SEARCH

SEARCH BY CITATION

Keywords:

  • osteoporosis;
  • ESRα;
  • interaction;
  • alcohol drinking;
  • Japanese

Abstract

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We hypothesized that environmental factors might affect the relationship between genetic predisposition and the risk of bone mineral density (BMD) loss. Cases were 114 Japanese women with a confirmed diagnosis of postmenopausal osteoporosis and controls were 171 general Japanese women. Genetic risk of SNPs in the estrogen receptors was analyzed by a case–control study. The interaction between gene and environmental factors for osteoporosis were assessed by a case-only design. Significant increases in osteoporosis risk were observed with minor alleles of rs2077647 located in the first exon and rs2234693 located in the first intron of estrogen receptor α (ESRα). Haplotype CC at these risk SNPs was strongly associated with osteoporosis risk (odds ratio [OR] = 3.15, 95% confidence interval [CI] = 1.83–5.41). There was a statistically significant interaction between haplotype CC and alcohol drinking; moderate alcohol consumption decreased genetic risk of osteoporosis (OR = 0.22, 95%CI = 0.05–0.83). © 2012 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 30:1529–1534, 2012

In 2010, 23.1% of Japanese were at least 65-years of age.1 The 2009 life expectancy of Japanese women at birth was 86.44 years; the highest in the world.2 Since the osteoporosis risk increases with age, 35% of Japanese women over 50-years of age are osteoporotic at the spine.3 Osteoporosis is the major cause of bone fracture in elderly women. Therefore, the prevention of osteoporosis in Japan remains a challenge.

Susceptibility to postmenopausal osteoporosis is influenced greatly by various genetic factors such as estrogen receptor α (ESRα), lipoprotein receptor-related protein 5 (LRP5), transforming growth factor (TGF-β1), bone morphogenic proteins (BMPs), vitamin D receptor (VDR), core-binding factor A1 (CBFA1), and collagen type I α I (COLIA1).4 Moreover, a low body mass index (BMI), early menopause, and other environmental factors are often associated with bone mineral density (BMD).5, 6 We hypothesized that environmental factors might alter the effect of genetic risk and the subsequent loss of BMD. For the isolation of genetic factors, especially the single nucleotide polymorphisms (SNPs) in ESRα and β, we compared cases and controls and studied the gene–environment interaction in cases only.

METHODS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Subjects

One hundred and fourteen Japanese women who were diagnosed with postmenopausal osteoporosis between 1990 and 2008 in the Sapporo Medical University Hospital were included as cases. Osteoporosis diagnoses were made according to established Japanese criteria, including low BMD below 70% of the young adult mean (T score < −2.5 measured by dual energy X-ray absorptiometry) or the presence of atraumatic spinal fracture.7 Patients with a history of any disease or medication known to affect bone metabolism were excluded from the study. The osteoporotic women had a mean (range) age of 67.5 (46.0–86.0) years. One hundred and seventy one Japanese women were used as controls. They were selected at random from the staffs who received a medical checkup between 2004 and 2005 at the Sapporo Medical University. They had a mean (range) age of 39.0 (18.0–63.0) years.

This study was approved by the ethics committee of the Sapporo Medical University and informed consent was obtained from all subjects prior to participation.

DNA Extraction and Genotyping Assays

For purposes of the genetic case–control study, blood samples were collected from all subjects, and stored at −80°C. DNA was extracted from frozen blood samples according to the frozen blood protocol using KURABO Genomic Automation NA-3000. SNP genotyping was done using the TaqMan® SNP Genotyping Assays with an Applied Biosynthesis 7500 Real-Time PCR System.

