Associations Between Baseline Risk Factors and Vertebral Fracture Risk in the Multiple Outcomes of Raloxifene Evaluation (MORE) Study

Authors

  • Olof Johnell,

    1. Department of Orthopaedics, Malmö General Hospital, Malmö, Sweden
    Search for more papers by this author
  • John A Kanis MD, PhD,

    Corresponding author
    1. WHO Centre for Metabolic Bone Diseases, University of Sheffield, Sheffield, United Kingdom
    • WHO Collaborating Centre for Metabolic Bone Diseases, University of Sheffield Medical School, Beech Hill Road, Sheffield S10 2RX, UK
    Search for more papers by this author
  • Dennis M Black,

    1. Division of Clinical Epidemiology, University of California, San Francisco, California, USA
    Search for more papers by this author
    • Drs Pavo, Zhou, and Sarkar received corporate appointments and stock from Eli Lilly and Company. Dr Black serves as a speaker for and receives research contracts from Merck, Novartis, and NPS Pharmaceuticals. Dr Kanis served as a consultant for Eli Lilly and Company. All other authors have no conflict of interest.

  • Adam Balogh,

    1. Department of Obstetrics and Gynecology, Medical and Health Science Center, University of Debrecen, Debrecen, Hungary
    Search for more papers by this author
  • Gyula Poor,

    1. National Institute of Rheumatology and Physiotherapy, Budapest, Hungary
    Search for more papers by this author
  • Somnath Sarkar,

    1. Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA
    Search for more papers by this author
    • Drs Pavo, Zhou, and Sarkar received corporate appointments and stock from Eli Lilly and Company. Dr Black serves as a speaker for and receives research contracts from Merck, Novartis, and NPS Pharmaceuticals. Dr Kanis served as a consultant for Eli Lilly and Company. All other authors have no conflict of interest.

  • Chunmei Zhou,

    1. Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA
    Search for more papers by this author
    • Drs Pavo, Zhou, and Sarkar received corporate appointments and stock from Eli Lilly and Company. Dr Black serves as a speaker for and receives research contracts from Merck, Novartis, and NPS Pharmaceuticals. Dr Kanis served as a consultant for Eli Lilly and Company. All other authors have no conflict of interest.

  • Imre Pavo

    1. Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA
    Search for more papers by this author
    • Drs Pavo, Zhou, and Sarkar received corporate appointments and stock from Eli Lilly and Company. Dr Black serves as a speaker for and receives research contracts from Merck, Novartis, and NPS Pharmaceuticals. Dr Kanis served as a consultant for Eli Lilly and Company. All other authors have no conflict of interest.


Abstract

Different risk factors may influence the effectiveness of osteoporosis therapies. The interaction of 30 baseline risk factors and the effectiveness of raloxifene in the MORE study were assessed. The efficacy of raloxifene in reducing vertebral fractures is largely independent of the presence of clinical risk factors for osteoporotic fractures.

Introduction: The aim of this analysis was to determine the effect of different risk factors on the effectiveness of raloxifene to reduce vertebral fractures in the Multiple Outcomes of Raloxifene Evaluation (MORE) study using logistic regression models.

Materials and Methods: The association was assessed using univariate analyses and a multivariate model between 30 potential risk factors at baseline and the risk of vertebral fractures after 3 years in the placebo group, as well as the interaction of risk factors with raloxifene therapy (at a dose of 60 or 120 mg/day).

Results and Conclusions: In the univariate analysis of the placebo group, after adjusting for baseline lumbar spine BMD (LS BMD), short stature (odds ratio [OR] = 1.18), age (OR = 1.38), years since menopause (OR = 1.38), impaired cognitive function, visuospatial capabilities (OR = 1.19), impaired musculoskeletal strength (OR = 1.23), low femoral neck BMD (OR = 1.21), and prior vertebral fracture (OR = 4.95) were significantly associated with the incidence of new vertebral fractures. In the univariate analysis, significant interactions were observed between raloxifene treatment and age (p = 0.04), serum triglycerides (p = 0.03), LS BMD (p = 0.08), and diabetes mellitus (p = 0.04). In the multivariate analysis, the effectiveness of raloxifene was independent of almost all risk factors, with the exception of baseline serum triglyceride level and LS BMD, suggesting an increased efficacy of raloxifene in patients with increased triglyceride levels (p = 0.006) and lower LS BMD values (p = 0.008) at baseline. These data suggest that the efficacy of raloxifene in reducing vertebral fractures is largely independent of the presence of clinical risk factors for osteoporotic fractures.

