CADTH Reimbursement Review

Romosozumab (Evenity)

Sponsor: Amgen Canada Inc.

Therapeutic area: Osteoporosis

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Clinical Review

Executive Summary

An overview of the submission details for the drug under review is provided in Table 1.

Introduction

Osteoporosis is a skeletal disorder characterized by low bone mass, compromised bone strength, and deterioration of bone quality, which results in an increased risk of fracture.^1,2 When osteoporosis-related fractures occur, patients experience pain, deformity, disability, loss of height, compromised health-related quality of life (HRQoL), and decreased life expectancy.^1,3 Osteoporosis affects 2 million Canadians, predominantly postmenopausal women.⁴

Different classes of drugs are indicated for prevention of osteoporosis-related fractures. Oral bisphosphonates are the most widely used anti-osteoporosis treatments in Canada.⁵ However, oral bisphosphonates are associated with important adverse events (AEs) such as gastrointestinal events^6,7 and are potentially associated with rare but serious adverse events (SAEs), including atrial fibrillation, osteonecrosis of the jaw (ONJ), and atypical femoral fractures. Alternative first-line medications include denosumab, zoledronic acid (an IV bisphosphonate), raloxifene, and teriparatide.

Romosozumab (Evenity) is a humanized monoclonal antibody that inhibits the action of sclerostin, a regulatory factor in bone metabolism. It increases bone formation and, to a lesser extent, decreases bone resorption.⁸ On June 17, 2019, romosozumab was approved by Health Canada for the treatment of osteoporosis in postmenopausal women at high risk for fracture, defined as a history of osteoporotic fracture, or with multiple risk factors for fracture.⁸ The reimbursement request by the sponsor for romosozumab is for the treatment of osteoporosis in postmenopausal women with a history of osteoporotic fracture and who are at very high risk for future fracture.

Romosozumab is available as a solution for subcutaneous injection in pre-filled syringes at a dose of 105 mg per 1.17 mL syringe. The recommended dosage of romosozumab is 210 mg administered once every month. Treatment duration of romosozumab is limited to 12 monthly doses. Once the patient has completed the 12-month romosozumab therapy, osteoporosis therapy with an antiresorptive agent should be considered.⁸

Stakeholder Perspectives

The information in this section is a summary of input provided by the patient and clinician groups who responded to CADTH’s call for input and from a clinical expert consulted by CADTH for the purpose of this review.

Patient Input

One patient group, Osteoporosis Canada, submitted input for this review. Osteoporosis Canada is a national organization dedicated to serving people who have or are at risk of developing osteoporosis.

The patient group indicated that the most important and feared consequence of osteoporosis is the risk of a fracture. From the patient perspective, the impact of a fracture can be substantial. Fractures in older populations are associated with acute and often chronic pain, changes in levels or loss of independence, decreased mobility, social isolation resulting in depression, or institutionalization as a result of a fragility fracture. Hip fractures in particular are associated with a significant decline in a patient’s ability to live independently and higher death rates resulting from complications. For younger seniors, this may result in time away from work, possibly with a financial impact.

Patients expressed that they would value a new treatment option that works differently from conventional options, particularly if it is easier to administer and comes with fewer side effects. Respondents to the Osteoporosis Canada survey reported the outcomes of most importance to them were preserving HRQoL, preventing fracture-related deaths, preventing admission to long-term care homes, preserving their ability to perform daily physical and social activities, preventing osteoporotic fractures, and avoiding serious side effects.

Clinician Input

Input From the Clinical Expert Consulted by CADTH

The clinical expert consulted by CADTH for this review indicated that, even though oral bisphosphonates are most commonly prescribed for the treatment of osteoporosis in postmenopausal women, they are associated with limitations such as inconvenient administration, gastrointestinal toxicities, and low absorption rates. Use of parenteral bisphosphonates is relatively limited because they are perceived as drugs to be prescribed by a specialist, and some patients are reluctant to receive IV infusions or subcutaneous injections. Treatment options for certain patient groups, such as those with renal insufficiency and renal failure, are limited.

In the expert’s opinion, romosozumab can be used as a first-line treatment for patients with the lowest bone mineral density (BMD) and greatest risk of fracture. It can also be used as a second-line treatment after patients fail on an antiresorptive agent and experience significant bone loss and fractures.

The expert stated that treatment response is assessed using change in BMD after treatment and suggested that measurement of BMD be conducted first at 12 months, when the patient transitions to antiresorptive therapy, and again 12 to 18 months later after a treatment change.

The expert indicated that romosozumab treatment should be discontinued if the patient experiences intolerable AEs. When a cardiovascular event occurs, the clinician should consider stopping the treatment.

Clinician Group Input

No input was provided by clinician groups.

Drug Program Input

In response to drug programs’ questions regarding the initiation of therapy in special subpopulations (e.g., patients with severe renal impairment), the clinical expert consulted by CADTH indicated that treatment of osteoporosis in patients with renal failure is complicated and should be managed by an osteoporosis specialist. Use of romosozumab is not always appropriate. For patients who are currently receiving active treatment for osteoporosis, clinicians would likely switch to romosozumab only if the patient was not responding satisfactorily to the current treatment and is still considered at high risk for fracture. Clinicians would consider use of romosozumab in patients who have received teriparatide therapy.

With respect to treatment continuation or renewal, the clinical expert indicated that, although continuous use of romosozumab beyond the recommended 12 months of treatment is not expected to be beneficial, intermittent treatment with romosozumab — for example, repeating a 12-month romosozumab course after a follow-on therapy of 1 year or more — may be considered. Should a dose of romosozumab be missed or treatment need to be temporarily stopped, the expert suggested that the patient receive a full 12 months of treatment.

The expert also stated that, although management and prescribing romosozumab would ideally involve clinicians with expertise in osteoporosis, in practice the drug may be prescribed by other health care providers and can be administered in an office.

Clinical Evidence

Pivotal Studies and Protocol-Selected Studies

Description of Studies

Two phase III studies (FRAME, N = 7,180; ARCH, N = 4,093) were included in the systematic review. The trials enrolled postmenopausal women (55 to 90 years of age) with osteoporosis.

The FRAME study was a double-blind, placebo-controlled randomized controlled trial (RCT) that assessed the efficacy and safety of romosozumab for the treatment of osteoporosis in postmenopausal women. Eligible patients were randomized to receive romosozumab 210 mg subcutaneously or placebo once a month for 12 months. After the 12-month double-blind treatment period, both groups received open-label denosumab 60 mg every 6 months for an additional 12 months. After the first 24-month treatment (12 months with romosozumab or placebo followed by 12 months with denosumab), patients entered a 12-month open-label extension period, during which they continued to receive denosumab 60 mg every 6 months. The co-primary efficacy end points were the incidences of new vertebral fractures at month 12 and month 24. Secondary efficacy end points included the incidence of various types of fractures and change from baseline in BMD T-scores. In this study, the mean age of the patients at baseline was 71 years and 41% of the patients had a historical fracture. The 10-year probability of a major osteoporotic fracture in this patient population at baseline was 13%, reflecting a moderate-risk population.

The ARCH study was a double-blind, active-controlled RCT that assessed the efficacy and safety of romosozumab for the treatment of osteoporosis in postmenopausal women with a high risk of fracture. Eligible patients were randomized to receive romosozumab 210 mg subcutaneously or oral alendronate 70 mg for 12 months. After the initial 12-month double-blind alendronate-controlled study period, both groups received open-label alendronate therapy 70 mg once a week for an additional 12 months. The primary efficacy end points in ARCH were the incidence of new vertebral fractures at month 24 and the incidence of clinical fractures (nonvertebral and clinical vertebral) during the primary analysis period, which refers to randomization to the time point at which clinical fractures were confirmed for at least 330 patients and all patients have had the opportunity to complete the month 24 study visit. Secondary efficacy end points included the incidence of various types of fractures and change from baseline in BMD T-scores. In this study, the mean age of the patients at baseline was 74 years and almost all patients had historical fracture. The 10-year probability of a major osteoporotic fracture in this patient population at baseline was 20%, reflecting a high-risk population.

Efficacy Results

Outcomes of fractures are relevant in clinical trials of osteoporosis. They were also identified by the clinicians and patient group as important outcomes. In FRAME, the risk of new vertebral fractures measured at the end of 1 year and 2 years of treatment were the primary efficacy end points. Treatment with romosozumab was associated with a 73% (95% confidence interval [CI], 53 to 84) reduction in the relative risk of a new vertebral fracture at month 12, and a 75% (95% CI, 60 to 84) relative risk reduction at month 24, compared to placebo. The between-group differences were statistically significant. According to the clinical expert consulted by CADTH, the benefit gained in the reduction of the risk of a new vertebral fracture is clinically meaningful. Results for a number of fracture-related outcomes (nonvertebral fractures, major nonvertebral fractures, new or worsening vertebral fractures, hip fractures, major osteoporotic fractures, and multiple new or worsening vertebral fractures) favoured romosozumab, and fewer patients in the romosozumab group developed these fractures compared to patients in the placebo group. Estimated differences in the risk of a fracture between romosozumab and placebo groups were statistically significant for clinical fractures, but failed to reach statistical significance for nonvertebral fractures. Firm conclusions for all other secondary end points cannot be made as the testing procedure was stopped after the failed test for nonvertebral fractures.

In ARCH, the risk of new vertebral fractures at month 24 and the risk of clinical fractures through the primary analysis study period (defined as the time from randomization to after clinical fractures had been confirmed in more than 330 patients and all patients completed the month 24 study visit) were co-primary efficacy end points. Treatment with 1 year of romosozumab followed by alendronate therapy for another year was associated with a statistically significantly reduced risk of new vertebral fractures (relative risk reduction = 50%; 95% CI, 34 to 62) through month 24, compared with treatment with alendronate for 2 years. Romosozumab was also associated with a statistically significantly reduced risk of clinical fractures (hazard ratio = 0.73; 95% CI, 0.61 to 0.88) through the primary analysis study period. The clinical expert indicated that the benefit gained in the reduction in the risk of new vertebral fractures or clinical fractures is clinically meaningful. Results for other fracture-related outcomes in this study (nonvertebral fractures, new vertebral fractures, clinical fractures, hip fractures, major nonvertebral fractures, major osteoporotic fractures, and all osteoporotic fractures) also favoured romosozumab over alendronate. The estimated difference in the rates of nonvertebral fractures was found to be statistically significant for patients on romosozumab compared to those on alendronate. Firm conclusions cannot be drawn for other fracture end points as no attempt was made to account for multiple comparisons.

One of the most important clinical outcomes considered by clinician and patient input was HRQoL, which was an exploratory outcome in both FRAME and ARCH. It was evaluated using the EuroQol 5-Dimensions 5-Levels questionnaire (EQ-5D-5L), which is a generic quality-of-life assessment tool, and the disease-specific Osteoporosis Assessment Questionnaire Short Version (OPAQ-SV). Results of the 2 studies did not show consistent or clinically meaningful changes between romosozumab and the comparators in any of these tools. A vertebral fracture is the most common clinical manifestation of osteoporosis, and approximately 2-thirds of these fractures are asymptomatic. This could explain why a deterioration or improvement in symptoms and quality of life may not be easily detected and a change in HRQoL may not be observed. Overall, the potential benefit of romosozumab on HRQoL remains unknown. The relationship between the gains from reduced fracture risk and improvement in patient’s HRQoL was unclear.

Change in BMD from baseline was measured in the lumbar spine, total hip, and femoral neck in both FRAME and ARCH. In ARCH, treatment with romosozumab was associated with a statistically significantly increased BMD from baseline at all 3 sites, compared to alendronate. Similar results were observed in the FRAME study when comparing romosozumab to placebo; however, BMD end points were not adjusted for multiple comparisons in this study. In general, the differences between romosozumab and placebo were numerically greater than those between romosozumab and alendronate. According to the clinical expert, the between-group differences in the ARCH study are clinically meaningful. These results were consistent with the change in incidence of fractures in the study population.

Harms Results

During the 24-month study period, the incidence of AEs was similar between romosozumab (month 12: 78%; month 24: 85%) and placebo (month 12: 80%; month 24: 86%) in FRAME, and between romosozumab (month 12: 76%; primary analysis period: 87%) and alendronate (month 12: 79%; primary analysis period: 89%). The incidence of SAEs was similar between romosozumab (month 12: 10%; month 24: 16%) and placebo (month 12: 9%; month 24: 15%) in FRAME, and between romosozumab (month 12: 13%; primary analysis period: 29%) and alendronate (month 12: 14%; primary analysis period: 30%) in ARCH. Treatment discontinuations due to AEs were similar between romosozumab (month 12: 3%; month 24: 3%) and placebo (month 12: 3%; month 24: 3%) in FRAME, and between romosozumab (month 12: 3%; primary analysis period: 7%) and alendronate (month 12: 3%; primary analysis period 24: 7%) in ARCH.

The incidence of fatal events was similar between romosozumab and placebo in the FRAME study, and between romosozumab and alendronate in the ARCH study, during the 2-year study period.

In terms of AEs of particular interest, the incidence of hypersensitivity and ONJ were similar between romosozumab and comparators at 12 and 24 months in both studies. The frequencies of cardiovascular events, particularly myocardial infarction and stroke, were higher with romosozumab versus alendronate at 12 and 24 months in the ARCH study.

Critical Appraisal

In the included studies, the completion rate at the end of 1 year of treatment was close to 90%, and more than 80% in FRAME and 77% in ARCH at the end of 2 years of treatment. The reasons for dropout were similar between treatment groups. The dropout rates, while consistent with those of other clinical trials of osteoporosis treatments, were still high and may affect the validity of the study results considering the proportion of data that needed to be imputed for analyses. In both studies, a last observation carried forward (LOCF) method was used to account for missing data for most efficacy end points. Sensitivity analyses that did not assume data were missing at random were conducted. These sensitivity analyses confirmed that the trial results were generally robust to the handling of missing data in the primary and secondary analyses.

In the FRAME and ARCH studies, a step-down procedure, with the primary and selected secondary outcome measures included, was used to control for multiplicity. Outcomes outside of the testing hierarchy, such as HRQoL (an exploratory outcome in both studies) and occurrence of cardiovascular events, need to be interpreted with consideration for the possibility of inflated type I error.

The clinical expert provided input on how to define high-risk patients. Risks of future fractures should be determined based on multiple factors, including patients’ demographic characteristics, history of fracture, sites of previous fracture, use of certain medications, Fracture Risk Assessment (FRAX) scores, BMD scores, and many others. For example, a hip fracture carries more weight than an ankle or wrist fracture when calculating future fracture risks. Subgroup analyses in the 2 studies examined the consistency of the primary analyses results across subgroup levels based on age, prevalent vertebral fracture status, history of fragility fracture, and baseline BMD T-scores, among others. More factors, such as effect of prior pharmacotherapy, patient compliance to prior treatment, and previous fractures that carry more weight (e.g., hip fractures or multiple fragility fractures), should also be considered.

There is limited direct evidence comparing romosozumab and relevant comparators, with only a comparison with alendronate in the ARCH study to inform the comparative efficacy and safety of romosozumab versus other osteoporosis medications.

Indirect Comparisons

Description of Studies

One sponsor-submitted indirect treatment comparison (ITC) was summarized and critically appraised.¹¹ The sponsor-submitted ITC aimed to evaluate the relative clinical efficacy of romosozumab and several treatments for osteoporosis, including denosumab, raloxifene, zoledronate, risedronate, and alendronate. Included studies enrolled postmenopausal women with primary osteoporosis or osteopenia who were at risk of developing fragility fractures. The 3 outcomes that were analyzed were sustaining vertebral, hip, and nonvertebral fragility fractures.

Efficacy Results

The sponsor-submitted ITC conducted a systematic review and used frequentist network meta-analysis (NMA) to evaluate the relative clinical efficacy of romosozumab to other treatments for osteoporosis.

The ITC reported that romosozumab was associated with a significant reduction in the risk of sustaining hip, nonvertebral, and vertebral fragility fractures compared to raloxifene, and a significant reduction in sustaining vertebral fractures compared to alendronate, risedronate, and raloxifene. There was no significant difference between romosozumab and any of alendronate, residronate, zoledronate, or denosumab in the risk of sustaining hip and nonvertebral fractures, and no significant difference was evident between romosozumab and either zoledronate or denosumab in the risk of sustaining vertebral fractures.

Harms Results

The sponsor-submitted ITC did not examine AEs, SAEs, withdrawal due to AEs, or deaths in the network of studies in the ITC.

Critical Appraisal

Critical appraisal points of the sponsor-submitted ITC involve the lack of reporting certain patient characteristics that would better address the certainty of the indirect evidence, including details of clinical heterogeneity in the included studies, effect modifiers and their influence on the results, construction of nodes in the ITC network, and details of assessments of essential NMA assumptions. The sponsor-submitted ITC could have used more sensitivity and subgroup analysis to satisfy the assumptions of transitivity and homogeneity, and a meta-regression that adjusted for effect modifiers could have influenced the results. Also, given that the sponsor-submitted ITC failed to provide a definition for “placebo,” double-counted hip fractures when analyzing hip and nonvertebral fracture outcomes, and did not distinguish between symptomatic and non-symptomatic vertebral fractures, the substantial uncertainty remaining undermines the internal and external validity of the ITC.

Other Relevant Evidence

Description of Studies

One long-term extension study, the FRAME Extension, provides longer-term evidence regarding the use of romosozumab to treat osteoporosis in postmenopausal women at high risk for fracture.¹² After the 24-month primary analysis of the FRAME study, eligible patients could enrol in the 12-month open-label extension period, during which they could continue to receive open-label denosumab 60 mg every 6 months; all patients in the extension phase of the FRAME trial were therefore treated with denosumab.

Efficacy Results

Through month 36, all fracture locations (new vertebral, clinical, nonvertebral, major nonvertebral, new or worsening vertebral, hip, major osteoporotic, and multiple new or worsening vertebral) showed an improved relative risk reduction in fractures among patients treated initially with romosozumab followed by denosumab (romosozumab/denosumab group), compared to patients treated initially with placebo followed by denosumab (placebo/denosumab group). The percent change in BMD at the lumbar spine, hip, and femoral neck from baseline to month 36 was also improved among patients in the romosozumab/denosumab group compared to patients in the placebo/denosumab group. The percent changes in bone turnover markers (BTMs) from baseline to month 36 were similar in both treatment groups when considering procollagen type 1 N-terminal propeptide (P1NP) and sclerostin. When considering serum C-telopeptide (sCTX), there was a −41 ng/L change in the placebo/denosumab group compared to a −14 ng/L change in the romosozumab/denosumab group; none of the BTM analyses (i.e., P1NP, sclerostin, or sCTX) indicated a difference between treatment groups (nominal P value > 0.05).

Harms Results

Reported AEs occurred in a similar proportion of both treatment groups (88% in the romosozumab/denosumab group and 89% in the placebo/denosumab group). For SAEs, the occurrence rate was 20% and 21% of patients in the romosozumab/denosumab and placebo/denosumab groups, respectively. Those AEs that led to treatment discontinuation were infrequently reported, occurring among 4% of patients in each treatment group. Few patients (2% in each treatment group) discontinued the study due to AEs. Fatal events were reported among 2% of patients in each treatment group.

Critical Appraisal

In the extension phase of the FRAME study, none of the analyses were adjusted for multiplicity, which may increase the likelihood of type I error. All results and analyses of the extension period of the FRAME trial should be considered supportive evidence. The FDA administered a warning related to cardiovascular AEs associated with the use romosozumab. None of the cardiovascular events that occurred during the extension phase suggested a greater risk among patients who were initially treated with romosozumab. The overall occurrence rate of cardiovascular AEs was generally low, at between less than 0.1% and 3.6% of patients. Analyses were conducted to determine the differences in the odds of cardiovascular events occurring between treatment groups; however, some of the sample sizes for specific cardiovascular AEs were low. The analyses designed to detect differences in the odds of cardiovascular events between the romosozumab/denosumab and placebo/denosumab groups were not powered or adjusted for multiplicity. Differences in the odds of cardiovascular AEs occurring in patients initially treated with romosozumab and those receiving placebo should be interpreted with caution.

