Phosphorylated 4E Binding Protein 1: A Hallmarkof Cell Signaling That Correlates With Survivalin Ovarian Cancer

Josep Castellvi, MD1

Angel Garcia, MD1

Federico Rojo, MD1

Carmen Ruiz-Marcellan, MD1

Antonio Gil, MD2

Jose Baselga, MD3

Santiago Ramon y Cajal, MD, PhD1

1 Department of Pathology, Vall d’Hebron Univer-sity Hospital, Barcelona, Spain.

2 Department of Gynecology, Vall d’Hebron Uni-versity Hospital, Barcelona, Spain.

3 Department of Oncology, Vall d’Hebron Univer-sity Hospital, Barcelona, Spain.

BACKGROUND. Growth factor receptors and cell signaling factors play a crucial

role in human carcinomas and have been studied in ovarian tumors with varying

results. Cell signaling involves multiple pathways and a myriad of factors that

can be mutated or amplified. Cell signaling is driven through the mammalian tar-

get of rapamycin (mTOR) and extracellular regulated kinase (ERK) pathways and

by some downstream molecules, such as 4E binding protein 1 (4EBP1), eukaryo-

tic initiation factor 4E, and p70 ribosomal protein S6 kinase (p70S6K). The objec-

tives of this study were to analyze the real role that these pathways play in

ovarian cancer, to correlate them with clinicopathologic characteristics, and to

identify the factors that transmit individual proliferation signals and are asso-

ciated with pathologic grade and prognosis, regardless specific oncogenic altera-

tions upstream.

METHODS. One hundred twenty-nine ovarian epithelial tumors were studied,

including 20 serous cystadenomas, 7 mucinous cystadenomas, 11 serous borderline

tumors, 16 mucinous borderline tumors, 29 serous carcinomas, 16 endometrioid

carcinomas, 15 clear cell carcinomas, and 15mucinous carcinomas. Tissue microar-

rays were constructed, and immunohistochemistry for the receptors epidermal

growth factor receptor (EGFR) and c-erb-B2 was performed and with phosphoryl-

ated antibodies for protein kinase B (AKT), 4EBP1, p70S6K, S6, and ERK.

RESULTS. Among 129 ovarian neoplasms, 17.8% were positive for c-erb-B2, 9.3%

were positive for EGFR, 47.3% were positive for phosphorylated AKT (p-AKT), 58.9%

were positive for p-ERK, 41.1% were positive for p-4EBP1, 26.4% were positive for

p70S6K, and 15.5% were positive for p-S6. Although EGFR, p-AKT, and p-ERK

expression did not differ between benign, borderline, or malignant tumors, c-erb-

B2, p-4EBP1, p-p70S6K, and p-S6 were expressed significantly more often in malig-

nant tumors. Only p-4EBP1 expression demonstrated prognostic significance

(P ¼ .005), and only surgical stage and p-4EBP1 expression had statistical signifi-

cance in the multivariate analysis.

CONCLUSIONS. In patients with ovarian carcinoma, significant expression of p-4EBP1

was associated with high-grade tumors and a poor prognosis, regardless other onco-

genic alterations upstream. This finding supports the study of this factor as a hallmark

or pivotal factor in cell signaling in ovarian carcinoma that may crucial in the transmis-

sion of the proliferation cell signal and may reflect the real oncogenic role of this path-

way in ovarian tumors. Cancer 2006;107:1801–11.� 2006 American Cancer Society.

KEYWORDS: cell signaling, ovarian neoplasms, prognostic factors, 4E binding protein 1.

M ost human neoplasms have broad clinical, pathologic, and mo-

lecular heterogeneity; and, in nearly all malignant solid

tumors, there is oncogenic activation of cell immortalization, cell

Supported by Grant SAF2002-02184 (to S.R.y.C.)from the Spanish Ministry of Education.

We thank Dr. Carlos Cordon-Cardo from MemorialSloan-Kettering Cancer Center (New York, NY)and Dr. Jaime Prat from Hospital de Sant Pau(Barcelona, Spain) for their critical review of thearticle. We also thank Jose Jimenez and SoniaRodriguez for technical assistance.

Address for reprints: Santiago Ramon y Cajal, MD,PhD, Department of Pathology, Vall d’Hebron Univer-sity Hospital, Pg. Vall d’Hebron, 119-129, 08035Barcelona, Spain; Fax: (011) 34 932746818; E-mail:sramon@vhebron.net

Received March 9, 2006; revision received June1, 2006; accepted July 5, 2006.

ª 2006 American Cancer SocietyDOI 10.1002/cncr.22195Published online 18 September 2006 in Wiley InterScience (www.interscience.wiley.com).

1801

signaling, cell cycle, apoptosis, and tumor-invasive

pathways.1 To date, > 350 cancer genes have been

identified, and hundreds of epigenetic alterations have

been described in tumors.2 In recent years, with the

study of messenger RNA (mRNA) arrays, new sets of

genetic alterations have been proposed in tumors,

some of them related to histologic type, prognosis,

and response to anticancer treatments.

