Lack of Estrogen Receptor Α Is Associated with Epithelial-Mesenchymal

Author Manuscript Published OnlineFirst on January 14, 2013; DOI: 10.1158/1078-0432.CCR-12-3039 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Low ERα associates with EMT and PI3Kinase alterations in endometrial carcinoma

Lack of Estrogen receptor α is associated with epithelial-mesenchymal

transition and PI3Kinase alterations in endometrial carcinoma.

Elisabeth Wik1,2, Maria B. Ræder 1,2, Camilla Krakstad1,2, Jone Trovik1,2, Even Birkeland1,2, Erling A. Hoivik1,2, Siv Mjos2, Henrica M.J. Werner1,2, Monica Mannelqvist3, Ingunn M. Stefansson3,4, Anne M. Oyan5,6, Karl H. Kalland5,6, Lars A. Akslen3,4, Helga B. Salvesen1,2

1Department of Obstetrics and Gynecology, Haukeland University Hospital, Bergen, Norway,

2Department of Clinical Medicine, University of Bergen, Norway

3The Gade Institute, Section for Pathology, University of Bergen, Norway

4Department of Pathology, Haukeland University Hospital, Bergen, Norway,

5 The Gade Institute, Section for Microbiology, University of Bergen, Norway

6Department of Microbiology and Immunology, Haukeland University Hospital

Running title: Low ERα associates with EMT and PI3Kinase alterations in endometrial

carcinoma

Keywords: Endometrial carcinoma, estrogen receptor, expression analyses, epithelial-

mesenchymal transition, PI3Kinase

Address for correspondence:

Helga B. Salvesen MD, PhD or Elisabeth Wik, MD

Department of Clinical Medicine, Section for Gynecology and Obstetrics

University of Bergen

Haukeland University Hospital, 5021 Bergen, Norway

Telephone: +47 55 97 42 00 Fax: +47 55 97 49 68

E-mail addresses: [email protected] or [email protected]

Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2013 American Association for Cancer Research. Author Manuscript Published OnlineFirst on January 14, 2013; DOI: 10.1158/1078-0432.CCR-12-3039 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Low ERα associates with EMT and PI3Kinase alterations in endometrial carcinoma

Potential conflicts of interest: None of the authors indicated potential conflicts of interest.

Manuscript content: 4417 words, 3 figures, 3 tables, 2 supplementary figures, 5

supplementary tables.

Translational Relevance

Estrogen receptor α is shown to be a strong prognostic marker and potentially predictive for

response to hormonal treatment in endometrial carcinoma, although, contrasting breast

cancer, not yet systematically implemented for tailoring therapy to endometrial cancer

patients. In one investigation series and three independent validation sets (in total 628

patients), we identify and validate an association between ERα negative endometrial

carcinomas and markers for epithelial-mesenchymal transition. Drug signature analyses

suggest testing of PI3K/mTOR inhibitors in ERα negative tumors in particular. Subsequently

we also find ERα negative tumors to be associated with several potential measures for

PI3Kinase pathways activation. This study confirms ERα as a strong and robust prognostic

marker in endometrial carcinoma and provides a rationale for clinical trials exploring the

potential of ERα as predictive marker for response to inhibitors of the PI3Kinase pathway and

epithelial-mesenchymal transition.

ABSTRACT

Lack of Estrogen receptor α is associated with epithelial-mesenchymal

transition and PI3Kinase alterations in endometrial carcinoma.

Purpose: We hypothesized that ERα status in endometrial carcinomas, associated with poor

prognosis, is reflected in transcriptional signatures suggesting targets for new therapy.

Experimental design: Endometrial carcinoma samples in a primary investigation cohort

(n=76) and three independent validation cohorts (n=155/286/111) were analyzed through

integrated molecular profiling. Biomarkers were assessed by IHC, DNA oligonucleotide

microarray, qPCR, SNP array and Sanger sequencing in the cohorts, annotated for

comprehensive histopathologic and clinical data, including follow-up.

Results: ERα immunohistochemical staining was strongly associated with mRNA expression

of the receptor gene (ESR1) and patient survival (both P< 0.001). ERα negativity associated

with activation of genes involved in Wnt-, Sonic Hedgehog- and TGF-β signaling in the

investigation cohort, indicating epithelial-mesenchymal transition (EMT). The association

between low ERα and EMT was validated in three independent data sets. Furthermore,

PI3Kinase- and mTOR inhibitors were amongst the top ranked drug signatures negatively

correlated with the ERα negative tumors. Low ERα was significantly associated with PIK3CA

amplifications but not mutations. Also, low ERα was correlated to high expression of

Stathmin, a marker associated with PTEN loss, and a high PI3K activation signature.

Conclusion: Lack of ERα in endometrial cancer is associated with epithelial-mesenchymal

transition and reduced survival. We present a rationale for investigating ERα’s potential to

predict response to PI3K/mTOR inhibitors in clinical trials and also suggest EMT inhibitors

to ERα negative endometrial carcinomas.

INTRODUCTION

Endometrial cancer is the most common gynecologic malignancy in developed countries and

accounts yearly for 50 000 deaths worldwide (1). Although three quarters of these cancers are

treated at an early stage, 15-20% recur (2). The most common basis for determining risk of

recurrent disease is the categorization into type I and II endometrial carcinoma (3). Type I

tumors are most frequent, characterized by endometrioid histology, low histologic grade, low

FIGO stage and favorable prognosis. In contrast, type II tumors are characterized by non-

endometrioid histology, high histologic grade and stage, and a tendency to recur, even when

treated at early stage (4). However, the prognostic value of this distinction is limited, as up to

20% of type I cancers recur, while half of type II tumors do not (4).

The molecular basis for type I and II cancers is only partially understood. Hyperestrogenic

risk factors and positivity for estrogen receptor α (ERα) are common in type I in contrast to

type II cancers, and ERα status is reported to be a prognostic marker in endometrial cancer (5,

6). Contrasting breast cancer treatment, hormone receptor status in endometrial cancer has

not been clinically implemented in a systematic manner to tailor hormonal therapy (7, 8) or

other targeted treatments, e.g. PI3K, mTOR or VEGF-A inhibitors (4).

Epithelial-mesenchymal transition (EMT) enables epithelial cells to become more like

mesenchymal cells with increased motility and the ability to invade locally, intravasate,

circulate, extravasate and resist apoptosis (9-11). Markers of EMT are known to be associated

with more aggressive tumors in different organs (12, 13). In endometrial carcinomas,

alterations of EMT-related markers like E-cadherin, P-cadherin, α- and β-catenin, have been

associated with metastatic disease and reduced survival (14-16). The EMT process is

suggested to be a potential target for new cancer therapies (17, 18). Importantly, data linking

EMT to drug resistance have been reported, indicating that EMT targeting may be relevant for

cancer treatment (18).

In the present study, we hypothesized that ERα negative endometrial carcinomas are

associated with an aggressive clinico-pathologic phenotype and are likely to be distinguished

from ERα positive tumors by underlying genetic alterations and differences in gene

expression, related to key processes in cancer development and progression. Based on data

from four independent patient series, we report for the first time that ERα negative

endometrial carcinomas are associated with increased epithelial-mesenchymal transition.