INTRODUCTION

Measurement of BMD is the cornerstone for the diagnosis of postmenopausal osteoporosis and consequently provides information of value for making decisions about medical treatment. Low bone mass is a strong predictor of fracture risk, although other skeletal and nonskeletal factors also contribute. Some of these risk factors affect fracture probability independently of BMD, such as age, previous fractures, glucocorticoid therapy, high bone turnover, a family history of hip fractures, poor visual capacity, low body weight, neuromuscular disorders, and smoking.(1) Acknowledging the contribution of some of these independent factors, the National Osteoporosis Foundation (NOF) guideline for the treatment of osteoporosis considers risk factors in addition to BMD for the setting of intervention thresholds.(2) In addition, the International Osteoporosis Foundation (IOF) recommends identifying individuals for testing with BMD measurements based on strong risk factors, such as prior fragility fractures, corticosteroid use, and low body mass index (BMI).(3) The use of risk factors that are independent of BMD improves the sensitivity of fracture risk assessment without loss of specificity.(4) In the future, therefore, it is likely that the clinician will more often initiate anti-osteoporotic medical treatment based on a combination of low BMD and the presence of other risk factors rather than on BMD alone.

There are now many agents available for the treatment of osteoporosis that have been shown to decrease the risk of fractures.(5) The majority of phase III studies have examined effects on vertebral fracture risk in patients selected based on osteoporosis or low bone mass. If patients are also to be selected based on risk factors other than BMD, it is important to determine whether the presence of such factors influences the responsivity to therapeutic intervention.

One of the antiresorptive agents used for the prevention and treatment of postmenopausal osteoporosis is raloxifene. Raloxifene is a selective estrogen receptor modulator (SERM) with estrogen agonist effects on the skeleton. In the Multiple Outcomes on Raloxifene Efficacy (MORE) study, raloxifene increased BMD and reduced bone turnover and the risk of vertebral fracture in postmenopausal women with osteoporosis.(6) In this study, a variety of clinical parameters, including potential risk factors for osteoporotic fractures, were recorded. The aim of this study was to explore the relationship between the antifracture effect of raloxifene and potential clinical risk factors for osteoporotic fractures. As a separate part of the analysis, we also report the effect of these potential risk factors on the vertebral fracture risk in the placebo group of the MORE population.

MATERIALS AND METHODS

Patients

The study design, complete inclusion and exclusion criteria, and the analysis of the 3-year vertebral fracture data (the primary endpoint) of the MORE trial have been described previously.(6) Briefly, the study included 7705 women who were at least 2 years postmenopausal and had osteoporosis, defined as low BMD or a radiographically apparent vertebral fracture. The women were randomized to receive placebo or raloxifene HCl at a dose of 60 or 120 mg/day. In addition, all women received daily supplements of 400-600 IU cholecalciferol (vitamin D3) and 500 mg calcium. Patients receiving the two different doses of raloxifene were pooled for the analysis to increase the sample size, because the outcome of interest (vertebral fracture risk) did not differ significantly between the two groups.(6)

Baseline characteristics

Variables recorded at baseline and analyzed for association with fracture risk included anthropometric, patient history and lifestyle-related factors, hormonal, biochemical measurements, and neurocognitive function. BMD and vertebral fracture assessment were also performed before treatment.

Anthropometric parameters recorded included height, weight, BMI, age, and years since menopause. At the baseline visit, patients completed a questionnaire regarding their medical history and lifestyle habits, including a family history of osteoporosis (whether or not the patient's mother, sister, or daughters were known to have osteoporosis), previous hormone replacement therapy (HRT; ever used HRT: yes or no), hysterectomy, current smoking status (ever smoked: yes or no), current alcohol use (more than three drinks of alcohol per week: yes or no), presence of diabetes (as indicated by a baseline fasting blood glucose >7.8 mM or a pre-existing condition of diabetes and baseline use of a hypoglycemic agent: patients with type 1 diabetes mellitus were excluded from the study), history of falls and near falls (falls or near falls to the ground in the past year: yes or no), and geographic location during the study. Patients in participating countries were pooled into geographic categories: United States, Scandinavia, other parts of Europe, and other (Latin America and Asia). Serum levels of parathyroid hormone (PTH), estradiol (estradiol level < or ≥ 5 pM), 25-hydroxyvitamin D (calcidiol), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglycerides were also determined in all patients. Markers of bone turnover, including serum bone-specific alkaline phosphatase (bone ALP; Tandem-R Ostase; Hybritech, San Diego, CA, USA),(7) osteocalcin (ELSAOSTEO; CIS Biointernational, Gif sur Yvette, France),(8) and urinary type I collagen C-telopeptide excretion, corrected for urinary creatinine excretion (U-CTX; CrossLaps; Osteometer A/S, Herlev, Denmark),(9) were measured in a subset of 2443 women.

