.034

0.037 0.140 0.164 0.212

−1.719 −1.533 −1.477 −1.751

0.694 0.703 0.720 0.768

0.014 0.031 0.042 0.024

0.040 0.080 0.112 0.143

B) Postmenopausal women Unadjusted −0.383 Age and GFR −0.743 Base −0.494 b Multivariable −0.475

0.397 0.399 0.414 0.432

0.337 0.064 0.235 0.274

0.007 0.098 0.161 0.223

−0.690 −1.197 −0.840 −0.748

0.497 0.501 0.523 0.536

0.167 0.018 0.111 0.166

0.014 0.096 0.151 0.241

−0.720 −1.077 −0.838 −0.979

0.616 0.632 0.666 0.694

0.245 0.090 0.211 0.161

0.010 0.063 0.101 0.171

C) Men Unadjusted Age and GFR Base Multivariablec

0.227 0.235 0.247 0.264

0.012 0.015 0.014 0.019

0.037 0.037 0.057 0.115

−0.701 −0.687 −0.725 −0.712

0.299 0.310 0.317 0.340

0.020 0.028 0.023 0.038

0.031 0.036 0.091 0.146

−0.483 −0.543 −0.650 −0.453

0.398 0.413 0.425 0.450

0.227 0.190 0.128 0.316

0.009 0.012 0.066 0.146

β

−0.576 −0.579 −0.614 −0.629

The enter method was applied to this model using annualized BMD changes at each proximal femur site as the dependent variable and the serum total homocysteine level as the independent variable. The p values indicate the statistical significance regarding the independent effect of serum total homocysteine on annualized BMD changes at various proximal femur sites in each adjustment model. The R2, which is a coefficient of determination, provides the information about the goodness of fit of each multivariable regression model after adopting various confounders. Confounding independent variables were selected on the basis of being clinically applicable and/or additionally on the basis of their statistical significance in linear univariate regression models (inclusion criterion: p b 0.2). Base model: adjusted for age, BMI, baseline BMD, GFR, smoking and drinking habits, physical exercise, and dairy product consumption. Serum total homocysteine concentrations were logarithmically transformed because the distribution was positively skewed. BMD, bone mineral density; BMI, body mass index; GFR, glomerular filtration rate; BP, blood pressure; ALP, alkaline phosphatase; GGT, gamma glutamyltransferase. a Multivariable model for premenopausal women: adjusted for waist circumference, systolic BP, corrected calcium, fasting glucose, triglycerides, and GGT as well as the factors included in the base model. b Multivariable model for postmenopausal women: adjusted for waist circumference, diastolic BP, corrected calcium, fasting glucose, total cholesterol, triglycerides, HDL cholesterol, ALP, and GGT as well as the factors included in the base model. c Multivariable model for men: adjusted for waist circumference, systolic BP, diastolic BP, fasting glucose, uric acid, total cholesterol, triglycerides, GGT, and total bilirubin as well as the factors included in the base model. Bold means that values are statistically significant.

60

B.-J. Kim et al. / Bone 52 (2013) 56–62

only after adjusting for age and GFR. In men, a high serum tHcy concentration was also independently associated with accelerated bone loss at the total femur and femur neck, even after adjusting for all potential confounders. However, serum tHcy did not contribute to annualized BMD change at the trochanter of men in any of the adjusted models. When we divided men into two groups, based on the median age, and performed subgroup analyses, the detrimental effect of serum tHcy on bone loss at the total femur and femur neck after

adopting the multivariable adjustment model was more prominent in men ≥ 55 years old than in those b55 years old (Supplementary Table 1). Multivariate-adjusted least-square mean annualized BMD changes according to serum tHcy quartiles were estimated after considering potential confounding factors (Fig. 2). The overall rates of bone loss at the total femur, femur neck, and trochanter were − 0.43%/year, − 0.36%/year, and − 0.73%/year, respectively, in premenopausal