Ovarian carcinoma is the gynecologic malignancy

associated with the highest mortality in industrialized

countries, with a 5-year survival rate of < 30% reported

in some series.3 The prognosis for patients with ovar-

ian cancer is determined by conventional factors, such

as histologic type, histologic grade, and surgical stage;

however, to our knowledge to date, no single molecu-

lar profile has helped to identify the most aggressive

tumors. In addition, little is known regarding the mo-

lecular mechanisms involved in malignant transfor-

mation or the biochemical alterations of the cell cycle,

cell signaling, apoptosis, and angiogenesis-related

pathways affected. The genes involved in these path-

ways show great redundancy, and many genetic altera-

tions have been described. All carcinomas show

activated growth factor receptor-cell signaling path-

ways with genetic alterations that involve either the

receptor, or some of the factors that drive the pro-

liferation signal downstream, or both. Although nor-

mal cell growth depends on 2 types of stimuli, 1 type

from growth factors and another type from the pre-

sence of nutrients, malignant cells acquire so-called

‘‘self sufficiency in growth signals,’’ which means that

malignant cells harbor mutation or gene amplification

either in the growth factor receptor family (Ras, phos-

phatase and tensin homolog, phosphoinositide-3 ki-

nase [PI3K]) or in other factors involved in the

labyrinthine routes of cellular pathways downstream.4

Most stimuli follow different pathways that converge

in the mammalian target of rapamycin (mTOR),5

which plays a central role in cell growth control (Fig.

1), and this factor can phosphorylate p70 ribosomal

protein S6 kinase (p70S6K) and 4EBP1, which clearly

are related to protein synthesis and cell proliferation.

4EBP1 dimerizes with eukaryotic initiation factor 4E

(eIF4E), blocking the formation of the initiation com-

plex. When 4EBP1 is phosphorylated, eIF4E is re-

leased, and translation can begin. The principal

substrate of the kinase p70S6K is ribosomal protein S6,

which is part of the 40S ribosomal unit. It is involved

in the translation of proteins encoded by 50 terminal

oligopyrimide genes. Most of these proteins are re-

gulators of protein synthesis and can act as proto-

oncogenes.6 It is noteworthy that 4EBP1 has 8

phosphorylation sites7 and also can be phosphoryl-

ated by other kinases, including PI3K, ataxia telangiec-

tasia mutated kinase, mitogen-activated protein kinase/

extracellular signal-regulated kinase (MEK), and others

that remain unknown. Wild-type p53 can modulate

4EBP1 through cdc2 or other unidentified phospha-

tases.8 Other p70S6K targets also include transcription

factors or translation regulators.9 Moreover, p70S6K

and eIF4E can be activated by pathways that are inde-

pendent of mTOR, including the Ras-Raf-MEK-ERK

pathway.10

In previous studies, it was observed that epider-

mal growth factor receptor (EGFR), c-erb-B2, protein

kinase B (AKT), and ERK were expressed frequently

in ovarian neoplasms11–14; however, downstream

effectors have not been characterized well in these

tumors. We hypothesized that very few factors should

exist downstream that could relay the oncogenic sig-

nal to the ribosomes or nucleus, regardless of the

individual oncogenic alterations upstream. Then,

these final factors would be transmitters of the pro-

liferation signal, regardless the specific oncogene

alteration upstream, and could be like a choke point

for these pathways.

FIGURE 1. The signaling pathways involved in cell growth control are illu-strated in this diagram, which shows the main cell signaling pathways stud-

ied. 4E binding protein 1 (4EBP1) is proposed as a central factor in which

diverse oncogenic signals may converge, providing a possible explanation for

the high percentage of positive tumors and its significant clinical correlation.

TK indicates tyrosine kinase, PI3K, phosphoinositide-3 kinase; PTEN, phos-

phatase and tensin homolog; MEK, mitogen-activated protein kinase/ERK

kinase; AKT, protein kinase B; ATM, ataxia telangiectasia mutated; PLD, phos-

pholipase D; Aa, amino acids; mTOR, mammalian target of rapamycin; ERK,

extracellular signal-regulated kinase; MNK, mitogen-activated protein kinase-

interacting kinase; eIF4E, eukaryotic initiation factor 4E; 50 TOP, 50 terminaloligopyrimide; CAP, a structure that stabilizes the RNA; FGF, fibroblast growth

factor; VEGF, vascular endothelial growth factor.

1802 CANCER October 15, 2006 / Volume 107 / Number 8

To investigate this hypothesis, we studied activa-

tion of the receptors EGFR and c-erb-B2, the 2 main

biochemical pathways AKT and mitogen-activated

protein kinase ERK, and the downstream factors

4EBP1, p70S6K, and S6 by immunohistochemistry

using specific phosphorylated antibodies. Expression

of these factors was measured in ovarian tumors,

and the results were correlated with patient survival

and tumor recurrence to determine which, if any, of

these factors fulfilled the concept of a ‘‘hallmark fac-

tor’’ of the pathway and was associated with clinical

prognosis regardless the oncogenic alterations

upstream.

MATERIALS AND METHODSPatients and Tissue CharacteristicsOver 5 years, from January 1994 to December 1998,

129 patients underwent surgery for primary epithelial

ovarian tumors at Hospital Vall d’Hebron (Barcelona,

Spain). The tumors included 20 serous cystadeno-

mas, 7 mucinous cystadenomas, 11 serous borderline

tumors, 16 mucinous borderline tumors, 29 serous

adenocarcinomas, 15 mucinous adenocarcinomas, 16

endometrioid adenocarcinomas, and 15 clear cell

adenocarcinomas. The patients ranged in age from

20 years to 87 years (mean age, 55 years). All Interna-

tional Federation of Gynecology and Obstetrics

(FIGO) stages of ovarian carcinomas were repre-

sented; however, because of the small number of

patients in some subgroups and the uneven distribu-

tion of patients, only the 4 major groups (Stages I

through IV) were used in the analysis. Thus, the se-

ries included 29 Stage I tumors, 5 Stage II tumors, 37

Stage III tumors, and 4 Stage IV tumors. Follow-up

information included disease-free survival, overall

survival, and cancer-related death. All data collected

were entered prospectively into the ovarian cancer

registry data base. The mean follow-up was 31

months (range, 24–80 months).

The tumors had been fixed in neutral formalin

and embedded in paraffin. In all patients, the diag-

nosis was based on a light microscopy examination

using conventional hematoxylin and eosin stain. For

immunohistochemical staining, tissue microarrays

from all patients were constructed, including 3 cores

that measured 2 mm in greatest dimension for each

patient.

Cell Lines and Western Blot AnalysisThe human breast MDA-MB-435 lung-2 cell line15

was grown in Dulbecco Medium Essential Medium

(GibcoBRL) containing 10% fetal calf serum (Bio-

Whittaker Europe). To validate the specificity of the

antibodies used in this study, Western blot analysis

was performed in the cell line and in lysates of fro-

zen tissue from 15 human ovarian tumors (Fig. 2).

Protein was quantified (Biorad assay; Biorad, Mun-

chen, Germany), and lysates were processed for Wes-

tern blot analysis as reported previously16,17 by using

the same antibodies that were used for immuno-

histochemistry (Table 1). Immunocomplexes were

washed extensively and resuspended in sample

buffer � 5. Antibody detection was achieved by using

enhanced chemiluminescence (Amersham Pharma-

cia, Uppsala, Sweden). Antibody expression also was

studied in paraffin embedded histologic specimens

from the same tumors (Fig. 2).

Antibodies and ImmunohistochemistryImmunohistochemical staining using the avidin-bio-

tin-peroxidase technique was carried out for each

antibody. Five-micrometer sections were cut from

the tissue specimens and placed on poly-L-lysine-

coated glass slides. Sections were deparaffined in xy-

lene and rehydrated in graded alcohol. Endogenous

peroxidase was blocked by immersing the sections in

0.1% hydrogen peroxide in absolute methanol for

20 minutes. For antigen retrieval, the tissue sections

were heated in a pressure cooker in 10 mM citric

acid monohydrate (pH 6.0) for 5 minutes and then

incubated with the primary antibody at room tem-

perature. The primary antibodies, dilutions, and

incubation times used are shown in Table 1. Immu-

nostaining was performed with the EnVision sys-

tem (DakoCytomation, Glostrup, Denmark). All slides

were counterstained with hematoxylin, dehydrated,

and mounted. Negative controls were performed by

FIGURE 2. Western blot validation of immunohistochemical stainingsshowed a good correlation between Western blot analysis and immunohisto-

chemical analysis for phosphorylated 4E binding protein 1 in serous carci-

noma (SC), endometrioid carcinoma (EC), clear cell carcinoma (CCC), serous

borderline tumor (SBLT), and mucinous borderline tumor (MBLT). The positive

control (C þ) for Western blot analysis was the breast carcinoma cell lineMDA-MB-435.

Phospo-4EBP1 in Epithelial Ovarian Neoplasms/Castellvi et al. 1803

omitting the primary antibody. All primary antibodies

were tested first by Western blot to evaluate the

specificity of the staining in both the cell lines and

the tumor samples.

To evaluate immunohistochemical staining, we

scored the percentage of positive cells and intensity

of the staining, which was assessed semiquantita-

tively. Samples that showed no immunostaining were

considered negative, and samples that showed any

positivity were grouped together for statistical pur-

poses. Some markers, such as phosphorylated 4EBP1

(p-4EBP1) and p-p70S6K, showed 2 types of positive

expression: a weak, uniform nuclear staining (Fig. 3A)

and a rough, moderate, or intense staining (Fig. 3D).

Moderate-to-strong positivity was identified easily at

low magnification (� 40), whereas weak positivity was

visible at high magnification (� 400). Some of the

samples that had intense nuclear staining also showed

cytoplasmic, granular positivity. Because weak positiv-

ity was observed in benign ovarian tumors, normal

skin samples, and other normal tissues, such as lobu-

lar breast epithelium, we considered this intensity

pattern for p-4EBP1 and p-p70S6K, as well as the ab-

sence of staining, as negative. For EGFR, we consid-

ered samples positive if they had membrane staining

in > 25% of cells; and, for c-erb-B2, we used the same

scoring system that is used commonly for breast

tumors and only considered samples positive if they

showed complete membrane staining (2 þ and 3 þ).

When the 3 cores from the same patient showed dif-

ferent positivity results, then the highest score was

considered valid.

Statistical AnalysisStatistical analysis was performed with the Statistical

Package for Social Science (SPSS version 13.0; SPSS

Inc., Chicago, IL). Categorical variables were ana-

lyzed by using cross-tabulation, and differences were

evaluated by using the chi-square test. A 2-sided P

value � .05 was considered indicative of a statisti-

cally significant difference. The effects of the various

parameters on survival, including histologic grade

and type, FIGO stage, and immunostaining results,

were tested with the Kaplan–Meier method, and dif-

ferences were compared by using the log-rank test.

Multivariate analysis for overall survival was per-

formed with the Cox proportional hazards method.

