Early Changes in Biochemical Markers of Bone Turnover Predict the Long-Term Response to Alendronate Therapy in Representative Elderly Women: A Randomized Clinical Trial

Authors

  • Susan L. Greenspan,

    Corresponding author
    1. Division of Bone and Mineral Metabolism, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    2. Division of Gerontology, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    3. Charles A. Dana Research Institute, General Clinical Research Center, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    • Susan L. Greenspan, M.D., Beth Israel Deaconess Medical Center, 330 Brookline Avenue (GZ-800), Boston, MA 02215 U.S.A.
    Search for more papers by this author
  • Robert A. Parker,

    1. Charles A. Dana Research Institute, General Clinical Research Center, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    2. Biometrics Center, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Lauren Ferguson,

    1. Division of Bone and Mineral Metabolism, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Harold N. Rosen,

    1. Division of Bone and Mineral Metabolism, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    2. Division of Gerontology, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Lauri Maitland-Ramsey,

    1. Division of Bone and Mineral Metabolism, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • David B. Karpf

    1. Roche Global Development, Palo Alto, California, U.S.A.
    Search for more papers by this author

Abstract

Although the antiresorptive agent alendronate has been shown to increase bone mineral density (BMD) at the hip and spine and decrease the incidence of osteoporotic fractures in older women, few data are available regarding early prediction of long-term response to therapy, particularly with regard to increases in hip BMD. Examining short-term changes in biochemical markers incorporates physiologic response with therapeutic compliance and should provide useful prognostic information for patients. The objective of this study was to examine whether early changes in biochemical markers of bone turnover predict long-term changes in hip BMD in elderly women. The study was a double-blind, placebo-controlled, randomized clinical trial which took place in a community-based academic hospital. One hundred and twenty community-dwelling, ambulatory women 65 years of age and older participated in the study. Intervention consisted of alendronate versus placebo for 2.5 years. All patients received appropriate calcium and vitamin D supplementation. The principal outcome measures included BMD of the hip (total hip, femoral neck, trochanter, and intertrochanter), spine (posteroanterior [PA] and lateral), total body, and radius. Biochemical markers of bone resorption included urinary N-telopeptide cross-linked collagen type I and free deoxypyridinoline; markers of bone formation included serum osteocalcin and bone-specific alkaline phosphatase. Long-term alendronate therapy was associated with increased BMD at the total hip (4.0%), femoral neck (3.1%), trochanter (5.5%), intertrochanter (3.8%), PA spine (7.8%), lateral spine (10.6%), total body (2.2%), and one-third distal radius (1.3%) in elderly women (all p < 0.01). In the placebo group, bone density increased 1.9–2.1% at the spine (p < 0.05) and remained stable at all other sites. At 6 months, there were significant decreases in all markers of bone turnover (–10% to –53%, p < 0.01) in women on alendronate. The changes in urinary cross-linked collagen at 6 months correlated with long-term bone density changes at the hip (r = –0.35, p < 0.01), trochanter (r = –0.36, p < 0.01), PA spine (r = –0.41, p < 0.01), and total body (r = –0.34, p < 0.05). At 6 months, patients with the greatest drop in urinary cross-linked collagen (65% or more) demonstrated the greatest gains in total hip, trochanteric, and vertebral bone density (all p < 0.05). A 30% decrease in urinary cross-linked collagen at 6 months predicted a bone density increase of 2.8–4.1% for the hip regions and 5.8–6.9% for the spine views at the 2.5-year time point (p < 0.05). There were no substantive associations between changes in biochemical markers and bone density in the placebo group. Alendronate therapy was associated with significant long-term gains in BMD at all clinically relevant sites, including the hip, in elderly women. Moreover, these improvements were associated with early decreases in biochemical markers of bone turnover. Early dynamic decreases in urinary cross-linked collagen can be used to monitor and predict long-term response to bisphosphonate therapy in elderly women. Future studies are needed to determine if early assessment improves long-term patient compliance or uncovers poor compliance, thereby aiding the physician in maximizing the benefits of therapy.

