2771 BRCA1 and BRCA2 mutations in 1691 epithelial ovarian tumors identify subgroups with distinct molecular characteristics

S554

Abstracts

of pLV-sFLT-1 and the treatment with recombinant VEGFR-1 induced cytotoxicity, but didn’t increase caspase-3 activity, the number of TUNELpositive cells, and Annexin V-positive cells. The activation levels of the extracellular signal regulated kinase (ERK) and c-Jun N-terminal kinase (JNK) in cell lysate didn’t decrease by transfected with sFLT-1. By FACS for Annexin V and propidium iodide (PI), transfection of sFLT1 increased the percentage of necrotic cells compared to the non-transfected group. Conclusions: In this study, non-apoptotic cell damage was found to be induced by sFLT1. Soluble FLT1 could be a future therapeutic candidate for several cancers. No conflict of interest. 2769 POSTER Trabectedin in women with advanced ovarian carcinoma and BRCAness phenotype − Exploratory analysis C. Hinojo Gonzalez1 , C. Orbegoso Aguilar1 , M. Mangas Izquierdo2 , R. Jimeno Mate1 , C. Lopez Lopez1 , E. Martinez De Castro1 , V. Martinez Callejo3 , A. De Juan Ferre1 . 1 Marques de Valdecilla Universitary Hospital, Medical Oncology, Santander, Spain; 2 Galdakao Hospital, Medical Oncology, Bilbao, Spain; 3 Marques de Valdecilla Universitary Hospital, Pharmacy, Santander, Spain Background: BRCA1/2 defective or “BRCAness” phenotype in ovarian cancer (OC) identifies a subgroup of high-grade serous tumors characterized by its exquisite sensitivity to platinum and prolonged survival. Trabectedin (T) binds to minor groove of DNA and it is particularly active in OC patients with a defective homologous recombination system. In addition, Trabectedin’s unique mechanism of action, could explain the reversal of platinum resistance when it is used in subsequent lines. Material and Methods: Women with advanced OC and BRCAness phenotype treated with T in our department from July 2012 to May 2015 were retrospectively analyzed. Median follow up time was 13 months (range 1−23). Overall survival (OS) was defined as the interval from date of starting T treatment to date of death from any cause. N (%) Tumor location Ovarian Fallopian tube Histology Papillary serous Other Grade: High-grade Stage at diagnosis I II III IV Platinum free interval (months) <6 6−12 Results T clinical benefit rate PFS (months) Subsequent Platinum Subsequent Platinum clinical benefit

9 (75%) 3 (25%) 8 (67%) 4 (33%) 9 (75%) 2 1 6 3

(17%) (8%) (50%) (25)

2 (17%) 10 (83%) 6 (50%) 7.33 (1−14) 8 (67%) 3 (37%)

Results: 9 women with OC and 3 Fallopian tube cancers were treated with T. 67% were papillary serous carcinomas and 75% were high-grade. Mean age was 63 years (range 47−78). 2 patients (pts) had BCRA mutation confirmed and a variant of unknown significance was identified in other 2 pts. T was combined with pegylated liposomal doxorubicin in 75% of pts and 25% received T as monotherapy. All pts were ECOG 0−1. 83% had a platinum progression-free interval of 6−12 months. 50% of pts received 3 or more previous lines and 42% only one. Progression-free survival was 7,3 months (range 1−14). Average number of cycles administered was 5 (range 1−12). Clinical benefit was achieved in 6 pts (50%). 67% (8 pts) received a subsequent line with platinum with a response rate of 37%. 67% of pts were alive at the data cut-off time. Mean OS for the entire cohort was 17,8 months (95% CI: 11−23); pts who did not receive subsequent platinum had an OS of only 3 months (95% CI: 0,4−9) whereas OS for pts platinum-rechallenged has not been reached. Most common g3−4 toxicities reported were neutropenia (42%); thrombocytopenia (25%); asthenia (25%); emesis (17%) and constipation (8%). Dose reduction was necessary in 57% of pts.