Selection of Genes and Polymorphisms

Estrogen receptor genes that might be associated with BMD were selected based on prior research.8–11 ESRα is a nuclear receptor and a ligand-activated transcription factor that mediates the effects of the steroid hormone, 17β-oestradiol, in both males and females. ESRβ is highly homologous to ESRα and binds estrogens with similar affinity to that of ESRα. The genes that encode ESRα and ESRβ are located on chromosomes 6 and 14, respectively. Tag SNPs with a minor allele frequency of more than 30% in Japanese were selected from the HapMap data and the SNP databases of National Center for Biotechnology Information. In total, four SNPs in the ESRα and two SNPs in the ESRβ were analyzed. The SNPs ID located in ESRα were as follows: rs2077647 in exon1, rs2234693 in intron1, rs1801132 in exon4, and rs2228480 in exon8. The SNPs ID located in ESRβ included rs1256049 in Exon5 and rs1152579 in Intron8.

Questionnaire

For the analysis of the interaction between genetic and lifestyle factors, self-administered questionnaire were administered to all osteoporotic women. The questionnaire assessed the following items: height and weight at age 20's and at diagnosis, type and frequency of physical activity in exercise at age 10's, frequency of alcohol consumption, duration of smoking, and number per day of tobacco, age of menopause, number of pregnancies and frequency of dietary intake of milk, dairy products, and soybean products at age 10's.

Statistical Analysis

The osteoporotic risk of genetic factors was assessed using a case–control design. Controls were not matched to cases with age, because SNPs were not influenced by the age. The genotypes of cases were compared with those of general Japanese population as controls. The interaction between SNPs and life style factors environmental factors was assessed using a case-only design.

With odds ratios (ORs) of 1.8–2.0, a total sample size of 130–170 is required to achieve at least 90% statistical power at the 5% significance level. ORs of 1.8–2.0 came from the result of a meta-analysis for SNPs in ESRα by Ioannidis et al.12 The total sample size in this study was 311, which was sufficient to detect significant ORs for SNPs of genes. The case-only study which was the powerful design was used to analyze the interaction between SNPs and environmental factors.13–15 A case–control study with 100 cases would have low power to detect possible interaction effects with OR < 5.0, whereas a case-only study with 100 cases would have moderate power if the exposure prevalence were >15%.16 A sample size of 114 cases in this study was sufficient to detect the gene–environment interaction.

BMI at age 20 and at diagnosis were calculated as weight divided by the square of height in meters (kg/m2) and categorized into underweight (<18.5) and normal (≥18.5). The metabolic equivalents (MET/d) were used as a measure of physical activity. One MET equals approximately 3.5 ml/kg/min of oxygen consumed, or the cost of sitting at rest. Higher activities levels are represented in multiples of this value. The MET values for various physical activities were provided by Ainsworth et al.17 Physical activity at age 10 and 20 were categorized into <3METs/d and ≥3METs/d. Other lifestyle factors were dichotomized as follow: smoking at diagnosis (yes and former/no), alcohol consumption (every day and several times a week or month/never), age at menopause (≤45 and >45 years), number of births (≤1 and ≥2), milk intake at age 10 (every day and several times a week/several times a month or never), and soy products intake at age 10 (every day and several times a week/several times a month or never).

All SNPs were tested for the Hardy Weinberg equilibrium (HWE) prior to statistical analysis. Genotypes of each SNP were classified into three categories as follows: the homozygote of major alleles, the heterozygote of major and minor alleles, and the homozygote of minor alleles. A major allele homozygote was used as the reference category. Diplotypes (the combination of haplotypes) and genetic risk score (the number of minor alleles) in two risk SNPs at ESRα were classified into following four categories; both the homozygote of major alleles in two SNP (diplotype TT/TT, score 0), either the homozygote of major alleles (diplotype TT/TC, TT/CT, TC/TC, and CT/CT, score 1), both the heterozygote of major and minor alleles (diplotype TT/CC, score 2), and both the homozygote of minor alleles and others (diplotype TC/CC, CT/CC, and CC&CC, score ≥3). The linkage disequilibrium coefficient (D′) of rs2077647 and rs2234693 among Japanese is high (D′ = 0.946) and diplotype which consists of the genotypes of heterozygote of major and minor alleles at two SNPs is TT/CC by HapMap Project data. Although genetic risk score of diplotype TC/TC and CT/CT was 2, they were included in the second category, because either one of two SNPs was the genotypes of homozygote of major alleles. Diplotype TT/TT was used as the reference category. Furthermore, these four categories were broadly summarized into low- and high-risk diplotype groups. The high-risk group was compared with the low-risk group.