Tests of cognitive function, similar to those used by the Consortium to Establish a Registry for Alzheimer's Disease,(10) were administered in a standard order by trained study personnel as reported elsewhere.(11) The Short Blessed Test(12) assesses orientation, concentration, and memory. The Trail Making Test (Part A) measures visuospatial scanning, sequential processing, motor speed, executive function, and attention.(13) This test involves connecting a series of consecutively numbered circles; shorter times to completion (in seconds) indicate better performance. The Word List Fluency Test(14) measures verbal production, semantic memory, and language. In this test, patients name as many animals as possible in 60 s, and higher scores indicate better performance. Balance Score and Standing Time Test (the time needed to stand and sit five times) measure musculoskeletal function.(15)

Spine and femoral neck BMD were measured at baseline by DXA. A central reading facility provided correction factors to adjust for intersite differences.(16)

Participants underwent vertebral radiography at baseline and at 24 and 36 months. Vertebral fractures were analyzed using a combination of semiquantitative (SQ) readings and quantitative morphometric analysis.(16,17) The criterion for diagnosis of a new vertebral fracture was based on a reduction in the anterior, middle, and/or posterior vertebral height of at least 20% and at least 4 mm compared with the baseline radiograph. An incident fracture was defined as a grade change of at least 1.

Statistical analysis

The baseline variables for the placebo and raloxifene (patients treated with either 60 or 120 mg/day raloxifene) groups were compared using ANOVA for continuous variables and Pearson's χ2 test for categorical data.

Logistic regression analysis was used for modeling, with the presence of any incident vertebral fractures recorded as a binomial (yes/no) response. The relationship between each baseline factor and the risk of having at least one new vertebral fracture over 3 years was first evaluated for placebo patients. The results were subsequently adjusted for baseline lumbar spine (LS) BMD (primary analysis) or age. Odds ratios (ORs) for continuous variables are given as the gradient of risk for a 1 SD increase.

To identify individual factors that might influence raloxifene efficacy, logistic regression models were fitted, with treatment (raloxifene or placebo), individual risk factors (covariates), and the treatment-by-covariate interaction as fixed effects. A significant interaction indicates that the treatment effect is dependent on the covariate, whereas a nonsignificant interaction suggests that the treatment effect on the new fracture risk is not influenced by the covariate. The interaction was tested at the 10% significance level.

The interaction identified by the single covariate and treatment model did not account for the influence of other important predictors and their interaction with treatment. Thus, the next step was to identify all variables having an association with new vertebral fractures and to evaluate their interaction with treatment. Because data on bone ALP, osteocalcin, and U-CTX were available for only ∼35% of patients, these variables were excluded from multivariate model development.

To develop the single model incorporating all of the main effect variables, a step-wise model-building strategy was used to select a set of main effect covariates associated with vertebral fracture risk among all study patients. A significance level of 20% was used as the criterion for inclusion of a variable in the model. After the main effect covariates had been identified, pair-wise interactions between these covariates were selected using the step-wise method, with a significance level of 10%. This phase constituted development of the preliminary final model.

For validation purposes, a backward elimination method was used for selection of the final multivariate model. This phase confirmed the model derived from the step-wise method.

After development of the final multivariate model, the influence of extremely high and low covariate values was evaluated. The adequacy and predictive accuracy of the model was also assessed using the Hosmer and Lemeshow Goodness-of-Fit Test and c statistics.(18) Statistical analysis was performed using the SAS system.

RESULTS

Thirty baseline factors were originally identified as variables of potential clinical relevance for the analysis either as continuous (Table 1) or categorical variables (Table 2). No significant differences existed between the two groups (placebo and raloxifene) for any variables. The subset of patients (35.8%) for whom measurements of biochemical bone markers were available differed from those study participants without these measurements. Patients who had bone marker measurements were significantly older and taller, had lower BMI, and had higher LS BMD compared with those without these measurements (data not shown). In addition, patients in whom bone markers were measured had a lower vertebral fracture event rate (8.3% versus 11.1%, p = 0.03) and lower fracture risk reduction compared with the rest of the MORE population (RR = 0.70; 95% CI, 0.51, 0.97 versus RR = 0.51; 95% CI, 0.41, 0.64; p = 0.13).

