80.1

Introduction

Sclerostin is an osteocyte glycoprotein that was originally discovered in the rare genetic diseases of sclerosteosis and Van Buchem disease, both characterized by the absence of the sclerostin (SOST) gene, and associated with very high bone mass and resistance to fracture . Sclerostin plays a role in transmitting effects of mechanical load on bone; with reduced mechanical loading, osteocyte sclerostin secretion increases. Secreted sclerostin travels through the dendritic processes of the osteocyte where it ultimately binds to the LRP5/6 receptor on osteoblasts, lining cells and osteoblast precursor cells, and prevents the association of LRP5/6 with the Wnt/Frizzled coreceptor complex. This leads to the degradation of cytoplasmic beta-catenin and ultimately inhibits Wnt signaling . Sclerostin therefore reduces the production, activity and survival of osteoblasts, and results in reduced bone formation. Sclerostin might also have a role in limiting the wall width of bone packets formed in each remodeling unit. Sclerostin also has an autocrine function on neighboring osteocytes to stimulate the production of RANK ligand and reduce the production of osteoprotegerin . This leads to an increase in the formation and function of osteoclasts, ultimately resulting in increased bone resorption .

Romosozumab is a humanized monoclonal antibody that binds and inhibits sclerostin activity, resulting in increased conversion of bone lining cells to osteoblasts, increased differentiation of osteoblast precursor cells to osteoblasts, and increased activity of mature osteoblasts. The result is a substantial increase in rate and extent of bone formation. Romosozumab also inhibits bone resorption by reducing RANK ligand production and increasing osteoprotegerin secretion, resulting in inhibition of osteoclast formation and activity. With romosozumab, bone formation occurs primarily on quiescent surfaces and is largely modeling based , in contrast to the formation seen with proremodeling agents such as teriparatide and, perhaps to a lesser extent, abaloparatide, which stimulate more remodeling-based bone formation.

Lifelong rodent studies did not reveal any evidence of carcinogenicity, including development of osteosarcoma, with romosozumab .

80.2

Early clinical studies

In Phase 1 human studies, single and then escalating doses of romosozumab were well tolerated and resulted in increased biochemical markers of bone formation and reduced markers of bone resorption with marked rapid increases in bone mineral density (BMD) .

In the Phase 2 clinical trial, 419 postmenopausal women (aged 55–85) with low bone mass or osteoporosis (T-score at spine, total hip, or femoral neck ≤−2 and ≥−3.5) were randomized to one of five different romosozumab doses versus placebo, alendronate or teriparatide . BMD increased significantly with all romosozumab doses, but increases were largest with the highest dose of 210 mg subcutaneously once per month (with 1 year of treatment, 11.3% in the spine, 4.1% in the total hip, and 3.7% in the femoral neck). BMD of the radius declined slightly with romosozumab, similarly to the decline seen with placebo. With the 210 monthly romosozumab dose, spine and hip BMD increased significantly more than seen with either alendronate or teriparatide . Volumetric BMD measurements of the integral spine and hip by quantitative computerized tomography (QCT) and estimated bone strength of both spine and hip by finite element modeling were all larger with romosozumab 210 mg compared to teriparatide.

During the second year of this Phase 2 study, BMD continued to increase with romosozumab, though the rate of BMD gain was about half that seen with the first year of treatment. In a third-year extension to this same study, patients were randomized to receive either denosumab or placebo. In the denosumab group, BMD continued to increase, at a rate similar to that seen during the second year of romosozumab, but in the placebo group, BMD declined as expected. These observations led to the regimen taken forward into the Phase 3 studies (210 once monthly for 1 year).

In a fourth-year extension to this trial, after a year of placebo or denosumab, patients received retreatment with romosozumab for 1 year. In the placebo group, retreatment with romosozumab resulted in spine and total hip BMD gains that were very similar to those seen with the first year of romosozumab treatment. In the denosumab group, romosozumab retreatment produced an increase in spine BMD and stability in hip BMD . This latter response compares favorably with the effect of denosumab transition to teriparatide, where rapid loss of hip BMD is observed . A fifth-year extension to the Phase 2 study is evaluating the use of intravenous zoledronic acid to maintain the BMD increases seen with romosozumab.

A second Phase 2 study performed in 252 Japanese women with osteoporosis, mean age 68, tested three different monthly doses of romosozumab versus placebo . Bone formation marker levels peaked at month 1 and were below placebo levels at the end of the year. BMD increases were greatest with the 210 mg monthly dose: in the spine 16.9%, total hip 4.7%, and femoral neck 3.8% above baseline.