Conclusions: BRCAness phenotype in OC pts is a biologically distinct entity that identifies a better prognostic subgroup of patients with a striking sensitivity to Trabectedin in platinum previously treated pts. Our results are consistent with the ones observed in the phase II trial (MITO 15) for women with BRCA mutation OC and BRCAness phenotype treated with Trabectedin. No conflict of interest. 2770 POSTER Routine germline BRCA testing in serous ovarian cancer: The West of Scotland experience P. Spiliopoulou1 , S. Gibson2 , R. Davidson3 , R. Glasspool1 , I. McNeish4 . 1 Beatson West of Scotland Cancer Centre, Medical Oncology, Glasgow, United Kingdom; 2 Southern General Hospital, Clinical Genetics, Glagsow, United Kingdom; 3 Southern General Hospital, Clinical Genetics, Glasgow, United Kingdom; 4 Institute of Cancer Sciences, University of Glasgow, Glasgow, United Kingdom Background: Germline mutations in BRCA1/2 (gBRCA) are strongly linked to the development of non-mucinous ovarian cancer (OC), in particular high grade serous OC (HGSOC). Up to 18% unselected patients with HGSOC may harbor mutations. Since Jul 13, all patients with non-mucinous OC in the West of Scotland (WoS) have been offered germline BRCA1/2 testing regardless of family history. All patients received counselling prior to testing. Methods: Sequencing results, derived using chain termination method, were collated with clinical data. Progression-free (PFS) and overall (OS) survivals in gBRCA mutation carriers and non-carriers were compared using Gehan-Breslow-Wilcoxon test. Results: From Jul 13-Nov 14, 154 patients were referred to genetics and 120 accepted testing. gBRCA mutations were identified in 22/120 (18.3%) − 6 BRCA1, 16 BRCA2. An additional 4/120 (3.3%) had variants of unknown significance (VUS). In HGSOC, the mutation rate was 21/92 (22.8%) (6 BRCA1, 15 BRCA2). 50% mutations occurred in women with Manchester score (MS) 15: these would have been missed if family history alone determined decision to test. gBRCA mutation rate was comparable to that (20%) found in the highly-selected cohort of patients tested in the preceding years (Mar 11-Jun 13). gBRCA1 mutation carriers were significantly younger than gBRCA2 (52 vs 62 years, p = 0.026, gBRCA1: 35−84years, BRCA2: 32−78years). OS for gBRCA2 carriers with stage III/IV HGSOC was significantly better than gBRCA wild-type (HR: 2.25 95% CI 1.05–4.79, p = 0.02, gBRCA wild-type median 63.2m, BRCA2 carriers median undefined). Median PFS of gBRCA2 carriers was also significantly better than gBRCA wild-type (27.9m vs 15.1m, HR 1.8, 95% CI 1.08–3.26 p = 0.02). For gBRCA1carriers both PFS and OS did not differ significantly from non-carriers. Conclusions: Our data confirm that routine germline BRCA1/2 mutation testing is acceptable and feasible in women with OC. Whole population screening is more effective than selection based on family history. c.20% women with HGSOC carry a gBRCA mutation. In WoS, we observe higher prevalence of BRCA2 mutations (16.3%) than previously reported (6−8%). In our cohort, gBRCA2 mutation confers better PFS and OS than wild-type BRCA1/BRCA2 status. No conflict of interest. 2771 POSTER BRCA1 and BRCA2 mutations in 1691 epithelial ovarian tumors identify subgroups with distinct molecular characteristics T. Herzog1 , J. Xiu2 , R. Bender3 , Z. Gatalica4 , S. Reddy2 . 1 University of Cincinnati, College of Medicine, Cincinnati, USA; 2 Caris Life Sciences, Medical Affairs, Phoenix, USA; 3 Caris Life Sciences, CMI Lab Operations, Phoenix, USA; 4 Caris Life Sciences, Pathology, Phoenix, USA Background: Increasing data has shown that ovarian cancer patients carrying BRCA1 and BRCA2 mutations, both somatic and germline, experience a survival advantage and are more sensitive to DNA damaging agents such as platinum and PARP inhibitors. We aim to identify BRCA1 and 2 mutations in various histological subtypes of epithelial ovarian cancer (EOC) tumor samples and to identify molecular differences in EOCs with or without BRCA1 or 2 mutations. Material and Methods: A total of 1691 EOC tumors were submitted to Caris Life Sciences between April, 2014 and March, 2015 (63% serous, 5% clear cell, 4% carcinosarcoma, 3% endometrioid, 16% unannotated or carcinoma not-otherwise specified). Multiplatform molecular analysis included next generation sequencing, immunohistochemistry (IHC) of protein expression, and/or gene amplification (FISH/CISH). Results: Deleterious BRCA1 mutations were seen in 9% (n = 155) of EOC tumors [10.5% in serous, 3% in carcinosarcoma, 3% in clear cell and 2%

