The Protective Effect of Hormone-Replacement Therapy on Fracture Risk Is Modulated by Estrogen Receptor α Genotype in Early Postmenopausal Women


  • Presented in part as an abstract at the 21st Annual Meeting of the American Society for Bone and Mineral Research, St. Louis, Missouri, U.S.A., 1999


Genetic factors regulate bone mineral density (BMD) and possibly development of osteoporosis. It has been suggested that estrogen receptor α (ERα) genotype is associated with BMD, but the association between ERα genotype, fracture risk, and postmenopausal hormone replacement therapy (HRT) has not been studied. Therefore, we evaluated whether ERα polymorphism is associated with fracture risk in a 5-year trial with HRT in a population-based, randomized group of 331 early postmenopausal women. The participants consisted of two treatment groups: the HRT group (n = 151) received a sequential combination of 2 mg of estradiol valerate (E2Val) and 1 mg of cyproterone acetate with or without vitamin D3, 100-300 IU + 93 mg calcium as lactate per day; and the non-HRT group (n = 180) received 93 mg of calcium alone or in combination with vitamin D3, 100-300 IU/day. All new symptomatic, radiographically defined fractures were recorded. Pvu II restriction fragment length polymorphism of the ERα was determined using polymerase chain reaction (PCR). In all, 28 women sustained 33 fractures during the approximately 5.1-year follow-up. In the HRT group, the ERα genotype (PP, Pp, and pp) was not significantly associated with fracture risk (p = 0.138; Cox proportional hazards model). When the genotype was dichotomized (PP + Pp vs. pp), the incidence of new fractures in the HRT group was significantly reduced in women with the P allele (p = 0.046) with the relative risk (HR) of 0.25 (95% CI, 0.07-0.98), in comparison with the non-P allele group. After adjustment for time since menopause and previous fracture, the association between the dichotomous genotype and fracture risk persisted with HR of 0.24 (95% CI, 0.06-0.95; p = 0.042). In the non-HRT group, the ERα genotype was not significantly associated with fracture risk. During HRT, women with the pp genotype have a greater fracture risk than those with the P allele. The results suggest that the pp genotype is a relatively hormone-insensitive genotype, and it appears that women with the P allele may benefit more from the protective effect of HRT on fracture risk than women with the pp genotype.


OSTEOPOROSIS IS a common disease characterized by reduced bone mineral density (BMD), deterioration of the skeletal microarchitecture, and increased fracture risk. Twin studies suggest that up to 80% of BMD may be heritable,(1,2) and that BMD is under multigene control.(3) Several candidate gene polymorphisms have been studied with increasing interest. The effects of vitamin D receptor (VDR), estrogen receptor α (ERα), type I collagen (COLIA1), apolipoprotein E (apoE), and calcitonin receptor (CTR) gene polymorphisms on BMD have been investigated in previous studies. (4–16) However, the association between genetic factors and fracture risk is less understood. Results of studies on VDR gene polymorphism and the risk of vertebral fractures are controversial.(17,18) Associations between fracture risk and COLIA1,(13,14) apoE,(19) and CTR(16) gene polymorphisms have been suggested. An influence of ERα genotype on fracture risk has not been previously reported.

Hormone replacement therapy (HRT) can prevent bone loss in estrogen-deficient women. (20–22) Additionally, HRT may prevent fractures among postmenopausal women.(23,24) Among postmenopausal women, there are those who can be classified as fast and slow bone losers,(25,26) and ERα gene polymorphism at least in part may explain the differences in bone loss rates.(9,11) However, the influence of genetic factors on the response of bone to HRT is still relatively poorly understood.

To determine whether ERα gene polymorphism predicts fracture risk with and without HRT, the influence of the ERα genotype on fracture risk was evaluated in a 5-year trial with HRT in a placebo-controlled, population-based, randomized group of nonosteoporotic early postmenopausal women.



The population of the present study was a subgroup of the Kuopio Osteoporosis Risk Factor and Prevention (OSTPRE) Study. In 1989, a postal inquiry was sent to all 14,200 women aged 47-56 years of the Kuopio Province, Eastern Finland, to investigate risk factors for osteoporosis among perimenopausal women.(27) A total of 13,100 women responded. A stratified sample of 3222 women underwent densitometry in 1989-1991. All 464 postmenopausal volunteers out of those 3222 women who had had their last menstrual period within 6-24 months before the study and who were not osteoporotic were selected for the 5-year clinical trial. Exclusion criteria were restricted to general contraindications for HRT including history of estrogen-dependent cancer, thromboembolic diseases, and medication-resistant hypertension. These women were randomized to four treatment groups: an estradiol valerate/cyproterone acetate (E2Val/CPA) group, a vitamin D3 + calcium lactate group, an E2Val/CPA + vitamin D3 + calcium lactate group, and a calcium lactate group. Random allocation to study groups was carried out blockwise with a computer—the block size being 4, 8, or 12. The personnel involved were unaware of the group allocation. However, after randomization, the study was open for all treatment groups. The study design has been described elsewhere in detail.(22) As reported recently, the low-dose vitamin D supplementation used alone or in combination with HRT did not influence BMD.(22, 28, 29) Therefore, to increase the power of this study, the two hormone groups and the two nonhormone groups were combined. The following were the final study groups.