80.3

Pivotal fracture trials: FRAME

The FRAME trial enrolled 7180 postmenopausal women between the ages of 55 and 90 with osteoporosis defined by T-score ≤−2.5 at the spine, hip, or femoral neck . Women with T-scores below −3.5 or with a severe vertebral or multiple moderate vertebral fractures or a history of hip fracture were excluded, to avoid the risk of treating these more severe patients with placebo. Participants were randomized to receive once monthly injections of romosozumab 210 mg subcutaneously versus placebo in a double-blind fashion for 1 year followed by transition to denosumab for 1 year in both groups. Denosumab was continued for a second year during the FRAME study extension. Patients received supplemental calcium and vitamin D, with doses determined by each investigator. Clinical fractures were assessed throughout the study and spine X-rays performed at 6, 12, and 24 months for assessment of vertebral fractures. BMD was measured annually in all patients and every 6 months in a subset. Bone turnover marker levels were also measured in a subset of patients.

A fixed statistical testing sequence was preplanned. First tested were the coprimary end points, group differences in new vertebral fracture incidence (radiographically diagnosed) at 12 and 24 months. The next end point tested was clinical fracture occurrence at 12 months. Clinical fractures were defined as nonvertebral fractures and clinically symptomatic vertebral fractures presenting with acute back pain (only about 20% of the total number of vertebral fractures). More than 85% of the clinical fractures were nonvertebral fractures, which excluded fractures due to high trauma and those of the fingers, toes, face, and skull. The third end points in the testing sequence were nonvertebral fracture occurrence at 12 and at 24 months, which were tested in parallel. Other fracture end points were tested subsequently.

At baseline, the groups were well matched with a mean age of 71, and about a third of patients were 75 and older. A large proportion of participants (40%) were from Latin America, specifically Colombia. Mean bone densities were all in the osteoporosis range, consistent with enrollment criteria. Fewer than 22% of subjects had a history of nonvertebral fracture and just below 19% had a radiographically confirmed vertebral fracture, most of which were mild. About 90% of participants completed 1 year and just under 85% completed the 2-year study.

The biochemical marker of bone formation (PINP) increased rapidly with peak levels at 2 weeks and returned to baseline within 9 months. The level of the CTX bone resorption marker declined within 2 weeks and stayed below baseline for the duration of the study. When tested at the 3- and 6-month time points, transient increases in PINP and decreases in CTX levels were seen, though the magnitude of these changes was greatest after the first month of romosozumab treatment. Bone density increases were rapid and substantial during the year of romosozumab treatment, with no change in the placebo group. With denosumab, BMD increased in both groups. After the 2-year sequence of romosozumab followed by denosumab, mean BMD increased 18% in the spine, 9% in the total hip, and almost 7% in the femoral neck.

Over 1 year, 1.8% of patients randomized to placebo had a new vertebral fracture, compared to 0.5% of patients randomized to romosozumab, a 73% risk reduction in romosozumab treated women ( P <.001). For the coprimary end point of new vertebral fracture at 24 months, after all women from both groups had received denosumab for a year, 2.5% of patients who were on placebo first had a new vertebral fracture, compared to only 0.6% of patients who received romosozumab first, a 75% risk reduction ( P <.001). There were very few new vertebral fractures in year 2 in patients assigned to romosozumab followed by denosumab (5 patients in the group that started with romosozumab had a new vertebral fracture in year 2 vs 25 in the group that started with placebo). In the placebo group, 2.5% of participants had clinical fractures, compared to 1.6% of the romosozumab group, a 36% risk reduction ( P =.008).

At 12 months, nonvertebral fractures had occurred in 2.1% of placebo-treated participants and 1.6% of romosozumab-treated women ( P =.096). Because this end point was not statistically significant, all other 12-month fracture outcomes tested later in the statistical hierarchy were considered exploratory. In prespecified subgroup analyses a significant geographical region-by-treatment effect was seen . With posthoc exploration, it was determined that the Latin American cohort (40% of the population) had a very low background baseline fracture risk (1.6% over the year) and no apparent reduction in nonvertebral fracture incidence with romosozumab (although the effect on vertebral fracture was preserved). In the rest of the world FRAME population (grouped post hoc), romosozumab reduced nonvertebral fracture risk by 42% ( P =.012). The prior nonvertebral fracture history in Latin America was about half that of the rest of the world (15% in Latin America vs 27% in the rest of the world population). Consistent with that observation, the 10-year major fracture probability estimate by FRAX was 7.3% for Latin America versus 14.5% for the rest of the world. One plausible explanation for these findings is that T-score calculations based on an NHANES reference population were inaccurately low for Colombian Latin American participants due to unique body/bone size features and/or other genetic factors, similar to issues previously determined for Japanese individuals .