Also, inhibitors of the PI3Kinase/mTOR pathways are suggested as relevant for the treatment

of ERα negative endometrial cancers in particular.

PATIENTS AND METHODS

Patients and follow-up

Four independent endometrial carcinoma patient cohorts (including both endometrioid and

non-endometrioid histologic subtypes) were studied to explore the associations between

hormone receptor status, prognosis and alterations in transcriptional signatures identifying

molecular phenotypes and potential targets for new therapy (Supplementary Figure 1): 1) A

primary investigation cohort of 76 snap frozen, prospectively collected endometrial

carcinomas with high tumor purity was studied by DNA oligonucleotide microarray for

mRNA expression levels, and Single Nucleotide Polymorphism (SNP) array and Sanger

sequencing for detection of PIK3CA amplification and PIK3CA mutations, respectively. 2) A

prospective validation cohort of 155 snap frozen endometrial cancers was subjected to qPCR

validation of candidate markers picturing findings in the primary investigation set. PIK3CA

Sanger sequencing was also applied for tumors in this series. Both prospective series were

collected sequentially, taking into account tumor purity (see below). 3) A retrospective

population based validation cohort of 286 patients with formalin fixed paraffin embedded

(FFPE) tumor tissue was used for clinical validation. Protein expression of candidate markers

and vascular invasion was assessed by immunohistochemistry (IHC), as previously reported

(5, 15, 19, 20). For all these cohorts, biopsies were sampled from primary tumors in

hysterectomy specimens. 4) Finally, publically available external gene expression array data

from 111 patients were investigated for external validation (Supplementary table 1).

Patients in series 1 through 3 (Supplementary Table 2) were treated at the Department of

Obstetrics and Gynecology, Section of Gynecological Cancer, Haukeland University

Hospital, Bergen, Norway, a referral hospital for patients in Hordaland County, representing

approximately 10% of the Norwegian population and with a similar incidence rate and

prognosis as the total Norwegian population of endometrial cancers (15).

Comparison of clinico-pathologic variables for the different cohorts is presented in

Supplementary Table 3. The primary investigation set largely reflects the general population

of endometrial cancers from the region, while the prospective validation series and the

external data set are more enriched for aggressive cancer subtypes.

The primary investigation set was subjected to formal revision. For the prospective validation

series, we applied the histopathologic diagnoses derived from the routine reports. The

retrospective series (1981-90) was also subjected to formal histopathologic revision for

adequate sub-typing and grading in previous studies (15, 20).

Oligonucelotide DNA microarray analyses

RNA was extracted from snap frozen, surgically removed tumor biopsies from clinically well-

defined endometrial carcinomas with > 80% neoplastic tissue for the vast majority (only 4

cases, 3 endometrioid and one serous carcinoma, had tumor purity of 50%). The RNA was

obtained according to standard protocols using the RNeasy Mini Kit protocol (Qiagen,

Hilden, Germany) and hybridized to Agilent Whole Human Genome Microarrays 44k

(Cat.no. G4112F), according to manufacturer's instructions (www.agilent.com). Arrays were

scanned using the Agilent Microarray Scanner Bundle. Microarray signal intensities were

determined using BRB-ArrayTools (http://linus.nci.nih.gov/BRB-ArrayTools/). Mean spot

signal data were used as intensity measure. The expression data were normalized, using

median over entire array. The DNA oligonucleotide arrays were performed around the same

time, in the same lab, by the same personnel and in one batch.

Genes differentially expressed between ERα positive and negative tumors were considered

significantly differentially expressed if FDR <0.05; the test was performed by Class

comparison analysis in BRB Array Tools (21). Global hierarchical clustering was performed

using weighted average linkage and Pearson correlation as similarity measures. Gene Set

Enrichment Analyses (GSEA - www.broadinstitute.org/gsea) and BRB Array Tools were

used to explore pathways and transcription factor gene sets differentially expressed between

ERα positive and negative tumors (22, 23). Signature scores for gene sets in GSEA were

created by using a pathways activation score as implemented in the GSEA application for R

(www.broadinstitute.org/gsea).

Connectivity Map

The correlation between global gene expression pattern and potential new therapeutics for

ERα negative tumors was assessed in the primary investigation cohort and validated in the

external gene expression data set, using the drug signatures database Connectivity Map (24).

The drug signatures in Connectivity Map are based on genes differentially expressed between

cell lines treated with various drugs/compounds compared to the corresponding untreated cell

lines. As such, the database drug signatures represent gene expression alterations presumed to

reflect drug effects. Genes differentially expressed (FDR <0.05) between subsets with low

and high ERα protein and ESR1 mRNA expression levels were included in the signatures that

were the basis for the analyses in Connectivity Map.

Quantitative PCR (qPCR)

cDNA was synthesized from 1µg RNA from 155 tumor samples in the prospective validation

series, using the High capacity RNA to cDNA kit (Applied Biosystems Inc, Foster City, CA,

USA). mRNA expression of ESR1 and a selection of cell adhesion-/EMT markers were

determined using the TaqMan gene expression assays ESR1-Hs00174860_m1 (ERα), CDH1-

Hs01013958_m1 (E-cadherin), CDH2-Hs00983056_m1 (N-cadherin), CDH3-

Hs00999915_m1 (P-cadherin), CTNNB1Hs00170025_m1 (beta-catenin), CTNNA2-

Hs00189285_m1 (α-catenin) and CTNND1-Hs00931670_m1 (p120 catenin). For mRNA

expression of SERPINE-1, the assay SERPINE1-Hs00167155_m1 (Serine peptidase inhibitor,

clade E 1, member 1) was used. All samples were run on micro fluidic card with GAPDH-

Hs99999905_m1 as endogenous control and a sample of atypical endometrial hyperplasia as

calibrator. Microfluidic cards were run as previously reported (19) and described by the

manufacturer. All analyses were performed in RQ manager (Applied Biosystems software),

using the ∆∆Ct based method for calculation of RQ values (25).

EMT signature scores

In the prospective validation cohort, the mRNA expression values (by qPCR) for a selection

of cell adhesion markers known from the literature to be related to EMT (11, 26, 27) were

assessed to calculate an mRNA signature score reflecting the EMT process. The expression

values were mean normalized and scaled to the same standard deviation (19). The sum of the

expression values of the genes down-regulated (CDH1, CTNNB1, CTNNA2, CTNND1) were

subtracted from the sum of the expression values of genes up-regulated (CDH2, CDH3).

Immunohistochemical staining

FFPE tumor specimens from the three patient cohorts from Haukeland University Hospital

were prepared in regular slides or tissue microarrays (TMA) and stained and scored for ERα,

Stathmin and cell adhesion-/EMT markers (E-cadherin, P-cadherin, β-catenin and p120

catenin), as well as evaluated for vascular invasion and myometrial infiltration, as previously

described (5, 15, 19, 20, 28). ERα IHC data were available for all cases in the primary

investigation set, for 153 cases in the prospective validation series and for 266 cases in the

retrospective validation cohort.

