This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Marcus, R. Articles by Stock, J. L. Search for Related Content PubMed PubMed Citation Articles by Marcus, R. Articles by Stock, J. L.

Antiresorptive Treatment of Postmenopausal Osteoporosis: Comparison of Study Designs and Outcomes in Large Clinical Trials with Fracture as an Endpoint

Department of Medicine, Stanford University School of Medicine, and the Musculoskeletal Research Laboratory, Geriatric Research, Education, and Clinical Center, Veterans’ Affairs Medical Center (R.M.), Palo Alto, California 94304; and Lilly Research Laboratories, Eli Lilly and Company (M.W., H.H.III, J.L.S.), Indianapolis, Indiana 46285

Correspondence: Address all correspondence and requests for reprints to: Robert Marcus, M.D., Professor of Medicine, Stanford University, Geriatric Research, Education, and Clinical Center 182-B, Veterans’ Affairs Medical Center, 3801 Miranda Avenue, Palo Alto, California 94304. E-mail: marcuse{at}Stanford.edu

	Abstract

Top Abstract I. Introduction II. Calcium and Vitamin... III. Pharmacological Therapies... IV. Discussion References

 
Antiresorptive treatments for postmenopausal osteoporosis have been studied extensively, but due to the volume of published data and lack of head-to-head trials, it is difficult to evaluate and compare their fracture reduction efficacy. The objective of this review is to summarize the results from clinical trials that have fracture as an endpoint and to discuss the factors in study design and populations that can affect the interpretation of the results. Although there are numerous observational studies suggesting that estrogen and hormone replacement therapies may reduce the risk of vertebral and nonvertebral fractures, there is no large, prospective, randomized, placebo-controlled, double-blind clinical trial demonstrating fracture efficacy. The effects of raloxifene, alendronate, risedronate, and salmon calcitonin on increasing bone mineral density (BMD) and decreasing fracture risk have been shown in randomized, placebo-controlled, double-blind clinical trials of postmenopausal women with osteoporosis. Although the increases in lumbar spine BMD vary greatly in these trials, the decrease in relative risk of vertebral fractures is similar among therapies. However, nonvertebral fracture efficacy has not been consistently demonstrated. Combined administration of two antiresorptive therapies results in greater BMD increases, but the effects on fracture risk are unknown. Direct comparisons of clinical trial results should be considered carefully, given the differences in study design and populations. Differences in study design that may influence the efficacy of fracture risk reduction include calcium and vitamin D supplementation, primary fracture endpoints, definition of vertebral deformity or fracture, discontinuation rates, and statistical power. Factors in the study population that may influence fracture efficacy include the age of the population and the proportion of subjects with prevalent fractures. The use of surrogate endpoints such as BMD to predict fracture risk should be approached with caution, as the relationship between BMD changes and fracture risk reduction with antiresorptive therapies is uncertain. Consideration of these results from clinical trials can contribute to clinical judgment in selecting the best treatment option for postmenopausal osteoporosis.

I. Introduction

A. Clinical endpoint of fracture

B. Surrogate markers of efficacy

C. Methodology and statistical techniques used in clinical trials

II. Calcium and Vitamin D as Baseline Therapy for Osteoporosis

III. Pharmacological Therapies for Osteoporosis

A. Estrogen and hormone replacement therapies

B. Selective estrogen receptor modulators (SERMs)

C. Bisphosphonates

D. Salmon calcitonin nasal spray

E. Other antiresorptive therapies

F. Antiresorptive therapies in combination

G. Risks and extraskeletal benefits

IV. Discussion

	I. Introduction

Top Abstract I. Introduction II. Calcium and Vitamin... III. Pharmacological Therapies... IV. Discussion References

 
THE INCIDENCE OF osteoporosis, its associated morbidity, and the health-care costs connected with its prevention and treatment will rise dramatically in the future as an increasing proportion of the world’s population lives to old age (1, 2). In 1941, Albright et al. (3) postulated that postmenopausal estrogen deficiency plays a primary role in development of osteoporosis, which until recently was thought to be an inevitable consequence of aging and regarded as untreatable. More recent research has progressively improved understanding of bone biology and how calcium and vitamin D deficiency, lack of exercise, and hormone deficiency and excess contribute to the emergence of osteoporosis in aging women and men (1). Important advances in the diagnosis, characterization, and management of osteoporosis have been translated into clinical practice since the 1970s (4).

At this writing, the U.S. Food and Drug Administration has approved several drugs for management of postmenopausal osteoporosis. Estrogen alone [estrogen replacement therapy (ERT)] has been used for management of postmenopausal symptoms for over 50 yr and, along with hormone replacement therapy [HRT (estrogen combined with progestin)], is approved for osteoporosis prevention. In the past few years, reductions in fracture incidence demonstrated in long-term, prospective, double-blind, placebo-controlled, randomized clinical trials (RCTs) have become necessary for regulatory approval of newer therapies for treatment of osteoporosis. In the United States, raloxifene, alendronate, and risedronate are indicated for osteoporosis prevention and treatment, whereas salmon calcitonin is approved for osteoporosis treatment. In addition, etidronate, tibolone, and active vitamin D analogs are used in other countries. For the purposes of clinical trial design and drug registration, prevention of osteoporosis has been interpreted as the use of therapy to prevent further bone loss in subjects who do not yet have osteoporosis, although the ultimate goal of an osteoporosis therapy is prevention of fractures. Treatment of osteoporosis has been interpreted as the use of therapy to prevent new fractures in patients who already have osteoporosis diagnosed by bone mineral density (BMD) definitions or as a result of prevalent fractures.

It is difficult to compare the antifracture efficacy of the available therapeutic options, as there have been relatively few trials with fracture endpoints and no head-to-head trials comparing the efficacy of approved drugs on fracture risk reduction. However, many clinical trials have measured surrogate endpoints such as BMD. It has become common to compare changes in surrogate markers among antiresorptive drugs and to infer apparent differences in therapeutic efficacy on fracture risk reduction. Several reviews have described the effects of osteoporosis therapies on BMD and fracture risk (4, 5, 6). Although techniques have emerged to subject clinical trial data to formal systematic review (7), application of such techniques to the field of osteoporosis remains problematic, as systematic reviews do not generally elucidate many of the factors that can impact trial results. Several RCTs studied the effects of alendronate and risedronate on fracture incidence, and the effects of salmon calcitonin nasal spray and raloxifene on fracture incidence were reported from single, large RCTs, but RCT data for estrogen therapies are sparse.

This review examines critically the available data on fracture risk reduction for calcium and vitamin D and for pharmaceutical skeletal antiresorptive drugs that are currently approved in the United States or used in countries other than the United States in patients with postmenopausal osteoporosis. This review discusses differences in study design and population characteristics that may influence interpretation of clinical trials with fracture as an endpoint in women with postmenopausal osteoporosis. For example, findings in RCTs often differ from those obtained in observational or case-controlled studies (8). Calcium and vitamin D supplementation, primary fracture endpoints, definitions of vertebral deformity or fracture, discontinuation rates, and statistical power may vary across trials and affect outcomes. Because none of the major clinical trials included representative numbers of women from ethnic minorities, published data and conclusions may not necessarily apply to minority populations. This problem is compounded by the controversy regarding the need for ethnic-specific BMD databases, as the World Health Organization definition of osteoporosis is based on the prevalence in white women (9).