All ORs and 95%CIs were calculated by cross table. All p values <0.05 were considered significant. Permutation tests were applied for p values of ORs of interaction in cases in order to adjust the statistical α errors. In addition, age adjusted ORs and 95%CI of interaction were calculated by Cochran Mantel–Haenszel cross table. Analyses were performed using SPSS software v.16.0 and SAS software v.9.2.

RESULTS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Significant increases in osteoporosis risk were observed with minor alleles of rs2077647 located in the first exon (TC: OR = 3.27, 95% CI = 1.83–5.85, CC: OR = 2.48, 95% CI = 1.20–5.16) and rs2234693 located in the first intron (TC: OR = 2.99, 95% CI = 1.65–5.44, CC: OR = 2.52, 95% CI = 1.21–5.23) of ESRα (Table 1). Other SNPs in ESRα and ESRβ were not significantly associated with osteoporosis risk. At two risk SNPs of ESRα, diplotypes which consisted of haplotype CC (genetic risk score ≥2) were significantly associated with osteoporosis risk (TT/CC: OR = 3.56, 95% CI = 1.88–6.73, TC/CC, CT/CC and CC/CC: OR = 2.53, 95% CI = 1.22–5.24) (Table 2). Further, the broad category of high-risk diplotypes, that is, haplotype CC was strongly associated with osteoporosis risk (OR = 3.15, 95% CI = 1.83–5.41) (Table 3).

Table 1. Risk of Osteoporosis Associated with the SNPs of ESRα
dbSNPLocation (Amino Acid Position)GenotypeCases (%) n = 114Controls (%) n = 171ORs (95%CIs)p
rs2077647Exon 1 (Ser10Ser)TT22 (19.3%)72 (42.1%)1.00 
 TC70 (61.4%)70 (40.9%)3.27 (1.83–5.85)<0.0001
 CC22 (19.3%)29 (17.0%)2.48 (1.20–5.16)0.015
rs2234693Intron 1TT20 (17.5%)65 (38.0%)1.00 
 TC70 (61.4%)75 (43.9%)3.03 (1.67–5.51)<0.0001
 CC24 (21.1%)31 (18.1%)2.52 (1.21–5.23)0.013
Table 2. Risk of Osteoporosis Associated with the Combination of Haplotypes at rs2077647 and rs2234693
DiplotypeCases (%) n = 114Controls (%) n = 171Genetic Risk ScoreORs (95%CIs)p
  • *

    In diplotype TC/TC and CT/CT, either one of two SNPs was the homozygote of major alleles, though the genetic risk score was 2.

TT/TT18 (15.8%)59 (34.5%)01.00 
TT/TC4 (3.5%)12 (7.0%)   
TT/CT2 (1.8%)5 (2.9%)   
TC/TC0 (0.0%)1 (0.9%)1*1.04 (0.36–2.98)0.95
CT/CT0 (0.0%)1 (0.9%)   
TT/CC63 (44.7%)58 (33.9%)23.56 (1.88–6.73)<0.0001
TC/CC5 (4.9%)7 (4.1%)   
CT/CC3 (2.6%)5 (2.9%)≥32.53 (1.22–5.24)0.01
CC/CC19 (16.7%)23 (13.5%)   
Table 3. Risk of Osteoporosis Associated with Low- Versus High-Risk Diplotypes
DiplotypeCases (%) n = 114Controls (%) n = 171OR (95%CI)p
  • *

    Haplotype was CC and genetic risk score was ≥2.

Low risk diplotype (TT/TT, TT/TC, TT/CT, TC/TC and CT/CT)24 (21.1%)78 (45.6%)1.00 
High risk diplotype * (TT/CC, TC/CC, CT/CC and CC/CC)90 (78.9%)93 (54.4%)3.15 (1.83–5.41)<0.0001

Characteristics of osteoporotic women are shown in Table 4. There was a statistically significant interaction between high-risk diplotype and alcohol drinking (Table 5). Specifically, the risk for osteoporosis with high-risk diplotype decreased to about 1/5 by the consumption of alcohol (OR = 0.22, 95% CI = 0.05–0.83). There was no interaction in other lifestyle factors. The confounding factors to alcohol drinking were not admitted because there was no association between alcohol drinking and other environmental factors.