INTRODUCTION

OSTEOPOROSIS IS A MAJOR public health problem that primarily affects elderly women.1,2 In addition to hormone replacement therapy,3–5 several antiresorptive agents have recently become available to prevent or treat osteoporosis.6–9 Alendronate, a potent aminobisphosphonate, has been shown to increase bone mineral density (BMD) at the spine, hip, and total body, and reduce fractures of the vertebrae, hip, and wrist by ∼50% in women with osteoporosis or low bone mass.6,8,9 However, few data are available regarding early prediction of long-term responses to this therapy in older women.

Biochemical markers of bone turnover provide information on bone resorption and formation, correlate with the rate of bone loss,10 and predict the likelihood of hip fracture.11 One investigation in postmenopausal women (mean age 63 ± 6 years) found an association between baseline biochemical markers and spinal BMD after 2 years on alendronate or placebo.12 There was also an association between changes in biochemical markers at 6 months and increased spinal BMD at 2 years. However, there are no data with regard to prediction of long-term hip BMD outcomes based on levels of markers at baseline or early changes in biochemical markers. Because the hip is the site associated with the greatest morbidity, mortality, and cost in terms of postmenopausal osteoporosis,1,2 and because hip BMD has the greatest predictive value for hip fractures,13 such data are needed.

Although a static baseline assessment provides information on bone turnover prior to therapy, we postulated that early changes in biochemical markers represent a dynamic response to treatment and would predict long-term changes in hip bone density in older women. Bisphosphonates, such as alendronate, are poorly absorbed and have a somewhat complex administration schedule.14 Thus, an examination of dynamic changes after 6 months of therapy incorporates therapeutic compliance with the patient's physiologic response. To investigate this hypothesis, we examined early changes in biochemical markers of bone turnover in a double-blind, placebo-controlled, randomized clinical trial of alendronate versus placebo in elderly women. To allow for greater relevance to the general population of older women, participants were not selected with strict health or bone density criteria.

MATERIALS AND METHODS

Subjects

We recruited 120 unselected healthy, ambulatory, community-dwelling women age 65 years of age and older from the Greater Boston area via advertisement. Entry criteria were not based on BMD. Potential subjects were excluded if they had a history of any illness affecting bone and mineral metabolism (e.g., renal failure, hepatic failure, active malignancy, current hyperthyroidism or hyperparathyroidism, or malabsorption), were currently taking medications known to affect bone metabolism (e.g., glucocorticoids, anticonvulsants), or had been treated for osteoporosis with bisphosphonates, hormone replacement therapy, or calcitonin within 1 year of screening. This study was approved by the Committee on Clinical Investigations at the Beth Israel Deaconess Medical Center. Subjects were advised of the nature of the study and informed consent was obtained.

Study design

For this 2–1/2 year, double-blind trial, women were randomly assigned to receive placebo or alendronate, 5 mg/day. In November 1993, when data became available that the 10 mg dose of alendronate produced greater increases in bone density than the 5 mg dose, we increased the treatment group's dose of alendronate from 5 mg/day to 10 mg/day for the final year of the study. A food frequency questionnaire was utilized to estimate dietary calcium intake.15 Subjects with daily dietary intakes of <1000 mg received one or more daily supplemental tablets (Oscal + D) containing 250 mg of elemental calcium and 125 IU of vitamin D to provide a total daily calcium intake of at least 1000 mg.

Outcome variables

Bone mineral density:

BMD of the hip (femoral neck, trochanter, intertrochanter, and total hip), lumbar spine (posteroanterior [PA] and lateral), total body, and radius (total, ultradistal, mid- and one-third distal radius) were measured by dual-energy X-ray absorptiometry (QDR-2000, Hologic, Inc., Waltham, MA, U.S.A.) at baseline and every 6 months thereafter for a total of 30 months. As previously reported, the coefficients of variation (CVs) of BMD in elderly women (mean age 71 ± 7 years) on this densitometer are 1.7, 1.5, 1.2, and 1.9% for the lateral spine, PA spine, total hip, and femoral neck, respectively.16,17

Clinical characteristics:

Height was measured to the nearest millimeter at baseline and every 6 months thereafter using a Harpenden stadiometer (Holtain Ltd., Crymych, Dyfed, U.K.). At each visit, we obtained three measurements and calculated the average. Weight was measured by an ACME digital in-bed scale (Model 0501; ACME Medical Scale Co., San Leandro, CA, U.S.A.) at baseline and every 6 months thereafter. Body mass index was calculated as kilograms per square meter.