We searched MEDLINE for full English-language publications from 1966 to February 2001 containing the key word "fracture" and the names of the individual antiresorptive drugs. The search results were supplemented by references cited in these publications and by abstracts when full publications of the clinical trial results were unavailable. Tables 1 and 2 summarize RCTs with calcium and/or vitamin D, and ERT or HRT, respectively. None of these trials enrolled large numbers of women with an inclusion criterion of documented osteoporosis. The efficacy of ERT and HRT for osteoporosis therapy has been supported primarily by nonrandomized observational studies, which are included for comparison (see Table 3). The main part of this review (see Tables 4–6) summarizes RCTs that each enrolled at least 500 postmenopausal women with osteoporosis, had fracture as an endpoint, and studied pharmaceutical antiresorptive agents approved for osteoporosis prevention and/or treatment in the United States. Many other published studies present the effects of antiresorptive therapies on the surrogate endpoints of BMD and biochemical markers of bone turnover in postmenopausal women with or without established osteoporosis. Data on biochemical markers of bone turnover will not be described in detail in this review, as these markers are not widely used in clinical practice for diagnosis of osteoporosis or monitoring response to antiresorptive therapy. To relate the changes in the surrogate endpoint of BMD to the ultimate clinical outcome, this review will limit the discussion of BMD to data from studies that had fracture as an endpoint.

Definitions and methods of assessment of nonvertebral fractures vary among clinical trials. The annual rate of vertebral fractures is approximately 3 times that of hip fractures, with the latter occurring in women at a mean age of 75 yr (2, 20). Due in part to the relative infrequency of hip fractures in typical clinical trial subjects, it has been difficult for trials of antiresorptive therapies to demonstrate antifracture efficacy in the hip, and the confidence intervals (CIs) around relative risk (RR) estimates are wide. In reporting the total incidence of new nonvertebral fractures, some clinical trials have included fractures that result from traumatic injury, such as a traffic collision or beating, or fractures at non-weight-bearing sites, such as the skull or digits. It is uncertain whether traumatic fractures or fractures at non-weight-bearing sites are clinically relevant outcomes in osteoporosis.

B. Surrogate markers of efficacy
Because lengthy and expensive clinical trials are necessary to determine the effects of antiresorptive drugs on fractures, many studies used surrogate markers for antifracture efficacy. The role of BMD measurements in detecting osteoporosis and predicting fracture risk is well established (21). After adjustment for age, low baseline BMD and prevalent (radiographic) vertebral fractures both predict increased risk of subsequent vertebral fractures (22, 23). Although several techniques are available to estimate BMD, dual-energy x-ray absorptiometry is the method of choice for measuring BMD in the axial skeleton and is used to diagnose osteoporosis. The World Health Organization criterion for a BMD diagnosis of osteoporosis is a spine and/or hip BMD that is 2.5 SD values or more below the young adult mean value (T-score -2.5). Women who have a spine or hip BMD T-score between -1 and -2.5 are considered to have osteopenia or low bone mass and are at above-average risk of a future fracture (24). Therapeutic intervention may be considered for women with osteopenia, depending on the presence of additional risk factors such as age, family history, medication or life-style exposure, or a past history of fracture (4). Most skeletal sites used for BMD measurement have similar overall predictive ability for fractures, with a 1 SD decrease in BMD associated with an approximately 2-fold increase in fracture risk (25).

The known relationship between low baseline BMD and increased fracture risk (25) has often been extrapolated to relate treatment-induced increases in BMD to fracture risk reduction, although this latter relationship has not been completely defined or quantitated (26). Treatment-associated changes in BMD reportedly accounted for 4–67% of the reduction in vertebral fracture risk observed in five clinical trials with various antiresorptive drugs and in the Study of Osteoporotic Fractures (27). For raloxifene therapy, changes in lumbar spine and hip BMD, respectively, accounted for approximately 14% and 9% of the observed reduction in vertebral fracture risk (28), which was similar irrespective of the changes in femoral neck BMD (29). With alendronate therapy, the proportion of the vertebral fracture risk reduction that was explained by an increase in lumbar spine BMD was 17–27% in women with prevalent vertebral fractures and 13–20% in women without prevalent vertebral fractures (30). A meta-analysis of 13 clinical trials with various antiresorptive therapies calculated a 1% increase in spine BMD to be associated with a 3.3% decrease in vertebral fracture risk, yet the observed fracture risk reduction was twice as large as was expected from these calculations (31).

In contrast, two other analyses suggest that a stronger relationship does exist between increased BMD and decreased fracture risk with antiresorptive therapy. The relationship between BMD and subsequent vertebral fracture was evaluated in 2,984 women who received alendronate therapy and had total hip BMD measurements (32). Compared with women who had no gain or a loss in BMD at 2 yr, women who had a 3% or greater gain in BMD with alendronate had a significant 50–60% decreased vertebral fracture risk, but the vertebral fracture reduction in women who had BMD gains of less than 3% was not significantly different from the reduction in risk in those women who had no gain or a loss in BMD. An analysis of 13 clinical trials predicted that an antiresorptive agent that increases spine BMD by 8% would produce a 13% decrease in vertebral fracture risk not related to BMD changes, whereas an agent that produces no change in BMD was predicted to reduce vertebral risk by 20–22% (33).

There are factors, such as changes in bone turnover, bone microarchitecture, and propensity to fall, that may contribute to fracture risk reduction observed with antiresorptive therapy (see Table 7). The above analyses suggest that fracture risk reduction with antiresorptive therapy may result from changes in these other risk factors in addition to BMD increment. The clinical utility and predictive power of biochemical markers of bone turnover remains controversial (34). A recent review stated that there was no conclusive evidence relating changes in biochemical markers to fracture outcomes in patients treated with antiresorptive therapy (35). Although changes in surrogate markers can verify antiresorptive activity of a treatment, reduction in new fracture risk is the ultimate determinant of the clinical efficacy.

If subjects leave a RCT before study completion for any reason either related or unrelated to treatment, study outcomes may be biased if the analyses are performed on only the remaining subjects. To adjust for this potential bias, the outcomes for all subjects with at least one post-baseline measurement are included in the analysis, using a technique referred to as "last observation carried forward." In last observation carried forward analysis, patient data are analyzed on an intention-to-treat basis, according to their original study group assignment irrespective of whether they actually received their assigned study drug or completed the study. Statistical analyses should be specified in the study design a priori before data collection so that the results can be interpreted to answer the study question. Exploratory post hoc statistical analyses are performed after data collection to answer questions suggested by the data and are considered less scientifically rigorous than a priori analyses. If a large number of post hoc analyses are performed, statistically significant results that are not biologically plausible may be achieved by chance. Data collected to answer the initial study question may not be appropriate to address these new relationships, and the results of post hoc analyses may be difficult to interpret (36, 37).

The incidence of fracture is the proportion of women in each group who experience a new fracture during the study and is expressed as a percentage. The difference between the treatment and control groups in the fracture incidence is the absolute risk reduction. The number of women needing treatment (NNT) to prevent one fracture over a specified time period is the reciprocal of the absolute risk reduction. Because NNT is dependent on the prevalence of the disease as well as drug efficacy, a high NNT value may be due, in part, to the rare occurrence of a disease. The ratio of the risk of a fracture occurring in one group compared with another group within a certain time period is the RR, which is used in RCTs, or relative hazard (RH), which is used in observational studies for hazard analysis of survival. The probability of a new fracture divided by one minus the probability of a new fracture is the odds of a new fracture. The ratio of the odds that a fracture would occur in one group compared with the odds for another group is the odds ratio (OR), which can be used to estimate the RR when the incidence rate is very low. The CI indicates the precision of the fracture risk point estimates, expressed as RR, RH, or OR. Wider intervals indicate less precision, and if the upper or lower limits of the CI do not overlap 1.0, then the result is considered to be statistically significant. The statistical power of a study, specified in the study design, is the ability to detect a specified difference between groups and is dependent on the sample size, frequency of the outcome event, and the magnitude of the treatment effect. Interpretation of treatment effects is more difficult if a large number of study subjects discontinue the study, because the statistical power to detect a difference is diminished even though the study may have had adequate statistical power at the beginning.

