Article Text

Download PDFPDF

Extended report
Raloxifene for prevention of glucocorticoid-induced bone loss: a 12-month randomised double-blinded placebo-controlled trial
  1. Chi Chiu Mok1,
  2. King Yee Ying2,
  3. Chi Hung To1,
  4. Ling Yin Ho1,
  5. Ka Lung Yu1,
  6. Hon Kit Lee3,
  7. Kwok Man Ma4
  1. 1Department of Medicine, Tuen Mun Hospital, Hong Kong, SAR China
  2. 2Department of Medicine, Princess Margaret Hospital, Hong Kong, SAR China
  3. 3Department of Clinical Pathology, Tuen Mun Hospital, Hong Kong, SAR China
  4. 4Department of Nuclear Medicine, Tuen Mun Hospital, Hong Kong, SAR China
  1. Correspondence to Dr Chi Chiu Mok, Department of Medicine, Tuen Mun Hospital, Tsing Chung Koon Road, New Territories, Hong Kong, SAR China; ccmok2005{at}


Objectives To study the efficacy of raloxifene in preventing bone mineral density (BMD) loss in women receiving long-term glucocorticoids (GC). The study took the form of a parallel-group randomised double-blinded placebo-controlled trial.

Methods Postmenopausal women without hypercoagulability risk factors who were prevalent GC users were randomised to receive either raloxifene (60 mg/day) or placebo (1 tablet/day) on top of calcium (1000 mg/day) and calcitriol (0.25 μg/day). BMD of the hip and spine (primary outcome), bone turnover markers and new vertebral fractures (secondary outcomes) at month 12 were assessed.

Results Between December 2006 and December 2008, 114 patients were recruited (age 55.3±7.7 years). The duration and dose of prednisolone received was 62.2±64 months and 6.7±5.9 mg/day, respectively. Baseline vertebral fracture was present in six (5%) patients. In all, 57 patients were allocated to each of the treatment arms. Demographic data, osteoporotic risk factors and BMD at various sites were similar between the two groups of patients. At month 12, a significant gain in the lumbar spine (+1.3±0.4%; p=0.004) and total hip BMD (+1.0±0.4%; p=0.01) was observed in patients treated with raloxifene but a significant decrease in BMD of the lumbar spine (−0.9±0.4%; p=0.045) and hip (−0.8±0.3%; p=0.01) occurred in the placebo group. The femoral neck BMD did not change significantly in favour of raloxifene. Three new fractures developed exclusively in the patients treated with placebo. Bone formation (serum osteocalcin and procollagen type I N-terminal) and resorption (urine deoxypyridinoline and type I collagen) markers decreased significantly in the raloxifene group but not in patients treated with placebo. Leg cramps were numerically more frequent in the raloxifene group (7% vs 0%) but thromboembolism was not reported in any patients.

Conclusions In postmenopausal women receiving long-term GCs, raloxifene is well tolerated and significantly increases spinal and hip BMD after 12 months of treatment.

Statistics from


Glucocorticoid (GC) is an important cause of secondary osteoporosis. The chronic use of GCs increases the risk of fragility fracture at a much higher threshold of bone mineral density (BMD) than postmenopausal osteoporosis, indicating an additional deleterious effect of GC on bone quality.1 More than one-third of patients who are postmenopausal receiving long-term GC treatment developed asymptomatic vertebral fractures.2 3 A general practice research database confirmed an increased relative risk of vertebral and hip fractures in chronic GC users,4 with fracture risk proportional to the daily dose of GC. In another database study, intermittent use of high-dose GC and the cumulative GC dose was associated with an augmented risk of osteoporotic fracture.5

