Raloxifene for postmenopausal women with systemic lupus erythematosus: A pilot randomized controlled study




To study the effects of raloxifene on disease activity and bone mineral density (BMD) in postmenopausal women with systemic lupus erythematosus (SLE).


Postmenopausal women with osteopenia and inactive SLE were randomly assigned to receive either raloxifene (60 mg/day) plus elemental calcium (1,200 mg/day) or elemental calcium alone (control). Patients with a history of thromboembolism or antiphospholipid antibody positivity were excluded. BMD at various sites was serially measured, and lupus activity was serially assessed using the Safety of Estrogens in Lupus Erythematosus: National Assessment–Systemic Lupus Erythematosus Disease Activity Index (SELENA–SLEDAI).


The study group comprised 33 patients (16 assigned to receive raloxifene and 17 controls, mean ± SD age 53.8 ± 5.3 years). Age, body mass index, and baseline BMD values did not differ significantly between the 2 groups of patients. All patients were receiving low-dose prednisolone. After 12 months, femoral neck BMD (mean ± SD −2.6 ± 1.0%; P = 0.02) and lumbar spine BMD (−3.3 ± 0.8%; P = 0.001) decreased significantly in the controls but not in the raloxifene group. No patient had a major flare of lupus, but mild/moderate flares occurred in 4 raloxifene-treated and 6 control patients (P = 0.79). The total area under the curve of SELENA–SLEDAI scores was not significantly different between the 2 groups. A significant increase in the high-density lipoprotein cholesterol level and a reduction in the low-density lipoprotein cholesterol level were observed in the raloxifene group but not in controls. One patient in the raloxifene group (6%) withdrew from the study because of hot flushes. No thromboembolic events were reported.


Raloxifene was well tolerated in Chinese patients with SLE who had inactive disease and in whom hypercoagulability was not identified. Raloxifene maintained femoral neck and spinal BMD in patients receiving corticosteroids.

Systemic lupus erythematosus (SLE) is an autoimmune disease that predominantly affects women of childbearing age. Aberrations in estrogen metabolism, the binding capacity of the estrogen receptors, and expression of estrogen receptor transcripts have been hypothesized to contribute to the pathogenesis of SLE (1). In murine models of lupus, removal of the endogenous source of estrogens retards progression of autoimmunity, whereas administration of exogenous estrogens has the opposite effect (1). Treatment of female (NZB × NZW)F1 mice with the estrogen antagonist tamoxifen significantly reduces anti-DNA production, ameliorates immune complex–mediated glomerulonephritis, and prolongs survival (2, 3). In female MRL-lpr/lpr mice, tamoxifen alleviates disease activity, and treatment with the selective estrogen receptor modulator LY139478 improves survival and retards histologic progression of glomerulonephritis (4, 5). Administration of tamoxifen or an antiestradiol antibody improved hematologic and renal manifestations in an experimental murine lupus model (6).

The selective estrogen receptor modulators belong to a class of chemically diverse compounds that bind to estrogen receptors and exhibit different selectivity in their estrogenic actions in different tissues. Little information has been available regarding the effect of selective estrogen receptor modulators on disease activity in human SLE. An open-label study of 11 patients did not demonstrate any benefits of tamoxifen in ameliorating the clinical and serologic activity of SLE (7).

Raloxifene is a selective estrogen receptor modulator that has been extensively studied. A large randomized, controlled trial (the Multiple Outcomes of Raloxifene Evaluation [MORE] study) demonstrated that raloxifene significantly reduced the incidence of vertebral fractures in postmenopausal women with osteoporosis (8). Patients with SLE are prone to osteoporotic fractures because of multiple factors, including corticosteroid therapy, premature menopause, avoidance of sunshine, renal insufficiency, disabling arthritis, and failure to achieve peak bone mass. No studies of raloxifene in corticosteroid-treated patients have been performed. This prompted the current pilot study, in which we evaluated the effects of raloxifene on disease activity and bone mineral density (BMD) in postmenopausal patients with SLE.


The study group comprised postmenopausal patients with SLE who attended the medical clinics of Tuen Mun Hospital, Hong Kong. Inclusion criteria were as follows: fulfillment of at least 4 of the American College of Rheumatology criteria for a diagnosis of SLE (9); sustained amenorrhea for ≥12 months with documented postmenopausal levels of estradiol and follicle-stimulating hormone (FSH); clinically inactive SLE for ≥6 months prior to the study, with a stable dosage of prednisone (≤10 mg/day); baseline T score for BMD of < −1.0 at either the hip or the lumbar spine; and provision of informed consent. Exclusion criteria were as follows: personal/family history of thromboembolism; presence of antiphospholipid antibodies (either lupus anticoagulant or anticardiolipin antibodies); receipt of any form of antiresorptive treatment or vitamin D supplementation within 12 months prior to study entry; and current use of hormone replacement therapy (HRT). This study was approved by the Research and Ethics Committee of our hospital.

