Introduction

Vitamin D is vital for bone and mineral health [1], and a deficiency/insufficiency of vitamin D is associated with several musculoskeletal clinical conditions [2]; therefore, guidelines for the proper management of serum vitamin D levels have been published worldwide [2,3,4,5,6,7,8]. Moreover, it has been reported that serum 25-hydroxy vitamin D (25[OH]D) levels may be able to influence the effects of anti-osteoporosis treatments, such as bisphosphonates or selective estrogen receptor modulators, with low 25(OH)D associated with inadequate treatment responses [9,10,11,12,13,14].

In many clinical trials for anti-osteoporosis drugs, participants are supplemented with vitamin D and/or calcium (Ca) to achieve sufficient efficacy [15,16,17,18,19,20,21,22,23,24]. However, the optimal dosages for vitamin D and Ca supplementation are unclear, and the amounts used have differed widely between trials: 0–1200 IU/day (vitamin D) and 200–1500 mg/day (Ca) [15,16,17,18,19,20,21,22,23,24].

Denosumab is a fully human monoclonal IgG2 antibody against receptor activator of nuclear factor-kB ligand (RANKL) [25] that prevents the survival and activation of osteoclasts by inhibiting RANK activation. In the FREEDOM trial in postmenopausal women, subjects received denosumab (60 mg every 6 months) or placebo, alongside vitamin D (at least 400–800 IU/day) and Ca (at least 1000 mg/day). Denosumab was found to increase bone mineral density (BMD) and decrease bone turnover markers and fracture risk significantly, compared with placebo [26]. In the Japanese, phase 3 Denosumab Fracture Intervention Randomized Placebo-Controlled (DIRECT trial), denosumab (60 mg every 6 months) was administered alongside vitamin D (at least 400 IU/day) and Ca (at least 600 mg/day), and treatment benefits were observed in subjects with osteoporosis who had prevalent mild or moderate vertebral fractures [27].

Given the various dosages of vitamin D/Ca used between the international and Japanese studies, the optimal dose of these supplements alongside denosumab remains to be determined. It is also unclear whether the doses of vitamin D and Ca need to be adjusted according to the baseline 25(OH)D level. In the FREEDOM trial, vitamin D supplementation varied according to serum 25(OH)D: at least 800 IU/day for subjects with ≤ 20 ng/mL serum 25(OH)D, and at least 400 IU/day for subjects with > 20 ng/mL of serum 25(OH)D [26]; however, in the DIRECT trial, at least 400 IU/day vitamin D was administered regardless of serum 25(OH)D concentration [27]. Ca dosage was not adjusted in either study based on 25(OH)D. Furthermore, whether baseline serum vitamin D levels have an impact on denosumab effects, such as BMD increase and fracture risk reduction, is unknown. Using data from the DIRECT trial, this study assessed the changes in serum 25(OH)D level and the effects of denosumab co-administered with vitamin D and calcium on BMD and fracture risk according to baseline 25(OH)D level.

Materials and methods

Study design

The DIRECT trial (NCT00680953) was a randomized, double-blind, placebo-controlled trial. The complete study methodology has been described previously [27]. In brief, denosumab 60 mg or matching placebo was administered subcutaneously every 6 months during a 24-month period (at 0, 6, 12, and 18 months) in a double-blind manner. All subjects who received denosumab or placebo also received daily supplements containing Ca ≥ 600 mg and vitamin D ≥ 400 IU throughout the study period. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki, the International Conference on Harmonization E6 guideline for Good Clinical Practice, and all applicable national ordinances. Approval was obtained from the Institutional Review Board at each study center, and informed consent was obtained from all individual participants included in the study.

The detailed inclusion and exclusion criteria have been described [27]. Subjects with one to four prevalent vertebral fractures and a BMD T-score of less than − 1.7 (Young Adult Mean in Japan 80%) at the lumbar spine or − 1.6 (Young Adult Mean in Japan 80%) at the total hip by dual-energy X-ray absorptiometry (DXA) based on the diagnostic criteria of primary osteoporosis in Japan [28] were included in the DIRECT trial. Subjects with serum 25(OH)D levels < 12 ng/mL; with more than two moderate (grade 2) and/or any severe (grade 3) vertebral fractures on lateral spine radiographs by semiquantitative grading [29]; or with evidence of conditions such as hyperparathyroidism, hypoparathyroidism, hypercalcemia, hypocalcemia, rheumatoid arthritis, or Paget’s disease of the bone were excluded.

