Purpose: Although 67% of high-grade serous ovarian cancers (HGSOC) express the estrogen receptor (ER), most fail antiestrogen therapy. Because MAPK activation is frequent in ovarian cancer, we investigated if estrogen regulates MAPK and if MEK inhibition (MEKi) reverses antiestrogen resistance.

Experimental Design: Effects of MEKi (selumetinib), antiestrogen (fulvestrant), or both were assayed in ER-positive HGSOC in vitro and in xenografts. Response biomarkers were investigated by gene expression microarray and reverse phase protein array (RPPA). Genes differentially expressed in two independent primary HGSOC datasets with high versus low pMAPK by RPPA were used to generate a “MAPK-activated gene signature.” Gene signature components that were reversed by MEKi were then identified.

Results: High intratumor pMAPK independently predicts decreased survival (HR, 1.7; CI > 95%,1.3–2.2; P = 0.0009) in 408 HGSOC from The Cancer Genome Atlas. A differentially expressed “MAPK-activated” gene subset was also prognostic. “MAPK-activated genes” in HGSOC differ from those in breast cancer. Combined MEK and ER blockade showed greater antitumor effects in xenografts than monotherapy. Gene set enrichment analysis and RPPA showed that dual therapy downregulated DNA replication and cell-cycle drivers, and upregulated lysosomal gene sets. Selumetinib reversed expression of a subset of “MAPK-activated genes” in vitro and/or in xenografts. Three of these genes were prognostic for poor survival (P = 0.000265) and warrant testing as a signature predictive of MEKi response.

Conclusions: High pMAPK is independently prognostic and may underlie antiestrogen failure. Data support further evaluation of fulvestrant and selumetinib in ER-positive HGSOC. The MAPK-activated HGSOC signature may help identify MEK inhibitor responsive tumors. Clin Cancer Res; 22(4); 935–47. ©2015 AACR.

Translational Relevance

Novel therapies are urgently needed for high-grade serous ovarian cancers (HGSOC). Although most HGSOC express the estrogen receptor (ER), responses to antiestrogen trials have been poor. We found high intratumor pMAPK by RPPA independently predicts decreased HGSOC survival. Combined MEK and ER blockade showed greater antiproliferative and antitumor effects in vitro and in xenografts than either drug alone. Gene enrichment analysis and RPPA showed dual therapy targets DNA replication, signaling, cell-cycle drivers, and cell death. We identified a gene signature characteristic of high pMAPK in HGSOC. This signature is different from MAPK-activated genes identified in breast cancer. A subset of the HGSOC high-pMAPK genes whose expression is reversed by MEK inhibitor predicts poor survival and warrants further testing as a predictor of MEKi response. High pMAPK is not only prognostic but may also underlie antiestrogen failure. Data support further evaluation of combined fulvestrant and selumetinib therapy in clinical trials for ER-positive HGSOC.

Ovarian cancer is the seventh leading cancer worldwide (1). Most patients present with advanced disease (2), and 80% relapse within 5 years (3), ultimately leading to death. Understanding tumor biology is critical for successful implementation of targeted therapies.

Molecular and epidemiologic data implicate estrogen as an ovarian cancer driver. Estrogen receptor (ER) protein was expressed in 67% of high-grade serous ovarian cancers (HGSOC; n = 338) from The Cancer Genome Atlas (TCGA; ref. 4), confirming findings of a meta-analysis of >2,000 cancers (5). Estrogen stimulates ovarian cancer (OVCA) proliferation in vitro and tumorigenicity in vivo (4, 6–8). Moreover, epidemiologic data implicate estrogens in OVCA risk. When initially reported, the prospective Women's Health Initiative (WHI) analysis of estrogen and progestin replacement showed a nonsignificant trend for increased OVCA risk (9). However, a recent, retrospective meta-analysis that included the WHI data showed a modest increase in ovarian cancer risk by 1.28 (95% CI, 1.20–1.36) in women who used hormone replacement therapy compared with never users (10).

Although ER protein is frequently expressed in OVCAs (4, 5), antiestrogens have been disappointing in the clinic. Despite their importance in breast cancer (11), the predictive value of hormone receptors and impact of postsurgical (adjuvant) endocrine therapy have not been well studied in OVCA. A review of 20 OVCA trials (695 patients) of the selective ER modulator, tamoxifen, showed a 13% overall response rate and 40% disease stabilization in heavily pretreated recurrent disease (12). The ER blocker, fulvestrant, yielded stable disease in 50% of heavily pretreated recurrent OVCA in a phase II trial (13). Aromatase inhibitors showed similar short-term efficacy (12). Most endocrine trials involved recurrent, heavily pretreated, platinum-resistant OVCA, and failed to ascertain tumor ER status. Antiestrogens have not been systematically prospectively evaluated and their role in advanced HGSOC may be underestimated because the responsive target population is not defined.

Estrogen mediates mitogenic effects in OVCAs by activating ER target genes including: cathepsin D (14), c-FOS (15), TGFα (16), c-MYC (17), and PGR (15). We showed that in addition to ER-mediated gene activation, estrogen stimulates ER:Src binding, and rapidly activates Src in OVCA cells via nongenomic ER action (4). The Src inhibitor, saracatinib, resensitized OVCA cells to ER blockade in vitro and showed synergistic antitumor effects in vivo in part through p27-mediated cell-cycle arrest (4).

The Ras/Raf/MEK/MAPK (MAPK cascade) is activated by gene copy number aberrations or mutations in HGSOC (18) and importantly regulates proliferation, survival, and chemoresistance (19). Selumetinib (AZD6244) is a noncompetitive MEK1/2 inhibitor that suppresses OVCA xenograft growth (20). A recent phase II selumetinib trial in recurrent low-grade serous OVCA showed a 15% response rate and 63% disease stabilization, surpassing the efficacy of cytotoxic therapies (21). MEK inhibitors (MEKi) have not been systematically evaluated in HGSOCs in the clinic to date.

Because ER activates oncogenic signaling in most OVCA, and because the MAPK cascade is frequently activated in HGSOCs (18), we investigated the therapeutic potential of combined ER blockade and selumetinib. Retrospective analysis of HGSOC showed MAPK activation is an independent poor prognostic factor. In vitro and in vivo studies show greater efficacy of combined MEK and ER blockade than either monotherapy. Gene expression and proteomic analysis showed dual therapy more profoundly affected DNA replication, cell cycle, signaling, and autophagy than either drug alone. Genes differentially expressed in high versus low pMAPK HGSOCs from two patient cohorts identified a highly prognostic “MAPK activation signature.” A subset of these genes were repressed by MEKi and warrant further investigation as potential predictors of MEKi responsiveness in the clinic.

Drugs

Selumetinib (AZD6244) and fulvestrant (both from Astra Zeneca) were dissolved in dimethyl sulfoxide (DMSO) and ethanol, respectively. Growth inhibitory selumetinib concentrations were titrated. For xenografts studies, selumetinib was suspended in 0.5% hydroxypropyl methyl cellulose and Tween 80.

Cell culture, cell-cycle analysis, and viability assays

OVCA lines PEO1R, an antiestrogen-resistant variant of PEO1 (4), and BG-1 (ref. 22; from Ken Korach, NIEHS, Research Triangle Park, NC) were cultured in RPMI with 10% FBS and authenticated per ATCC guidelines. OCI-E1P (Ovarian-Carcinoma-Ince-Endometrioid-Primary-1, or E1P) was cultured from a primary ER-positive OVCA in OCMI-L media (4, 23). OCMI-L medium and OCI-E1P are available from the Sylvester Comprehensive Cancer Center Live Tumor Culture Core at UM Miller School of Medicine, Miami, FL (contact email: LTCC@med.miami.edu). Asynchronous cultures were treated with vehicle, 10−6 M fulvestrant, selumetinib, or both for 48 hours. BG-1 was grown with 0.1% cFBS for 48 hours then treated with 10−8 M 17-β-estradiol (E2) ± selumetinib, fulvestrant, or both for 18 hours and analyzed by flow cytometry as in ref. 24.

OCI-E1P was treated with vehicle, selumetinib, fulvestrant, or both for 72 hours followed by CellTitre-Blue cell viability assays (Promega).

Immunoblotting, immunoprecipitation, kinase assay, and PCR

Westerns (20 μg protein/lane) were performed and quantitated by densitometry (24). Antibodies used were as follows: Cyclin E antibody (HE-12); ERα, Cdk2, MAPK, pMAPK from Santa Cruz; MEK and pMEK from Cell Signaling; p27 from BD transduction; and β-actin (Sigma).