Table Table 1.. Baseline Characteristics for Continuous Variables*
original image
Table Table 2.. Baseline Characteristics for Categorical Variables*
original image

Fracture risk in the placebo group

In the univariate analysis of the placebo group, seven variables were associated with an increased incidence of new vertebral fractures after adjusting for LS BMD (Table 3). These were low height measurement, age, years since menopause, cognitive function measured using the Trail Making Test, impaired musculoskeletal strength characterized by the Standing Time Test, low femoral neck BMD, and previous vertebral fracture. Other putative risk factors for fracture such as body weight, low BMI, high PTH, increased biochemical markers of bone turnover, family history of osteoporosis, smoking, and a history of falls were not significantly associated with increased risk of vertebral fractures. When univariate analysis of the placebo group was repeated after adjusting for age instead of LS BMD, all significant associations listed above remained (data not shown). In addition to the results of the Trail Making Test (OR = 1.24, p = 0.0005), impairment of cognitive function identified by the Short Blessed Test also reached the level of statistical significance (OR = 1.14, p = 0.03).

Table Table 3.. Univariate Analysis Results for New Vertebral Fracture Risk in Placebo Patients, Adjusted for LS BMD*
original image

Efficacy of raloxifene

Table 4 summarizes the significant (10% significance level) and some of the nonsignificant (10-20% significance level) univariate interactions between individual risk factors and the vertebral fracture risk reduction efficacy of raloxifene. In summary, serum triglycerides (p = 0.03), age (p = 0.04), and LS BMD (p = 0.08) influenced the fracture efficacy of raloxifene, indicating that patients with a high serum triglyceride level, who were younger, or had the lower LS BMD had the greater vertebral fracture risk reduction with raloxifene therapy. In addition, based on a small number of patients (n = 169, 2.5% of the study population) with type 2 diabetes, raloxifene also showed a higher efficacy in these patients compared with those without diabetes (p = 0.04). Furthermore, all nonsignificant interactions between the efficacy of raloxifene and risk factors, with the exception of less years after menopause, previous HRT use, and hysterectomy, were characterized with <0.1 risk reduction differences between either the 25% and 75% values (for continuous variables) or the two different categories (for categorical variables).

Table Table 4.. Univariate Interactions Between Individual Risk Factors and the Vertebral Fracture Risk Reduction Efficacy of Raloxifene
original image

In the non-representative subset of the study population with bone marker measurements, the baseline level of both bone formation markers (osteocalcin and bone ALP) interacted significantly with the vertebral fracture risk reduction of raloxifene, suggesting a higher fracture risk reduction on raloxifene in patients with high baseline values of these markers compared with patients with low values. In summary, the p interaction value for bone ALP was 0.066 (at the median value of 15.7 ng/ml, OR = 0.70; 95% CI, 0.51, 0.98), whereas the interaction p value for osteocalcin was 0.005 (at the median value of 24.3 ng/ml, OR = 0.72; 95% CI, 0.52, 0.99). The bone resorption marker U-CTX did not interact with the efficacy of raloxifene (p = 0.53).

Of the 30 potential risk factors identified at baseline, the 27 factors assessed in all patients were considered in the multivariate analysis; the biochemical markers of bone turnover (bone ALP, osteocalcin, and U-CTX), measured only in a subset of patients, were omitted. Out of these 27, 13 variables were selected for the final multivariate model as the result of the step-wise model building. These variables were height, age, years since menopause, serum PTH level, vitamin D level, HDL-C and LDL-C levels, triglyceride levels, Standing Time Test, LS BMD, family history of osteoporosis, geography, and previous vertebral fracture. In the outcome of the final multivariate model, triglyceride level and LS BMD demonstrated a significant interaction with treatment, indicating that the efficacy of raloxifene is dependent on these two baseline factors; an elevated triglyceride level (p = 0.006) and low LS BMD values (p = 0.008) were associated with increased vertebral fracture risk reduction with raloxifene therapy (Figs. 1 and 2). Other putative risk factors that might influence raloxifene efficacy based on the univariate analysis, such as height, age, years since menopause, serum PTH level, vitamin D level, HDL-C and LDL-C levels, Standing Time Test, family history of osteoporosis, geography, and previous vertebral fracture, did not significantly interact with the effect of raloxifene to reduce the risk of vertebral fractures.