The present report describes a sub-analysis of subjects with primary osteoporosis from the DIRECT trial [27], focusing on serum levels of 25(OH)D at baseline; this analysis was prespecified at the time the DIRECT trial was planned. The subjects in the denosumab and placebo arms in the DIRECT trial were analyzed in this study. Both arms were stratified into three groups based on serum 25(OH)D concentration at baseline, according to current Japanese criteria [2]: deficient (< 20 ng/mL); insufficient (≥ 20 to < 30 ng/mL); and sufficient (≥ 30 ng/mL).

Study measures

The endpoints of this sub-analysis were: absolute serum 25(OH)D level at baseline and after 12 and 24 months of treatment; change in serum 25(OH)D status from baseline; serum 25(OH)D concentrations in the denosumab and placebo groups at the last measurement by baseline levels; percent change in BMD from baseline to 24 months at the lumbar spine (L1–L4), total hip, femoral neck, and distal one-third of the radius by baseline serum 25(OH)D levels; and the incidence of new vertebral fractures at 24 months stratified by baseline serum 25(OH)D levels.

Details of sampling have been described elsewhere [27]. Serum samples were transported to LSI Medience Co., Ltd. (formerly Mitsubishi Chemical Medience Co., Ltd., Tokyo Japan), where the measurement of serum 25(OH)D concentrations was performed by competitive protein binding assay. BMD was evaluated by DXA (QDR® instrument; Hologic Inc., Marlborough, MA, USA) measured at the lumbar spine (L1–L4), total hip, femoral neck, and distal one-third of the radius at baseline, 3 months (except for the distal radius), and 6, 12, 18, and 24 months. Quality control and BMD scan analysis were performed centrally (Synarc Inc., Portland, OR, USA).

As previously reported [27], vertebral fractures were identified from anteroposterior and lateral spine radiographs, taken at baseline and then every 6 months. The definition applied for a new vertebral fracture was an increase of ≥ 1 semiquantitative grading scale in a vertebral body that was normal at baseline, and ≥ 20% loss of height from baseline at the anterior, posterior, or central vertebra.

Statistical analyses

As analysis targets, the safety analysis set (comprising all subjects who received at least one dose of the study drug) was used for the summary of 25(OH)D. The full analysis set (comprising all randomized subjects, with the exception of those who did not have osteoporosis at screening, those who did not receive the study drug or those who had no available efficacy data after the first dose of the study drug) was used for the analysis of BMD and new vertebral fractures. Missing data were not imputed for 25(OH)D, and the last observation carried forward method was used for BMD data. All tests were performed at a two-sided 5% significance level.

The change in 25(OH)D was displayed as a transition chart over time or scatter chart. Student’s t-test was used to compare the percent changes from baseline in BMD at 24 months between the placebo and denosumab groups and among each subgroup stratified by baseline 25(OH)D levels (< 20 ng/mL, ≥ 20 to < 30 ng/mL, or ≥ 30 ng/mL). The incidences of new vertebral fractures at 24 months were displayed as a bar chart and stratified by baseline 25(OH)D levels. A log-rank test was used to compare the placebo and denosumab groups with a hazard ratio (HR) based on the proportional hazards model stratified by baseline 25(OH)D levels. The grouped survival data approach was applied for the log-rank test, and the estimation of the HR was used to statistically analyze new vertebral fracture data as most new vertebral fractures were observed at scheduled visits as previously described [27]. A post hoc analysis was performed using a Student’s t-test to compare 25(OH)D concentrations between baseline and each time point. The software used for statistical analysis was SAS version 9.1.3 (SAS Institute Inc., Cary, NC, USA).

Results

Subject characteristics

The safety analysis set from the DIRECT trial, which comprised 481 placebo-treated subjects and 475 denosumab-treated subjects, was used for the analyses in this study. Baseline characteristics, according to 25(OH)D levels (deficient, insufficient, and sufficient), are shown in Table 1. The background characteristics at baseline were similar between the placebo and the denosumab groups and across each subgroup except for serum 25(OH)D concentration. These background characteristics are in accordance with the data reported in the main study [27].