HRT group (n = 232):

Sequential combination of 2 mg E2Val (days 1-21) and 1 mg CPA (days 12-21; Climen; Schering AG, Berlin, Germany), with a treatment-free interval (days 22-28) alone or in combination with vitamin D3, 100-300 IU + calcium lactate (D-Calsor; Orion Corp., Espoo, Finland), 500 mg/day, and equivalent to 93 mg Ca2+.

Non-HRT group (n = 232):

Calcium lactate (Calcium Lactate; Rohto Ltd., Tampere, Finland), 500 mg/day, equivalent to 93 mg Ca2+ alone or in combination with vitamin D3, 100-300 IU/day. Study design and formation of the present study population are depicted by the flow diagram (Fig. 1). Written informed consents were obtained from the participants and the study design was approved by the Ethics Committee of the Kuopio University Hospital.

Figure FIG. 1.

Trial profile.

A total of 331 women (71.3%) were included in the final analysis. During the 5-year follow-up period, 96 (20.7%) women out of the 464 dropped out. Prospectively defined stopping rules were the same as the exclusion criteria. Because the study was designed to examine the preventive effect of the HRT on osteoporosis, 6 osteoporotic women were excluded at baseline and considered as dropouts. Additionally, 37 women (8.0%) were excluded because of missing laboratory data, poor quality of BMD scans, or the presence of measurement artifacts (e.g., severe spondylarthrosis).

Analysis of the ER gene polymorphism

Genomic DNA was extracted and purified from EDTA blood samples using QIAamp Blood Kit (Qiagen GmbH, Hilden, Germany). Genomic DNA was amplified in 20 μl of a buffer solution: 10 mM Tris-HCl, pH 8.8, 50 mM KCl, 1.5 mM MgCl2, 0.08% NP-40, 200 μM each of the four deoxyribonucleotides, 1.2 U of Taq polymerase (Fermentas, Vilnius, Lithuania), and 1.0 μM of each oligonucleotide primer (forward, 5′-CTGCCACCCTATCTGTATCTTTTCCTATTCACC-3′ and reverse, 5′-TCTTTCTCTGCCACCCTGGCGTCGATTATCTGA-3′). Polymerase chain reaction (PCR) was performed with the following steps: 94°C, 5 minutes; and then 94°C, 1 minute; 60°C, 1 minute; and 72°C, 1 minute for 30 cycles; and a final extension 72°C, 7 minutes (Biometra, Göttingen, Germany). The product contains parts of intron 1 and exon 2 of the ERα gene. After amplification, the PCR product was digested with Pvu II restriction endonuclease (Fermentas) and electrophoresed in 2.0% agarose gel. The polymorphic Pvu II site(30) was detected in this 1.3-kilobase (kb) fragment.


BMD of the lumbar spine (L2-L4) and the left proximal femur (femoral neck) was measured by trained personnel using dual X-ray absorptiometry (DXA; Lunar, Madison, WI, U.S.A.) at Kuopio University Hospital before and after 5 years of treatment, as described previously.(11) Body weight and height were measured in the connection of each BMD measurement.

During the follow-up, each subject visited the outpatient clinic once a year. Fasting venous blood samples were obtained, and information on compliance and occurrence of new fractures was collected simultaneously. No attempt was made to exclude high-energetic fractures. However, there were no fractures caused by car accidents or other major trauma. Reported symptomatic fractures were validated by medical and radiographic reports. At baseline, daily calcium intake was calculated as the sum of calcium intake from milk, sour milk, yogurt (120 mg/dl), and cheese (87 mg/slice). The weekly duration of physical activity and the smoking and drinking habits were recorded. The baseline concentrations of serum follicle-stimulating hormone (FSH), estradiol (E2), calcium (Ca), and phosphate (Pi) were assayed by standard laboratory methods used at the Kuopio University Hospital. All blood samples were taken in the morning after an overnight fast.

Statistical analysis

Statistical analyses were carried out using SPSS for Windows statistical package (SPSS Inc., Chicago, IL U.S.A.). The analyses were limited to fractures that had occurred after the trial medication was initiated. The value of p < 0.05 was considered statistically significant. Data analysis was performed on valid cases, that is, women who had complied with the trial medication for at least 80% of the follow-up time (at least 4 years).

