Further Nonvertebral Fracture Reduction Beyond 3 Years for Up to 10 Years of Denosumab Treatment

Abstract

Context

Evidence for further nonvertebral fracture (NVF) reductions with long-term antiresorptive therapy in osteoporosis is lacking.

Objective

To evaluate NVF risk reduction in subjects receiving ≤10 years of denosumab treatment.

Design

Phase 3, randomized, placebo-controlled, 3-year Fracture Reduction Evaluation of Denosumab in Osteoporosis Every 6 Months (FREEDOM) trial (NCT00089791) and its open-label 7-year extension (NCT00523341).

Setting

One hundred seventy-two study centers worldwide.

Patients

Women 60 to 90 years, lumbar spine or total hip bone mineral density T-scores <−2.5 (≥−4.0 at both).

Interventions

Subjects randomly assigned 1:1 denosumab 60 mg SC Q6M (long-term) or placebo (crossover) in FREEDOM; eligible subjects could enroll in the extension to receive denosumab 60 mg SC Q6M.

Main Outcome Measures

NVF Exposure-adjusted subject incidence (per 100 subject-years) during denosumab treatment years 1 to 3 and 4 to 7 (all subjects) and years 4 to 10 (long-term only), and rate ratios (RRs) for years 4 to 7 or 4 to 10 vs 1 to 3.

Results

Among 4074 subjects (2343 long-term, 1731 crossover), NVF rates (95% CI) in all subjects were 2.15 (1.90 to 2.43) during years 1 to 3 and 1.53 (1.34 to 1.75) during years 4 to 7 of denosumab treatment [RR (95% CI) = 0.72 (0.61 to 0.86); P < 0.001]; in long-term only were 1.98 (1.67 to 2.34) during years 1 to 3 and 1.44 (1.24 to 1.66) during years 4 to 10 [RR = 0.74 (0.60 to 0.93); P = 0.008]. combined osteonecrosis of the jaw and atypical femoral fracture rate was 0.06.

Conclusions

Long-term denosumab treatment, >3 and ≤10 years, was associated with further reductions in NVF rates compared with the first 3 years.

Clinical fractures cause burdens to patients with osteoporosis (1–4). Nonvertebral fractures (NVFs) represent the majority of osteoporosis-related fractures (5–7) and are associated with morbidity, economic cost, and negative impact on health-related quality of life (7–10). Although antiresorptive therapies, including bisphosphonates and denosumab, reduce NVF incidence in women with postmenopausal osteoporosis, as shown in pivotal clinical trials (11–14), evidence of further reduction in NVF rates with prolonged therapy is limited. In long-term placebo-controlled extension studies with bisphosphonates, no further reduction in NVF incidence beyond the initial treatment in the original study has been demonstrated overall. Furthermore, in the 5-year alendronate Fracture Intervention Trial Long-Term Extension (FLEX) trial [Fracture Intervention Trial (FIT) Extension], 2-year risedronate Vertebral Efficacy With Risedronate Therapy Extension, and 3-year zoledronic acid Health Outcomes and Reduced Incidence With Zoledronic Acid Once Yearly Extension, subjects who received additional years of bisphosphonate treatment showed a similar NVF risk as those who switched to placebo in the extensions (15–17).

In the phase 3, randomized, placebo-controlled Fracture Reduction Evaluation of Denosumab in Osteoporosis Every 6 Months (FREEDOM) trial, postmenopausal women with osteoporosis were treated with denosumab or placebo for 3 years (13). Subjects who missed no more than one dose of the investigational product (IP) could enroll in the 7-year open-label FREEDOM Extension, during which all subjects were to receive denosumab (18). During FREEDOM, subjects randomly assigned to denosumab had reduced vertebral, nonvertebral, and hip fracture rates compared with placebo (13). With ≤10 years of denosumab treatment in the extension, continuous bone mineral density (BMD) gains at both the spine and hip were accompanied by a sustained and low incidence of NVFs (18). In these subjects, we previously reported that the rate of NVFs in year 4 was lower than the rate during the first 3 years in FREEDOM (19). After completion of the 10-year FREEDOM Extension trial, we now extend and complete this analysis for both the long-term and crossover groups, who received ≤10 and 7 years of continuous denosumab therapy, respectively, and balance the long-term fracture risk reduction against the rate of rare skeletal adverse events observed, specifically osteonecrosis of the jaw (ONJ) and atypical femoral fracture (AFF).