Table 1 Subject characteristics by serum 25(OH)D level at baseline
Full size table

Absolute serum 25(OH)D level after 12 and 24 months of treatment

Mean serum 25 (OH)D levels were significantly increased by supplementation in both placebo- and denosumab-treated subjects, even with 400 IU vitamin D supplements. The mean ± standard deviation (SD) serum 25(OH)D concentrations of the placebo group were 20.6 ± 5.9 ng/mL, 32.9 ± 8.9 ng/mL, and 30.9 ± 7.0 ng/mL at baseline, 12 months, and 24 months, respectively. Mean ± SD serum 25(OH)D concentrations of the denosumab group were 21.0 ± 6.1 ng/mL, 32.2 ± 8.9 ng/mL, and 31.0 ± 6.8 ng/mL at baseline, 12 months, and 24 months, respectively.

The serum 25(OH)D concentrations at baseline, 12 months, and 24 months were similar between the two groups (Fig. 1). In both groups, the mean value of serum 25(OH)D concentrations at 12 and 24 months were all ≥ 30 ng/mL. The mean values of serum 25(OH)D concentrations at 12 or 24 months were significantly higher than at baseline in both groups (P < 0.0001). No difference was observed between the 25(OH)D concentrations at 12 and 24 months (Fig. 1).

Fig. 1
figure 1

Mean (± standard deviation) serum 25(OH)D concentrations during the study. 25(OH)D 25-hydroxy vitamin D

Full size image

Changes in serum 25(OH)D status from baseline

In both groups, the majority of subjects (placebo 58.3%, denosumab 57.1%) had sufficient levels of 25(OH)D at 24 months (Fig. 2a, b).

Fig. 2
figure 2

Change in serum 25(OH)D status and concentrations from baseline. Change in serum 25(OH)D in the placebo (a) and denosumab groups (b) stratified by serum 25(OH)D status. Scatter charts illustrate the change in serum 25(OH)D concentrations at the last measurement from baseline in the placebo (c) and denosumab groups (d). 25(OH)D 25-hydroxy vitamin D

Full size image

At the last measurement, 25(OH)D was increased compared with baseline in 89.6% (395/441) of subjects in the placebo group and 88.0% (387/440) in the denosumab group (Fig. 2c, d). The serum 25(OH)D status at baseline increased from deficient to insufficient or sufficient at the last measurement in 20.4% (90/441) and 26.8% (118/441) of subjects in the placebo group, and 19.3% (85/440) and 29.8% (131/440) of subjects in the denosumab group, respectively (Table 2). The serum 25(OH)D status at baseline increased from insufficient to sufficient at the last measurement in 26.5% (117/441) of subjects in the placebo group and 21.1% (93/440) of subjects in the denosumab group (Table 2).

Table 2 Proportions of subjects in each serum 25(OH)D level category at last measurement according to baseline serum 25(OH)D levels
Full size table

Changes in BMD from baseline by baseline serum 25(OH)D level

The percent changes in BMD for 24 months by serum 25(OH)D status at baseline (deficient, insufficient, and sufficient) in the denosumab group at three sites were as follows: 9.10%, 9.10%, and 9.29% at the lumbar spine (L1–L4); 4.60%, 4.52%, and 5.19% at the total hip; and 3.93%, 3.66%, and 5.71% at the femoral neck, respectively (Fig. 3a–c). All changes were greater than the corresponding placebo groups (0.27%, − 0.12%, and − 0.13% at the lumbar spine [L1–L4]; − 0.98%, − 1.24%, and − 0.99% at the total hip; and − 0.97%, − 1.36%, and − 0.69% at the femoral neck).

Fig. 3
figure 3

Change in BMD from baseline to 24 months by baseline serum 25(OH)D levels. The percent change in BMD from baseline to 24 months stratified by baseline serum 25(OH)D levels according to site: lumbar spine (a), total hip (b), femoral neck (c), and distal 1/3 radius (d); and difference in percent change from placebo in the denosumab groups at 24 months (e). For panels (a–c), and (d), the bars show 95% CIs of the mean values at each time point. For panel (e), P < 0.0001 for all comparisons vs placebo was based on a Student’s t-test. Bars show 95% CIs of the difference from placebo at 24 months. 25(OH)D 25-hydroxy vitamin D, BMD bone mineral density, CI confidence interval

Full size image

At the distal one-third of the radius, the percent changes in BMD from baseline to 24 months by serum 25(OH)D status at baseline (deficient, insufficient, and sufficient) in the denosumab groups were 0.60%, 0.46%, and 0.07%, respectively, showing a slight increase in BMD, while decreases were observed in the corresponding placebo groups (− 1.82%, − 1.53%, and − 3.44%) (Fig. 3d).