Cyclin E was precipitated after 48 hours with either DMSO control, 200 nM selumetinib, 10−6 M fulvestrant or both drugs, and Cyclin E–Cdk2 complexes were immunoblotted for associated proteins or assayed for kinase activity and quantitated as described (24).

qRT-PCR was performed for AXL, ADRA2A, and ACTA2 as in ref. 4. For primer sets, see Supplementary Methods.

Tumor xenografts

Estradiol pellets (0.36 mg/90 days; Innovative Research) were implanted subcutaneously in all 5-week-old female NOD/SCID mice (Charles River). PEO1R cells (2 × 106 in 0.1 mL matrigel) were injected as in ref. 25 and mice were treated when tumors reached 70 to 100 mm3. Animals were weighed and tumor volume evaluated weekly. In a first experiment, selumetinib was titrated using 5, 20, and 40 mg/kg daily by oral gavage. To assay dual therapy, mice were treated in four groups (10 mice/group): (1) control (no drug), (2) selumetinib (oral gavage, 5 mg/kg/day), (3) fulvestrant weekly subcutaneous injection (3.75 mg/wk), and (4) both drugs. Mice were sacrificed 75 days post-implantation or when morbidity required euthanasia per IRB and Institutional Animal Use and Care Committee approved procedures. Tumors were recovered at sacrifice for immunohistochemistry (IHC) and RNA sequencing as described (see Supplementary Methods).

IHC of xenograft tumors

Xenograft tumors were immunostained for p27 (p27mAb, 1:500; BD Bioscience) and for Ki67 (Ki-67PAb, 1:500; Abcam) and hematoxylin counterstained as in ref. 26. Nuclei were scored for 80 to 120 cells from at least five high-power fields/tumor and mean % nuclei stained was calculated. Cleaved caspase-3 was detected with mAb (1:1,000; Cell Signaling).

Evaluation of pMAPK as a prognostic factor in HGSOC

The Cancer Proteome Atlas (TCPA) reports 178 proteins and phosphoproteins assayed by reverse phase protein arrays (RPPA). Kaplan–Meier (KM) survival curves were generated based on median expression of MAPKpT202pY204 (hereafter pMAPK) in RPPA data from TCGA/TCPA HGSOC cohort and survival distributions were compared by the log-rank test for all patients and the ER-positive subset. Univariate and multivariate models evaluated clinic–pathological variables: race, time of initial diagnosis, age at diagnosis, tumor grade, stage, size and residual disease location, patient age, and ER status using Cox proportional-hazards regression. Analyses used R software version 3.0.2 (www.Rproject.org) and coxph function from R “survival” package. For details see Supplementary Methods.

Defining a MAPK-activated gene signature in HGSOC

TCPA reports RPPA data for 408 of 485 untreated newly diagnosed HCSOC from The Cancer Genome Atlas (TCGA; ref. 27) and for 130 tumors from the 300 patient JAPAN HGSOC dataset (28). Patients were never treated with MEK inhibitors. HGSOC from TCGA/TCPA and JAPAN cohorts were classified by median pMAPK expression from RPPA as “high MAPK” or “low MAPK.” Gene expression differences between “high MAPK” and “low MAPK” HGSOCs were determined using the Student t test, and P values were permutation adjusted. Gene expression differences between “high MAPK” and “low MAPK” ovarian cancers were obtained using the R “Limma” package, and P values were false-discovery-rate adjusted.

One hundred and twenty-six unique genes were commonly differentially expressed (110 over and 16 underexpressed) between “high MAPK” and “low MAPK” TCGA and the Japan OVCA datasets. The 110 genes concordantly overexpressed in high-MAPK versus low-MAPK OVCA were compiled using Oncomine premium edition. Oncomine Concept analysis revealed significant association (P ≤ 0.01) between overexpression of these “poor outcome” genes in 11 analyses, and underexpression of these genes with one “poor outcome” analysis (representing five unique datasets; refs. 18, 29–32).

The top 20 MAPK-activated signature genes most significantly overexpressed in poor outcome cancers from six Oncomine OVCA datasets were further evaluated for prognostic significance in 485 TCGA HGSOCs by KM survival analysis using the log-rank test. In a leave-one-out analysis, individual genes were excluded from the ovarian cancer MAPK-activated gene signature, and clustering and survival analysis of TCGA OVCA was performed. Genes whose individual removal improved the prognostic significance of the signature were identified, and a leave-one-out signature was constructed. See also Supplementary Methods.

Gene array and RPPA analysis of PEO1R in vitro

RPPA and Illumina gene expression analyses were carried out on PEO1R with and without drug treatments for 48 hours in vitro from three biologic repeat samples. RPPA data processing used SuperCurve (SuperCurve Package R package version 1.4.1.2011) as described (27). Gene expression analysis of these samples used Affymetrix U133A GeneChips (12,042 genes) per MIAME guidelines of the Microarray Gene Expression Data Society as in ref. 18.

Statistics on in vitro and in vivo data

Cell-cycle, kinase assays, and IP/Western assays were done at least thrice and mean ± SEM graphed. Selumetinib dose titrations for OCI-E1P and BG-1 were duplicate assays. One- or two-way ANOVA tests assessed differences among means. For 2 × 2 factorial experiments, interactions were tested by two-way ANOVA followed by Tukey's honesty significance test, with P ≤ 0.05 indicating a significant difference. Statistical analysis used Graph Pad Prism software. Effects of dual or monotherapy on cell cycle and % viability were analyzed using the median-effect method of Chou and Talalay (33). For analysis of synergy between fulvestrant and selumetinib in vitro, combination index number (CIN) values were calculated using Calcusyn software (Biosoft). CIN values <1.0 indicate synergistic drug interaction.

Differences between drug effects on xenograft growth were compared by two-way ANOVA of mean tumor volumes weekly after 2 weeks. Analysis of potential synergy between fulvestrant and selumetinib on xenografts in vivo also used the combination ratio (34) comparing expected over observed fractional tumor volumes (see Supplementary Methods).

pMAPK is independently prognostic in HGSOCs

Of 408 HGSOCs from TCGA/TCPA, 293 (73%) were ER-positive and 115 (27%) failed to express detectable ER protein by RPPA. By RPPA, 76% (310/408) of HGSOC expressed activated, phosphorylated MAPK1 (MAPKpT202pY204 or pMAPK). To test the prognostic significance of pMAPK versus outcome, tumors were dichotomized by median RPPA pMAPK value as “high-pMAPK” or “low-pMAPK”. In all HGSOCs (Fig. 1A, left) and in the ER-positive HGSOC subset (Fig. 1A, right), “high pMAPK” cancers had a worse overall survival (OS) compared with “low pMAPK” HGSOC. Median OS was 39 months versus 52 months in high-pMAPK and low-pMAPK HGSOC, respectively (P = 0.0001 by log-rank test). Although MAPK hyperactivation in breast cancer is strongly associated with negative ER status (35, 36), this was not true in OVCA. pMAPK had a similar survival impact in ER-positive HGSOC, with a median OS of 40 months versus 52 months in high-pMAPK and low-pMAPK, respectively (P = 0.01 by log-rank test). High pMAPK, residual disease, macroscopic disease after surgery, and age were prognostic of poor OS upon multivariate analysis (Fig. 1B).

Figure 1.

pMAPK is an independent poor prognostic factor for HGSOC. A, KM curves show survival in high-grade serous OVCA with high pMAPK (above median pMAPK on RPPA) versus low pMAPK (n = 407; P = 1.47e–04, left). High pMAPK was also prognostic in the ER-positive OVCA subset (n = 293; P = 0.0113, right). B, multivariate analysis in this TCGA cohort.

Figure 1.

pMAPK is an independent poor prognostic factor for HGSOC. A, KM curves show survival in high-grade serous OVCA with high pMAPK (above median pMAPK on RPPA) versus low pMAPK (n = 407; P = 1.47e–04, left). High pMAPK was also prognostic in the ER-positive OVCA subset (n = 293; P = 0.0113, right). B, multivariate analysis in this TCGA cohort.