Study protocol.

Upon study entry, participants were assigned, by simple randomization, into 2 treatment arms. Those assigned to the raloxifene group received raloxifene (60 mg/day) together with elemental calcium (1,200 mg/day) in the form of calcium carbonate (caltrate). Patients assigned to the control group received elemental calcium (1,200 mg/day) only. Other medications were used as indicated, except for the statins, which were prohibited during the study period.

Baseline evaluation and serial monitoring.

Information regarding the patient's age, duration of SLE, corticosteroid use, duration of menopause, body mass index (BMI), and disease activity scores was obtained at the time of study entry. At baseline, all patients underwent dual x-ray absorptiometry (DXA) scanning to determine BMD and plain radiography (anteroposterior and lateral) of the thoracic and lumbar spine to identify preexisting fractures. Blood cell counts and renal and liver function tests were performed, plasma levels of estradiol and FSH were determined, a lipid profile was obtained, and anti–double-stranded DNA and complement levels were evaluated.

BMD and the lipid profile were monitored at 6 and 12 months. Disease activity scores were recorded at 2-month intervals. Plain radiographs of the spine were obtained again at 12 months to screen for the occurrence of new fractures. Disease flares, thromboembolic events, and adverse effects of raloxifene were noted. In patients in whom severe SLE flares were thought to be temporally related to administration of raloxifene, the drug was discontinued and usual treatment was instituted. Raloxifene was also discontinued if a thromboembolic event occurred at any time during the study period.

DXA procedure.

Lumbar spine (L2–L4), nondominant femoral (femoral neck, trochanter, intertrochanter, Ward's triangle), and whole body BMD were measured by the DXA technique, using a Delphi densitometer (Hologic, Bedford, MA). For patients who had avascular bone necrosis of the hip or had undergone hip replacement surgery, the BMD of the other hip was used. The precision of the bone densitometer in measuring BMD at the lumbar spine, nondominant femoral neck, total hip, and whole body was 1.14%, 1.91%, 1.03%, and 0.77%, respectively. The reference ranges for the T scores for the hip were derived from the Third National Health and Nutrition Examination Survey database, and those for the lumbar spine were derived from the data set provided by the manufacturer of the device (10). The technician who was responsible for measuring BMD in all participants was blinded to the treatment arm and the clinical outcome assessments.

Assessment of SLE activity and definitions of disease flares.

SLE disease activity was measured using the Safety of Estrogens in Lupus Erythematosus National Assessment–Systemic Lupus Erythematosus Disease Activity Index (SELENA–SLEDAI) instrument. The SELENA–SLEDAI is a reliable tool for assessing lupus activity and flares and was used in the multicenter randomized controlled SELENA trial for the safety of estrogen use in patients with SLE (11). SELENA–SLEDAI scores were obtained at study entry and at 2-month intervals. Flares of SLE were classified as severe or mild/moderate, according to the definitions of the SELENA–SLEDAI instrument as described elsewhere (11). Patients were followed up throughout the study by the same group of physicians, who were briefed about the SELENA–SLEDAI instrument before the study. Disease flares were verified by at least 2 physicians from the group of principal investigators. The SELENA–SLEDAI score over time (area under the curve [AUC-S]) was summated using the following formula:

equation image

where z = the number of clinical assessments (1, 2, … z), and SELENA–SLEDAI(n) = SELENA–SLEDAI score at time n.

Statistical analysis.

In this intent-to-treat analysis, between-group (raloxifene versus control) comparisons of BMD at baseline (month 0) were performed using the independent Student's t-test. BMD values at different time points (month 6 and month 12) were compared, after adjusting for baseline BMD, BMI, age, duration of menopause, cumulative doses of prednisolone, and AUC-S of the SELENA–SLEDAI scores, by one-way analysis of covariance (ANCOVA). Within-group comparisons (baseline versus month 6 or month 12) were performed by Student's paired t-test. Two-tailed P values less than 0.05 were considered significant. All statistical analyses were performed using SPSS version 11.5 software for Windows XP (SPSS, Chicago, IL).