The extension phase of the FRAME study provided insight into the long-term effects of initial treatment with romosozumab followed by treatment with an antiresorptive agent such as denosumab. These long-term data can help inform patients and physicians about the long-term effects (i.e., 36 months) of treatment with romosozumab followed by denosumab. Longer-term data (i.e., ≥ 10 years) may be more useful for patients and clinicians for determining long-term fracture risks after treatment with romosozumab. While the extension phase provided some indication of the fracture risk and change in BMD associated with romosozumab, it is not possible to know how patient outcomes will be affected in the future.

The differences in fracture incidence between patients initially treated with romosozumab and placebo was an end point of the extension phase of the FRAME study; results indicated a greater risk reduction in all analyzed sites among patients initially treated with romosozumab compared to patients initially treated with placebo. It is uncertain how initial treatment with romosozumab followed by long-term intervention with denosumab may compare to other treatments patients receive in clinical practice.

Conclusions

Two phase III double-blind RCTs, 1 placebo-controlled (FRAME) and 1 active-controlled (ARCH), provided evidence supporting the efficacy of romosozumab for the treatment of osteoporosis in postmenopausal women. The evidence from ARCH was considered more relevant to this review because it was the only study with an active comparator (alendronate) and more closely represents the target population for treatment with romosozumab (postmenopausal women with osteoporosis who are at high risk for fracture). Compared to alendronate, patients who were treated with monthly subcutaneous injections of 210 mg of romosozumab experienced benefits in reducing the risk of new fractures and increasing BMD. Changes in the incidence of new vertebral fractures and clinical fractures at month 12 and month 24 were considered statistically significant and likely clinically relevant. However, whether treatment with romosozumab is associated with any HRQoL benefit remains uncertain. The incidence rates of AEs, SAEs and treatment discontinuation due to AEs were similar between romosozumab and alendronate; however, a signal for a potential increased risk of cardiovascular-related AEs, particularly myocardial infarction and stroke, was noted with romosozumab compared with alendronate.

The results from 1 sponsor-submitted ITC suggest that romosozumab therapy may offer a beneficial effect in reducing the risk of sustaining nonvertebral fractures compared to some current treatments. Results of this indirect comparison are associated with a substantial risk of bias due to limitations, such as extensive heterogeneity, that have not been adequately accounted for.

The extension phase of the FRAME study, in which all participants received denosumab, suggest that the treatment effect from romosozumab in reducing the risk of fracture and increasing BMD was maintained. The frequency of AEs was generally similar between patients in the romosozumab/denosumab and placebo/denosumab groups. However, limitations of this extension study, such as the lack of a comparator group and a lack of patients from a high-risk population, contribute uncertainty to the results.

Introduction

Disease Background

Osteoporosis is a generalized skeletal disorder characterized by low bone mass, compromised bone strength, and deterioration of bone quality, which results in an increased risk of fracture.^1,2 Patients with osteoporosis usually have no clinical manifestations until there is a fracture. When osteoporosis-related fractures occur, patients suffer from pain, deformity, disability, loss of height, compromised HRQoL, and decreased life expectancy.^1,3 Vertebral compression fracture is the most common clinical manifestation of osteoporosis. Approximately 2-thirds of these fractures occur asymptomatically and are diagnosed as incidental findings on chest or abdominal radiographs. Symptomatic vertebral fractures, including height loss, are also observed. Hip fractures affect up to 15% of women by 80 years of age.¹³ The major source of morbidity and mortality from osteoporosis is attributed to hip fractures. They are not only associated with an increased mortality risk, but also influence long-term function and independence. In Canada, mortality among women was reported to be as high as 28% in the first year following a hip fracture.¹⁴ Fifty percent of women who suffer a hip fracture do not return to their previous functional state and become dependent on others for assistance with daily activities.¹⁵ Osteoporotic fractures can also occur at the distal forearm and proximal humerus.¹³ After the first fracture, the risk of subsequent fracture is increased and is highest in the first 1 to 2 years.¹⁶ While some researchers indicated that 10 years after a fracture, there is no added risk of future fracture compared to someone who has not fractured,¹⁶ other researchers reported that the relative risk of a subsequent fracture after the first fracture remained increased over 15 years.¹⁷

Osteoporosis affects 2 million Canadians, predominantly postmenopausal women due to decreased production of estrogen following menopause.⁴ Among Canadians aged 50 years and older, osteoporosis is associated with a substantial economic cost, accounting for $2.3 billion, or 1.3% of health care budgets, and the acute care cost of managing osteoporotic fractures was $1.2 billion, or 50% of total costs, based on data for fiscal year 2007 to 2008.¹⁸

According to the clinical expert consulted by CADTH, a diagnosis of osteoporosis should be considered for anyone who has had a fragility fracture. However, in most patients it is diagnosed before any fractures occurring through an assessment of BMD by dual-energy X-ray absorptiometry (DXA). A BMD scan provides a 2-dimensional record of mineral content in the region of interest (lumbar spine, total hip, and femoral neck), which, when combined with anthropometric parameters (height and weight), provides information about the strength and other mechanical properties of bone. The risk for most fractures is inversely proportional to BMD, with women typically having a lower baseline value than men. The WHO uses the “T-score” to define diagnostic thresholds for low bone mass and osteoporosis based on BMD measurements compared with those from a young adult reference population. Normal BMD is defined as a value within 1 standard deviation (SD) of the mean value in the reference population (values of −1 or higher); a T-score that is 1 to 2.5 SDs below the young adult mean (values of −1 to −2.5) is termed low bone mass, which was previously osteopenia; and a T-score that is 2.5 SDs or more below the young adult mean BMD (values of lower than −2.5) is defined as osteoporosis, provided that other causes of low BMD have been ruled out.¹ A clinical diagnosis of osteoporosis may be made in the presence of a fragility fracture (those occurring spontaneously or from minor trauma), particularly at the spine, hip, wrist, humerus, rib, and pelvis, without measurement of BMD.¹ In postmenopausal women, a clinical diagnosis of osteoporosis may also be made if there is a high risk for fracture as determined by fracture risk models. Organizations such as the Canadian Association of Radiologists recommend using an individual’s 10-year risk of fracture as the threshold for intervention.¹⁹ The 2010 Clinical Practice Guidelines for the Diagnosis and Management of Osteoporosis in Canada recommends the FRAX tool, a computer-based calculator that estimates the probability of major osteoporotic fracture,^20,21 or the Canadian Association of Radiologists and Osteoporosis Canada (CAROC) fracture risk assessment tool to evaluate an individual’s absolute 10-year fracture risk, accounting for risk factors such as age, history of fracture, and glucocorticoid use.²² In addition to femoral neck BMD, age, gender, fracture history, and glucocorticoid use, FRAX also takes into account other clinical risk factors, such as individual body mass index, smoking and alcohol use, and any comorbid conditions that may contribute to bone loss.¹ According to the clinical expert consulted by CADTH, the FRAX model is used globally and has country-specific databases to adjust fracture risks for ethnicity (country of origin), as opposed to the CAROC model, which has no capability of such adjustments.

Standards of Therapy

Lifestyle measures such as adequate calcium and vitamin D intake, exercise, smoking cessation, counselling on fall prevention, avoidance of heavy alcohol use, and avoidance of use of medications that increase bone loss (e.g., glucocorticoids), should be adopted universally to reduce bone loss in postmenopausal women.²³ For women with a high risk of fracture as determined by a combination of BMD and clinical risk factors (such as a prior spine or hip fracture, multiple fragility fractures, a T-score between −1 and −2.5, or with a 10-year risk of fracture of greater than 10%), potential pharmacotherapies include antiresorptive agents (e.g., bisphosphonates) that decrease bone turnover and anabolic agents (e.g., teriparatide or romosozumab) that stimulate bone formation.^15,22-25 According to the clinical expert, pharmacotherapy is required for patients with a greater than 20% risk of fracture over 10 years, while for those with a moderate risk of fracture (between 10% to 20%), pharmacotherapy is optional.

In Canada, different classes of drugs are indicated for osteoporosis, including 4 bisphosphonates (alendronate, risedronate, etidronate, and zoledronic acid), various forms of postmenopausal hormone therapy, a selective estrogen-receptor modulator (raloxifene), a biologic (denosumab), and a parathyroid hormone analogue (teriparatide). The 2010 Canadian clinical practice guidelines recommend the bisphosphonates alendronate and risedronate as first-line treatments, with high-quality evidence supporting benefits in the prevention of hip, nonvertebral, and vertebral fractures.^5,22 Among these, alendronate and risedronate accounted for the vast majority of all dispensed oral osteoporosis medications.^26,27 However, bisphosphonates are associated with reduced adherence due to gastrointestinal AEs^6,7 and rare SAEs, including atrial fibrillation, ONJ, and atypical femoral fractures.^28,29 Alternative first-line medications include denosumab, zoledronic acid, and raloxifene. Although teriparatide is also listed, the high cost of this anabolic therapy and the need for daily injections for 2 years restrict access to the medication. Raloxifene is only recommended for the prevention of vertebral fractures and therefore is not considered for use in most patients, except for those early in menopause, when the spine BMD begins to drop but hip bone density is usually preserved. Evidence supporting the efficacy of cyclic etidronate with calcium supplementation is weaker. For patients at high risk of fractures residing in long-term care, raloxifene and cyclic etidronate with calcium supplementation are not recommended.²²

Drug

Romosozumab is a humanized monoclonal antibody that inhibits the action of sclerostin, a regulatory factor in bone metabolism. It increases bone formation and, to a lesser extent, decreases bone resorption. Results of animal studies show that romosozumab stimulates new bone formation on trabecular and cortical bone surfaces by stimulating osteoblastic activity, resulting in increases in trabecular and cortical bone mass and improvements in bone structure and strength.⁸

On June 17, 2019, romosozumab was approved by Health Canada for the treatment of osteoporosis in postmenopausal women at high risk for fracture, defined as a history of osteoporotic fracture, or multiple risk factors for fracture.⁸ The reimbursement request by the sponsor for romosozumab is for the treatment of osteoporosis in postmenopausal women with a history of osteoporotic fracture and who are at very high risk for future fracture.

Romosozumab is available as a solution for subcutaneous injection in pre-filled syringes at a dose of 105 mg per 1.17 mL syringe. The recommended dosage of romosozumab is 210 mg administered once every month as 2 consecutive subcutaneous injections of 105 mg each using single-dose pre-filled syringes for 12 doses. Patients should be adequately supplemented with calcium and vitamin D. The treatment regimen for romosozumab is limited to 12 monthly doses. Once the patient has completed the 12-month romosozumab therapy, osteoporosis therapy with an antiresorptive agent should be considered. In the absence of a follow-on antiresorptive therapy, BMD gains typically trend toward pre-treatment levels following cessation of romosozumab.⁸

Romosozumab carries a serious warning regarding the potential risk of myocardial infarction, stroke, and cardiovascular death. Romosozumab should be discontinued in patients who experience a myocardial infarction or stroke. Table 3 provides details of the mechanism of action, indication, route and dose of administration, and adverse effects of romosozumab, alendronate, and denosumab.

Stakeholder Perspectives

Patient Group Input

This section was prepared by CADTH staff based on the input provided by patient groups.

About the Patient Group and Information Gathered

One patient group, Osteoporosis Canada, submitted the patient input for this review. Osteoporosis Canada is a national organization dedicated to serving people who have or are at risk of developing osteoporosis. The organization works to educate, empower, and support individuals and communities in the risk reduction of osteoporosis and related fractures. Its vision is a Canada without osteoporotic fractures and to this end it educates Canadians about osteoporosis, advocates for optimal osteoporosis care, and invests in osteoporosis research.

Osteoporosis Canada spoke directly with patients who have either completed 1 year of romosozumab or have been on it for several months, after obtaining permission from treating physicians who were on the Scientific Advisory Council of Osteoporosis Canada. From February 3 to 15, 2021, Osteoporosis Canada interviewed 4 patients with an average age of 70 in the Greater Toronto Area regarding their personal experience with romosozumab. Osteoporosis Canada recognized that the information it gathered on direct patient experience with this drug is limited. There have been no further fractures in patients interviewed at very high risk of fracture, who tolerated the drug well with no or minimal side effects and valued the opportunity to be treated on this new drug. In 2018, Osteoporosis Canada conducted a survey of patients living with osteoporosis to provide input in updating the 2010 Clinical Practice Guidelines for the Diagnosis and Management of Osteoporosis in Canada. In this survey, information on the issues and health outcomes that was deemed important to the patients was collected from more than 1,000 members of the Canadian Osteoporosis Patient Network. Some patient input from this survey was used to inform this submission.

Disease Experience

The patient group indicated that it recognizes that the most important and feared consequence of osteoporosis is the risk of fracture. Fractures can occur at several sites but are of significant concern when affecting the spine, hips, wrists, or shoulders. From the patient’s perspective, the impact of a fracture can be substantial; fractures in older populations are associated with acute and chronic pain, changes in levels or loss of independence, decreased mobility, social isolation resulting in depression, or institutionalization as a result of a fragile fracture. It is estimated that up to 40%, or even 50%, of elderly individuals who sustain a hip fracture will experience a significant decline in their ability to live independently. Additionally, 28% of women and 37% of men who suffer a hip fracture will die within a year from complications. Even a relatively simple wrist fracture will interfere with a person’s daily activities. For younger seniors, this may result in time away from work, possibly with a financial impact. In many cases, it results in increased care requirements from family members and/or other caregivers. According to Osteoporosis Canada, 1.5 million work-days are lost annually in Canada by fracture patients and 400,000 days are lost by caregivers.

Everyday activities can be severely compromised for those with fractures. If an individual has osteoporosis affecting the spine and is at high risk of fracture, an activity as ordinary as making a bed or the act of bending forward without caution can cause a fracture. A cough or sneeze can break a rib. Intimate relations are compromised. Activities such as golf or tennis, or picking up a grandchild, may have to be avoided because of the possibility of inducing a fracture. The knowledge that bones may break with minimal trauma results in significant fear of falling, which further limits independence and mobility.

Experiences With Treatments

Four patients who have experience with romosozumab were interviewed. The first patient was a 65-year-old female who was diagnosed with osteoporosis several years ago and was intolerant of bisphosphonate and denosumab. She received a 9-month treatment of romosozumab in the doctor’s office, administered by a nurse. This patient indicated that romosozumab worked well for her without serious side effects, except for some tolerable joint, muscle, hip, and groin pain. The second patient was a 77-year-old female who continued to be at very high risk of an osteoporotic fracture despite prior exposure to other, unspecified drugs. She completed 12 months of romosozumab therapy and reported no side effects, and no pain or irritation from the treatment. The third patient was a female in her mid-70s who was diagnosed with osteoporosis several years ago and suffered osteoporotic fractures but for whom other, unspecified drugs were contraindicated. She completed 6 months of romosozumab therapy. This patient was grateful for being offered this treatment as other osteoporosis medications had either failed or are contraindicated. She experienced an injection-site reaction on the day of and the day after the injection, but these were considered inconsequential. The fourth patient was a male in his early 60s who was diagnosed in May 2019. He was prescribed 12 months of romosozumab because of the severity of his osteoporosis. After the treatment, significant increases in BMD of both the hip and spine were observed. Minimal side effects, such as soreness at the injection-site and mild leg pain, were reported. None of these patients had problems with the administration of romosozumab.

Many common themes emerge from these interviews. The patients did not have an adequate response to or could not tolerate the conventional available drugs. They continued to fracture. These patients valued a new treatment option that works differently from conventional options, 1 that both builds new bone and prevents bone loss, is easier to administer, and has fewer side effects. The patients said that this new drug offers hope that new fractures will be prevented and that quality of life, including functionality and independence, will be maintained.

Patients expressed appreciation for the opportunity to choose an anabolic therapy option other than teriparatide. They appreciated that romosozumab involves 12 monthly injections rather than daily injections for 24 months (as is the case with teriparatide).

Improved Outcomes

Respondents to the Osteoporosis Canada survey reported that the outcomes that were of most importance to them were preserving quality of life (for example, improved mobility and independence), preventing fracture-related deaths, preventing admission to long-term care homes, preserving their ability to perform daily physical and social activities, preventing all fractures related to osteoporosis, and avoiding serious side effects.

Respondents expressed the desire for a choice of treatment regimens when unable to tolerate conventional medications, experiencing inadequate response to other medications, or having severe osteoporosis and continuing to fracture. The patients who were interviewed indicated that, compared to teriparatide, romosozumab may be more acceptable to many patients due to its less-frequent (monthly) injection schedule and shorter treatment course. Some patients received the injection from a health care professional, which was convenient; and some learned to self-inject.

Patients interviewed expressed appreciation for a novel therapy that both increases bone formation and bone resorption. The treatment under review holds great promise for patients for whom conventional therapies are contraindicated or not working.

Clinician Input

Input From the Clinical Expert Consulted by CADTH

All CADTH review teams include at least 1 clinical specialist with expertise in the diagnosis and management of the condition for which the drug is indicated. Clinical experts are a critical part of the review team and are involved in all phases of the review process (e.g., providing guidance on the development of the review protocol, assisting in the critical appraisal of clinical evidence, interpreting the clinical relevance of the results, and providing guidance on the potential place in therapy). The following input was provided by a clinical specialist with expertise in the diagnosis and management of osteoporosis.

Unmet Needs

Although oral bisphosphonates are most commonly prescribed for the treatment of osteoporosis, they are associated with inconvenient administration, gastrointestinal toxicities, and low absorption rates. Parenteral bisphosphonates offer greater benefit and fewer adverse effects, but are more costly than oral formulations. Primary care physicians are less likely to prescribe IV formulations, which are perceived as drugs to be prescribed by a specialist, and some patients are reluctant to receive IV infusions or subcutaneous injections. In addition, although patients with renal insufficiency and renal failure often suffer from substantial bone loss and fragility, current treatment options for this patient group are limited. Only denosumab is considered safe in this population at present. There is a substantial unmet need for medications that can preserve bone mass and reduce the risk of fracture but are safe in the renal-failure population.

Place in Therapy

Romosozumab would be used as a first-line treatment for patients with the lowest BMD and greatest risk of fracture. Romosozumab should be followed with an antiresorptive medication to maintain any gains in BMD and bone strength. Romosozumab would also be used as a second-line treatment after patients fail on an antiresorptive agent and have significant bone loss and fractures. Romosozumab can be used again if the subsequent antiresorptive use fails to maintain the bone mass and strength achieved through the initial romosozumab therapy.

According to the clinical expert, if romosozumab were affordable and covered by the public drug benefit programs, it would cause a significant shift in the current treatment paradigm in postmenopausal women with osteoporosis. Treatment with romosozumab increases bone formation and improves bone strength, and subsequent antiresorptive use maintains raised BMD levels, reducing the risk of fracture in the future.

Patient Population

Romosozumab shows treatment effects in males and females of all adult age groups that have been studied, and at all levels of initial BMD. The expert indicated that all patients with osteoporosis would benefit from the increased bone formation and strength associated with romosozumab, and those at the highest risk of fracture and/or with the lowest BMD would benefit most. These patients can be identified through a combination of fracture risk estimation using FRAX, documentation of prior fragility fractures (by radiographic imaging and patient’s medical history), and routine BMD testing.

Patients at low risk of fracture are least suitable for treatment with romosozumab.

Assessing Response to Treatment

In clinical practice, treatment response is assessed using change in BMD. Bone typically responds slowly to any treatment, such that an 18- to 24-month interval is more suitable when treating with a bisphosphonate or denosumab, but a significant improvement in BMD may be observed after a full year of treatment with romosozumab. The clinical expert suggests measuring BMD at 12 months, when the patient transitions to antiresorptive therapy, and again 12 to 18 months later after a treatment change.

It is also possible to measure the bone formation marker P1NP at baseline and a month later. An increase in serum P1NP at 1 month from baseline may predict a BMD response at 1 year. However, measurement of this marker is not widely available in Canada, and a change in P1NP did not correlate well with the net change in BMD at 1 year in a post hoc analysis of the STRUCTURE study.

