Lumbar spine (L2-L4) BMD measurements by dual energy X-ray absorptiometry (DXA) were performed after surgery and within 2 weeks prior to randomization and repeated every 6 months for the first 2 years and annually thereafter until 5 years after enrollment. The primary scanning site was the lumbar spine (L2 to L4) and the secondary scanning site was the total hip. All DXA scans were evaluated by a central reader.

The safety population consisted of only patients who took randomized therapy for at least one day. The number of patients in each treatment arm who had completed 2 years of the study and had centrally assessed measurements of lumbar spine or total hip BMD.

Lumbar spine (L2-L4)BMD measurements by dual energy X-ray absorptiometry (DXA) were performed after surgery and within 2 weeks prior to randomization and repeated every 6 months for the first 2 years and annually thereafter until 5 years after enrollment. The primary scanning site was the lumbar spine (L2 to L4) and the secondary scanning site was the total hip. All DXA scans were evaluated by a central reader.

The safety population consisted of only patients who took randomized therapy for at least one day. The analysis at 5 years included all patients enrolled and who had centrally assessed measurements of lumbar spine and/or total hip BMD.

Total hip BMD measurements by dual energy X-ray absorptiometry (DXA) were performed after surgery and within 2 weeks prior to randomization and repeated every 6 months for the first 2 years and annually thereafter until 5 years after enrollment. All DXA scans were evaluated by a central reader.

The safety population consisted of only patients who took randomized therapy for at least one day. The analysis of BMD at 5 years included all patients enrolled with centrally assessed measurements of total hip BMD.

Bone turnover markers (fasting serum procollagen-I extension peptide [P1NP], C-telopeptide [CTX], skeletal bone-specific alkaline phosphatase [BSAP, N-telopeptide [NTX]) were measured at baseline/screening and at each visit thereafter. A central laboratory was used to analyze the samples. The analysis of bone markers was based on analysis of variance of the regression slopes calculated for each individual patient and each bone marker over time. In the following summary, only the median treatment group percent change from baseline (and range) at 5 years is presented for each bone marker.](streamdown:incomplete-link)

The safety population consisted of only patients who took randomized therapy for at least one day. During different time points, participants with observations at that time point were included in the analysis. The analysis of bone markers over time consisted of patients with measurements of specific markers at each time point.

Serum lipid profile (fasting serum cholesterol [total, HDL and calculated LDL], triglycerides, and lipoprotein [a]) were measured at baseline/screening and at each visit thereafter. A central laboratory was used to analyze the samples. The analysis of serum lipids was on the treatment group median percent change from baseline (and range) at 5 years.

The safety population consisted of only patients who took randomized therapy for at least one day. During different time points, participants with observations at that time point were included in the analysis.

Serum lipid profile (fasting serum cholesterol [total, HDL and calculated LDL], triglycerides, and lipoprotein [a]) were measured at baseline/screening and at each visit thereafter. A central laboratory was used to analyze the samples. Numbers are not additive, as patients could be included in multiple rows.

The safety population consisted of only patients who took randomized therapy for at least one day. The number of patients with pre-defined clinically relevant changes in serum lipids over the course of 5 years in each treatment arm is presented.

Disease-free survival was defined as the interval between randomization and earliest confirmed event of loco-regional recurrence, distant metastases, invasive contralateral breast cancer, or death from any cause.

Analysis of disease-free survival was based on the ITT principle, with all enrolled (and randomized) patients included. The Kaplan-Meier product-limit approach was used.

Overall survival was measured from date of randomization to date of death.

All randomized patients constituted the ITT Population, unless withdrawal of consent occurred after randomization but before the start of study treatment assigned.