Abstracts

S555

in endometrioid]. In addition, 4.2% (72) of BRCA1 mutations characterized as presumed pathogenic (PP) or variants of unknown significance (VUS) were found. Deleterious BRCA2 mutations were seen in 5% (86) of EOC tumors [6% in serous, 6% in carcinosarcoma, 2% in endometrioid and 1% in clear cell]. In addition, 8% (136) of PP or VUS BRCA2 mutations were seen in EOC tumors. No deleterious BRCA1 and BRCA2 mutations were found to co-occur. In the 249 EOC tumors that carried deleterious BRCA1 or 2 mutations, TP53 mutation was significantly more frequent than the wild type tumors (83% vs. 68%, p < 0.0001). Wild type EOC tumors carried significantly higher mutation rates of KRAS (10% vs. 2%, p < 0.0001), PIK3CA (8% vs. 3%, p = 0.003), CTNNB1 (3% vs. 1%, p = 0.03) and more frequent expression of cMET (16% vs. 7%, p < 0.0001). In the 169 serous ovarian tumors with deleterious BRCA1 or 2 mutations, similar to the complete EOC cohort, a significantly higher TP53 mutation (85% vs. 67% p < 0.0001), lower PIK3CA (0.6% vs. 4%, p = 0.02) and KRAS mutation rates (1% vs. 7%, p = 0.002) were seen compared to the wild type serous ovarian tumors. In the 3 clear cell ovarian tumors that carried deleterious BRCA1 or 2 mutations, 100% (3) expressed ER while in the 61 BRCA1/2 wild type clear cell tumors, only 3% (2) showed ER expression (p = 0.0002) Conclusion: Deleterious BRCA1 and 2 mutations are seen in 9% and 5% of epithelial ovarian tumors, respectively, with the highest mutation rate observed in serous histology, followed by carcinosarcoma. Oncogenic drivers including KRAS and PIK3CA are more likely to mutate in BRCA1/2 wild type EOC tumors while TP53 mutations significantly co-occur with BRCA1/2. High expression of ER in clear cell ovarian tumors with BRCA1/2 deleterious mutation suggests potential combination of hormone therapy with PARP inhibitors in this rare cancer type. Conflict of interest: Advisory Board: (Dr. Thomas Herzog) Genentech, Morphotech, Caris Life Sciences.

disease (48% vs 19%, p = 0.004 Fisher exact test) and Fagotti score 8 (61% vs 56%, p = 0.8) were more frequent in pts with CTC 1 than <1. The same trend was seen for a C1 CTC count 2. CTC count and CA-125 correlated poorly at each time point (R = 0.17 at C1, 0.33 at C2 and 0.29 at IDS). Five CTC kinetic profiles were identified (“increasing”, “decreasing”, “peak”, “nadir”, “null”): 5 (6%) pts had an increasing CTC count, 23 (26%) a decreasing count, 12 (14%) had a peak at C2 and 2 (2%) had a nadir at C2. These profiles will be used for further exploratory analysis of the complete resection rate (primary endpoint). Preliminary safety analysis showed no significant difference between CTC negative (<1) and positive (1) pts. Conclusions: CTC are detectable at low levels in about half of pts with untreated high stage OC. CTC counts correlated with clinical stage, but poorly with CA-125 values, suggesting they may generate additional information to CA-125. CTC profiles can be determined with serial sampling that may serve as potential predictors of response. No conflict of interest.

2772 POSTER Circulating tumor cells kinetics in ovarian carcinoma: correlation with baseline characteristics and CA-125 levels in the randomized ANTHALYA trial evaluating bevacizumab in the neoadjuvant setting

Background: The survival rate for ovarian cancer has remained largely unchanged over the past three decades, despite the rapid advancement of our knowledge of the molecular and genetic mechanisms underlying most of the subtypes of ovarian cancer. There is, therefore, a need to rapidly translate this knowledge into improved clinical outcomes for patients with ovarian cancer. The Australian Ovarian Cancer Assortment Trial (ALLOCATE) study was designed to screen tumour samples from patients with relapsed ovarian cancer, using next-generation sequencing technology (NGS), with the aim of sorting patients into various targeted treatment arms based on the genetic make up of their tumours. A panel of 29 genes that are commonly mutated in ovarian cancer and potentially therapeutically targetable was developed using an Illumina TruSeq Custom Amplicon Panel. A panel of 10 genes with common copy number changes in ovarian cancer was also developed using a novel, multiplex amplicon-based gene dosage method. Both panels were validated and accredited. We aimed to determine the clinical utility of the ALLOCATE test, as well as identify potential hurdles in translating a multi-gene panel into clinical practice. Material and Methods: 30 patients with advanced ovarian cancer with unselected tumour (17 archival formalin-fixed paraffin-embedded (FFPE), 12 snap-frozen and 1 ascites) and matching blood samples were screened using the validated panels. Results: We identified at least one pathogenic mutation in 19 of the 21 assessable tumours, which included 5 cases with potentially actionable BRCA1 and BRCA2 mutations, a PIK3CA E545K mutated case, and a KRAS G12V mutated case. The KRAS mutated tumour was reclassified to low-grade, altering the prognosis of the patient and allowing the patient to be enrolled in the low-grade arm of the Paragon trial. There was a 23.3% sample quality control failure rate with 4 cases (2 frozen and 2 FFPE) having insufficient tumour purity (<20%) and 3 FFPE cases having DNA that was too fragmented for accurate analysis. Conclusions: We demonstrated clinical utility of the ALLOCATE test as a means of stratifying treatment for ovarian cancer patients, however, there are still a number of challenges in relation to pre-analytical sample quality and selection that need to be overcome. No conflict of interest.