Discontinuing Treatment

Romosozumab treatment should be discontinued if the adverse effects, such as an allergic response with hives or anaphylaxis, are intolerable. When a cardiovascular event occurs, the clinician should consider stopping the treatment if it is unknown whether the event is drug-related or not.

Prescribing Conditions

The involvement of a specialist is not required to diagnose, treat, or monitor patients who receive romosozumab. The drug can be administered in a primary care physician or nurse practitioner’s office, outpatient or specialty clinic, or by a community health nurse.

Clinician Group Input

No input was provided by clinician groups.

Drug Program Input

The drug programs provide input on each drug being reviewed through CADTH’s reimbursement review processes by identifying issues that may affect their ability to implement a recommendation. The implementation questions and corresponding responses from the clinical experts consulted by CADTH are summarized in Table 4.

Clinical Evidence

The clinical evidence included in the review of romosozumab is presented in 3 sections. The first section, the systematic review, includes pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those studies that were selected according to an a priori protocol. The second section includes indirect evidence from the sponsor and indirect evidence selected from the literature that met the selection criteria specified in the review. The third section includes sponsor-submitted long-term extension studies and additional relevant studies that were considered to address important gaps in the evidence included in the systematic review.

Systematic Review (Pivotal and Protocol-Selected Studies)

Objectives

To perform a systematic review of the beneficial and harmful effects of romosozumab (solution for injection, 105 mg per 1.17 mL syringe) for the treatment of osteoporosis in postmenopausal women at high risk for fracture, defined as a history of osteoporotic fracture, or multiple risk factors for fracture.

Methods

Studies selected for inclusion in the systematic review include pivotal studies provided in the sponsor’s submission to CADTH and Health Canada, as well as those meeting the selection criteria presented in Table 5. Outcomes included in the CADTH review protocol reflect those considered to be important by patients, clinicians, and drug plans.

The literature search for clinical studies was performed by an information specialist using a peer-reviewed search strategy according to the Peer Review of Electronic Search Strategies checklist (https://www.cadth.ca/resources/finding-evidence/press).³²

Published literature was identified by searching the following bibliographic databases: MEDLINE All (1946‒) via Ovid and Embase (1974‒) via Ovid. The search strategy comprised both controlled vocabulary, such as the National Library of Medicine’s MeSH (Medical Subject Headings), and keywords. The main search concept was Evenity (romosozumab). Clinical trials registries searched included the US National Institutes of Health’s clinicaltrials.gov, WHO’s International Clinical Trials Registry Platform search portal, Health Canada’s Clinical Trials Database, and the European Union Clinical Trials Register.

No filters were applied to limit the retrieval by study type. Retrieval was not limited by publication date or by language. Conference abstracts were excluded from the search results. Appendix 1 provides detailed search strategies.

The initial search was completed on March 10, 2020. Regular alerts updated the search until the meeting of the CADTH Canadian Drug Expert Committee on September 22, 2021.

Grey literature (literature that is not commercially published) was identified by searching relevant websites from the Grey Matters: A Practical Tool For Searching Health-Related Grey Literature checklist (https://www.cadth.ca/grey-matters).³³ Included in this search were the websites of regulatory agencies (US FDA and European Medicines Agency). Google was used to search for additional internet-based materials. Appendix 1 provides more information on the grey literature search strategy.

These searches were supplemented by reviewing bibliographies of key papers and through contacts with appropriate experts. In addition, the sponsor of the drug was contacted for information regarding unpublished studies.

Findings from the Literature

Two studies were identified from the literature for inclusion in the systematic review (Figure 1). The included studies are summarized in Table 6.

Figure 1: Flow Diagram for Inclusion and Exclusion of Studies

Description of Studies

FRAME (N = 7,180) was a phase III, multi-centre, double-blind, placebo-controlled RCT to evaluate the efficacy and safety of romosozumab for the treatment of osteoporosis in postmenopausal women. Eligible patients were randomized 1:1 to receive romosozumab 210 mg by subcutaneous injection or placebo once a month for 12 months. Randomization was stratified by age group (< 75 years versus ≥ 75 years) and prevalent vertebral fracture. Patients, investigators, and the sponsor remained blind to the initial treatment assignment. After the 12-month double-blind treatment period, both groups received open-label denosumab 60 mg every 6 months for an additional 12 months. During this period, patients remained blinded to the initial treatment allocation. After the first 24-month treatment (12 months with romosozumab or placebo followed by 12 months with denosumab), patients entered a 12-month open-label extension period, during which they continued to receive denosumab 60 mg every 6 months (Figure 2).

The co-primary efficacy end points were the incidence of new vertebral fracture at month 12 and at month 24.

Figure 2: FRAME Study Flow Chart

D and d = day; M = month; QM = every month; Q6M = every 6 months; SC = subcutaneously.

Note: The n = 3,300 per treatment group refers to planned enrolment.

Source: Clinical Study Report for FRAME.⁹

The ARCH study (N = 4,093) was a phase III, multi-centre, double-blind, active-controlled RCT to evaluate the efficacy and safety of romosozumab in postmenopausal women with a high risk of fracture. Eligible patients were randomized in a ratio of 1:1 to receive romosozumab subcutaneously or oral alendronate 70 mg for 12 months. Randomization was stratified by age (< 75 years versus ≥ 75 years). After the initial 12-month double-blind alendronate-controlled study period, both groups received open-label alendronate 70 mg once a week for an additional 12 months, while remaining blind to their initial treatment assignment of romosozumab or alendronate (Figure 3).

The primary efficacy end points in ARCH were the incidence of new vertebral fracture at month 24 and incidence of clinical fracture (nonvertebral fracture and clinical vertebral fracture) during the primary analysis study period, which refers to randomization to the time point at which clinical fractures were confirmed for at least 330 patients and all patients have had the opportunity to complete the month 24 study visit.

Figure 3: ARCH Study Flow Chart

ALN = alendronate; D = day; IU = international unit; M = month; PO = orally; QM = every month; QW = every week; SC = subcutaneously.

Note: Study completion constituted completion of primary analysis if superiority was proven for nonvertebral fractures or when at least 440 patients experienced a nonvertebral fracture.

Source: Clinical Study Report for ARCH.¹⁰

Populations

Inclusion and Exclusion Criteria

In FRAME, postmenopausal women aged 55 to 90 years at randomization (post-menopause was defined as no vaginal bleeding or spotting for 12 consecutive months before screening) were screened. Patients were eligible if they had a BMD T-score of −2.50 or lower at the total hip or femoral neck, as assessed by the central imaging vendor at the time of screening, based on DXA scans and using data for Caucasian women from the 1998 National Health and Nutritional Examination Survey. At least 2 vertebrae in the L1 through L4 region and at least 1 hip were evaluable by DXA, as assessed by the principal investigator, based on lateral spine X-rays. Patients were excluded if they had a BMD T-score of no more than −3.50 at the total hip or femoral neck at baseline, a hip fracture at any time, any severe or more than 2 moderate vertebral fractures, severe metabolic or bone diseases, and significant laboratory abnormalities.

In ARCH, postmenopausal women aged 55 to 90 years who met at least 1 of the following criteria were eligible: a BMD T-score of −2.5 or less at the total hip or femoral neck and either 1 or more moderate or severe vertebral fractures or 2 or more mild vertebral fractures; or a BMD T-score of −2.0 or less at the total hip or femoral neck and either 2 or more moderate or severe vertebral fractures or a fracture of the proximal femur sustained 3 to 24 months before randomization. The BMD T-scores at the time of screening were assessed by the central imaging vendor based on DXA scans and using data for Caucasian women from the 1998 National Health and Nutritional Examination Survey. Vertebral fractures at the time of screening were assessed by the central imaging vendor based on lateral spine X-rays. Histories of proximal femur fractures were assessed by the principal investigator based on a discharge summary, radiology report, or comparable documentation of type and date of fracture. Patients were excluded if the glomerular filtration rate was < 35 mL/min/1.73 m³ of body surface area.

In both studies, patients with recent use of certain osteoporosis drugs affecting bone metabolism were excluded. Details of inclusion and exclusion criteria in FRAME and ARCH are provided in Table 6.

Baseline Characteristics

In both studies, baseline characteristics were generally similar between treatment groups (Table 7). Patients in FRAME were approximately 3 years younger than those in ARCH. The mean age of the patients was 71 years in FRAME and 74 years in ARCH. More White patients were enrolled in ARCH (70% in ARCH versus 57% in FRAME). The ARCH study enrolled a patient population with higher risks for fracture: 96% of patients in ARCH with at least 1 prevalent vertebral fracture at baseline and 18% of those in FRAME had at least 1 prevalent vertebral fracture, BMD T-scores were lower in ARCH, and the 10-year probability of a major osteoporotic fracture calculated by FRAX was 20% in ARCH but 13% in FRAME.

Interventions

In FRAME, eligible patients were randomized to receive romosozumab subcutaneously or matched placebo for 12 months. After the initial 12-months double-blind placebo-controlled period, all patients received open-label denosumab 60 mg subcutaneously every 6 months, while remaining blinded to their initial treatment assignment of romosozumab or placebo. After the 24-month study period, eligible patients who entered the extension period received an additional 12 months of open-label treatment with denosumab, for a total treatment period of 36 months.

In ARCH, eligible patients were randomized to receive subcutaneous romosozumab or oral alendronate 70 mg for 12 months. During this period, patients also received matched placebo for either alendronate or romosozumab. After the initial 12-month double-blind alendronate-controlled study period, both groups received open-label alendronate 70 mg once a week for an additional 12 months, while remaining blinded to their initial treatment assignment of romosozumab or alendronate. No dosing adjustments for romosozumab/placebo, alendronate/placebo, or alendronate were permitted.

In both studies, patients received daily calcium and vitamin D supplementation from screening to end of study at a minimum of 500 mg to 1,000 mg of elemental calcium and 600 IU to 800 IU of vitamin D. In addition, patients with a serum 25 (hydroxide) vitamin D level of 20 ng/mL or greater and no more than 40 ng/mL at screening received an initial loading dose of 50,000 IU to 60,000 IU vitamin D after randomization. Patients with a serum 25 (hydroxide) vitamin D level of greater than 40 ng/mL at screening may have received the vitamin D loading dose at the principal investigator’s discretion.

Outcomes

A list of efficacy end points identified in the CADTH review protocol that were assessed in the clinical trials included in this review is provided in Table 8. These end points are summarized below. A detailed discussion and critical appraisal of the outcome measures are provided in Appendix 2.

New Fractures (Vertebral, Hip, and Others)

The incidence of various fracture types was measured in both FRAME and ARCH. The different types of fracture are defined below.

New vertebral fractures occurred when there was an increase of at least 1 grade from the previous grade of 0 in any vertebra from T4 to L4 on the Genant semiquantitative scoring method.

Clinical fractures included clinical vertebral and nonvertebral fractures (excluding skull, facial, mandible, cervical vertebrae, thoracic vertebrae, lumbar vertebrae, metacarpus, finger phalanges, and toe phalanges) that were associated with signs and/or symptoms indicative of a fracture.

A nonvertebral fracture was defined as a fracture present on radiographs or other diagnostic images such as CT or MRI confirming the fracture within 14 days of the reported fracture image date on an electronic case report form, and/or documented in a radiology report, surgical report, or discharge summary, excluding skull, facial, mandible, cervical vertebrae, thoracic vertebrae, lumbar vertebrae, metacarpus, finger phalanges, and toe phalanges. Fractures associated with high trauma severity or pathologic fractures were excluded.

The major nonvertebral fracture type was a subset of nonvertebral fractures, including pelvis, distal femur (i.e., femur excluding hip), proximal tibia (i.e., tibia excluding ankle), ribs, proximal humerus (i.e., humerus excluding elbow), forearm, and hip.

The hip fracture type was a subset of nonvertebral fractures, including femur neck, femur intertrochanter, and femur subtrochanter.

Major osteoporotic fractures included hip, forearm, or humerus fractures that were not associated with a pathologic fracture regardless of trauma severity, and clinical vertebral fractures.

Health-Related Quality of Life

An exploratory outcome measure in both studies, HRQoL was measured with the OPAQ-SV and EQ-5D-5L at specific time points throughout the 24-month study period in FRAME and ARCH.

Mortality

Death rates were reported in both studies. In both FRAME and ARCH, an attempt was made to classify each death as cardiovascular- or non-cardiovascular-related based on predefined criteria. In the event a death could not be classified as either cardiovascular- or non-cardiovascular-related, it was classified as undetermined. In ARCH, all undetermined deaths were presumed to be cardiovascular-related for completeness.

Change in Bone Mineral Density

The percent changes from baseline in BMD at the lumbar spine, total hip, and femoral neck were compared between treatment groups in both studies.

Change in Bone Turnover Markers

Increases in bone turnover have been associated with aging and diseases such as osteoporosis, contributing to increased risks of a fracture. Change in BMTs from baseline was assessed in substudies in both FRAME and ARCH. The evaluated bone formation markers included P1NP, bone-specific alkaline phosphatase (BSAP) and osteocalcin, and the evaluated bone resorption marker was sCTX.

Hospitalization Related to Fractures

Hospitalization related to fractures was not assessed in either study.

Statistical Analysis

FRAME Study

A sample size of 6,600 patients was anticipated to provide 91% power to detect a 40% decrease in the risk of nonvertebral fractures after romosozumab treatment for 12 months and 88% power to detect an overall 30% decrease in the risk of nonvertebral fractures in the 24-month study period using a log-rank test (alpha = 0.05).

A fixed-sequence testing procedure was used for multiplicity adjustment of the co-primary and secondary efficacy end points to maintain the overall significance level at 0.05. If the testing sequence was stopped at a particular step, the remaining end points in the testing sequence were not formally tested for statistical significance, and the corresponding P values were considered descriptive. The P values for the analyses of other secondary, exploratory, and substudy end points were nominal without adjusting for multiplicity. All P values were 2-sided. The following sequence was used in the testing procedure:

Step 1: Subject incidence of new vertebral fracture through month 12 and through month 24 (co-primary end points; nominal P value for each end point < 0.05)

Step 2: Subject incidence of clinical fracture through month 12

Step 3: Subject incidence of nonvertebral fracture through month 12 and through month 24, using the Hochberg procedure at a 0.05 level within the step

Step 4: Subject incidence of clinical fracture through month 24

Step 5: Subject incidence of major nonvertebral fracture through month 12 and through month 24, using the Hochberg procedure at a 0.05 level within the step

Step 6: Subject incidence of new or worsening vertebral fracture through month 12 and through month 24, using the Hochberg procedure at a 0.05 level within the step

Step 7: Subject incidence of hip fracture through month 12 and through month 24, using the Hochberg procedure at a 0.05 level within the step.

For the co-primary end points in this study, the primary analysis set for vertebral fractures was used for analyses of patient incidence of new vertebral fracture through month 12 and through month 24. The number and percentage of patients with a new vertebral fracture was summarized by treatment group. The significance of the treatment effect between romosozumab and placebo (month 12) and between romosozumab/denosumab and placebo/denosumab (month 24) was assessed using a logistic regression model with treatment as the main effect and age and prevalent vertebral fracture strata as covariates. The odds ratio, corresponding 95% CI, and P value of the score test were provided. Statistical significance for the co-primary end points was declared if the P values for testing the treatment effects through month 12 and month 24 were both less than 0.05. Point estimates of absolute risk reduction (difference in proportions, control – treatment), risk ratio (ratio of proportions, treatment over control), and the corresponding 95% CIs were also calculated using the Mantel-Haenszel method, adjusting for age and prevalent vertebral fracture strata.

For the secondary fracture end points, analyses of new or worsening vertebral fractures and multiple new or worsening vertebral fractures were performed on the primary analysis set using the logistic regression analysis described for the co-primary end points. The significance of the treatment effect between romosozumab and placebo (month 12) or romosozumab/denosumab and placebo/denosumab (month 24) was assessed using a stratified Cox proportional hazards model that controlled for age and prevalent vertebral fracture strata and used treatment as the independent variable. The estimated hazard ratio, corresponding 95% CI, and the P value of the score test were provided. The point estimate of the adjusted risk difference (difference in Kaplan–Meier estimates; control arm − treatment arm) and the corresponding 95% CIs were also provided using the inverse variance-weighted method. Adjusted P values based on the testing sequence for secondary fracture end points were provided in addition to the nominal P values from the statistical tests.

Subgroups analyses were performed for new vertebral fractures, clinical fractures, and nonvertebral fractures at months 12 and 24 using subgroup age categories (younger than 65 and 65 years or older, younger than 75 years and 75 years or older), as well as categories based on prevalent vertebral fracture status (yes, no), history of fragility fracture (yes, no), history of nonvertebral fracture at age ≥ 55 years (yes, no), race (White, non-White), geographic region (Western Europe; Australia and New Zealand; Central and Eastern Europe; Asia Pacific; North America; and Central or Latin America), baseline lumbar spine BMD T-score (≤ −3, > −3 and ≤ −2.5, > −2.5), baseline total hip or femoral neck BMD T-score (≤ −3 versus both total hip and femoral neck BMD T-score > −3), baseline BMI, and FRAX score for major osteoporotic fractures. The treatment-by-subgroup interaction was tested quantitatively. If the P value for the quantitative interaction was no greater than 0.05, qualitative interaction testing was performed.

Missing data for the vertebral fracture rate were imputed using the LOCF method.

For BMD end points, the primary efficacy analysis set for BMD (month 12 and month 24) was used for analyses of percent change from baseline in BMD. The treatment comparisons of the BMD in the lumbar spine, total hip, and femoral neck at months 12 and 24 were analyzed using an analysis of covariance (ANCOVA) model with LOCF imputation as the primary analysis. The ANCOVA model included treatment, age, prevalent vertebral fracture strata, and baseline value of the end point. Additional covariates of machine type and machine type-by-baseline value interactions were also included. The least squares mean of the treatment difference (treatment − control), the corresponding 95% CI, and P value were used to compare treatments at each time point.

ARCH Study

The total planned sample size of approximately 4,000 patients (2,000 patients per treatment group) was determined based on the clinical fracture and new vertebral fracture end points. The dropout rate was assumed to be 10% for the first year and 8% per year thereafter. Because the study design required all patients to complete the month 24 visit for the ascertainment of the new vertebral fracture end point, the minimum follow-up for an individual subject was at least 24 months. Additionally, all patients were to be followed until nonvertebral fracture events were confirmed in 440 patients for the final analysis, but this follow-up was not required as superiority of the nonvertebral fracture end point was achieved at the primary analysis.

The statistical significance for the primary end points (new vertebral fractures through month 24; clinical fractures through primary analysis) and selected key secondary end points (lumbar spine, total hip, and femoral neck BMD at month 24 and month 12; nonvertebral fractures through the primary analysis), were controlled using a sequential testing procedure to maintain the overall significance level for the study at 0.05, as determined by the Hochberg procedure. With this procedure, formal inferential testing was performed for a step only when statistical significance was declared for all end points tested in the previous steps. If the testing sequence stopped at a particular step, the remaining end points in the testing sequence were not formally tested for statistical significance and the corresponding P values were considered descriptive. The P values for the analyses of other secondary, exploratory, and substudy end points were nominal without adjusting for multiplicity. All P values were 2-sided.

The following testing sequence was used if both the co-primary end points were significant at 0.05 (2-sided):

1. percent change from baseline in BMD at lumbar spine at month 24

2. percent change from baseline in BMD at total hip at month 24

3. percent change from baseline in BMD at femoral neck at month 24

4. percent change from baseline in BMD at lumbar spine at month 12

5. percent change from baseline in BMD at total hip at month 12

6. percent change from baseline in BMD at femoral neck at month 12

7. if all preceding end points were significant, the nonvertebral fracture end point would be tested using a group sequential approach at the primary analysis and the final analysis based on a 1-sided test (alpha = 0.025).