Table 4. Characteristics of Osteoporotic Post-Menopausal Women
Characteristicsn = 114
Age at diagnosis
 45–494
 50–5919
 60–6942
 70–8649
BMI (kg/m2) at age 20's
 <18.521
 ≥18.582
 Missing11
BMI (kg/m2) at diagnosis
 <18.512
 ≥18.565
 Missing37
Smoking at diagnosis
 No (never)91
 Yes (ever or former)16
 Missing7
Alcohol drinking at diagnosis
 No (never)97
 Yes (every day and several times a week or month)10
 Missing7
Age at menopause
 ≤4523
 >4580
 Missing11
Number of births
 ≤139
 ≥268
 Missing7
Physical activity in exercise at age 10's
 <3 METs/d85
 ≥3 METs/d21
 Missing8
Milk intake at age 10's
 No (several times a month or never)73
 Yes (every day and several times a week)34
 Missing7
Dairy products intake at age 10's
 No (several times a month or never)81
 Yes (every day and several times a week)26
 Missing7
Soybean products intake at age 10's
 No (several times a month or never)74
 Yes (every day and several times a week)33
 Missing7
Table 5. Interaction of Environmental Factors and Low- Versus High-Risk Diplotypes in Predicting Osteoporosis Risk among Osteoporotic Post-Menopausal Women
Environmental FactorsLow Risk DiplotypeHigh Risk DiplotypeORs (95%CIs)Age Adjusted ORs (95%CIs)p*
  • *

    p values of ORs were adjusted by permutation test.

BMI (kg/m2) at age 20's
 <18.56151.001.00 
 ≥18.515670.56 (0.19–1.68)0.59 (0.19–1.77)0.37
BMI (kg/m2) at diagnosis
 <18.5481.001.00 
 ≥18.511540.41 (0.10–1.59)0.35 (0.09–1.46)0.23
Smoking at diagnosis
 No19721.001.00 
 Yes5180.95 (0.31–2.89)0.94 (0.31–2.88)1.0
Alcohol drinking at diagnosis
 No17801.001.00 
 Yes550.21 (0.06–0.82)0.22 (0.05–0.83)0.03
Age at menopause
 ≤455181.001.00 
 >4517631.01 (0.33–0.32)1.01 (0.34–3.21)1.0
Number of births
 ≤19301.001.00 
 ≥214541.12 (0.45–2.99)1.15 (0.45–2.98)1.0
Physical activity in exercise at age 10's
 <3 METs/d19631.001.00 
 ≥3 METs/d3212.11 (0.57–7.86)2.42 (0.63–9.35)0.37
Milk intake at age 10's
 No14591.001.00 
 Yes8260.77 (0.29–2.06)0.78 (0.29–2.10)0.62
Dairy products intake at age 10's
 No16651.001.00 
 Yes6200.82 (0.25–2.38)0.81 (0.28–2.36)0.77
Soybean products intake at age 10's
 No17571.001.00 
 Yes5281.67 (0.56–4.99)1.62 (0.54–4.87)0.45