Participants and Methods

Study design

The FREEDOM trial (NCT00089791) and its extension (NCT00523341) have previously been described (13, 18). Briefly, in FREEDOM, subjects were randomly assigned 1:1 to receive denosumab 60 mg subcutaneously (SC) every 6 months (Q6M) or placebo for 3 years. At the end of 3 years, subjects who missed no more than one dose of IP could enroll in the open-label extension to receive denosumab 60 mg SC Q6M for an additional 7 years. Subjects randomly assigned to denosumab in FREEDOM could receive ≤10 years of denosumab treatment (long-term group); subjects randomly assigned to placebo in FREEDOM could receive ≤7 years of denosumab treatment (crossover group). FREEDOM and its Extension were conducted in accordance with the principles set out in the Declaration of Helsinki and were formally approved by the appropriate institutional review board, ethical review committee, or equivalent, at each study site.

Study population

This analysis included subjects who received ≥3 years of denosumab treatment. For the long-term group, this included subjects who enrolled in FREEDOM Extension, all of whom did not miss more than one dose of denosumab during the 3 years of FREEDOM, based on the extension enrollment criteria. For the crossover group, this included subjects who enrolled in FREEDOM Extension and did not miss more than one dose of denosumab during the first 3 years of FREEDOM Extension. Informed consent was obtained from the participants, and an institutional human research committee approved the investigations.

Outcome measures

NVFs were compared within groups, between time periods based on years of denosumab treatment, rather than years on study. NVF rates were calculated for the following durations of denosumab treatment:

• Years 1 to 3: represent the first 3 years of exposure to denosumab in either FREEDOM (long-term group) or the extension (crossover group).

• Years 4 to 7: represent the next 4 consecutive years of exposure to denosumab in the extension in both groups.

• Years 4 to 10: represent the entire duration of exposure to denosumab in the 7-year extension in the long-term group.

An NVF rate ratio was calculated between years 4 to 7 of denosumab treatment and years 1 to 3 of denosumab treatment in the combined (long-term and crossover) group as well as between years 4 to 10 of denosumab treatment and years 1 to 3 of denosumab treatment in the long-term group only (Fig. 1). NVF rate ratios were also calculated between years 4 to 7 and years 1 to 3 of denosumab treatment by femoral neck and total hip BMD T-score attained after 3 years of denosumab therapy: ≤–2.5, >–2.5 and <–1.0, ≥–1.0, ≤–2.0 and >–2.0, and ≤–1.5 and >–1.5. This approach of using within-group rate ratios has previously been used to compare long-term and short-term treatment in the absence of a placebo group (19). Subject incidence of individual NVF types (hip, distal femur, forearm, wrist, humerus, clavicle or rib, pelvis, and lower leg) was also calculated for the combined long-term and crossover groups during years 1 to 3 and years 4 to 7. Subject incidence of ONJ and AFF during the FREEDOM Extension was calculated. ONJ cases were reviewed by an independent external adjudication committee based on a prespecified list of Medical Dictionary for Regulatory Activities preferred terms; AFF cases were assessed by a panel at the central radiographic vendor (BioClinica, Newark, CA) based on criteria established by the American Society for Bone and Mineral Research 2010 Task Force for Atypical Subtrochanteric and Diaphyseal Femoral Fractures until 30 October 2013, after which the panel used the updated criteria by the American Society for Bone and Mineral Research 2013 Task Force for Atypical Subtrochanteric and Diaphyseal Femoral Fractures.