Fig. 2. Annualized BMD changes at various proximal femur sites after adjusting for confounding factors, according to the serum total homocysteine quartiles in pre- and postmenopausal women and men. The p values for trend were generated using multiple linear regression analysis. Multivariate-adjusted least-square mean annualized BMD changes, in terms of the serum total homocysteine quartiles, were estimated using analysis of covariance (ANCOVA). ⁎Determined to be statistically and significantly different from the lowest quartile (Q1) by ANCOVA after adjusting for confounding variables. aConfounding factors for premenopausal women included age, BMI, baseline BMD, GFR, smoking and drinking habits, physical exercise, dairy product consumption, waist circumference, systolic BP, corrected calcium, fasting glucose, triglycerides, and GGT levels. bConfounding factors for postmenopausal women included age, BMI, baseline BMD, GFR, smoking and drinking habits, physical exercise, dairy product consumption, waist circumference, diastolic BP, corrected calcium, fasting glucose, total cholesterol, triglycerides, HDL cholesterol, ALP, and GGT levels. cConfounding factors for men included age, BMI, baseline BMD, GFR, smoking and drinking habits, physical exercise, dairy product consumption, waist circumference, systolic BP, diastolic BP, fasting glucose, uric acid, total cholesterol, triglycerides, GGT, and total bilirubin levels. BMD, bone mineral density; BMI, body mass index; GFR, glomerular filtration rate; BP, blood pressure; ALP, alkaline phosphatase; GGT, gamma glutamyltransferase.

B.-J. Kim et al. / Bone 52 (2013) 56–62

women, − 1.16%/year, − 1.19%/year, and − 1.48%/year, respectively, in postmenopausal women, and − 0.29%/year, − 0.37%/year, and − 0.59%/year, respectively, in men. After adjusting for confounders in the multivariable models, the rates of bone loss at each of the proximal femur sites in premenopausal women and at the total femur and femur neck in men were significantly accelerated in a dose–response manner across increasing tHcy quartiles. However, there was no trend associated with annualized BMD changes at any sites and the tHcy quartiles in postmenopausal women. Consistently, compared with subjects in the lowest tHcy quartile, premenopausal women in the third and/or highest tHcy quartile and men in the highest tHcy quartile showed significantly faster bone loss at all proximal femur sites (p = 0.015–0.048) and at the total femur and femur neck (p = 0.008–0.013), respectively, even after adjusting for all potential confounders. In contrast, there was no difference in terms of bone loss among the tHcy quartiles in postmenopausal women. Partial correlation analysis revealed that serum tHcy concentrations were positively correlated with serum total ALP levels after adjusting for age, BMI, GFR, smoking and drinking habits, physical exercise, and dairy product consumption in premenopausal women (γ = 0.175; p = 0.034) and men (γ = 0.211; p = 0.006), whereas there was no association between these factors in postmenopausal women (γ = − 0.005; p = 0.950). Discussion In this longitudinal study of 460 healthy Koreans, we found that the rates of bone loss over 3 years at multiple proximal femur sites significantly accelerated in a dose–response manner as the tHcy concentration increased, even after adjusting for potential confounding factors in premenopausal women and men. These data provide the first clinical evidence that increased tHcy levels could be an independent risk factor for future deterioration of bone mass in premenopausal women and men. Although there are accumulating data that support the clear roles of homocysteine on bone metabolism, there have been few clinical studies relating tHcy to age-related bone loss that have been carried out using a longitudinal design. Based on previously published studies, to the best of our knowledge, we could only identify 2 well-designed studies. In the Framingham Osteoporosis Study, which included 684 participants (mean age: 75.3 years) who provided plasma tHcy information and 4-year follow-up BMD, the mean percentage of annual bone loss at the femur neck is not statistically different depending on the plasma tHcy levels [14]. On the other hand, Zhu et al. [13] showed that a high tHcy level is associated with greater bone loss at the total hip in a 5-year cohort study that included 1213 women aged 70–85 years. Consequently, not only are the results of these cited studies opposite, but both also included only elderly subjects. Therefore, the existing studies are not enough to draw strong conclusions regarding the effects of tHcy on the changes of bone mass. We believe that the present study has important implications, in that it was performed on subjects across a wide range of ages and included both sexes, and demonstrates that the detrimental effects of tHcy on bone loss are significant even in relatively younger subjects, compared with other longitudinal studies. Furthermore, the fact that we intentionally applied strict exclusion criteria based on medical history and routine laboratory findings and made careful adjustments for factors known to affect bone metabolism in order to appropriately investigate the pure physiological effects of tHcy, and that the association between tHcy and annualized BMD changes was consistent across multiple sites, improves the likelihood that tHcy may be one of the risk factors for future low bone mass. It is interesting that serum tHcy levels were higher in postmenopausal women than premenopausal women, which is in agreement with previous studies showing the association between low estrogen levels and hyperhomocystinemia [18], while the effect of tHcy on bone loss was observed only in premenopausal women. The abrupt