Patients with chronic rheumatic disorders are prone to osteoporosis and its complications. This is the result of multiple factors such as immobility, avoidance of sunshine, premature menopause, renal insufficiency and failure to achieve a peak bone mass, in addition to the chronic use of GCs. Persistent elevation of cytokines such as tumour necrosis factor α (TNFα) and interleukin 6 (IL-6) as a result of activity of the underlying diseases also aggravates BMD loss.6 7 This is illustrated by registry studies of rheumatoid arthritis that the prevalence of osteoporosis at the spine and the hip was increased by twofold when compared to age and gender matched population controls.8 In addition to non-pharmacological measures, calcium, vitamin D and the bisphosphonates are recommended as the preferred agents for GC-induced osteoporosis.9 Randomised placebo-controlled trials have shown efficacy of alendronate and risedronate in reducing vertebral fractures in GC users.10 11 A recent 12-month comparative study demonstrated superiority of intravenous zoledronic acid to risedronate in increasing spinal and hip BMD in GC-induced osteoporosis.12 However, teriparatide has been shown to be more effective than alendronate in reducing new vertebral fractures in chronic GC users.13

Raloxifene is a selective oestrogen receptor modulator (SERM) that binds to the oestrogen receptors and exhibits different selectivity in their oestrogenic actions in different tissues. A large randomised controlled trial (Multiple Outcomes of Raloxifene Evaluation (MORE)) in 7705 postmenopausal women demonstrated a significant increase in BMD of the femoral neck and lumbar spine by 2.1% and 2.6%, respectively, with raloxifene treatment for 3 years.14 Compared to placebo, raloxifene significantly reduced the risk of incident vertebral fractures and its antifracture efficacy was sustained up to 4 years.15 There is little information on the efficacy of raloxifene in GC-induced osteoporosis. In a controlled trial, we demonstrated that calcium combined with raloxifene (60 mg/day) maintained femoral and lumbar spinal BMD at 12 months compared to calcium alone in patients with systemic lupus erythematosus (SLE) who were postmenopausal without thrombophilia risk factors receiving long-term prednisolone.16 In view of the favourable preliminary result, we conducted a larger 12-month randomised double-blinded placebo-controlled study of raloxifene in patients who are postmenopausal receiving long-term GCs.

Patients and methods

Study population

Adult woman patients who were receiving long-term GC treatment for various rheumatic diseases in the outpatient rheumatology clinics of Tuen Mun and Princess Margaret Hospitals, Hong Kong, were invited to participate in this study. Inclusion criteria were: (1) postmenopausal for ≥12 months, (2) having been receiving a stable dose of corticosteroids (prednisone ≤10 mg/day or equivalent) for ≥6 months prior to entry, (3) patients expected to be on corticosteroid treatment throughout the study period, (4) written consent could be obtained. Exclusion criteria were: (1) patients with hypercoagulability risk factors (eg, positive anti-phospholipid antibodies) or a history of thromboembolism; (2) history of allergic reactions or intolerance to raloxifene or other SERMs; (3) patients receiving bisphosphonates, parathyroid hormone, SERMs, anticonvulsants or anticytokine therapies within 6 months prior to entry; (4) patients with known bone disorders such as osteomalacia, renal osteodystrophy and hyperparathyroidism; (5) patients with undiagnosed uterine bleeding; (6) patients with serum creatinine level of ≥200 μmol/litre.

The protocol was approved by the Research and Ethics Committee of our hospital and registered in the US database (no. NCT00371956). Reporting of severe adverse events and interim analyses on safety to our Research and Ethics Committee was mandatory.

Randomisation and treatment protocol

This was a 12-month parallel-group, randomised, double-blind, placebo-controlled study. At study entry, all patients received elemental calcium (1000 mg/day) and calcitriol (0.25 μg/day) and were randomised by blocks of four to receive either raloxifene (60 mg/day) or placebo (1 tablet/day). Patients, investigators and outcome assessors were blinded for the treatment assignment. The raloxifene and placebo (with identical appearance to raloxifene) tablets were supplied by Eli Lily Asia and managed by a designated pharmacist in our pharmacy department. The randomisation code was kept sealed until the completion of the study. Demographic data and risk factors for osteoporosis of the participants, and the cumulative doses of prednisolone received at each study visit during the study period were collected.