RESULTSAntibody Validation by Western Blot AnalysisTo validate the paraffin results, we correlated the im-

munostaining obtained in paraffin sections with the

intensity and specificity obtained by Western blot

analysis of fresh tumors. In the 15 tumors that were

TABLE 1Primary Antibodies and Dilutions Used

Protein

Phosphorylation

site Source

Antibody

type Dilution

Incubation

time, minutes

c-erb-B2 — Biogenex Monoclonal 1/20 60

EGFR — DakoCytomation Monoclonal 1/100 60

p70S6K Thr 389 Cell Signaling Tech. Monoclonal 1/100 60

S6 Ser 240/Ser 244 Cell Signaling Tech. Polyclonal 1/50 30

AKT Ser 473 Cell Signaling Tech. Polyclonal 1/50 120

4EBP1 Thr 70 Cell Signaling Tech. Polyclonal 1/100 60

ERK1/ERK2 Thr 202/Tyr 204 Cell Signaling Tech. Polyclonal 1/50 120

EGFR indicates epidermal growth factor receptor; Thr, threonine; p70S6K, p70 ribosomal protein S6 kinase; AKT, protein kinase B; Ser, serine; 4EBP1, 4E binding

protein P1; ERK, extracellular signal-regulating kinase; Tyr, tyrosine.

FIGURE 3. Immunohistochemical staining for phosphorylated 4E bindingprotein 1 is observed in benign, low-grade, and high-grade tumors, including

(A) ovarian mucinous cystadenoma, (B) serous cystadenoma, (C) mucinous bor-

derline tumor with mild nuclear staining, and (D) serous carcinoma with intense

nuclear and cytoplasmic staining (original magnification � 400 in A-D).

1804 CANCER October 15, 2006 / Volume 107 / Number 8

studied, a good correlation was found between the

intensity of the band on Western blot analysis and

the immunoreactivity of paraffin tissue (Fig. 2).

EGFR and c-erb-B2 expression patternsBoth EGFR and c-erb-B2 proteins were expressed in

tumor cell membranes, and granular cytoplasmic

positivity also was observed in some tumors (Fig. 4).

EGFR was observed in < 10% of tumors and without

significant differences between benign, borderline,

and malignant tumors, although benign tumors

rarely expressed the protein (Tables 2 and 3). No cor-

relation was observed with histologic type, grade, or

surgical stage. Conversely, only malignant tumors

were positive for c-erb-B2. High-grade carcinomas

showed c-erb-B2 expression more frequently than

low-grade carcinomas, although no differences were

observed among surgical stages. When associations

between the expression of c-erb-B2 and the expres-

sion the other markers were assessed, we observed

no significant statistical correlation with p-AKT

(P ¼ .162) or p-ERK (P ¼ .807), and we observed sig-

nificant correlations with p-4EBP1 (P ¼ .016) and p-

p70S6K (P ¼ .030). EGFR expression did not correlate

with the expression of any other marker.

p-AKT expression patternsp-AKT was expressed in either the nucleus or the

cytoplasm, and both staining patterns were consid-

ered positive. Most tumors with nuclear positivity

also had some cytoplasmic expression of p-AKT.

Approximately 50% of tumors were positive for p-

AKT, and there were no differences in the percentage

of positive cells or the intensity of expression

between benign, borderline, or malignant tumors.

Mucinous carcinomas and low-grade carcinomas

were positive for p-AKT less often. Some tumors

showed intense nuclear positivity of stromal cells

with no relation to epithelial expression or morphol-

ogy. With regard with downstream effectors, p-AKT-

positive tumors showed positivity for p-4EBP1 in 34

patients (55%; P ¼ .002) and for p-p70S6K in 24

patients (39%; P ¼ .002). In 23 patients (37%), both

markers were negative (P ¼ .003).

p-ERK expression patternsImmunohistochemistry was positive for p-ERK in ei-

ther the nucleus or the cytoplasm, although only nu-

clear staining positive was considered positive (Fig. 4).

Like p-AKT, no significant differences were observed

in p-ERK expression between benign, borderline, or

malignant tumors. Clear cell carcinomas were positive

more often. Some tumors had strong stromal cell cyto-

plasmic and nuclear staining. Tumors that expressed

p-ERK were positive for p-4EBP1 in 40 patients (53%;

P ¼ .002) and positive for p-p70S6K in 25 patients

(33%; P ¼ .067). In 22 patients (29%), both markers

were negative (P ¼ .011).

p-4EBP1 expression patternsp-4EBP1 immunostaining was mainly nuclear and

was expressed in 53 tumors (41.1%), with significantly

lower expression observed in mucinous tumors. Ma-

lignant tumors were positive for 4EBP1 more often

than benign or borderline tumors, and there was a sig-

nificant correlation with histologic grade (P < .0001)

(Fig. 5). Cytoplasmic staining was observed in 18

tumors (25%) and was correlated with higher histologic

FIGURE 4. These photomicrographs illustrate immunohistochemical stain-ing for c-erb-B2, epidermal growth factor receptor (EGFR), phosphorylated

protein kinase B (p-AKT), phosphorylated extracellular signal-regulated kinase

(p-ERK), phosphorylated p70 ribosomal protein S6 kinase (p-p70S6K), and

phosphorylated ribosomal protein S6 (p-S6) in ovarian tumors. (A) c-erb-B2

membrane positivity is shown in a high-grade clear cell carcinoma. (B) EGFR

expression is shown in a serous carcinoma. (C) This endometrioid carcinoma

has moderate nuclear and cytoplasmic expression of p-AKT. (D) This high-

grade serous carcinoma has strong cytoplasmic and nuclear staining for

ERK. (E) This high-grade serous carcinoma has nuclear expression of p-

p70S6K. Some cytoplasmic staining also is observed. (F) Diffuse cytoplasmic

expression of p-S6 is observed in a high-grade endometrioid carcinoma (ori-

ginal magnification � 280 (A,C,E,F); � 400 (D).