	II. Calcium and Vitamin D as Baseline Therapy for Osteoporosis

Top Abstract I. Introduction II. Calcium and Vitamin... III. Pharmacological Therapies... IV. Discussion References

 
Calcium and vitamin D are frequently recommended as dietary supplements for postmenopausal women, although their formal approval for osteoporosis prevention or treatment in the United States has been neither required nor sought. To assume adequate nutritional status and bone mineralization, calcium and vitamin D are often used concomitantly in clinical trials of other antiresorptive therapies. Several studies have examined the effects of calcium and/or vitamin D on fracture incidence and BMD, with varying outcomes possibly due to differences in study populations, treatment regimens, and baseline nutritional status of the subjects (Table 1).

Calcium was given alone in several randomized, placebo-controlled, prospective studies, but the effects on bone loss and fracture rate reduction were inconsistent. Postmenopausal women with prevalent vertebral fractures who were given a calcium supplement had reduced forearm BMD loss and a substantially lower incidence of new vertebral fractures (Ref. 38 and Table 1). In contrast, women who did not have prevalent fractures did not have significant reduction in fracture risk. This was possibly related to the imbalance in the number of placebo (n = 59) and calcium-treated (n = 40) patients, the lower rate of incident vertebral fractures, and fewer individuals with nutritional calcium deficiency in the group without prevalent fractures. In a second study, elderly women and men with or without prevalent hip fractures were randomized to receive placebo or calcium after treatment with vitamin D to correct any underlying deficiency (39). The incidence of vertebral fractures was 2 times higher in patients with prevalent hip fractures than in those without hip fractures, even with calcium supplementation. In subjects without hip fractures who received calcium supplements, there was no significant decrease in the new vertebral fracture rate, although the data were not expressed as number of patients with new vertebral fractures and there was no difference in lumbar spine BMD from placebo, but femoral shaft BMD was increased by 2.0% above placebo (Table 1).

In a 4-yr trial, calcium-treated postmenopausal women who completed the study had a significantly decreased incidence of symptomatic nonvertebral fractures (Ref. 40 and Table 1). However, fracture was not a primary endpoint of this study, and the analysis, which was based on fracture rates calculated from a small number of events, did not present data on the smaller number of patients who sustained at least one new nonvertebral fracture. Therefore, as noted by the authors, these results cannot be regarded as definitive evidence of the efficacy of calcium supplementation for fracture prevention. In this study, calcium administration reduced BMD loss by approximately 1–2% at various sites in each year, with the greatest effects on BMD in the first year.

Some population-based studies also suggest that high dietary calcium intake throughout life (41) or in adulthood (42) reduces the risk of hip fractures. Cumming and Nevitt (43) systematically reviewed randomized and nonrandomized trials, observational epidemiological studies, and case control studies to assess the effect of dietary or supplemental calcium on fracture risk in postmenopausal women. Most of the studies analyzed did not restrict the study population to postmenopausal women with osteoporosis. The reviewers performed a meta-analysis of 16 observational studies of dietary calcium and hip fractures and concluded that increased calcium intake was associated with a small reduction in fracture risk.

Prospective studies of combined calcium and vitamin D administration in postmenopausal women have also yielded mixed results (Table 1). Institutionalized French women treated with calcium and vitamin D (800 IU/day) had significant reductions in hip and nonvertebral fractures and significantly increased femoral neck (1.1%) and trochanter (5.4%) BMD at 18 months compared with placebo (44, 45). These effects may actually reflect the benefits of treating vitamin D deficiency and secondary hyperparathyroidism likely present in many of the women in this French study. Calcium and vitamin D supplementation (700 IU/day) also reduced the incidence of the first osteoporotic nonvertebral fracture in 213 healthy, independently living women (46). However, fracture was a secondary endpoint in this study of men and women, and only 32 women sustained at least one new nonvertebral fracture. Additionally, although an intention-to-treat analysis was performed, only those subjects evaluated at the final study visit (87% of those enrolled) were included. Those women who received calcium and vitamin D had a significantly increased total body BMD (1.1% above placebo) and small nonsignificant increases in lumbar spine (0.6%) and femoral neck (0.3%) BMD.

A relatively low dose of vitamin D (400 IU/day) did not decrease the incidence of hip and peripheral fractures in independent elderly Dutch men and women (Ref. 47 and Table 1). In contrast, in an open study with randomization determined by birthdays, elderly Finnish women who received annual im injections of 150,000–300,000 IU of vitamin D for up to 5 yr had significantly fewer clinical fractures than did a control group who did not receive injections (48). The variation in results of these vitamin D studies may be related to the different doses of vitamin D used or population differences such as baseline vitamin D status.

Thus, available studies suggest small effects of calcium and/or vitamin D supplementation to decrease the risk of some fractures in elderly women, possibly related to underlying mild or moderate nutritional deficiencies. The inconsistency of the fracture risk-reduction data may be related to the underlying nutritional status and/or different regimens that were used. Because calcium and/or vitamin D supplements have been used in placebo and treatment groups in most large RCTs of pharmacological therapies for osteoporosis, any fracture efficacy of the therapies observed in these trials is over and above the efficacy that might be derived from the supplements.

In addition to vitamin D given as a nutritional supplement to women at risk for osteoporosis, synthetic forms of activated vitamin D are used as pharmacological agents for osteoporosis therapy in some countries. Those vitamin D analogs will be discussed in Section III.E with other antiresorptive therapies.

	III. Pharmacological Therapies for Osteoporosis

Top Abstract I. Introduction II. Calcium and Vitamin... III. Pharmacological Therapies... IV. Discussion References

 
The U.S. Food and Drug Administration has approved several antiresorptive therapies for the prevention and/or treatment of postmenopausal osteoporosis. These therapeutic agents include ERT and HRT, a selective estrogen receptor modulator (SERM), bisphosphonates, and salmon calcitonin.

A. Estrogen and hormone replacement therapies
Although ERT or HRT have been widely accepted as the agents of choice for osteoporosis management, data from RCTs investigating the effects of ERT or HRT on fracture risk in postmenopausal women with osteoporosis are sparse. Studies demonstrating increased BMD and retrospective case-control and prospective cohort studies are often cited as evidence for efficacy of ERT or HRT for osteoporosis treatment due to the lack of well-designed large RCTs. Interpretation of results from such studies is difficult, as there is no control over composition of the study population or the results obtained. Women who use ERT or HRT tend to have better overall health and closer medical follow-up than women who do not use these drugs (49), and it is difficult to match for such baseline variables. Also, the results collected in such studies may be influenced by the subjects’ recall ability. Estrogen reduces bone turnover by inhibiting osteoclast recruitment and activity, and progestin is added to estrogen in HRT to prevent the endometrial hyperplasia that occurs with unopposed estrogen. After a review of the evidence, Lindsay (50) concluded that progestins given at normal doses neither have a direct effect on osteoporosis prevention, nor have additive effects with estrogen when used in HRT. There are no data showing that progestins reduce fracture risk.