Close modal

Defining a gene signature in high pMAPK HGSOC

Differential gene expression analysis in high pMAPK versus low pMAPK HGSOC identified a gene signature characteristic of MAPK activation. In the 408 TCPA/TCGA HGSOC, gene expression analysis (GEA) identified 1,712 genes significantly, differentially expressed in high pMAPK versus low pMAPK cancers. This same stratification in a second HGSOC dataset, the Japan Ovarian Cohort (n = 130), identified 802 differentially expressed genes. A subset of 126 differentially expressed genes (110 up- and 16 down-regulated, P = 0.05), common to both datasets, defined a “MAPK-activated gene signature” (Fig. 2A and Supplementary Table S1).

Figure 2.

Activated pMAPK gene signature is prognostic in HGSOC. A, TCGA OVCA cohort (N = 408) and Japan Ovarian cohort (N = 130) were dichotomized based on median pMAPK expression, and genes differentially expressed and common to both are shown. B, the top 20 genes overexpressed in the MAPK-activated signature (left) were significantly overrepresented in patients with poor survival across 11 different analyses (#1–11) in Oncomine from five different ovarian cancer datasets (P < 0.01). (1) top 10% genes overexpressed in OVCA patients dead at 3 years (29); (2) top 10% genes overexpressed in OVCA patients dead at 1 year (29); (3) top 5% genes overexpressed in HGSOC dead at 3 years (18); (4) top 10% genes overexpressed in OCVA dead at 5 years (29); (5) top 10% genes overexpressed in HGSOC with recurrence at 1 year (30); (6) top 10% genes overexpressed in HGSOC dead at 1 year (18); (7) top 10% genes overexpressed in ovarian adenocarcinoma dead at 5 years (31); (8) top 10% genes overexpressed in ovarian adenocarcinoma recurrent at 5 years (31); (9) top 10% genes overexpressed in HGSOC dead at 1 year (30); (10) top 10% genes overexpressed in HGSOC dead at 5 years (18); (11) top 10% genes overexpressed in OVCA dead at 1 year (32). C, KM curves for OS of TCGA HGSOCs that overexpress the top 20 MAPK-activated signature genes (top graph; P = 0.0431); a leave-one-out analysis identified eight most prognostically significant pMAPK-activated genes (bottom graph, P = 7.1e-04).

Figure 2.

Activated pMAPK gene signature is prognostic in HGSOC. A, TCGA OVCA cohort (N = 408) and Japan Ovarian cohort (N = 130) were dichotomized based on median pMAPK expression, and genes differentially expressed and common to both are shown. B, the top 20 genes overexpressed in the MAPK-activated signature (left) were significantly overrepresented in patients with poor survival across 11 different analyses (#1–11) in Oncomine from five different ovarian cancer datasets (P < 0.01). (1) top 10% genes overexpressed in OVCA patients dead at 3 years (29); (2) top 10% genes overexpressed in OVCA patients dead at 1 year (29); (3) top 5% genes overexpressed in HGSOC dead at 3 years (18); (4) top 10% genes overexpressed in OCVA dead at 5 years (29); (5) top 10% genes overexpressed in HGSOC with recurrence at 1 year (30); (6) top 10% genes overexpressed in HGSOC dead at 1 year (18); (7) top 10% genes overexpressed in ovarian adenocarcinoma dead at 5 years (31); (8) top 10% genes overexpressed in ovarian adenocarcinoma recurrent at 5 years (31); (9) top 10% genes overexpressed in HGSOC dead at 1 year (30); (10) top 10% genes overexpressed in HGSOC dead at 5 years (18); (11) top 10% genes overexpressed in OVCA dead at 1 year (32). C, KM curves for OS of TCGA HGSOCs that overexpress the top 20 MAPK-activated signature genes (top graph; P = 0.0431); a leave-one-out analysis identified eight most prognostically significant pMAPK-activated genes (bottom graph, P = 7.1e-04).

Close modal

To identify prognostically important MAPK-activated genes, the “MAPK-activated gene signature” was evaluated in six independent ovarian cancer datasets from Oncomine with 11 distinct outcome analyses (18, 29–32). The top 20 most highly expressed MAPK-activated signature genes that were significantly associated with poor prognosis across five of six Oncomine datasets make up a subsignature of the MAPK-activated genes (P < 0.01; see table in Fig. 2B). Unsupervised hierarchical clustering of HGSOC from TCGA cohort based on expression of these 20 genes, and subsequent KM survival analysis, revealed that patients overexpressing the 20 gene “high-MAPK” subsignature had significantly worse OS (P = 0.0431; see Fig. 2C, top), as expected based on the significant association of high pMAPK protein expression with poor survival. A leave one out analysis further refined the most prognostic MAPK-activated genes, yielding 8 genes (AXL, PTGIS, RAB31, LRP1, DCN, CLN8, GPR124, and TIMP3) as the most robust outcome predictors (P = 0.00071; Fig. 2C, bottom). Median OS was 35 months versus 48 months in high-pMAPK and low-pMAPK HGSOC, respectively, as determined by this eight MAPK-activated gene signature (P = 0.0007 by log-rank test).

MAPK activation (high pMAPK by RPPA) is more frequent than mutational activation of the RAF/MEK/ERK cascade in HGSOC. Only 14% (45/312) of TCGA HGSOC show mutational activation of one or more of 42 MAPK pathway-associated genes (Supplementary Fig. S1). Analysis of high and low pMAPK HGSOCs showed MAPK-associated genes were overexpressed, amplified, and mutated at similar rates (Supplementary Table S1); however, the actual genes deregulated in high pMAPK cancers were of interest. High pMAPK HGSOC showed more frequent genetic activation of certain pathways including VEGF, PDGF, FGFR1, KDR, and the EGFR family, than the low pMAPK group (Supplementary Fig. S2). In high pMAPK HGSOC, FGFR1 was overexpressed in nearly 14%, and genetic changes that upregulate EGFR family member activity were frequent. No ESR1 mutations were found in these newly diagnosed HGSOC.

MEKi increases response to ER blockade in endocrine-resistant ER-positive ovarian cancer in vitro

The frequent MAPK activation in HGSOG indicates the potential for efficacy of MEK inhibitors (MEKi) in this disease. Because MEKi monotherapy leads to bypass pathway activation and drug resistance (37), MEKi may be most clinically useful when combined with other drugs. Because MAPK is frequently activated in ER-positive HGSOC, we posited that this may underlie resistance of ER-positive OVCA to antiestrogen therapies. We thus tested if MEKi would increase response to ER blockade in ER-positive OVCA models in vitro. PEO1R is an ER-positive, antiestrogen-resistant HGSOC line (4) with high ER and low pMAPK levels (Supplementary Fig. S3A). OCI-E1P is a weakly ER-positive, primary ovarian cancer culture with high pMAPK (Supplementary Fig. S3A). Cell-cycle effects of the MEKi, selumetinib (AZD6244) were titrated on PEO1R. Selumetinib caused dose-dependent pMAPK inhibition, increased the Cdk2 inhibitor, p27, and induced G0–G1 arrest in PEO1R over 48 hours (100–250 nM; Supplementary Fig. S3B and S3C). Early passage OCI-E1P showed de novo selumetinib resistance, with concentrations up to 1,000 nM causing little cell-cycle effect (Supplementary Fig. S3D).

Our prior work showed estrogen activates Src in OVCA cells (4). Estrogen also stimulates MEK/MAPK activation in PEO1R (Fig. 3A). Asynchronous PEO1R and OCI-E1P were treated with selumetinib, fulvestrant, or both for 48 hours. PEO1R cell cycle was unaffected by fulvestrant (10−6 M), partly inhibited by 200 nM selumetinib (S-phase 52% in controls vs. 25%, P ≤ 0.001) and more inhibited by combination treatment, with the %S-phase decreasing from 51% to 13% over 48 hours; P < 0.001) compared with untreated cells. Synergistic cell-cycle inhibition by these two drugs was demonstrated by a CIN of 0.811. Dual therapy also had the greatest inhibitory effect on pMAPK (Fig. 3B and C). p27 levels increased more with combination than with either monotherapy (1.46 ± 0.12-fold with both vs. 1.0 ± 0.005-fold with fulvestrant and 1.2 ± 0.05-fold with selumetinib; Fig. 3C, top). Densitometry showed Cyclin E-Cdk2 bound p27 increased little (1.05 ± 0.007-fold) with fulvestrant alone, but increased 1.2 ± 0.07-fold with selumetinib alone (P < 0.05), and 1.36 ± 0.09-fold with dual therapy (P < 0.04; Fig. 3C, bottom). Although fulvestrant alone failed to inhibit Cyclin E-Cdk2, both drugs together caused greater kinase inhibition than either alone (Fig. 3D).