Biochemical markers:

Following an overnight fast, we obtained a serum sample and a second-void 2-h urine collection (6:00–8:00 a.m.). All samples were frozen at –20°C after collection unless otherwise specified. Urinary N-telopeptide cross-linked collagen type I (NTx, nmol bone collagen equivalents [BCE]/mmol creatinine [Cr]) was measured with an enzyme-linked immunosorbent assay (Osteomark, Ostex International, Seattle, WA, U.S.A.; intra-assay CV 5–19%),18 as was urinary-free deoxypyridinoline (Dpd, nmol/mmol creatinine; Pyrilinks-D, Metra Biosystems, Mountain View, CA, U.S.A.; intra-assay CV 4.3–8.4%).19

Serum tests for bone formation included intact osteocalcin (OC, Novocalcin, ng/ml; Metra Biosystems; intra-assay CV 4.8–10.0%) and bone-specific alkaline phosphatase (BAP, Alkphase-B, U/L; Metra Biosystems; intra-assay CV 3.9–5.8%).20 Parathyroid hormone (pg/ml) was measured by Allegro Immunoradiometric Assay (Nichols Allegro, San Juan Capistrano, CA, U.S.A.; intra-assay CV 1.8–3.4%) on serum that had been frozen separately at –80°C immediately after collection. Additional serum measurements included calcium, albumin, phosphate, and 25-hydroxyvitamin D.

Statistical analysis:

Descriptive statistics are presented as mean ± SD unless otherwise noted. Comparisons over time within subjects were assessed using the Wilcoxon signed rank test, and comparisons between groups at specific time points used the Wilcoxon rank sum test for two groups and the Kruskal-Wallis test for more than two groups. Patients were analyzed with an intention-to-treat analysis. For patients discontinuing during the study, we carried the last available measurement forward to subsequent time points, an approach that would tend to underestimate the true change occurring at 2.5 years in the treatment group. Similar results were found when we did not carry measurements forward (i.e., used only the actual data at a time point as per protocol). Spearman rank correlation was used for preliminary assessment of the relationship between variables. Following this, we classified all alendronate-treated patients by the percentage decrease in NTx at 6 months into small (≤45% decrease), moderate (45–65% decrease, exclusive), and large (≥65% decrease) to assess the change in BMD after 30 months of treatment. Patients not followed for at least 6 months were not included in these analyses. We also calculated the predicted change in BMD after 30 months for a 30% decrease in NTx based on a standard linear regression model. Confidence intervals for these estimates are given for the predicted change in a population.

RESULTS

There were no statistically significant differences in baseline clinical characteristics or biochemical markers of bone turnover between the treatment and control groups (Table 1). Baseline BMD measurements were also similar, with the exception of mean intertrochanteric BMD, which was higher in the placebo group compared with the alendronate group (0.930 ± 0.112 vs. 0.885 ± 0.164 g/cm2 placebo vs. alendronate, respectively, p < 0.05; Table 1). Baseline BMD characteristics have previously been reported17 and reveal that over half of these women had osteoporosis by World Health Organization (WHO) criteria, or BMD > 2.5 SD below peak bone mass (55% using femoral neck BMD, 66% using lateral spine BMD). Of the 120 women who were enrolled in the study, 75% completed the study (77% treatment group, 73% placebo group).

Table TABLE 1. BASELINE CLINICAL CHARACTERISTICS
original image

Following treatment with alendronate, mean BMD increased by 4.0% at the total hip, 3.1% at the femoral neck, 5.5% at the trochanter, 3.8% at the intertrochanter, 7.8% at the PA spine, 10.6% at the lateral spine, 2.2% at the total body, and 1.3% at the one-third distal radius (all p < 0.01; Figs. 1 and 2). In the placebo group, which received calcium plus vitamin D, mean BMD remained stable at the total body, radius and all regions of the hip; significant increases of 2.1% and 1.9% occurred at the PA and lateral spine, respectively (p < 0.01 and 0.05). Treatment with alendronate significantly increased mean BMD compared with placebo at all sites (all p < 0.01; Figs. 1 and 2). Furthermore, hip BMD decreases were noted in only 10% of alendronate-treated patients compared with 50% of patients on calcium and vitamin D alone (p < 0.01). As expected, there were no significant differences in the report of wrist or hip fractures between the two groups over 2.5 years. There was one hip fracture in the placebo group but none in the alendronate group; no wrist fractures occurred in subjects on placebo, but three were reported in subjects on alendronate. Vertebral fractures were not assessed because spinal X-rays were not performed due to the a priori low power of demonstrating a treatment effect.