Validation in external gene expression data set

Publicly available gene expression data corresponding to tumors resected from endometrial

carcinoma tissue were obtained from the Expression Project for Oncology (expO:

http://expo.intgen.org/geo/home.do). GEO accession numbers and information on clinico-

pathologic data for these 111 tumors are listed in Supplementary Table 1. The GEO data sets

were run at Affymetrix U133+2 arrays. For these data sets, individual probes were sequence-

matched against Aceview (NCB135) to construct transcript level probe sets (19). The two

different data sets were not merged since the GEO data set was used for validation of findings

in the primary investigation series. Calibration to get comparable expression levels was

therefore not necessary.

DNA analyses

Genomic DNA was extracted from surgically dissected, fresh frozen primary tumors. SNP

array and Sanger sequencing of PIK3CA (exon 9 and exon 20) was performed for 230 and

215 cases, respectively, as previously described (19, 29), overlapping with the primary

investigation set and the prospective validation series.

Statistics

Data were analyzed using PASW 18 (Predictive Analytics SoftWare, version 18.0, IBM, New

York, USA). Probability of <0.05 was considered statistically significant except for the

oligonucleotide microarray analyses (see above). Categories were compared using Pearson’s

chi-square or Fisher’s exact test when appropriate. Spearman’s correlation coefficient was

reported for bivariate correlation between continuous variables. Mann-Whitney U test was

used for analysis of continuous variables between categories. Univariate survival analyses of

time to death due to endometrial carcinoma (disease specific survival) and time to recurrence

for patients without metastases at time of diagnosis (recurrence free survival) were performed

using the Kaplan–Meier method (log-rank test). Entry date was the time of primary surgery.

Patients who died from other causes were censored at the date of death in the analyses of

disease specific survival. Variables included in the Cox’ proportional hazards regression

analyses were examined by log-log plot before incorporated in the regression models. When

categorizing continuous variables (e.g. mRNA expression values of genes, gene signature

scores, IHC score indices) cut-off points were based on median or quartile values, considering

also the frequency distribution for each marker. Groups with similar survival were merged.

For gene expression data (gene signatures and single genes), upper quartile defined “high

expression”. ERα negative cases in the primary investigation and retrospective validation

series included cases with staining index (SI) 0. In the prospective validation series, low ERα

included SI ≤2 cases.

Estimation of sample size was done by chi-square test using software East4, 2005 Cytel

Software Corp. To reach 90% power detecting a 30% difference in 5-year survival (90% for

patients with markers within normal range versus 60% with pathological markers) at a 5%

level of significance, 65 patients were needed assuming a positive to negative ratio of the

markers of 1:3.

RESULTS

Lack of ERα is associated with aggressive endometrial cancer.

ERα protein and ESR1 mRNA expression estimated by microarray and qPCR were

significantly associated in the primary investigation set (Figure 1A and B). The strong

correlation between mRNA and protein expression was confirmed in the prospective

validation cohort (Table 1). Low ERα protein and ESR1 mRNA expression and

transcriptional clusters containing the majority of ERα negative cases were significantly

correlated with an aggressive clinico-pathologic phenotype and poor survival in all patient

cohorts studied (Table 1, Figure 1C and D, Supplementary Figure 2). Low ESR1 mRNA was

associated with reduced survival in the primary investigation set (multivariate analyses,

adjusted for age, FIGO stage, histologic type and histologic grade, HR=1.4, 95% CI 1.05-1.8,

P=0.02). In the prospective validation set a significant interaction between ESR1 mRNA and

FIGO stage (p=0.03) was detected. The association between ESR1 mRNA levels and survival

was therefore further assessed separately for FIGO I/II and FIGO III/IV tumors (adjusted for

age, histologic grade and histologic type). A borderline statistically significant association

between low ESR1 mRNA and reduced survival was seen for FIGO I/II cases (HR 2.9, 95%

CI 0.9-9.2, P=0.07), but not in the stratified analysis of the small subset (n=33) of more

aggressive FIGO III/IV cases (p=0.7). Further, an independent association with survival for

ERα protein expression was confirmed in multivariate survival analyses for the subgroups of

endometrioid tumors, adjusting for age, FIGO stage and histologic grade (Supplementary

Table 4), aiding in the identification of high-risk patients within this clinically challenging

patient group expected to have a good prognosis. Also for the whole cohort, ERα status

maintained prognostic impact adjusted for the same variables and histologic subtype

(footnotes, Supplementary Table 4).

Taken together, this clearly supports that ERα protein and ESR1 mRNA expressions are

strongly correlated, robust and reproducible measures for identification of aggressive

endometrial carcinomas.

Integrated analyses associates lack of ERα with epithelial-mesenchymal transition

To explore potential biological processes contributing to the aggressive phenotype of ERα

negative endometrial carcinomas, we examined transcriptional differences between ERα

positive and negative tumors. In the primary investigation cohort, supervised analyses of the

oligonucleotide DNA microarray expression data revealed 303 unique genes differentially

expressed between ERα positive and negative tumors; expression of 60 unique genes were

significantly higher in ERα negative as compared to ERα positive tumors. Six of these genes

were previously demonstrated to be linked to signaling pathways and processes related to

cancer invasion (Supplementary table 5A), including SERPINE-1, suggested to be a marker

for TGF-β signaling (30).

Clustering of the mRNA expression (microarray data) of ESR1 and a selection of EMT

related transcription factors (SNAI1, SNAI2, TWIST1, ZEB1 and ZEB2) indicated an

association between low ESR1 expression and high expression of the EMT transcription

factors (Supplementary Figure 2C). Further, in gene set enrichment analyses (GSEA), gene

sets indicating activation of Sonic Hedgehog (SHH) (31) , Wnt (32) and TGF-β signaling (33)

were found amongst the most significantly enriched gene sets in ERα negative compared to

positive tumors (FDR<0.25). Down-regulation of MTA3, potentially leading to increased

expression of the transcription factor Snail (34) (www.biocarta.com), was also found in ERα

negative tumors. From the literature, these pathways are known to be involved in EMT

activation (10, 34). To assess the gene expression signatures’ correlation to survival and to

visualize the associations between gene signatures and ERα expression, signature scores were

generated for each gene set, based on expression of the genes defining the pathways (for

details see http://www.broadinstitute.org/gsea/msigdb). High SHH-, TGF-β- and Wnt-

signature scores were significantly correlated with ERα negative tumors (P<0.001, P=0.001,

P=0.007, respectively) and survival (P<0.001, P=0.002, P=0.009 respectively) (Figure 2A).

Also for the endometrioid subgroup of tumors, High SHH-, TGF-β and Wnt- signature scores

were significantly correlated with ERα negative cancer (P<0.001, P=0.007, P=0.01,

respectively) and survival (P=0.004, P=0.02 and P=0.04, respectively). In transcription

factors (TFs) gene sets analyses (primary investigation cohort) based on ERα protein status,

several TFs related to EMT activation and TGF-β signaling, were identified (Supplementary

Table 5B), further supporting the association between ERα and EMT in endometrial

carcinoma. In sum, these data indicate an association between ERα negative endometrial

cancer and EMT, and the association with survival further support biological relevance of the

signatures assessed. In contrast, in the ERα positive tumors, two pathways related to estrogen

and estrogen receptor activity were amongst the top ranked gene sets enriched in the ERα

positive tumors (GSEA, FDR<0.025): A Biocarta pathway related to estrogen receptor

activity (“Pelp1 Modulation of Estrogen Receptor Activity”), and a gene set generated from

identified estrogen-regulated genes from breast cancer tumors and cell lines (35).

