Results
Participant disposition through the core study and extensions is shown in Figure 1. Extension II enrolled 1,732 women; 1,301 completed extension II (23% and 85% of dosed women in the core study and in extension II, respectively), and 1,530 received one or more doses of study medication and were included in the safety population. The baseline and demographic characteristics of extension II participants were similar to those of the overall study population, with no significant differences between BZA-treated and PBO-treated women ( Table 1 ).
(Enlarge Image)
Figure 1.
Flow chart of women throughout the core study and extensions. Randomized and received ≥1 does of study drug (n = 7,492) is the sum of the 4 original treatment arms and will not match the sum of the 3 arms in this study. BZA, bazedoxifene; PBO, placebo.Excludes observational substudy women who discontinued during extension II and 31 women who did not enter extension II and had inaccurate termination records.Does not account for all deaths reported to sponsor.Overall P < 0.05.
The mean compliance rate (ie, percentage of scheduled medication capsules consumed) across 7 years was 94.8%, 94.9%, and 95.1% for the BZA, BZA20, and PBO groups, respectively. Compliance for years 6 to 7 was 95.0%, 94.9%, and 95.9%, respectively. During extension II, a significantly (P < 0.01) higher proportion of PBO-treated women (18.6%) used concomitant bone-active nonstudy medications compared with the BZA (11.7%) and BZA20 (11.8%) groups.
Vertebral Fractures
The cumulative rate of new radiographically confirmed vertebral fractures (Kaplan-Meier estimate) at 7 years was significantly lower for the BZA (6.4%) and BZA20 (7.6%) groups than for the PBO group (9.9%; Figure 2A), corresponding to relative risk reductions of 36.5% and 30.4%, respectively (Figure 2B). Estimated cumulative rates of new vertebral fractures were similar when women who received bone-active nonstudy medications (BZA, n = 180; BZA20, n = 90; PBO, n = 128) were excluded (6.5%, 7.7%, and 10.1%, respectively; P < 0.001 for BZA and P < 0.05 for BZA20 vs PBO).
(Enlarge Image)
Figure 2.
Kaplan-Meier estimates of new vertebral fracture rates (A) and cumulative incidence of new vertebral fractures (B) at 7 years in the modified intent-to-treat population. BZA, bazedoxifene; PBO, placebo; RRR, relative risk reduction; HR, hazard ratio. Log rank P < 0.05 for BZA and BZA 20 mg versus PBO.
The Kaplan-Meier cumulative rate estimate of new clinical vertebral fractures was 1.28% (95% CI, 0.82-1.97) in the BZA group and 1.92% (95% CI, 1.20-3.08) in the PBO group across 7 years (P = 0.18). Seven women (0.21%) in the BZA group and 1 woman (0.06%) in the PBO group had worsening vertebral fractures; these low numbers were insufficient for a definitive statistical comparison.
Among women with prevalent vertebral fractures at baseline, BZA significantly reduced the cumulative rate of new vertebral fractures compared with PBO (BZA, 6.9%; BZA20, 8.3%; PBO, 11.4%). The relative risk reductions were 39.0% (HR, 0.61; 95% CI, 0.43-0.86; P = 0.004) for the BZA group and 32.6% (HR, 0.67; 95% CI, 0.45-1.00; P = 0.048) for the BZA20 group. Among women with no prevalent fractures at baseline, the BZA and BZA20 groups had nonsignificant reductions in the incidence of vertebral fractures compared with PBO (5.8%, 6.8%, and 8.2%, respectively), corresponding to PBO (5.8%, 6.8%, and 8.2%, respectively), corresponding to 1.05) and 26.2% (HR, 0.74; 95% CI, 0.44-1.23).
The absolute numbers of women who developed new vertebral fractures during extension II were 20 of 1,032, 14 of 506, and 15 of 535 for the BZA, BZA20, and PBO groups (Kaplan-Meier rate estimates of 2.4%, 3.4%, and 3.4%, respectively). The BZA group showed a 32.6% statistically insignificant relative risk reduction for years 6 to 7 compared with PBO (HR, 0.67; 95% CI, 0.35-1.32); the rate for BZA20 was similar to that for PBO (relative risk reduction, 2.7%; HR, 0.97; 95% CI, 0.47-2.02).
Nonvertebral Fractures
Within the overall safety population, the cumulative incidence of nonvertebral fractures was similar among treatment groups (Kaplan-Meier rate estimates: BZA, 11.2%; BZA20, 12.0%; PBO, 10.8%). In the higher-risk subset (n = 1,324), the cumulative rate of nonvertebral fractures was numerically, but not significantly, lower for the BZA (12.8%; HR, 0.72; 95% CI, 0.48-1.07) and BZA20 (12.8%; HR, 0.68; 95% CI, 0.42-1.10) groups.