P. Cottu1 , T. De La Motte Rouge2 , P. Pautier3 , M. Provansal4 , A. Floquet5 , F. Selle6 , M.A. Mouret-Reynier7 , M. Fabbro8 , E. Kalbacher9 , C. Piprot10 , P. Follana11 , A. Lesoin12 , J. Medioni13 , V. Menguy14 , Y. Ghazi15 , F.C. Bidard1 , C. Dubot2 , F. Joly16 . 1 Institut Curie, Medical Oncology, Paris, France; 2 Institut Curie, Medical Oncology, Saint Cloud, France; 3 Gustave Roussy Cancer Campus, Medical Oncology, Villejuif, France; 4 Institut ´ Paoli Calmette, Medical Oncology, Marseille, France; 5 Institut Bergonie, Medical Oncology, Bordeaux, France; 6 CHU Tenon, Medical Oncology, 7 Paris, France; Centre Jean Perrin, Medical Oncology, Clermont Ferrand, France; 8 Centre Val D’Aurelle, Medical Oncology, Montpellier, France; 9 CHU Besan¸con, Medical Oncology, Besan¸con, France; 10 CHU Amiens, Medical Oncology, Amiens, France; 11 Centre Antoine Lacassage, Medical Oncology, Nice, France; 12 Centre Oscar Lambret, Medical Oncology, Lille, ˆ ´ France; 13 Hopital Europeen Georges Pompidou, Medical Oncology, Paris, France; 14 Roche SAS, Clinical Operations France, Boulogne Billancourt, 15 France; Roche SAS, Medical Affairs, Boulogne Billancourt, France; 16 Centre Fran¸cois Baclesse, Medical Oncology, Caen, France Background: Circulating tumor cells (CTC) are detected in 12−30% of advanced stage/relapsing ovarian carcinoma (OC) and are thought to indicate a poor prognosis. The dynamics of CTC counts during neoadjuvant chemotherapy is unknown. We evaluated CTC counts in patients (pts) from ANTHALYA (Roche), an open-label randomized, multicenter, phase II study evaluating bevacizumab plus neoadjuvant carboplatin-paclitaxel in pts with unresectable FIGO stage IIIc/IV ovarian, tubal or peritoneal adenocarcinoma. Patients and Methods: We obtained CTC counts before cycles 1 and 2 (C1, C2) and interval debulking surgery (IDS). CTC thresholds of 1 and 2 CTC/7.5 mL blood were considered. We report the dynamic profiles of CTC counts and their correlation with baseline characteristics and CA-125 levels (using Pearson correlation coefficient [R]). Results: All 95 included pts had CTC counts at one time point (88 pts at C1 and C2; 70 pts at IDS). 67 (70%) pts had FIGO stage IIIc, 84 (94%) pts were high grade, and 46 (59%) pts had a Fagotti score 8. 88 pts had CTC counts at 2 time points and 63 (72%) pts at 3 time points (full profile). Mean CTC count was 1.2 (±2.9) at C1, 0.5 (±1.0) at C2 and 0.6 (±3.2) at IDS. Maximum values were 15 at C1, 5 at C2, and 26 at IDS. A 0 CTC count was seen in 46 (52%) pts at each time point (“null” profile, see below). Conversely, 29 (33%) pts had a CTC count 1 at C1, 22 (25%) at C2, and 6 (9%) at IDS (Median and range are shown in table 1). A CTC count 2 was seen in 20 (23%) pts at C1, 9 (10%) at C2, and 6 (9%) at IDS. Baseline C1 CTC count and clinical characteristics were partly related: Stage IV

2773 POSTER ALLOCATE: sorting ovarian cancer patients into treatment categories based on genetic characteristics of their tumours O. Kondrashova1 , S. Lunke1 , L. Mileshkin2 , K. Alsop3 , C. Scott4 , A. Hamilton2 , S. Ananda5,6 , M. Quinn5 , D. Bowtell3 , O. McNally5 , T. Cowie1 , M. Wakefield1,4 , A. Hsu1 , G. Taylor1 , P. Waring1 . 1 University of Melbourne, Pathology, Melbourne, Australia; 2 The Peter MacCallum Cancer Centre, Department of Medical Oncology, Melbourne, Australia; 3 The Peter MacCallum Cancer Centre, Cancer Genetics and Genomics Laboratory, Melbourne, Australia; 4 Walter and Eliza Hall Institute, Stem Cells and Cancer, Melbourne, Australia; 5 The Royal Women’s Hospital, Gynaeoncology Service, Melbourne, Australia; 6 The Royal Melbourne Hospital, Medical Oncology, Melbourne, Australia