Results

Patient disposition and characteristics

A total of 30 patients were enrolled: 6 in the ROMO group and 24 in the BP group. However, 2 patients in the BP group withdrew their consent and were therefore excluded from all analyses. The baseline characteristics of patients are summarized in Table 1. The age of the ROMO group was significantly higher than that of the BP group (78.5 yr vs 64.5 yr; p = .001). There was no significant difference in the male-to-female ratio between the 2 groups. Although not statistically significant, the daily GC dose tended to be higher in the ROMO group (8.0 mg vs 5.0 mg; p = .19), and the duration of BP administration before the start of the study tended to be longer in the BP group (25.0 mo vs 58.5 mo; p = .47). The history of vertebral fractures prior to the start of the study was significantly more common in the ROMO group. Baseline BMD of the LS was comparable between the 2 groups; however, the ROMO group showed lower BMD of the FN and TH. Urinary pentosidine levels were significantly higher in the ROMO group (p < .0001). Baseline values for other bone turnover markers and molecules regulating bone metabolism were similar between the 2 groups.

 | ROMO group (n = 6) | BP group (n = 22) | p value
Age (years) | 78.5 [74.5-82.3] | 64.5 [49.8-71.5] | .001
Male/female | 0/6 | 3/19 | .56
Postmenopausal women (%) | 6 (100) | 16 (84.2) | .55
BMI (kg/m 2) | 19.8 [16.5-21.8] | 20.9 [19.0–22.9] | .17
Daily prednisolone dose (mg) | 8.0 [5.0-14.6] | 5.0 [5.0–6.3] | .19
Duration of bisphosphonates use (months) | 25.0 [6.0-170.8] | 58.5 [17.8-110.3] | .47
Prior vertebral fracture, no. (%) | 2 (33.3) | 0 (0) | .04
Prior nonvertebral fracture, no. (%) | 0 (0) | 1 (4.2) | >.99
Diagnosis, no. (%) |  |  | 
Vasculitis syndrome | 3 (50.0) | 5 (20.8) | 
Systemic lupus erythematosus | 1 (16.7) | 6 (25.0) | 
Systemic sclerosis | 1 (16.7) | 1 (4.2) | 
Rheumatoid arthritis | 1 (16.7) | 0 (0) | 
Adult-onset Still’s disease | 0 (0) | 2 (8.3) | 
IgG4 related disease | 0 (0) | 2 (8.3) | 
Polymyositis/Dermatomyositis | 0 (0) | 2 (8.3) | 
Behçet’s disease | 0 (0) | 1 (4.2) | 
Sjögren’s disease | 0 (0) | 1 (4.2) | 
Mixed connective tissue disease | 0 (0) | 1 (4.2) | 
Relapsing polychondritis | 0 (0) | 1 (4.2) | 
Serum markers |  |  | 
eGFR (mL/min/1.73m 2) | 71.4 [64.6-78.0] | 64.6 [56.4-75.5] | .37
25OHD (ng/mL) | 15.5 [8.2–18.3] | 15.9 [12.7-17.8] | .59
P1NP (μg/L) | 19.0 [13.6-70.8] | 23.6 [15.7-34.6] | .97
OC (ng/mL) | 4.8 [3.5–6.6] | 6.8 [5.7-11.6] | .05
TRACP-5b (mU/dL) | 279.5 [225.5-340.3] | 181.0 [132.3-266.3] | .08
Sclerostin (pg/mL) | 42.1 [26.3-74.4] | 97.3 [46.7-154.4] | .07
Dkk-1 (pg/mL) | 388.4 [258.2-495.6] | 443.4 [303.5-597.7] | .49
Wnt3a (ng/mL) | 6.2 [5.0–8.0] | 4.2 [2.7-5.1] | .01
Wnt3a/sclerostin | 0.16 [0.08–0.32] | 0.04 [0.02-0.09] | .02
Wnt3a/Dkk-1 | 0.02 [0.01-0.02] | 0.01 [0.01–0.02] | .03
RANKL (pg/mL) | 220.5 [20.7-248.4] | 137.1 [46.8-293.8] | >.99
OPG (pg/mL) | 3384 [3028-4197] | 2848 [2102-3309] | .06
RANKL/OPG | 0.07 [0.01–0.01] | 0.04 [0.02-0.11] | .77
Urine pentosidine (pmol/mg Cr) | 18.3 [13.9-26.5] | 10.0 [6.9–11.4] | <.0001
BMD (g/cm 2) |  |  | 
LS | 0.88 [0.79-1.07] | 0.85 [0.82-1.00] | .60
FN | 0.54 [0.43-0.62] | 0.63 [0.59-0.72] | .03
TH | 0.64 [0.45-0.75] | 0.76 [0.73-0.84] | .01

Changes in BMD

This study was analyzed using the FAS; consequently, 2 patients in the BP group who withdrew their consent by month 6 and had no subsequent data available were excluded from the analysis. There was no significant difference in cumulative annual GC dose between the 2 groups, but it tended to be higher in the ROMO group (median [25th**-75th percentile]: ROMO group: 2442 [1755-3293] mg; BP group: 1823 [1820-**2226] mg; p = .72).