Figure 3e shows the differences from placebo in mean percent change in BMD from baseline to 24 months in the denosumab group by baseline 25(OH)D levels (deficient, insufficient, and sufficient) as follows: 8.83%, 9.22%, and 9.42% at the lumbar spine (L1–L4); 5.58%, 5.76%, and 6.19% at the total hip; 4.89%, 5.02%, and 6.40% at the femoral neck; and 2.43%, 1.99%, and 3.51% at the distal one-third of the radius, respectively. Compared with placebo, denosumab significantly (all, P < 0.0001) increased bone density regardless of baseline serum 25(OH)D level at the four sites evaluated.

Fracture risk by baseline serum 25(OH)D level

The incidences of new vertebral fractures at 24 months, stratified by baseline 25(OH)D levels (deficient, insufficient, and sufficient), were 1.2%, 3.4%, and 2.2% in the denosumab group and 7.8%, 9.0%, and 2.9% in the placebo group, respectively (Fig. 4). Denosumab-treated subjects with deficient or insufficient 25(OH)D at baseline had a significantly lower risk of a new vertebral fracture compared with placebo-treated subjects (HR 0.149, 95% confidence interval [CI] 0.044–0.504, P = 0.0004, and HR 0.365, 95% CI 0.145–0.919, P = 0.0256, respectively). For subjects with sufficient 25(OH)D at baseline, there was no significant difference observed between denosumab and placebo in the risk of a new vertebral fracture (HR 0.789, 95% CI 0.049–12.629, P = 0.8667).

Fig. 4
figure 4

Incidence of new vertebral fracture at 24 months by baseline serum 25-hydroxy vitamin D levels

Full size image

Discussion

Although the potential influence of serum 25(OH)D levels on treatment responses to anti-osteoporosis treatments are well-recognized [9,10,11,12,13,14], to date, no studies have analyzed the effect of denosumab according to baseline serum 25(OH)D concentration. This study is the first to reveal the effects of denosumab on BMD and fracture prevention stratified by serum 25(OH)D level at baseline.

In this subgroup analysis of the randomized DIRECT trial [27], the impact of denosumab on BMD increase and fracture prevention seemed unaffected by baseline 25(OH)D levels in subjects supplemented with ≥ 400 IU/day vitamin D and ≥ 600 mg/day calcium. According to the present data, there is no need to increase the dose of supplemented vitamin D based on 25(OH)D sufficiency at baseline to obtain the effects of denosumab. Furthermore, the efficacy of denosumab was confirmed in subjects receiving supplementation with just 400 IU/day vitamin D, which is lower than in the FREEDOM trial [26]. Furthermore, in the present study, no difference was observed between the groups even when stratified by the serum 25(OH)D concentration after 24 months (Online Resource 1). In addition, after 24 months, there was no increase in the incidence of new vertebral body fractures in the denosumab group according to the serum 25(OH)D concentration (Online Resource 2). If subjects are supplemented with vitamin D ≥ 400 IU/day and calcium ≥ 600 mg/day, the effect of denosumab is likely to be obtained regardless of the serum 25(OH)D concentration.

Our results are consistent with the results of previous reports evaluating the impact of baseline 25(OH)D on efficacy outcomes with anti-osteoporotic treatment agents plus vitamin D supplementation [30, 31], although it must be remembered that supplementation dosage varied in each trial. The MORE study, which evaluated the effect of raloxifene on BMD and fractures in postmenopausal women with osteoporosis, supplemented subjects with 400–600 IU/day vitamin D and 500 mg/day Ca throughout the study [21, 30]. The FIT study, which evaluated the efficacy of alendronate in women with low BMD, supplemented subjects with 250 IU/day vitamin D and 500 mg/day Ca throughout the study [15, 31]. Both studies showed that baseline 25(OH)D level did not influence the efficacy of raloxifene or alendronate [30, 31]. In contrast, in studies in which no vitamin D supplementation was administered, low baseline levels of 25(OH)D were found to correlate with inadequate response to treatment [9, 10, 12, 14]. In bisphosphonates studies, the effects of treatment have been evaluated without the administration of vitamin D and Ca supplementation. In contrast, no studies, as far as we know, have examined the association between baseline serum 25(OH)D and the effect of denosumab without vitamin D and Ca supplementation. Therefore, while we consider that vitamin D and Ca supplementation may have contributed to the effect of denosumab independent of the baseline serum 25(OH)D concentration, we cannot draw a definitive conclusion based on these findings.

