(Roche). In a 2-stage statistical analysis, data from 1 center were used to discover the lowest sAFP level 100% sensitive for IT; this cutoff point was validated using data from the other 2 centers. Discriminating properties were determined and compared for the novel sAFP cutoff point alone, RFS alone, and the 2 methods used together. Results: In the discovery set, sAFP exceeded 4.15 ng/mL in all 5 patients with pure IT and 2 of 24 with BCT (100% sensitivity, 92% specicity, area under receiver/operator curve = 0.984). In validation sets, sAFP exceeded the new cutoff value of 4.15 ng/mL in 10 of 13 patients with IT (sensitivity 77%). All 3 cases with sAFP b 4.15 ng/mL had stage 1A grade 1 IT. Of 123 BCT patients in validation sets, 12 had sAFP values over 4.15 ng/mL (specicity 91%). Among the subset of 84 patients in validation sets for whom RFS results were available, RFS was 100% specic but only 63.6% sensitive for malignancy. However, when combining RFS with sAFP at the 4.15 ng/mL cutoff value, sensitivity increased to 90.9% with little loss of specicity (90.4%). Of 4 patients with malignant tumors but negative RFS, 3 had sAFP over 4.15 ng/mL, while the fourth patient had sAFP of 1.7 ng/mL and grade 1 stage 1A disease. Conclusion: Our data demonstrate the limitations of RFS and suggest that use of preoperative sAFP with the novel cutoff point of 4.15 ng/mL discovered and validated in this study aids intraoperative decision making given the improved sensitivity for IT of the combined RFS/sAFP approach. doi:10.1016/j.ygyno.2017.03.203 176 - Poster Session Combination of targeted SMAC mimetic SW IV-134 and standard chemotherapy improves cell death and survival in ovarian cancer murine model P.S. Binder a , Y.M. Hashim b , S. Vangveravong a , M.A. Powell a , D.G. Mutch a , W. Hawkins a , D. Spitzer a . a Washington University School of Medicine in St. Louis, St. Louis, MO, USA, b Cedars-Sinai Medical Center, Los Angeles, CA, USA Objective: Chemotherapy resistance and toxicity can be a signicant hurdle in the treatment of ovarian cancer. The sigma-2/SMAC drug conjugate (SW IV-134) employs the sigma-2 ligand/receptor interaction to selectively deliver intrinsic death pathway agonist second mitochondria-derived activator of caspases (SMAC) to cancer cells. Our objective was to test the cancer cell killing ability of SW IV-134 in combination with cisplatin in vivo and in vitro in ovarian cancer. Method: OVCAR-3 and ID8 mouse ovarian surface epithelial cell (MOSEC) lines were used. In vitro luminescent cell viability assays were performed. Pathway analyses were performed by assessing caspase activation using luminescent assays and apoptotic cell death using ow cytometry. C57BL/6 and NON SCID mice were used. Subtherapeutic single-agent treatment regimens were com- pared to combination therapy in an ID8 mouse model and patient- derived xenograft (PDX) model. One- and 2-way ANOVA were used to compare differences in tumor burden and change over time. Kaplan-Meier method and log rank test were used to compare survival. Results: The cell death induced by sublethal doses of combination SW IV-134 and cisplatin was greater than that observed with either SW IV-134 or Cisplatin alone in OVCAR-3 and ID8 cells. On a mechanistic level, the activation of caspase-3, -8, and -9 and percentage of Annexin-V positive cells during cell death execu- tion was higher in the combination group than in the individual treatment groups. The percentage of Annexin-V positive cells was 0.9%, 2.5%, 4.1%, and 19.3% in the untreated, cisplatin, SW IV-143 and combination groups, respectively. Combination therapy was signicantly superior in decreasing tumor burden and improving survival of animals in the allograft and PDX model. Furthermore, the PDX mice were cured of their tumors and lived to full life- expectancy. Toxicity including weight loss was temporary during treatment. The difference in blood counts, renal function, and liver function was not signicant in the different treatment groups. Conclusion: Combining cancer-targeted SW IV-134 and cisplatin enhances their tumor-killing effectiveness. Results from these preclinical studies support future clinical applications. doi:10.1016/j.ygyno.2017.03.204 177 - Poster Session Mutations and transcript expression of maternal embryonic leucine zipper kinase in endometrial cancer patients E.R. Hope a , C.R. Miller b , C. Tian c , T.P. Conrads d,e , J.I. Risinger f , C.A. Hamilton d , G.L. Maxwell d,e,g , K.M. Darcy d . a Women's Health Integrated Research Center, Annadale, VA, USA, b Brooke Army Medical Center, Fort Sam Houston, TX, USA, c Gynecologic Cancer Center of Excellence, Bethesda, MD, USA, d Gynecologic Cancer Center of Excellence, Murtha Cancer Center, Walter Reed National Military Medical Center, Uniformed Services University of the Health Sciences, Annandale, VA, USA, e Inova Schar Cancer Institute, Fairfax, VA, USA, f Michigan State University, Grand Rapids, MI, USA, g Inova Fairfax Hospital, Falls Church, VA, USA Objective: Maternal embryonic leucine zipper kinase (MELK) is a serine/threonine kinase commonly overexpressed in cancers that regulates tumor growth, aggressiveness, and poor prognosis. Previ- ous immunohistochemistry (IHC) studies revealed ubiquitous high- intensity MELK expression in endometrial cancer (EC), and that perinuclear localization of MELK rather than intensity was associated with worse progression-free survival (PFS) and overall survival (OS). An integrated analysis of mutations and transcript expression in MELK was performed to test the hypothesis that mutations or overexpression were associated with aggressive disease and worse outcomes in EC patients. Method: Normalized level 3 RNA sequencing and clinical data were downloaded from The Cancer Genome Atlas. Relationships between MELK transcript expression and clinical factors, previously published biomarkers, and outcomes were evaluated. Results: There were 518 evaluable EC patients in this study. Mutations in MELK were rare, but overexpression of MELK transcription was common in advanced-stage disease, serous cancers, and higher tumor grade (P b 0.001). Transcript expression of MELK was inversely correlated with estrogen receptor (ESR1) and progesterone receptor (PGR) status. EC patients with MEL- K overexpression had the worse PFS (test for trend, P b 0.001, Fig. 1) and OS (test for trend, P = 0.002, Fig. 2). An incremental increase in risk of disease progression (P b 0.001) and death (P = 0.002) was observed with increasing levels of MELK transcript expression, and these relationships were maintained after multivariate correc- tion (P b 0.04). Conclusion: MELK appears to be an independent prognostic factor for aggressive disease and poor outcomes in EC patients. These ndings validate our prior IHC studies of MELK conrming that this biomarker is associated with worse PFS and OS. Additional studies are warranted to determine whether MELK is also a therapeutic target in EC patients. Abstracts / Gynecologic Oncology 145 (2017) 2220 87