Figure 1 shows the median percent change from baseline in BMD of the LS, FN and TH at 6 and 12 mo in each group. The median change in LS BMD [25**-75th percentile] showed a slight decrease in the ROMO group at 6 mo (−0.3 [−5.4−1.1] %), but an increase at 12 mo (2.6 [0.8-**3.4] %). In contrast, the BP group showed an increase at 6 mo (0.2 [−0.8−2.0] %) and at 12 mo (1.6 [−1.9−4.3] %) (Figure 1A). However, the percent change in LS BMD at both time points was not statistically significantly different from baseline in either group. The ROMO group exhibited a greater percentage increase in LS BMD at 12 mo compared to the BP group; however, this difference was not statistically significant.

Figure 1: Percent changes in BMD from baseline in patients with glucocorticoid-induced osteoporosis in each treatment group of the LS (A), FN (B), and TH (C). Changes in the ROMO group (closed circles), and BP group (open quadrangles) from base line. Data are expressed as medians with interquartile ranges. Abbreviations: BP, bisphosphonate; ROMO, romosozumab.

The median change in FN BMD was 5.5 [−0.6−9.3] % at 6 mo and 1.4 [−0.5−5.4] % at 12 mo in the ROMO group. Increases were also observed in the BP group at 6 mo (2.5 [−2.1−5.5] %) and 12 mo (1.0 [−4.7−5.6] %), but no significant difference compared to baseline was observed, nor was there a significant difference between the 2 groups (Figure 1B). The median change in TH BMD increased at 6 mo (4.2 [−8.1−8.3] %) and at 12 mo (1.1 [−7.6−6.5] %) in the ROMO group. The BP group showed decreases at both time points (−0.9 [−6.7−2.2] % and −1.8 [−6.4−1.8] %, respectively) (Figure 1C). There were no significant differences from baseline or between the 2 groups at either time point. However, numerically, the ROMO group showed a greater percentage increase in BMD of the FN and TH at all time points compared to the BP group.

Given that the ROMO group was significantly older than the BP group, we compared the percent changes in BMD from baseline to 12 mo between the 2 groups using an age-adjusted linear regression model (Table 2). The ROMO group did not show significantly different percent changes in BMD from baseline to 12 mo compared with the BP group at any site, including the LS (β = .002, 95% CI −0.052 to 0.056, p = .941), FN (β = −.030, 95% CI −0.103 to 0.043, p = .427), and TH (β = .041, 95% CI −0.023 to 0.106, p = .222). Nonetheless, the percent changes in BMD of the LS and TH tended to be slightly greater in the ROMO group, while the FN showed a slightly smaller percent change in the ROMO group.

Outcome | Variable | β | 95% CI | p value
LS | ROMO vs BP | 0.002 | −0.052 to 0.056 | .94
Age | 0.000 | −0.001 to 0.002 | .73
FN | ROMO vs BP | −0.030 | −0.103 to 0.043 | .43
Age | 0.002 | 0.000 to 0.004 | .02
TH | ROMO vs BP | 0.041 | −0.023 to 0.106 | .22
Age | −0.001 | −0.003 to 0.001 | .25

Bone turnover markers

Serum levels of P1NP, a marker of bone formation, increased in the ROMO group at 3 mo but declined to approximately baseline values from 6 mo onward. In contrast, P1NP levels decreased in the BP group. The increase was significantly greater in the ROMO group compared to the BP group at 3 and 12 mo (Figure 2A). Serum levels of OC, another bone formation marker, increased in the ROMO group from 3 mo, but decreased from 9 mo onward. Conversely, the BP group showed no significant changes from 3 mo onward. The increase was significantly greater in the ROMO group compared to the BP group at 3, 6, and 12 mo (Figure 2B).

Figure 2: Percent changes from baseline in serum P1NP (A), OC (B), TRACP-5b (C), and urine pentosidine (D) levels in each treatment group of patients with GC-induced osteoporosis. Changes in the ROMO group (closed circles), and BP group (open quadrangles) from base line. Data are expressed as medians with interquartile ranges. *p < .05, **p < .01 vs the baseline within each treatment group by Dunn’s multiple comparison test. #ROMO vs BP (#p < .05; ##p < .01) by Bonferroni’s multiple comparison test. Abbreviations: BP, bisphosphonate; OP, osteocalcin; ROMO, romosozumab; TRACP-5b, tartrate-resistant acid phosphatase 5b.