In the only RCT studying the effects of HRT (Table 2) with a fracture endpoint in postmenopausal women with osteoporosis, the HRT group had fewer vertebral fractures and a 5.1% increase in lumbar spine BMD compared with the placebo group (51). There may have been insufficient time for significant BMD changes to occur in the femoral neck due to the 1-yr duration of this study, which enrolled only 75 subjects. This study is often cited as evidence of the vertebral fracture efficacy of HRT, despite several important concerns about its design and analysis. Women with a calcium intake of less than 800 mg/day were instructed to increase their dietary calcium to this level, but no calcium or vitamin D supplements were provided. Additionally, this study defined a vertebral fracture by a 15% decrease in vertebral height. Because more recent studies of other antiresorptive therapies used a 20% decrease to define a vertebral fracture, there is dispute as to whether this change is an adequately stringent criterion to define a true vertebral fracture (52) and to whether the 15% criterion is related to clinically relevant outcomes. Finally, reporting the vertebral fracture results as events per person-year has been criticized because of the tendency of these fractures to occur in clusters within individuals (53). The apparent drug effect was not significant when results were expressed as number of women who sustained at least one fracture (53). Thus, although the observed changes in BMD in this study are consistent with an antiresorptive effect, this study did not offer compelling evidence for antifracture efficacy of estrogen.

There are other RCTs of ERT or HRT with fracture as an endpoint in populations of women without documented osteoporosis (Table 2). Metacarpal bone mineral content was maintained in oophorectomized women who received mestranol for a median of 9 yr, whereas those in the placebo group lost bone (54). Bone mineral content in the metacarpal and radius was significantly greater in the mestranol group compared with placebo. A spinal score, calculated by weighted scoring of wedged and collapsed vertebrae, was significantly lower in the mestranol group (Table 2) than with placebo, suggesting fewer fractured vertebrae. Although these women had all undergone oophorectomy, it is unknown whether the double blind could be maintained for as long as 9 yr, given the symptoms associated with HRT therapy.

Clinical fractures were assessed as a secondary outcome in the Heart and Estrogen/Progestin Replacement Study (HERS) (8), the only large completed, prospective trial of HRT in postmenopausal women. The primary endpoint of this study was the occurrence of cardiovascular events in women with coronary artery disease. The women were at low risk for osteoporosis and their BMD status was generally unknown (8). Additionally, they did not receive calcium supplements. The RH of any fracture, in the hip or other regions, was not significantly different between the HRT and placebo groups (Table 2). This result could be due to inadequate statistical power in this population of patients who were selected on the basis of cardiovascular disease and not osteoporosis, the short duration of HRT use, or lack of efficacy of this HRT preparation (8). Trials with other antiresorptive agents (55, 56) did not show efficacy in risk reduction for nonvertebral fractures in women without documented osteoporosis.

Komulainen et al. (57) studied the effects of HRT, with or without vitamin D supplementation, on the incidence of nonvertebral fractures in postmenopausal women. A preplanned intention-to-treat analysis showed a nonsignificant reduction in nonvertebral fracture risk in the HRT group (Table 2). The risk of nonvertebral fractures was significantly reduced by 57% (95% CI, 3–81%) after pooling the HRT groups and adjusting for baseline BMD and previous fracture. A high discontinuation rate may have introduced a conservative bias, as 72 of the 96 women (75%) who withdrew from the study were in the HRT group.

A 10-yr RCT of high-dose HRT (including 2.5 mg/day conjugated equine estrogen) was conducted in 84 pairs of postmenopausal women who were residents of a chronic-care hospital (58). Seven fractures occurred in the control group and none in the HRT group, although fractures were not a prespecified endpoint and statistical interpretation was not performed. This study had methodological issues, including the high dose of estrogen used, lack of calcium and/or vitamin D supplementation, and the difficulty of maintaining a double blind for 10 yr in women receiving high-dose HRT.

Most observational trials suggest that postmenopausal women who used ERT or HRT have reduced fracture risk compared with case controls (Table 3). In the prospective cohort Study of Osteoporotic Fractures, Cauley et al. (59) found that current or prolonged use of ERT or HRT was associated with decreased fracture risk, but cautioned that the conclusions drawn from such trials may be affected by important study design aspects. For example, information on the initiation and duration of estrogen use and on the doses and preparations used were based upon self-report. In addition, observational trials are subject to selection bias, because postmenopausal women who use ERT or HRT for long periods have associated clinical, socioeconomic, and lifestyle variables that are independently related to decreased fracture risk (49). The evidence summarized above, along with many studies that show increased BMD, suggest that ERT or HRT are effective in reducing fracture risk. The skeletal efficacy results with ERT and HRT from large and well-designed RCTs, such as the Women’s Health Initiative (60), are anticipated.

B. Selective estrogen receptor modulators (SERMs)
Raloxifene, a benzothiophene SERM, is approved for prevention and treatment of postmenopausal osteoporosis and acts as an estrogen receptor agonist in bone and on serum lipid concentrations and as an estrogen receptor antagonist in the breast and uterine tissues (61). The Multiple Outcomes of Raloxifene Evaluation (MORE) trial (62) studied the effects of 60 or 120 mg/day raloxifene on the risk of fractures in postmenopausal women with osteoporosis, defined by femoral neck or lumbar spine BMD T-score less than or equal to -2.5 or radiographically apparent pre-existing vertebral fractures (Table 4). Those with pre-existing fractures were also required to have osteoporosis by BMD criteria unless they had at least two moderate fractures. The incidence of new vertebral fractures was the primary endpoint of the MORE study, with the incidence of new nonvertebral fractures as a secondary endpoint.

At 3 yr, 60 mg/day raloxifene decreased the cumulative risk of new morphometric vertebral fractures by 55% in women with low BMD and no prevalent vertebral fractures, and by 30% in women with prevalent vertebral fractures (63). In the group of women who had a low prevalence of baseline vertebral fractures, the absolute risks of new vertebral fractures were decreased by 2.2% and 1.7% with 60 and 120 mg/day raloxifene, respectively (Table 5). In contrast, the absolute risks of new vertebral fractures were reduced by 6.5% and 10.5% with 60 and 120 mg/day raloxifene, respectively, in the group of women who had a high prevalence of baseline vertebral fractures (Table 5). Raloxifene treatment reduced the risk of new vertebral fractures to a similar extent when the study population was divided into subgroups based upon tertiles of age, tertiles of baseline femoral neck or lumbar spine BMD, hysterectomy status, or prior use of HRT. Raloxifene also significantly decreased the risk of new multiple vertebral fractures by 50–80% compared with placebo. In the pooled raloxifene groups at 3 yr, the risk of painful clinical vertebral fractures was reduced by 60% compared with placebo (Table 5). Clinical vertebral fractures were defined as incident fractures confirmed through unscheduled radiographs performed after subjects reported symptoms suggestive of fractures. A post hoc analysis found that 60 mg/day raloxifene at 12 months decreased the risk of clinical vertebral fractures by 68% (95% CI, 20–87%) compared with placebo (64). The cumulative risk of nonvertebral fractures at 3 yr was not significantly different in the pooled raloxifene groups compared with placebo (Table 6). Compared with placebo, lumbar spine and femoral neck BMD were increased significantly by 2–3% with both doses of raloxifene (62).