For the analysis of primary end points, the number and percentage of patients with 1 or more new vertebral fractures through month 24 were summarized by randomized treatment group. The incidence of new vertebral fractures through month 24 between the randomized treatment groups was compared using a logistic regression model based on the primary efficacy analysis set for vertebral fractures. The adjusted odds ratio and the corresponding 95% CI were also derived from the model. The primary analytical model for comparing the incidence of clinical fractures at primary analysis between the randomized treatment groups used a stratified Cox proportional hazards model based on the full analysis set. The hazard ratio of romosozumab compared with that of alendronate and the CI were based on the model. The cumulative incidence of fractures was summarized using Kaplan–Meier estimates. To demonstrate the robustness of the results from the primary analytical models, additional supportive analyses were performed for the analysis subset and for new vertebral fractures through month 24 based on time to first new vertebral fracture using the stratified Cox proportional hazards model.

Missing data for the vertebral fracture rate were imputed using the LOCF method.

For secondary fracture end points, analyses of fracture end points (nonvertebral fractures, clinical fractures, clinical vertebral fractures, all fractures, major nonvertebral fractures, major osteoporotic fractures, and hip fractures) were performed on the full analysis set. For other fracture end points (new vertebral fractures, new or worsening vertebral fractures, and multiple new or worsening vertebral fractures), the incidences were compared using a logistic regression model. The number and percentage of patients with at least 1 of these vertebral fractures were summarized by treatment group at month 12 and month 24. The adjusted odds ratio and the corresponding 95% CI were also provided. The number and percent of patients with a new vertebral fracture between month 12 and month 24 were summarized by treatment group and a comparison performed using a logistic regression model.

Subgroups analyses were performed for the incidences of new vertebral fractures, clinical fractures, and nonvertebral fractures for the subgroup categories of age (< 75 and ≥ 75 years), presence or absence of severe vertebral fractures at baseline, number of prevalent vertebral fractures at baseline (0 to 1, 2, and 3 or more), race (White, non-White), geographic region (Western Europe, Australia and New Zealand; Central and Eastern Europe and the Middle East; Asia Pacific and South Africa; North America; Central and Latin America; and all regions excluding Central and Latin America), baseline lumbar spine BMD T-score (≤ −3, > −3 and ≤ −2.5, > −2.5), baseline total hip or femoral neck BMD T-score of −3 or lower versus both total hip and femoral neck BMD T-score of higher than −3, baseline BMI, FRAX score for major osteoporotic fracture, and history of nonvertebral fracture at age 55 years or older (yes, no). The odds ratios of romosozumab compared with alendronate, and their respective 95% CIs, were summarized for each category of subgroup. For the incidence of new vertebral fractures, the treatment-by-subgroup interaction was tested using logistic regression and included individual subgroup and treatment-by-subgroup interactions. For clinical and nonvertebral fractures, the treatment-by-subgroup interaction was tested using a stratified Cox proportional hazards model. If the P value for the interaction was less than 0.05, qualitative interaction testing was performed using the Gail and Simon test.

For BMD end points, the primary efficacy subset for BMD was used for the analyses of percent changes from baseline in lumbar spine, total hip, and femoral neck by DXA at months 12, 24, and 36. The treatment comparisons were analyzed using the ANCOVA model with LOCF imputation as the primary analysis. Descriptive summaries of observed DXA values at month 36 were provided without LOCF. Graphical displays to assess the model included response versus fitted values, response versus standardized residuals, and a normal probability plot of the standardized residuals. The sensitivity analyses included the repeated measures model, in which each body site is estimated using a separate repeated measures model. Absolute values and change from baseline for BMD were also summarized descriptively by time point and machine type.

Analysis Populations

FRAME Study

The primary analysis set for vertebral fractures included all randomized patients with a baseline and at least 1 post-baseline evaluation of vertebral fracture at or before the time point under consideration, including patients with missing baseline Genant semiquantitative scores whose first post-baseline spinal radiograph showed no fracture on the same vertebrae.

The full analysis set included all randomized patients.

The per-protocol analysis sets included patients who were compliant with the protocol (i.e., did not violate any of the predefined important inclusion or exclusion criteria for patient eligibility at baseline) and who met minimum investigational product-exposure requirements.

Primary efficacy analysis set for BMD and patient- and clinician-reported outcome end points included all randomized patients who had a baseline and at least 1 post-baseline evaluation at or before the time point under consideration in the study period.

The safety analysis set included all randomized patients who received at least 1 dose of the investigational product in the 12-month double-blind period. The incidence rates for the 24-month study period included all events that occurred in the double-blind period, and all events that occurred in the open-label period for those patients who received at least 1 dose of denosumab.

ARCH Study

The primary efficacy analysis set for vertebral fractures included all randomized patients who had a baseline and at least 1 post-baseline evaluation of vertebral fractures at or before the time point under consideration. Patients in this subset were analyzed according to their randomized treatment assignment, regardless of the treatment received. This analysis set was used as the primary analysis set for the following end points: new, new or worsening, and multiple new or worsening vertebral fractures.

The full analysis set included all randomized patients. Patients in the full analysis set were analyzed according to their randomized treatment assignments, regardless of treatment received. The full analysis set was used as the primary analysis set for the following end points: nonvertebral fractures, clinical fractures, clinical vertebral fractures, all fractures, major nonvertebral fractures, major osteoporotic fractures, and hip fractures.

The per-protocol analysis set was only used to analyze the following end points: incidence of clinical fractures, new vertebral fractures, and nonvertebral fractures through month 12 as sensitivity analyses. This subset includes patients who received an active investigational product and did not violate any important inclusion or exclusion criteria for patient eligibility at enrolment per important protocol deviation documents.

The safety analysis set included all randomized patients who received at least 1 active dose of the investigational product in the 12-month, double-blind, alendronate-controlled study period. This analysis set was used to analyze safety data for the double-blind study period, primary analysis period, and overall study period.

The primary analysis set for BMD, height, and patient- and clinician-reported outcome end points for the double-blind period and primary analysis period included all randomized patients who had a baseline and at least 1 post-baseline evaluation at or before the time point under consideration.

Results

Patient Disposition

In FRAME, a total of 7,180 patients were randomized into the study, with 3,589 patients randomized to the romosozumab group and 3,591 patients randomized to the placebo group. Eighty-nine percent of patients completed the double-blind period (romosozumab, 88.7%; placebo, 89.3%) and 83.9% of patients completed the 24-month study period (romosozumab/denosumab: 83.4%; placebo/denosumab: 84.4%). Throughout the study, withdrawn consent was the most frequently reported reason for withdrawal from study in both treatment groups (romosozumab/denosumab: 9.6%; placebo/denosumab: 8.7% in the 24-month study period).

In ARCH, a total of 4,093 patients were randomized to the romosozumab 210 mg monthly (2,046 patients) and alendronate 70 mg weekly (2,047 patients) treatment groups. A total of 89.3% of enrolled patients completed the 12-month double-blind period (romosozumab: 89.5%; alendronate: 89.1%) and 77.0% of them completed the primary analysis period (romosozumab/alendronate: 76.9%; alendronate/alendronate: 77.0%). Throughout the study, withdrawn consent and death were the most frequently reported reasons for study discontinuation in both treatment groups (6.3% and 1.3%, respectively, in the romosozumab group, and 6.8% and 1.2%, respectively, in the alendronate group during the 12-month double-blind period; 12.9% and 4.6%, respectively, in the romosozumab/alendronate group and 12.3% and 4.7%, respectively, in the alendronate/alendronate group during the primary analysis period).

Details of patient disposition for FRAME and ARCH are provided in Table 9.

Exposure to Study Treatments

FRAME

The median number of doses received during the double-blind period was 12.0 (range = 1 to 12) in both groups; 75.3% of the romosozumab group and 75.1% of the placebo group received 12 doses. The mean cumulative romosozumab exposure was 2,260.3 mg (SD = 612.8).

ARCH

During the 12-month double-blind period, the median number of double-blind period subcutaneous injections of romosozumab or matching placebo received was 12.0 (range = 1 to 12) in both groups. In the romosozumab group, 74.8% of patients received 12 doses of subcutaneous romosozumab and, in the alendronate treatment group, 75.2% received subcutaneous injections of matching placebo. In the romosozumab treatment group, the median cumulative romosozumab exposure was 2,520.0 mg (range = 70 mg to 2,520 mg). In the alendronate group, the median cumulative alendronate exposure during the 12-month double-blind period was 3,640 mg (range = 70 mg to 4,410 mg).

The distribution for the duration of follow-up (in weeks or months) per treatment group was not reported for either FRAME or ARCH.

Efficacy

Only those efficacy outcomes and analyses of subgroups identified in the review protocol are reported below. Appendix 3 provides detailed efficacy data.

New Fractures (Vertebral, Hip, and Others)

In the FRAME study, romosozumab therapy reduced the risk of new vertebral fractures compared with placebo through month 12 (odds ratio = 0.27; 95% CI, 0.15 to 0.47; P < 0.001), with a relative risk reduction of 73% (risk ratio = 0.27; 95% CI, 0.16 to 0.47). Over the 24-month study period, treatment with romosozumab for 12 months followed by denosumab for 12 months reduced the risk of new vertebral fractures compared with treatment with placebo for 12 months followed by denosumab for 12 months (odds ratio = 0.24; 95% CI, 0.15 to 0.39; P < 0.001), with a relative risk reduction of 75% (risk ratio = 0.25; 95% CI, 0.16 to 0.40) (Table 11).

Subgroup analyses based on lumbar spine BMD T-scores at baseline (≤ −3 versus > −3 and ≤ −2.5 versus > −2.5), total hip and femoral neck BMD T-scores at baseline (≤ −3 versus > −3), historical fragility fracture (yes versus no) and FRAX scores at baseline (≤ 8.21 versus > 8.21 and ≤ 14.27 versus > 14.27) were performed for the primary efficacy end points. In general, the subgroup analyses results are consistent with the primary analysis results; patients treated with romosozumab had a numerically lower risk of new vertebral fractures at month 12 and 24 compared to the patients treated with placebo (Table 35).

The co-primary and selected secondary efficacy end points were tested in a fixed-sequence approach to control for multiplicity. Romosozumab reduced the risk of clinical fractures (nonvertebral and clinical [i.e., symptomatic] vertebral fractures) by 36% (95% CI, 11 to 54) compared with placebo through month 12 (P = 0.008). Clinical fractures through 24 months were not a part of the testing structure. The next step in the testing sequence was the simultaneous comparison of the fracture incidence for the secondary end points of nonvertebral fractures at months 12 and 24. The 25% (95% CI, −5 to 47) relative risk reduction with romosozumab for nonvertebral fractures through month 12 was not statistically significant (P = 0.096) and the P value for nonvertebral fractures through month 24 did not meet a statistical significance threshold after applying the Hochberg procedure to adjust for multiple comparisons (P = 0.057) (Table 12). No further statistical conclusions can be made for other secondary end points as the testing sequence was halted or the secondary end points were not adjusted for multiple comparisons.

In ARCH, through month 24, treatment with romosozumab 210 mg monthly for 12 months followed by alendronate 70 mg weekly for 12 months reduced the risk of new vertebral fractures by 50% compared with alendronate alone (relative risk = 0.50; 95% CI, 0.38 to 0.66; odds ratio = 0.48; 95% CI, 0.36 to 0.64; adjusted P < 0.001). Through the primary analysis (after clinical fractures had been confirmed in at least 330 patients), treatment with romosozumab 210 mg monthly for 12 months followed by alendronate 70 mg weekly reduced the hazard ratio of clinical fractures (nonvertebral fracture and clinical vertebral fracture) by 27% compared with alendronate alone (hazard ratio = 0.73; 95% CI, 0.61 to 0.88; adjusted P < 0.001), based on 464 patients with clinical fractures.

Subgroup analyses based on lumbar spine BMD T-scores at baseline (≤ −3 versus > −3 and ≤ −2.5 versus > −2.5), total hip and femoral neck BMD T-score at baseline (≤ −3 versus > −3) and the FRAX score at baseline (≤ 14.22 versus > 14.22 and ≤ 22.47 versus > 22.47) were performed for the primary efficacy end points. In general, the subgroup analyses results are consistent with the primary analysis results in that patients treated with romosozumab had a numerically lower risk of a new vertebral fracture and clinical fracture at month 24, compared to the patients treated with alendronate (Table 36).

Nonvertebral fractures were a secondary efficacy outcome and accounted for multiple comparisons via a sequential testing structure. In the primary analysis, treatment with romosozumab 210 mg monthly for 12 months followed by alendronate 70 mg weekly reduced the hazard ratio of nonvertebral fractures by 19% (hazard ratio = 0.81; 95% CI, 0.66 to 0.99; P = 0.019, 1-sided nominal P value) based on 395 patients with nonvertebral fractures. Under the group sequential design, with an information fraction of 90% (395 of 440), the significance level at primary analysis was determined to be 0.0233 (1-sided). This result is significant as 0.019 is less than 0.0223; consistent with this, the multiplicity-adjusted 2-sided P value (adjusting for the fixed-sequence and the group sequential testing) was significant at 0.040.

No additional fracture end points were included in the testing structure for ARCH. Results for all other fracture end points should be considered as supportive evidence. During the primary analysis period, treatment with romosozumab for 12 months followed by alendronate was associated with a numerically smaller risk of hip fractures, major nonvertebral fractures, major osteoporotic fractures, and all osteoporotic fractures.

At month 12, treatment with romosozumab for 12 months followed by alendronate was associated with a numerically smaller risk of new vertebral fractures, clinical fractures, all osteoporotic fractures, new or worsening vertebral fractures, and major nonvertebral fractures.

At month 24, treatment with romosozumab for 12 months followed by alendronate was associated with a numerically smaller risk of clinical fractures, clinical vertebral fractures, new or worsening vertebral fractures, and multiple new or worsening vertebral fractures (Table 15).

Health-Related Quality of Life

In FRAME, no consistent or clinically meaningful differences in patient- or clinician-reported outcome end points (OPAQ-SV, EQ-5D-5L, Limited Activity Days survey, or Brief Pain Inventory worst pain score) were identified between the romosozumab and placebo groups during the 12-month double-blind period, or between the romosozumab/denosumab and placebo/denosumab groups during the 24-month study period.

Similarly in ARCH, no consistent or clinically meaningful differences in patient- or clinician-reported outcome end points (OPAQ-SV, EQ-5D-5L, Limited Activity Days, or Brief Pain Inventory worst pain scores) were identified between the romosozumab and alendronate groups during the 12-month double-blind period or between romosozumab/alendronate and alendronate/alendronate groups during the primary analysis period.

Mortality

The numbers of patients who died during the study period in the FRAME and ARCH studies are presented in Table 16. The section on harms provides detailed mortality data from the 2 studies.

Change in Bone Mineral Density

FRAME Study

Treatment with romosozumab significantly increased BMD at the lumbar spine (between-group difference = 12.7%; 95% CI, 12.4 to 12.9), total hip (between-group difference = 5.8%; 95% CI, 5.6 to 6.0), and femoral neck (between-group difference = 5.2%; 95% CI, 4.9 to 5.4) compared with placebo at month 12. In addition, romosozumab followed by denosumab was associated with increases in BMD at the lumbar spine, total hip, and femoral neck at month 24. Mean differences in percent change from baseline to month 24 in BMD were greater for patients in the romosozumab/denosumab group compared with the placebo/denosumab group (P < 0.001 at each site). The BMD end points were not included in the sequential testing procedure for multiplicity adjustment.

ARCH Study

The BMD end points were adjusted for multiple comparisons at month 24 and month 12 and included in the testing sequence. At month 12, romosozumab significantly increased BMD compared with alendronate at the lumbar spine, total hip, and femoral neck, with least squares mean differences of 8.7% (95% CI, 8.3 to 9.1), 3.3% (95% CI, 3.0 to 3.6), and 3.2% (95% CI, 2.9 to 3.5), respectively (adjusted P < 0.001 for all 3 sites). At month 24, romosozumab for 12 months followed by alendronate for 12 months, significantly increased BMD compared with alendronate alone at the lumbar spine, total hip, and femoral neck, with mean differences of 8.1% (95% CI, 7.6 to 8.6), 3.8% (95% CI, 3.4 to 4.1), and 3.8% (95% CI, 3.4 to 4.1), respectively (adjusted P < 0.001 for all 3 sites).

Change in Bone Turnover Markers

FRAME Study

A total of 130 randomized patients were enrolled in the BTM and biomarker substudy, representing all regions except Asia Pacific; 129 patients were included in the BTM and biomarker substudy efficacy analysis set (64 patients in the romosozumab/denosumab group and 65 patients in the placebo/denosumab group). At least 90% of patients in each treatment group completed the 24-month study period. No adjustment for multiple comparisons was made for the analysis of this substudy.

For the bone formation markers, the median percent change from baseline in P1NP was lower in the romosozumab group compared with the placebo group at the end of double-blind treatment at month 12 (romosozumab group: −17%; placebo group: −3%; P = 0.006). Similar to P1NP, serum concentrations of BSAP (romosozumab group: −13%; placebo group; −10%; P = 0.34) and osteocalcin (romosozumab group: −19%; placebo group: −12%; P = 0.48) peaked early in romosozumab-treated patients and declined thereafter, returning to baseline by month 12.

For the bone resorption markers during double-blind treatment, the median percent change from baseline in sCTX was lower in the romosozumab group than in the placebo group at day 14, month 1, month 3 plus 14 days, month 6 plus 14 days, month 9, and month 12 (romosozumab group: −53%; placebo group: −29%; P < 0.001). Changes in sCTX were assessed through month 24 after the end of double-blind treatment. In both treatment groups, denosumab caused similar reductions in sCTX through month 24.

ARCH Study

A total of 298 randomized patients enrolled in the imaging and BTM substudy, including 157 patients in the romosozumab/alendronate group and 141 patients in the alendronate/alendronate group. Patients from all regions participated in the substudy. A total of 136 patients (86.6%) in the romosozumab/alendronate group and 125 patients (88.7%) in the alendronate/alendronate group completed the double-blind study period. No adjustment for multiple comparisons was made for the analysis of this substudy.

For the bone formation markers, the median percent change from baseline in P1NP was greater in the romosozumab group than in the alendronate group at all time points evaluated during double-blind treatment: at month 12, the median percent change was −30% in the romosozumab group compared to −63% in the alendronate group (P < 0.001). After month 12, when all patients received open-label alendronate, the median P1NP value in the romosozumab/alendronate group decreased below baseline at month 15 and remained consistently reduced through month 36. Similar to P1NP, serum concentrations of BSAP (−7% at month 12 in the romosozumab group compared to −35% in the alendronate group) and osteocalcin (−18% at month 12 in the romosozumab group compared to −50% in the alendronate group) peaked at month 1 in romosozumab-treated patients and declined through month 12 (the last assessment time point for these BTMs), returning to baseline by month 9.

Levels of the bone resorption marker sCTX showed a decrease in the romosozumab group at month 1 (the time point of first assessment) and remained reduced but above the levels in the alendronate group through month 12 (romosozumab group: −38%; alendronate group: −70%; P < 0.001). After month 12, when all patients received open-label alendronate, the median sCTX value in the romosozumab/alendronate group decreased at months 15, 18, and 24 to levels similar to those in the alendronate/alendronate group. At month 36, the median sCTX value in the romosozumab/alendronate group increased but remained below baseline (nominal P = 0.042 versus the alendronate/alendronate group).

Harms

Only those harms identified in the review protocol are reported below.

Table 18 provides detailed harms data.

Adverse Events

FRAME Study

The incidence of AEs was similar in patients who received romosozumab (78.4%) and patients who received placebo (79.7%). Frequently reported AEs in the romosozumab group and the placebo group during the 12-month double-blind period were arthralgia (13.0% and 12.0%, respectively), nasopharyngitis (12.8% and 12.2%, respectively), and back pain (10.5% and 10.6%, respectively). At month 24, commonly reported in the respective 2 treatment groups were arthralgia (16.3% and 15.8%), nasopharyngitis (15.6% and 15.3%), back pain (12.9% and 14.4%), fall (10.8% and 12.8%), and pain in extremity (10.1% and 10.2%).