Serum levels of TRACP-5b, a bone resorption marker, decreased in both groups from 3 mo onward. Notably, the ROMO group demonstrated statistically significant reductions at 6 and 9 mo relative to baseline values and compared to the BP group (Figure 2C). Additionally, urinary pentosidine, a marker inversely associated with bone quality, decreased in both groups following the 3-mo period (Figure 2D). However, no significant differences were detected in percent change from baseline either within each group or between the 2 groups.

Molecules regulating bone metabolism

We assessed the levels of Wnt signaling inhibitors, sclerostin and Dkk-1, along with the Wnt signaling ligand Wnt3a. Serum sclerostin levels showed a significant increase in the ROMO group starting at 3 mo and remained elevated thereafter (Figure 3A). In the BP group, serum sclerostin levels increased gradually over time, but no significant difference was observed compared to baseline. Comparing the ROMO and BP groups, the rate of increase in the ROMO group was significantly higher at 6 and 12 mo. Serum Dkk-1 levels declined from 6 mo onward in the ROMO group, whereas in the BP group, they decreased at 6 mo but increased at 12 mo (Figure 3B). The rate of increase in each group, when compared to their respective baseline levels or between the 2 groups, showed no significant difference. Serum Wnt3a levels increased from 3 mo in the ROMO group and at 12 mo in the BP group; however, the increase did not reach statistically significance. Although the ROMO group exhibited a higher rate of increase compared to the BP group, the difference was not statistically significant (Figure 3C). The ratio of serum Wnt3a to sclerostin significantly decreased from 6 mo in the ROMO group, whereas no significant change was observed in the BP group. In the ROMO group, this ratio was significantly lower than that of the BP group from 6 mo (Figure 3D). The ratio of serum Wnt3a to Dkk-1 increased in both groups from 3 mo; however, it did not show a significant increase compared to baseline throughout the study period (Figure 3E).

Figure 3: Percent changes from baseline in serum sclerostin (A), Dkk-1 (B), Wnt3a (C), Wnt3a/sclerostin (D), and Wnt3a/Dkk-1 (E) levels in each treatment group of patients with GC-induced osteoporosis. Changes in the ROMO group (closed circles), and BP group (open quadrangles) from base line. Data are expressed as medians with interquartile ranges. *p < .05 vs the baseline within each treatment group by Dunn’s multiple comparison test. #ROMO vs BP (#p < .05; ####p < .0001) by Bonferroni’s multiple comparison test. Abbreviations: BP, bisphosphonate; Dkk-1, Dickkopf-1; ROMO, romosozumab.

We also measured serum levels of RANKL, a molecule involved in osteoclast differentiation and maturation, and OPG, a RANKL inhibitor. Serum levels of RANKL increased slightly from 6 mo in the ROMO group, whereas they decreased from 6 mo in the BP group (Figure 4A). However, neither group showed statistically significant changes compared to baseline, and no significant differences were observed between the 2 groups at any time point. Regarding serum levels of OPG, a non-significant decline was observed from 6 mo in the ROMO group and at 12 mo in the BP group; however, no statistically significant differences were detected between the groups (Figure 4B). The ratio of serum RANKL to OPG decreased from 6 mo onward in the ROMO group. In contrast, the BP group showed a reduction at 3 mo, followed by a return to baseline levels thereafter (Figure 4C). These changes did not reach statistical significance when compared to baseline within each group or between groups.

Figure 4: Percent changes from baseline in serum RANKL (A), OPG (B), and RANKL/OPG (C) levels in each treatment group of patients with GC-induced osteoporosis. Changes in the ROMO group (closed circles), and BP group (open quadrangles) from base line. Data are expressed as medians with interquartile ranges. Abbreviations: BP, bisphosphonate; OPG, osteoprotegerin; ROMO, romosozumab.

New fractures

During the 12-mo follow-up, 1 new vertebral fracture was observed only in the BP group. No non-vertebral fractures were reported in either group.

Adverse events

A single adverse event was observed in each of the ROMO and BP groups. In the ROMO group, 1 patient—a 73-yr-old female diagnosed with vasculitis syndrome—experienced a cerebral infarction following the administration of 7 doses of ROMO, leading to discontinuation of the treatment; however, the causal association with ROMO is undetermined. In the BP group, 1 patient developed femoral head necrosis, with the causal relationship to the treatment also remaining unclear. No treatment-related deaths were observed.