Statistical analyses

NVF rates per 100 subject-years (95% CI) and rate ratios (95% CI) based on generalized estimating equation Poisson regression were adjusted for age, total hip BMD T-score, weight, and history of NVF at the beginning of denosumab treatment (19). Treatment group was included in the model for the combined (long-term and crossover) group analysis only. Exposure-adjusted subject incidence per 100 subject-years and 95% CI of ONJ and AFF during FREEDOM Extension were calculated for subjects included in this analysis and all subjects who received one or more doses of denosumab during FREEDOM Extension.

Results

Subject characteristics

This analysis included 4074 subjects (2343 long-term and 1731 crossover). There were 476 additional subjects from the crossover group who enrolled in the extension but were excluded from this analysis because they discontinued or missed more than one dose of denosumab during the first 3 years of the extension. Overall, mean age at FREEDOM baseline was 71.9 and 71.3 years in the long-term and crossover groups, respectively, with 23.9% and 21.3% of subjects having prevalent vertebral fractures. BMD T-scores at FREEDOM baseline in the long-term and crossover groups were –2.83 and –2.85 at the lumbar spine and –1.85 and –1.83 at the total hip, respectively (Table 1). Subjects included in this analysis did not differ from those excluded from this analysis with respect to age, prevalent vertebral fractures, BMD T-score (both lumbar spine and total hip), and baseline bone turnover markers (20). Extension baseline characteristics in long-term and crossover subjects included in this analysis can also be found in Table 1.

NVF rates and rate ratios

For reference, the NVF rate per 100 subject-years (95% CI) in the crossover group after they received placebo for 3 years was 2.65 (2.23 to 3.15). In the combined denosumab group (i.e., n = 4074), the NVF rate was 2.15 (1.90 to 2.43) during years 1 to 3 and 1.53 (1.34 to 1.75) during years 4 to 7 of denosumab treatment [Fig. 2(a) and Table 2]. Hence, the fracture rate ratio (95% CI) was 0.72 (0.61 to 0.86; P < 0.001) for years 4 to 7 vs years 1 to 3 of denosumab treatment [Fig. 2(a)], consistent with previous data in the long-term group only during the same period (19).

In the long-term group only (n = 2343), the fracture rate was 1.98 (1.67 to 2.34) during years 1 to 3 and 1.44 (1.24 to 1.66) during years 4 to 10 with denosumab treatment [Fig. 2(b) and Table 2]. Hence, the NVF rate ratio was 0.74 (0.60 to 0.93; P = 0.008) for years 4 to 10 vs years 1 to 3 of denosumab treatment [Fig. 2(b)].

Pooled data from the combined long-term and crossover groups indicated a fracture risk reduction in individual NVF types (i.e., hip, distal femur, forearm, wrist, humerus, clavicle or rib, pelvis, and lower leg) from years 1 to 3 to years 4 to 7 of denosumab exposure (Fig. 3).

NVF risk by BMD subgroup after 3 years of denosumab treatment

NVF risk reduction was also evaluated by BMD attained after 3 years of denosumab treatment. Among subjects whose femoral neck BMD T-scores remained ≤–2.5 (i.e., osteoporotic), there was a trend (borderline significance) in NVF reduction in years 4 to 7 vs years 1 to 3 of denosumab treatment: rate ratio (95% CI) = 0.70 [0.49 to 1.01; P = 0.056; Fig. 4(a)]. The long-term reduction in NVF was significant in the larger group of subjects with femoral neck BMD T-scores between –2.5 and –1.0 (i.e., osteopenic): rate ratio = 0.73 (0.57 to 0.93; P = 0.009). In contrast, no further reduction was observed in subjects with femoral neck BMD T-scores ≥–1.0: rate ratio = 1.17 (0.62 to 2.18; P = 0.631). Because the subgroup of subjects with femoral neck BMD T-scores between –2.5 and –1.0 was the largest, we further evaluated subjects within that category with T-scores up to vs greater than –2.0 and –1.5. There was a significant reduction in NVFs in subjects who attained a T-score ≤–2.0 [rate ratio = 0.68 (0.52 to 0.88); P = 0.003] but not >–2.0, and ≤–1.5 [rate ratio = 0.71 (0.57 to 0.88); P = 0.001] but not >–1.5 [0.94 (0.62 to 1.44); P = 0.786; Fig. 4(a)]. Similar results were observed by total hip BMD T-scores attained after 3 years of denosumab treatment [Figs. 4(b) and 5(b)]. The P for interaction was of borderline significance only for total hip BMD ≤–1.5 or >–1.5.