ARCH Study

The incidence of AEs was similar in patients who received romosozumab (75.7%) and patients who received alendronate (78.6%). Frequently reported AEs during 12-month double-blind period were nasopharyngitis (10.4% and 10.8%) and back pain (9.1% and 11.3%) in the romosozumab group and the alendronate group, respectively. In the primary analysis study period, commonly reported AEs were nasopharyngitis (17.8% and 18.5%), back pain (16.1% and 19.5%), arthralgia (16.1% and 18.0%), fall (13.8% and 16.7%), hypertension (11.1% and 11.2%), pain in extremity (11.0%, and 12.2%), upper respiratory tract infection (9.8% and 10.9%), osteoarthritis (9.6% and 10.1%), and urinary tract infection (9.1% and 12.1%), respectively, in the 2 treatment groups.

Serious Adverse Events

FRAME Study

During the 12-month double-blind period, SAEs were reported for 344 patients (9.6%) who received romosozumab and 312 patients (8.7%) who received placebo. The most frequently reported SAEs (> 0.2% of patients in either treatment group) were pneumonia (romosozumab: 0.5%; placebo: 0.3%), chronic obstructive pulmonary disease (romosozumab: 0.2%; placebo: 0.4%), and osteoarthritis (romosozumab: 0.2%; placebo: 0.4%).

During the 24-month study period, SAEs were reported for 565 patients (15.8%) in the romosozumab/denosumab group and 540 patients (15.1%) in the placebo/denosumab group. The most frequently reported SAEs (≥ 0.4% of patients in either treatment group) were pneumonia (0.9% for the romosozumab/denosumab group; 0.6% for the placebo/denosumab group), hypertension (romosozumab/denosumab: 0.4%; placebo/denosumab: 0.3%), chronic obstructive pulmonary disease (romosozumab/denosumab; 0.4%; placebo/denosumab: 0.6%), and atrial fibrillation (0.4% each).

ARCH Study

During the 12-month double-blind period, SAEs were reported for 262 patients (12.8%) who received romosozumab and 278 (13.8%) who received alendronate. The most frequently reported SAEs (> 0.5% of patients in either treatment group) were pneumonia (romosozumab: 0.8%; alendronate: 0.8%), femur fracture (romosozumab: 0.5%; alendronate 0.6%), radius fracture (romosozumab: 0.4%; alendronate: 0.6%), chronic obstructive pulmonary disease (romosozumab: 0.2%; alendronate: 0.5%), and femoral neck fracture (romosozumab: 0.2%; alendronate:0.6%).

During the primary analysis period, SAEs were reported for 586 patients (28.7%) in the romosozumab/alendronate group and 605 (30.0%) in the alendronate/alendronate group. The most frequently reported SAEs (> 1.0% of patients) in the romosozumab/alendronate group were pneumonia (2.3% versus 2.1% in the alendronate/alendronate group), femur fracture (1.9% versus 2.3%), and chronic obstructive pulmonary disease (1.1% versus 1.4%).

Withdrawal Due to Adverse Events

FRAME Study

At month 12, AEs led to withdrawal from the study drug for 103 patients (2.9%) who received romosozumab and 94 patients (2.6%) who received placebo during the 12-month double-blind period. The AEs that led to discontinuation of the study drug for at least 0.2% of patients in either treatment group were pain in extremity (romosozumab: 0.2%; placebo: < 0.1%) and musculoskeletal pain (romosozumab: < 0.1%; placebo: 0.2%).

During the 24-month study period, AEs led to withdrawal from the study drug for 122 patients (3.4%) who received romosozumab/denosumab and 110 patients (3.1%) who received placebo/denosumab. The AEs that led to discontinuation of the study drug for ≥ 0.2% of patients in either treatment group were pain in extremity (romosozumab/denosumab: 0.3%; placebo/denosumab: < 0.1%), arthralgia (romosozumab/denosumab: 0.2%; placebo/denosumab: 0.1%), and musculoskeletal pain (romosozumab/denosumab: < 0.1%; placebo/denosumab: 0.2%).

ARCH Study

At month 12, AEs led to discontinuation of the study drug for 70 patients (3.4%) in the romosozumab group and 64 patients (3.2%) in the alendronate group. The most frequently reported AEs (≥ 0.2% in either treatment group) leading to discontinuation of the study drug were nausea (romosozumab: 0.2%; alendronate: < 0.1%), dyspepsia (romosozumab: 0.2%; alendronate: < 0.1%), abdominal pain upper (romosozumab: 0.1%; alendronate: 0.2%), and myalgia (romosozumab: 0.2%; alendronate: < 0.1%)

In the primary analysis period, AEs led to withdrawal of the study drug for 133 patients (6.5%) who received romosozumab/alendronate and 146 patients (7.2%) who received alendronate/alendronate during the primary analysis period. The AEs that led to discontinuation of the study drug for more than 0.2% of patients in either treatment group were dyspepsia (romosozumab/alendronate: 0.4%; alendronate/alendronate: 0.2%), abdominal pain upper (romosozumab/alendronate: 0.3%; alendronate/alendronate: 0.3%), and myalgia (romosozumab/alendronate: 0.3%; alendronate/alendronate: < 0.1%).

Mortality

FRAME Study

At month 12, fatal AEs occurred in 29 patients (0.8%) in the romosozumab group and 23 patients (0.6%) in the placebo group during double-blind treatment. The AEs that occurred in at least 0.1% of patients in either treatment group resulted in death (0.1% in each group) and malignant lung neoplasm (romosozumab: 0.1%; placebo: 0%). The AEs resulting in death that were considered by the investigator to be related to investigational product included 1 event of deep vein thrombosis in the romosozumab group and 1 event of sudden death in the placebo group.

At month 24, additional AEs leading to death occurred in 23 patients (0.7%) in the romosozumab/denosumab group and 24 patients (0.7%) in the placebo/denosumab group during the open-label denosumab treatment period, resulting in a cumulative total of 52 deaths (1.5%) in the romosozumab/denosumab group and 47 deaths (1.3%) in the placebo/denosumab group. Over the 24-month study period, AEs resulting in death that occurred in at least 0.1% of patients in either treatment group were malignant lung neoplasm (romosozumab/denosumab: 0.2%; placebo/denosumab: < 0.1%), death (romosozumab/denosumab: 0.3%; placebo/denosumab: 0.1%), and sudden death (romosozumab/denosumab: 0%; placebo/denosumab: 0.1%). Two (< 0.1%) and 1 (< 0.1%), respectively, were considered by the investigator to be related to treatment.

ARCH Study

At month 12, fatal AEs occurred in 30 patients (1.5%) in the romosozumab group and 21 patients (1.0%) in the alendronate group during the 12-month double-blind treatment period. The AEs resulting in death that occurred in at least 2 patients in either treatment group were acute myocardial infarction (romosozumab: 3 patients; alendronate: 0 patients), cardiac failure (romosozumab: 2 patients; alendronate: 1 patient), urosepsis (romosozumab: 2 patients; alendronate: 0 patients), pneumonia (romosozumab: 1 patient; alendronate: 3 patients), cerebrovascular accident (romosozumab: 1 patient; alendronate: 2 patients), death (romosozumab: 1 patient; alendronate: 2 patients), and sudden death (romosozumab: 1 patient; alendronate: 2 patients). During the 12-month double-blind treatment period, the AEs resulting in death that were considered by the investigator to be related to the investigational product were reported for no patients in the romosozumab group and 3 patients in the alendronate group (1 event each of pneumonia, cervix carcinoma stage IV, and cerebrovascular accident).

Through the primary analysis period, AEs leading to death occurred in 60 patients (2.9%) in the romosozumab/alendronate group and 69 patients (3.5%) in the alendronate/alendronate group during the open-label alendronate treatment period, resulting in a cumulative total of 90 deaths (4.4%) in the romosozumab/alendronate group and 90 deaths (4.5%) in the alendronate/alendronate group. Over the primary analysis period, the AEs resulting in death that occurred in at least 4 patients (0.2%) in either treatment group were acute myocardial infarction (romosozumab/alendronate: 5 patients; alendronate/alendronate: 4 patients), cardiac failure (romosozumab/alendronate: 3 patients; alendronate/alendronate: 4 patients), cardiac failure congestive (romosozumab/alendronate: 2 patients; alendronate/alendronate: 4 patients), pneumonia (romosozumab/alendronate: 7 patients; alendronate/alendronate: 8 patients), cerebrovascular accident (romosozumab/alendronate: 4 patients; alendronate/alendronate: 3 patients), death (romosozumab/alendronate: 13 patients; alendronate/alendronate: 17 patients), and sudden death (romosozumab/alendronate: 4 patients; alendronate/alendronate: 3 patients).

Notable Harms

The incidence of AEs of particular interest was similar between treatment groups in FRAME and ARCH.

At month 12 in FRAME, hypersensitivity occurred in 6.8% of patients in the romosozumab group and 6.9% of patients in the placebo group. Serious cardiovascular events were reported in 1.2% of patients treated with romosozumab and 1.1% of patients treated with placebo. At month 24, hypersensitivity occurred in 8.8% of patients in the romosozumab group and 9.3% of patients in the placebo group. Serious cardiovascular events were reported in 2.3% of patients treated with romosozumab and 2.2% of patients treated with placebo.

At month 12 in ARCH, hypersensitivity occurred in 6.0% of patients in the romosozumab group and 5.9% of patients in the alendronate group. Serious cardiovascular events were reported in 2.5% of patients treated with romosozumab and 1.9% of patients treated with alendronate. At month 24, hypersensitivity occurred in 10.0% of patients in the romosozumab group and 9.2% of patients in the alendronate group. Serious cardiovascular events were reported in 6.5% of patients treated with romosozumab and 6.1% of patients treated with alendronate. Of the major adverse cardiac events, ischemic cardiac events and cerebrovascular events were more frequent in the romosozumab group.

Cases of ONJ were rare in both studies.

Critical Appraisal

Internal Validity

Both FRAME and ARCH were large, phase III, double-blind RCTs. The ARCH study had an active control arm of oral bisphosphonate. The sample sizes of the 2 studies provided sufficient power to detect the differences between treatment groups. Appropriate methods were used to randomize patients to treatments and conceal treatment allocation. These studies reported on objective outcomes that are deemed reasonable for clinical trials of postmenopausal women with osteoporosis (i.e., incidence of fracture and change in BMD), limiting the possibility of bias.

In general, patients’ characteristics appear to be balanced at baseline between groups. The study results are unlikely to be biased by imbalanced baseline patient characteristics. The completion rate at the end of 1 year of treatment was close to 90%, and greater than 80% in FRAME and 77% in ARCH at the end of 2 years of treatment. The reasons for dropouts were similar between treatment groups. The clinical expert stated that dropout rates were acceptable and consistent with other clinical trials of osteoporosis.

The dropout rates in both FRAME and ARCH were high, which could affect the study results. In ARCH, there was a concern for the potential for bias from the exclusion of patients from the primary analysis set for vertebral fractures through month 24. The inclusion of these patients in an analysis based on the full analysis set was also problematic when they were considered as not having a vertebral fracture and subsequently lowered the incidence rates. In addition, missing data on post-baseline vertebral fracture status due to missing spinal X-ray assessments were imputed using the LOCF method. In response to a request from Health Canada, sensitivity analyses using various methods (e.g., LOCF approach or multiple imputation with the missing-at-random assumption) were conducted to assess the robustness of the primary analysis results. However, the sensitivity analysis using a multiple imputation methodology under the missing-at-random assumption was still a concern, given that the primary reasons for discontinuation from the study included consent withdrawal, death, and AEs. Additional sensitivity analyses that did not assume that the data were missing at random were requested by Health Canada. The results showed that the magnitude of the treatment effect was reduced (compared to the original analyses and as the analyses using multiple imputation under the missing-at-random assumption) but remained statistically significant. The results were robust to the handling of missing data.³⁷ In addition, the main reasons for study discontinuation were withdrawn consent and death. No sensitivity analyses were performed based on these reasons.

Multiplicity was controlled for in FRAME and ARCH based on a step-down procedure, with the primary and selected secondary outcome measures included. Outcomes outside of the testing hierarchy, such as HRQoL (an exploratory outcome in these 2 studies) or occurrence of cardiovascular events, need to be interpreted with caution due to the possible inflated type I error.

The clinical expert provided input on defining high-risk patients. Risks of future fractures should be determined based on multiple factors, including patients’ demographic characteristics, history of fracture, sites of previous fracture, use of certain medications, FRAX scores, BMD scores, and many others. For example, a hip fracture carries more weight than an ankle or wrist fracture for future fracture risk. The clinical expert also indicated that fracture risk models continue to be revised with time and the latest versions should be used; forthcoming iterations of FRAX are anticipated to incorporate the effects of prior pharmacotherapy, type 1 and type 2 diabetes, how recent a fracture is, and the effects of multiple fractures. Subgroup analyses were performed in the 2 studies to examine the consistency of the primary analyses results across subgroup levels, based on age, prevalent vertebral fracture status, history of fragility fracture and baseline BMD T-scores, among others. These subgroups were not adjusted for multiple comparisons and therefore their results are only considered as supportive evidence. Moreover, it would be desirable to consider more factors, such as the effect of patient compliance to prior treatment and previous fractures that carry more weight (e.g., hip fractures or multiple fragility fractures). As well, previous pharmacologic treatment for osteoporosis may be a treatment effect modifier and was a subgroup of interest specified in the protocol of this review. In both FRAME and ARCH, patients were generally excluded based on previous osteoporosis treatment within specified time periods of study enrolment. For example, patients were excluded if they had received any dose of an oral bisphosphonate within 3 months before randomization, more than 1 month of cumulative use between 3 and 12 months before randomization, or more than 3 years of cumulative use unless the last dose received was at least 5 years before randomization. Additionally, a washout period of at least 35 days was required for patients who were previously treated with osteoporosis drugs. However, antiresorptive drugs such as oral bisphosphonates may have a long half-life in bone, ranging from 1 to 10 years, depending on the rate of bone turnover. It is unclear to what extent there were carry-over effects of previous treatments and whether these were distributed differently between groups. As a result, the potential impact of this factor could not be evaluated.

Results of the FRAME study are difficult to interpret at month 24 because patients treated with romosozumab followed by denosumab received 2 years of active treatment while those in the placebo followed by denosumab group received only 1 year of active treatment. The European Medicines Agency and FDA guidelines suggest a minimum of 24 months of active treatment data to assess fracture outcomes and bone safety due to the delayed effects of anti-osteoporosis drugs, particularly antiresorptive drugs such as denosumab.^39,40 The comparison and duration of treatment for the comparison in FRAME are therefore problematic.

For ARCH, it is unclear whether the treatment response in the alendronate group is consistent with what has been observed in similar patient populations in other trials. Specifically, assay sensitivity of the ARCH study is difficult to determine in the absence of a placebo group and between-trial heterogeneity. It is therefore unclear whether the magnitude of the treatment effect with romosozumab versus alendronate will be observed in clinical practice.

External Validity

A high proportion of patients (57% in FRAME and 76% in ARCH) failed the screening. The main reasons for the screening failure were not meeting the BMD T-score or fracture requirement and vitamin D insufficiency (Table 9).

According to the clinical expert consulted by CADTH, the inclusion and exclusion criteria for the 2 pivotal studies were generally consistent with clinical practice. Based on patients’ baseline characteristics, the study populations reflect a typical Canadian population that would receive romosozumab in practice. However, patients were excluded from FRAME if they had a BMD T-score of −3.5 or lower at the total hip or femoral neck, a hip fracture at any time and/or any severe vertebral fracture, or more than 2 moderate vertebral fractures. Although patients with these characteristics were likely excluded because the comparator in FRAME was placebo, these patients represent those at a higher risk for fracture and the patient population for which romosozumab is intended to treat. The ARCH study did not exclude patients with these characteristics and is therefore more similar to the patient population intended for treatment with romosozumab. Compared to FRAME, patients in the ARCH study were older, almost all had historical fractures and prevalent vertebral fractures, and they had lower BMD T-scores at different body sites and a higher FRAX score at baseline, which is consistent with patients with high risk of osteoporotic fracture.

The clinical expert also indicated that, in practice, most patients with osteoporosis are untreated due to fear of side effects and potential long-term adverse effects, such as ONJ and atypical hip fractures, as well as failure to recognize the condition. Across North America, less than 20% and, in clinical trials of osteoporosis, less than 10% of hip fracture patients were treated for osteoporosis when they were surveyed a year after the incident fracture. Therefore, in the included study (FRAME), using placebo as a comparator to romosozumab reflects the current clinical practice in Canada and would provide evidence for the effect of the study drug in preventing future fractures in the study population.

Because alendronate was the only active comparator available for this review, there is a lack of direct evidence to demonstrate comparative efficacy and safety of romosozumab versus other osteoporosis medications, and anabolic drugs in particular.

The FRAME and ARCH studies were designed to include a treatment-naive population. However, there is likely a percentage of patients who have been treated with medications, such as antiresorptive therapies, and who may be prescribed romosozumab. The results from the FRAME and ARCH studies do not adequately inform on this sequencing of treatments.

Indirect Evidence

Objectives and Methods for the Summary of Indirect Evidence

No head-to-head comparison of romosozumab against other relevant treatments for osteoporosis, other than alendronate, was available for this review. CADTH conducted a literature search to identify potentially relevant ITCs in people with osteoporosis at high risk for fracture, in addition to reviewing the sponsor’s ITC submission to CADTH. A focused but unlimited literature search for NMAs dealing with osteoporosis was run in MEDLINE All (1946–) on March 09, 2021. Titles, abstracts, and full-text articles were screened for inclusion by 1 reviewer based on the population, intervention, comparator, and outcome criteria outlined in Table 5. One sponsor-submitted ITC was summarized and critically appraised.¹¹ The sponsor-submitted ITC was used to inform the pharmacoeconomic model.

Description of Indirect Comparisons

The sponsor-submitted ITC, which is an NMA, aimed to evaluate the relative clinical efficacy of romosozumab to teriparatide, denosumab, raloxifene, zoledronate, risedronate, ibandronate, alendronate, strontium ranelate, vitamin D plus calcium, bazedoxifene, tibolone, hormone therapy, calcitonin, lasofoxifene, vitamin D, calcium, parathyroid hormone 1 to 84, and abaloparatide in postmenopausal women with primary osteoporosis or osteopenia at risk for developing fragility fractures. The sponsor performed a systematic review to identify relevant studies for inclusion in the ITC. The 3 outcomes analyzed were occurrence of vertebral, hip, and nonvertebral fragility fractures.

The population, intervention, comparators, outcomes, and design of studies included in the sponsor’s ITC are provided in Table 19.

Methods of Sponsor-Submitted Indirect Treatment Comparison

Objectives

The primary objective of the sponsor’s ITC was to compare the efficacy of various drugs in the treatment of primary osteoporosis or osteopenia for the prevention of fragility fractures in postmenopausal women.

Study Selection Methods

A literature of multiple electronic databases, including MEDLINE through the Ovid interface, Embase, Cochrane Central Register of Controlled Trials (CENTRAL), ISI Web of Science, and Scopus, were searched. The search was not restricted by the language of the publication or the country of origin of the study.

Studies were eligible and included if they were RCTs, enrolled postmenopausal women with primary osteoporosis or osteopenia at risk for developing fragility fractures, compared 1 or more of the interventions of interest to each other or to placebo, and reported the outcomes of interest (vertebral, hip, and nonvertebral fragility fractures) as a primary or secondary outcome or as an AE. The authors of the sponsor-submitted ITC did not provide a set of exclusion criteria for the systematic review.

Titles and abstracts were reviewed for potential study inclusion. Two reviewers independently evaluated study eligibility based on their titles and abstracts. If at least 1 reviewer determined that an article was potentially eligible, its full-text version was retrieved, and pairs of reviewers assessed its eligibility. Conflicts and discrepancies between the 2 reviewers were resolved through discussion and consensus.

A standardized and piloted data extraction form using an online reference management system (Distiller SR, Ottawa, Canada) was used. Pairs of reviewers extracted data independently and resolved disagreements by discussion and consensus. The following variables were extracted: baseline characteristics, patient demographics, type of interventions, and outcome data. The outcome data extracted corresponded to the number of patients with the outcome of fracture, unless only the number of fractures (not patients) was reported. If the fracture was reported as a clinical fracture and if it was assessed by radiography, the radiographic fracture was extracted. Some studies reported fractures by location without the “nonvertebral” label; in these cases, all the fractures, including hip and/or pelvis fractures, were considered nonvertebral. For studies that report hip and pelvis fractures separately, only the hip fracture outcome was characterized as hip fracture.