Figure FIG. 1.

The estimated effect of raloxifene on the risk of new vertebral fracture at different baseline triglyceride levels, controlled for LS BMD (median LS BMD = 0.80 g/cm2). Error bars denote 95% CI.

Figure FIG. 2.

The estimated effect of raloxifene on the risk of new vertebral fracture at different baseline LS BMD values, controlled for triglyceride level (median serum triglyceride = 1.09 mM). Error bars denote 95% CI.

Figure 3 shows the relationship between serum triglyceride level and LS BMD and the effect of raloxifene. Raloxifene more effectively reduced vertebral fracture risk in patients with lower baseline LS BMD values and/or higher triglyceride levels. Neither influenced meaningfully the absolute fracture risk of the raloxifene-treated patients. In contrast, placebo patients with elevated triglyceride combined with low LS BMD had a markedly elevated absolute fracture risk compared with those with the lower triglyceride and higher LS BMD values.

Figure FIG. 3.

Relationship between LS BMD and triglyceride level at baseline and the incidence of vertebral fractures after 3 years in placebo and raloxifene-treated patients. For example, in high-risk patients, with low LS BMD (0.71 g/cm2) and high triglyceride levels (2.1 mM), the risks of new vertebral fractures were (A) 0.081 in the placebo group and (B) 0.023 in the raloxifene (pooled doses) group, which correspond to an absolute risk reduction of 0.058 and a relative risk reduction of 71.6%. In low-risk patients, with high LS BMD (0.95 g/cm2) and low triglyceride levels (0.99 mM), the risks of new vertebral fractures were 0.024 and 0.018 in the (C) placebo and (data not shown) raloxifene groups, respectively, which correspond to an absolute risk reduction of 0.006 and a relative risk reduction of 25%.

Patients with type 2 diabetes mellitus had a significantly higher serum level of triglycerides (1.8 mM) than those without diabetes (1.3 mM, p < 0.0001).

DISCUSSION

Fracture risk in the placebo group

The univariate analysis of the placebo group confirmed the impact of some potential clinical risk factors for vertebral fractures(19,20) in the MORE cohort, such as older age, a long time since menopause, impaired cognitive and musculoskeletal function, low BMD, and previous vertebral fractures. Other potential risk factors did not increase the vertebral fracture risk, such as a high level of circulating PTH, a low vitamin D level, smoking, alcohol intake, type 2 diabetes (p = 0.07), and a low endogenous estradiol level, possibly because of either no effect or a small effect and insufficient power of the current analysis. Interestingly, being shorter, but not low weight or BMI, increased the risk of vertebral fractures after adjusting for LS BMD. In the subset of placebo patients with biochemical bone marker measurements, none of the investigated markers (bone ALP, osteocalcin, or U-CTX) influenced fracture risk. This lack of influence is in conflict with the results of prospective cohort studies, where elevated levels of both formation and resorption markers were associated with increased vertebral and nonvertebral fracture risk, as independent risk factors.(21,22)

The association between cognitive or musculoskeletal impairment and an increased risk of vertebral fractures has not been described previously. These associations were independent of age, suggesting that these measurements may be of clinical importance. The strength of this analysis was that a variety of known and potential baseline risk factors for osteoporotic fractures were collected in a well-defined study population, and the influence of these risk factors on fracture efficacy of an antiresorptive agent was prospectively examined. A limitation of this analysis is that all potential clinical risk factors for fractures were not recorded at the study baseline, so the results are not comprehensive. Patients receiving glucocorticoid therapy or those with endocrine diseases (with the exception of type 2 diabetes), for example, were excluded from the study. Patients were selected either based on low BMD or prevalent fracture status.

Efficacy of raloxifene

Data from this analysis suggest that the efficacy of raloxifene in the reduction of vertebral fractures is largely independent of the existence of clinical risk factors for osteoporotic fractures. These are among the first data showing that the efficacy of an antiresorptive agent does not interact with a variety of known and potential risk factors of osteoporotic fractures. Using the same population (the MORE study database), some individual clinical risk factors have already been investigated. In those investigations, current smoking(23) and serum estradiol level(24) did not alter the efficacy of raloxifene to reduce vertebral fracture risk. The results obtained with the current multivariate model confirm these findings. Some other factors, such as younger age(25) and low LS BMD,(26) have been shown to increase the efficacy of raloxifene based on the results of univariate analyses of the MORE study. These findings are consistent with this analysis, but in the multivariate model, younger age was no longer a significant factor in affecting the efficacy of raloxifene.