Clinical assessments

BMD (lumbar spine, hip and whole body) and markers of bone turnover (urine deoxypyridinoline (DPD), serum osteocalcin, total serum procollagen type I N-terminal propeptide (P1NP) and C-terminal crosslinked telopeptide collagen degradation product of type I collagen (β-CTX)) were assessed at baseline, months 6 and 12. Plain radiographs of the thoracic and lumbar spine for fractures were taken at baseline and month 12. Fasting serum lipid level (total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol and triglyceride) was also measured at baseline, months 6 and 12.

Outcomes of interest and sample size calculation

The primary outcomes of the study were changes in BMD at the lumbar spine and hip at month 12 compared to baseline. Secondary outcomes included changes in levels of bone turnover markers over time, lipid profile, occurrence of new vertebral fractures and adverse events.

Assuming that the mean baseline BMD at the spine of the participants was 0.80 g/cm2 (estimated from our preliminary study of raloxifene in patients with SLE),16 with a SD of 0.08 g/cm2, and there was an expected 2% gain or loss in BMD, respectively, in the raloxifene and placebo groups at month 12, a sample size of 77 in each arm was required to detect the difference between the two groups with an α error of 5% and a power of 80%.

BMD and vertebral fracture assessment

BMD at various body sites (lumbar spine (L2–4), non-dominant hip, femoral neck and trochanter and whole body) were measured by the dual-energy x-ray absorptiometry technique using Delphi densitometer (Hologic, Bedford, Massachusetts, USA). For patients with avascular bone necrosis of the hip or joint replacement, the BMD of the other hip was used. The reference ranges for the T scores were derived from the third National Health and Nutrition Examination Survey (NHANES III) database (hip) and the device manufacturer's dataset (lumbar spine).17 The technician who was responsible for measuring BMD was blinded for the details of the study.

The radiographs of the thoracic and lumbar vertebrae were examined for deformities by visual inspection. Baseline vertebral fracture was defined when there was a loss of at least 25% of the vertebral height. Incident / new vertebral fractures at month 12 were diagnosed when there were distinct alterations in the morphology of the vertebral bodies that resulted in a loss of at least 25% of vertebral height in previously normal vertebrae or worsening of previously deformed vertebrae.

Assay of bone turnover markers

Markers of bone resorption (serum β-CTX and urine DPD/creatinine) and bone formation (serum osteocalcin and P1NP) were studied. Serum β-CTX was assayed by electrochemiluminescence (Roche Diagnostics, GmbH, Mannheim, Germany) and urine DPD was measured by chemiluminescence (Siemens Medical Solutions Diagnostics, Deerfield, Illinois, USA) using commercially available kits. Serum osteocalcin and P1NP was also studied by electrochemiluminescence using commercial kits (Roche Diagnostics, GmbH, Mannheim, Germany). All blood and urine samples were collected at 9:00 h in the morning, with fasting for at least 8 h.

Monitoring and follow-up

Participants were followed up at 3-month intervals with special attention to disease flares, thromboembolic events and intolerance to raloxifene. In case of severe flares of the underlying diseases thought to be temporally related to the study drug, the medication would be discontinued and usual treatment would be given. The study drug would also be discontinued in case of thromboembolic events.

Data analyses and statistical methods

Unless otherwise stated, values in this study were expressed as mean±SD (SD). Between-group (raloxifene and placebo) comparison of BMD at baseline was performed by the independent Student t test. BMD at month 6 and month 12 between the raloxifene and placebo groups were compared with adjustment of baseline BMD values, age, body mass index (BMI) and the cumulative doses of prednisolone received by one-way analysis of covariance. Within-group (between baseline and months 6 or 12) comparison was made by the paired Student t test. Statistical significance was defined as a p value <0.05, two tailed. All statistical analyses were performed using the SPSS program, V.11.5 (SPSS, Chicago, Illinois, USA) for Windows.