Phospo-4EBP1 in Epithelial Ovarian Neoplasms/Castellvi et al. 1805

grades (P ¼ .037). Moreover, cytoplasmic staining was

observed only in malignant tumors. No correlation was

observed between nuclear or cytoplasmic expression of

the protein and the surgical stage of the tumor. Tumors

that were positive for p-4EBP1 expressed p-AKT in 34

tumors (64%; P ¼ .002) and expressed p-ERK in 40

tumors (75%; P ¼ .001). Only 10 tumors (19%) were

negative for both p-AKTand p-ERK (P < .0001).

p-p70S6K expression patternsNuclear immunohistochemistry for p-p70S6K was posi-

tive in 30% of tumors, and additional cytoplasmic

expression was found in 15% of tumors. Malignant

tumors showed greater nuclear expression of p-p70S6K

than benign or borderline tumors. Cytoplasmic staining

was observed only in malignant tumors, although there

was no correlation with histologic grade. Among the

ovarian carcinomas, high-grade tumors showed stron-

ger nuclear immunostaining, although the difference

was not significant (P ¼ .109). Clear cell carcinomas

were the most frequently positive tumors, and muci-

nous carcinomas were the least frequently positive

tumors. Tumors that were positive for p-p70S6K also

were positive for p-ERK in 25 tumors (73.5%; P ¼ .067)

and were positive for p-AKT in 24 tumors (70.6%;

P ¼ .002). Only 5 tumors (14.7%) did not express either

TABLE 3Immunohistochemical Results and Clinicopathologic Characteristics in Patients withEpithelial Ovarian Carcinoma

Characteristic

No. of patients (%)

Total

no. c-erb B2 EGFR p-AKT p-ERK p-4EBP1 p-p70S6K p-S6

Age

< 60 y 34 11 (34.4) 3 (9.4) 20 (62.5) 24 (75.0) 22 (68.8) 14 (43.8) 8 (25.0)

> 60 y 41 12 (27.9) 5 (11.6) 19 (44.2) 22 (51.2) 19 (44.2) 14 (32.6) 10 (23.3)

Histologic type

Serous carcinoma 29 8 (27.6) 3 (10.3) 15 (51.7) 19 (65.5) 18 (62.1) 10 (34.5) 5 (17.2)

Mucinous carcinoma 15 1 (6.7) 1 (6.7) 3 (20.0) 6 (40.0) 2 (13.3) 2 (13.3) 4 (26.7)

Endometrioid carcinoma 16 7 (43.8) 3 (18.8) 10 (62.5) 8 (50.0) 9 (56.3) 7 (43.8) 4 (25.0)

Clear cell carcinoma 15 7 (46.7) 1 (6.7) 11 (73.3) 13 (86.7) 12 (80.0) 9 (60.0) 5 (33.3)

Histologic grade

Grade 1 11 1 (9.1) 0 (0.0) 2 (18.2) 4 (36.4) 1 (9.1) 2 (18.2) 2 (18.2)

Grade 2 13 3 (23.1) 4 (30.8) 7 (53.8) 7 (53.8) 5 (38.5) 3 (23.1) 5 (38.5)

Grade 3 51 19 (37.3) 4 (7.8) 30 (58.8) 35 (68.6) 35 (68.6) 23 (45.1) 11 (21.6)

Disease stage

Stage I-II 34 7 (20.6) 3 (8.8) 17 (50.0) 18 (52.9) 16 (47.1) 14 (41.2) 8 (23.5)

Stage III-IV 41 16 (39.0) 5 (12.2) 22 (53.7) 28 (68.3) 25 (61.0) 14 (34.1) 10 (24.4)

EGFR indicates epidermal growth factor receptor; p-AKT, phosphorylated protein kinase B; p-ERK, phosphorylated ERK; p-4EBP1, phosphorylated 4E binding

protein 1; p-p70S6K, phosphorylated p70 ribosomal protein S6 kinase; p-S6, phosphorylated ribosomal protein S6.

TABLE 2Summary of Immunohistochemical Results Comparing Benign, Borderline, and Malignant Ovarian Tumors

Tumor type

No. of patients (%)

Total

no. c-erb-B2 EGFR p-AKT p-ERK p-4EBP1 p-p70S6K p-S6

Benign 27 0 (0.0) 1 (3.7) 9 (33.3) 14 (51.9) 5 (18.5) 2 (7.4) 1 (3.7)

Borderline 27 0 (0.0) 3 (11.1) 13 (48.1) 16 (59.3) 7 (25.9) 4 (14.8) 1 (3.7)

Malignant 75 23 (30.7) 8 (10.7) 39 (52.0) 46 (61.3) 41 (54.7) 28 (37.3) 18 (24.0)

Total 129 23 (17.8) 12 (9.3) 61 (47.3) 76 (58.9) 53 (41.1) 34 (26.4) 20 (15.5)

P < .0001 .529 .248 .691 .001 .003 .007

EGFR indicates epidermal growth factor receptor; p-AKT, phosphorylated protein kinase B; p-ERK, phosphorylated ERK; p-4EBP1, phosphorylated 4E binding

protein 1; p-p70S6K, phosphorylated p70 ribosomal protein S6 kinase; p-S6, phosphorylated ribosomal protein S6.