DISCUSSION

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Estrogens maintain the balance between resorption and formation of bone, and thus influence bone turnover in postmenopausal women. Estrogens exert pro-apoptotic effects on osteoclasts but anti-apoptotic effects on osteoblasts of mature bone cells. Therefore, loss of estrogen leads to increased rate of remodeling and tilts the balance between bone resorption and formation in favor of the former.18 Estrogens exert their anti-apoptotic signal through binding and activation of ESRs (ESRα and ESRβ). Genetic polymorphisms in these receptors play a major role in the etiology of osteoporosis risk. Specifically, ESRα is an important functional candidate for the regulation of bone mass.4 TA repeat upstream from the first exon19–21 and two SNPs (rs9340799 XbaI and rs2234693 PvuII)12, 22 in the first intron are associated with BMD, though the evidence has not been consistent. Of note, rs9340799 and rs2234693 are in linkage disequilibrium in Caucasian and Japanese.23 Interestingly, rs2077647 of the first exon analyzed in this study was at a position near the rs9340799 and rs2234693 block. The linkage disequilibrium coefficient (D′) of rs2077647 and rs2234693 among Japanese is high (D′ = 0.946) by HapMap Project data and haplotype CC in these SNPs are 57.8%. Haplotype CC is observed in 54.4% of the healthy control subjects in this study, which corresponds to the percentage of that in the HapMap project data. Since osteoporosis is a common disease in women, the rate of subjects having risk alleles at both of the two SNPs is high. There may be a potential misclassification bias due to the younger age distribution in the control group compared to the relatively older age distribution in the case group. However the bias should be toward to the Null, which we under-estimated the osteoporosis risk effect due to the ESRα polymorphisms.

Risk factors for a decrease in BMD that are difficult to modify include female gender, menopause, family history, and low body weight.24 There is good evidence for female gender, menopause and low body weight as risk factors for BMD. Many women in this study reported experiencing an early menopause, which is of concern given that early menopause (<45 years old) is a significant risk factor for osteoporosis.25, 26 Due to a significant amount of missing data, the rate of low body weight (BMI < 18.5) at diagnosis was difficult to ascertain. Although Waugh et al. reported inconsistent evidence, a meta-analysis conducted by Kanis et al.27 found that a parental history of fracture conferred an increased risk of osteoporotic fracture. A lack of dietary calcium and vitamin D, an inactive lifestyle, smoking, and excess alcohol consumption have all been suggested to increase risk of osteoporosis.24 However, the data supporting many of these factors is inconsistent.24

There is reasonable evidence that moderate alcohol consumption (<150 g/week or <12 drinks/week) is not associated with lower BMD24 and insufficient evidence for an association between past or high alcohol consumption and BMD.24 However, recent studies report that moderate alcohol consumption may protect BMD.28–30 This protective effect has been hypothesized to be mediated by effects of alcohol on adrenal androgens or estrogen concentrations. The effect may be stronger among postmenopausal women than in premenopausal women as the low estrogen concentrations are more readily increased.31 Resveratrol found in wine can play an estrogenic role in protecting bone loss in postmenopausal women, while silicon found in beer may promote bone formation.32, 33 Although it was uncertain in this study design whether alcohol drinking decreased the risk of osteoporosis because alcohol consumption in controls was not investigated, this study suggested that alcohol drinking might decrease the genetic risk of osteoporosis to about 1/5 (OR = 0.22), though the amount and type of alcohol were unclear.

In this study, the osteoporotic risk of genetic factors was assessed using a case–control study while the interaction between a genotype and environmental factors was examined among case subjects only. The case only study is used when both genotype and environmental exposure data from cases but only genotype or only environment data from the controls are collected. This is done, for example, when it is difficult to select normal matched-age controls given that osteoporosis is so prevalent. Although the case only study is a powerful design to assess gene–environment interaction, it requires that the gene and environment factors occur independently, such as the association between SNPs in ESRα and alcohol drinking. Environment factors do not influence SNPs and the possibility that SNPs in ESRα influence the drinking behavior may be extremely low. Under this condition the OR of the interaction is the effect as measured in a traditional case–control study under a multiplicative model.12–15

In conclusion, our findings suggest that ESRα may be associated with osteoporosis risk and that moderate alcohol drinking may decrease this genetic predisposition. However, our study included some limitations that warrant mention. First, due to the retrospective design and the use of a self-administered questionnaire, recall bias, and nonresponse were both possible. Secondly, since few of the osteoporotic women were alcohol drinkers, it may be necessary to verify our findings using larger epidemiological studies. Third, the genetic risk of osteoporosis in a case–control study may not be accurate, because controls were not matched to cases for age and osteoporosis.

Acknowledgements

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

This work was supported by grants from the Sapporo Medical University.

REFERENCES

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES