Background. Most ER+ tumors will eventually escape endocrine treatments. PI3K/mTOR pathway targeting has improved clinical results in endocrine treatments resistant (ETR) patients, however with unsustained responses. Our laboratory has developed several models of ETR luminal breast cancer (LBC) patient derived xenografts (PDX) suitable for biological studies (Cottu et al, AACR 2012 & IMPAKT 2012). The aims of the present study were 1) to analyze the biological factors associated with acquired ETR in relevant models of LBC; 2) to determine the therapeutic interest of combining everolimus with different ET models resistant to different endocrine treatments (ET).

Methods. Implementation of LBC models has been reported (Cottu et al, BCRT 2012). ETR models were derived by selecting tumors growing under continuous ET, subsequently re-engrafted and rechallenged with endocrine therapies (tamoxifen, letrozole, fulvestrant), and with everolimus alone and combined with ETs. When possible, initial sensitive tumors were simultaneously treated with the same protocols. At progression, or after at least 120 days of continuous therapies, tumor tissue was retrieved. Tissue microarrays (TMA) were performed to analyze the biological modifications associated to ET and mTOR targeting. Triplicate immunohistochemistry of P-AKT/AKT, P-mTOR/mTOR and P-S6 were performed in addition to standard markers. Transcriptomic and RTPCR analyses are also being conducted.

Results. Three models resistant to tamoxifen and 2 models resistant to surgical ovariectomy used a surrogate to estrogen deprivation, were developed from 3 initial PDX, and were maintained as independent hormone-resistant variants. All these models retained a luminal phenotype, based on IHC and transcriptomic analyses and had a wt PI3KCA. ETR models had an expression profile very similar to the initial PDX, suggesting that minor changes may be associated with the acquisition of an ETR phenotype. Three of these models have been tested. One tamoxifen resistant model confirmed a resistance specific to tamoxifen only and a high level of sensitivity to a fulvestrant everolimus combination. A second PDX yielded 2 other models resistant to tamoxifen and ovariectomy. These 2 models developed a resistance to all ET modalities. Therapeutic testing with combinations of everolimus and ETs were highly efficient, but everolimus alone was as efficient in these 2 models with complete and durable responses. Activation of the PI3K pathway was demonstrated in all cases by IHC analyses (expression of pAKT and pS6), and complementary ongoing RTPCR analyses. Everolimus based effects were associated with tumor fibrosis, a strong inhibition of Ki67 and pS6, but not always pAKT. The expression of ER was not modified by treatments, except in the fulvestrant-treated mice where it was strongly inhibited.

Conclusion. We have developed and validated a platform of ETR models derived from our LBC original PDX. These models retained a luminal phenotype although highly resistant to ET. Activation of the PI3K pathway has been confirmed in ETR models, and everolimus alone, or combined with ET confirms high and durable efficacy. Updated mechanistic analyses performed at different time points will be presented at the meeting.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P6-04-14.