Risk of bias was assessed using the Cochrane risk-of-bias tool.¹¹ Studies with a loss to follow-up of greater than 10% were qualified as having a high risk of attrition bias. The Grading of Recommendations, Assessment, Development, and Evaluation framework was used to rate the certainty in the estimates of the direct and indirect comparisons as part of the NMA. Certainty in the estimates was rated down because of methodological limitations of the trial, indirectness, inconsistency, imprecision, and reporting bias. The confidence in the estimates was rated as high, moderate, low, or very low.

The clinical efficacy outcomes assessed were risks of sustaining vertebral, hip, and nonvertebral fragility fractures.

Indirect Treatment Comparison Analysis Methods

The authors of the submitted ITC used a frequentist NMA approach. Direct (head-to-head) comparisons were conducted using the random-effects model as described by DerSimonian and Laird⁴¹ to estimate pooled relative risks and 95% CIs. A multivariate random-effects NMA was also conducted to combine the direct and indirect comparisons of drugs using a frequentist consistency model.^42,43

The random-effects model was chosen to account for the anticipated heterogeneity between studies. For the random-effects model of direct comparisons, between-trial heterogeneity was assessed using the I² statistic, for which a value greater than 50% suggests substantial between-trial heterogeneity.

Network inconsistency between direct and indirect comparisons was assessed using the node-splitting method by comparing the direct and indirect estimates and conducting z tests using a significance level of 5%.

Potential publication bias was assessed using the Egger regression test and visual inspection of funnel plots.

Sensitivity analyses were conducted for the effects of zoledronate and calcitonin based on dosage and administration route.

All statistical analyses were performed using version 15 of STATA (StatCorp LP, College Station, TX).

Table 20 below present a summary of the methods used for the ITC.

Results of Sponsor-Submitted Indirect Treatment Comparison

Summary of Included Studies

A total of 107 RCTs were included in the ITC. The studies were published between 1979 and 2017; this included 27 studies published before 2000. A total of 193,987 postmenopausal women with a mean age of 66.2 years were included. The majority (55.1%) of participants were White. The duration of the trials ranged between 3 and 120 months, with a median of 27.7 months. The sample size was between 36 and 36,282 patients in treatment arms. Most of the included trials involved comparisons with placebo. Hormone therapy and alendronate were the most commonly tested agents, followed by calcium and vitamin D. When calcium and/or vitamin D were given to both arms of a trial, their effect was considered neutralized and equal in both arms, allowing the arm that received calcium and/or vitamin D plus placebo to be considered placebo. A summary of the study characteristics and patient’s baseline characteristics for each of the included studies are presented in Table 37 and Table 38.

The risk of bias was assessed using the Cochrane risk-of-bias tool. Of the 107 included RCTs, 49 had concealment of a treatment allocation, 102 had well-balanced baseline prognostic factors, 76 reported blinded patients, 77 reported blinded caregivers, 21 reported blinded outcome assessors, 11 reported blinded data collectors, and 7 reported blinded data analysts. The quality and appropriateness of the randomization procedures of the included RCTs were not provided. Overall, 55 RCTs had a high risk of bias. The risk of bias was high in most of the RCTs involving calcitonin, calcium, and vitamin D, as well as in some older bisphosphonate trials. No RCTs were excluded based on the assessment of bias.

Results

Hip Fractures

In all, 107 studies and 19 treatments were included in the hip fracture network (Figure 4). For studies reporting hip and pelvis fractures separately, only the hip fracture outcome was analyzed as hip fracture.

Figure 4: Network of Trials Included in the Hip Fracture Analysis

Source: Moreno et al.,⁴⁴ copyright 2019. This work is licensed under the Attribution 4.0 International Public Licence. Full text available at https://figshare.com/articles/dataset/Appendix_to_network_meta-analysis_Fragility_Fractures/7629344.

Treatment with romosozumab was associated with a significant reduction in the risk of sustaining hip fragility fractures compared to raloxifene. No significant difference in the risk of sustaining hip fractures was seen when romosozumab was compared with alendronate, risedronate, zoledronate, or denosumab.

Nonvertebral Fractures

The nonvertebral fracture network included 107 studies and 20 treatments (Figure 5). Some studies in the evidence network reported fractures by location without the “nonvertebral” label; in these cases, all the fractures, including hip and/or pelvis fractures, were considered nonvertebral.

Figure 5: Network of Trials Included in the Nonvertebral Fracture Analysis

PTH = parathyroid hormone.

Source: Moreno et al.,⁴⁴ Copyright 2019. This work is licensed under the Attribution 4.0 International Public Licence. Full text available at https://figshare.com/articles/dataset/Appendix_to_network_meta-analysis_Fragility_Fractures/7629344.

Treatment with romosozumab was associated with a significant reduction in the risk of sustaining nonvertebral fragility fractures compared to raloxifene. There was no significant difference between romosozumab versus alendronate, risedronate, zoledronate, or denosumab in the risk of sustaining nonvertebral fractures.

Vertebral Fractures

The vertebral fracture network included 107 studies and 20 treatments (Figure 6). Vertebral fractures are characterized as symptomatic or non-symptomatic, with 70% being non-symptomatic. The sponsor-submitted ITC did not explicitly state whether vertebral fractures defined and assessed in the studies were symptomatic or non-symptomatic.

Figure 6: Network of Trials Included in the Vertebral Fracture Analysis

PTH = parathyroid hormone.

Source: Moreno et al.,⁴⁴ copyright 2019. This work is licensed under the Attribution 4.0 International Public Licence. Full text available at https://figshare.com/articles/dataset/Appendix_to_network_meta-analysis_Fragility_Fractures/7629344.

Treatment with romosozumab was associated with a significant reduction in the risk of sustaining vertebral fragility fractures compared to alendronate, risedronate and raloxifene. There was no significant difference between romosozumab versus zoledronate or denosumab in the risk of sustaining vertebral fractures.

Sensitivity Analyses Based on the Route of Administration

Sensitivity analyses of the effects of zoledronate and calcitonin based on dosage and administration route were conducted. No statistically significant difference was found between 5 mg zoledronate and other doses of zoledronate in the effect on nonvertebral fractures. A sensitivity analysis of a single 5 mg dose of zoledronate found a statistically significant reduction in vertebral fractures. A sensitivity analysis of intranasal calcitonin demonstrated a statistically significant reduction in vertebral fractures, whereas the effects of injectable or oral calcitonin were not statistically significant. The results are presented in Figures 7 through 12 in Appendix 4.

Critical Appraisal of the Sponsor-Submitted Indirect Treatment Comparison

The sponsor’s rationale and objectives for conducting the ITC were clearly reported. According to the clinical expert consulted by CADTH for this review, the comparators and the dosages used were generally similar to those used in clinical practice. A comprehensive systematic review was performed with a 2-stage, dual-selection process. The language of publication was not restricted, thereby reducing publication bias. The clinical efficacy outcomes assessed were risks of sustaining vertebral, hip, and nonvertebral fragility fractures. Key efficacy outcomes for HRQoL or symptoms were not included in the analysis. Safety outcomes (i.e., AEs and SAEs), mortality, withdrawals due to AEs, and notable harms were not reported. Overall, 55 out of the 107 included RCTs had a high risk of bias. The risk of bias was high in most of the RCTs involving calcitonin, calcium, and vitamin D, as well as in some older bisphosphonate trials. No RCTs were excluded based on the assessment of bias.

Clinical heterogeneity was present in the analysis due to varying study duration, blinding, dosage, fracture risk assessment, publication date, patient demographic and clinical characteristics, and clinical-effect modifiers. Heterogeneity could have been reduced by specifying additional inclusion criteria for studies, such as requiring data on the blinding status, dosage, fracture risk assessment, and follow-up duration. The clinical expert consulted for this review indicated that many additional factors are considered potential effect modifiers, such as ethnicity, country of origin, calcium and vitamin D intake, weight-bearing physical activity, falls and conditions that increase the risk of falls, overweight or obesity, glucocorticoid use, lactose intolerance, gastrointestinal disorders (Crohn or celiac disease, ulcerative colitis, and bowel resections), systemic inflammatory disorders such as rheumatoid arthritis and lupus, diabetes (type 1 or 2), early menopause, and lack of postmenopausal hormone replacement therapy. The potential impact of these were not evaluated and adjusted for (where appropriate) in the ITC. Furthermore, the Canadian FRAX model and CAROC assume that the patient is Caucasian. In the US, the FRAX model distinguishes African-American from Hispanic and Caucasian patients. As such, a patient’s country of origin is important, and Asians tend to have a much lower fracture risk than Caucasians at the same level of BMD (the former can be low-risk when the latter are high-risk, as experts have seen in many patients). According to the clinical expert consulted for this review, ethnicity and how it is used in fracture risk assessment tools such as FRAX and CAROC have substantial effects on the risk of sustaining fragility fractures. The sponsor-submitted ITC did not incorporate or adjust for these effects when summarizing the ITC results. In addition, in the sponsor-submitted ITC the trials examining hormone replacement therapy tended to enrol younger women, and trials examining denosumab, romosozumab, teriparatide, and abaloparatide tended to enrol women who were more likely to have prevalent fractures or an increased risk for fracture. Also, the 2 trials involving romosozumab differed significantly in patient population and treatment dosage, as described in previous sections, providing further evidence that the ITC had substantial heterogeneity. The sponsor-submitted ITC did not perform sensitivity analyses by removing outlier studies that had effect modifiers that could potentially influence the results. The sponsor-submitted ITC also did not conduct a meta-regression to attempt to account for effect modifiers that could potentially minimize their influence on the results.

While the authors of the sponsor-submitted ITC did assess the risk of bias for the included studies, they did not provide a plan to investigate the impact of studies that were considered to be of low quality or have a high risk of bias.

Some studies in the evidence network reported fractures by location without the “nonvertebral” label; in these cases, all the fractures, including hip and/or pelvis fractures, were considered nonvertebral. This resulted in the same hip-fracture incidences being used in the calculations of 2 different outcomes: the risk of sustaining hip fractures and the risk of sustaining nonvertebral fractures. This double-counting thereby reduced the comparability of the results synthesized for different outcomes and overall internal validity of the ITC.

Vertebral fractures are characterized as symptomatic or non-symptomatic, with 70% being non-symptomatic, according to the clinical expert consulted for this review. It was not explicitly stated in the sponsor-submitted ITC whether vertebral fractures defined and assessed in the studies were symptomatic or non-symptomatic, which limits the generalizability of the synthesized results for vertebral fractures and the overall external validity of the ITC.

Sample size was large for most studies included in the ITC, the number of trials in relation to the number of nodes in the network was sufficiently large, and the total number of studies in the network was large (107). The degree of uncertainty in treatment effect estimates (risk ratios) was therefore not substantial, as demonstrated by the relatively narrow 95% CIs.

As a definition for “placebo” was not provided, it was not possible to determine whether this intervention was defined consistently across studies.

Summary

The sponsor-submitted ITC suggested that romosozumab was superior to alendronate and risedronate in preventing vertebral fractures and superior to raloxifene for all 3 outcomes evaluated (hip, vertebral, and nonvertebral fractures). No statistically significant differences were reported between romosozumab and zoledronate or denosumab on any of the outcomes evaluated. However, the sponsor-submitted ITC lacked: reporting of details for clinical heterogeneity in the included studies; effect modifiers and their influence on the results; construction of nodes in the ITC network; and details of assessments of essential NMA assumptions, including transitivity, consistency, and heterogeneity, that would better inform on the certainty of the indirect evidence. The ITC also did not conduct sensitivity and subgroup analyses to sufficiently assess homogeneity, or conduct a meta-regression to adjust for effect modifiers that could potentially influence the results. In addition, the majority of the studies in the evidence network were rated as having a high risk of bias, but no sensitivity analysis was conducted to examine the influence of these studies. Furthermore, given that the sponsor-submitted NMA did not provide a definition for “placebo,” double-counted hip fractures when analyzing hip and nonvertebral fracture outcomes, and did not distinguish between symptomatic and non-symptomatic vertebral fractures, there is uncertainty around the ITCs that undermines the internal and external validity of the ITC.

Other Relevant Evidence

This section includes submitted long-term extension studies and additional relevant studies in the sponsor’s submission to CADTH that were considered to address important gaps in the evidence included in the systematic review.

One long-term extension study, the FRAME Extension, was summarized to provide long-term evidence regarding romosozumab for the treatment of osteoporosis in postmenopausal women at high risk for fracture. The extension phase of the trial did not meet the eligibility criteria pre-specified for the CADTH systematic review as the main interventions of the trial did not involve romosozumab. However, the trial provided long-term efficacy data for patients treated with denosumab for 2 years after initially being treated with either romosozumab or placebo.

Long-Term Extension Study

The FRAME study is described in the Clinical Evidence section of this report. Briefly, the goal of the trial was to determine if 12 months of treatment with romosozumab was effective in reducing the incidence of new vertebral fractures, and if 12 months of romosozumab followed by 12 months of denosumab was effective in reducing the incidence of new vertebral fractures, compared to placebo, among postmenopausal women with osteoporosis.¹² After the 24-month primary analysis of the trial, eligible patients could enrol in the 12-month open-label extension period, during which patients could continue to receive open-label denosumab at 60 mg every 6 months. The following section discusses results from the extension phase of the trial.¹²

Methods

Methodological details of the FRAME trial are described previously in this report. Figure 2 of this report illustrates all phases of the FRAME Trial.

Populations

Eligibility criteria of the FRAME study are descried elsewhere in this CADTH report. Briefly, inclusion criteria specified patients who were postmenopausal women between the ages of 55 and 90 years of age with a BMD T-score of no more than −2.50 at the total hip or femoral neck. Exclusion criteria included patients with a BMD T-score of no more than −3.50 at the total hip or femoral neck, a hip fracture at any time, any severe or more than 2 moderate vertebral fractures, severe metabolic or bone diseases, significant laboratory abnormalities, and the use of drugs that can affect bone metabolism. However, some therapies were permitted for use during off-treatment periods before trial randomization.¹²

Interventions

Patients who received open-label denosumab in the extension phase continued to receive 60 mg of denosumab subcutaneously every 6 months. Patients who continued into the extension phase remained blinded to their initial treatment assignment.¹²

Outcomes

All outcomes assessed in the extension portion of the FRAME study were based on the final analysis (36-month) study period. A summary of outcome objectives and specific end points assessed is included in Table 22.

Statistical Analysis

All analyses of efficacy were performed on patients by the treatment group to which they were randomized, regardless of the treatment they actually received. Descriptive statistics (i.e., mean, SD, minimum, 25th percentile, median, 75th percentile, maximum, and number of non-missing observations) were used to summarize continuous outcomes. Frequencies and percentages were used to summarize nominal and ordinal categorical outcomes; unless otherwise specified, percentages were based on the patients with non-missing observations. Kaplan–Meier event rates were used to capture time-to-event outcomes. P values were 2-sided. For exploratory and substudy end points, including this extension phase, P values were nominal and were not adjusted for multiplicity.¹²

A summary of statistical methods for outcomes is provided in Table 23.

Safety Analysis

The safety analysis set was used for all analyses of safety. No formal statistical testing was planned for safety analyses. Safety data were summarized based on actual treatment received by patients during the double-blind period of the FRAME trial. Patient incidence rates for the extension phase (i.e., the 36-month study period) included all events occurring during both the main double-blind phase of the trial and the extension phase for all patients receiving at least 1 dose of denosumab. All AEs were coded using the Medical Dictionary for Regulatory Activities version 19.1.¹²

All deaths and cardiovascular-related deaths were submitted for adjudication to an external independent committee of physicians with expertise in cardiology, osteoporosis, internal medicine, or neurology. Adjudicated cardiovascular events of death, cardiac ischemic events, cerebrovascular events, noncoronary revascularization, heart failure, and peripheral vascular events not requiring revascularization were summarized using patient incidence rates, odds ratios, and 95% CIs. No statistical tests for safety analyses were performed.¹²

Substudy Analyses of Bone Turnover Markers and Biomarkers

No stratification (i.e., by age strata and prevalent vertebral fracture strata) occurred during analyses of BTM and biomarkers. The level of significance was 0.05 and all P values were considered nominal and calculated from a Wilcoxon rank sum test. No adjustments were made for multiplicity.¹²

Analysis Sets

• Primary analysis set for vertebral fractures: all randomized patients with a baseline and at least 1 post-baseline evaluation of vertebral fractures at or before the time point under consideration, including patients with missing baseline Genant semiquantitative scores whose first post-baseline spinal radiograph shows no fracture on the same vertebrae.

• Full analysis set: all randomized patients. The full analysis set was the primary analysis set for the following fracture efficacy end points: nonvertebral, clinical, major nonvertebral, major osteoporotic, and hip.¹²

• Primary efficacy analysis set for BMD end points: all randomized patients who had a baseline and at least 1 post-baseline evaluation at or before the time point under consideration in the study period.¹²

• Safety analysis set: all randomized patients who received at least 1 dose of the investigational product in the 12-month double-blind period. The patient incidence rates for the 36-month study period included all events that occurred in the double-blind period and, in addition, all events that occurred in the open-label and extension periods for those patients who received at least 1 dose of denosumab.¹²

• Extension period analysis set: all randomized patients who entered the extension period at month 24. This set was used to summarize baseline characteristics and demographics for patients who entered the extension study.¹²

• BTM and biomarker substudy efficacy analysis set: all randomized patients who enrolled in the BTM substudy and had a baseline and at least 1 reported post-baseline reported BTM result.¹²

Patient Disposition

Eligibility for the 12-month extension phase was reported to follow the overall enrolment of the main phase of the trial, as there were no noticeable changes in proportions of patients remaining in the study. The study was conducted across 222 centres in 25 countries.¹²

A summary of patients completing and withdrawing from the FRAME extension study is reported in Table 24. A total of 7,180 patients were randomized into the main phase of the trial; 3,589 patients to the romosozumab group and 3,591 patients to the placebo group. A total of 6,045 patients completed the 24-month study period and entered the extension phase. Of the 7,180 randomized patients, 80% completed the 36-month study period, which was completed by 79.4% of patients within the romosozumab/denosumab group and 80.5% of patients in the placebo/denosumab group. Approximately 20% of patients discontinued from the 36-month study period, most commonly due to withdrawal of consent (10.3%).¹²

Important protocol deviations during the extension phase of the FRAME trial involved 256 patients (3.6%); these included deviations related to eligibility criteria that were considered important.¹²

Exposure to Study Treatments

A summary of baseline characteristics for the FRAME study was reported in the Baseline Characteristics subsection of the Clinical Evidence section of this clinical report. Demographic and disease baseline characteristics of patients during the extension phase of the trial were similar to demographic characteristics of patients during the main phase of the trial. Baseline characteristics were similar across both treatment groups.¹²

The following exposures to treatment are reported for the final analysis at month 36. Exposure to treatment during the primary 24 months of the trial are reported in the Exposure to Study Treatment subsection of the Clinical Evidence section of this report. A total of 3,581 patients received a minimum of 1 dose of romosozumab (210 mg monthly) and 3,576 patients received placebo; these patients were included in the safety analysis set. Of patients who entered the 12-month extension period, 3,087 received a minimum of 1 dose of denosumab in the romosozumab/denosumab group compared to 3,112 patients in the placebo/denosumab group.¹² Most patients in both the romosozumab/denosumab group (90.6%) and the placebo/denosumab group (91.2%) received all 4 doses of denosumab. The mean cumulative exposure of denosumab was similar across both treatment groups at 229.2 mg (range = 60 mg to 240 mg) in the romosozumab/denosumab group and 230.5 mg (range = 60 mg to 240 mg) in the placebo/denosumab group.¹²

Efficacy

None of the analyses for the FRAME study were adjusted for multiplicity; all results summarized below are therefore considered descriptive. A summary of fracture end points through to month 36 of the FRAME study are reported in Table 25.