Because of the prespecified statistical testing sequence, P values for tests of other fracture end points were all adjusted. The incidence of clinical fracture over 24 months was 33% less common in the romosozumab/denosumab group (nominal P =.002 but adjusted P <.1). Similarly, the 25% reduction in incidence of nonvertebral fracture in the romosozumab/denosumab group was nominally significant ( P <.03) but after adjustment was not ( P <.06). Hip fractures occurred in 22 participants who originally received placebo and only 11 in those who originally received romosozumab (nominal P =.059, adjusted P =.12).

80.4

Pivotal fracture trials: ARCH

In the multinational ARCH study, 4093 women, age range 55–90, with prevalent osteoporotic fracture were enrolled and randomized to receive romosozumab 210 subcutaneously once monthly or alendronate 70 mg once weekly orally for 1 year in a double-blind fashion . Subsequently, all participants received open-label alendronate for the remainder of the study, which had a median treatment period of 2.7 years (33 months). This was an event-driven trial that was stopped after 330 women had fractures and after all participants had completed 24 months. The coprimary end points were new vertebral fractures at 24 months and clinical fractures at the primary analysis (end of study). Secondary end points included incidence of nonvertebral and hip fractures at primary analysis. BMD was assessed yearly and every 6 months in a subset of participants. Biochemical marker levels were assessed in a subset.

Because there was no placebo group in this study and all women were on active therapy throughout the trial, patients with more severe osteoporosis were recruited. In women who had a moderate or severe vertebral fracture or at least two mild vertebral fractures, T-score ≤−2.5 in the total hip or femoral neck was also required. A T-score of ≤−2 was allowed in women with two or more moderate or severe vertebral fractures or a recent hip fracture (within the preceding 2 years). Eighty-nine percent of subjects completed 1 year and 77% completed through the primary analysis.

Baseline characteristics were balanced between groups with a mean age of 74, and 52% of the population was at least 75 years of age. Ninety-six percent of subjects had a prevalent vertebral fracture, of which about 65% were severe and 27% were moderate; 9% of participants had had a hip fracture in the preceding 2 years. Mean T-scores were −3, −2.8, and −2.9 at the spine, total hip, and femoral neck, respectively.

For the coprimary end points, at 24 months in the alendronate group, 11.9% of women had new vertebral fractures, compared with 6.2% in the romosozumab/alendronate group, a 48% relative risk reduction ( P <.001). At the time of primary analysis (33 months) the coprimary end point of clinical fractures occurred in 13% of women on alendronate versus 9.7% of women on romosozumab/alendronate (27% risk reduction; P <.001). Similarly, there was a 19% reduction in nonvertebral fracture ( P <.04) and a 38% reduction in hip fracture risk ( P <.02) in the romosozumab/alendronate group compared to the alendronate alone group. Fracture risk reductions were already apparent at 12 months: new vertebral fracture incidence was reduced by 37%, clinical fractures reduced by 28%, and nonvertebral fractures reduced by 26% ( P =.06).

BMD increments were substantially higher in patients who received romosozumab versus alendronate in year 1. In the romosozumab group the BMD increments were very similar to those seen in FRAME. In ARCH, BMD increased further after transition to alendronate; however, BMD increments at 36 months were not as high as those seen in FRAME when women transitioned from romosozumab to denosumab. In the FRAME study after 1 year of romosozumab and 2 years of denosumab, mean BMD increased 18.1% in the spine and 9.4% in the total hip, whereas in the ARCH study, after 1 year of romosozumab and 2 years of alendronate, mean BMD increased 15% in the spine and 7% in the total hip.

80.5

Other phase 3 studies: STRUCTURE

The STRUCTURE study was a bone imaging end point study that enrolled 436 postmenopausal women with osteoporosis who had all been treated with oral bisphosphonates for at least 3 years, with at least the last year on oral alendronate 70 mg once weekly . Women in STRUCTURE were between 55 and 90 years old, and all had a history of adulthood nonvertebral fracture or radiographically confirmed vertebral fracture. All patients had BMD at spine, hip, or femoral neck in osteoporosis range.

