ABSTRACT Background Local production of estrogen using aromatase at tumor site was thought to be the direct target of aromatase inhibitor (AI). However recent our works (Honma N, Can Sci 2011, Takada M, Breast 2012) and others (Lonning PE, Clin Can Res 2011) postulated that not local aromatase but whole body aromatase (possibly correlates to serum estrone (E1) level) would be the critical target of this drug. Patients and methods Among the 116 postmenopausal patients who enrolled to the JFMC 34-0601 phase II clinical trial of neo-adjuvant 24 weeks (wks) treatment with AI, exemestane, serums from 60 patients at pre-treatment, at 4wks and at the end of treatment (24wks) were subjected to this study. Estradiol (E2), E1, dehydroepiandrosterone (DHEA) and androstenedione (A) were measured using LC-MS/MS analysis for precise measurement, and E1- sulfate (E1S) was by RIA. Results Mean pre-treatment serum levels of E2, E1 and E1S were 2.29, 12.86 and 198.87 pg/ml, respectively. All these estrogens decreased to under-detection level (0.5 pg/assay for E1 and E2, 5 pg/assay for E1S) at 4wks of treatment, and it maintained to the end of the treatment in almost all patients irrespective of clinical response. Patients with higher concentration of pre-treatment E2 and E1 (cutoff = mean value) had more chance to obtain clinical objective response (p = 0.0245 and p = 0.0448, respectively). However in multivariate analysis, only E1 is the independent predictive factor for objective response (Odds 5.97, p = 0.0113), probably due to positive correlation of E2 to BMI, which is also predictive factor for AI response. When comparing pre-treatment level and those at 24wks (or at termination of treatment), patients with progressive disease showed significant increase of DHEA and A (p = 0.0313 for both), although patients with objective response maintained those at the same level. Conclusion Pretreatment serum concentration of E1, not E2, is the independent predictive factor of clinical response to neo-adjuvant exemestane. Patients with disease progression showed significant increase of androgens during the treatment. Serum steroid concentrations using precise LC-MS/MS analysis may have the informative value for the AI treatment, and the increased androgens would be the important subject to be studied further. (UMIN trial ID; 00345) Disclosure S. Saji: S.S. has a honorarium for lecture from Pfizer. M. Toi: M.T. has a research grant and honorarium for lecture from Pfizer. All other authors have declared no conflicts of interest.