High-dose corticosteroids, used for many medical conditions, are associated with rapid bone loss from sites such as the vertebrae, and compression fractures can be observed within months. Recent trials suggest treatment with bisphosphonates or active vitamin D analogs can reduce bone loss and the risk of fracture associated with glucocorticoids, but few studies have directly compared such agents. We conducted a randomized, multicenter, open-label trial to compare the efficacy of alendronate, calcitriol, and simple vitamin D in prevention and treatment of glucocorticoid-induced bone loss. A total of 195 subjects (134 females and 61 males) commencing or already taking glucocorticoids were randomized to one of three groups: calcitriol, 0.5 to 0.75 μg/day; simple vitamin D (ergocalciferol, 30,000 IU weekly) plus calcium carbonate (600 mg daily); or alendronate, 10 mg/day plus calcium carbonate (600 mg daily). Over 2 years, mean lumbar bone mineral density change was +5.9% with alendronate, −0.5% with ergocalciferol, and −0.7% with calcitriol (p < 0.001). At the femoral neck, there was no significant difference in bone mineral density change between the treatments over 2 years: alendronate (+0.9%), ergocalciferol (−3.2%), and calcitriol (−2.2%). Lumbar bone loss varied according to whether patients were starting or receiving chronic glucocorticoids, and there was a significant treatment × prior glucocorticoid use interaction effect. Six of 66 calcitriol subjects, 1 of 61 ergocalciferol subjects, and 0 of 64 alendronate subjects sustained new vertebral fractures. These data do not suggest any difference between simple vitamin D and calcitriol but do show that alendronate was superior to either treatment for glucocorticoid induced bone loss.
Glucocorticoids are widely used for many inflammatory diseases; the associated glucocorticoid-induced osteoporosis is a common clinical problem. The association between osteoporosis and glucocorticoid therapy is well known and was identified shortly after the first use of these drugs in humans in the 1950s. Glucocorticoids are known to affect bone through multiple pathways, influencing mainly bone formation, and to a lesser extent, bone resorption.(1, 2) When high-dose glucocorticoid therapy is used, the loss of bone from sites such as the vertebrae can be rapid, with compression fractures occurring within a few months.
Randomized trials in glucocorticoid bone suggest that treatment with bisphosphonates(3–5) or vitamin D analogs(6–8) can prevent or restore bone loss caused by glucocorticoids. However, there have been few comparative studies of potentially active agents, with no direct comparison of simple versus active vitamin D metabolites or bisphosphonates.
In this study, we compared the efficacy of calcitriol versus simple vitamin D, using ergocalciferol plus calcium, or alendronate plus calcium in the prevention and treatment of glucocorticoid bone loss. Given that bone loss is most rapid within the first 4-6 months, subjects were stratified into two different therapeutic situations: (1) primary prevention in patients commencing glucocorticoids and (2) treatment of patients on chronic glucocorticoids.
MATERIALS AND METHODS
Subjects and treatment
The study was a multicenter, randomized, open-label, parallel group study in four Australian centers: two in Sydney, New South Wales; one in Cottontree, South Queensland; and one in Geelong, Victoria. Patients were allocated to one of three groups for 2 years of treatment: (1) calcitriol (Rocaltrol 0.5-0.75 mg/day); (2) ergocalciferol (Ostelin) 0.25 mg (10,000 IU) three times per week plus 600 mg calcium daily as the carbonate (Caltrate); or (3) alendronate (Fosamax) 10 mg/day plus 600 mg calcium carbonate daily (Caltrate). The dose of calcitriol was 0.75 μg in those receiving >10 mg of prednisolone/day (based on our original findings suggesting this may be the optimal dose in corticosteroid-treated patients(6)) and was reduced to 0.5 μg/day at or below that level (to reduce the risk of hypercalcemia). The dose of alendronate used was greater than the 5 mg daily dose approved by the Food and Drug Administration (FDA), but it is a dose that has been recommended for treating postmenopausal women receiving corticosteroids.(9) The dose of ergocalciferol is greater than that commonly used in postmenopausal osteoporosis, but it was chosen in an attempt to enhance the efficacy of plain vitamin D in comparison with the activated formulation. Patients treated with calcitriol were not given calcium supplements.
Patients of either sex ages 20-80 years receiving or starting glucocorticoids (≥5 mg prednisone) were recruited. Exclusion criteria included medical conditions known to affect bone metabolism or treatment with medications known to affect bone metabolism (apart from glucocorticoids).
