DFS defined as time from randomization to earliest documentation of breast cancer relapse or death from any cause. DFS at Month 36 post-randomization was defined as probability of participants alive and disease-free at 36 months after the randomization. Participants withdrawn from the study for any reason in the absence of relapse were censored at the date they were last seen. Relapse was categorized as follows: loco-regional: ipsilateral breast or axillary nodal relapse; distant: distant relapse, including supraclavicular nodes; second primary breast cancer: contralateral breast cancer, excluding ductal carcinoma in situ.

Intent-to-treat (ITT) population included all participants assigned to the treatment group to which they were randomized, irrespective of the treatment they actually received.

OS was defined as the duration from randomization to death (due to any cause). OS at Month 36 post-randomization was defined as probability of participants' survival at 36 months after the randomization. For participants who were alive, OS was censored at the last available assessment. Probability of OS at Month 36 post-randomization was reported using Kaplan-Meier estimates at Month 36 post-randomization based on 120-month follow-up data.

ITT population included all participants assigned to the treatment group to which they were randomized, irrespective of the treatment they actually received.

Number of events of second primary breast cancer in contralateral breast (excluding ductal carcinoma in situ) were reported.

ITT population included all participants assigned to the treatment group to which they were randomized, irrespective of the treatment they actually received.

BMD measurements for Lumbar spine (LS) and Proximal Femur (Total Hip [TH]) were performed using dual energy X-ray absorptiometry (DXA) for participants who entered the bone-metabolism sub-study. 'N' (number of participants analyzed) signifies those participants who were evaluable for this outcome measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

Evaluable population for bone metabolism substudy: all treated participants who did not violate any exclusion criteria, received treatment for at least 9 months and had baseline and at least on-treatment Month 12 and/or Month 24 assessment available for parameter to be analyzed. Participants were analyzed according to treatment actually received.

BMD measurements for femoral neck (FN) and femoral wards (FW) were performed using dual energy X-ray absorptiometry (DXA) for participants who entered the bone-metabolism sub-study. 'N' (number of participants analyzed) signifies those participants who were evaluable for this outcome measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

Evaluable population for bone metabolism substudy: all treated participants who did not violate any exclusion criteria, received treatment for at least 9 months and had baseline and at least on-treatment Month 12 and/or Month 24 assessment available for parameter to be analyzed. Participants were analyzed according to treatment actually received.

BMD measurements for Lumbar spine (LS) and Proximal Femur (Total Hip [TH]) were performed using dual energy X-ray absorptiometry (DXA) for participants who entered the bone-metabolism sub-study. Results were scored as T-score. T-score indicated how many standard deviations higher or lower participant's value was when compared to the young normal reference mean. Using the World Health Organization (WHO) criteria for osteoporosis, a T-score of greater than or equal to (>=)-1.0 was classified as normal, a T-score of greater than -2.5 to less than -1.0 as osteopenic, and a T-score less than or equal to (<=)-2.5 as osteoporotic. Here 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

Evaluable population for bone metabolism substudy: all treated participants who did not violate any exclusion criteria, received treatment for at least 9 months and had baseline and at least on-treatment Month 12 and/or Month 24 assessment available for parameter to be analyzed. Participants were analyzed according to treatment actually received.

Bone specific alkaline phosphatase (BAP) serum concentration analyzed using enzyme immuno assay (EIA) at post-baseline time points was expressed as percentage of baseline BAP serum concentration. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

As treated population for bone metabolism sub-study included all treated participants, irrespective of the treatment duration and allocated to the group that corresponded to the treatment they actually received. Analysis population for biomarkers included as treated population with baseline and at least 1 on-treatment assessment available.

C-terminal telopeptide (CTX) serum concentration analyzed using competitive enzyme-linked immunosorbent assay (ELISA) at post-baseline time points was expressed as percentage of baseline CTX serum concentration. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

As treated population for bone metabolism sub-study included all treated participants, irrespective of the treatment duration and allocated to the group that corresponded to the treatment they actually received. Analysis population for biomarkers included as treated population with baseline and at least 1 on-treatment assessment available.

Osteocalcin (OC) serum concentration analyzed using ELISA and procollagen T1 c-peptide (PICP) serum concentration analyzed using sandwich EIA at post-baseline time points was expressed as percentage of baseline OC serum concentration and baseline PICP serum concentration, respectively. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points, for each group respectively.

As treated population for bone metabolism sub-study included all treated participants, irrespective of the treatment duration and allocated to the group that corresponded to the treatment they actually received. Analysis population for biomarkers included as treated population with baseline and at least 1 on-treatment assessment available.