Figure 3.

Dual therapy effects on signaling and cell cycle. A, Western in PEO1R before and 1 hour after E2 addition (10−8 M). B–D, asynchronous PEO1R treated with fulvestrant (Fulv 10−6 M), selumetinib (MI, 200 nM), or both for 48 hours and recovered for cell-cycle analysis. B, the cell-cycle effect of dual therapy was synergistic (CIN 0.81 for mean % S phase, **** for P < 0.0001 for dual therapy vs. control and *** for P < 0.001 for either monotherapy vs. dual therapy). C, Western analysis (top). Cyclin E bound p27 protein (bottom). D, Cyclin E precipitates were assayed for kinase activity and radioactivity in Histone H1 substrate is graphed as % max activity, * P < 0.001 for dual therapy versus either drug alone. E, asynchronous OCI-E1P cells treated with Fulv (10−6 M), selumetinib (MI 200 nM), or both for 72 hours were assayed with CellTiter-Blue cell viability assay; * for P < 0.05. F, Western blot after 48 hours of drug treatment in OCI-E1P. All data are from ≥3 biologic assays and graphed as mean ± SEM or representative data shown.

Figure 3.

Dual therapy effects on signaling and cell cycle. A, Western in PEO1R before and 1 hour after E2 addition (10−8 M). B–D, asynchronous PEO1R treated with fulvestrant (Fulv 10−6 M), selumetinib (MI, 200 nM), or both for 48 hours and recovered for cell-cycle analysis. B, the cell-cycle effect of dual therapy was synergistic (CIN 0.81 for mean % S phase, **** for P < 0.0001 for dual therapy vs. control and *** for P < 0.001 for either monotherapy vs. dual therapy). C, Western analysis (top). Cyclin E bound p27 protein (bottom). D, Cyclin E precipitates were assayed for kinase activity and radioactivity in Histone H1 substrate is graphed as % max activity, * P < 0.001 for dual therapy versus either drug alone. E, asynchronous OCI-E1P cells treated with Fulv (10−6 M), selumetinib (MI 200 nM), or both for 72 hours were assayed with CellTiter-Blue cell viability assay; * for P < 0.05. F, Western blot after 48 hours of drug treatment in OCI-E1P. All data are from ≥3 biologic assays and graphed as mean ± SEM or representative data shown.

Close modal

Dual MEK and ER blockade also decreased early passage primary ovarian cancer OCI-E1P viability. Although each monotherapy caused a modest but not significant viability loss (25% for fulvestrant, P > 0.05, and 30% for selumetinib, P > 0.05 vs. control), both together synergistically decreased viability (60% reduction vs. control, P < 0.05; CIN = 0.33; see Fig. 3E). Similarly, p27 increased most with dual therapy in OCI-E1P (Fig. 3F).

The BG-1 OVCA line, derived from a primary ER-positive HGSOC, was resistant to 250 to 1,000 nmol/L selumetinib more than 48 hours (Supplementary Fig. S4A). When BG-1 was serum and estradiol (E2)-deprived in 0.1% cFBS for 48 hours, addition of 10−8 M E2 increased %S phase from 15% to 35% within 18 hours (Supplementary Fig. S4). Thus, BG-1 is estrogen sensitive. Fulvestrant or selumetinib each modestly decreased S-phase entry after E2 stimulation of E2-starved cells. Both drugs together synergistically inhibited S-phase entry (11% compared with 35%, P < 0.001, with a CIN of 0.93).

Gene expression arrays show dual therapy affects cell cycle, DNA replication, FOXM1 and autophagy pathways

To further evaluate differences between dual and monotherapies, gene expression profiles were assayed in PEO1R before and after 48 hours selumetinib, fulvestrant, or both. Of 11,647 genes assayed, fulvestrant altered only a small subset, MEKi a larger subset, whereas combination therapy generated the greatest differences compared with controls (Fig. 4A). Gene set enrichment plots show genes governing cell cycle, DNA replication, and the FOXM1 pathway were the top three gene sets downregulated and autophagy genes were most upregulated by combination therapy (Fig. 4B and C). The top 10 most significant genes in each gene set are illustrated in heat maps (Fig. 4B).

Figure 4.

Gene expression analysis in treated PEO1R. A–D, asynchronous PE01R were not treated (control, C) or were treated with fulvestrant (10−6 M), selumetinib (200 nM), or both for 48 hours, in vitro. A, clustering analysis shows genes differentially expressed from among 11,647 genes, P < 0.05. B, cell cycle, DNA replication, and FOXM1 pathways were among the top 20 gene sets downregulated; the autophagy gene set was among the top 10 gene sets upregulated. Heat maps were generated using R package “ggplot2,” P < 0.05. Significant gene names are represented on the x-axis. C, gene set enrichment plots of cell cycle, DNA replication, FOXM1, and autophagy compare combination therapy versus controls with normalized enrichment score (NES) and false discovery rate (FDR) Q values shown in respective plots. D, volcano plots of differentially expressed individual genes between treatment groups versus control and combination therapy versus MEK inhibition only. Gray = adjusted P value < 0.001; blue = adjusted P value < 0.001; Log2 FC ≤ −1 (down > 2-fold); red = adjusted P-value < 0.001; Log FC change >1 (up > 2-fold); and black = genes not differentially expressed between comparison groups.

Figure 4.

Gene expression analysis in treated PEO1R. A–D, asynchronous PE01R were not treated (control, C) or were treated with fulvestrant (10−6 M), selumetinib (200 nM), or both for 48 hours, in vitro. A, clustering analysis shows genes differentially expressed from among 11,647 genes, P < 0.05. B, cell cycle, DNA replication, and FOXM1 pathways were among the top 20 gene sets downregulated; the autophagy gene set was among the top 10 gene sets upregulated. Heat maps were generated using R package “ggplot2,” P < 0.05. Significant gene names are represented on the x-axis. C, gene set enrichment plots of cell cycle, DNA replication, FOXM1, and autophagy compare combination therapy versus controls with normalized enrichment score (NES) and false discovery rate (FDR) Q values shown in respective plots. D, volcano plots of differentially expressed individual genes between treatment groups versus control and combination therapy versus MEK inhibition only. Gray = adjusted P value < 0.001; blue = adjusted P value < 0.001; Log2 FC ≤ −1 (down > 2-fold); red = adjusted P-value < 0.001; Log FC change >1 (up > 2-fold); and black = genes not differentially expressed between comparison groups.

Close modal

Drug effects on differential expression of individual genes and differences between groups are shown by volcano plots (Fig. 4D). These show individual differentially expressed genes based on a false discovery rate of P ≤ 0.001 (genes represented in grey) and those that were differentially expressed with a log fold change >1 (represented in red) or <−1 (represented in blue). Genes in black were not differentially expressed. Only seven genes were differentially expressed with fulvestrant alone, 1,413 with MEKi alone, and 1,852 genes following combination therapy.

Expression of key drivers of cell cycle (MYC, PCNA, MCM3, MCM4, and E2F2) and mitogenic signaling were more affected by combination therapy than either monotherapy. ER-regulated genes GAL, CA2, and HSPSC111 were also maximally affected by dual therapy (Supplementary Fig. S5).

Proteomic analysis shows dual therapy affects cell cycle, signaling, and apoptosis

To determine if selumetinib effects on gene expression led to protein changes, and to assay consequences on signaling activities, drug effects were assayed by RPPA. RPPA confirmed cell-cycle effects, and showed dual therapy inhibited mitogenic signaling and activated cell death more than either monotherapy. Combination therapy over 48 hours markedly increased CDK inhibitors p21 and p27, decreased cell-cycle activators, and reduced total or phosphorylated levels of signaling kinases and targets more than either drug alone (log fold change × 100 vs. controls). Apoptotic mediators were increased and antiapoptotic Bcl2 was most powerfully reduced by dual therapy (Fig. 5). Representative RPPA data are also presented as heat maps in Supplementary Fig. S6, to show variability within triplicate repeat assays.

Figure 5.

RPPA shows greater effect of dual than monotherapy on cell cycle, signaling, and apoptosis. PE01R controls, C, were treated with fulvestrant (10−6 M), selumetinib (200 nM), or both for 48 hours in vitro and recovered for RPPA. Normalized RPPA data were used for complete clustering analysis. The bar plots represent protein expression differences between groups compared with controls, expressed as log fold change. The top 20 downregulated or upregulated proteins were identified and classified by function, and representative proteins are shown (see also Supplementary Fig. S6 for variability within samples).