1806 CANCER October 15, 2006 / Volume 107 / Number 8

protein (P ¼ 0.010). Conversely, 12 tumors (35%) that

were positive for p-p70S6K expressed p-S6 (P ¼ .001).

p-S6 expression patternsp-S6 was expressed in cytoplasm, and immunohisto-

chemical results were positive in only 15.5% of

tumors. Benign and borderline tumors mostly were

negative (P ¼ .004), although there were no differ-

ences among the histologic types. Sixty percent of

p-S6-positive tumors showed p-p70S6K expression

(P ¼ .009).

Because p70S6K and 4EBP1 can be activated by

both the AKT and ERK pathways, we determined the

effect of activation of either or both pathways on

expression of the phosphorylated forms of these pro-

teins. We observed that p-p70S6K expression was

more frequent in tumors in which both AKT and ERK

were phosphorylated (58%) compared with tumors in

which only 1 protein was expressed (26%) or in

which both proteins were negative (14%; P ¼ .026).

These findings were more significant for 4EBP1

expression (i.e., 58% of tumors that were positive for

4EBP1 demonstrated coexpression of p-AKT and p-

ERK, 22% were positive only for 1 protein, and 19%

were negative for both proteins [P < .0001]).

Survival AnalysisOf 75 patients with ovarian carcinoma, 22 died of

disease, and 14 survived with progressive disease.

One of 27 patients who had borderline tumors had

died of disease at the time of last follow-up, and

another was alive with progressive disease. The mean

survival was 55 months (range, 13–75 months). Eight

patients (5 patients with carcinomas and 3 patients

with borderline tumors) died of unrelated causes. A

survival analysis of patients with carcinoma in rela-

tion to activation of the proteins studied showed a

significant log-rank value only for p-4EBP1 nuclear

expression (Table 4) when overall survival was

assessed (Fig. 6); however, similar to the other mar-

kers, p-4EBP1 was not significant for progression-free

survival (P ¼ .052). Tumors that were negative for

TABLE 4Overall Survival Results of the Signaling Proteins Studied

Signaling protein

Mean survival, months

P (Log-rank)Negative patients Positive patients

c-erb-B2 44.87 55.13 .808

EGFR 52.84 55.53 .820

p-AKT 54.04 53.27 .370

p-ERK 61.12 49.47 .151

p-4EBP1 70.55 46.87 .005*

p-p70S6K 58.75 43.38 .197

p-S6 56.35 43.97 .906

EGFR indicates epidermal growth factor receptor; p-AKT, phosphorylated protein kinase B; p-ERK,

phosphorylated extracellular signal-regulated kinase; p-4EBP1, phosphorylated 4E binding protein 1;

p-p70S6K, phosphorylated p70 ribosomal protein S6 kinase; p-S6, phosphorylated ribosomal protein S6.

* p-4EBP1 was the only marker that had clear prognostic significance.

FIGURE 6. Overall survival curves are shown in relation to phosphorylated4E binding protein 1 (p-4EBP1) expression.

FIGURE 5. Phosphorylated 4E binding protein 1 (p-4EBP1)-positive tumorsare illustrated according to histologic grade in patients with ovarian carcino-

mas. Most tumors with p-4EBP1 expression were high-grade carcinomas,

whereas only 1 well differentiated carcinoma (2%) was positive.

Phospo-4EBP1 in Epithelial Ovarian Neoplasms/Castellvi et al. 1807

p-4EBP1 accounted for only 7.4% of deaths, whereas

tumors that were positive for p-4EBP1 represented

43.9% of deaths. In the multivariate analysis with

Cox regression that included histologic grade, surgi-

cal stage, and p-4EBP1 expression, only surgical

stage (P ¼ .010) and p-4EBP1 expression (P ¼ .028)

were statistically significant.

DISCUSSIONStudy of the signaling pathways involved in cell

growth, the cell cycle, and apoptosis has contributed

greatly to our understanding of the molecular

mechanisms of carcinogenesis in human tumors.

The interest in this line of study is enhanced further

when there are potential or known therapeutic tar-

gets involved, prognostic factors, or factors predictive

of resistance to therapy. These signaling pathways

are crucial in many malignancies, including breast,

prostate, and gastric carcinomas, lymphomas, and

multiple myelomas.18–20 Few studies have evaluated

the role of these molecules in ovarian carcinomas,

and most of the work has been done in cell lines. For

this reported, we evaluated the main components of

the cell signaling pathways, including the epithelial

growth factor receptors EGFR and c-erb-B2, the ERK

and AKT pathways, and their downstream factors,

p70S6K, S6 and 4EBP1, in a large series of ovarian

tumors. The primary objective of this study was to

correlate the expression of these factors with patho-

logic features and the patient’s clinical prognosis

and, ultimately, to identify factors that may reflect

the real oncogenic role of these important biochem-

ical pathways in ovarian tumors regardless the onco-

genic alterations present upstream.