In the univariate interaction analysis, a significantly greater vertebral fracture risk reduction was observed in patients with higher values of baseline bone formation markers (bone ALP and osteocalcin). Baseline levels of bone resorption marker U-CTX did not correlate with fracture reduction. These results are consistent with a previous report describing the relationship between the changes in bone turnover marker and vertebral fracture reduction in this subset of the MORE population. In this study,(27) a greater 12-month reduction of bone ALP and osteocalcin but not U-CTX was significantly correlated with a greater vertebral fracture reduction after 3 years on raloxifene therapy.

Numerous risk factors interact with each other, so it is not possible to determine from a univariate analysis which factors interact with treatment independently of any other factors. For this reason, a multivariate model was established and used to analyze all possible interactions between risk factors that affected the vertebral fracture risk reduction achieved on raloxifene therapy. A total of 13 of the 27 factors qualified for inclusion in the model. Based on the results of the systematic model, only serum triglyceride level and LS BMD could be considered as independent determinants of the efficacy of raloxifene in reducing vertebral fracture risk.

Raloxifene was more effective in patients with higher baseline triglyceride levels and/or lower baseline LS BMD values. Raloxifene, however, significantly reduced the risk of vertebral fractures across the spectrum of the triglyceride and LS BMD values measured in the MORE study. Thus, no pretreatment measurements of triglycerides and LS BMD are necessary to predict the drug efficacy. The fact that patients with low LS BMD, who if untreated have a higher risk of vertebral fracture, seem to benefit more from raloxifene would seem to be a desirable treatment effect.

Interestingly, the effect of both the triglyceride and LS BMD levels on the absolute fracture incidence was negligible in patients treated with raloxifene but was very pronounced in placebo-treated patients. Consequently, the influence of elevated triglyceride and lower LS BMD values on risk reduction was mainly because of the higher fracture incidence in the placebo group. The higher fracture incidence in placebo patients with the lower LS BMD values is plausible, but the relationship between higher fracture incidence and elevated serum triglycerides has not been previously described. It is possible that elevated triglyceride levels could reflect a metabolic state such as increased insulin resistance and type 2 diabetes mellitus associated with higher risk for fragility fractures. Indeed, in a meta-analysis, diabetes mellitus was associated with increased risk of peripheral fractures.(28) Additional results in this study also support this view. Patients with diabetes mellitus in the placebo group (N = 45) had an increased vertebral fracture risk, the OR being 2.17 (p = 0.07, see Table 3). In the univariate efficacy analysis, raloxifene was more effective in patients with type 2 diabetes mellitus. In addition, patients with diabetes in the MORE study had significantly elevated serum triglyceride levels. Thus, the increased efficacy of raloxifene in patients with elevated triglyceride levels may be because of greater efficacy of raloxifene in patients with increased insulin resistance and type 2 diabetes mellitus.

Among the different types of osteoporotic fractures, the clinical significance of vertebral fractures is increasingly acknowledged, because these are associated with significant pain, disability, and impairment of quality of life. Vertebral fractures have also been shown to be associated with increased mortality in postmenopausal women.(29–31) Some risk factors apply to all of the major types of osteoporotic fractures (i.e., femoral neck, forearm, vertebra), but their predictive value differs for each location.(31) Specifically, each SD decrease in LS BMD is associated with an ∼2-fold increase in vertebral fracture risk.(33) Prevalent fractures are one of the strongest predictors of subsequent vertebral fractures.(34) Other risk factors associated with vertebral fracture are age, late menarche, early menopause, a short duration of fertility, low consumption of cheese and yogurt, a history of osteoporosis, a reduction in height, and low levels of physical activity.(35–37) Interestingly, in elderly women, increased body weight and BMI were associated with a reduced risk of vertebral fractures in the European Prospective Osteoporosis Study (EPOS) study.(19)

Within this framework, these data suggest that the efficacy of raloxifene to reduce vertebral fracture risk is largely independent of the existence of clinical risk factors for osteoporotic fractures.

Acknowledgements

The MORE study was funded by Eli Lilly and Company.

Ancillary