Figure 5.

RPPA shows greater effect of dual than monotherapy on cell cycle, signaling, and apoptosis. PE01R controls, C, were treated with fulvestrant (10−6 M), selumetinib (200 nM), or both for 48 hours in vitro and recovered for RPPA. Normalized RPPA data were used for complete clustering analysis. The bar plots represent protein expression differences between groups compared with controls, expressed as log fold change. The top 20 downregulated or upregulated proteins were identified and classified by function, and representative proteins are shown (see also Supplementary Fig. S6 for variability within samples).

Close modal

Combination fulvestrant and selumetinib most effectively reduced ovarian cancer xenograft growth

Antitumor efficacy of selumetinib, fulvestrant, or both was assayed in PEO1R xenografts in vivo. MEKi or fulvestrant alone each modestly decreased in tumor growth over time versus controls. Growth inhibition was greater with dual therapy than with either monotherapy (ANOVA P < 0.001 after week 3; Fig. 6A). Potential drug synergy in vivo was analyzed using the combination ratio of expected over observed fractional tumor volumes (FTV). A combination ratio greater than 1 indicates a synergistic effect. The combination ratio was 1.58 at week 1 and rose to 1.68 at week 3, indicating synergistic antitumor effects of fulvestrant and selumetinib (see Supplementary Table S3). Animal weights did not differ significantly between groups, demonstrating tolerability of combination therapy (data not shown). One animal in each of the dual therapy and MEK inhibitor monotherapy groups died without evidence of prior ill health in the second last week of the experiment.

Figure 6.

Selumetinib and fulvestrant have synergistic antitumor activity and reverse MAPK-activated OVCA signature genes. A, treatment effects on PEO1R xenograft mean tumor volume ± SEM. ANOVA showed tumor volumes differed significantly in dual therapy versus controls at all times after 2 weeks, **** for P < 0.0001, and also differed significantly between dual therapy versus either drug alone, *** for P < 0.001 at all times after week 3 (see also Supplementary Table S2). B, the % xenograft nuclei positive for p27 and Ki67 by IHC graphed as mean ± SEM, * for P < 0.05, ** for P < 0.01, *** for P < 0.001. C, representative IHC (40× magnification). D, table of MAPK-activated signature genes altered by MEKi in PEO1R xenografts in vivo or after 48 hours in vitro, ranked by log fold change after MEKi alone or in combination. E, changes in expression of MAPK-activated genes in PE01R after MEKi treatment detected in vivo or in vitro were validated by RT-PCR and triplicate assays graphed as shown. F, KM curve showing high ADRA2A, ACTA2, and AXL expression levels associate with worse OS in 485 TCGA HGSOC, P = 2.65e-04.

Figure 6.

Selumetinib and fulvestrant have synergistic antitumor activity and reverse MAPK-activated OVCA signature genes. A, treatment effects on PEO1R xenograft mean tumor volume ± SEM. ANOVA showed tumor volumes differed significantly in dual therapy versus controls at all times after 2 weeks, **** for P < 0.0001, and also differed significantly between dual therapy versus either drug alone, *** for P < 0.001 at all times after week 3 (see also Supplementary Table S2). B, the % xenograft nuclei positive for p27 and Ki67 by IHC graphed as mean ± SEM, * for P < 0.05, ** for P < 0.01, *** for P < 0.001. C, representative IHC (40× magnification). D, table of MAPK-activated signature genes altered by MEKi in PEO1R xenografts in vivo or after 48 hours in vitro, ranked by log fold change after MEKi alone or in combination. E, changes in expression of MAPK-activated genes in PE01R after MEKi treatment detected in vivo or in vitro were validated by RT-PCR and triplicate assays graphed as shown. F, KM curve showing high ADRA2A, ACTA2, and AXL expression levels associate with worse OS in 485 TCGA HGSOC, P = 2.65e-04.

Close modal

IHC showed dual therapy increased p27 and decreased Ki67 more than either drug alone and all values differed from untreated controls (Fig. 6B and C). Thus, prolonged xenograft treatment in vivo has the same effect as short-term treatment in vitro to inhibit cell cycle through upregulation of p27. Notably, caspase-3 activation was minimal (3–5% of cells) in all xenograft groups, thus not all changes observed with short-term treatment in vitro are observed in vivo.

Defining a MAPK activation signature predictive of MEKi response in vitro and in vivo

Prognostic factors are of less value than predictive factors in patients with limited treatment options. Although high pMAPK is prognostic of poor HGSOC outcome, phosphoprotein analysis is challenging in the clinic. Having identified a prognostic MAPK-activated gene signature for HGSOC, we next sought to identify a gene signature that might predict MEK inhibitor response. We tested if any MAPK-activated signature genes were altered by short-term in vitro therapy with MEKi alone or with fulvestrant or after prolonged xenograft treatment in vivo. Genes most significantly affected by MEKi or by dual therapy are indicated in Fig. 6D. Notably, three genes that were reduced by MEKi in vitro or in vivo in PEO1R, ADRA2, ACTA2, and AXL (Fig. 6D), were also present in the top 20 MAPK-activated prognostic signature genes in HGSOC (Fig. 2). RT-PCR confirmed their repression by MEKi and also that of SRPX2 (Fig. 6E). SRPX2 is a MEKi responsive gene, but was not prognostic in Fig. 2. Because ADRA2A, ACTA2, and AXL were all reduced by MEKi and were of potential clinical significance as part of the top 20 MAPK-activated gene signature, we examined the prognostic potential of this three gene subsignature. Hierarchical clustering revealed ADRA2A, ACTA2, and AXL overexpression in TCGA HGSOC correlated with significantly poorer survival (P = 0.000265; Fig. 6F), with a median OS of 32 months with high versus 45 months with low AXL, ADRA2A, and ACTA2 expression, respectively (P = 0.0002). Thus, these three MAPK signature genes are downregulated by MEKi and also correlate with poor patient outcome. Whether these genes identify responsiveness to MEK inhibition in the clinic warrants further prospective evaluation.

OVCA survival has improved minimally over the last decade (3) and better, rationally targeted treatments are clearly needed. Despite evidence that estrogens increase OVCA risk (10, 38), drive OVCA proliferation (4, 6, 12, 39), and that most HGSOCs express ERα (4, 5), drug resistance limits the benefit of endocrine therapy for this cancer (12). The current TCGA analysis shows 74% of HGSOCs are ER-positive, similar to prior reports for all OVCA (5) and HGSOC (4).

Antiestrogens have been used primarily in heavily pretreated, recurrent OVCA, usually without evaluation of ER status. Tamoxifen in this setting showed modest activity (13% ORR and 35% SD rate) with higher responses (OR, 26%) earlier in treatment (12). Few studies have tested whether antiestrogens add to the benefit of chemotherapy after surgery in HGSOC. One such study showed no benefit of tamoxifen given with chemotherapy for stage III/IV patients; however, tamoxifen was given for only 36 weeks after surgery (40). Endocrine therapy for prevention or treatment of HGSOC recurrence has not been systematically evaluated and its clinical application has been hampered by lack of characterization of ER expression and ER-activated pathways in OVCA.

The frequent activation of the MEK/MAPK pathway by oncogenic receptor tyrosine kinase (RTK) activation, or mutation of RAS, RAF, or downstream kinases in cancers has stimulated development of MEK inhibitors, several of which are currently in clinical trials (19, 37, 41, 42). Despite limited efficacy in unselected cancers, single-agent MEKi activity is seen in malignancies harboring RAS or B-RAF mutations, such as melanoma (37). Although low-grade serous ovarian cancers account for only 6% of all epithelial OVCAs, four MEK1/2 inhibitors are currently being evaluated (clinical trials.gov), because 68% of these tumors have activating B-RAF or K-RAS mutations (43). Notably, in the one phase II trial reporting single-agent selumetinib efficacy in recurrent low-grade OVCA (15% overall response and 63% stable disease, with acceptable toxicity), response was independent of gene mutation status (21). There have been no MEKi trials in HGSOCs reported to date and these account for >80% of epithelial OVCAs.

