Search

5036 Background: Current xenograft and transgenic models of ovarian cancer are mainly homogeneous and poorly predict response to therapy. Use of patient tumors may represent a better model for tumor biology and offer potential to test personalized medicine approaches, but poor take rates and questions of recapitulation of patient tumors have limited this approach. We have developed a protocol for improved feasibility of such a model and examined its similarity to the patient tumor. Methods: Under IRB and IACUC approval, 23 metastatic ovarian cancer samples were collected at the time of tumor reductive surgery. Samples were implanted either subcutaneously (SQ), intraperitoneally (IP), in the mammary fat pad (MFP), or in the subrenal capsule (SRC) and monitored for tumor growth. Cohorts from 8 xenolines were treated with combined carboplatin and paclitaxel or vehicle, and response to therapy compared between xenografts and patients. Expression of tumor-initiating cell (TIC) markers ALDH1, CD133, and CD44 was assessed by immunohistochemistry in tumors from patients and treated and untreated xenografts. Results: At least one SQ implanted tumor developed in 91.3% of xenografts, significantly higher than in the MFP (63.6%), IP (23.5%), or SRC (8%). Xenografts were similar in expression of putative TIC’s compared to patient tumors. The patients and the xenografts also have similar responses to chemotherapy in that xenografts from patients with a partial response responded more slowly than those from patients achieving a complete response (45 vs 21 days, p=.004). Treated xenografts were more densely composed of TICs. ALDH1 increased to 36.1% from 16.2% (p=0.002) and CD133 increased to 33.8% from 16.2% (p=0.026). Conclusions: Xenoline development can be achieved at a high rate when tumors collected from metastatic sites are implanted SQ. These xenografts are similar to patient tumors with regard to chemotherapy response and TIC expression.. This model may be a more accurate model for in vivo pre-clinical studies as compared to current models. Also, as treated xenografts become chemoresistant, this model is well positioned to evaluate targeted therapies aimed at the most aggressive populations in a heterogeneous tumor.

e15502 Background: Poor long-term survival in high-risk endometrial cancer patients attests to a need for new therapeutic strategies. The prevalence of estrogen (ER) and progesterone receptors (PR) among high-risk endometrial carcinomas has not been systematically studied and is a target to which individualized therapies may be tailored. Methods: 41 women with high grade, stage III/IV and/or aggressive histology endometrial cancer who received post-hysterectomy chemotherapy were identified. Patient, tumor and treatment characteristics were recorded. Immunohistochemical staining of ER and PR was scored on a 0-4 scale incorporating intensity and extent of expression. Receptor-positivity was defined as a score of >= 2. Univariate and multivariate logistic regression analysesand the Cox proportional hazards model were performed to examine hormone receptor expression and recurrence risk. Results: Median age at surgery was 59 years (range 34-79). Median follow up time was 39 months (range 1-82). Recurrence occurred in 21 women (51%) with median time to first recurrence of 19 months (range 1-71). 22 cancers (54%) were ER+ and 21 (51%) PR+. Median ER and PR expression was 2 and 2, respectively (range 0-4). While ER and PR were not independent factors for recurrence, on multivariate logistic regression analysis ER and PR as continuous values showed recurrence associated with greater ER expression (p = 0.0836) and lesser PR expression (p = 0.0986). Cox proportional hazards model supported the association between increasing ER and decreasing PR with shorter recurrence free survival (p=0.0034, p = 0.0021). An increase in ER score of 1 corresponded to a hazard ratio (HR) of 2.647 (95% CI 1.38-5.08); an increase in PR score of 1 corresponded to a HR of 0.362 (0.19-0.693). Conclusions: More than 50% of high-risk endometrial cancers in this small retrospective cohort receiving cytotoxic chemotherapy expressed positive hormone receptors. Multivariate analyses of hormone receptors suggest increased recurrence with discordant expression of ER and PR. Further investigation is warranted to verify whether differential hormone receptor expression predicts poorer response to cytotoxic systemic therapy.