Purpose We aimed to provide long-term bone mineral density (BMD) data on early breast cancer patients of the BREX (Breast Cancer and Exercise) study. The effects of exercise and adjuvant endocrine treatment 10 years after randomization were analyzed, with special emphasis on aromatase inhibitor (AI) therapy discontinuation at 5 years.

Methods The BREX study randomized 573 pre- and postmenopausal breast cancer patients into a 1-year supervised exercise program or a control group. 372 patients were included into the current follow-up analysis. BMD (g/cm2) was measured by dual-energy X-ray absorptiometry at lumbar spine (LS), left femoral neck (FN), and the total hip. Separate groups were displayed according to baseline menopausal status, and whether the patient had discontinued AI therapy at 5 years or not.

Results The BMD change from 5 to 10 years did not significantly differ between the two randomized arms. AI discontinuation at 5 years had statistically significant BMD effects. The FN BMD continued to decrease in patients who discontinued AI therapy during the first 5-year off-treatment, but the decrease was three-fold less than in patients without AI withdrawal (− 1.4% v. − 3.8%). The LS BMD increased (+ 2.6%) in patients with AI withdrawal during the first 5 years following treatment discontinuation, while a BMD decrease (-1.3%) was seen in patients without AI withdrawal.

Conclusion This study is to our knowledge the first to quantify the long-term impact of AI withdrawal on BMD. Bone loss associated with AI therapy seems partially reversible after stopping treatment. Trial registration: http://www.clinicaltrials.gov/ (Identifier Number NCT00639210).