At enrollment, alendronate was stopped and patients were randomized to receive either open-label romosozumab 210 once monthly or teriparatide 20 μg daily for 1 year. The end points of the STRUCTURE study were BMD by DXA and volumetric BMD by QCT, with assessments done at both 6 and 12 months. The main focus was on the hip because prior studies indicated that hip BMD declined in patients switched from bisphosphonates to teriparatide . The QCT data were further analyzed using finite element analysis (FEA) models. The primary end point for this study was the group difference in total hip BMD change (DXA) looking at the mean BMD between 6 and 12 months. Secondary end points included group differences and within-group changes in DXA-based BMD at femoral neck and spine sites at 6 and 12 months, and integral and cortical volumetric BMD by QCT and FEA strength estimates at 6 and 12 months.

Baseline characteristics were well matched between groups. Mean age was just under 72. The mean duration of prior bisphosphonate use and prior alendronate use was 6.2 years and 5.7 years, respectively, so most of the prior bisphosphonate exposure was with alendronate. The groups were also well matched for the various bone density and imaging outcomes. Mean biochemical marker levels were in the low normal premenopausal range, consistent with the use of alendronate at the time of enrollment in the study.

Total hip BMD (DXA) declined slightly with teriparatide (−0.6%), consistent with all of the other studies showing a switch from bisphosphonate treatment to teriparatide . In contrast, in patients who switched to romosozumab, total hip BMD increased 2.6% ( P <.0001). In the FN at 6 and 12 months, respectively, BMD decreased 1.1% and 0.2%, respectively, with teriparatide but increased 2.1% and 3.2%, respectively, with romosozumab. In the spine, BMD increased at both 6 and 12 months with teriparatide, but the increases seen with romosozumab were about twice the magnitude of those with teriparatide at both time points.

Results for integral volumetric BMD (which includes trabecular and cortical compartments) by QCT followed the same pattern as seen with total hip BMD. In the trabecular compartment alone, both teriparatide and romosozumab increased volumetric trabecular BMD, though increments were greater with romosozumab. The biggest difference between teriparatide and romosozumab was seen in the cortical QCT compartment. Cortical BMD declined progressively over the year with teriparatide but increased progressively with romosozumab. Consistent with all of these findings, hip strength by FEA declined with teriparatide (−1% and −0.7% at 6 and 12 months, respectively) but increased with romosozumab (2.1% and 2.5% at 6 and 12 months, respectively). Interestingly, the degree of change in hip strength seen with both agents is incredibly similar to the magnitude of change seen with both agents on total hip BMD by DXA. In fact, the total hip BMD level achieved on medication does appear to be an excellent surrogate for determining the risk of further fractures after or on treatment.

The rates of all AEs and serious AEs were similar between the groups and nasopharyngitis and arthralgias were the most common AEs reported in the romosozumab-treated patients. The most common injection site reaction was pain. Hypercalcemia was far more common with teriparatide, as expected.

80.6

Other phase 3 studies: BRIDGE

The BRIDGE study enrolled 245 men with osteoporosis between the ages of 55 and 90 and randomized participants 2:1 to blinded romosozumab versus placebo for 1 year, with a subsequent 3-month follow-up period . The primary end point was group difference in spine BMD at 1 year, and the overall study concept was to determine if the BMD differences between romosozumab and placebo in men were similar to those in women. Inclusion criteria were T-score at the spine, total hip, or femoral neck ≤−2.5, but above −3.5, or ≤−1.5 with an osteoporosis-related fracture.

At baseline, groups were well matched with mean age 72, and mean T-scores −2.3, −1.9, and −2.3 in the spine, total hip, and femoral neck, respectively. About half the patients had had a prior fracture. At 12 months, mean BMD increases in the romosozumab versus placebo groups were 12.1% versus 1.2% in the spine, 2.5% versus −0.5% in the total hip, and 2.2% versus −0.2% in the femoral neck (all P <.001 group differences). Group differences were seen at 6 months at all sites. AEs and serious AEs were balanced between the groups. There was a slightly higher number of adjudicated serious cardiovascular events in the romosozumab compared with placebo group (8 events or 4.9% vs 2 events or 2.5%).

80.7

Conclusion

Romosozumab stimulates bone formation, suppresses bone resorption, and produces dramatic bone density increases in both spine and hip. Fracture risk reductions compared to both placebo and alendronate are rapid throughout the skeleton. Bone density improvements in both hip and spine are also seen with romosozumab upon transition from alendronate. Questions remain around the significance of cardiovascular safety signals seen in the ARCH and BRIDGE studies. A treatment sequence of romosozumab for 1 year followed by denosumab or other potent antiresorptive therapy is a new treatment option for patients with osteoporosis at high risk for fracture.