For validation of the associations in tumors of low ERα expression, at first we examined a

publicly available external DNA microarray data set for associations between low ESR1

mRNA expression and the gene sets identified from the primary investigation set. ESR1

expression was inversely correlated with the Sonic Hedgehog, TGF-β and Wnt signature

scores (P<0.001, rs= -0.46, P<0.001, rs= -0.37 and P=0.002, rs= -0.30, respectively) and

correlated with the MTA3 signature score (P<0.001, rs= 0.60), also supporting the association

between ERα negative endometrial cancer and EMT identified in our primary investigation

set.

This association was further validated in the prospectively collected cohort of 155 snap frozen

primary tumor specimens. Expression levels for cell adhesion-/EMT markers were estimated

by qPCR. Tumor mRNA ESR1 expression was significantly correlated with E-cadherin,

catenin p120, catenin α and β-catenin mRNA levels (all P<0.001). ERα protein expression

was correlated to the same markers (all P≤0.02), and also negatively correlated to N-cadherin

mRNA expression (P=0.035). Further, ERα protein expression was significantly associated

with high mRNA expression of the suggested TGF-β signaling marker SERPINE-1 (=PAI-1),

both in the primary investigation set (P<0.002, Figure 2E) and prospective validation series

(P<0.004). SERPINE-1 mRNA expression was also associated with high TGF- β signature

score (Figure 2F).

Finally, in the additional independent, retrospective population based cohort of 286

endometrial carcinomas, ERα negative tumor status confirmed to be significantly correlated to

pathologic protein expression of the EMT markers E-cadherin, β-catenin and catenin p120,

the EMT related P-cadherin (36) as well as invasive properties reflected by vascular invasion

and deep myometrial infiltration (Table 2).

A signature for EMT is associated with prognosis in endometrial cancer

In the prospective validation cohort, we generated an EMT gene expression signature by

combining a panel of cell adhesion markers with known involvement in EMT (see Patients

and Methods), and we explored the signature’s relation to prognosis in endometrial cancer.

High EMT signature score (upper quartile of the score) was strongly associated with poor

survival (Fig 2G) and recurrence free survival (P=0.002), also for the subsets of endometrioid

cases only (P=0.006).

To further assess the prognostic value of the EMT signature score compared to the mRNA

levels of single EMT markers, we included separately the EMT signature score and markers,

one at a time, in Cox’ multivariate analyses together with age, FIGO stage, histologic subtype

and histologic grade. Here, the EMT score had independent prognostic impact with HR=1.4

(95% CI: 1.02-1.7, P=0.005), contrasting the individual EMT markers, of which none had

significant independent prognostic impact.

The EMT signature score was negatively correlated to ERα (P<0.001) and ESR1 mRNA

expression (Figure 2H). In Cox’ survival analyses, including EMT score and ERα expression,

both markers had independent prognostic impact on disease specific survival with HR 2.6

(95%CI: 1.03-6.6, P=0.04) and 5.0 (95%CI: 1.9-12.8, P=0.001), respectively. In sum, this

validates that EMT, measured by an EMT gene expression signature score, is associated with

ERα negative tumors and adds prognostic information in addition to ERα status in

endometrial carcinoma.

Integrated analyses identify PI3Kinase- and mTOR inhibitors as potential drugs for

patients with ERα negative tumors.

Connectivity Map version 02 (24) was queried for drug signatures negatively correlated to the

gene expression signature of ERα negative tumors. The gene lists acquired from the class

comparison analyses based on ERα protein status were analyzed. Amongst 1309 small

molecules represented in Connectivity Map, one mTOR- and two PI3Kinase inhibitors as well

as a retinoic acid receptor agonist were the 4 top ranked compounds with signatures most

significantly negatively correlated with ERα negative tumors, as listed in Table 3. In the

external endometrial cancer data set, we confirmed PI3Kinase- and mTOR inhibitors to be the

top ranked compounds significantly negatively correlated to the signature defined by low

ESR1 mRNA expression (Table 3). These data support that targeting the PI3Kinase-AKT-

mTOR pathway, may be particularly relevant in ERα negative endometrial cancers.

We further explored the association between low ERα expression and potential measures for

PI3Kinase activation: PIK3CA mutations and amplifications, PIK3CA, PTEN and AKT1

mRNA levels, Stathmin protein expression, a suggested surrogate marker for PTEN loss and

PI3Kinase activation (19, 37), and a publicly available PI3Kinase activation score (38). Low

ERα expression was significantly associated with amplifications of the 3q26 region, harboring

PIK3CA (P=0.02, Figure 3A), and also with high mRNA expression of the PI3Kinase

signaling family members PIK3CA and AKT1 as well as low PTEN expression (P=0.05,

P=0.01 and P=0.005, respectively). ERα-low tumors were not significantly associated with

PIK3CA mutation status (P=0.8, Figure 3B), neither with any statistical trend for the

association when investigating the correlation with exon 9 and exon 20 separately (P=0.6 and

P=0.9, respectively). However, high PI3Kinase activation score was negatively correlated to

ESR1 mRNA expression (P=0.001, rs= -0.4).

In line with this, a high PI3Kinase score was found to be correlated with high SHH- and Wnt

signature scores (P<0.001, rs= -0.5 and P<0.001, rs= -0.4, respectively) and negatively

correlated with MTA3 signaling (P=0.01, rs= -0.3), but not with the TGF-β signature score.

Also, low ERα protein expression was significantly associated with high Stathmin protein

expression (P=0.03). High Stathmin expression was also associated with markers for EMT as

well as deep myometrial growth, vascular invasion (Table 2) and high EMT signature score

(P=0.003). However, although our results are based only on statistical associations, the

findings from this study suggest a potential for targeting PI3K/mTOR in ERα negative

endometrial carcinomas and suggest presence of epithelial-mesenchymal transition in tumors

with PI3Kinase pathways alterations.

DISCUSSION

Current clinical decision making in the treatment of endometrial carcinomas mainly relies on

surgical FIGO stage, histologic subtype and histologic grade. This risk-stratification is

suboptimal and may lead to over- and under-treatment. The distinction between type I and

type II endometrial cancers is clinically important, although with considerable morphologic

and molecular overlap (4). Improved identification of patients at risk of recurrences and poor

outcome is needed, and molecular markers may be helpful in this setting. We wanted to apply

a population-based approach in the present study, aiming to associate estrogen receptor α

status with potential targets for new therapy, across all histologic subtypes and by applying

state of the art risk categories for treatment stratification including surgical FIGO 2009 stage,

histologic subtyping and histologic grading.

Here, we confirm a strong prognostic impact of ERα in endometrial cancer, for several

independent patient cohorts, as well as in subgroups of endometrioid cases. The prognostic

importance of ERα was confirmed in multivariate survival analyses including clinico-

pathologic factors. This may potentially aid in the identification of patients at risk of poor

outcome within the particularly challenging patient groups expected to have a good prognosis

based on standard markers applied in the clinic. We clearly confirm the role of ERα as a

strong prognostic marker in endometrial cancer, as indicated for around 30 years (5, 6, 39,

40). Still, implementation of this biomarker for tailoring systemic therapy has been slow (4,

8). Importantly, we found that ERα negative tumors are associated with epithelial-

mesenchymal transition and PI3Kinase pathway alterations, providing potential targets for

treatment of patients with this subgroup of endometrial carcinoma.