Thirty-three patients were recruited; 16 were assigned to the raloxifene group, and 17 were assigned to the control group. The mean ± SD age of the patients was 53.8 ± 5.3 years, and the mean ± SD duration of SLE was 95.6 ± 86 months. Thirty patients had natural menopause, and 3 had cyclophosphamide-induced premature menopause. The mean ± SD plasma estradiol and FSH levels were 65.2 ± 31 pmoles/liter (postmenopausal range <201 pmoles/liter) and 78.7 ± 27 IU/liter (postmenopausal range 25.8–135 IU/liter), respectively. All patients had clinically inactive SLE (mean ± SD SLEDAI score 1.4 ± 1.6 [median 0]; interquartile range 2.5) and were receiving low-dose prednisolone (mean 4.2 ± 2.8 mg/day). Nineteen patients (58%) had osteoporosis of the lumbar spine, and 3 patients (9%) had preexisting lumbar fractures at baseline.

Table 1 shows the characteristics of the patients at the time of recruitment into the study. No significant differences between the 2 groups were observed. No patients were known to be habitual smokers or drinkers.

Table 1. Characteristics of patients at the time of recruitment*
CharacteristicRaloxifene (n = 16)Controls (n = 17)P
  • *

    Except where indicated otherwise, values are the mean ± SD. SLE = systemic lupus erythematosus; LDL = low-density lipoprotein; HDL = high-density lipoprotein; SELENA–SLEDAI = Safety of Estrogens in Lupus Erythematosus National Assessment–SLE Disease Activity Index.

Age, years52.2 ± 5.655.3 ± 4.70.10
Years since onset of menopause6.3 ± 4.35.5 ± 4.30.52
Duration of SLE, months100.8 ± 8590.8 ± 890.75
Parity2.1 ± 2.02.5 ± 1.50.52
Body mass index, kg/m223.3 ± 4.924.2 ± 4.60.57
No. (%) with preexisting vertebral fractures2 (13)1 (6)0.96
Bone mineral density, gm/cm2   
 Total hip0.760 ± 0.080.737 ± 0.070.46
 Femoral neck0.669 ± 0.070.660 ± 0.050.73
 Hip trochanter0.557 ± 0.060.518 ± 0.060.12
 Lumbar spine (L2–L4)0.786 ± 0.090.821 ± 0.120.28
 Whole body0.955 ± 0.061.017 ± 0.250.39
Serum lipid level, mmoles/liter   
 Total cholesterol4.93 ± 1.14.64 ± 0.90.42
 LDL cholesterol2.78 ± 0.92.49 ± 0.70.34
 HDL cholesterol1.53 ± 0.41.38 ± 0.40.34
 Triglycerides1.37 ± 0.81.81 ± 1.00.16
SELENA–SLEDAI score1.6 ± 1.71.1 ± 1.50.37
Daily prednisolone dose, mg4.8 ± 2.83.5 ± 2.80.19
Use of other medications, no. (%)   
 Hydroxychloroquine6 (38)6 (35)1.00
 Azathioprine6 (38)9 (53)0.37

Changes in BMD.

Table 2 shows the changes in BMD at various sites between baseline, month 6, and month 12. The changes in BMD of the femoral neck and lumbar spine are shown in Figure 1. A significant decrease in BMD of the femoral neck (mean ± SD −2.6 ± 1.0%; P = 0.02) and lumbar spine (−3.3 ± 0.8%; P = 0.001) was noted in the control patients at month 12. However, no significant decrease in BMD of the total hip (−1.3 ± 0.7%; P = 0.07), femoral neck (−0.4 ± 0.9%; P = 0.71), and lumbar spine (−0.5 ± 0.8%; P = 0.58) was observed in the raloxifene-treated patients. By ANCOVA (with adjustment for baseline BMD, BMI, age, duration of menopause, cumulative dose of prednisolone received, and the AUC-S of SELENA–SLEDAI scores), the difference in BMD of the lumbar spine between the 2 groups was significant at 12 months. No new fractures occurred in any patient. The cumulative dose of prednisolone received did not differ significantly between the 2 groups (mean ± SD 1.79 ± 1.0 gm in the raloxifene group versus 1.21 ± 1.1 gm in controls; P = 0.12). Changes in the BMI in either group of patients at month 12 compared with baseline were statistically insignificant (mean ± SD −1.6 ± 1.2% in the raloxifene group [P = 0.21] and −0.2 ± 1.5% in the control group [P = 0.92]).