Here we show HGSOCs from TCGA show frequent MAPK pathway activation, and high pMAPK by RPPA is independently prognostic for poor survival upon multivariate analysis. Furthermore, 14% of HGSOCs have gene mutations activating the MEK/ERK pathway. Although the rates of mutation, amplification, and overexpression of 42 different MAPK-associated genes did not differ significantly between high and low MAPK HGSOC groups, pMAPK-activated cancers showed more frequent genetic activation of certain MAPK driver pathways, including VEGFA, PDGF, FGFR, KDR, and the EGFR family, than did the low pMAPK group. Notable among these were the >14% BRAF amplification and overexpression, 17% FGFR overexpression or FGF1 amplification, and frequent overexpression, amplification, and mutation of genes encoding multiple EGFR family members in high MAPK-activated cancers.

Present data also demonstrate that estrogen stimulates MAPK activation in both the established PEO1R ovarian line and the primary patient tumor culture, OCI-E1P. Selumetinib had antiproliferative activity in PEO1R, a line which lacks activating BRAF or KRAS mutations (44). As observed in our ER-positive OVCA models, MAPK activation in ER-positive OVCA may arise in part through cross-talk with ER signaling. Notably, despite their relative resistance to selumetinib monotherapy, selumetinib combined with fulvestrant significantly affected both the OCI-EIP primary culture and BG-1.

Although mechanisms of antiestrogen resistance have been extensively studied in breast cancer (24, 45–47), few studies have focused on HGSOC (4, 12). In breast cancer, endocrine resistance can result from cross-talk between activated RTKs and ER signaling (46, 48). HER2/MAPK and estrogen signaling cross-talk is also reported in OVCA (49). We showed estrogen stimulates, ER-pSrc binding, and Src activation, revealing cross-talk between these pathways in OVCA (4). Targeting Src restored antiestrogen sensitivity (4). Antiestrogen responses were restored by MEK inhibition with re-expression of ERα in a subset of ER-negative breast cancer cells (35). Similarly, MEK inhibition with PD0325901 increased ER, sensitizing the SKOV3 OVCA line to ER blockade (50).

Present data indicate that resistance to ER blockade is effectively reversed by targeting MEK with selumetinib in vitro and in vivo. Selumetinib cooperates with fulvestrant to arrest growth of ER-positive OVCA lines (PEO1R and BG-1) and the primary human ovarian cancer culture OCI-EIP through p27 mediated cell-cycle arrest, as in breast cancer (24, 45). Validation in primary tumor OCI-EIP cells strengthens this study, because most OVCA lines are genomically unlike HGSOCs from TCGA (23, 51), and OCI-EIP mirrors its parent tumor genomically (23). Most importantly, combination therapy proved superior to either monotherapy in vivo: selumetinib and fulvestrant synergistically inhibited PEO1R tumor growth, supporting evaluation of this combination in future clinical trials. Xenograft analysis showed p27 was increased and Ki67 reduced most by dual therapy, revealing long-term drug effects to arrest growth in vivo are similar to short-term effects in culture.

The enhanced growth arrest and antitumor activity of dual therapy in vitro and in vivo was reflected in changes in both gene expression and proteomic analysis. Gene sets for cell-cycle progression, DNA replication, and FOXM1 were downregulated, and autophagy pathway genes were upregulated by combined ER/MEK blockade compared with control. MEKi treatment was the dominant driver in decreasing gene expression (1,174 genes affected). The addition of fulvestrant to MEKi decreased another 78 genes reflecting potential cross-talk between ER and MAPK signaling. RPPA analysis confirmed that cell cycle and signaling mediators were affected more profoundly by dual therapy than by either drug alone.

Our investigation revealed that pMAPK is a powerful independent prognostic factor in HGSOC. Emerging data support the use of proteomic profiling to identify molecular signatures of drug response (52), and tests to detect phospho-proteins are moving toward clinical application (53). In an effort to define a clinically usable tool, we further evaluated genes differentially expressed between “high” and “low” pMAPK HGSOCs. One hundred and twenty-six genes were differentially expressed between low and high pMAPK cancers from two independent HGSOC cohorts and defined a pMAPK-activated gene signature. Of this 126 pMAPK-activated gene signature, genes most highly overexpressed in poor prognostic tumors from five of six independent tumor-derived OVCA datasets from Oncomine (18, 29–32) defined a simplified 20 gene prognostic pMAPK signature and a leave one out analysis further refined the profile to an eight gene subset. Interestingly, the 126 pMAPK-activated HGSOC gene signature does not overlap significantly with a hMAPK gene profile of known and validated MAPK-regulated genes previously identified in breast cancer (36, 54). Indeed, the breast cancer hMAPK gene signature (36, 54) was not prognostic in HGSOCs (data not shown). Thus, MAPK signaling in different hormonally responsive cancers appears to drive distinct cancer specific downstream gene expression patterns, and may reflect different upstream MAPK activators.

A predictive test that can proactively identify patients most likely to respond to MEKi therapy might be of greater potential utility than a prognostic marker. Unlike melanoma, where B-RAF mutation predicts response to MEK inhibitors, selumetinib response in a recent phase II trial was not linked to these mutations in low grade serous OVCA (21). To develop a gene set that might predict MEKi responsiveness, we identified MAPK-activated genes that are reversed by MEKi both in vitro and in xenografts. Most MAPK-activated genes downregulated by short-term MEKi therapy differed from those downregulated by prolonged treatment in vivo. SRPX2 alone, which promotes tumor migration and is a poor prognostic factor in gastric cancer (55), was downregulated by both short- and long-term MEKi treatment. MEKi mediated changes in AXL, ACTA2, ADRA2A, and SRPX2 expression were validated by PCR. AXL is a RTK that promotes OVCA metastasis (56). ACTA2 (alpha smooth muscle actin) drives metastasis and poor prognostis in lung cancer (57). ADRA2A (adrenoreceptor alpha 2A) stimulates proliferation and is associated with MAPK activation (58). Notably, elevated expression of these three MAPK-activated genes (AXL, ACTA2, ADRA2A) whose expression is reversed by selumetinib was strongly predictive of poor patient outcome (P = 0.000265). These highly prognostic genes that are upregulated in MAPK-activated HGSOCs and downregulated with MEKi in the PEO1R model provide a three gene set for further validation in vitro that could be evaluated prospectively in clinical trials as a signature predictive of clinical MEKi response.

In summary, this work suggests that MEK inhibitor utility in OVCA may not be limited to low-grade cancers. HGSOCs in TCGA show frequent MAPK activation by RPPA and high pMAPK is a strong independent prognostic factor for OS on multivariate analysis (P = 0.0009187). We have identified a prognostic pMAPK-activated OVCA gene signature and a gene subset whose predictive potential for response to MEK inhibition warrants further study. The median survival of patients in both low and high MAPK HGSOC groups treated with standard therapy is short, underlining the need for novel treatment approaches. Present work supports clinical evaluation of combined fulvestrant and MEKi for ER-positive HGSOCs, and raises the intriguing possibility that further MEKi combinations with FGFR, PDGFR, or pan-EGFR inhibitors may warrant investigation.

No potential conflicts of interest were disclosed.

Conception and design: K.E. Hew, P.C. Miller, D. El-Ashry, G. Zhang, J.M. Slingerland, F. Simpkins

Development of methodology: K.E. Hew, P.C. Miller, J. Sun, Z. Wei, G.B. Mills, J.M. Slingerland, F. Simpkins

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Sun, A.H. Besser, T.A. Ince, W. Gao, F. Simpkins

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.E. Hew, P.C. Miller, D. El-Ashry, M. Gu, Z. Wei, G. Zhang, P. Brafford, W. Gao, G.B. Mills, J.M. Slingerland, F. Simpkins

Writing, review, and/or revision of the manuscript: K.E. Hew, P.C. Miller, D. El-Ashry, Z. Wei, G. Zhang, Y. Lu, G.B. Mills, J.M. Slingerland, F. Simpkins

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.E. Hew, P.C. Miller, Z. Wei, G.B. Mills, F. Simpkins

Study supervision: Z. Wei, J.M. Slingerland, F. Simpkins

Other (performed reverse phase protein array analysis on this study): Y. Lu

Other (obtained project funding): J.M. Slingerland, F. Simpkins

The authors thank Amanjot Riar for ANOVA analysis, Toni Yeasky for assistance with xenograft experiments and IHC analysis, and the Slingerland and El-Ashry laboratories for helpful discussions.