We have explored the transcriptional differences between ERα high and low cases. The

transcriptional differences between ERα high and low cases are explored by high-throughput

analyses, known to be associated with a risk of false discovery. To adjust for potential errors

related to multiple testing, validations of the findings in independent patient series and with

alternative methods are recommended. Our conclusions are strengthened by the use of several

independent data sets to validate both the association between ERα and EMT, and the

association between low ERα and potential measures for PI3K activation.

ERα has been suggested as a predictive marker for response to hormonal therapy in

endometrial carcinoma (7). Thigpen et al found significantly better response to

medroxyprogesterone in patients with ERα and PR positive compared to negative tumors

(41). Still, a Cochrane Review from 2010 on hormonal treatment in advanced endometrial

cancer concludes that in sum, there is insufficient evidence that hormonal treatment improves

the survival for patients with advanced or recurrent endometrial cancer. However, the lack of

stratification according to hormone receptor status represents a significant problem when

evaluating therapy response (8). Our findings suggest that treatment protocols systematically

implementing ERα levels may have a yet unexplored potential in endometrial carcinoma.

The current study suggests the presence of EMT as an important feature in ERα negative

endometrial carcinomas. ERα loss has previously been associated with increased expression

of Snail, an E-cadherin repressor, in these tumors (42), and a functional link between ERα

loss and EMT activation has not been found in endometrial cancer cell line studies, although

in cell lines from other cancer types (34, 43). Functional studies have also linked loss of ERα,

PI3K signaling and activation of STAT3 to EMT activation through TGF-β signaling (44).

This, together with our finding of increased expression of a TGF- β signaling marker,

SERPINE-1 (30), in ERα negative tumors suggest that TGF-β is involved in the EMT process

in ERα negative endometrial carcinomas. A recent study identifying TGF-β as a key factor for

invasion in endometrial cancer adds support to this (45). Further studies of functional

implications of both TGF-β signaling and ESR1 loss in relation to EMT and PI3K activation

in endometrial carcinoma would be highly relevant.

Individual markers of EMT have been reported to have prognostic impact in endometrial

carcinomas and other cancer types (15, 46), and estimation of Cadherin E to P- and E to N-

switches has indicated additional prognostic impact in various cancers (15, 47). EMT is

activated through different signaling pathways (9, 10, 13), and an EMT signature may better

reflect the EMT process than individual markers (11). The strong prognostic impact of the

EMT gene expression signature score in our data set is in line with this, supporting the

importance of EMT in aggressive endometrial carcinomas (16). Also, the EMT process is

hypothesized as a potential target for therapy in cancer (18). Therapy targeting TGF-β and

other EMT-related pathways are currently being evaluated in cancers (17, 18), and TGF-β and

N-cadherin inhibitors are presently in phase II studies (www.clinicaltrials.gov, August 2012).

Given the observed association between low ERα expression and EMT, the potential role of

EMT inhibiting therapies should also be explored for patients with ERα negative endometrial

carcinomas.

The possibilities for targeted therapy in endometrial carcinoma are limited (4) and in

particular for ERα negative tumors. PI3Kinase- and mTOR inhibitors are today in phase I/II

trials in advanced endometrial cancer (www.clinicaltrials.org, August 2012), based on

molecular changes reported in aggressive endometrial carcinomas (4, 19). Data from a clinical

phase II trial of mTOR inhibition are promising (48). Although the number of compounds

examined in Connectivity Map is limited (~1300), and other relevant drugs not included may

thus be undetected in the present study, our data add knowledge by suggesting that

PI3Kinase-/mTOR inhibitors may be particularly relevant for ERα negative endometrial

carcinomas, and an association with EMT is also indicated. PTEN loss of function and

PIK3CA mutations appear to be more frequent in type I cancers, while type II cancers are

characterized by ER loss, PIK3CA amplification and PI3K signaling activation (4, 19, 49).

Our finding that amplification of the PIK3CA region and not PIK3CA mutations are

associated with ERα loss appears to be in line with this. Recent studies on other cancer types

support that PI3Kinase signaling activation may be related also to PIK3CA amplifications

(50-52).Taken together, this suggests a potential for PI3Kinase inhibitors for ERα negative

tumors in particular. Still, subgroups within ERα positive tumors with PI3K alterations or

PTEN loss may also be relevant for such treatment.

Our results are partially based on indirect measures for drug effects, through assessment of

associations with effects demonstrated in functional studies. Further mechanistic studies of

PI3K/mTOR and EMT inhibition related to ERα levels will be relevant to elucidate the

potential of ERα as a predictive marker for response to these inhibitors. Assessing ERα in pre-

treatment samples from patients included in clinical trials with these inhibitors is also a

relevant ‘next step’ before conduction of randomized controlled trials with stratification

according to ERα status to further define predictive markers for the response to PI3K

inhibitors in endometrial cancer.

CONCLUSION

By integrated profiling of endometrial carcinomas we found that ERα negative tumors are

associated with the process of EMT as well as PI3K activation, and the two appears to be

related. Our study provides a rational for clinical trials exploring the effects of PI3K/mTOR

and EMT inhibitors in endometrial carcinoma, and in particular assessing the potential of ERα

to predict therapy response.

Acknowledgements: We thank Bendik Nordanger, Gerd Lillian Hallseth, Britt Edvardsen,

Mari Kyllesø Halle, Pål Christian Njølstad and Erlend Njølstad for technical assistance, and

the biostatisticians Roy M. Nilsen and Kjell Petersen for advice regarding statistical analyses

and microarray analyses, respectively.

Grant Support: Helse Vest, Research Council of Norway and The Norwegian Cancer Society,

Harald Andersens legat supported this study.

