Methods
Study Design
The SMART-1 trial was a 2-year multicenter, double-blind, randomized, placebo- and active-controlled, phase 3 study that enrolled 3,397 postmenopausal women with an intact uterus. Full details of the study methodology and clinical trial results have been reported elsewhere. In brief, women were randomized to receive daily oral treatment with one of six doses of BZA/CE (BZA 10 mg/CE 0.45 mg, BZA 20 mg/CE 0.45 mg, BZA 40 mg/CE 0.45 mg, BZA 10 mg/CE 0.625 mg, BZA 20 mg/CE 0.625 mg, and BZA 40 mg/CE 0.625 mg), raloxifene 60 mg, or placebo. All participants were supplemented with calcium and vitamin D, as needed, to achieve a daily intake of 1,000 to 1,600 mg and 200 to 400 IU, respectively. The study protocol was approved by an institutional review board or an independent ethics committee at each participating site before the start of the study. This trial was conducted according to the Declaration of Helsinki and in compliance with Good Clinical Practice. Before enrollment, all participants completed a written informed consent form.
Participants
To be eligible for enrollment in the SMART-1 trial, women had to be aged between 40 and 75 years and postmenopausal—defined as having finished the last natural menstrual cycle at least 1 year before screening and having serum follicle-stimulating hormone concentrations of 30 mIU/mL or higher and serum 17β-estradiol concentrations of 183.5 pmol/L or lower. Women were also required to have an intact uterus and acceptable endometrial biopsy readings at screening. Women who had used oral estrogen-, progestin-, androgen-, or selective estrogen receptor modulator-containing medication or any vaginal or transdermal hormone product within 8 weeks of screening were excluded. Women who had used any injectable or pelleted hormone product within 24 weeks of screening were also excluded.
Study Assessments
The primary endpoint of the SMART-1 trial was the incidence of endometrial hyperplasia at 1 year; the secondary endpoint was the percent change in BMD at the lumbar spine and total hip at 2 years. Additional endpoints included evaluations of safety, metabolic parameters, vaginal atrophy, uterine bleeding, vasomotor symptoms, and quality of life.
Lumbar spine and total hip BMD were measured by dual-energy x-ray absorptiometry at screening and at months 6, 12, 18, and 24. Serum levels of the bone markers C-telopeptide (CTx) and osteocalcin (OC) were measured at screening and at months 6, 12, 18, and 24. Participants recorded the total number and severity of hot flushes in their daily diaries.
Statistical Analysis
The primary analysis population for assessment of BMD was the modified intent-to-treat (MITT) population, which included women who took at least one dose of the study drug and had baseline BMD assessment and at least one on-therapy BMD assessment. Hot flush analysis was based on data from symptomatic women who had seven or more moderate to severe hot flushes per day (or ≥50 per week) at screening, baseline BMD assessment, and at least one on-therapy BMD assessment.
In this post hoc exploratory analysis, mean percent changes from baseline in BMD for each treatment group at month 24 were compared using tertiles of baseline serum CTx and OC in the overall population. The analysis was performed using an analysis of covariance model that included treatment and center as factors and baseline BMD and years since menopause as covariates by CTx and OC tertiles. These covariates were chosen because they were found to be significant according to stepwise regression analysis. Because the BMD/bone marker level analysis was performed in the overall population, it used combined data for BZA 20 mg/CE 0.45 mg and BZA 20 mg/CE 0.625 mg (doses intended for clinical development).
The hot flush composite scores for women in the moderateto-severe-hot-flush population were calculated using the following formula adapted from Jin et al:
Hot flush composite score =
(number of mild hot flushes per day x 1) +
(number of moderate hot flushes per day x 2) +
(number of severe hot flushes per day x 3)
Pearson's partial correlations between mean percent changes from baseline in BMD at the lumbar spine and total hip at month 24 and mean percent changes from baseline in hot flush composite score at week 12, while controlling for the effects of baseline BMD and baseline hot flush score, were determined in the population that had moderate to severe hot flushes. These time points were chosen to see whether the early response in hot flushes to treatment could predict later changes in BMD. Because they were performed in the subset with moderate to severe hot flushes, these analyses used combined data for all doses of BZA/CE to increase the available data points.
In addition, a linear multivariate regression (stepwise model selection process with a 0.15 significance level as selection criterion) was performed to investigate the relationship between baseline hot flush composite score and other baseline characteristics (including age, years since menopause, smoking, race, weight, body mass index [BMI], and BMD) in both the MITT population and the moderate-to-severe-hot-flush subset. There is no interaction term included in the regression model. Pearson's correlation was also performed to evaluate these relationships.