Figure FIG. 1.

Mean change in BMD from baseline at (A) total hip, (B) femoral neck, (C) trochanter, and (D) intertrochanter in women receiving alendronate or placebo for 30 months. Results as mean ± SEM. *p < 0.05, †p < 0.01 from baseline, ‡p < 0.05, §p < 0.01 alendronate versus placebo.

Figure FIG. 2.

Mean change in BMD from baseline at (A) PA spine, (B) lateral spine, (C) total body, and (D) one-third distal radius in women receiving alendronate or placebo for 30 months. Results as mean ± SEM.*p < 0.05, †p < 0.01 from baseline, ‡p< 0.05, §p < 0.01 alendronate versus placebo.

At 6 months, there were significant decreases in mean NTx (–53%), Dpd (–10%), OC (–20%), and BAP (–24%) in the alendronate-treated group (all p < 0.01). Corresponding changes in the placebo group were less impressive (Fig. 3), although there were significant decreases in NTx (–14%, p < 0.01) and BAP (–9%, p< 0.01) at 30 months and OC (–8%, p < 0.05) at 24 months (Fig. 3). Significant differences between mean levels of biochemical markers in the alendronate and placebo groups occurred at 6 months (all p < 0.01), a difference that continued for 2.5 years (Fig. 3).

Figure FIG. 3.

Mean change from baseline in markers of bone turnover for (A) NTx, (B) Dpd, (C) OC, and (D) BAP in women receiving alendronate or placebo for 30 months. Results as mean ± SEM. *p < 0.05, †p < 0.01 from baseline, ‡p < 0.05, §p < 0.01 alendronate versus placebo.

There were relatively few significant correlations between baseline biochemical markers and subsequent changes in BMD. However, baseline levels of NTx were associated with long-term changes in total body, vertebral, and trochanteric BMD in women on alendronate. Baseline levels of NTx were correlated with total body BMD at 18 months (r = 0.36, p < 0.01), 24 months (r = 0.34, p < 0.01), and 30 months (r = 0.32, p< 0.05). Baseline levels of NTx were also correlated with PA spine BMD at 12 months (r = 0.28, p < 0.05) and lateral spine at 18 months (r = 0.27, p < 0.05). Baseline NTx and trochanteric BMD at 24 months were also positively correlated (r = 0.28, p < 0.05).

In contrast, changes in several biochemical markers at 6 months were consistently associated with changes in BMD over 2.5 years in women treated with alendronate. Decreases in urinary NTx significantly correlated with long-term changes at the total hip (r = –0.35, p < 0.01), trochanter (r = –0.36, p < 0.01), femoral neck (r = –0.28, p< 0.05), PA spine (r = –0.41, p < 0.01), and total body BMD (r = –0.34, p < 0.05, Table 2). Decreases at 6 months in OC correlated with total hip, trochanter, intertrochanter, PA spine, and lateral spine BMD increases at 2.5 years (r = –0.31 to –0.43, p < 0.05; Table 2). In the placebo group, there were no consistent associations. However, decreases in Dpd at 6 months were correlated with BMD changes at the intertrochanter (r = –0.29, p < 0.05) and decreases in BAP were correlated with the long-term changes in total body BMD (r = –0.32, p < 0.05).