This work was supported by the Gynecological Cancer Foundation Mary-Jane Welker Ovarian Cancer Research Grant, 7K08CA151892-04.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

2.
Ozols
RF
,
Bundy
BN
,
Greer
BE
,
Fowler
JM
,
Clarke-Pearson
D
,
Burger
RA
, et al
Phase III trial of carboplatin and paclitaxel compared with cisplatin and paclitaxel in patients with optimally resected stage III ovarian cancer: a Gynecologic Oncology Group study
.
J Clin Oncol
2003
;
21
:
3194
200
.
3.
Winter
WE
 III
,
Maxwell
GL
,
Tian
C
,
Carlson
JW
,
Ozols
RF
,
Rose
PG
, et al
Prognostic factors for stage III epithelial ovarian cancer: a Gynecologic Oncology Group Study
.
J Clin Oncol
2007
;
25
:
3621
7
.
4.
Simpkins
F
,
Hevia-Paez
P
,
Sun
J
,
Ullmer
W
,
Gilbert
CA
,
da Silva
T
, et al
Src inhibition with saracatinib reverses fulvestrant resistance in ER-positive ovarian cancer models in vitro and in vivo
.
Clin Cancer Res
2012
;
18
:
5911
23
.
5.
Rao
BR
,
Slotman
BJ
. 
Endocrine role in ovarian cancer
.
Endocr Relat Cancer
1996
;
3
:
309
26
.
6.
Langdon
SP
,
Crew
AJ
,
Ritchie
AA
,
Muir
M
,
Wakeling
A
,
Smyth
JF
, et al
Growth inhibition of oestrogen receptor-positive human ovarian carcinoma by anti-oestrogens in vitro and in a xenograft model
.
Eur J Cancer
1994
;
30A
:
682
6
.
7.
Langdon
SP
,
Hirst
GL
,
Miller
EP
,
Hawkins
RA
,
Tesdale
AL
,
Smyth
JF
, et al
The regulation of growth and protein expression by estrogen in vitro: a study of 8 human ovarian carcinoma cell lines
.
J Steroid Biochem Mol Biol
1994
;
50
:
131
5
.
8.
Armaiz-Pena
GN
,
Mangala
LS
,
Spannuth
WA
,
Lin
YG
,
Jennings
NB
,
Nick
AM
, et al
Estrous cycle modulates ovarian carcinoma growth
.
Clin Cancer Res
2009
;
15
:
2971
8
.
9.
Anderson
GL
,
Judd
HL
,
Kaunitz
AM
,
Barad
DH
,
Beresford
SA
,
Pettinger
M
, et al
Effects of estrogen plus progestin on gynecologic cancers and associated diagnostic procedures: the Women's Health Initiative randomized trial
.
JAMA
2003
;
290
:
1739
48
.
10.
Beral
V
. 
Ovarian cancer and hormone replacement therapy in the Million Women Study
.
Lancet
2007
;
369
:
1703
10
.
11.
Jordan
VC
. 
Selective estrogen receptor modulation: concept and consequences in cancer
.
Cancer Cell
2004
;
5
:
207
13
.
12.
Simpkins
F
,
Garcia-Soto
A
,
Slingerland
J
. 
New insights on the role of hormonal therapy in ovarian cancer
.
Steroids
2013
;
78
:
530
7
.
13.
Argenta
PA
,
Thomas
SG
,
Judson
PL
,
Downs
LS
 Jr.
,
Geller
MA
,
Carson
LF
, et al
A phase II study of fulvestrant in the treatment of multiply-recurrent epithelial ovarian cancer
.
Gynecol Oncol
2009
;
113
:
205
9
.
14.
Rochefort
H
,
Chalbos
D
,
Cunat
S
,
Lucas
A
,
Platet
N
,
Garcia
M
. 
Estrogen regulated proteases and antiproteases in ovarian and breast cancer cells
.
J Steroid Biochem Mol Biol
2001
;
76
:
119
24
.
15.
Albanito
L
,
Madeo
A
,
Lappano
R
,
Vivacqua
A
,
Rago
V
,
Carpino
A
, et al
G protein-coupled receptor 30 (GPR30) mediates gene expression changes and growth response to 17beta-estradiol and selective GPR30 ligand G-1 in ovarian cancer cells
.
Cancer Res
2007
;
67
:
1859
66
.
16.
Simpson
BJB
,
Langdon
SP
,
Rabiasz
GJ
,
Macleod
KG
,
Hirst
GL
,
Bartlett
JMS
, et al
Estrogen regulation of transforming growth factor-alpha in ovarian cancer
.
J Steroid Biochem Mol Biol
1998
;
64
:
137
45
.
17.
Chien
CH
,
Wang
FF
,
Hamilton
TC
. 
Transcriptional activation of c-myc proto-oncogene by estrogen in human ovarian cancer cells
.
Mol Cell Endocrinol
1994
;
99
:
11
9
.
18.
Cancer Genome Atlas Research Network
. 
Integrated genomic analyses of ovarian carcinoma
.
Nature
2011
;
474
:
609
15
.
19.
De
LA
,
Maiello
MR
,
D'Alessio
A
,
Pergameno
M
,
Normanno
N
. 
The RAS/RAF/MEK/ERK and the PI3K/AKT signalling pathways: role in cancer pathogenesis and implications for therapeutic approaches
.
Expert Opin Ther Targets
2012
;
16
Suppl 2
:
S17
27
.
20.
Bartholomeusz
C
,
Oishi
T
,
Saso
H
,
Akar
U
,
Liu
P
,
Kondo
K
, et al
MEK1/2 inhibitor selumetinib (AZD6244) inhibits growth of ovarian clear cell carcinoma in a PEA-15-dependent manner in a mouse xenograft model
.
Mol Cancer Ther
2012
;
11
:
360
9
.
21.
Farley
J
,
Brady
WE
,
Vathipadiekal
V
,
Lankes
HA
,
Coleman
R
,
Morgan
MA
, et al
Selumetinib in women with recurrent low-grade serous carcinoma of the ovary or peritoneum: an open-label, single-arm, phase 2 study
.
Lancet Oncol
2013
;
14
:
134
40
.
22.
Geisinger
KR
,
Kute
TE
,
Pettenati
MJ
,
Welander
CE
,
Dennard
Y
,
Collins
LA
, et al
Characterization of a human ovarian carcinoma cell line with estrogen and progesterone receptors
.
Cancer
1989
;
63
:
280
8
.
23.
Ince
TA
,
Sousa
AD
,
Jones
MA
,
Harrell
JC
,
Agoston
ES
,
Krohn
M
, et al
Characterization of twenty-five ovarian tumour cell lines that phenocopy primary tumours
.
Nat Commun
2015
;
6
:
7419
32
.
24.
Cariou
S
,
Donovan
JC
,
Flanagan
WM
,
Milic
A
,
Bhattacharya
N
,
Slingerland
JM
. 
Down-regulation of p21WAF1/CIP1 or p27Kip1 abrogates antiestrogen-mediated cell cycle arrest in human breast cancer cells
.
Proc Natl Acad Sci U S A
2000
;
97
:
9042
6
.
25.
Chen
Y
,
Alvarez
EA
,
Azzam
D
,
Wander
SA
,
Guggisberg
N
,
Jorda
M
, et al
Combined Src and ER blockade impairs human breast cancer proliferation in vitro and in vivo
.
Breast Cancer Res Treat
2010
;
128
:
69
78
.
26.
Catzavelos
C
,
Bhattacharya
N
,
Ung
YC
,
Wilson
JA
,
Roncari
L
,
Sandhu
C
, et al
Decreased levels of the cell-cycle inhibitor p27Kip1 protein: prognostic implications in primary breast cancer
.
Nat Med
1997
;
3
:
227
30
.
27.
Hennessy
BT
,
Lu
Y
,
Gonzalez-Angulo
AM
,
Carey
MS
,
Myhre
S
,
Ju
Z
, et al
A technical assessment of the utility of reverse phase protein arrays for the study of the functional proteome in non-microdissected human breast cancers.
Clin Proteomics
2010
;
6
:
129
51
.
28.
Yoshihara
K
,
Tsunoda
T
,
Shigemizu
D
,
Fujiwara
H
,
Hatae
M
,
Fujiwara
H
, et al
High-risk ovarian cancer based on 126-gene expression signature is uniquely characterized by downregulation of antigen presentation pathway
.
Clin Cancer Res
2012
;
18
:
1374
85
.
29.
Bonome
T
,
Levine
DA
,
Shih
J
,
Randonovich
M
,
Pise-Masison
CA
,
Bogomolniy
F
, et al