Randomization was performed by a contract research organization (Datapharm, Sydney, Australia) using adaptive assignment into three treatment groups based on age (≤40 or >40 years), sex, baseline glucocorticoid dose (≤10 or >10 mg/day prednisone), duration of therapy (≤6 or >6 months), and presence or absence of vertebral fractures.
The primary efficacy endpoint was bone mineral density (BMD), measured by DXA of the lumbar spine at 0, 6, 12, 18, and 24 months. Secondary endpoints were BMD of the femoral neck and total body, which were performed at the same times. The densitometer technician was blinded to treatment group allocation.
X-rays of the thoracic and lumbar spine were performed at 0, 12, and 24 months. Vertebral fracture was assessed centrally by an experienced radiologist (CS) who was blinded to treatment allocation using a semiquantitative scale(10) : 0, a normal vertebra; 1, a mild deformity with a 25% reduction in anterior, middle, or posterior height (or all three); 2, a moderate deformity with a 25-40% reduction any height; and 3, a severe deformity with a reduction of more than 40% in any height. A new vertebral fracture was deemed to have occurred when any grade progressed to a higher grade between visits. The study was not powered to examine incident fractures as an outcome.
Free deoxypyridinoline was measured in urine with an automated solid phase chemiluminescent enzyme labeled immunoassay (Metra Biosystems, Mountain View, CA, USA). Serum osteocalcin was measured with an automated solid phase chemiluminescent enzyme-labeled immunoassay (DPC, Los Angeles, CA, USA). Serum 25OH vitamin D was measured after extraction by a radioimmunoassay procedure using a specific antibody to 25OH vitamin D metabolites (Diasorin, Dusseldorf, Germany).
It was expected that patients starting glucocorticoids would have more rapid bone loss than chronic users. Based on percent change in lumbar BMD as the primary endpoint and a sample size of 26 per group to detect independent treatment group differences, 156 subjects were determined as the required sample size (26 × 3 × 2). This was increased to 196 to allow for dropouts.
The primary analysis was based on an intention-to-treat analysis population and included all available patient data up to 2 years, where missing observations were replaced using the last observation carried forward technique. For 6, 12, 18, and 24 months, the BMD data were analyzed by repeated measured analysis of covariance (ANCOVA). Center, treatment, a center-by-treatment interaction, and prior duration of treatment with glucocorticoids (≤6 or >6 months treatment) were included in the model as factors. Baseline BMD, baseline dose of glucocorticoid, and cumulative dose of glucocorticoids while on study were included in the model as covariates. A treatment-by-glucocorticoid use interaction was included in the model to test for a treatment difference between these subgroups of patients. The compound symmetry variance-covariance structure was selected as the most appropriate to model the data. The assumptions of the model were checked by a residual analysis. Results from the ANCOVA are presented as adjusted means for BMD, where ANCOVA adjusts for disparities in covariates.
There were 195 patients included in the analysis population: 134 females and 61 males, ranging from age 19 to 80 years. One patient who was subsequently diagnosed as having humoral hypercalcemia of malignancy was excluded from the analysis population. The indications for glucocorticoid therapy are shown in Table 1. Twenty-six of the females were premenopausal, and 108 were postmenopausal. The baseline data for patients in each group are shown in Table 2.
Table Table 1. Disease Indications for Glucocorticoid Therapy
Table Table 2. Baseline Demographics (Mean ± SD)
Of the 195 patients, 177 completed the first year, and 142 completed the full 2 years of the study. The reasons for discontinuation were death (2), adverse events (11), noncompliance (15), switched to other therapy (3), and prednisone dose <5 mg per day (22), and there were no apparent differences between groups in withdrawals, adverse events, or deaths.
More patients were randomized as having received glucocorticoids for greater than 6 months than those who had received them for ≤6 months (Table 3). Mean daily prednis(ol)one dose at the commencement of study and cumulative glucocorticoid doses during the first 12 months of the study were significantly greater in the calcitriol than other groups, although these differences were smaller in the second year and were adjusted.