Table TABLE 2. CORRELATIONS OF DECREASED BIOCHEMICAL MARKERS AT 6 MONTHS WITH INCREASED BMD AT 2.5 YEARS IN WOMEN ON ALENDRONATE
original image

When patients treated with alendronate were separated into three groups, based on decreases in urinary NTx at 6 months, those with the largest decrease in NTx (65% or greater) had the greatest gain in BMD at the total hip, trochanter, and PA spine compared with patients who had smaller decreases in NTx (all p < 0.05; Fig. 4). There was a similar (but not statistically significant) trend at the lateral spine (Fig. 4). Linear regression for NTx decreases at 6 months versus BMD increases at 2.5 years in the treatment group predicted increases of 2.8% at the total hip, 4.1% at the trochanter, 5.8% at the PA spine, and 6.9% at the lateral spine (all p < 0.05) for a 30% decrease in urinary NTx (Table 3).

Table TABLE 3. PREDICTED BMD CHANGES AT 2.5 YEARS FOR A 30% DECREASE IN NTX AT 6 MONTHS
original image
Figure FIG. 4.

Change in BMD at the (A) total hip and trochanter and (B) PA and lateral spine after 2.5 years of alendronate therapy grouped by percentage decreases in NTx at 6 months. Results as mean ± SEM. Group 1: ≤45% decrease NTx, mean = –27 ± 3%, n = 12; Group 2: 45–65% (exclusive) decrease NTx, mean = –57 ± 1%, n = 19; Group 3: ≥65% decrease NTx, mean = –76 ± 1%, n = 24; comparison of Groups 1–3: total hip, trochanter, PA spine all p < 0.05; lateral spine p = 0.13.

By 2.5 years, the average height of subjects in the alendronate group decreased by 0.3 ± 0.6 cm (mean ± SD, p < 0.01 from baseline); the average height of subjects in the placebo group decreased by 0.5 ± 0.8 cm (p < 0.01 from baseline), but there were no differences between the groups. Weight remained stable in both groups. In general, alendronate was well tolerated with a 47% incidence of gastrointestinal complaints in subjects on alendronate versus 43% on placebo.

DISCUSSION

Our goal was to test the hypothesis that early changes in biochemical markers of bone turnover could predict the long-term improvements in hip BMD usually seen in elderly women on alendronate therapy. The data are consistent with this hypothesis, demonstrating that a minimum decrease of 30% in urinary NTx at 6 months predicts increases of between 2.8% and 4.1% at the total hip and trochanter and increases of 5.8% and 6.9% at the PA and lateral spine after 2.5 years of therapy. In addition, greater decreases in urinary NTx at 6 months were associated with greater long-term increases in BMD at the hip and spine. We also observed significant associations between early decreases in serum OC and long-term increases in hip BMD. In subjects on placebo, there were no significant trends between early changes in biochemical markers and long-term BMD. Furthermore, we observed significant improvements in BMD at all skeletal sites for women on therapy; this included regions rich in both trabecular bone (PA/lateral spine) and cortical bone (total body, one-third distal radius). Finally, our results are relevant to a broad spectrum of elderly women because subjects were enrolled without strict health criteria or without bone density criteria for osteoporosis.

Previous placebo-controlled, double-blind, randomized, multicenter trials with strict BMD and health criteria have demonstrated similar increases in hip and spine BMD in older women on alendronate therapy.8,9,21 Studies have also reported early decreases in biochemical markers of bone turnover shortly after the initiation of therapy.22,23 Bauer and colleagues24 reported associations between baseline urinary NTx or serum OC versus spinal BMD at 3 years in older women (mean age 71 years). In contrast to the present study, they did not find an association with hip BMD. The authors did not report whether changes in biochemical markers following treatment were associated with follow-up BMD measurements. In a multicenter study of 359 women aged 60–85 years, Bone and colleagues reported that the magnitude of effect on biochemical markers was not predictive of BMD gains after 2 years on alendronate therapy.21 They also found that baseline NTx was correlated with spine BMD at 2 years. However, an earlier study by Garnero and colleagues, which included younger postmenopausal women on placebo or alendronate therapy, did find an association between changes in markers at 3 months and spine BMD at 2 years12 when all women were included in the analysis. Information on the ability of markers to predict changes in hip BMD only in women on therapy was not provided. By comparison, the present study found that changes in urinary NTx following therapy were predictive of improvements at the hip, spine, and total body in elderly women.

