Abstract Background: Aromatase inhibitors (AI) are accepted as adjuvant endocrine therapy for hormone receptor-positive (HR+) postmenopausal early breast cancer (EBC) with superior efficacy to tamoxifen. However, because of the increased bone loss associated with the use of AI, three international randomised studies Z-FAST, ZO-FAST and E-ZO-FAST were performed to evaluate the bone protective effects of zoledronic acid (ZOL). These studies showed that the bone mineral density (BMD) loss associated with AI therapy can effectively be prevented by ZOL. In addition, these studies included disease-free survival (DFS) as a secondary endpoint, although the studies were not powered to detect a difference between treatment arms. Following the results from ABCSG 12 indicating a powerful effect of ZOL in premenopausal women receiving endocrine therapy, exploratory analyses of the individual trials have been done. However, this is the first report of DFS outcome for all three studies.Materials and Methods: Across the three studies 2194 patients(pts) with HR+ EBC were randomised to either immediate ZOL 4mg every 6 months or delayed ZOL where bone protection was started if either the BMD T-score fell below < -2.0 SD or a non-traumatic fracture occurred. DFS was defined as time to first appearance of breast cancer recurrence or death from any cause. Pts in ZO-FAST and E-ZO-FAST, but not in Z-FAST, were followed for recurrence and survival after stopping study drug. DFS was summarized by the Kaplan-Meier method and compared between the two groups using the log-rank test, stratified by baseline BMD and adjuvant chemotherapy. A Gail-Simon test was performed to assess interactions between treatment and study on DFS.Results: Median follow up times for this analysis are 54, 48 and 36 months respectively for Z-FAST, ZO-FAST and E-ZO-FAST. The proportion of pts with a follow-up lag time of ≥ 6 months was 29.7% Z-FAST, 8.3% ZO-FAST and 8.7% E-ZO-FAST. The pt-years of time on study for both arms combined were calculated; Z-FAST 2709; ZO-FAST 4260 and E-ZO-FAST 1580. The percent of pts starting ZOL in the delayed ZOL arm was: Z-FAST 26%, ZO-FAST 25%, and E-ZO-FAST 17%. The Gail-Simon test was statistically significant for a quantitative interaction between studies (p-value 0.047). Therefore, combining the results was not considered statistically appropriate. Numbers of recurrences and individual trial hazard ratios are shown below.Table 1 No of PatientsImmediateDelayedHR (95% CI)p valueZ-FAST60222260.80 (0.45-1.41)NSZO-FAST106532530.59 (0.38-0.92)0.0176E-ZO-FAST52719111.76 (0.83-3.69)NS Conclusion: This exploratory analysis of DFS has several limitations: a small number of pt events, differences between trials in the method of collecting pt follow up data on DFS and length of median follow-up times, a wide range of pt-years on study, and a statistically significant interaction between studies, reflecting the heterogeneity of results. Also, the use of a ZOL delayed arm as active control limits a direct assessment to that of no ZOL. Additional analyses are planned when the studies are completed. Results from ongoing appropriately powered randomised studies are awaited to determine the true role of adjuvant ZOL in EBC. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 4082.