Unlike previous studies such as MORE and FIT, our study also examined the effect of baseline 25(OH)D on the incidence of fractures during the study. We believe this is the first such published analysis, and our data confirm that the number of new vertebral fractures was reduced regardless of baseline vitamin D status. The incidence rate of new vertebral fracture was significantly suppressed in the denosumab group compared with the placebo group in subjects with deficient/insufficient 25(OH)D at baseline. Conversely, there was no significant difference between treatment groups in subjects with sufficient 25(OH)D at baseline; however, this is likely because the number of subjects with sufficient baseline vitamin D who developed fractures was too small to evaluate the difference between the treatment groups properly.

Supplementation with calcium and vitamin D is aimed at preventing hypocalcemia and promoting an adequate response to anti-osteoporosis drugs. In the DIRECT trial, hypocalcemia, as an adverse event, occurred in two subjects (0.4%) in the denosumab group. However, these events were not serious adverse events [27]. Given that in the present study, the 25(OH)D status at baseline was deficient in about half of the subjects, it is considered that supplementation with ≥ 400 IU/day vitamin D and ≥ 600 mg/day calcium effectively reduced the incidence of hypocalcemia.

Reportedly, among elderly Japanese women, subjects with sufficient vitamin D had a lower incidence of limb and vertebral fractures [32]. Furthermore, other studies reported a correlation between vitamin D insufficiency and fracture risk [33,34,35,36]. Similarly, the subjects in the placebo group with sufficient baseline vitamin D in our study tended to have a lower incidence of new vertebral fractures than subjects in the placebo group with deficient or insufficient vitamin D. However, as previously discussed, the number of subjects in the sufficient group was too small to draw any definitive conclusions.

The proportions of subjects with deficient (< 20 ng/mL), insufficient (≥ 20 to < 30 ng/mL), and sufficient (≥ 30 ng/mL) levels of 25(OH)D [2] at baseline in our study were 53.1%, 37.1%, and 9.9%, respectively, in the denosumab group and 50.9%, 42.0%, and 7.1%, respectively, in the placebo group. These proportions are similar to those reported in a recent cohort study of Japanese women aged ≥ 50 years, in which 52% were classified as deficient, 38% as insufficient, and 10% as sufficient [37]. These proportions are also similar to those reported in a Chinese study for postmenopausal women [38]. However, in a US study, the proportion of subjects with a deficient status (< 20 ng/mL) among women ≥ 70 years old was 33.6%, and the combined percentage of subjects who were deficient and insufficient (< 30 ng/mL) was 76.7% [39]. Both of these incidences are lower than those reported in the placebo group in this study (50.9% and 92.9%, respectively).

Although > 90% of subjects in our study were classified as having deficient or insufficient vitamin D levels at baseline, at 24 months, the mean ± SD serum 25(OH)D level in both treatment groups was > 30 ng/mL (30.9 ± 7.0 ng/mL in the placebo group and 31.0 ± 6.8 ng/mL in the denosumab group), which is classed as sufficient vitamin D, according to Japanese guidelines [2]. Although an increase in 25(OH)D was observed, the amounts of vitamin D (≥ 400 IU/day) and Ca (≥ 600 mg/day) supplemented in this study were lower than those used in non-Japanese studies of denosumab [26] and several other anti-osteoporotic agents [17, 20, 22, 23].

Finally, we noted that the BMD of the distal one-third of the radius in the placebo group decreased over 24 months in the subgroup with sufficient 25(OH)D compared with the subgroups with deficient or insufficient 25(OH)D. In the placebo group, the baseline absolute BMD of the distal one-third of the radius was similar among subgroups stratified by baseline serum 25(OH)D level (data not shown). A greater reduction in BMD in the subgroup with sufficient 25(OH)D was seen in this research. However, we cannot draw a definitive conclusion regarding this finding because of the small number of subjects in that subgroup.

The main limitations of this study were the exclusion of subjects who had < 12 ng/mL of serum 25(OH)D and the fact that the study was conducted only in Japanese subjects; thus, the results may not be directly applicable to other ethnicities. Additionally, the DIRECT study was not powered to investigate nonvertebral fractures. As the number of subjects with nonvertebral fractures was extremely small, the efficacy data are not quantifiable and are not described.

In conclusion, in subjects supplemented with vitamin D (at least 400 IU/day) and Ca (at least 600 mg/day), the baseline sufficiency status of serum 25(OH)D may not affect denosumab efficacy outcomes such as BMD increase and fracture prevention.