This study has several strengths. All participants were enrolled at a single clinical site using the same bone densitometer, decreasing the intersite and interinstrument variability in bone mass measurements that frequently occurs and requires statistical adjustments.6,8 Second, in contrast to the early alendronate trials which required use of relatively healthy subjects,8,21,23 the inclusion criteria for the present study were more relaxed, allowing for characteristic medical problems closer to those that exist in the general population of community-dwelling elderly women. Third, in addition to assessing the relevant sites of the hip (femoral neck, trochanter, intertrochanter, and total hip), BMD of the spine (PA/lateral), total body, and radius were measured to allow examination of sites rich in both trabecular and cortical bone. While the sites with the largest increase in BMD were those with the greatest component of trabecular bone (PA and lateral spine, trochanteric region of hip), even peripheral sites with a large composition of cortical bone, such as the radius, responded to therapy.

This study also has several limitations. The investigation was not large enough to examine fracture outcomes, but rather focused on hip BMD. Several studies that have demonstrated the association between hip BMD and fractures required large sample sizes or a cohort of fallers.13,25,26 Large multicenter trials involving alendronate have clearly demonstrated reductions in hip, spine, and wrist fractures in treated patients.6,8,9 Even with the much smaller sample size of the present study, we saw improvements in BMD similar to some larger trials.21 Second, our results apply to elderly women in general and are not specific to early postmenopausal or elderly osteoporotic women. However, based on WHO BMD criteria for the diagnosis of osteoporosis, we have previously reported that over half of these older women are in the osteoporotic range.17 Third, the results may have been confounded when the dose of alendronate was increased from 5 to 10 mg after 1.5 years; the results may be different for women on 5 mg for 2.5 years or those on 10 mg for 2.5 years. Thus, the results of this study likely underestimate the BMD efficacy of 10 mg of alendronate.

Of the resorption markers examined, the percentage decrease in urinary NTx had the greatest association with long-term increases in BMD. We and others have previously reported that NTx is a more specific marker of bone resorption in patients treated with bisphosphonates compared with other markers.27,28 While Dpd decreased significantly after 6 months of alendronate therapy, it was not associated with long-term changes in BMD. This may be because of the more modest decrease in free Dpd's (10%) versus urinary NTx (53%). Furthermore, OC and BAP (markers of bone formation) significantly decreased at 6 months, but only the changes in OC were associated with long-term changes in BMD. OC, a noncollagenous protein synthesized by the osteoblasts, may have a different physiologic response to bisphosphonates than BAP, an isoenzyme localized in the plasma membrane of the osteoblast.29

In summary, elderly women who received calcium and vitamin D supplementation maintained bone mass at the hip, forearm, and total body over 30 months, and showed small increases in spine BMD. Within this group, there were no consistent correlations between bone mass after 2.5 years and early (6-month) changes in biochemical markers of bone turnover. In contrast, 2.5 years of therapy with alendronate in unselected and representative elderly women produced significant and substantial increases in BMD of the hip, spine, and total body that can be predicted by early decreases in urinary NTx, a specific marker of bone resorption.

By this and other reports,6,8,9 the majority of women respond to treatment with alendronate. However, this study suggests that specific markers of bone turnover may be clinically useful in at least two ways. Patients with osteoporosis who have not fractured are usually asymptomatic and will not feel better after treatment; thus, long-term compliance with therapy may be compromised. The ability to assess an early decrease in bone turnover that predicts increases in spine and hip BMD might provide the motivation some patients need to remain compliant. Second, a minimal or absent decrease in urinary NTx may indicate poor compliance with the dosing regimen (e.g., taking a bisphosphonate with food) or decreased absorption. Both of these uses could aid the physician in maximizing the benefits of antiresorptive therapy and deserve further study.

Acknowledgements

We are indebted to the nursing staff of the General Clinical Research Center at the Beth Israel Deaconess Medical Center-East Campus; Meryl Poku and Michelle MacCallum for their assistance with recruitment and bone densitometry imaging; and Dawn Griffiths for preparation of the manuscript. We also acknowledge Ostex International and Metra Biosystems for their support of this study. Support was also provided by CDC Grant No. CC102550, National Institutes of Health Grant No. RR01032, Harvard-Thorndike General Clinical Research Center, Beth Israel Deaconess Medical Center, and Merck Research Laboratories, Rahway, NJ, U.S.A.

Ancillary