Deoxy-pyridinoline (DPD) urine concentration (adjusted for urinary creatinine) analyzed using competitive EIA at post-baseline time points was expressed as percentage of baseline DPD urine concentration. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

As treated population for bone metabolism sub-study included all treated participants, irrespective of the treatment duration and allocated to the group that corresponded to the treatment they actually received. Analysis population for biomarkers included as treated population with baseline and at least 1 on-treatment assessment available.

N-telopeptide of Type 1 collagen (NTX) urine concentration (adjusted for urinary creatinine) analyzed using competitive inhibition EIA at post-baseline time points was expressed as percentage of baseline NTX urine concentration. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

As treated population for bone metabolism sub-study included all treated participants, irrespective of the treatment duration and allocated to the group that corresponded to the treatment they actually received. Analysis population for biomarkers included as treated population with baseline and at least 1 on-treatment assessment available.

As treated population for bone metabolism sub-study included all treated participants, irrespective of the treatment duration and allocated to the group that corresponded to the treatment they actually received.

The TOI was defined as the sum of 23 items based on following Functional Assessment of Cancer Therapy - Breast version [FACT-B] subscales: Physical well-being (7 items), Functional well-being (7 items), Breast cancer subscale (9 items). Each item was scaled from 0='Not at all' to 4='Very much'. Total TOI score ranged from 0 to 92, where higher TOI score indicated better health-related quality of life (QoL). A change of five points in the TOI scores was considered clinically meaningful. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

ITT population for QoL sub-study included all randomized participants with available data for any given endpoint and were grouped according to randomized treatment, irrespective of whether they were actually treated or not.

The FACT-ES assessed health-related QoL in participants with breast cancer. ES subscale comprised of 18 items (hot flushes,cold sweats,night sweats, vaginal discharge,vaginal irritation,vaginal bleeding,vaginal dryness,discomfort with intercourse,lost interest in sex,gained weight,light headed/dizzy,vomiting,had diarrhea,headaches,felt bloated,breast tenderness,mood swings, felt irritable).Participants indicated how true a statement was for them using a 5-point scale from 0 (not at all) to 4 (very much). For items that were negatively framed, the scores were reversed for the analysis so that higher scores equated to a good QoL. Total FACT-ES score was calculated as sum of all the 18 items and ranged from 0 to 72, where higher score indicated better QoL. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

FACT-GBE assessed health-related quality of life (QoL) in participants with breast cancer. It consisted of 56 items,summarized to 7 subscales(subscale 1 to 6 constituted total FACT-B and subscale 7 constituted total ES):physical well-being(7 items), social/family well-being(7 items),relationship with doctor (2 items),emotional well-being(6 items),functional well-being(7 items),breast cancer subscale(9 items),endocrine symptoms(18 items). Participants indicated how true a statement had been for them using 5-point scale from 0(not at all) to 4(very much). For items that were negatively framed,scores were reversed for analysis so that higher scores equated to good QoL. Total FACT-GBE score=sum of all 56 items(range 0 to 224, where higher score indicated better QoL. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

The PWB subscale assessed physical well-being related QoL in participants with breast cancer. PWB subscale comprised of 7 items (energy lack, nausea, family needs, pain, side effects, felt ill, forced to stay in bed). Participants indicated how true a statement had been for them using a 5-point scale from 0 (not at all) to 4 (very much). For items that were negatively framed, the scores were reversed for the analysis so that higher scores equated to a good QoL. Total PWB score was calculated as the sum of all the 7 items and ranged from 0 to 28, where higher score indicated better physical well-being related QoL. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

The SWB subscale assessed social/family well-being related QoL in participants with breast cancer. SWB subscale comprised of 7 items (distant from friends, emotional support, support from friends, family acceptance, family communication, close to main support, sexual satisfaction). Participants indicated how true a statement had been for them using a 5-point scale from 0 (not at all) to 4 (very much). For items that were negatively framed, the scores were reversed for the analysis so that higher scores equated to a good QoL. Total SWB score was calculated as the sum of all the 7 items and ranged from 0 to 28, where higher score indicated better social/family well-being related QoL. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

The RWD subscale assessed relationship with doctor in participants with breast cancer. RWD subscale comprised of 2 items (confidence in doctors, doctor answered questions). Participants indicated how true a statement had been for them using a 5-point scale from 0 (not at all) to 4 (very much). Total RWD score was calculated as the sum of the 2 items and ranged from 0 to 8, where higher score indicated better relationship with doctor. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

The EWB subscale assessed emotional well-being related QoL in participants with breast cancer. EWB subscale comprised of 6 items (felt sad, proud of coping, lost hope, felt nervous, worried about dying, worried about condition worsening). Participants indicated how true a statement had been for them using a 5-point scale from 0 (not at all) to 4 (very much). For items that were negatively framed, the scores were reversed for the analysis so that higher scores equate to a good QoL. Total EWB score was calculated as the sum of the 6 items and ranged from 0 to 24, where higher score indicated better emotional well-being related QoL. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