Patient enrolment and disposition

Between December 2006 and December 2008, 127 postmenopausal women who were receiving low-dose GCs were approached but 11 patients were reluctant to participate. Two patients were excluded because of a history of minor stroke. Finally, 114 patients consented for the study. Although this number of patients did not meet our target sample size, it was the best we could achieve because of the difficulty in recruitment and time constraints. The mean age of the participants was 55.3±7.7 years. A total of 57 patients were randomised to receive raloxifene while the other 57 patients were randomised to placebo. At month 12, 50 (88%) patients in the raloxifene group and 55 (96%) patients in the placebo group completed the study. Seven (12%) patients in the raloxifene group withdrew from the study because of non-compliance (N=4), skin rash (N=1), leg cramps (N=1) and generalised aching (N=1). Six of these patients did not complete the clinical assessments at month 12. In the placebo arm, two (4%) patients withdrew from the study because of non-compliance (N=1) and skin rash (N=1). One of these patients did not have clinical assessments at month 12. Figure 1 summarises the patient enrolment and disposition of the current study.

Figure 1

Patient enrolment and disposition.

Underlying rheumatic diseases and baseline characteristics

The underlying rheumatic diseases of the participants were as follows: SLE (54%), rheumatoid arthritis (27%), inflammatory myopathies (4%), systemic sclerosis (4%) and systemic vasculitides (3%). There was no significant difference in the frequency of these diseases between the raloxifene and placebo groups of patients. At study entry, 29% patients were naive to calcium and 75% patients were naive to calcitriol. In all, 92% of the participants were naive to any antiosteoporotic medications including bisphosphonates. The duration and dose of prednisolone received by the participants was 62.9±61 months and 6.9±5.8 mg/day, respectively. The proportion of participants receiving different daily doses of prednisolone was as follows: ≤2.5 mg (13%), >2.5–5.0 mg (54%), >5.0–7.5 mg (18%) and >7.5 mg (16%). The mean age at menopause was 47.1±5.2 years and the mean duration of menopause was 8.5±7.7 years. The mean parity was 2.1±1.3. The BMI of the participants was 23.7±3.5 kg/m2 (4% of patients had BMI <18 kg/m2). Seven patients (6%) were chronic smokers, but no patients were habitual drinkers. Osteopenia or osteoporosis (T scores <−1.0) of the lumbar spine and the hip occurred in 76% and 70% of the patients, respectively, at baseline. Pre-existing vertebral fracture was present in six (5%) patients. The baseline demographic data, osteoporotic risk factors, lipid levels, BMD at various sites and the cumulative dose of prednisolone received during the 12-month study period were not significantly different between the two groups of patients (table 1).

Table 1

Baseline clinical characteristics of the recruited patients

Changes in BMD over time

Data from 51 patients in the raloxifene group and 56 patients from the placebo group were available for efficacy analyses. Figure 2A–D shows the changes in BMD of the lumbar spine, total hip, femoral neck and whole body in the two groups of patients. At month 12, a significant gain in BMD at the lumbar spine (from 0.868 to 0.879 g/cm2; +1.3±0.4%; p=0.005) and the total hip (from 0.757 to 0.765 g/cm2; +1.0±0.4%; p=0.01) compared to baseline was observed in the patients treated with raloxifene. Conversely, BMD of the lumbar spine (from 0.849 to 0.842 g/cm2; −0.9±0.4%; p=0.045) and hip (from 0.779 to 0.773 g/cm2; −0.8±0.3%; p=0.01) decreased significantly in the placebo group of patients. The differences in spinal and hip BMD between the two groups of patients at month 12 were statistically significant with adjustment for baseline BMD values, age, BMI and cumulative prednisolone dose received during the study period (p<0.001 in both). The changes in femoral neck BMD, however, were not significant in both groups of patients.

Figure 2

Changes in bone mineral density (BMD) from baseline to 12 months (N=107). (A) Lumbar spine, (B) hip, (C) femoral neck, (D) whole body. Error bars represent SEM. p Values refer to within-group comparison.

At month 12, three new vertebral fractures developed in the placebo group but no new fractures were observed in the patients treated with raloxifene. The difference in the incidence of new vertebral fractures between the two groups of patients was not statistically significant (p=0.24).