One-way analysis of variance was used to test the significance of differences in continuous variables between the ERα genotype groups at baseline. Most baseline parameters were not distributed normally. Therefore, the nonparametric Kruskal-Wallis test was used. The χ2 test was used to compare the proportions of smokers, physically active persons, and women with previous fracture in genotype groups. Fracture data were analyzed with the Cox proportional hazards model. In the analysis, the period of time to the first fracture was used as the dependent variable. The results were evaluated using unadjusted analysis as well as after adjustment for previous fracture, time since menopause, spinal and femoral baseline BMD, 1-year E2 change, smoking, height, and weight.


Characteristics of the study population

The allelic frequencies of ERα gene (42% for the P allele and 58% for the p allele) and the distribution of the genotypes were similar to those reported earlier in different populations.(7,8) The most common ERα genotype was Pp (n = 164, 49.5%). Genotype pp was more frequent (n = 110, 33.2%) than PP (n = 57, 17.2%; Table 1).

Table Table 1.. ER Genotype and Baseline Characteristics of 331 Postmenopausal Women According to HRT Use and Genotypea
original image

At baseline, no significant differences in lumbar or femoral neck BMDs between the genotypes and HRT use were observed (Table 1). There were significant differences between women with different ERα genotypes and HRT used in the variables FSH, E2, 1-year FSH change, and 1-year E2 change. The concentrations of serum calcium (Ca) and phosphorus (Pi) were similar between the groups (data not shown). During the 5-year follow-up, the relative weight increases were similar (1.9-4.8%) between the different treatment groups and the ERα genotypes (data not shown).

Association between the genotype and fractures

In all, 33 symptomatic fractures were sustained by 28 women during the average 5.1-year follow-up period. In the HRT group, 8 women reported 1 fracture and 2 women reported 2 fractures. In the non-HRT group, 15 women reported 1 fracture and 3 women reported 2 fractures (Table 2).

Table Table 2.. New Fractures and Site of Fracture in Relation to HRT Use and ERα Genotypea
original image

In the HRT group, the Cox proportional hazards model showed that the trichotomous ERα genotype (PP, Pp, and pp) did not significantly predict fracture risk (p = 0.138; Table 3). When the genotype was dichotomized (PP + Pp vs. pp), the reduction in the incidence of new fractures was significant in women with the P allele (p = 0.046; Fig. 2). The estimated relative risk of new fractures among women with the P allele was 0.25 (95% CI, 0.07-0.98) compared with the non-P allele group (Table 3). The risk of fracture did not change after adjustment for potentially confounding factors such as previous fracture, time since menopause, spinal and femoral baseline BMD, 1-year E2 change, smoking, height, and weight in the Cox proportional hazards model (Table 3).

Figure FIG. 2.

Cumulative fracture-free survival as a function of the period of time to the first fracture. In the HRT group, the reduction in fracture risk was significant in women with the P allele (p = 0.046). In the non-HRT group, the ERα genotype was not significantly associated with fracture risk (p = 0.395, Cox proportional hazards model).

Table Table 3.. Fracture Risk Values and the Estimated Risk for New Fractures According to ERα Genotype in the HRT Group
original image

In the non-HRT group, the trichotomous genotype variable did not significantly predict fracture risk (p = 0.456). The relative risk of fracture among women with the PP genotype was 0.27 (95% CI, 0.03-2.16) and among those with the Pp genotype was 0.78 (0.30-2.05) compared with the pp genotype (p = 0.215 and 0.613, respectively). Similarly, the dichotomous genotype was not significantly associated with fracture risk (p = 0.395) with a relative risk of 0.66 (95% CI, 0.26-1.71) in the P allele group compared with the non-P allele group.


The results of our longitudinal study show for the first time that the ERα gene allelic variation is associated with fracture risk in nonosteoporotic early postmenopausal white women who use HRT. During the 5-year follow-up, the reduction in the incidence of new fractures was significant in women with the P allele, in comparison with women without the P allele (pp genotype). In the non-HRT group, the ERα genotype was not significantly associated with fracture risk.

Studies on transgenic ERα knockout mice have shown that both female and male knockout mice have 20-25% lower BMD than wild-type mice.(31) Also, a man with a loss-of-function mutation of the ER gene has shown severe osteoporosis and incomplete epiphyseal closure.(32) However, these studies do not imply that other mutations may cause mild estrogen resistance; they only show that the absence of a functional ER has an effect. Results of studies concerning the relationship between ERα and BMD changes in postmenopausal women have not been consistent.(9, 11, 33, 34) We have previously reported that ERα gene polymorphism (Pvu II) is associated with the magnitude of long-term lumbar BMD changes in this same study population.(11) During the 5-year follow-up, women with the P allele lost significantly more bone than women without the P allele.(11) Recently, the effect of COLIA1 gene polymorphism on fracture risk has been studied(13,14) and the results suggest that this polymorphism is associated with BMD and fracture risk. However, there have not been previous studies concerning ERα genotype and fracture risk. In addition, there have not been studies concerning ERα gene polymorphism and the HRT effect on fracture risk. The positive influence of postmenopausal HRT on BMD is well established and its antifracture effect is widely accepted even though it is based mostly on observational studies. There have been only two randomized trials showing the positive effect of HRT on fracture risk.(23,24) Our 5-year trial showed that HRT prevents nonvertebral fractures in this population.(24) However, it seems that densitometric nonresponders to HRT exist.(35,36) The results of this study, showing that the ER genotype may modify the HRT effect on fracture risk, may give an explanation of why some women do not have favorable bone response to HRT.