The FWB subscale assessed functional well-being related QoL in participants with breast cancer. FWB subscale comprised of 7 items (able to work, work fulfilled, able to enjoy life, acceptance of illness, sleeping well, enjoyed normal fun activities, contented with QoL). Participants indicated how true a statement had been for them using a 5-point scale from 0 (not at all) to 4 (very much). Total FWB score was calculated as the sum of the 7 items and ranged from 0 to 28, where higher score indicated better functional well-being related QoL. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

The BCS subscale assessed health related QoL in participants with breast cancer. BCS subscale comprised of 9 items (short of breath, self-conscious dress, tender/swollen arms, sexually attractive, bothered by hair loss, worried about familial risk, worried about family stress, bothered by weight change, able to feel like a woman). Participants indicated how true a statement had been for them using a 5-point scale from 0 (not at all) to 4 (very much). For items that were negatively framed, the scores were reversed for the analysis so that higher scores equated to a good QoL. Total BCS score was calculated as the sum of the 9 items and ranged from 0 to 36, where higher score indicated better QoL. 'N' (number of participants analyzed) signifies those participants who were evaluable for this measure and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

Participants indicated prevalence of an endocrine subscale items using a 5-point scale, where 0 (not at all), 1 (a little bit), 2 (somewhat), 3 (quite a bit), 4 (very much). Endocrine items were grouped in five categories vasomotor (hot flushes, cold sweats, night sweats, sleeping difficulties), neuropsychological (lack of energy, nervous feeling, lightheaded/dizzy, headaches, mood swings, feeling irritable), gastrointestinal symptoms (nausea, gained weight, vomiting, diarrhea, bloated feeling), gynecological symptoms (vaginal discharge, vaginal irritation, vaginal bleeding, vaginal dryness, discomfort with intercourse, lost interest in sex, breast tenderness) and other symptoms (pain, feeling ill, side effects). Number of participants who reported severe endocrine symptoms (defined as response categories "quite a bit" and "very much") were presented.

ITT population for QoL sub-study. Results for 30, 36, 48, 60 months were not reported because data for these time points was only summarized as graphical presentation.

Endometrial thickness was assessed using transvaginal ultrasound examination. 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

Evaluable population for endometrial sub-study included all treated participants who did not violate any exclusion criteria, received treatment for at least 2 years, and had on-treatment endometrial ultrasound examination performed between 22 and 26 months from treatment start. Analysis was based on actual treatment received.

Endometrial thickness was assessed using transvaginal ultrasound examination. 'N' (number of participants analyzed) signifies those participants who were evaluable for this outcome and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

Evaluable population for endometrial sub-study included all treated participants who did not violate any exclusion criteria, received treatment for at least 2 years, and had on-treatment endometrial ultrasound examination performed between 22 and 26 months from treatment start. Analysis was based on actual treatment received.

Uterine volume (UV) and ovarian volume was estimated using ultrasonography. Uterine volume = (longitudinal diameter * transverse diameter * anteroposterior diameter of uterus)/(2*1000). Ovary volume = [(longitudinal diameter * transverse diameter * anteroposterior diameter of ovary) * 3.14]/(6*1000). Overall ovary volume (OV) is calculated as the sum of the right and left ovary volume. 'N' (number of participants analyzed) signifies those participants who were evaluable for this outcome and 'n' signifies those participants who were evaluable for this measure at given time points for each group, respectively.

Evaluable population for endometrial sub-study included all treated participants who did not violate any exclusion criteria, received treatment for at least 2 years, and had on-treatment endometrial ultrasound examination performed between 22 and 26 months from treatment start. Analysis was based on actual treatment received.

Number of participants with presence of polyps (POL) and fibroids (FIB) at post-baseline time points compared to the baseline (BL) status of 'yes', 'no' or 'missing' (that is, participants reporting POL/FIB at post-baseline time points who had yes, no or missing POL/FIB status at baseline, respectively) were presented. Result for number of participants with ovarian cysts was not analyzed at post-baseline time points as very few participants reported ovarian cysts at baseline.

Evaluable population for endometrial sub-study. Analysis was based on actual treatment received. 'n' signifies those participants who were evaluable for this measure at given time points for each group,respectively.

Gynecological symptoms included bleeding/spotting, pelvic pain, leucorrhoea and vaginal itching.

As treated population included all treated participants, irrespective of the treatment duration and allocated to the group that corresponded to the treatment they actually received.

Results were not reported for this outcome measure because no data was collected as per change in planned analysis.