New Vertebral Fractures Through Month 36

A total of 32 patients (1.0%) in the romosozumab/denosumab group had a new vertebral fracture compared to 94 patients (2.8%) in the placebo/denosumab group. A relative risk reduction of 66% (95% CI, 49 to 77; risk ratio = 0.34) was observed for patients in the romosozumab/denosumab compared with patients in the placebo/denosumab group.¹²

Clinical Fractures (Nonvertebral Fractures and Clinical Vertebral Fractures) Through Month 36

In the romosozumab/denosumab group, 143 patients (4.0%) in the romosozumab/denosumab group and 196 patients (5.5%) in the placebo/denosumab group had a clinical fracture resulting in a relative risk reduction of 27% (95% CI, 10 to 41; hazard ratio = 0.73) for romosozumab/denosumab compared with placebo/denosumab.¹²

Nonvertebral Fractures Through Month 36

A total of 139 patients (3.9%) and 176 patients (4.9%) in the romosozumab/denosumab group and placebo/denosumab groups, respectively, had a nonvertebral fracture. The relative risk reduction for romosozumab/denosumab compared with placebo/denosumab was 21% (95% CI, 1 to 37; hazard ratio = 0.79).¹²

Major Nonvertebral Fractures Through Month 36

One hundred patients (2.8%) in the romosozumab/denosumab group compared to 138 (3.8%) in the placebo/denosumab group had a major nonvertebral fracture. A relative risk reduction of 27% (95% CI, 6 to 44; hazard ratio = 0.73) was observed for patients in the romosozumab/denosumab group compared with patients in the placebo/denosumab group.¹²

New or Worsening Vertebral Fractures Through Month 36

Through month 36, 33 patients (1.0%) in the romosozumab/denosumab group and 94 (2.8%) in the placebo/denosumab group had a new or worsening vertebral fracture; the relative risk reduction for romosozumab/denosumab compared with placebo/denosumab was 65% (95% CI, 48 to 76; risk ratio = 0.35).¹²

Hip Fractures Through Month 36

A total of 18 patients (0.5%) in the romosozumab/denosumab group had a hip fracture versus 31 (0.9%) in the placebo/denosumab group. The relative risk reduction for romosozumab/denosumab compared with placebo/denosumab was 41% (95% CI, −5 to 67; hazard ratio = 0.59).¹²

Major Osteoporotic Fractures Through Month 36

A major osteoporotic fracture was observed in 103 patients (2.9%) in the romosozumab/denosumab group and 147 (4.1%) in the placebo/denosumab group. The relative risk reduction was 30% (95% CI, 10 to 45; hazard ratio = 0.70) for patients in the romosozumab/denosumab group compared with patients in the placebo/denosumab group.¹²

Multiple New or Worsening Vertebral Fractures Through Month 36

Two patients (< 0.1%) in the romosozumab/denosumab group and 20 patients (0.6%) in the placebo/denosumab group had multiple new or worsening vertebral fractures. The relative risk reduction for romosozumab/denosumab compared with placebo/denosumab was 90% (95% CI, 57 to 98; risk ratio = 0.10).¹²

Percent Change from Baseline in Bone Mineral Density by Dual-Energy X-ray Absorptiometry

Table 26 reports the change in BMD from baseline to month 36 of the study. Mean BMD values for the lumbar spine, total hip, and femoral neck for patients in the romosozumab/denosumab group were more than double the BMDs of the same locations for patients in the placebo/denosumab group.¹²

Bone Turnover Marker and Biomarker Substudy

Bone Formation Marker (Procollagen type 1 N-terminal Propeptide): The mean percent change from baseline in P1NP was −56% in the romosozumab/denosumab group and −57% in the placebo/denosumab group (Table 27).¹²

Bone Resorption Marker (Serum C-telopeptide): In the romosozumab/denosumab group, the mean percent change from baseline in sCTX was −14% compared to −41% in the placebo/denosumab group (Table 27). The mean percent change of sclerostin from baseline was 13.0% in the romosozumab/denosumab group versus 14.7% in the placebo/denosumab group.¹²

Harms

Frequently Reported Adverse Events

Adverse events occurred in 88.1% and 89.0% of patients in the romosozumab/denosumab and placebo/denosumab groups, respectively. A summary of frequently reported AEs (occurrence rate ≥ 5%) is provided in Table 28. The most commonly occurring AEs in the respective romosozumab/denosumab and placebo/denosumab groups were arthralgia (18.7% versus 18.6%), nasopharyngitis (18.2% versus 17.4%), back pain (14.5% versus 16.1%), fall (13.7% versus 15.2%), pain in extremity (11.7% versus 11.4%) and hypertension (11.0% versus 11.8%). All AEs were reported at a similar rate across both treatment groups.¹²

Serious Adverse Events

A summary of SAEs is reported in Table 29. In general, SAEs were infrequently reported and similar across both treatment groups. Most SAEs were reported among less than 1% of patients.¹²

Treatment Discontinuation and Withdrawal Due to Adverse Events

Treatment was discontinued due to AEs for 3.9% of patients in the romosozumab/denosumab group compared to 3.6% of patients in the placebo/denosumab group. In general, treatment discontinuations due to AEs were infrequently reported among all patients (occurring at a rate of < 0.2%). In the romosozumab/denosumab group, the most common AEs resulting in treatment discontinuation were pain in extremity (0.3%) and arthralgia (0.2%). In the placebo/denosumab group, the most common AEs resulting in treatment discontinuation were arthralgia and musculoskeletal pain (0.2% each) (Table 30).¹²

Withdrawal from the study due to AEs was reported in 64 patients (1.8%) in each treatment group. Adverse events that led to discontinuation from the trial were reported in no more than 0.1% of patients.¹²

Deaths

Over the 36-month study period, fatal AEs occurred in 72 patients (2.0%) and 85 patients (2.4%) in the romosozumab/denosumab and placebo/denosumab groups, respectively. The occurrence of treatment-related AEs resulting in death (indicated by the investigator that there was a possibility they may have been caused by the study drug) were generally infrequent, occurring in less than 0.1% of patients.¹²

Cardiovascular-related deaths were also captured and adjudicated separately by an independent committee of physicians. A summary of adjudicated positive cardiovascular deaths is supplied in Table 31. In total, 43 patients (1.2%) in the romosozumab/denosumab group and 50 patients (1.4%) in the placebo/denosumab group died from a cardiovascular-related cause. Approximately half of cardiovascular deaths in each treatment group were labelled “undetermined” but were counted among cardiovascular deaths. Fatal serious cardiovascular AEs were uncommon, occurring in no more than 0.2% of patients in each group. Overall, cardiovascular deaths were reported at similar rates across both treatment groups, although the number of patients who died from a cerebrovascular event was numerically higher in the placebo/denosumab group.¹²

Adverse Events of Special Interest

Certain AEs of special interest were specified in the CADTH systematic review protocol. Adverse events of special interest are reported in Table 32, and include adjudicated ONJ and adjudicated cardiovascular events, including cardiac ischemic events, heart failure, noncoronary revascularization, cerebrovascular events, and peripheral vascular events not requiring vascularization. Adverse events of special interest occurred at similar rates among both treatment groups.

Adjudicated Cardiovascular Events: Cardiovascular SAEs were reported among 128 patients (3.6%) in the romosozumab/denosumab group and 124 patients (3.5%) in the placebo/denosumab group. In general, there were no differences in the incidence of adjudicated cardiovascular events (e.g., cardiac ischemic event, heart failure, noncoronary revascularization, cerebrovascular event, and peripheral vascular events not requiring revascularization) between the 2 treatment groups.

Adjudicated Positive Osteonecrosis of the Jaw: Two events (< 0.1%), both occurring in the romosozumab/denosumab group, were adjudicated to positively show ONJ. No events were considered serious. All events occurred during the 24-month period of the trial; no additional positively adjudicated AEs of ONJ were identified during the 12-month extension phase of the study.¹²

Critical Appraisal

Internal Validity

During the open-label extension period all patients were assigned to the same treatment. Due to knowledge of treatment assignment, open-label trials may have a greater likelihood of biasing outcomes in favour of novel treatments compared to placebo or traditional standards of care. However, as all patients were assigned the same treatment during the extension phase, it is unlikely that biases related to open-label trials affected the performance of patients and analyses of efficacy outcomes. Further, patients remained blinded to initial treatment assignment through the extension phase of the trial. It was unclear whether investigators and study personnel also remained blinded to initial treatment assignment.

All trial outcomes and statistical methods were pre-specified. However, none of the analyses were adjusted for multiplicity. Therefore, the lack of adjustment may increase the likelihood of type I error and of detecting a statistical difference when it may not be present. All results and analyses of the extension period of the FRAME trial should be considered descriptive.

Exposure to denosumab was similar across both the romosozumab/denosumab and placebo/denosumab treatment groups. More than 90% of patients in both the groups received all 4 doses of denosumab and had a mean cumulative exposure of 229.2 mg (range = 60 mg to 240 mg) and 230.5 mg (range = 60 mg to 240 mg), respectively.¹² Any differences in treatment exposure of denosumab are therefore not likely to confound overall treatment effects based in initial treatment with romosozumab or placebo.

Important protocol deviations were reported for 122 patients (3.4%) in the romosozumab/denosumab group and 134 patients (3.7%) in the placebo/denosumab group. Important protocol deviations occurred at similar frequency between both treatment groups of the trial and were due to patients receiving the incorrect treatment or dose (2.1% versus 2.2%), entering the trial without satisfying eligibility criteria (0.7% versus 1.0%), receiving a concomitant medication excluded from eligibility criteria (0.3% versus 0.3%), missing data (0.1% versus 0.3%), and developing withdrawal criteria without being withdrawn from the trial (0.3% versus 0.1%).¹² Eligibility criteria of the trial included a list of concomitant medications patients could be administered throughout the trial. Important protocol deviations pertaining to patients who received excluded concomitant medications were reported for 24 patients (12 in each treatment group). As equal numbers of patients in each treatment group received concomitant medications not permitted within the trial, and because the number of patients receiving these medications was so small, it is unlikely that such deviations had a substantial impact on any efficacy or safety analyses.¹²

Data on BTMs were reported as part of the substudy of the FRAME extension. Results of BTMs in the extension phase of the FRAME trial showed a similar mean percent change from baseline in P1NP and sclerostin levels. The sample sizes for the BTM substudy were small (N = 62 for each treatment group), and the substudy was not powered to detect differences between treatment groups. In addition, the analyses for BTMs were not adjusted for multiplicity, increasing the likelihood of type I error.

Health Canada and the FDA have included safety warnings related to cardiovascular AEs (potential risk of myocardial infarction, stroke, and cardiovascular death) in the respective product monograph and product label for romosozumab. Analyses were conducted to determine the differences in odds of cardiovascular events occurring between treatment groups; however, some of the sample sizes for specific cardiovascular AEs were small (and are not reported here). The analyses for detecting differences in the odds of a cardiovascular event between the romosozumab/denosumab and placebo/denosumab groups was not powered or adjusted for multiplicity. Differences in the odds of cardiovascular AEs occurring between patients initially treated with romosozumab and placebo should be interpreted with caution as the study was not adequately designed to evaluate the excess risk of cardiovascular events in the treatment groups and comparisons have not been controlled for multiple comparisons.

External Validity

Patient baseline demographic and clinical characteristics were generally balanced across both treatment groups. Baseline characteristics of patients during the extension phase of the trial remained similar to baseline characteristics at the beginning of the FRAME trial. Therefore, as the characteristics of patients remained stable throughout the 36-month length of the trial, it is unlikely that imbalances either between treatment groups or between patients analyzed at different time points in the trial resulted in any confounding of results. Clinical experts also confirmed that patients in the FRAME trial would be considered candidates for treatment with romosozumab.

The extension phase of the FRAME trial provided longer-term efficacy and safety data pertaining to initial treatment with either romosozumab or placebo, and long-term use of denosumab. Denosumab may be administered to patients for as long as it remains effective.³¹ However, romosozumab is recommended for use once a month for a maximum of 12 months.⁸ Therefore, this trial provides insight into the long-term effects of initial treatment with romosozumab followed by treatment with an antiresorptive agent, such as denosumab. This long-term data can help inform patients and physicians about the long-term effects (i.e., 36 months) of treatment with romosozumab followed by denosumab. Longer-term data (i.e., ≥ 10 years) may be more useful for patients and clinicians for determining long-term fracture risks after treatment with romosozumab. Differences in fracture incidence between patients initially treated with romosozumab and placebo was an end point of the extension phase of the FRAME study; results indicated a greater risk reduction in all analyzed sites among patients initially treated with romosozumab compared to patients initially treated with placebo.

Discussion

Summary of Available Evidence

Two phase III studies (FRAME, N = 7,180; ARCH, N = 4,093) submitted by the sponsor are included in this systematic review. The trials enrolled postmenopausal women (55 to 90 years of age) with osteoporosis.

The FRAME study was a double-blind, placebo-controlled RCT that assessed the efficacy and safety of romosozumab for the treatment of osteoporosis in postmenopausal women. Eligible patients were randomized to receive romosozumab 210 mg subcutaneously or placebo once a month for 12 months. After the 12-month double-blind treatment period, both groups received open-label denosumab 60 mg every 6 months for an additional 12 months. After the first 24-month treatment, patients entered a 12-month open-label extension period (currently ongoing), during which they continued to receive denosumab 60 mg every 6 months. The co-primary efficacy end points were incidence of new vertebral fractures at month 12 and at month 24.

The ARCH study was a double-blind, active-controlled RCT that assessed the efficacy and safety of romosozumab for the treatment of osteoporosis in postmenopausal women with a high risk of fracture. Eligible patients were randomized to receive romosozumab 210 mg subcutaneously or oral alendronate 70 mg for 12 months. After the initial 12-month double-blind alendronate-controlled study period, both groups received open-label alendronate therapy 70 mg once a week for an additional 12 months. The primary efficacy end points in ARCH were the incidence of new vertebral fractures at month 24 and the incidence of clinical fractures (nonvertebral fractures and clinical vertebral fractures) during the primary analysis period, which refers to randomization to the time point that clinical fractures were confirmed for at least 330 patients, and at which point patients have had the opportunity to complete the month 24 study visit.

A limitation of the direct evidence provided by the 2 trials is a lack of comparative evidence between romosozumab and other active treatments, such as another anabolic drug. One sponsor-submitted ITC evaluated the relative clinical efficacy of romosozumab to other active treatments such as teriparatide, denosumab, raloxifene, zoledronate, risedronate, and abaloparatide in postmenopausal women with primary osteoporosis or osteopenia at risk for developing fragility fractures.

Interpretation of Results

Efficacy

Outcomes of fracture are relevant in clinical trials of osteoporosis. They were also identified by the clinicians and patients as important clinical outcomes. In FRAME, the risks of new vertebral fractures measured at the end of 1 year and 2 years of treatment were the primary efficacy end points. Treatment with romosozumab was associated with a statistically significant relative risk reduction in new vertebral fractures at month 12 and at month 24 compared to placebo. According to the clinical expert consulted by CADTH, the benefit gained in reduction in the risk of new vertebral fracture is likely to be clinically meaningful. Results on a number of fracture-related outcomes evaluated as secondary end points (nonvertebral fractures, major nonvertebral fractures, new or worsening vertebral fractures, hip fractures, major osteoporotic fractures, and multiple new/worsening vertebral fractures) favoured romosozumab, with fewer patients in the romosozumab group developing these fractures compared to patients receiving placebo. The difference in the risk of clinical fractures was shown to be statistically significant for romosozumab compared to placebo at months 12 and 24. Results were not statistically significant for comparisons of nonvertebral fractures, and formal testing was not conducted for the remaining end points as the statistical testing sequence was stopped. The lack of statistical differences between romosozumab and placebo, according to the clinical study report for the FRAME study, may be partly attributed to a higher percentage of the FRAME study population coming from Central or South America. A statistically significant treatment-by-region interaction for the analysis of nonvertebral fractures was reported through month 12. The Central or South America subgroup made up 43.0% of the randomized population in the study, but the frequency of nonvertebral fractures in this population was lower than expected. The observed frequency of nonvertebral fractures in the first 12 months in the placebo group was 1.2%, and in the romosozumab group it was 1.5%, while the expected nonvertebral fracture frequency was 3.5% (used for sample-size calculations). In comparison with the rest of the world, the nonvertebral fracture frequency was 2.7% in the placebo group and 1.6% in the romosozumab group (post hoc relative risk reduction = 42%; 95% CI, 11 to 63; P = 0.012). Although subgroup analyses by world region were pre-specified, the analyses to explore the impact of the Central or South America results versus the rest-of-world results were post hoc.

In general, the results from the FRAME study are difficult to interpret, particularly the sequencing of romosozumab followed by denosumab, because of the placebo comparator and differences between groups in the study. This point was raised in the Health Canada reviewer’s report, which stated, “The comparison of romosozumab/denosumab to placebo/denosumab at month 24 is not a valid comparison given that patients in the placebo/denosumab treatment group have only been exposed to one year of active treatment versus the romosozumab/denosumab treatment group who have been exposed to 2 years of active treatment.”³⁷ As well, it was noted previously that patients with characteristics of those at higher risk of fracture were excluded from the FRAME study. As a result, there is a high degree of uncertainty that the results from FRAME at month 24 are interpretable or representative of the expected effects in the population targeted for treatment with romosozumab.

In the ARCH study, the risk of new vertebral fractures at month 24 and the risk of clinical fractures through the primary analysis study period were co-primary efficacy end points. Treatment with 1 year of romosozumab followed by alendronate therapy for another year was associated with a statistically significantly reduced risk of new vertebral fractures through month 24, compared with treatment with alendronate for 2 years. Romosozumab was also associated with a statistically significantly reduced risk of clinical fractures through the primary analysis study period. The clinical expert indicated that the benefit gained in reduction in the risk of new vertebral fractures and clinical fractures is likely clinically meaningful. Results on other fracture-related outcomes in this study (nonvertebral fractures, new vertebral fractures, clinical fractures, hip fractures, major nonvertebral fractures, major osteoporotic fractures, and all osteoporotic fractures) also favoured romosozumab over alendronate. The comparison for nonvertebral fractures was statistically significant in favour of romosozumab. Firm conclusions cannot be drawn for all other fracture end points as they were not adjusted for multiple comparisons. As with FRAME, the results of this study are somewhat difficult to interpret because of the design of the study, which used a relevant treatment as a comparator. However, it was difficult to assess the assay sensitivity of the trial because it lacked a third group treated with placebo, which would be particularly important because alendronate has a different mechanism of action and it may take 2 to 3 years for the fracture preventive benefits to be detectable. Nonetheless, the treatment groups were likely more comparable than in the FRAME study and the enrolled patient population is more representative of the target patient population. As well, the results are biologically plausible as, according to the clinical expert consulted by CADTH, the preferred approach to treating patients with osteoporosis at risk of fracture is to enhance bone building (using a drug such as romosozumab) then prevent bone degeneration (using an antiresorptive or antiremodelling drug). The results of the ARCH study therefore appear to represent improved fracture risk reduction with romosozumab/alendronate versus alendronate/alendronate, although the exact magnitude of the treatment effect is unclear.

Among the important clinical outcomes by clinicians and patient group was HRQoL, which was an exploratory outcome in both FRAME and ARCH. It was evaluated using the generic EQ-5D-5L and a disease-specific questionnaire, OPAQ-SV. Results of the 2 studies did not show consistent or clinically meaningful changes in any of these tools between romosozumab and the comparators. A vertebral fracture is the most common clinical manifestation of osteoporosis. Among them, approximately 2-thirds of these fractures are asymptomatic. This could explain why a deterioration or improvement in symptoms and quality of life may not be easily detected, and therefore why a change in HRQoL may not be observed.

Overall, the potential benefit of romosozumab on HRQoL remains unknown. The relationship between the gains from reduced fracture risk and improvement in patient’s HRQoL was unclear.

The incidence of fatal events was similar between romosozumab and placebo in the FRAME study, and between romosozumab and alendronate in the ARCH study, during the 2-year study period.