The mechanisms by which the different ERα gene alleles affect bone loss rate and fracture risk are not known. The ERα gene has been localized to chromosome 6q25.1 by Menasce and coworkers.(37) The other genes within the same locus do not have any known effects on bone metabolism. However, the genes for bone morphogenic protein 5 and 6 are located in chromosome 6.(38) According to our previous studies, women with the P allele of ERα gene seem to undergo fast bone loss after menopause and long-term HRT seems to prevent development of the ERα genotype-related differences in the BMD.(11) It is not known whether this effect is mediated through BMD gain only or whether there are other tissues that ERα gene polymorphism affects, such as the central nervous system and the muscle tissue, in which it could affect fall risk and reduction of muscle strength. The number of fractures in our study was too low to examine the influence of BMD, bone markers, or muscle strength on fracture risk in different ER genotype groups.

At baseline, serum FSH and E2 concentrations were associated with ERα genotype and HRT use. Especially in the non-HRT group, women with the pp genotype had the longest time since menopause, highest serum FSH, and lowest serum E2 values. The situation with the PP genotype was vice versa. The reason why pp genotype had the lowest serum E2 values may be related to the length of time since menopause. The level of serum E2 values drops dramatically after menopause and simultaneously the level of serum FSH starts to rise. The statistical difference in the 1-year FSH and E2 changes is caused by HRT use, and with annual FSH measurements, compliance to HRT can be estimated. In the HRT group, the annual FSH levels and the FSH changes during the 5-year follow-up were similar between those women who sustained a fracture compared with those who did not, showing that compliance was good (data not shown).

As we have reported previously, there were differences in the HRT group in the baseline lumbar BMD between the ERα genotypes, but the significance (p = 0.039) disappeared when the result was adjusted for the variable time since menopause (p = 0.052).(11) This baseline difference in the lumbar BMD, which persisted through the follow-up, may in part explain the genotype-related differences in the fracture risk.(11) Thus, these ERα genotype-related differences in the fracture risk appear to be only partly dependent on BMD. It has been previously reported that hip BMD, heel broadband ultrasound attenuation (BUA), and bone resorption markers are independent predictors of hip fracture in elderly women.(39,40) Thus this genotype-related difference in fracture risk may, in addition, also be caused by increased bone resorption, different architectural characteristics of the trabecular bone, fall risk, decreased muscle strength, or a mixture of those. In a recent study by Han et al., association between ERα genotypes (Pvu II and Xba I restriction endonucleases used) and bone turnover markers in postmenopausal women was not observed.(10)

These results mostly reflect the effect of HRT, but it is still possible that vitamin D has a minor independent effect, although the used low-dose vitamin D supplementation alone or in combination with HRT does not influence BMD in this study population.(22, 28, 29) In a 1-year substudy of this trial, vitamin D supplementation increased circulating 25-hydroxyvitamin D [25(OH)D] concentrations significantly, but did not affect 1,25-dihydroxyvitamin D [1,25-(OH)2D] levels.(41) It is therefore unlikely that the low-dose vitamin D supplementation was effective in the prevention of bone loss and fractures, because our study subjects may not have been deficient in vitamin D.

This study has the advantage of being a long-term, population-based trial with a homogenous group of nonosteoporotic early postmenopausal women with a good compliance rate compared with other HRT trials. In our previous report, we concluded that women with the P allele will derive more benefit from long-term HRT than women without the P allele.(11) However, in the present study, the numbers of fractures were relatively small, probably because the subjects were relatively young and nonosteoporotic. In the non-HRT group, the statistical power of our study was not strong enough to show the possible influence of the ERα genotype on fracture risk during the natural menopause.

To our knowledge, this is the first clinical trial investigating the association between the ERα gene polymorphism and the protective effect of HRT on fracture risk. A new and clinically important finding was that women with the P allele appear to benefit from the fracture preventive effect of HRT, whereas women without the P allele do not seem to gain similarly from HRT.


This work was supported in part by the Finnish Graduate School in Musculo-Skeletal Problems (TULES), the Academy of Finland, the Kuopio University Hospital, and Schering AG.