(average: 6.7 months) prior to B+C was predictive of response to B+C. The 6-month PFS for the group with longer TFI was 76.8% vs. 64.1% in those with shorter TFI (p=0.004). On multivariate analysis, TFI remained as an independent prognostic factor for response to B + C (HR = 0.72; 95% CI = 0.57-0.90; p = 0.003). Conclusions: Longer treatment-free interval prior to bevacizumab combined with chemotherapy was associated with a greater like- lihood of response to bevacizumab combined with chemotherapy in recurrent ovarian cancer. However, the number of prior regimens and other clinical prognostic factors were not important. This prognostic index may have significant implications toward individualizing therapies for ovarian cancer patients. doi:10.1016/j.ygyno.2011.12.235 235 Outcomes of ovarian cancer patients after bevacizumab and chemotherapy in primary vs. recurrent setting – A multi-institutional study J. Chan 1 , A. Alvarez-Secord 2 , T. Tillmanns 3 , K. Moore 4 , K. Bevis 5 , A. ElNaggar 6 , L. Perry 4 , C. McClung 7 , C. Tian 8 , W. Huh 5 . 1 UCSF Helen Diller Family Comprehensive Cancer Center, San Francisco, CA, 2 Duke University Medical Center, Durham, NC, 3 West Clinic, Memphis, TN, 4 University of Oklahoma, Oklahoma City, OK, 5 University of Alabama at Birmingham, Birmingham, AL, 6 University of Tennessee West Clinic, Memphis, TN, 7 Stanford University Medical Center, Stanford, CA, 8 Precision Therapeutic, Inc., Pittsburgh, PA. Objective: To compare the outcomes of advanced ovarian cancer patients treated with bevacizumab with chemotherapy in primary vs. recurrent setting. Methods: Retrospective data were abstracted from six academic institutions. Chi-square, Kaplan-Meier and Cox regression model were used for statistical analyses. Results: Of 486 patients with recurrent ovarian cancer, the median age was 59 (range: 19–85). The majority of patients (84.7%) underwent primary surgery and most (96.9%) had adjuvant che- motherapy with paclitaxel and carboplatin. 21.9% had bevacizumab combined with chemotherapy in adjuvant setting while 37.3% had bevacizumab at initial recurrence. In the overall group, the median PFS after primary treatment was 11 months. The overall 5-year survival of the study group was 50%. The 5 year overall survival of those who underwent bevacizumab with chemotherapy in primary setting was 54.7% compared to 48.1% in those who had bevacizumab and chemotherapy at first recurrence (p=0.16). Conclusions: In this retrospective multi-institutional study of advanced ovarian cancer patients, our data suggest that there is no survival difference in those treated with bevacizumab and che- motherapy in primary compared to the recurrent setting. doi:10.1016/j.ygyno.2011.12.236 236 DNA methylation profiles distinguish serous, mucinous and clear cell but not endometrioid epithelial ovarian cancers S. Murphy 1 , H. Yang 1 , K. Yamaguchi 1 , Z. Huang 1 , I. Konishi 2 , A. Berchuck 1 . 1 Duke University Medical Center, Durham, NC, 2 Kyoto University, Kyoto, Japan. Objective: Altered DNA methylation is frequent in epithelial ovarian cancer, and plays a prominent role in deregulated expression of tumor suppressors and oncogenes. It is also involved in modulating chemotherapeutic response, and there is interest in using DNA methylation profiles as diagnostic and prognostic tools, yet little is known about such genome-wide profiles of ovarian cancers. Our objective was to determine if DNA methylation profiles can distinguish the four major histologic types of epithelial ovarian cancer. Methods: DNA from serous (Se; N=53), endometrioid (En; 11), clear cell (CC; 13) and mucinous (Mu; 8) ovarian cancers was bisulfite modified using the Zymo EZ DNA Methylation kit and conversion was confirmed by pyrosequencing. The Illumina Infinium 450 k Methylation Beadchip was used to quantify methylation at N 485,000 CpG sites. The raw data was adjusted for color balance, background corrected and transformed to M values, where M = log2(methylated/unmethylated signal). T-test statistics were used for comparisons using the R package limma. GATHER was used for gene ontology analysis. Results: Se, Mu and CC ovarian cancers each exhibit distinct methylation profiles as compared to the other histologic types at a q value threshold of 0.0001 (1,857, 16,740 and 3,373 differentially methylated, or DM probes, respectively) while En cancers exhibit only 26 DM probes at q = 0.01. Mu tumors showed the most pronounced differences in methylation relative to the other histologic types, particularly across chromosome 19 (Fisher's exact test p=2.153e-10). Differentially methylated genes included those involved in stimulus response (hypomethylated in Mu, p b 0.0001 but hypermethylated in CC, p =0.0002), protein phosphorylation (hypo in CC, p = 0.0003 but hyper in Se, p0.003) and organogenesis (hypo in Se, p=0.0002 but hyper in Mu, p b 0.0001). Conclusions: This is the first report on use of the Illumina 450 k methylation platform in primary ovarian cancers. Methylation profiles distinguish three of the four major histologies, but not En tumors. These surprising results suggest a very different epigenetic program in En tumors. In addition, gene ontologies targeted by hypomethylation vs. hypermethylation differ by histologic type. These results may be instructive for development of methylation biomarkers for diagnostic and prognostic use and have implications for the potential utility of epigenetic modulatory drugs in treating ovarian cancers of different histologies. doi:10.1016/j.ygyno.2011.12.237 237 High prevalence of primary peritoneal serous carcinoma among Jewish BRCA mutation carriers may limit the efficacy of screening ultrasound for early detection I. Cass, C. Saad, A. Li, C. Walsh, J. Gross, B. Karlan. Cedars-Sinai Medical Center, Los Angeles, CA. Objective: BRCA mutations cause a spectrum of pelvic serous carcinomas including primary peritoneal, tubal and ovarian cancers. The frequency and characteristics of primary peritoneal serous carcinoma (PPC) are not well characterized. Our objectives were to compare the clinical characteristics between BRCA- associated and sporadic PPC cases (non-mutation carriers) and to describe the prevalence of PPC in a tertiary care center with a large Jewish population. Methods: Tumor registry and pathology databases were queried for all gynecologic serous carcinoma cases including BRCA mutation status from 1992–2011 in an IRB approved study. Clinical and histologic information was abstracted from patient interviews and medical records. PPC diagnosis was confirmed by reviewing operative and pathology reports using GOG criteria. Results: Six hundred nineteen women with serous carcinoma were identified of whom 115 patients had PPC. Jewish womenwere more likely to have PPC than non-Jews, 25% vs. 15% (p=0.001). Among all BRCA mutation carriers with pelvic serous carcinoma,16% had PPC,15% had tubal and 69% had ovarian primaries. Jewish BRCA founder mutations tended to Abstracts / Gynecologic Oncology 125 (2012) S3–S167 S99