Changes in markers of bone turnover

Figure 3A–D shows the changes in bone turnover markers in the two groups of patients. A significant drop in the levels of urine DPD/creatinine, serum β-CTX, serum osteocalcin and P1NP was observed in the raloxifene-treated group but not in the placebo group of patients. In patients treated with raloxifene, the levels of β-CTX, osteocalcin and P1NP were all within the premenopausal ranges at month 12 (73% and 98% of patients, respectively, had P1NP and osteocalcin levels reduced to the lower half of the premenopausal ranges). Conversely, in the placebo group, reduction of bone markers to premenopausal ranges occurred in 77% (β-CTX), 96% (osteocalcin) and 72% (P1NP) of patients, respectively (41% and 89% of patients, respectively, had P1NP and osteocalcin levels reduced to the lower half of the premenopausal ranges).

Figure 3

Changes in bone markers from baseline to 12 months (N=107). (A) Urine deoxypyridinoline (DPD)/creatinine (Cr) (nmol/mmol), (B) serum C-terminal telopeptide collagen degradation product of type 1 collagen (CTX) (ng/ml), (C) serum procollagen type 1 N-terminal propeptide (P1NP) (ng/ml), (D) serum osteocalcin (ng/ml). Error bars represent SEM. p Values refer to within-group comparison.

Changes in lipid profile

The baseline levels of total cholesterol, LDL cholesterol, HDL cholesterol and triglyceride did not differ significantly between the raloxifene and placebo groups of patients (table 1). There were no significant changes in each of above lipid parameters from baseline to month 12 in both groups of patients. In patients with baseline LDL cholesterol levels of ≥2.6 mmol/litre (N=35 for raloxifene and N=29 for placebo), a significant drop in total cholesterol level (p=0.04) and a non-significant drop in LDL cholesterol (p=0.08) at month 12 was observed in the patients treated with raloxifene whereas no significant changes in any of the lipid parameters were observed in the patients treated with placebo (table 2).

Table 2

Changes in lipid level in patients with baseline LDL cholesterol ≥2.6 mmol/litre

Adverse events

Table 3 shows the frequency of adverse events. Leg cramps were numerically more common in patients treated with raloxifene (7% vs 0%; p=0.13) while other adverse events occurred at similar frequency between the two groups. Nine patients withdrew from the study because of non-compliance to treatment (N=5) and adverse events (skin rash N=2, leg cramp N=1 and generalised aching N=1). Study medications were stopped in these patients. None of the participants developed arterial or venous thromboembolism and no patients had severe flares of their underlying rheumatic diseases that led to study withdrawal.

Table 3

Adverse events experienced by participants


This was a pilot randomised placebo-controlled study on the efficacy of raloxifene in preventing BMD loss in postmenopausal women receiving long-term GCs. We demonstrated that raloxifene (60 mg/day) treatment for 12 months was effective in increasing the lumbar spine and total hip BMD by 1.3% and 1.0%, respectively. The effect of the drug on femoral neck BMD, however, was not statistically significant. Raloxifene significantly suppressed markers of bone turnover as compared to placebo. New vertebral fractures developed exclusively in patients treated with placebo. Raloxifene was well tolerated, with only a non-significant increase in the frequency of leg cramps. Arterial or venous thromboembolism was not reported.

The mechanisms of GC-induced osteoporosis are complex. GCs reduce intestinal absorption of calcium, increase urinary excretion of phosphate and calcium and suppress the production of sex steroids and insulin-like growth factor 1.18 GCs downregulate the expression of osteoprotegerin in osteoblasts, which leads to increased receptor activator for nuclear factor κ B ligand (RANKL) expression and hence enhanced activity and survival of the osteoclasts.19 GCs also exhibit direct inhibitory effects on osteoblast activity and promote apoptosis of osteoblasts and osteocytes through a reduction of the Wnt signalling, increase in peroxisome proliferator-activated receptor γ2 expression and activation of caspase 3.20 21 High-dose GC treatment leads to an early and transient phase of increased bone resorption, followed by decreased bone formation on long-term administration.