Several aspects of the study design and outcomes of the MORE trial may have contributed to the nonvertebral fracture results. The MORE study did not have statistical power to detect reductions in nonvertebral fracture risk at any site. Women were required to discontinue the MORE study if they developed excessive BMD loss and/or more than two new vertebral fractures. The early discontinuation rule removed high-risk women from the study cohort, particularly from the placebo group, which had 3 times more women withdraw from the study than did the raloxifene groups (62). The impact of this early discontinuation rule on the incidence and RRs of vertebral and nonvertebral fractures is unknown. Thus, the MORE trial showed that raloxifene effectively reduces the risk of vertebral fractures in postmenopausal women with osteoporosis, but efficacy in reducing the risk of site-specific nonvertebral fractures was not demonstrated.

C. Bisphosphonates
Bisphosphonates are stable pyrophosphate analogs that bind to hydroxyapatite in bone and inhibit bone resorption by decreasing the number and activity of osteoclasts (65). Alendronate and risedronate are the bisphosphonates currently approved in the United States for the prevention and treatment of postmenopausal osteoporosis (66). The first-generation bisphosphonate etidronate is not approved for osteoporosis indications in the United States, although it is approved for these indications in other countries and will be discussed in Section III.E (67).

1. Alendronate.
Several RCTs have examined the effects of alendronate on fracture endpoints in postmenopausal women with osteoporosis (Table 4). In the first published RCT, Liberman et al. (68) examined the effects of alendronate in a dose-ranging study of postmenopausal women with osteoporosis defined by low BMD, 20% of whom had prevalent vertebral fractures. The Fracture Intervention Trial [FIT (69)] of alendronate enrolled postmenopausal women with prevalent vertebral fractures and women with low femoral neck BMD and no prevalent vertebral fractures (55). Some of the women without prevalent vertebral fractures who were enrolled in the FIT on the basis of low femoral neck BMD did not have osteoporosis according to BMD criteria from the Third National Health and Nutrition Examination Survey (69A ), published after initiation of the FIT. Post hoc analyses were performed to measure the effects of treatment on the women who had osteoporosis by these new criteria. The fourth RCT was a multinational study that assessed the efficacy of 10 mg/day alendronate on BMD in postmenopausal women with osteoporosis defined as low lumbar spinal BMD (70). We will summarize the data from each study for the effects of alendronate on vertebral, clinical, and nonvertebral fracture risks separately to obtain an overall perspective. Clinical vertebral fractures were defined as those that came to medical attention, reported to the clinical center by participants, and confirmed by the study radiologists.

In the study by Liberman et al. (68), the cumulative RR of new morphometric vertebral fractures was reduced by 48% (95% CI, 5–72%) in the alendronate-treated women compared with placebo-treated women. The RR of new vertebral fracture was still decreased when the study population was stratified according to age, presence or absence of previous vertebral fractures, or alendronate dose. Alendronate at 5 and 10 mg/day reduced the absolute risks of new vertebral fractures by 3.3% and 3.4%, respectively. In the FIT (69), alendronate decreased the cumulative risk of new morphometric vertebral fractures by 47% in women with prevalent vertebral fractures and by 44% in women with no prevalent vertebral fractures, compared with placebo (Table 5 and Ref. 55). Alendronate therapy reduced the absolute risk of new vertebral fractures by 7.0% in women with prevalent vertebral fractures and by 1.7% in women without prevalent vertebral fractures in the FIT (Table 5). Vertebral fractures were not radiographically assessed at defined intervals in the multinational alendronate trial by Pols et al. (70).

Alendronate decreased the incidence of new vertebral fractures associated with clinical symptoms by 55% in women with prevalent vertebral fractures in the FIT (Table 5 and Ref. 69) but did not significantly reduce the risk of total clinical fractures, including clinical vertebral fractures, in women with no prevalent vertebral fractures (55). However, women in this latter arm of the FIT who had a femoral neck BMD T-score of less than or equal to -2.5 experienced a significant 36% reduction in the risk of any clinical fracture with alendronate, whereas women with higher femoral neck BMD T-scores did not have a significant reduction (55). An analysis of 3,658 women with prevalent vertebral fractures and/or baseline femoral neck BMD T-scores less than or equal to -2.5 from both arms of the FIT found that the cumulative risk of clinical vertebral fractures was reduced by 59% (P < 0.001) and 45% (P = 0.003) at 1 and 3 yr, respectively, in the alendronate group, compared with placebo (71, 72). Alendronate treatment reduced the incidence of multiple (two or more) new vertebral fractures per subject by 90% (95% CI, 78–95%) in women with prior vertebral fractures, but the incidence of multiple vertebral fractures was not significantly reduced in women without prevalent vertebral fractures (55). Alendronate treatment also had beneficial effects on measures related to vertebral fractures. In the study of Liberman et al. (68), women in the alendronate group had less progression in the Spinal Deformity Index. Two trials reported a decrease in average height loss with alendronate treatment compared with the placebo group (68, 69).

Although alendronate therapy consistently decreased the risk of new vertebral fractures, the effect of alendronate on nonvertebral fracture risk reduction varied across trials (Table 6). The cumulative incidence of reported symptomatic nonvertebral fractures was not significantly decreased in the study of Liberman et al. (68). In the multinational study by Pols et al., which captured clinical nonvertebral fractures as adverse events, the risk of new clinical nonvertebral fractures was significantly reduced by 47% at 1 yr with alendronate (70). In the FIT, alendronate did not significantly reduce the risk of total nonvertebral fractures in women with prevalent vertebral fractures (69) or in women with no prevalent vertebral fractures (Ref. 55 and Table 6). Although the total nonvertebral fracture risk was not significantly changed, women with prevalent vertebral fractures treated with alendronate had a significant 51% (P = 0.047) decreased risk of hip fracture (69). Among women in the FIT with no prevalent vertebral fractures, the risks for any nonvertebral fracture and for hip fractures in the alendronate group were not significantly different from placebo (Tables 5 and 6 and Ref. 55). However, a post hoc analysis found a significant 56% (95%, CI 3–82%) decrease in the risk of hip fractures in alendronate-treated women who had baseline femoral neck BMD T-scores less than or equal to -2.5, with no risk reduction in women whose femoral neck BMD T-scores were greater than -2.5 (55). The risk of nonvertebral fractures was decreased by 27% (95% CI, 13–39%), and the risk of hip fractures was decreased by 53% (95%CI, 21–74%) with alendronate treatment in women with osteoporosis combined from both arms of the FIT who had prevalent vertebral fractures or baseline femoral neck BMD T-scores less than or equal to -2.5 (71, 72). Therefore, hip fracture efficacy was demonstrated in analyses of subgroups of women with high fracture risk, although efficacy of alendronate on the risk of total nonvertebral fractures was not demonstrated in all trials.

In the dose-ranging study of Liberman et al. (68), 10 mg of daily alendronate increased BMD at the lumbar spine, femoral neck, and total body BMD by 8.8%, 5.9%, and 2.5%, respectively, at 3 yr compared with placebo. In women with and without prevalent vertebral fractures enrolled in the FIT (55, 69), alendronate treatment for an average of 2.9–4.2 yr significantly increased lumbar spine BMD by 6.2–6.6%, femoral neck BMD by 4.1–4.6%, and total body BMD by 1.8–2.0% compared with placebo. In the multinational study by Pols et al. (70), alendronate therapy increased BMD in the lumbar spine and femoral neck after 1 yr by 4.9% and 2.4%, respectively, compared with placebo.