Subject incidence of skeletal adverse events

Among subjects included in this analysis (n = 4074), the exposure-adjusted subject incidence per 100 subject-years (95% CI) was 0.05 (0.03 to 0.09) for ONJ, 0.01 (0.00 to 0.03) for AFF, and 0.06 (0.03 to 0.10) for ONJ and AFF combined during the FREEDOM Extension, corresponding to a total of 12 and 2 cases of ONJ and AFF, respectively (Table 3).

Discussion

Rates of NVFs in postmenopausal women with prolonged denosumab treatment during FREEDOM Extension (i.e., ≥3 and ≤10 years of denosumab treatment) were significantly lower than rates during FREEDOM (i.e., years 1 to 3 of denosumab treatment). By contrast, in extension trials with bisphosphonate therapy, no further reduction in NVF rates was observed compared with the benefits observed in the initial, pivotal trials. Subjects who received an additional 5 years of alendronate during the FLEX extension trial, beyond the initial 5 years during the placebo-controlled FIT trial, had statistically similar NVF incidences: 18.9% (FLEX) vs 11.9% (FIT) (11, 16). Similar trends were observed in the extensions vs the pivotal placebo-controlled trials for zoledronic acid (8.2% at the end of 6-year extension period vs 8.0% at the end of placebo-controlled trial) and risedronate (5.2% vs 5.2%, respectively) (12, 14, 15, 17).

In this aging postmenopausal population of the FREEDOM Extension trial, the NVF incidence would be expected to increase over time if subjects were left untreated. Data from population-based cohort studies indicate that age is a major risk factor for hip fractures (21), with an increasing 2-year incidence of hip, major, and all fractures observed in women with osteoporosis by increasing 5-year age group from 65 years to ≥85 years (22). In contrast, in the current analysis, long-term (i.e., more than 3 years) denosumab treatment resulted in significantly lower rates than the first 3 years of denosumab treatment. Although there was a significant reduction in NVF risk with 3 years of denosumab treatment compared with placebo in FREEDOM (13), the lower NVF risk becomes more evident at the fourth year (and beyond) of denosumab treatment (19). As previously reported, long-term denosumab treatment in the FREEDOM Extension trial for ≤10 years resulted in continuous gains in BMD at the total hip and femoral neck, from mean BMD T-scores of –1.9 and –2.2 at baseline to –1.2 and –1.5, respectively, by year 10 in the long-term group (18).

The findings in the current analysis are consistent with the notion that achieving higher BMD T-scores with denosumab treatment is associated with further reduction in NVF risk (23, 24). This may be because of temporal differences in the effect of denosumab therapy on trabecular and cortical bone, with rapid improvements in trabecular bone and slower improvements in cortical bone (25). With long-term treatment, continued increase in cortical bone volume and decrease in cortical porosity may further contribute to a decrease in NVF risk. Biochemical markers of bone turnover decreased with therapy and remained low through 10 years. Transiliac bone biopsies, which evaluated bone turnover at the tissue level, similarly did not show progressive changes in indices, with continued low turnover. Interestingly, there was evidence of increased matrix mineralization through 5 years of treatment, reaching levels near the physiologic threshold (1.2 g/cm³), which remained unchanged through 5 additional years of treatment (26). Based on the findings from the biopsy analysis, it is possible that there is a limit to the maximal amount of mineral that can be “packed” into bone, after which a steady state of bone mineralization ensues. However, with long-term denosumab treatment and, consequently, nearly complete inhibition of bone resorption in the endosteal envelope, residual modeling-based bone formation at loaded surfaces (such as the hip) has been suggested to contribute to the progressive increase in BMD and strength observed at this site (27, 28).