Raloxifene is a SERM that binds to the oestrogen receptors of bone tissues. In vitro studies show that raloxifene inhibits osteoclastic resorptive activity and reduces production of TNFα and IL-6.22 In randomised controlled trials,23 raloxifene was shown to significantly reduce bone turnover markers after 24 months of treatment. These markers included bone-specific alkaline phosphatase (15% decrease), osteocalcin (30% decrease), CTX (40% decrease). In the MORE study, raloxifene was demonstrated to lower the levels of all bone turnover markers by 30% to 40% to the premenopausal ranges during the first year of treatment, as compared to placebo.14 The results from the current study that bone formation and resorption markers were reduced by 19% to 44% with raloxifene treatment for 12 months are consistent with previous findings. Moreover, in the MORE study, the vertebral fracture rate reduction with raloxifene treatment was much higher than that expected from the gain in the lumbar spine BMD,24 suggesting that raloxifene might also improve bone quality.

Raloxifene has been shown to reduce total and LDL cholesterol level without significant change in HDL, triglycerides or C reactive protein in controlled trials.14 25 26 The Raloxifene Use for the Heart trial, which was a randomised placebo-controlled trial for the cardiovascular benefits of raloxifene, did not demonstrate an increase in the risk of coronary events and stroke in raloxifene users compared to placebo after a median observation of 5.6 years, although the absolute incidence of fatal stroke was increased.27 28 In our previous study of SLE,16 we demonstrated that raloxifene reduced LDL cholesterol level but in the current study which involved a larger number of patients, we were unable to demonstrate this effect. One possible explanation is that a significant proportion of our recruited patients did not have elevated cholesterol levels at baseline. In a post hoc subgroup analysis of patients with baseline LDL cholesterol levels of ≥2.6 mmol/litre, we were able to demonstrate efficacy of raloxifene in lowering the total cholesterol level.

In our previous study of SLE,16 raloxifene was well tolerated with no serious events such as thromboembolism and major disease flares reported. The low incidence of venous thromboembolism was consistent with a randomised controlled study of 483 Asian women in which no venous thrombosis was reported with the use of raloxifene.29 Once again, the current study confirmed that no participants developed venous thromboembolism during the 12-month treatment period. However, it should be noted that in our previous study16 and the present one, subjects with hypercoagulability risk factors and history of thromboembolism were excluded. Moreover, a longer period of observation is necessary to confirm the safety of raloxifene in terms of thromboembolic risk. Nevertheless, leg cramp was the only adverse effect that was numerically more frequent in patients treated with raloxifene in this study, but only one patient withdrew from the study for this reason.

Finally, the baseline vertebral fracture rate was low (5%) in our study compared with others involving postmenopausal women receiving long-term GCs.2 3 In our locality, only those patients with fragility fractures were financially subsidised by public hospitals to receive antiosteoporotic medications. As a result, there was difficulty in recruiting patients with pre-existing fractures not on bisphosphonates. Many patients who agreed to participate in this study were treatment naïve because they could not afford antiosteoporotic medications.

In summary, this is the first randomised controlled trial of raloxifene in chronic GC users. Our results demonstrate efficacy of raloxifene in raising the spinal and hip BMD after 1 year of treatment. Raloxifene is well tolerated and is an option for the prevention of BMD loss in postmenopausal women receiving long-term GC treatment, especially when bisphosphonates are contraindicated or cannot be tolerated, or during bisphosphonate drug holiday. However, in osteoporosis interventional trials, the change in BMD is only a surrogate endpoint of antifracture efficacy. Randomised controlled trials of a larger sample size addressing vertebral fracture as the primary outcome are necessary to confirm the efficacy of raloxifene. The long-term safety of raloxifene such as the risk of thromboembolism has also to be assessed.


This was an investigator-initiated study supported by Eli Lily Asia, which provided the investigators raloxifene and placebo tablets and a lump sum sponsorship of HK$700 00 for the assay of the bone turnover markers. Eli Lily Asia was not involved in the design and conduct of this study. The investigators would also like to thank Miss Susanna So Mui Chiu and Fondy Shun Chi Ng of the Department of Clinical Pathology, Tuen Mun Hospital, Hong Kong for their effort in measuring the bone turnover markers.



  • Competing interests The principal investigator is independent of the sponsoring drug company with regard to the data analysis. He had full access to all of the data and takes responsibility for the integrity of the data and the accuracy of the data analysis.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Research and Ethics Committee of Tuen Mun Hospital.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.