Thus, the clinical trial evidence shows that alendronate reduces the risk of vertebral fractures in postmenopausal women with osteoporosis defined by prevalent vertebral fracture or low BMD. However, the reductions in risk of total nonvertebral fractures with alendronate were variable. Although the risk of total nonvertebral fractures was not significantly reduced in women with or without prevalent vertebral fractures in FIT (55, 69), further analyses found significant reductions in hip fracture risk for women with baseline femoral neck BMD T-score less than -2.5 or with prevalent fractures (72). The risk of nonvertebral fractures was significantly decreased by 27% in the combined group of women with osteoporosis defined by prevalent vertebral fracture or a baseline femoral neck BMD T-score less than -2.5 (72). Alendronate significantly decreased the incidence of nonvertebral fractures by 47% in the multinational study of Pols et al. (70) and by 27% in a meta-analysis of several smaller alendronate studies (73).

2. Risedronate.
The 3-yr Vertebral Efficacy with Risedronate Therapy (VERT) trial (Table 4 and Refs. 74 and 75), conducted at North American and multinational (European and Australian) sites, investigated the efficacy of risedronate in the prevention of new vertebral fractures in women with prevalent vertebral fractures. Several differences in the study population and design of the VERT study distinguish it from similar trials with alendronate and raloxifene. Women enrolled in the VERT trial had more prevalent vertebral fractures per person (74, 75) than women in the prevalent fracture arms of the raloxifene or alendronate trials (55, 62, 68, 69). The VERT study protocol did not exclude women with recent peptic ulcers, ulcers that required hospitalization, or dyspepsia requiring daily treatment, unlike the alendronate fracture trial. Spinal radiographs were obtained at baseline and annually during the VERT study, unlike the alendronate and raloxifene studies, in which radiographs were performed at baseline and annually after 2 yr. The risedronate trials used a 15% decrease in vertebral height to define a fracture, unlike the raloxifene, alendronate, and calcitonin trials, which defined a fracture as a decrease of at least 20%. The discontinuation rate was 42% for both risedronate trials, which was higher than that observed in the raloxifene and alendronate trials, which ranged from 11% to 23% (Table 4).

The results from the North American and multinational sites were similar (Tables 5 and 6). In prespecified analyses, risedronate therapy decreased the cumulative RRs of new morphometric vertebral fractures by 61–65% in the first year and by 41–49% at 3 yr (74, 75). Risedronate therapy decreased the absolute risks of new vertebral fractures by 4.0% in the first year and by 5.0% at 3 yr in the North American study, and by 10.9% in the first year and by 7.4% at 3 yr in the multinational study (Table 5). The risedronate group had less height loss than the placebo group in the multinational trial. At 3 yr, risedronate significantly decreased the risk of nonvertebral fractures compared with placebo by 39% in the North American study, but the 33% decreased risk observed in the multinational study was not significant. The multinational study population was half the size of the North American population, and combined with the high discontinuation rate, there may have been insufficient numbers of women in the multinational study to detect a significant reduction in nonvertebral fracture risk. The risks of fractures at individual nonvertebral sites were not significantly decreased in either the North American or multinational study. Risedronate treatment increased lumbar spine BMD by 4.3–5.9% and femoral neck BMD by 2.8–3.1%, respectively, compared with placebo.

A separate trial examined the effects of risedronate on hip fracture risk in 5,445 women aged 70–79 yr with osteoporosis defined by low femoral neck BMD, and 3,886 women over 80 yr of age who had at least one clinical risk factor for hip fracture (56). These risk factors included low femoral neck BMD or nonskeletal risk factors such as advanced age, difficulty standing, poor gait, previous fall-related injury, poor hand-eye coordination, smoking, previous hip fracture or maternal history of hip fracture, or long hip axis length. The preplanned analyses and statistical power were specified for each dose of risedronate as in the VERT trial, but the analysis plan was subsequently changed to pool women assigned to either dose. Risedronate significantly decreased the overall cumulative risk of hip fracture by 30% (Table 6). The risk of hip fracture was reduced by 40% in the women with osteoporosis (mean age, 74 yr; femoral neck BMD T-score -2.9 to -2.7 by the Third National Health and Nutrition Examination Survey criteria). Further analyses revealed a 60% decrease in the risk of hip fractures in women who had osteoporosis by BMD definition and prevalent vertebral fractures, but no significant reduction was observed in women who had osteoporosis by BMD definition without prevalent vertebral fractures. There was also no significant reduction in risk for hip fractures in the older women (mean age of 83 yr) who were enrolled on the basis of clinical risk factors for hip fracture, such as frequent falling (56). This was the first large prospective trial designed specifically with adequate statistical power to investigate hip fracture efficacy. However, issues in study design limit the conclusions. For example, only 31% of subjects in the older group had BMD measurements, and there were no data on falls. Also, the discontinuation rates were 43% in the younger group and 59% in the older group of women.

In summary, risedronate has vertebral fracture efficacy in women with prevalent vertebral fractures and hip fracture efficacy in a subset of elderly women with confirmed osteoporosis. However, the effects of risedronate on total nonvertebral fracture risk were not demonstrated in all trials. The efficacy of risedronate on clinical vertebral fractures and on vertebral, nonvertebral, or clinical vertebral fractures in women with osteoporosis but without prevalent vertebral fractures has not been reported.

D. Salmon calcitonin nasal spray
Salmon calcitonin, administered by nasal spray, slows bone loss by inhibiting osteoclast-mediated bone resorption (76). It is indicated for the treatment of osteoporosis in women who are at least 5 yr postmenopausal and refuse or cannot tolerate estrogens, or in whom estrogens are contraindicated. The Prevent Recurrence of Osteoporotic Fractures [PROOF (Ref. 77 and Tables 4 and 5)] trial studied the effects of salmon calcitonin nasal spray (100, 200, and 400 IU/day) on fracture risk in postmenopausal women with low lumbar spine BMD and 1–5 prevalent vertebral fractures. Subsequent adjudication of the radiographs revealed that 269 (21%) of the women enrolled had no prevalent vertebral fracture, and 65 women had more than 5 fractures. In addition, other aspects of the PROOF study design may have affected the study results (77, 78).

Although 200 IU/day calcitonin reduced the cumulative RR of new morphometric vertebral fractures by 33% at 5 yr (Table 5), the 100- and 400-IU/day doses did not significantly decrease the vertebral fracture risk. Similar results were observed in those women who received calcitonin for at least 3 yr or who had a fracture during the first 3 yr of treatment. Calcitonin at 200 IU/day decreased the absolute risk of new vertebral fractures by 8.1%. None of the calcitonin doses significantly decreased the risk of two or more new vertebral fractures. The effects of calcitonin on clinical vertebral fractures have not been reported. The 100-IU/day group had significant risk reductions of 36% and 90% for total nonvertebral and hip fractures, respectively, whereas the risks for these fractures were not significantly changed in the 200- and 400-IU/day groups (Table 6 and Ref. 77). These results may be related to chance, given the small numbers of hip and nonvertebral fractures, the high discontinuation rate, and the absence of a significant effect in the 200-IU/day and 400-IU/day groups in this study.