Table Table 3. Glucocorticoid Use (Mean ± SD)
The adjusted means for BMD at the lumbar spine and femoral neck are shown in Fig. 1 for the 24 months of the study. Over 2 years, there was a significant difference between the three treatment groups at the lumbar spine (p < 0.001). Patients treated with alendronate gained bone at the lumbar spine in comparison with the other two groups, but there was no difference observed between calcitriol or ergocalciferol. After 2 years, observed mean lumbar spine BMD percent change was +5.9% with alendronate, −0.5% with ergocalciferol, and −0.7% with calcitriol. In the repeated measures ANCOVA, there were significant effects at the spine related to prior duration of glucocorticoids (p = 0.042), baseline glucocorticoid dose (p = 0.021), and baseline BMD (p < 0.001). There was also a significant interaction between the treatment effect and prior duration of glucocorticoid therapy (p = 0.017). The estimated mean differences in lumbar BMD between the treatment groups relative to duration of glucocorticoid were higher for both calcitriol and alendronate for glucocorticoid users of ≤6 months duration compared with >6 month users. In contrast, ergocalciferol demonstrated a lower BMD of the lumbar spine for glucocorticoid use of ≤6 month duration compared with >6 month users.
Over 2 years, there was no significant difference between the three treatment groups at the femoral neck. The observed mean femoral neck BMD percent change after 2 years was +0.9% with alendronate, −3.2% with ergocalciferol, and −2.2% with calcitriol. However, there were significant effects at the femoral neck related to prior duration of glucocorticoid treatment (p = 0.003), cumulative glucocorticoid dose in the first year (p = 0.008), and baseline BMD (p < 0.001). Similarly, there was no significant differences between treatments at the total body by repeated measures ANCOVA. The mean BMD percent change observed at 2 years was +1.6% with alendronate, −0.4% with ergocalciferol, and −0.6% with calcitriol. In the repeated measures ANCOVA, there were significant effects for total body BMD related to cumulative glucocorticoid dose in the first year (p = 0.02) and baseline BMD (p < 0.001).
Group differences were also observed in BMD change at both lumbar spine and femoral neck over 24 months, according to whether they had received glucocorticoids for >6 or ≤6 months at baseline (Table 4). In the two vitamin D groups, bone loss was more rapid in patients starting glucocorticoids compared with those who had received glucocorticoids for >6 months, where there was little or no loss observed.
Table Table 4. Percent Change in BMD From Baseline to 24 Months According to Duration of Steroid Use at Baseline (Unadjusted for Covariates)
The results of changes in serum 25 hydroxyvitamin D, serum osteocalcin, and urinary deoxypyridinoline/creatinine are shown in Table 5. Serum 25 hydroxyvitamin D increased in the ergocalciferol group as expected. For serum osteocalcin, there was a decrease in the alendronate treated group, but no comparable reduction in the ergocalciferol- or calcitriol-treated groups. For deoxypyridinoline, there seemed to be a tendency to an increase over time in the ergocalciferol- and calcitriol-treated groups with no consistent increase in the alendronate group. However, the values were not particularly high at baseline and the wide range, particularly in the deoxypyridinoline, could explain why there was no apparent and consistent decrease in any group with treatment.
Table Table 5. Biochemical Markers (Mean ± SD)
The study medications were well tolerated with relatively few adverse effects. In the alendronate group, 14% experienced dyspepsia compared with 13% with ergocalciferol and 7% with calcitriol. Hypercalcemia occurred in 7% calcitriol patients compared with 5% ergocalciferol patients and 0% of the alendronate-treated patients. In the three ergocalciferol patients, the hypercalcemia was borderline and considered unlikely to be associated with vitamin D therapy and resolved without intervention.
Vertebral deformities were present at baseline in 22% of patients (14 of 64 alendronate-, 15 of 64 ergocalciferol-, and 13 of 67 calcitriol-treated patients). Seven patients sustained new vertebral fractures during the study (six in the calcitriol group and one in the ergocalciferol group). These seven fractures occurred in four postmenopausal women and three men aged 51, 64, and 76 years. Six of the seven fractures occurred in patients who had been treated with glucocorticoids for more than 6 months at study entry.
There have been a number of recently reported trials in glucocorticoid osteoporosis. However, this is the first to compare potentially active agents. Although open label, subjects were informed that all three treatments were potentially active, based on published studies. The data show that alendronate is superior to simple vitamin D in the treatment of glucocorticoid lumbar spine bone loss with no significant difference between simple vitamin D and calcitriol at the doses used. However an interaction effect between treatment and prior exposure to glucocorticoids suggested calcitriol was associated with less bone loss than simple vitamin D in patients starting glucocorticoids. It is not clear whether the higher cumulative glucocorticoid doses in the calcitriol group may have affected the efficacy results, even after adjustment in the ANCOVA, for this agent. In previous calcitriol studies, in patients using doses up to 1.0 and 0.75 μg, protective effects of calcitriol were observed, albeit with a moderate incidence of hypercalcemia.(6, 11) The lower calcitriol dosage in our study was associated with a lower risk of hypercalcemia but reduced efficacy at the spine and femoral neck compared with these earlier studies.(6, 11) A dose-dependent effect would be another explanation for this discrepancy.