61

decrease in estrogen levels during menopause is the most important biological event that causes enormous changes in homeostasis in women. Regarding bone metabolism, substantial evidence indicates that estrogen deficiency cannot only directly affect bone cells [19], but can also involve diverse, indirect mechanisms that result in accelerated bone turnover and bone loss, such as ROS, cytokines, and growth factors [20,21]. In the present study, although we cannot determine how tHcy differently contributes to bone mass depending on the menopause status, we assume that the effects of tremendous changes caused by estrogen deficiency in postmenopausal women may be strong enough to countervail those of tHcy. Because most men do not develop sudden hypogonadism with aging, in contrast to women, and the 87.8% of men in our cohort had the age ≤ 65 years old, which is a threshold for generally being considered as the elderly people, we performed the analyses as one group without dividing men. Consequently, we did observe the deleterious effects of tHcy on bone loss in men after adjusting for the multivariable factors including age. Meanwhile, when we divided men in our cohort into two groups, for convenience based on the median age, the inverse association between serum tHcy and bone loss at the total femur and femur neck was more prominent in men ≥ 55 years old than in those b 55 years old. However, the men included in the older group were still marked younger (mean age = 61.0 years old) than those in other tHcy studies in men [14], implying that they may not be the representative of male hypogonadism, and thus we cannot conclude that the role of tHcy on bone metabolism is different depending on the male hormonal status. Further studies focusing on how sex hormones interact with tHcy in relation to bone metabolism may help to explain the sex and menopause difference that are associated with tHcy and low bone mass. Because information regarding bone turnover markers was not available for our cohort, we examined the association between serum tHcy and total ALP levels as an alternative line of investigation. ALP activity in circulation is normally contributed to by bone and liver isoforms in approximately equal amounts [22]. In general, serum total ALP is regarded as a useful marker which can reflect the degree of bone turnover in subjects without liver diseases [23]. Based on this background, we performed a partial correlation analysis and found that serum tHcy concentrations were positively correlated with serum total ALP levels after adjusting for confounders in premenopausal women and men. In normal adults, the three phases of the remodeling cycle (i.e., resorption, reversal, and formation) have different lengths. Resorption probably continues for about two weeks. The reversal phase may last up to four weeks, while formation can continue for four months until the new bone structural unit is fully formed. Because the resorption phase is first and relatively short, and the period required for osteoblastic replacement of the bone is long, any increase in the rate of bone remodeling will result in the loss of bone mass and skeletal fragility [20]. Therefore, the present result showing the positive association between serum tHcy and total ALP concentrations suggests that accelerated bone loss associated with a high tHcy level may result from an increased bone turnover rate in these subjects. Higher circulating tHcy is known to be a risk factor for osteoporotic fractures [8], even independently of the BMD value [6]. Furthermore, although a reduced bone mass is one of the strongest predictors for future fracture, other factors that cannot be detected by BMD measurement, including microarchitectural parameters, can contribute to bone fragility. Therefore, even if a high serum tHcy level in postmenopausal women was not a factor related to the decline in BMD in the present study, there is a possibility that it may be involved in the risk of fractures in postmenopausal women through the mechanisms not mediated by BMD. There are some limitations to this study. First of all, information about nutrient status of subjects was not available in our cohort. Low folate, cobalamin, and vitamin B6 levels, which are important cofactors in homocysteine metabolism, are known as the primary