Calcitonin treatment at doses ranging from 100 to 400 IU/day significantly increased lumbar spine BMD from baseline from 1.0% to 1.5%, whereas those in the placebo group had a 0.5% increase after 5 yr (77). These increases in lumbar spine BMD differed significantly from the placebo group only in the first 2 yr for all calcitonin doses and in the first 3 yr for the 400-IU/day dose. None of the calcitonin doses significantly increased femoral neck and trochanter BMD.

Several aspects in the design and conduct of the PROOF trial have been criticized and may have contributed to the inconsistent fracture risk results (78). The study discontinuation rate was substantial, with 59% of the 1,255 women enrolled discontinuing over 5 yr (77). For all fracture and BMD endpoints, statistical analyses were performed as intention-to-treat, with the last available post-baseline observation carried forward to the end of the 5-yr study. Given the high discontinuation rate, it is not known if results obtained in the early years of the trial and carried forward are reflective of the actual efficacy at the end of the trial (78). The PROOF investigators were not blinded to the BMD results, and because calcitonin produced small BMD changes, this apparent lack of efficacy may have caused some women to discontinue the study and to begin other therapy, contributing to the high overall discontinuation rate (78). The numbers of patients who discontinued due to "ineffective study drug" were similar in the placebo group and all treatment groups (77). These faults in the design and conduct of the PROOF trial make it difficult to draw any definitive conclusions about the efficacy of salmon calcitonin nasal spray on fracture risk.

E. Other antiresorptive therapies
There are several antiresorptive agents that, although not approved for osteoporosis prevention and/or treatment in the United States, are approved for osteoporosis indications and used widely in other parts of the world.

Etidronate, a first-generation bisphosphonate, was not approved in the United States because the 3-yr data from the largest RCT did not show efficacy in vertebral fracture reduction and also because of concerns about osteomalacia with long-term use (79). The pivotal trial studied the effects of 400 mg/day etidronate administered in 2-wk cycles every 91 d for 2 yr in 423 postmenopausal women with 1–4 baseline vertebral fractures, defined by a 20% reduction in vertebral height (80). Phosphate was administered to some patients for 3 d before the etidronate phase and 500 mg/day calcium was administered after the etidronate phase in each cycle. In this study, 9.3% of control patients sustained a new fracture, compared with 4.1% of etidronate patients (P = 0.004). The discontinuation rate at 2 yr was 14%. Compared with placebo, etidronate increased lumbar spine BMD by 3–4%, but there were no significant changes in femoral neck or forearm BMD. The addition of phosphate provided no additional benefit. There were few osteoporotic nonvertebral fractures. Of the original cohort, 68% entered an optional third year extension of this study, and the cumulative reduction in vertebral fractures was no longer significant in the entire cohort after 3 yr (81). The vertebral fracture rate was stable in the control subjects but increased in the etidronate-treated subjects in the third year. In a post hoc analysis, patients who were at high risk for fracture, having low spine BMD and three or more vertebral fractures at baseline, had significantly reduced risk of vertebral fractures. Several smaller studies of etidronate, although demonstrating increases in BMD, also failed to show significant reductions in vertebral fracture risk (82, 83). Etidronate continues to be widely prescribed outside the United States because of its excellent safety profile, ability to increase BMD, and low cost (79). As the pivotal study with etidronate apparently had insufficient statistical power for the fracture endpoint, vertebral fracture efficacy of etidronate cannot be concluded from the present clinical trial evidence (79, 84).

Tibolone is a steroid analog with estrogenic, progestogenic, and androgenic properties. It is available in some countries outside of the United States for relief of climacteric symptoms, such as hot flashes, and for the prevention of osteoporosis related to estrogen deficiency (85). As summarized recently in several studies, tibolone significantly increased BMD in postmenopausal women compared with either baseline or placebo (85). To date, there are no RCTs in postmenopausal women demonstrating any effects of tibolone on fracture risk.

Synthetic vitamin D analogs, including calcitriol (1,25-dihydroxyvitamin D) and alfacalcidol (1-hydroxyvitamin D), are used in some countries for osteoporosis indications. Clinical trials assessing the fracture efficacy of these vitamin D analogs in postmenopausal women with osteoporosis have produced conflicting results, due to the small size of the trials and different criteria for subject enrollment and efficacy. The largest trial evaluated the fracture efficacy of calcitriol, 0.25 µg twice daily for 3 yr, in a single-blind study of 622 women with at least one vertebral fracture at baseline. The rate of new vertebral fractures, expressed as fractures per 100 patient-years, was significantly reduced in the calcitriol group (9.3) compared with the calcium control group (25.0) at 2 yr, with similar fracture rates observed in each group at 3 yr (86). There are several weaknesses in the design of this study (87). The investigators and subjects were unblinded to treatment assignment after randomization, and this may have led to biases in assessing adverse events and study discontinuation, which was 31% at 3 yr. Additionally, biochemical markers of calcium metabolism, which may be affected by calcitriol, were not measured in all subjects during the study, and the criterion for a vertebral fracture was a 15% decrease in vertebral height. In smaller trials, which enrolled fewer than 100 postmenopausal women in each study, calcitriol therapy did not significantly reduce vertebral fracture rates after 2 or 3 yr of therapy (88, 89, 90, 91). The efficacy of alfacalcidol was examined in two studies, which enrolled approximately 80 women each. Alfacalcidol increased BMD and decreased the vertebral fracture rate after 1 (92) or 2 (93) yr of therapy. Except for the trial by Tilyard et al. (86), none of the other trials with calcitriol or alfacalcidol had adequate statistical power to determine vertebral fracture efficacy. Based upon the current clinical trial evidence, the efficacy of synthetic vitamin D analogs on reduction of risk for vertebral fractures remains uncertain.

Phytoestogens are plant compounds with biological activities similar to estrogen and are widely available naturally or as nonprescription dietary supplements. They consist of two main classes: isoflavones and lignans (94). The incidence of osteoporosis-related fractures is lower in Asian countries than in Western countries (95), and some have attributed this difference to the phytoestrogen-rich Asian diet. Very few prospective clinical trials have studied the effects of phytoestrogens on postmenopausal osteoporosis. As reviewed by Tham et al. (94), several clinical studies showed that the isoflavone derivative ipriflavone also prevented bone loss in postmenopausal women. In contrast, a recent RCT found ipriflavone to have no effects on BMD in 474 postmenopausal women with low bone mass (96). To date, there is no clinical trial evidence that phytoestrogens in any form reduce the risk of osteoporotic fractures (97).

F. Antiresorptive therapies in combination
Recent RCTs have assessed some effects of antiresorptive agents administered in combination in postmenopausal women. The efficacy of combining HRT with other antiresorptive agents has been reviewed (98). Only one of the prospective RCTs of antiresorptive agents in combination had a prespecified secondary endpoint of fracture (99), and none of these small studies had adequate statistical power to demonstrate fracture risk reduction.

A 2-yr study of alendronate and conjugated equine estrogens alone and in combination found greater increases in lumbar spine and femoral neck BMD in the combination group compared with each agent alone (100). The incidences of clinical fractures were not significantly different between groups, possibly due to the small number of fracture events. Compared with women additionally given placebo, postmenopausal women with osteoporosis who were already taking HRT experienced greater increases in lumbar spine and femoral neck BMD when additionally treated with alendronate for 1 yr (101). Again, there was no significant difference in the small number of fractures between groups. In a 1-yr study of postmenopausal women with osteoporosis, raloxifene and alendronate in combination increased lumbar spine and femoral neck BMD to a greater extent than either agent alone, but fractures were not assessed (102). Cyclic etidronate combined with percutaneous HRT (103) in healthy early postmenopausal women, or with daily oral HRT in older postmenopausal women (99), increased lumbar spine and femoral neck BMD to a greater extent than either therapy alone. The rate of new vertebral fracture in the etidronate/oral HRT group was not significantly different from that of the control group.