References

1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin. 2005 Mar- Apr;55(2):74-108. 2. Amant F, Moerman P, Neven P, Timmerman D, Van Limbergen E, Vergote I. Endometrial cancer. Lancet. 2005 Aug 6-12;366(9484):491-505. 3. Bokhman JV. Two pathogenetic types of endometrial carcinoma. Gynecol Oncol. 1983 Feb;15(1):10-7. 4. Salvesen HB, Haldorsen IS, Trovik J. Markers for individualised therapy in endometrial carcinoma. Lancet Oncol. 2012 Aug;13(8):e353-61. 5. Engelsen IB, Stefansson IM, Akslen LA, Salvesen HB. GATA3 expression in estrogen receptor alpha- negative endometrial carcinomas identifies aggressive tumors with high proliferation and poor patient survival. Am J Obstet Gynecol. 2008 Nov;199(5):543 e1-7. 6. Jongen V, Briet J, de Jong R, ten Hoor K, Boezen M, van der Zee A, et al. Expression of estrogen receptor-alpha and -beta and progesterone receptor-A and -B in a large cohort of patients with endometrioid endometrial cancer. Gynecol Oncol. 2009 Mar;112(3):537-42. 7. Decruze SB, Green JA. Hormone therapy in advanced and recurrent endometrial cancer: a systematic review. Int J Gynecol Cancer. 2007 Sep-Oct;17(5):964-78. 8. Kokka F, Brockbank E, Oram D, Gallagher C, Bryant A. Hormonal therapy in advanced or recurrent endometrial cancer. Cochrane Database Syst Rev. 2010(12):CD007926. 9. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. J Clin Invest. 2009 Jun;119(6):1420-8. 10. Thiery JP, Acloque H, Huang RY, Nieto MA. Epithelial-mesenchymal transitions in development and disease. Cell. 2009 Nov 25;139(5):871-90. 11. Zeisberg M, Neilson EG. Biomarkers for epithelial-mesenchymal transitions. J Clin Invest. 2009 Jun;119(6):1429-37. 12. Hung JJ, Yang MH, Hsu HS, Hsu WH, Liu JS, Wu KJ. Prognostic significance of hypoxia-inducible factor-1alpha, TWIST1 and Snail expression in resectable non-small cell lung cancer. Thorax. 2009 Dec;64(12):1082-9. 13. Micalizzi DS, Ford HL. Epithelial-mesenchymal transition in development and cancer. Future Oncol. 2009 Oct;5(8):1129-43. 14. Mell LK, Meyer JJ, Tretiakova M, Khramtsov A, Gong C, Yamada SD, et al. Prognostic significance of E-cadherin protein expression in pathological stage I-III endometrial cancer. Clin Cancer Res. 2004 Aug 15;10(16):5546-53. 15. Stefansson IM, Salvesen HB, Akslen LA. Prognostic impact of alterations in P-cadherin expression and related cell adhesion markers in endometrial cancer. J Clin Oncol. 2004 Apr 1;22(7):1242-52. 16. Mannelqvist M, Stefansson IM, Bredholt G, Hellem Bo T, Oyan AM, Jonassen I, et al. Gene expression patterns related to vascular invasion and aggressive features in endometrial cancer. Am J Pathol. 2011 Feb;178(2):861-71. 17. Iwatsuki M, Mimori K, Yokobori T, Ishi H, Beppu T, Nakamori S, et al. Epithelial-mesenchymal transition in cancer development and its clinical significance. Cancer Sci. 2010 Feb;101(2):293-9. 18. Krasnapolski MA, Todaro LB, de Kier Joffe EB. Is the epithelial-to-mesenchymal transition clinically relevant for the cancer patient? Curr Pharm Biotechnol. 2011 Nov;12(11):1891-9. 19. Salvesen HB, Carter SL, Mannelqvist M, Dutt A, Getz G, Stefansson IM, et al. Integrated genomic profiling of endometrial carcinoma associates aggressive tumors with indicators of PI3 kinase activation. Proc Natl Acad Sci U S A. 2009 Mar 24;106(12):4834-9. 20. Stefansson IM, Salvesen HB, Immervoll H, Akslen LA. Prognostic impact of histological grade and vascular invasion compared with tumour cell proliferation in endometrial carcinoma of endometrioid type. Histopathology. 2004 May;44(5):472-9. 21. Benjamini Y HY. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of The Royal Statistical Society. 1995;57(1):11. 22. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005 Oct 25;102(43):15545-50. 23. Efron B, Tibshirani R. On testing the significance of gene sets. Annals of applied statistics. 2007;1:22. 24. Lamb J, Crawford ED, Peck D, Modell JW, Blat IC, Wrobel MJ, et al. The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science. 2006 Sep 29;313(5795):1929-35. 25. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001 Dec;25(4):402-8. 26. Bellovin DI, Bates RC, Muzikansky A, Rimm DL, Mercurio AM. Altered localization of p120 catenin during epithelial to mesenchymal transition of colon carcinoma is prognostic for aggressive disease. Cancer Res. 2005 Dec 1;65(23):10938-45.

27. Benjamin JM, Nelson WJ. Bench to bedside and back again: molecular mechanisms of alpha-catenin function and roles in tumorigenesis. Semin Cancer Biol. 2008 Feb;18(1):53-64. 28. Trovik J, Wik E, Stefansson IM, Marcickiewicz J, Tingulstad S, Staff AC, et al. Stathmin Overexpression Identifies High-Risk Patients and Lymph Node Metastasis in Endometrial Cancer. Clin Cancer Res. 2011 May 10. 29. Birkeland E, Wik E, Trovik J, Werner HM, Mjos S, Kusanmano K, et al. KRAS gene amplification but not mutation associates with aggressive and metastatic endometrial cancer. British Journal of Cancer. Revision submitted Sept 2012. 30. Kutz SM, Higgins CE, Samarakoon R, Higgins SP, Allen RR, Qi L, et al. TGF-beta 1-induced PAI-1 expression is E box/USF-dependent and requires EGFR signaling. Exp Cell Res. 2006 Apr 15;312(7):1093-105. 31. Pomeroy SL, Tamayo P, Gaasenbeek M, Sturla LM, Angelo M, McLaughlin ME, et al. Prediction of central nervous system embryonal tumour outcome based on gene expression. Nature. 2002 Jan 24;415(6870):436-42. 32. Kenny PA, Enver T, Ashworth A. Receptor and secreted targets of Wnt-1/beta-catenin signalling in mouse mammary epithelial cells. BMC Cancer. 2005 Jan 10;5:3. 33. Verrecchia F, Chu ML, Mauviel A. Identification of novel TGF-beta /Smad gene targets in dermal fibroblasts using a combined cDNA microarray/promoter transactivation approach. J Biol Chem. 2001 May 18;276(20):17058-62. 34. Fujita N, Jaye DL, Kajita M, Geigerman C, Moreno CS, Wade PA. MTA3, a Mi-2/NuRD complex subunit, regulates an invasive growth pathway in breast cancer. Cell. 2003 Apr 18;113(2):207-19. 35. Nagai MA, Brentani MM. Gene expression profiles in breast cancer to identify estrogen receptor target genes. Mini Rev Med Chem. 2008 May;8(5):448-54. 36. Paredes J, Figueiredo J, Albergaria A, Oliveira P, Carvalho J, Ribeiro AS, et al. Epithelial E- and P- cadherins: Role and clinical significance in cancer. Biochim Biophys Acta. 2012 Dec;1826(12):297-311. 37. Saal LH, Johansson P, Holm K, Gruvberger-Saal SK, She QB, Maurer M, et al. Poor prognosis in carcinoma is associated with a gene expression signature of aberrant PTEN tumor suppressor pathway activity. Proc Natl Acad Sci U S A. 2007 May 1;104(18):7564-9. 38. Gustafson AM, Soldi R, Anderlind C, Scholand MB, Qian J, Zhang X, et al. Airway PI3K pathway activation is an early and reversible event in lung cancer development. Sci Transl Med. 2010 Apr 7;2(26):26ra5. 39. Creasman WT, McCarty KS, Sr., Barton TK, McCarty KS, Jr. Clinical correlates of estrogen- and progesterone-binding proteins in human endometrial adenocarcinoma. Obstet Gynecol. 1980 Mar;55(3):363-70. 40. Iversen OE, Utaaker E, Skaarland E. DNA ploidy and steroid receptors as predictors of disease course in patients with endometrial carcinoma. Acta Obstet Gynecol Scand. 1988;67(6):531-7. 41. Thigpen JT, Brady MF, Alvarez RD, Adelson MD, Homesley HD, Manetta A, et al. Oral medroxyprogesterone acetate in the treatment of advanced or recurrent endometrial carcinoma: a dose-response study by the Gynecologic Oncology Group. J Clin Oncol. 1999 Jun;17(6):1736-44. 42. Blechschmidt K, Kremmer E, Hollweck R, Mylonas I, Hofler H, Kremer M, et al. The E-cadherin repressor snail plays a role in tumor progression of endometrioid adenocarcinomas. Diagn Mol Pathol. 2007 Dec;16(4):222-8. 43. Dhasarathy A, Kajita M, Wade PA. The transcription factor snail mediates epithelial to mesenchymal transitions by repression of estrogen receptor-alpha. Mol Endocrinol. 2007 Dec;21(12):2907-18. 44. Wendt MK, Allington TM, Schiemann WP. Mechanisms of the epithelial-mesenchymal transition by TGF-beta. Future Oncol. 2009 Oct;5(8):1145-68. 45. Muinelo-Romay L, Colas E, Barbazan J, Alonso-Alconada L, Alonso-Nocelo M, Bouso M, et al. High- risk endometrial carcinoma profiling identifies TGF-beta1 as a key factor in the initiation of tumor invasion. Mol Cancer Ther. 2011 Aug;10(8):1357-66. 46. Tang D, Xu S, Zhang Q, Zhao W. The expression and clinical significance of the androgen receptor and E-cadherin in triple-negative breast cancer. Med Oncol. 2012 Jun;29(2):526-33. 47. Gravdal K, Halvorsen OJ, Haukaas SA, Akslen LA. A switch from E-cadherin to N-cadherin expression indicates epithelial to mesenchymal transition and is of strong and independent importance for the progress of prostate cancer. Clin Cancer Res. 2007 Dec 1;13(23):7003-11. 48. Oza AM, Elit L, Tsao MS, Kamel-Reid S, Biagi J, Provencher DM, et al. Phase II study of temsirolimus in women with recurrent or metastatic endometrial cancer: a trial of the NCIC Clinical Trials Group. J Clin Oncol. 2011 Aug 20;29(24):3278-85. 49. Dedes KJ, Wetterskog D, Ashworth A, Kaye SB, Reis-Filho JS. Emerging therapeutic targets in endometrial cancer. Nat Rev Clin Oncol. 2011 Jan 11. 50. Shayesteh L, Lu Y, Kuo WL, Baldocchi R, Godfrey T, Collins C, et al. PIK3CA is implicated as an oncogene in ovarian cancer. Nat Genet. 1999 Jan;21(1):99-102. 51. Courtney KD, Corcoran RB, Engelman JA. The PI3K pathway as drug target in human cancer. J Clin Oncol. 2010 Feb 20;28(6):1075-83.