62

B.-J. Kim et al. / Bone 52 (2013) 56–62

determinants of elevated tHcy concentrations in the elderly [24]. Because several epidemiological and in vitro studies suggest that low levels of these nutrients may be responsible for decreased bone mass and higher risk of osteoporotic fractures [14,25] and may directly affect osteoblasts and osteoclasts [26,27], there is the presence of the possibility that the observed association between serum tHcy and hip bone loss in the present study could be attributed to inadequate nutrient status. However, our cohort mainly consisted of healthy and relatively young aged subjects, who are less likely to have nutrient deficiencies [28], implying that our results may provide additional evidence supporting the causal involvement of homocysteine in bone metabolism. Second, the study population consisted of subjects who visited a health promotion center, and these subjects may not be representative of the general population, thus possibly resulting in selection bias. Third, although we attempted to consider as many confounding factors as possible, we cannot exclude the possibility that the observed associations could be attributed to uncontrolled factors that affect HCY and/or BMD, such as 25-hydroxyvitamin D levels or years since menopause. Lastly, the study population was exclusively Korean. Thus, we cannot be certain that our results will be applicable in other populations. In summary, an elevated serum tHcy concentration was associated with higher rates of bone loss at various proximal femur sites in healthy premenopausal women and men. The data presented here may have clinical implications for the prevention of osteoporosis and subsequent osteoporotic fracture through the identification of groups at high risk of future bone loss, especially in relatively younger subjects. Further interventional studies are needed to confirm the causal, detrimental role of homocysteine in subjects at risk of developing low bone mass. Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.bone.2012.09.022. Acknowledgments This study was supported by grants from the Korea Health Technology R&D Project and the National Project for Personalized Genomic Medicine, Ministry of Health & Welfare, Republic of Korea (project nos. A110536 and A111218-GM03, respectively). References [1] Lubec B, Fang-Kircher S, Lubec T, Blom HJ, Boers GH. Evidence for McKusick's hypothesis of deficient collagen cross-linking in patients with homocystinuria. Biochim Biophys Acta 1996;1315:159-62. [2] Herrmann M, Wildemann B, Claes L, Klohs S, Ohnmacht M, Taban-Shomal O, et al. Experimental hyperhomocysteinemia reduces bone quality in rats. Clin Chem 2007;53:1455-61. [3] Kim DJ, Koh J-M, Lee O, Kim NJ, Lee Y-S, Kim YS, et al. Homocysteine enhances apoptosis in human bone marrow stromal cells. Bone 2006;39:582-90. [4] Koh JM, Lee YS, Kim YS, Kim DJ, Kim HH, Park JY, et al. Homocysteine enhances bone resorption by stimulation of osteoclast formation and activity through increased intracellular ROS generation. J Bone Miner Res 2006;21:1003-11.

