Background: Despite an initial response to hormone therapy, luminal breast cancers (BC) are defined by a persisting long term risk of relapse. Preclinical models of estrogen-dependent human BC are therefore requested for a better understanding of estrogen receptor (ER) and hormone resistance biology. Because of a very low tumor take in immunodeficient mice, most in vivo models of estrogen-dependent human breast tumors are derived from human cancer cell lines. We report here the establishment and the characterization of new primary human luminal breast cancer xenografts directly obtained from fresh human tumor samples. Methods: As of December 2009, 453 fresh human BC samples have been engrafted in the interscapular fatpad of nude mice, of which 405 were retained for further studies (32 were non infiltrating or non-breast carcinoma, and 16 were axillary metastatic lymph node from a simultaneously engrafted primay tumor). ER was expressed in 313 tumors (77.3%), progesterone receptor in 175/291 informative tumors (60.1%), Her2 in 39/315 tumors (12.4%), and 60 tumors were triple negative. After transplantation, mice were kept in a dedicated unit and fed with water containing estrogen. Validation of the xenografts was obtained by a large phenotypic and genotypic profiling including: pathological and immunohistochemical (IHC) examination, dedicated gene expression (RT-qPCR), genomic (BAC CGH arrays) and transcriptomic (Affymetrix u133 RNA chips) analyses, and therapeutic assessment (estrogen deprivation, ovariectomy, LHRH agonists, letrozole, tamoxifen, fulvestrant). Results: Among the 405 human xenografted tumors, 8 luminal models have been established (2%), 7 from ER+/PR+ tumors and 1 from an axillary relapse of an ER-/HER2+ tumor. In all tumor/xenograft pairs, histopathological analyses showed an impressive morphological concordance. One had a strong mucinous component, and all of them were grade II/III tumors. Out of the 7 ER+/PR+ models, 3 were also HER2 positive. RNA expression by RT-qPCR confirmed ER, PR and HER2 status for the 7 ER+/ER+ tumors, and confirmed the ER+ status of the ER-/HER2+ derived tumorgraft. CGH arrays analyses demonstrated striking similarities of the genomic profile between the original tumors and their corresponding xenografts. Array CGH analyses were also performed at several passages, showing stable profile of the tumors during successive in vivo passages. Transcriptomic profile are still ongoing and will define each xenograft as either luminal A or B tumor subtypes. The tumor growth of all luminal BC models was estrogen dependent as demonstrated by estrogen supplementation and ovariectomy. Finally, therapeutic characterization of the xenografts showed that tamoxifen had a delayed but significant anti-tumor effect, whereas fulvestrant was the most efficient hormone therapy with durable complete responses observed in 3/3 evaluable models. Updated and extended results will be presented during the meeting. Conclusions: We have durably established and characterized 8 primary human luminal BC xenografts. In order to identify new therapeutic approaches of hormone resistant BC, we have now planed to decipher in these well-defined preclinical models the molecular variations associated with emergence of resistance to hormone therapies.