Bone Mineral Density
At 7 years, all groups showed significant increases in adjusted mean (SE) percentage changes from baseline in lumbar spine BMD: BZA, 2.95% (0.39%); BZA20, 2.73% (0.51%); PBO, 2.19% (0.49%) (P < 0.001 vs baseline for all; Figure 3A). Increases from baseline to year 7 in lumbar spine BMD were greater for the BZA and BZA20 groups than for PBO, but these differences were not statistically significant. On year 6, the BZA and BZA20 groups also showed greater, but not statistically significant, mean (SE) increases from baseline in lumbar spine BMD compared with PBO: 2.24% (0.31%), 1.92% (0.41%), and 1.80% (0.40%), respectively.
(Enlarge Image)
Figure 3.
Changes in bone mineral density at the lumbar spine (A) and total hip (B) across 7 years. Bone mineral densities at the lumbar spine and total hip were measured by dual-energy x-ray absorptiometry at the end of extension I and on months 72 and 84 during extension II. All dual-energy x-ray absorptiometry measurements were evaluated at a central facility (Synarc, San Francisco, CA). BZA, bazedoxifene; PBO, placebo. For (A) P < 0.01 versus PBO for BZA and BZA 20 mg at all time points, except for months 60, 72, and 84. For (B) P < 0.001 versus PBO for BZA and BZA 20 mg at all time points.
All groups showed statistically significant decreases from baseline in total hip BMD at 7 years: adjusted mean (SE) percentage changes: BZA, −1.15% (0.30%), P < 0.001; BZA20, −1.19% (0.39%), P = 0.002; PBO, −2.53% (0.38%), P < 0.001 (Figure 3B). On year 7, compared with PBO, the BZA and BZA20 groups showed significantly smaller BMD reductions from baseline in total hip (both P < 0.001), femoral neck (both P < 0.001), and femoral trochanter (both P < 0.01) assessments. Similar BMD results were seen at all four sites when data from women receiving bone-active nonstudy medications were excluded.
Safety and Tolerability
BZA treatment up to 7 years was associated with a favorable safety/tolerability profile. The incidences of AEs, serious AEs, and AE-related study discontinuations in the overall safety population were similar between the BZA group and the PBO group ( Table 2 ). The most common AEs (reported by ≥20% of women in at least one group) were abdominal pain, accidental injury, back pain, flu syndrome, headache, infection, pain, hypertension, constipation, and arthralgia. Compared with PBO, the BZA group showed higher incidences of hot flushes (13.3% vs 6.7%; P < 0.001) and leg cramps (14.1% vs 10.8%; P < 0.001) across 7 years. Hot flush incidence was highest during the first year of BZA and steadily decreased (129.4, 28.9, 9.6, 7.8, 10.1, 6.0, and 2.3 per 1,000 woman-years for years 1-7, respectively). Leg cramp incidence was highest during the first year, decreased through year 3, and leveled off (99.8, 40.5, 21.8, 30.8, 20.2, 33.6, and 13.9 per 1,000 woman-years for years 1-7). Similar trends were observed for the incidences of hot flushes and leg cramps with PBO. Safety findings for years 6 to 7 (extension II safety population) were similar to those observed across 7 years. BZA was not associated with increased incidences of ischemic cardiac disorders across 7 years or during extension II ( Table 2 ). There were no significant differences in the incidence of cerebrovascular events (HR, 1.06; 95% CI, 0.62-1.81) between the BZA group and the PBO group. Incidence of VTE was higher in the BZA group compared with the PBO group (HR, 1.62; 95% CI, 0.8-3.3) owing to a higher incidence of deep vein thrombosis (DVT; HR, 3.38; 95% CI, 1.01-11.39; P < 0.05). This finding is similar to that observed at 5 years. The incidence of DVT was highest during the first year (2.9 per 1,000 woman-years), with lower rates in subsequent years (1.4, 1.2, 1.0, and 0.6 for years 2-5) and no new cases during years 6 and 7. The overall HR for VTEs remained similar at 3 years (HR, 1.66; 95% CI, 0.75-3.66), 5 years (HR, 1.57; 95% CI, 0.77-3.21), and 7 years (HR, 1.62; 95% CI, 0.8-3.3).
At 7 years and during extension II, the incidences of breast carcinoma and other breast-related AEs were low and similar between the BZA group and the PBO group. There were no between-group differences in the overall incidence of endometrial hyperplasia and no new cases during years 6 to 7. The overall incidence of endometrial carcinoma was significantly lower in the BZA group than in the PBO group (0.1% and 0.4%, respectively; P = 0.02); no cases of endometrial carcinoma were reported with BZA20 (P = 0.008 vs PBO). During years 6 to 7, there were no new cases of endometrial carcinoma in the BZA group and one case of endometrial carcinoma in the PBO group. There was a numerical, but not statistically significant, difference in the incidence of ovarian carcinoma (BZA, 0.2%; PBO, 0.0%) across 7 years (P = 0.19); no new cases were reported during years 6 and 7.