[5] Mudd SH, Skovby F, Levy HL, Pettigrew KD, Wilcken B, Pyeritz RE, et al. The natural history of homocystinuria due to cystathionine beta-synthase deficiency. Am J Hum Genet 1985;37:1–31. [6] van Meurs JB, Dhonukshe-Rutten RA, Pluijm SM, van der Klift M, de Jonge R, Lindemans J, et al. Homocysteine levels and the risk of osteoporotic fracture. N Engl J Med 2004;350:2033-41. [7] McLean RR, Jacques PF, Selhub J, Tucker KL, Samelson EJ, Broe KE, et al. Homocysteine as a predictive factor for hip fracture in older persons. N Engl J Med 2004;350:2042-9. [8] Gjesdal CG, Vollset SE, Ueland PM, Refsum H, Meyer HE, Tell GS. Plasma homocysteine, folate, and vitamin B 12 and the risk of hip fracture: the Hordaland Homocysteine Study. J Bone Miner Res 2007;22:747-56. [9] Dhonukshe-Rutten RA, Pluijm SM, de Groot LC, Lips P, Smit JH, van Staveren WA. Homocysteine and vitamin B12 status relate to bone turnover markers, broadband ultrasound attenuation, and fractures in healthy elderly people. J Bone Miner Res 2005;20:921-9. [10] Herrmann M, Peter Schmidt J, Umanskaya N, Wagner A, Taban-Shomal O, Widmann T, et al. The role of hyperhomocysteinemia as well as folate, vitamin B(6) and B(12) deficiencies in osteoporosis: a systematic review. Clin Chem Lab Med 2007;45:1621-32. [11] Morris MS, Jacques PF, Selhub J. Relation between homocysteine and B-vitamin status indicators and bone mineral density in older Americans. Bone 2005;37:234-42. [12] Gjesdal CG, Vollset SE, Ueland PM, Refsum H, Drevon CA, Gjessing HK, et al. Plasma total homocysteine level and bone mineral density: the Hordaland Homocysteine Study. Arch Intern Med 2006;166:88-94. [13] Zhu K, Beilby J, Dick IM, Devine A, Soos M, Prince RL. The effects of homocysteine and MTHFR genotype on hip bone loss and fracture risk in elderly women. Osteoporos Int 2009;20:1183-91. [14] McLean RR, Jacques PF, Selhub J, Fredman L, Tucker KL, Samelson EJ, et al. Plasma B vitamins, homocysteine, and their relation with bone loss and hip fracture in elderly men and women. J Clin Endocrinol Metab 2008;93:2206-12. [15] Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31-41. [16] Jacques PF, Bostom AG, Wilson PW, Rich S, Rosenberg IH, Selhub J. Determinants of plasma total homocysteine concentration in the Framingham Offspring cohort. Am J Clin Nutr 2001;73:613-21. [17] Refsum H, Smith AD, Ueland PM, Nexo E, Clarke R, McPartlin J, et al. Facts and recommendations about total homocysteine determinations: an expert opinion. Clin Chem 2004;50:3–32. [18] Dimitrova KR, DeGroot K, Myers AK, Kim YD. Estrogen and homocysteine. Cardiovasc Res 2002;53:577-88. [19] Feng X, McDonald JM. Disorders of bone remodeling. Annu Rev Pathol 2011;6: 121-45. [20] Raisz LG. Pathogenesis of osteoporosis: concepts, conflicts, and prospects. J Clin Invest 2005;115:3318-25. [21] Weitzmann MN, Pacifici R. Estrogen deficiency and bone loss: an inflammatory tale. J Clin Invest 2006;116:1186-94. [22] Green S, Anstiss CL, Fishman WH. Automated differential isoenzyme analysis. II. The fractionation of serum alkaline phosphatases into “liver”, “intestinal” and “other” components. Enzymologia 1971;41:9–26. [23] van Straalen JP, Sanders E, Prummel MF, Sanders GT. Bone-alkaline phosphatase as indicator of bone formation. Clin Chim Acta 1991;201:27-33. [24] Selhub J, Jacques PF, Wilson PW, Rush D, Rosenberg IH. Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. JAMA 1993;270:2693-8. [25] Yazdanpanah N, Zillikens MC, Rivadeneira F, de Jong R, Lindemans J, Uitterlinden AG, et al. Effect of dietary B vitamins on BMD and risk of fracture in elderly men and women: the Rotterdam study. Bone 2007;41:987-94. [26] Herrmann M, Schmidt J, Umanskaya N, Colaianni G, Al Marrawi F, Widmann T, et al. Stimulation of osteoclast activity by low B-vitamin concentrations. Bone 2007;41:584-91. [27] Kim GS, Kim CH, Park JY, Lee KU, Park CS. Effects of vitamin B12 on cell proliferation and cellular alkaline phosphatase activity in human bone marrow stromal osteoprogenitor cells and UMR106 osteoblastic cells. Metabolism 1996;45:1443-6. [28] The Division of Chronic Disease Surveillance, Korea Centers for Disease Control and Prevention. The Fourth Korea National Health and Nutrition Examination Survey (KNHANES IV); 2009.