With regard to c-erb-B2 and EGFR, we observed

their expression in 17.8% and 9.3% of ovarian tumors,

respectively. In other studies, the percentage of c-erb-

B2-positive tumors ranged from 15% to 30%.12,21,22 It

is noteworthy that c-erb-B2 overexpression in our se-

ries was observed only in carcinomas: mostly in high-

grade carcinomas. Surprisingly, despite this correla-

tion with aggressive morphology, results of the survival

study did not indicate statistical significance. Although

the predictive value of c-erb-B2 in ovarian carcinomas

is not clear, some studies have shown a significantly

worse prognosis for patients with positive

tumors,11,12,21 whereas others did not observe this cor-

relation.23–25 Similarly, EGFR reportedly was overex-

pressed in 30% to 77% of ovarian carcinomas, and its

association with a poor prognosis, like in our series,

was not verified.11,25–27

Downstream of the growth factor receptors, there

are at least 2 main biochemical routes. It is known

that the PI3K-AKT-mTOR pathway and the Ras-Raf-

MEK-ERK pathways are activated in ovarian carcino-

mas. Regarding the PI3K-AKT-mTOR pathway, p-AKT

is found in 36% to 68% of tumors, and 55% of

tumors express mTOR.13,28 Our current results indi-

cated that AKT was activated in 47% of carcinomas,

but no significant differences were observed between

benign or borderline tumors or between histologic

grades. Moreover, the results did not demonstrate

that p-AKT expression had prognostic significance.

These results contrast with other reports, which

reported the prognostic value of AKT activation in

certain tumors, such as melanomas29 and lung30 or

prostate carcinomas,31 but they are in agreement

with studies in breast carcinoma in which no correla-

tion was observed.32 The downstream effectors of

this pathway, 4EBP1 and p70S6K, frequently are over-

expressed in ovarian carcinomas. p-4EBP1 is ex-

pressed in 55% of carcinomas and is more frequent

in high-grade tumors. It is worth noting that, in the

current study, the overexpression of p-4EBP1 was

correlated with a poor prognosis, even in the multi-

variate analysis, regardless of the surgical stage. In

contrast, p-p70S6K was overexpressed in 37% of car-

cinomas but did not correlate with histologic grade,

surgical stage, or survival. Only 35% of tumors that

were positive for p-p70S6K were also positive for p-

S6 positivity in our series, supporting the finding that

p70S6K has substrates other than the S6 ribosomal

protein. The 3 downstream molecules were expressed

more often in malignant tumors than in benign or

borderline tumors.

It has been demonstrated that the Ras-Raf-MEK-

ERK pathway also is important in ovarian carcinomas,

particularly in serous carcinomas, in which it carries

prognostic significance.14,33 K-ras and BRAF mutations

often are found in serous borderline tumors and in

low-grade serous carcinomas.34 These mutations sup-

port the existence of 2 different carcinogenic pathways

for serous carcinomas: a pathway for low-grade carci-

nomas, which probably results from progression of a

borderline tumor with K-ras and BRAF mutations, and

a pathway for high-grade serous carcinomas with a

low incidence of mutations in these molecules.35 K-ras

and BRAF mutations also are frequent in mucinous

tumors.36 The effector of these molecules is ERK, and

the phosphorylated form of the protein can be an indi-

cator of the overall status of this pathway. Moreover, it

has been demonstrated that active ERK is a good prog-

nostic marker of high-grade serous carcinomas by

other pathways that have not been identified well.14 In

the current study, no significant differences in p-ERK

expression were observed between borderline and

high-grade serous carcinomas. Activated ERK can

1808 CANCER October 15, 2006 / Volume 107 / Number 8

phosphorylate p70S6K at a different site than mTOR.7

In multiple myeloma cells, the phosphorylation of

both sites reportedly showed a synergistic effect.37 In

our series, positivity for p-p70S6K was stronger

in tumors with p-ERK and p-AKT coexpression than in

tumors in which only 1 or neither was expressed. Simi-

larly, 4EBP1 can be activated by the ERK pathway.38,39

Moreover, for p-4EBP1 expression, we observed the

same synergistic effect when ERK and AKT were

expressed (Fig. 7), suggesting that these 2 pathways

involved in translational control are interrelated. This

effect was not observed in relation to receptor status

(Fig. 7). In this sense, in a recent study, it was observed

that, by blocking both pathways simultaneously, there

was a synergistic effect on the inhibition of cell prolif-

eration.40

Because p-4EBP1 was the only studied markers

that was associated with prognosis and histologic

grade, it may represent a hallmark of the activation

status of at least 2 signaling pathways. Our results

and conclusions regarding 4EBP1 are supported by

the oncogenic role of eIF4E, which is released when

4EBP1 is phosphorylated. Previous data have de-

scribed 4EBP1-eIF4E in human tumors and animal

models.41–44 In fact, eIF4E is an essential component

of the malignant phenotype in breast carcinoma,45

and hyperphosphorylation of 4EBP1 is crucial in this

effect. The high cytoplasmic level of 4EBP1 observed

in a subset of high-grade tumors may reflect a hyper-

phosphorylated state and, indirectly, concomitant

oncogenic alterations upstream. This means that

4EBP1 can be phosphorylated by mTOR, which can

be activated by AKT,5 phospholipase D,46 and other

kinases and by other PI3K kinases of the Ras-Raf-

MEK-ERK pathway.39 It is noteworthy that, because

p53 mediates dephosphorylation of p-4EBP1,8 non-

functional or mutated p53 also can contribute

toward maintaining a hyperphosphorylated 4EBP1

protein and, thus, an activated eIF4E. Moreover, co-

localization of p-4EBP1 in the nucleus and cytoplasm

of high-grade carcinomas suggests a true oncogenic

role in these tumors, in which associated biochem-

ical and molecular factors need to be investigated.

eIF4E forms nuclear bodies and has an important

function in the translation of a subset of growth-pro-

moting mRNAs, including cyclin D1, Myc, ornithine

decarboxylase, fibroblastic growth factor, and vas-

cular endothelial growth factor.47 Future immunohis-

tochemical studies will be needed to validate eIF4E

expression in human tumors as a true prognostic

factor and to correlate it with 4EBP1.

