Background: Treatment with L can reduce bone mineral density (BMD) in patients (pts) with EBC. However, the addition of Z may protect against bone loss. Methods: We performed a 1:1:1 randomized phase III trial to compare the bone effects of adjuvant T, L or L+Z in pre- and postmenopausal pts with hormone-receptor positive EBC. Premenopausal pts also received triptorelin. Primary end point was T-score at lumbar spine (LST) measured by DXA-scan 1 year after randomization. The study had 80% power to detect a 0.35 effect size. Two comparisons were planned: L vs T and L+Z vs L; for each one, a linear model was applied to test treatment effect adjusted by age, menopausal status, previous chemotherapy and baseline LST. Two exploratory analyses were done to test whether baseline body mass index (BMI) and 6-month estradiol level were associated with L vs T effect. NCT00412022. Results: From March ‘04 to December ‘09, 483 patients were enrolled; 459 (247 pre- and 212 postmenopausal) were available for primary analyses. Median age was 50 (range 28-80); 320 (70%) pts had received adjuvant chemotherapy. At baseline 35%/23% of the pts were overweight/obese. At 1-year, mean (SD) change of LST from baseline was -0.27 (0.64) with T, -0.57 (0.66) with L and +0.02 (0.59) with L+Z. Both L vs T and L+Z vs L comparisons were highly statistically significant, p<0.0001. In premenopausal pts, but not in postmenopausal ones, bone effect of L (vs T) decreased with increasing values of baseline BMI (interaction test p=0.004 and p=0.47, respectively). There was no interaction between 6-month estradiol level and bone treatment effect, in both menopausal subgroups (p=0.31 and p=0.22, respectively). Conclusions: The HOBOE study confirms that L decreases BMD as compared with T, and that the addition of Z protects against this side effect. Interestingly, BMI seems predictive of L effect on bone health among premenopausal but not postmenopausal pts; these findings are consistent with previous observations. HOBOE has been extended and is now recruiting only premenopausal pts, to test disease-free survival effects. Novartis kindly supplied Z (for all pts) and L (for premenopausal pts). Partially supported by AIRC.