The change in BMD from baseline was measured at the lumbar spine, total hip, and femoral neck in FRAME and ARCH. In the ARCH study, treatment with romosozumab was associated with a statistically significantly increase in BMD from baseline at all 3 sites, compared to alendronate at both months 12 and 24. Results from the FRAME study showed similar trends in BMD when comparing romosozumab to placebo, although these comparisons did not account for multiple comparisons. Additionally, the same concerns regarding the similarity of the treatment groups in FRAME for making comparisons is highly uncertain. According to the clinical expert, the between-group differences in the ARCH trial are clinically meaningful. These results were consistent with the change in incidence of fracture in the study population.

Longer-term data from the FRAME extension study provided insight regarding the effects of initial treatment with romosozumab followed by treatment with an antiresorptive agent beyond the initial 24-month treatment period. Results of the extension study suggested that, for all fracture types (i.e., new vertebral, clinical, nonvertebral, major nonvertebral, new or worsening vertebral, hip, major osteoporotic, and multiple new or worsening vertebral), greater reductions in the risk of fractures were observed among patients initially treated with romosozumab followed by denosumab (romosozumab/denosumab group) compared to patients initially treated with placebo followed by denosumab (placebo/denosumab group). The gains in BMD were also maintained in the original romosozumab group. There is no control arm in the extension phase of FRAME, and all results and analyses during this period are considered descriptive. Large observational studies and pooled analysis have found that treatment-related BMD changes are strongly associated with fracture reductions across randomized trials of osteoporosis treatments with different mechanisms of action, and support the use of BMD as a surrogate outcome for fracture outcomes.^45-47 The use of BTMs may complement measurements of BMD in the management of osteoporosis, particularly among patients who have been or are currently taking therapies that are antiresorptive or result in bone formation. However, the correlation between BTMs and the occurrence of osteoporotic fractures may not be as strong as BMD.

The direct evidence comparing romosozumab to existing treatments in postmenopausal women with a high risk of fracture is limited to the results from the ARCH study. The sponsor submitted a single ITC to examine the relative clinical efficacy of romosozumab to other active treatments, such as denosumab, raloxifene, alendronate, risedronate, and zoledronate, in postmenopausal women with primary osteoporosis or osteopenia at risk for developing fragility fractures. However, the ITC did not inform the comparative evidence because of key limitations that precluded drawing conclusions from the results of the analysis. Limitations that were identified in the ITC included a lack of reporting of certain patient characteristics that would better inform on the certainty of the indirect evidence, the high risk of bias associated with most studies, evidence of extensive clinical heterogeneity between studies, a lack of appropriate statistical methods to adjust for effect modifiers that could potentially influence the study results, no clear definition for “placebo” in the included studies, and double-counting of hip fractures when analyzing hip and nonvertebral fracture outcomes. There is therefore uncertainty around the ITC that undermines the internal and external validity of this analysis.

Harms

During the 24-month study period, the incidence of AEs was similar between romosozumab (month 12: 78%; month 24: 85%) and placebo (month 12: 80%; month 24: 86%) in FRAME, and between romosozumab (month 12: 76%; primary analysis period: 87%) and alendronate (month 12: 79%; primary analysis period 24: 89%) in ARCH. The common AEs reported in the romosozumab groups were arthralgia, nasopharyngitis, back pain, and pain in extremities. The incidence of SAEs was similar between romosozumab (month 12: 10%; month 24: 16%) and placebo (month 12: 9%; month 24: 15%) in FRAME, and between romosozumab (month 12: 13%; primary analysis period: 29%) and alendronate (month 12: 14%; primary analysis period 24: 30%) in ARCH. In the romosozumab group, pneumonia, chronic obstructive pulmonary disease, osteoarthritis, femur fractures, radius fractures, and femoral neck fractures were considered SAEs. Treatment discontinuation due to AEs were also similar between romosozumab (month 12: 3%; month 24: 3%) and placebo (month 12: 3%; month 24: 3%) in FRAME, and between romosozumab (month 12: 3%; primary analysis period: 7%) and alendronate (month 12: 3%; primary analysis period 24: 7%) in ARCH. In terms of AEs of particular interest for the review, the incidence of hypersensitivity and ONJ were similar between romosozumab and placebo and alendronate. There was a potential signal of an association between cardiovascular AEs with romosozumab, when compared with alendronate in the ARCH study. The rates of cardiovascular serious AEs were similar between groups in FRAME at 12 and 24 months, but higher in patients in the romosozumab group (2.5%) than the alendronate group (1.9%) at month 12 and at month 24 (6.5% versus 6.1%, respectively). Myocardial infarction and stroke occurred at higher rates with romosozumab. The percentages were more similar at 24 months, but patients in both groups had been receiving alendronate alone for 12 months after romosozumab or placebo, and therefore the 24-month time point is likely not the optimal point to assess differences between groups. The potential increased risk of cardiovascular events has been noted by Health Canada in the product monograph and by FDA on the label for romosozumab, with both regulators adding warnings for the potential risk of myocardial infarction, stroke, and cardiovascular death.^8,38 The clinical expert consulted by CADTH also indicated that romosozumab treatment would not be started in patients with a history of myocardial infarction or stroke, and that it would be discontinued in patients who experience a myocardial infarction or stroke.

The balance between longer-term benefits and risks with romosozumab is also unclear. Longer-term data from the FRAME Extension study suggested AEs occurred at similar rates across both treatment groups (88% in the romosozumab/denosumab group and 89% in the placebo/denosumab group). All AEs of special interest were infrequently reported (< 4%), except for AEs, potentially related to hypersensitivity. Cardiovascular-related AEs were generally reported at similar rates in each treatment group and were rare (< 1%). However, the design of the study (all patients had been receiving denosumab) makes these data difficult to interpret, leading to increased uncertainty about the longer-term safety of romosozumab treatment.

The sponsor-submitted ITC did not examine the comparative safety of romosozumab versus other medications for the treatment of osteoporosis. The only comparative data between romosozumab and other active treatments for osteoporosis available therefore come from the comparison with alendronate in the ARCH study. Given input from patients regarding the importance of treatment options with fewer adverse effects and better tolerability, there are limited data to determine whether romosozumab fills this need, particularly considering the observed higher frequency of cardiovascular events with romosozumab versus alendronate.

Conclusions

Two phase III double-blind RCTs — 1 placebo-controlled (FRAME) and 1 active-controlled (ARCH) — provided evidence supporting the efficacy and safety of romosozumab for the treatment of osteoporosis in postmenopausal women with a high risk of fracture. Compared to alendronate or placebo, patients who were treated with a romosozumab subcutaneous injection 210 mg once a month showed benefits in reducing the risk of new fracture and increasing BMD. Changes in the incidence of new vertebral fracture and clinical fracture at month 12 and month 24 were considered statistically and clinically relevant. Changes in BMD from baseline to month 12 and month 24 as reported in the ARCH study were also statistically and clinically meaningful. However, whether treatment with romosozumab is associated with any HRQoL benefit remains uncertain. The incidence of AEs, SAEs, and treatment discontinuation due to AEs were similar across treatment groups in both the FRAME and ARCH studies. The risk of cardiovascular-related AEs was comparable between romosozumab therapy and placebo, and between romosozumab therapy and alendronate therapy.

Little direct evidence is available on the relative efficacy and safety of romosozumab versus current treatments for the target population. One sponsor-submitted ITC provided indirect data on the relative clinical efficacy of romosozumab to other active treatments. The results suggest that romosozumab therapy may have a beneficial effect in reducing the risk of sustaining nonvertebral fractures compared to some current treatments. Results of this ITC are associated with a substantial risk of bias due to limitations such as extensive heterogeneity that have not been adequately accounted for.

The longer-term efficacy and safety of romosozumab were evaluated in the extension phase of the FRAME study, in which all study participants received denosumab. This extension study provided insight regarding the long-term effects (3 years) of initial treatment with romosozumab followed by treatment with an antiresorptive agent. Findings of this extension study suggest that the treatment effects from romosozumab in reducing the risk of fracture and increasing BMD were maintained. The frequency of AEs was similar between patients who originally received romosozumab or placebo. However, limitations of this extension study, such as the lack of a comparator group and a lack of patients from a high-risk population, contribute uncertainty to the results.

References

1.Rosen HN, Drezner MK. Clinical manifestations, diagnosis, and evaluation ofosteoporosis in postmenopausal women. In: Post TW, ed. UpToDate. Waltham (MA): UpToDate; 2020: www.uptodate.com. Accessed 2021 Mar 01.

2.Allen S, Forney-Gorman A, Homan M, Kearns A, Kramlinger A, Sauer M. Diagnosis and treatment of osteoporosis. (Health care guideline). Bloomington (MN): Institute for Clinical Systems Improvement; 2017: https://www.icsi.org/wp-content/uploads/2019/01/Osteo.pdf. Accessed 2021 Jul 14.

3.Simpson EL, Martyn-St James M, Hamilton J, et al. Clinical effectiveness of denosumab, raloxifene, romosozumab, and teriparatide for the prevention of osteoporotic fragility fractures: a systematic review and network meta-analysis. Bone. 2020;130:115081. PubMed

4.Osteoporosis: impact report 2017-2018. Toronto (ON): Osteoporosis Canada; 2018: https://osteoporosis.ca/wp-content/uploads/EN-Osteo-Impact-Report-F2017-2018.pdf. Accessed 2021 Jul 14.

5.Lewiecki EM. Bisphosphonates for the treatment of osteoporosis: insights for clinicians. Ther Adv Chronic Dis. 2010;1(3):115-128. PubMed

6.Modi A, Fan CS, Tang J, Weaver JP, Sajjan S. Association of gastrointestinal events with osteoporosis treatment initiation and treatment compliance in Germany: an observational study. Bone Rep. 2016;5:208-213. PubMed

7.Siris ES, Fan CS, Yang X, Sajjan S, Sen SS, Modi A. Association between gastrointestinal events and compliance with osteoporosis therapy. Bone Rep. 2016;4:5-10. PubMed

8.Evenity (romosozumab): 105 mg/1.17 mL solution for subcutaneous injection with single-use prefilled syringe [product monograph]. Mississauga (ON): Amgen Canada Inc.; 2019 Jun 17.

9.Clinical Study Report: 20070337. A multicenter, international, randomized, double-blind, placebo-controlled, parallel-group study to assess the efficacy and safety of romosozumab treatment in postmenopausal women with osteoporosis [internal sponsor's report]. Thousand Oaks (CA): Amgen Inc.; 2016 Jun 03.

10.Clinical Study Report: 20110142. A multicenter, international, randomized, double-blind, alendronate-controlled study to determine the efficacy and safety of romosozumab in the treatment of postmenopausal women with osteoporosis [internal sponsor's report]. Thousand Oaks (CA): Amgen Inc.; 2017 Jul 26.

11.Barrionuevo P, Kapoor E, Asi N, et al. Efficacy of pharmacological therapies for the prevention of fractures in postmenopausal women: a network meta-analysis. J Clin Endocrinol Metab. 2019;104(5):1623-1630. PubMed

12.Supplemental Clinical Study Report: 20070337. A multicenter, international, randomized, double-blind, placebo-controlled, parallel-group Study to assess the efficacy and safety of romosozumab treatment in postmenopausal women with osteoporosis: 36-month final analysis [internal sponsor's report]. Thousand Oaks (CA): Amgen Inc.; 2017 May 29.

13.Braithwaite RS, Col NF, Wong JB. Estimating hip fracture morbidity, mortality and costs. J Am Geriatr Soc. 2003;51(3):364-370. PubMed

14.Osteoporosis Canada. Osteoporosis facts and statistics. 2021; https://osteoporosis.ca/fast-facts/. Accessed 2021 Aug 13.

15.Wells GA, Cranney A, Peterson J, et al. Alendronate for the primary and secondary prevention of osteoporotic fractures in postmenopausal women. Cochrane Database Syst Rev. 2008;1:CD001155. PubMed

16.Compston JE, Drake MT. Defining very high fracture risk: is FRAX fit for purpose? J Bone Miner Res. 2020;35(8):1399-1403. PubMed

17.van Geel TA, van Helden S, Geusens PP, Winkens B, Dinant GJ. Clinical subsequent fractures cluster in time after first fractures. Ann Rheum Dis. 2009;68(1):99-102. PubMed

18.Tarride JE, Hopkins RB, Leslie WD, et al. The burden of illness of osteoporosis in Canada. Osteoporos Int. 2012;23(11):2591-2600. PubMed

19.Siminoski K, Leslie WD, Frame H, et al. Recommendations for bone mineral density reporting in Canada. Can Assoc Radiol J. 2005;56(3):178-188. PubMed

20.Centre for Metabolic Bone Diseases UoS, UK. FRAX: Fracture Risk Assessment Tool. 2008; https://www.sheffield.ac.uk/FRAX/. Accessed 2021 Jul 21.

21.Fujiwara S, Nakamura T, Orimo H, et al. Development and application of a Japanese model of the WHO fracture risk assessment tool (FRAX). Osteoporos Int. 2008;19(4):429-435. PubMed

22.Papaioannou A, Morin S, Cheung AM, et al. 2010 clinical practice guidelines for the diagnosis and management of osteoporosis in Canada: summary. CMAJ. 2010;182(17):1864-1873. PubMed

23.Rosen HN, Drezner MK. Overview of the management of osteoporosis in postmenopausal women. In: Post TW, ed. UpToDate. Waltham (MA): UpToDate; 2020: www.uptodate.com. Accessed 2021 Mar 01.

24.Camacho PM, Petak SM, Binkley N, et al. American Association of Clinical Endocrinologists and American College of Endocrinology clinical practice guidelines for the diagnosis and treatment of postmenopausal osteoporosis - 2016--executive summary. Endocr Pract. 2016;22(9):1111-1118. PubMed

25.Tabacco G, Bilezikian JP. Osteoanabolic and dual action drugs. Br J Clin Pharmacol. 2019;85(6):1084-1094. PubMed

26.Cadarette SM, Carney G, Baek D, Gunraj N, Paterson JM, Dormuth CR. Osteoporosis medication prescribing in British Columbia and Ontario: impact of public drug coverage. Osteoporos Int. 2012;23(4):1475-1480. PubMed

27.Wysowski DK, Greene P. Trends in osteoporosis treatment with oral and intravenous bisphosphonates in the United States, 2002-2012. Bone. 2013;57(2):423-428. PubMed

28.Brown JP, Morin S, Leslie W, et al. Bisphosphonates for treatment of osteoporosis: expected benefits, potential harms, and drug holidays. Can Fam Physician. 2014;60(4):324-333. PubMed

29.Barrionuevo P, Gionfriddo MR, Castaneda-Guarderas A, et al. Women's values and preferences regarding osteoporosis treatments: a systematic review. J Clin Endocrinol Metab. 2019;104(5):1631-1636. PubMed

30.Fosamax (alendronate sodium): 70 mg oral tablets [product monograph]. Kirkland (QC): Merck Canada Inc.; 2017 May 26: https://www.merck.ca/static/pdf/FOSAMAX-PM_E.pdf?bcs-agent-scanner=e73f1b8a-f1fc-114d-ae9b-ad952724ebf1. Accessed 2021 Jul 12.

31.Prolia (denosumab injection): 60 mg/mL solution for injection in a prefilled syringe [product monograph]. Mississauga (ON): Amgen Canada Inc.; 2020 Dec 09: https://www.amgen.ca/~/media/1e79aee7d94340df88c3d97f5bb897c3.ashx?bcs-agent-scanner=84821fa9-e09a-2747-af02-c9fd21351f20. Accessed 2021 Sep 01.

32.McGowan J, Sampson M, Salzwedel DM, Cogo E, Foerster V, Lefebvre C. PRESS Peer Review of Electronic Search Strategies: 2015 guideline statement. J Clin Epidemiol. 2016;75:40-46. PubMed

33.Grey matters: a practical tool for searching health-related grey literature. Ottawa (ON): CADTH; 2019: https://www.cadth.ca/grey-matters. Accessed 2021 Feb 26.

34.Cosman F, Crittenden DB, Adachi JD, et al. Romosozumab treatment in postmenopausal women with osteoporosis. N Engl J Med. 2016;375(16):1532-1543. PubMed

35.Saag KG, Petersen J, Brandi ML, et al. Romosozumab or alendronate for fracture prevention in women with osteoporosis. N Engl J Med. 2017;377(15):1417-1427. PubMed

36.Drug Reimbursement Review sponsor submission: Evenity (romosozumab), 105 mg/1.17 mL solution for subcutaneous injection with single-use prefilled syringe [internal sponsor's package]. Mississauga (ON): Amgen Canada Inc.; 2021 Feb 25.

37.Health Canada reviewer's report: Evenity (romosozumab) [internal sponsor's report]. Ottawa (ON): Therapeutics Products Directorate, Health Canada; 2019.

38.Center for Drug Evaluation Research. Multidiscipline review(s). Evenity (romosozumab) subcutaneous injection with single-use prefilled syringe. Company: Amgen Inc. Application No.: 761062. Approval date: 04/09/2019 (FDA approval package). Silver Spring (MD): U.S. Food and Drug Administration (FDA); 2019: https://www.accessdata.fda.gov/drugsatfda_docs/nda/2019/761062Orig1s000TOC.cfm. Accessed 2021 Jul 29.

39.Committee for Medicinal Products for Human Use. Guideline on the evaluation of medicinal products in the treatment of primary osteoporosis. (European public assessment report). London (GB): European Medicines Agency; 2006: https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-evaluation-medicinal-products-treatment-primary-osteoporosis_en.pdf. Accessed 2021 Jul 29.

40.Osteoporosis: nonclinical evaluation of drugs intended for treatment: draft guidance for industry Silver Spring (MD): U.S. Food and Drug Administration (FDA); 2016: https://www.fda.gov/files/drugs/published/Osteoporosis---Nonclinical-Evaluation-of-Drugs-Intended-for-Treatment-Guidance-for-Industry.pdf. Accessed 2021 Jul 29.

41.DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177-188. PubMed

42.Higgins JP, Jackson D, Barrett JK, Lu G, Ades AE, White IR. Consistency and inconsistency in network meta-analysis: concepts and models for multi-arm studies. Res Synth Methods. 2012;3(2):98-110. PubMed

43.White IR, Barrett JK, Jackson D, Higgins JP. Consistency and inconsistency in network meta-analysis: model estimation using multivariate meta-regression. Res Synth Methods. 2012;3(2):111-125. PubMed

44.Moreno PB, Kapoor E, Asi N, et al. Data from: Barrionuevo P, Kapoor E, Asi N, et al. Efficacy of pharmacological therapies for the prevention of fractures in postmenopausal women: a network meta-analysis. J Clin Endocrinol Metab. 2019;104(5):1623-1630. Figshare. 2019 Jan 25. 10.6084/m9.figshare.7629344. Accessed 2021 Jul 16.

45.Black DM, Cauley JA, Wagman R, et al. The ability of a single BMD and fracture history assessment to predict fracture over 25 years in postmenopausal women: the study of osteoporotic fractures. J Bone Miner Res. 2018;33(3):389-395. PubMed

46.Black DM, Bauer DC, Vittinghoff E, et al. Treatment-related changes in bone mineral density as a surrogate biomarker for fracture risk reduction: meta-regression analyses of individual patient data from multiple randomised controlled trials. Lancet Diabetes Endocrinol. 2020;8(8):672-682. PubMed

47.Lee JH, Lee YH, Moon SH, Group TOPS. Association between bone mineral density and clinical consequences: cross-sectional study of Korean postmenopausal women in an orthopaedic outpatient clinic. J Korean Med Sci. 2014;29(8):1152-1160. PubMed

48.Alay I, Kaya C, Cengiz H, Yildiz S, Ekin M, Yasar L. The relation of body mass index, menopausal symptoms, and lipid profile with bone mineral density in postmenopausal women. Taiwan J Obstet Gynecol. 2020;59(1):61-66. PubMed

49.Ilic Stojanovic O, Vuceljic M, Lazovic M, et al. Bone mineral density at different sites and vertebral fractures in Serbian postmenopausal women. Climacteric. 2017;20(1):37-43. PubMed