52. Konopka B, Janiec-Jankowska A, Kwiatkowska E, Najmola U, Bidzinski M, Olszewski W, et al. PIK3CA mutations and amplification in endometrioid endometrial carcinomas: relation to other genetic defects and clinicopathologic status of the tumors. Hum Pathol. 2011 Nov;42(11):1710-9.

Figure legends

Figure 1. A and B, primary investigation set. A) ERα protein expression (by IHC) is

correlated to ESR1 mRNA expression. Upper insert picture shows an example of an ERα

positive tumor with strong staining for ERα in the majority of tumor nuclei in contrast to the

lower insert picture showing an ERα negative tumor with weak expression of ERα in only a

few tumor nuclei. B) ESR1 mRNA expression estimated by DNA oligonucleotide array and

by qPCR are highly correlated. Red and green colors indicate positive and negative ERα

status (IHC), respectively. Both ERα protein (C; retrospective validation cohort, D;

prospective validation cohort) and ESR1 mRNA expression (D: prospective validation cohort)

are strong predictors of disease specific survival in endometrial carcinoma. Survival curves

are estimated by the Kaplan-Meier method. For each category the number of cases is given

followed by the number of endometrial carcinoma deaths.

Figure 2. Correlations between ERα status, markers for EMT and outcome in endometrial

carcinomas. Primary investigation series: The signature scores of two of the gene sets

suggesting the presence of EMT and being enriched in ERα negative tumors (identified in

gene set enrichment analyses, FDR<0.25), were significantly associated with ERα IHC

expression (A and C) and disease specific survival (B and D). SERPINE-1, a suggested

marker for TGF-β signaling, relates to ERα IHC expression (E), and is also associated with a

TGF- β signature score (F).

Prospective validation series: The level for an EMT mRNA signature is a strong predictor of

disease specific survival (G) and is significantly associated with ESR1 mRNA expression (H).

Red and green color for high and low ERα IHC expression, respectively (F, H).

Survival curves are estimated by the Kaplan-Meier method according to level of the signature

scores. For each category the number of cases is given followed by the number of endometrial

carcinoma deaths. All the p-values are significant (<0.01) also after Bonferroni correction.

Figure 3. ESR1 mRNA expression levels in primary tumors according to (A) amplified

versus non-amplified 3q26 region and (B) PIK3CA mutated and wild-type tumors.

Figure 1

A C

6.0 ERα positive 1.0

5.0 ERα positive (198/39) 0.8

4.0 ERα negative 0.6 ERα negative (68/30) 3.0

0.4 2.0 ESR1 mRNA by qPCRby ESR1mRNA

1.0 0.2 Diseasespecific survival

0 P P < 0.001 0 < 0.001

ERα positive ERα negative 50 10 15 20 25 Time after diagnosis (years)

B D 1.0 High ERα IHC expression (118/9) 2.5 0.8 High ESR1 mRNA (116/15) 0 Low ESR1 mRNA (39/13) 0.6 -2.5 Low ERα IHC expression (35/15) 0.4 -5.0 r = 0.83

ESR1 mRNA (qPCR) ESR1mRNA s 0.2

-7.5 Diseasespecific survival P(IHC) <0.001 P < 0.001 P(mRNA) <0.001 -10.0 0

12.510.07.55.02.5 0 20 40 60 ESR1 mRNA (microarray) Time after diagnosis (months) Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2013 American Association for Cancer Research. Figure 2 30 A B Low TGF-β score (57/6) 2 1.0

20 Author Manuscript Published OnlineFirst on January 14, 2013; DOI: 10.1158/1078-0432.CCR-12-3039 Author manuscripts have been peer reviewed and accepted0.8 for publication but have not yet been edited.