One previous study has compared the efficacy of active vitamin D metabolites compared with simple vitamin D in patients on chronic glucocorticoid therapy.(12) Eighty-five patients on long-term glucocorticoid therapy were randomized to either 1 μg alphacalcidol plus calcium 500 mg daily or 1000 IU cholecalciferol plus 500 mg calcium. The two groups were similar in age, sex, underlying diseases, initial BMD (lumbar spine, mean T score −3.3), and rates of vertebral and nonvertebral fractures. During the 3-year study, a small but significant increase was seen in lumbar spine BMD in the alphacalcidol group (+2.0%, p < 0.0001), with no significant changes at the femoral neck. In the cholecalciferol group, there were no significant changes at either site. By the end of the study, 12 new vertebral fractures had occurred in 10 patients of the alphacalcidol group and 21 in 17 patients of the cholecalciferol group. This study suggests active vitamin D metabolites are superior to simple vitamin D in the treatment of established glucocorticoid osteoporosis. For comparison, the dose of alphacalcidol used in that study is equivalent to 0.5 μg of calcitriol. Consistent with this and other studies,(4, 8, 12) our data suggest calcium/vitamin D is able to minimize further bone loss in patients on chronic low dose glucocorticoid therapy (secondary prevention).
One previous study compared a bisphosphonate with an “active” vitamin D metabolite in glucocorticoid-treated patients.(13) In an open-label study comparing calcium and alphacalcidol treatment with cyclical etidronate in 48 patients after cardiac transplantation, the change in spine BMD at 6 months was −4.6% with alfacalcidol and −7.7% with etidronate, but there was no other “control” group. Two vertebral deformities occurred in the alphacalcidol group over 2 years compared with eight in the etidronate group.
There have been two previous studies of alendronate in glucocorticoid-treated patients. The combined results of these two trials(4) included 477 glucocorticoid-treated subjects who received treatment with alendronate or calcium/vitamin D (800-1000 mg daily plus 250-500 IU daily, respectively). Patients were stratified according to whether they had received glucocorticoids for <4, 4-12, or >12 months. The patients receiving glucocorticoids for <4 months could be considered similar to our ≤6 month stratum. Over 12 months of follow-up, the mean change in lumbar spine BMD in patients in this primary prevention group was +3.0% for alendronate 10 mg/day compared with +4.0% in our alendronate group over the first 12 months. In those who had received chronic glucocorticoids for >12 months, the increase with alendronate was +2.8%, which compares with +4.3% in our other alendronate group (>6 months baseline glucocorticoids) over the first 12 months. Alendronate 5 mg per day was compared with placebo in another randomized controlled study, but bone density was measured at the ultradistal radius site only.(14) After 12 months, mean BMD change was +0.8% with alendronate versus −4.5% with placebo. The effects of alendronate on fracture risk reduction may also be mediated by the inhibition of the apoptosis of osteoblasts and osteocytes as well as by increasing BMD.(15)
A post hoc analysis of incident vertebral fractures in the combined alendronate studies(4) significantly favored alendronate over calcium plus vitamin D in postmenopausal women. In our study, the vertebral fracture rate was higher with calcitriol; however, because the study was not powered for a fracture endpoint and the calcitriol group received a higher cumulative glucocorticoid dose, this must be interpreted with caution. Of interest, no vertebral fractures occurred in premenopausal women or younger men. This is consistent with other studies where no incident vertebral fracture occurred among premenopausal women in four glucocorticoid trials comprising 251 years of patient exposure.(3–6, 16)
The main strength of this study is that it is one of the few “active” comparator studies in the treatment of osteoporosis. The present findings, however, must be interpreted in the context of a number of possible limitations. Baseline lumbar BMD and glucocorticoid doses were significantly greater in the calcitriol than other groups. Although these differences were adjusted for in the analysis, any systematic baseline imbalances with respect to the main factors and covariates can affect the validity of the results.
To summarize, the present data suggest alendronate is superior to vitamin D in the treatment of glucocorticoid bone loss, particularly in the initial rapid bone loss phase from the lumbar spine. There was no significant benefit of calcitriol over simple vitamin D. Because evidence suggests that postmenopausal women and older men receiving glucocorticoids are at the greatest risk of rapid bone loss and consequent vertebral fracture, they should be actively considered for prophylactic measures.