We observed continued low but not progressively declining NVF rates through years 4 to 10 of denosumab treatment, despite continued improvements in BMD. This finding may also be explained by the nonlinear relationship between BMD T-score and NVF risk, whereby there is a steep relationship between increasing T-score and decreasing NVF risk up to a T-score of about –1.5 (24). Beyond this BMD threshold, additional gains in BMD may not substantially decrease the risk of NVF (24). In the current analysis, we confirm and extend these observations by demonstrating that more NVF risk reduction was observed in years 4 to 7 compared with years 1 to 3 in subjects with femoral neck BMD T-scores ≤–1.5 but not above (Fig. 5). However, our findings may be limited by the number of subjects with femoral neck BMD T-scores ≤–2.5 (n = 776) and ≥–1.0 (n = 489). Nevertheless, although a substantial proportion of subjects in this analysis approached the T-score threshold during the first 3 years of denosumab treatment, with 1 additional year of treatment, most subjects reached or crossed the T-score threshold of ≥–1.5; beyond the fourth year of treatment, additional BMD gains over time may not contribute to further reduction in NVF rates in the population as a whole. This hypothesis is supported by the fact that the average baseline T-score at the total hip in FREEDOM was >–2.5, limiting the proportion of subjects who might be expected to show the greatest decrease in NVF risk along the steepest part of the BMD T-score–NVF curve. Indeed, previous analyses from both bisphosphonate and denosumab trials have shown that further fracture risk reduction with long-term therapy is observed mainly in women with total hip BMD T-scores <–1.5 after 3 years of therapy, with lesser or no effects in those with total hip BMD T-scores above that level (19, 29). As mentioned previously, this finding may reflect the threshold of maximal mineral content that can be incorporated into bone matrix through 5 years of denosumab treatment; nevertheless, low fracture risk is maintained through 10 years (18, 26).

Alternatively, further improvements in bone mass and structure with long-term treatment may be counteracted by an increasing incidence of frailty and falls in aging patients. Eventually, there may be a proportion of women with severe alterations in bone strength who would need even longer exposure to therapy to see a further decline in their fracture risk. Finally, the NVF incidence may also decrease with time, because it is the natural history of the disease: fractures cluster in time (30). However, the prevalent fractures in FREEDOM were of unknown date before subject enrollment, suggesting that this cluster phenomenon is not relevant in this follow-up.

Long-term antiresorptive therapy has been associated with the development of adverse skeletal events, specifically AFF and ONJ. Although there were no cases of AFF and ONJ during the FREEDOM study, cases have been observed with denosumab treatment in the FREEDOM Extension study. Among subjects included in this analysis, the exposure-adjusted incidence rate for AFF was 0.83 per 10,000 subject-years and for ONJ was 5.01 per 10,000 subject-years. Of the 12 cases of ONJ among subjects in this analysis, 11 were American Association of Oral and Maxillofacial Surgeons stage 2 or below; all subjects with complete follow-up healed despite most subjects receiving at least one dose of denosumab after ONJ onset, and one case was ongoing at the end of the FREEDOM Extension study (31, 32). Our findings suggest that the benefit of long-term denosumab treatment on NVFs far outweighs the risk of these skeletal adverse events. Health care practitioners need to determine individual benefit/risk to determine the pharmacologic needs of each patient, considering the risk of not treating and the risk of treatment-emergent adverse events.