Table 2. Changes in bone mineral density (BMD) in patients treated with raloxifene (n = 16) and controls (n = 17)*
BMDMonth 0Month 6Month 12P
  • *

    Values are the mean ± SD gm/cm2.

  • Baseline versus month 12 (within-group comparison).

  • Raloxifene versus control at month 6 and month 12, by analysis of covariance, adjusted for baseline BMD, body mass index, age, duration of menopause, cumulative dose of prednisolone received, and cumulative disease activity scores.

Total hip    
 Raloxifene0.760 ± 0.080.757 ± 0.080.750 ± 0.080.07
 Control0.737 ± 0.080.731 ± 0.070.720 ± 0.070.07
Femoral neck    
 Raloxifene0.669 ± 0.070.659 ± 0.070.666 ± 0.080.71
 Control0.660 ± 0.050.648 ± 0.050.643 ± 0.050.02
Hip trochanter    
 Raloxifene0.557 ± 0.060.558 ± 0.060.556 ± 0.060.80
 Control0.518 ± 0.060.517 ± 0.070.513 ± 0.070.49
Lumbar spine    
 Raloxifene0.786 ± 0.090.784 ± 0.100.782 ± 0.100.58
 Control0.821 ± 0.120.808 ± 0.130.794 ± 0.120.001
Whole body    
 Raloxifene0.955 ± 0.060.961 ± 0.060.961 ± 0.060.18
 Control1.017 ± 0.250.961 ± 0.080.956 ± 0.080.28
Figure 1.

A, Changes in bone mineral density (BMD) of the femoral neck. B, Changes in BMD of the lumbar spine. Values are the mean and SD.

Changes in disease activity.

None of the patients had severe disease flares, but mild/moderate flares were noted in 4 raloxifene-treated patients (arthritis in 2 patients, mucosal ulceration in 1, new lupus rash and mild cytopenia in 1) and 6 control patients (arthritis in 3 patients, new lupus rash in 1, arthritis and skin rash in 1, arthritis and mild cytopenia in 1) at 12 months (P = 0.79). The AUC-S of the SELENA–SLEDAI scores over the 12-month period was not significantly different between the 2 groups of patients (mean ± SD 15.6 ± 15 point-months in the raloxifene group versus 20.1 ± 16 point-months in controls; P = 0.41).

Changes in lipid profile.

No significant changes in serum total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, or triglyceride levels occurred in the control group (Table 3). However, a significant increase in the HDL cholesterol level and a reduction in the LDL cholesterol level were noted in the raloxifene-treated patients. However, the difference in lipid profiles between the 2 groups of patients at 12 months was not significant after adjustment for baseline lipid values.

Table 3. Changes in lipid levels in patients treated with raloxifene (n = 16) and controls (n = 17)*
LipidMonth 0Month 6Month 12P
  • *

    Values are the mean ± SD mmoles/liter. LDL = low-density lipoprotein; HDL = high-density lipoprotein.

  • Baseline versus month 12 (within-group comparison).

  • Raloxifene versus control at month 6 and month 12, by analysis of covariance, adjusted for baseline lipid levels.

Total cholesterol    
 Raloxifene4.93 ± 1.14.74 ± 0.84.81 ± 0.70.52
 Control4.64 ± 0.94.76 ± 1.14.78 ± 1.00.73
LDL cholesterol    
 Raloxifene2.78 ± 0.92.50 ± 0.82.43 ± 0.70.048
 Control2.49 ± 0.72.51 ± 0.82.54 ± 0.80.73
HDL cholesterol    
 Raloxifene1.53 ± 0.41.63 ± 0.51.71 ± 0.60.02
 Control1.38 ± 0.41.45 ± 0.51.52 ± 0.50.09
 Raloxifene1.37 ± 0.81.35 ± 0.71.47 ± 0.60.35
 Control1.81 ± 1.01.72 ± 0.91.50 ± 0.70.07

Adverse events.

One patient (6%) in the raloxifene group withdrew because of hot flushes and leg cramps. All other patients completed the 12-month study, with satisfactory drug compliance. None of the raloxifene-treated patients experienced influenza-like symptoms, peripheral edema, or skin rash. Serious adverse effects such as venous thromboembolism were not observed.