The possible excessive suppression of bone turnover with combination therapy is a safety concern. Although high bone turnover before treatment has been correlated with an increased fracture risk, there are reported associations between low bone turnover and increased fracture risk in women (104). These observational studies are supported by animal studies showing that high doses of bisphosphonates are associated with suppression of bone turnover, accumulation of microdamage, and deficiencies in bone material properties (105). Severe decreases in bone turnover ("frozen bone") may lead to increased fracture rates, due to impaired remodeling and the accumulation of microdamage (106). It has been suggested that incorporation of bisphosphonates, which have a long skeletal half-life, may inhibit bone remodeling to the extent that bone damage is not repaired (107). Combination therapy, with more extensive suppression of bone turnover, might contribute further to this problem. This hypothetical concern has not been confirmed in clinical trials with any antiresorptive agent.

G. Risks and extraskeletal benefits
Selection of an antiresorptive agent depends on clinical trial efficacy data for skeletal benefits, as well as associated risks and extraskeletal benefits. HRT is often used to treat symptoms of estrogen deficiency, such as hot flashes and vaginal dryness (108). HRT improved the serum lipid profile in the Postmenopausal Estrogen/Progestin Interventions (PEPI) trial (109), but the HERS study (8) in women with established coronary disease failed to confirm the expected decrease in risk of coronary heart disease events with HRT. The benefits of ERT and HRT in preventing coronary heart disease events remain to be addressed by such large prospective studies as the Women’s Health Initiative (60). The effects of HRT on cognitive function (110), Alzheimer’s disease (111), and the risk of breast cancer (112) remain controversial. Grady and Furberg (113) noted an approximately 3-fold increased risk in venous thromboembolism with HRT. Several studies have reported increases in breast tenderness, vaginal bleeding, and endometrial hyperplasia with estrogen use (51). The risks of adverse events associated with estrogen use (114) may have possible effects on adherence (115).

The beneficial effects of raloxifene on risk factors for cardiovascular disease, such as serum lipids, coagulation factors, and homocysteine, have been demonstrated in RCTs (116, 117). The relationships of these surrogate risk factors to the prevention of cardiovascular disease events are being studied in the large prospective Raloxifene Use for the Heart trial (118). A 3-fold increased risk of venous thromboembolism, similar to the risk noted in the estrogen trials, was observed in the raloxifene fracture trial (62). Although raloxifene use increased the incidence of hot flashes and leg cramps, the incidence of vaginal bleeding and breast pain was not different from placebo (62, 119). The occurrence of breast cancer was monitored in the raloxifene fracture trial, and women receiving raloxifene had a 76% decreased risk for invasive breast cancer and a 90% decreased risk for invasive estrogen receptor-positive breast cancer at 40 months of follow-up (119). Similar results were observed at 48 months (119A ). The long-term effects of raloxifene on the risk of breast cancer are being directly compared with tamoxifen in the 5-yr Study of Tamoxifen and Raloxifene trial (120) of women with an increased risk for breast cancer.

There are no documented extraskeletal benefits of bisphosphonates. The most commonly reported adverse events associated with bisphosphonate therapy are upper gastrointestinal disorders, such as abdominal pain, esophagitis, or esophageal ulcer (68). In the alendronate fracture trials, the proportion of women who reported any upper gastrointestinal adverse event ranged from 20% to 47%, but the incidence of these events and the rates of discontinuation due to these events were not different between the placebo and alendronate groups in any trial (55, 69, 70). In contrast to the clinical trial findings, post-marketing studies have found that one-third of alendronate users reported gastrointestinal adverse events (121), and some patients may develop chemical esophagitis, including severe ulcerations, with alendronate use (122). The increased occurrence of gastrointestinal events in post-marketing use may be explained by exclusion of women with previous gastrointestinal disorders in the clinical trials and/or by poor adherence to administration directions by some patients in clinical practice. The actual incidence of gastrointestinal adverse events reported in general clinical use remains controversial (123).

In the risedronate fracture trials, the incidence of upper gastrointestinal adverse events ranged from 26% to 30% (74, 75). The incidence of these adverse events and overall discontinuation rates were not significantly different between the placebo and risedronate groups in these trials (124). It is difficult to compare of the incidence of gastrointestinal adverse events between bisphosphonates, as women with active gastrointestinal disease or who were taking certain medications were excluded from the alendronate trials (55, 69) but not from the risedronate trials (74). There are no post-marketing data on the occurrence of gastrointestinal adverse events with risedronate. The incidence of adverse events in the fracture trial of salmon calcitonin nasal spray was not different between groups (77). Although analgesia is an extraskeletal benefit associated with calcitonin use (125), this finding has not been confirmed in any large RCT.

	IV. Discussion

Top Abstract I. Introduction II. Calcium and Vitamin... III. Pharmacological Therapies... IV. Discussion References

 
There are no head-to-head studies comparing the antifracture efficacy of any antiresorptive therapies. Fracture efficacy comparisons of antiresorptive agents are fraught with problems, due to differences in study characteristics across clinical trials. Despite this, the observed reductions in vertebral fracture risk are similar across trials with different antiresorptive agents (Fig. 1). Quantitative comparisons of nonvertebral fracture efficacy are more problematic (Fig. 2), because the frequencies of these fractures at individual sites are very low (Table 6), and most of these trials were designed to have statistical power for vertebral fracture risk reduction. The fracture efficacy of ERT or HRT is primarily supported by observational studies (Table 3), and the results of these trials cannot be compared with the large RCTs of other antiresorptive agents. Study design characteristics that may contribute to the fracture efficacy outcomes for raloxifene, alendronate, risedronate, and calcitonin are outlined in Tables 4–6. Although it is challenging to compare and rank the large RCTs of raloxifene, alendronate, risedronate, and calcitonin on the basis of study design, the MORE and FIT trials probably have the best design. Interpretation of results for nonvertebral fracture risks is limited to the one trial designed to study nonvertebral fractures as a primary endpoint. The risedronate trials were not as well designed, as the VERT trials lacked a significant number of osteoporosis patients without prevalent vertebral fractures and had a high discontinuation rate. The risedronate hip fracture trial did not collect BMD data in most of the women who were enrolled based upon clinical risk factors for hip fracture and did not collect data regarding falls. The PROOF trial had the poorest design for reasons previously discussed.

View larger version (27K): [in this window] [in a new window]   Figure 1. Increases in lumbar spine BMD compared with placebo (left panel), obtained at the final study endpoint of large RCTs of various antiresorptive agents and corresponding to decreases in RR for incident radiological vertebral fractures (VFx; right panel). These trials were all designed with adequate statistical power to demonstrate a reduction in the risk of new radiological vertebral fractures. Across clinical trials, the effects of antiresorptive therapies on vertebral fracture risk are similar and are not directly associated with the increase in BMD. However, the data for calcitonin may be less reliable due to weaknesses in design of the PROOF trial.

View larger version (29K): [in this window] [in a new window]   Figure 2. RRs for incident nonvertebral fractures in large RCTs of various antiresorptive agents. In these trials, the occurrence of nonvertebral fractures was a secondary endpoint or was captured as an adverse event.