Subjects maintaining long-term treatment with denosumab experience further reductions in vertebral fractures and NVFs (18). It is important to note that the discontinuation of denosumab treatment, which is a soluble and reversible inhibitor of RANK ligand, is associated with an increase in vertebral fracture risk, rapidly reaching the level observed in an untreated population—with a slightly higher risk of multiple vertebral fractures—particularly in subjects with a previous vertebral fracture (33). If denosumab is discontinued, a short course of bisphosphonate therapy is recommended to prevent reversible bone loss (34, 35). Taking into consideration the risk of vertebral fractures upon premature treatment discontinuation with further fracture risk reduction with long-term therapy, these observations emphasize the benefits and necessity of maintaining long-term treatment in high-risk patients.

This analysis was limited by the lack of a placebo group in FREEDOM Extension. There was also expected attrition over time in the long-term and crossover groups during the extension. However, a previous analysis did not suggest that loss of susceptible subjects affected NVF efficacy outcomes (20). There was also a limited number of observations in the BMD subgroup T-score analysis, particularly in the femoral neck BMD T-score ≤–2.5 and ≥–1.0 and total hip BMD T-score ≤–2.5 subgroups.

Conclusion

Compared with the first 3 years of denosumab treatment, a longer duration of denosumab therapy was associated with further decrease in NVF rate from year 4 on therapy, which was sustained through 10 years of treatment, whereas skeletal adverse events were rare. These data further support the long-term administration of denosumab, a potent antiresorptive therapy, in postmenopausal women with osteoporosis and a hip BMD T-score ≤–1.5 to improve bone strength and minimize the risk of fragility fractures.

Acknowledgments

Jessica Ma (Amgen Inc.), provided medical writing assistance. Qualified researchers may request data from Amgen clinical studies. Complete details are available at https://wwwext.amgen.com/science/clinical-trials/clinical-data-transparency-practices/.

Financial Support: This study was sponsored by Amgen Inc.

Clinical Trial Information:  ClinicalTrials.gov no. NCT00089791 (registered 16 August 2004).

Additional Information

Disclosure Summary: S.F. has served as a consultant for Amgen, UCB, AgNovos, Radius, and Labatec; has received research funding from MSD, Amgen, UCB, Oscare, Alexion, and AgNovos; is a member of the Board of Directors or advisory committee for IOF; and has served on the speakers’ bureau for Amgen. P.W.B., A.T.W., S.H., and R.B.W. are employees and stockholders of Amgen Inc. D.L.K. has served as a consultant for Pfizer, Eli Lilly, and Amgen; has received research funding from Amgen, Eli Lilly, and AstraZeneca; has received honoraria from Amgen, Eli Lilly, and Pfizer; and is a member of the Board of Directors for Doctors of BC. P.D.M. is an employee of Panorama Orthopedics; has served as a consultant for Amgen, Radius Health, and Sandoz; has received research funding from Amgen, Alexion, Merck, and Radius; and has served on an advisory committee for Amgen, Radius, Johnson and Johnson, Sandoz, and NBHA. C.R. has served as a consultant for Amgen and UCB; has received research funding from Ultragenyx; and has received honoraria for symposia from MSD and Lilly. E.M.L. has served as a consultant for Amgen, Radius, Alexion, Ultragenyx, and Sandoz; has received research funding from Amgen, PFEnex, and Mereo; is an association member of the National Osteoporosis Foundation, International Society for Clinical Densitometry, and Osteoporosis Foundation of New Mexico; and has served on the speakers’ bureau of Alexion and Radius. No authors received honoraria for the development of this post hoc analysis manuscript.

Data Availability: The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Abbreviations:

Abbreviations:
 
• AFF

  atypical femoral fracture

 
• BMD

  bone mineral density

 
• FIT

  Fracture Intervention Trial

 
• FLEX

  Fracture Intervention Trial Long-Term Extension

 
• FREEDOM

  Fracture Reduction Evaluation of Denosumab in Osteoporosis Every 6 Months

 
• IP

  investigational product

 
• N/A

  not applicable

 
• NVF

  nonvertebral fracture

 
• ONJ

  osteonecrosis of the jaw

 
• Q

  quartile

 
• Q6M

  every 6 months

 
• SC

  subcutaneous(ly)

References and Notes