This was a pilot randomized controlled study of the effect of raloxifene on disease activity and BMD in a small cohort of postmenopausal Chinese patients with SLE who were receiving corticosteroids. Our results indicate that treatment with raloxifene maintained femoral neck and lumbar spine BMD in these patients. Major lupus flares and serious adverse events such as venous thromboembolism were not observed in the raloxifene group during the 12-month study period. The incidence of hot flushes (7%) and leg cramps (7%) was consistent with the incidence reported in previous trials of raloxifene in both Caucasian and Chinese women (8, 12). However, the current study has limitations. First, the trial was not performed in a double-blinded manner. Bias in terms of a patient's behavior in lifestyle modification and the physician's clinical assessment was possible. Second, because of the relatively low incidence of disease flares in menopausal lupus patients, a very large sample size is necessary to show a difference in lupus activity between patients receiving raloxifene and controls. Because of the limited number of patients who could be recruited from a single center, the current study was not powered to detect a difference in the frequency of disease flares between the 2 groups of patients. Therefore, the data presented should be viewed as being only preliminary and must be confirmed by larger-scale multicenter, double-blind, placebo-controlled trials.

Unlike the MORE study, which demonstrated an increase in lumbar spine and femoral neck BMD (2.4% and 2.1%, respectively) in postmenopausal women after 3 years of raloxifene treatment (8), we did not observe any significant increase in BMD in the corresponding sites in our patients. This may be attributable to the shorter duration of raloxifene therapy and the use of corticosteroids in our study. Because our patients were in early menopause and were receiving long-term corticosteroid therapy, a greater decrease in BMD was expected when compared with healthy postmenopausal women. This was illustrated by a significant drop in BMD (2.6% and 3.3% at the femoral neck and lumbar spine, respectively) in our control patients over 12 months. Because fracture reduction depends on improvement of both BMD and bone quality, the capacity of raloxifene to maintain BMD in patients with SLE relative to controls might already indicate its efficacy in fracture prevention.

In breast cancer trials, tamoxifen was shown to increase lumbar spine BMD (13) and to reduce spine, radius, and hip fractures (13, 14). In the MORE study, raloxifene significantly reduced incident vertebral fractures in patients with and those without preexisting vertebral fractures, for up to 4 years (8, 15). There is no information regarding the efficacy of raloxifene in corticosteroid-treated patients. Our study showed that raloxifene preserved BMD at the femoral neck and the lumbar spine in such patients, and a reduction in vertebral fractures might be obvious after a longer period of observation.

The effect of exogenous estrogens on SLE flares has been a subject of controversy. The SELENA study showed that HRT increased the incidence of mild/moderate lupus flares (11). Although the incidence of major lupus flares was not increased, the recent Women's Health Initiative study demonstrated that HRT increased the risk of invasive breast cancer as well as arterial and venous thromboembolism in healthy postmenopausal women (16). These issues, together with the fact that SLE itself is a risk factor for accelerated atherosclerosis, have rendered HRT out of favor in SLE patients, except in women with disabling climacteric symptoms.

Our study showed that major flares of lupus did not occur in patients who were receiving raloxifene. In fact, the incidence of mild/moderate disease flares in raloxifene-treated patients was lower than that in controls. This may be attributable to the fact that all patients had stable disease at the time of entry into the study, or to the small number of patients studied. Thus, close monitoring for disease flares is still necessary in SLE patients who are receiving raloxifene.

Raloxifene may have more potential roles in the management of SLE. Subanalyses of the MORE study results suggested that raloxifene might reduce the number of new cardiovascular events in patients who were at risk (17). A recent comparative study showed a higher incidence of arterial thromboembolism in Chinese patients with SLE compared with the incidence in African Americans and Caucasians (18). The positive effect of raloxifene on cholesterol levels, as shown in the current study, and its possible protective effect on the heart is a potential advantage of the drug in the treatment of osteoporosis, especially in Chinese patients with SLE.

Venous thromboembolism is a concern associated with raloxifene treatment. In the MORE study, treatment with raloxifene led to a 3-fold increased risk of venous thromboembolism (8). Venous thromboembolism is generally less common in Asians as well as in Chinese patients with SLE when compared with Caucasians (18, 19). Both the current study and a recent trial involving 483 postmenopausal Asian women (12) did not demonstrate any case of venous thromboembolism after 12 months of raloxifene treatment. Thus, venous thromboembolism is of less concern when raloxifene is used in Chinese patients, especially in those without risk factors for hypercoagulability.

In summary, we have shown in this preliminary study that raloxifene preserved BMD at the femoral neck and lumbar spine in postmenopausal patients with SLE who were receiving corticosteroids. Raloxifene was well tolerated in Chinese patients who had inactive SLE and who were not known to be hypercoagulable. Further multicenter double-blind placebo-controlled trials are necessary to reinforce our results.