A gene signature predicting for survival in suboptimally debulked patients with ovarian cancer
.
Cancer Res
2008
;
68
:
5478
86
.
30.
Tothill
RW
,
Tinker
AV
,
George
J
,
Brown
R
,
Fox
SB
,
Lade
S
, et al
Novel molecular subtypes of serous and endometrioid ovarian cancer linked to clinical outcome
.
Clin Cancer Res
2008
;
14
:
5198
208
.
31.
Lu
KH
,
Patterson
AP
,
Wang
L
,
Marquez
RT
,
Atkinson
EN
,
Baggerly
KA
, et al
Selection of potential markers for epithelial ovarian cancer with gene expression arrays and recursive descent partition analysis
.
Clin Cancer Res
2004
;
10
:
3291
300
.
32.
Bild
AH
,
Yao
G
,
Chang
JT
,
Wang
Q
,
Potti
A
,
Chasse
D
, et al
Oncogenic pathway signatures in human cancers as a guide to targeted therapies
.
Nature
2006
;
439
:
353
7
.
33.
Chou
TC
. 
Drug combination studies and their synergy quantification using the Chou-Talalay method
.
Cancer Res
2010
;
70
:
440
6
.
34.
Dings
RP
,
Yokoyama
Y
,
Ramakrishnan
S
,
Griffioen
AW
,
Mayo
KH
. 
The designed angiostatic peptide anginex synergistically improves chemotherapy and antiangiogenesis therapy with angiostatin
.
Cancer Res
2003
;
63
:
382
5
.
35.
Bayliss
J
,
Hilger
A
,
Vishnu
P
,
Diehl
K
,
El-Ashry
D
. 
Reversal of the estrogen receptor negative phenotype in breast cancer and restoration of antiestrogen response
.
Clin Cancer Res
2007
;
13
:
7029
36
.
36.
Miller
PC
,
Clarke
J
,
Koru-Sengul
T
,
Brinkman
J
,
El-Ashry
D
. 
A novel MAPK-microRNA signature is predictive of hormone-therapy resistance and poor outcome in ER-positive breast cancer
.
Clin Cancer Res
2015
;
21
:
373
85
.
37.
Zhao
Y
,
Adjei
AA
. 
The clinical development of MEK inhibitors
.
Nat Rev Clin Oncol
2014
;
11
:
385
400
.
38.
Pearce
CL
,
Chung
K
,
Pike
MC
,
Wu
AH
. 
Increased ovarian cancer risk associated with menopausal estrogen therapy is reduced by adding a progestin
.
Cancer
2009
;
115
:
531
9
.
39.
Rao
BR
,
Slotman
BJ
. 
Endocrine factors in common epithelial ovarian cancer
.
Endocr Rev
1991
;
12
:
14
26
.
40.
Schwartz
PE
,
Chambers
JT
,
Kohorn
EI
,
Chambers
SK
,
Weitzman
H
,
Voynick
IM
, et al
Tamoxifen in combination with cytotoxic chemotherapy in advanced epithelial ovarian cancer. A prospective randomized trial
.
Cancer
1989
;
63
:
1074
8
.
41.
Burotto
M
,
Chiou
VL
,
Lee
JM
,
Kohn
EC
. 
The MAPK pathway across different malignancies: a new perspective
.
Cancer
2014
;
120
:
3446
56
.
42.
Miller
CR
,
Oliver
KE
,
Farley
JH
. 
MEK1/2 inhibitors in the treatment of gynecologic malignancies
.
Gynecol Oncol
2014
;
133
:
128
37
.
43.
Singer
G
,
Oldt
R
 III
,
Cohen
Y
,
Wang
BG
,
Sidransky
D
,
Kurman
RJ
, et al
Mutations in BRAF and KRAS characterize the development of low-grade ovarian serous carcinoma
.
J Natl Cancer Inst
2003
;
95
:
484
6
.
44.
Estep
AL
,
Palmer
C
,
McCormick
F
,
Rauen
KA
. 
Mutation analysis of BRAF, MEK1 and MEK2 in 15 ovarian cancer cell lines: implications for therapy
.
PLoS One
2007
;
2
:
e1279
.
45.
Donovan
JC
,
Milic
A
,
Slingerland
JM
. 
Constitutive MEK/MAPK activation leads to p27(Kip1) deregulation and antiestrogen resistance in human breast cancer cells
.
J Biol Chem
2001
;
276
:
40888
95
.
46.
Chu
I
,
Blackwell
K
,
Chen
S
,
Slingerland
J
. 
The dual ErbB1/ErbB2 inhibitor, lapatinib (GW572016), cooperates with tamoxifen to inhibit both cell proliferation- and estrogen-dependent gene expression in antiestrogen-resistant breast cancer
.
Cancer Res
2005
;
65
:
18
25
.
47.
Schiff
R
,
Massarweh
SA
,
Shou
J
,
Bharwani
L
,
Mohsin
SK
,
Osborne
CK
. 
Cross-talk between estrogen receptor and growth factor pathways as a molecular target for overcoming endocrine resistance
.
Clin Cancer Res
2004
;
10
(
1 Pt 2
):
331S
6S
.
48.
Osborne
CK
,
Shou
J
,
Massarweh
S
,
Schiff
R
. 
Crosstalk between estrogen receptor and growth factor receptor pathways as a cause for endocrine therapy resistance in breast cancer
.
Clin Cancer Res
2005
;
11
(
2 Pt 2
):
865s
70s
.
49.
Mullen
P
,
Cameron
DA
,
Hasmann
M
,
Smyth
JF
,
Langdon
SP
. 
Sensitivity to pertuzumab (2C4) in ovarian cancer models: cross-talk with estrogen receptor signaling
.
Mol Cancer Ther
2007
;
6
:
93
100
.
50.
Hou
JY
,
Rodriguez-Gabin
A
,
Samaweera
L
,
Hazan
R
,
Goldberg
GL
,
Horwitz
SB
, et al
Exploiting MEK inhibitor-mediated activation of ERalpha for therapeutic intervention in ER-positive ovarian carcinoma
.
PLoS One
2013
;
8
:
e54103
.
51.
Domcke
S
,
Sinha
R
,
Levine
DA
,
Sander
C
,
Schultz
N
. 
Evaluating cell lines as tumour models by comparison of genomic profiles
.
Nat Commun
2013
;
4
:
2126
.
52.
Akbani
R
,
Ng
PK
,
Werner
HM
,
Shahmoradgoli
M
,
Zhang
F
,
Ju
Z
, et al
A pan-cancer proteomic perspective on The Cancer Genome Atlas
.
Nat Commun
2014
;
5
:
3887
.
53.
Akbani
R
,
Becker
KF
,
Carragher
N
,
Goldstein
T
,
de
KL
,
Korf
U
, et al
Realizing the promise of reverse phase protein arrays for clinical, translational, and basic research: a workshop report: the RPPA (Reverse Phase Protein Array) society
.
Mol Cell Proteomics
2014
;
13
:
1625
43
.
54.
Creighton
CJ
,
Hilger
AM
,
Murthy
S
,
Rae
JM
,
Chinnaiyan
AM
,
El Ashry
D
. 
Activation of mitogen-activated protein kinase in estrogen receptor alpha-positive breast cancer cells in vitro induces an in vivo molecular phenotype of estrogen receptor alpha-negative human breast tumors
.
Cancer Res
2006
;
66
:
3903
11
.
55.
Tanaka
K
,
Arao
T
,
Maegawa
M
,
Matsumoto
K
,
Kaneda
H
,
Kudo
K
, et al
SRPX2 is overexpressed in gastric cancer and promotes cellular migration and adhesion
.
Int J Cancer
2009
;
124
:
1072
80
.
56.
Rankin
EB
,
Fuh
KC
,
Taylor
TE
,
Krieg
AJ
,
Musser
M
,
Yuan
J
, et al
AXL is an essential factor and therapeutic target for metastatic ovarian cancer
.
Cancer Res
2010
;
70
:
7570
9
.
57.
Chen
R
,
Khatri
P
,
Mazur
PK
,
Polin
M
,
Zheng
Y
,
Vaka
D
, et al
A meta-analysis of lung cancer gene expression identifies PTK7 as a survival gene in lung adenocarcinoma
.
Cancer Res
2014
;
74
:
2892
902
.
58.
Perez
PC
,
Bruzzone
A
,
Sarappa
MG
,
Castillo
LF
,
Luthy
IA
. 
Involvement of alpha2- and beta2-adrenoceptors on breast cancer cell proliferation and tumour growth regulation
.
Br J Pharmacol
2012
;
166
:
721
36
.

Supplementary data