0.6 10 High TGF-β score (19/8) 0.4 0 0.2 Disease specific survival specific Disease TGF-β signature score x10 score signature TGF-β P<0.001 P -10 =0.001 0 ERα ERα 0 20 40 60 80 100 positive negative Time after diagnosis (months) C 2 D 12 1.0 Low SHH score (57/5) 10 0.8 8 0.6 6 0.4 4 High SHH score (19/9)

2 0.2 Disease specific survival specific Disease P<0.001 P 0 <0.001 0 Sonic Hedgehog sicnature score x10 score sicnature Hedgehog Sonic ERα ERα 0 20 40 60 80 100 Time after diagnosis (months) positive negative E 12 F 12

10 10 P<0.001 rs= 0.6

8 8

6 6

4 SERPINE-1 mRNA 4 SERPINE-1 mRNA SERPINE-1 mRNA P=0.004 2 2 ERα ERα -10 0 10 20 30 positive negative TGF-β signature score x102 G H 1.0 Low EMT score (117/11) 4 0.8 2 0.6 0 0.4 High EMT score (38/14) -2

0.2 score signature EMT Disease specific survival specific Disease -4 P<0.001 P<0.001 rs= -0.7 0 0 20 40 60 0 2 4 6 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2013 American Association for Cancer Time after diagnosis (months) Research. ESR1 mRNA Author Manuscript Published OnlineFirst on January 14, 2013; DOI: 10.1158/1078-0432.CCR-12-3039 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Figure 3

A B 14 14

13 13

12 12

11 11

ESR1 mRNA expression ESR1 mRNA expression mRNA ESR1

10 10 P = 0.02 P = 0.8

3q26 region PIK3CA UnamplifiedDownloaded from clincancerres.aacrjournals.org Amplified on September 28, 2021.Wild © 2013type American AssociationMutated for Cancer Research. Downloaded from Author manuscriptshavebeenpeerreviewedandacceptedforpublicationbutnotyetedited. Table 1: Clinico-pathologic variables in relation to low ERα (by IHC or mRNA expression). Author ManuscriptPublishedOnlineFirstonJanuary14,2013;DOI:10.1158/1078-0432.CCR-12-3039

Retrospective clincancerres.aacrjournals.org Primary investigation Prospective validation validation External data cohort (n=76) cohort (n=155) cohort (n=286)a set (n=111)

IHC mRNAb IHC mRNAc IHC mRNAb Low ERα neg. Low ESR1 ERα neg. ESR1 ERα neg. Low ESR1 Variable Category n(%) Pc Pd n(%) Pc Pd n(%) Pc Pd Agee ≤ 65 years 10 (25) 1.0 0.1 17 (21) 0.6 0.1 31 (23) 0.4 No data on September 28, 2021. © 2013American Association for Cancer > 65 years 9 (25) 18 (25) 37 (28) Research.

FIGO stage I/II 12 (18) 0.004 < 0.001 19 (16) <0.001 < 0.001 47 (22) 0.002 < 0.001 III/IV 7 (64) 16 (50) 21 (43)

Histologic type E 14 (21) 0.04 < 0.001 17 (13) <0.001 < 0.001 50 (21) < 0.001 < 0.001 NEf 5 (56) 18 (69) 18 (69)

Histologic grade 1+2 9 (16) 0.003 < 0.001 9 (9) < 0001 < 0.001 33 (16) < 0.001 < 0.001 3 10 (50) 26 (51) 35 (56)

ERα negative (IHC) < 0.001 < 0.001 No data

Poor survivalg 0.01 0.002 <0.001 < 0.001 < 0.001 No data

Downloaded from Author manuscriptshavebeenpeerreviewedandacceptedforpublicationbutnotyetedited. Abbreviations: neg: negative, FIGO: International Federation of Gynecology and Obstetrics, IHC: immunohistochemistry, E: Endometrioid, NE: Author ManuscriptPublishedOnlineFirstonJanuary14,2013;DOI:10.1158/1078-0432.CCR-12-3039 Non-endometrioid. a Part of the data in the retrospective validation cohort reported previously (5), now with extended follow-up time. b mRNA expression c c

clincancerres.aacrjournals.org determined by DNA oligonucleotide microarray. mRNA expression determined by qPCR. Chi square and Fisher’s exact tests when d e f appropriate. Mann-Whitney U test. Median age, prospective validation series, 64 years. Carcinosarcomas, serous, clear cell and undifferentiated subtypes. gKaplan Meier survival analyses; log rank test. Missing data: Primary investigation set, Grade: n=1, Prospective validation series, Grade; n=1, ERα status (IHC): n=2. Retrospective validation series, ERα status: n=20, FIGO stage: n=1.

on September 28, 2021. © 2013American Association for Cancer Research. Author Manuscript Published OnlineFirst on January 14, 2013; DOI: 10.1158/1078-0432.CCR-12-3039 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Table 2: Retrospective validation cohort: Correlation between expression of ERα and Stathmin and epithelial-mesenchymal transition markers

ERαa Stathmina Positive Negative Low High EMT markersa n (%) n (%) Pb n (%) n (%) Pb

E-cadherin 0.047 n.s. High 94 (80) 23 (20) 83 (77) 25 (23) Low 103 (70) 45 (30) 91 (69) 41 (31) P-cadherin 0.007 0.07 Low 148 (79) 39 (21) 129 (76) 41 (24) High 50 (63) 29 (37) 45 (64) 25 (36) β-catenin 0.001 0.02 High 103 (84) 20 (16) 88 (80) 22 (20) Low 95 (66) 48 (34) 87 (66) 44 (34) p 120 0.046 <0.001 High 155 (77) 45 (23) 140 (80) 36 (20) Low 43 (65) 23 (35) 34 (53) 30 (47) Vascular invasionc <0.001 <0.001 Neg 127 (82) 28 (18) 116 (82) 25 (18) Pos 71 (64) 40 (36) 59 (59) 41 (41) Myometrial infiltration 0.02 0.01 < 50% 107 (80) 26 (20) 95 (77) 29 (23) ≥ 50% 59 (67) 29 (33) 47 (59) 32 (41)

Missing values amongst cases with ERα staining data (n=266): 1 E-cadherin, 45 myometrial infiltration. a Protein expression of EMT markers, ERα and Stathmin assessed by IHC. b Chi square- or Fisher’s exact test, as appropriate. c Vascular invasion: Positive = Invasion in ≥2 vessels (20).

Table 3. Drug signatures negatively correlated to ERα negative endometrial cancer

Rank Name of compound Known target/action Na Pb

Primary investigation cohort 1 Sirolimus mTOR inhibitor 44 < 0.00001 2 LY-294002 PI3Kinase inhibitor 61 < 0.00001 3 Tretinoin Retinoic acid receptor agonist 22 0.00008 4 Wortmannin PI3Kinase inhibitor 18 0.00022

External data set 1 Sirolimus mTOR inhibitor 44 < 0.00001 2 Wortmannin PI3Kinase inhibitor 18 0.00004 3 LY-294002 PI3Kinase inhibitor 61 0.00004

a N = number of instances in which the compounds were tested in the Connectivity map b The expression changes from the compounds tested were scored according to the ERα/ESR1 expression signatures and the p-value for each compound represents the distribution of this score in the N instances as compared with the distribution of these scores amongst all compounds tested, using a permutation test (24).

Lack of Estrogen receptor α is associated with epithelial-mesenchymal transition and PI3Kinase alterations in endometrial carcinoma.

Elisabeth Wik, Maria B. Ræder, Camilla Krakstad, et al.

Clin Cancer Res Published OnlineFirst January 14, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-12-3039

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2013/01/15/1078-0432.CCR-12-3039.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2013/01/12/1078-0432.CCR-12-3039. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.