In the setting of cell signaling and in an attempt

to identify molecules that clearly reflect the onco-

genic role of a specific pathway in ovarian cancer,

FIGURE 7. Percentages of ovarian neoplasms with phosphorylated 4Ebinding protein 1 (p-4EBP1) expression are illustrated relative to upstream

markers. Note that, among p-4EBP1-positive (p-4EBP1 þ) tumors, there wasa marked predominance of c-erb-B2 overexpression compared with epider-

mal growth factor receptor (EGFR)-positive (EGFR þ) tumors. Conversely,phosphorylated protein kinase B (p-AKT) and phosphorylated extracellular

signal-regulated kinase (p-ERK) appear to contribute to 4EBP1 activation. It

is noteworthy that there is a synergistic effect on 4EBP1 when both path-

ways are activated.

Phospo-4EBP1 in Epithelial Ovarian Neoplasms/Castellvi et al. 1809

we found that 4EBP1 may represent a hallmark of

cell signaling and may act as a pivotal factor in the

oncogenic pathways. We propose the concept of

4EBP1 as a choke point at which many oncogenic

signals converge and at which, after their phospho-

rylation, many transcription and phosphorylation

factors are activated. The findings that p-4EBP1

expression is associated with tumor progression and

an adverse prognosis and that it can be detected by

using immunohistochemistry allows us to suggest its

study as a new molecular marker in ovarian tumors,

regardless of the status of other possible oncogenic

alterations. Investigation into factors like 4EBP1 in

human tumors may lead to the identification of mo-

lecular markers in each oncogenic pathway that, in

turn, may complement the pathology report with a

small set of factors that are hallmarks of the func-

tional molecular signature of the tumor. Even more

important, 4EBP1 and eIF4E may be molecular tar-

gets for clinical treatment.

REFERENCES1. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell.

2000;100:57–70.

2. Futreal PA, Coin L, Marshall M, et al. A census of human

cancer genes. Nat Rev Cancer. 2004;4:177–183.

3. De CL, Marchionni L, Gariboldi M, et al. Gene expression

profiling of advanced ovarian cancer: characterization of a

molecular signature involving fibroblast growth factor 2.

Oncogene. 2004;23:8171–8183.

4. Shamji AF, Nghiem P, Schreiber SL. Integration of growth

factor and nutrient signaling: implications for cancer biol-

ogy. Mol Cell. 2003;12:271–280.

5. Hay N, Sonenberg N. Upstream and downstream of mTOR.

Genes Dev. 2004;18:1926–1945.

6. Holland EC, Sonenberg N, Pandolfi PP, Thomas G. Signal-

ing control of mRNA translation in cancer pathogenesis.

Oncogene. 200419;23:3138–3144.

7. Wang X, Beugnet A, Murakami M, Yamanaka S, Proud CG.

Distinct signaling events downstream of mTOR cooperate

to mediate the effects of amino acids and insulin on initia-

tion factor 4E-binding proteins. Mol Cell Biol. 2005;25:

2558–2572.

8. Constantinou C, Clemens MJ. Regulation of the phospho-

rylation and integrity of protein synthesis initiation factor

eIF4GI and the translational repressor 4E-BP1 by p53.

Oncogene. 2005;24:4839–4850.

9. Fingar DC, Salama S, Tsou C, Harlow E, Blenis J. Mamma-

lian cell size is controlled by mTOR and its downstream

targets S6K1 and 4EBP1/eIF4E. Genes Dev. 2002;16:1472–

1487.

10. Scheper GC, Morrice NA, Kleijn M, Proud CG. The mito-

gen-activated protein kinase signal-integrating kinase

Mnk2 is a eukaryotic initiation factor 4E kinase with high

levels of basal activity in mammalian cells. Mol Cell Biol.

2001;21:743–754.

11. Berchuck A, Rodriguez GC, Kamel A, et al. Epidermal

growth-factor receptor expression in normal ovarian epi-

thelium and ovarian-cancer. 1. Correlation of receptor

expression with prognostic factors in patients with ovar-

ian-cancer. Am J Obstet Gynecol. 1991;164:669–674.

12. Camilleri-Broet S, Hardy-Bessard AC, Le Tourneau A, et al.

HER-2 overexpression is an independent marker of poor

prognosis of advanced primary ovarian carcinoma: a multi-

center study of the GINECO group. Ann Oncol. 2004;15:

104–112.

13. Altomare DA, Wang HQ, Skele KL, et al. AKT and mTOR

phosphorylation is frequently detected in ovarian cancer

and can be targeted to disrupt ovarian tumor cell growth.

Oncogene. 2004;23:5853–5857.

14. Hsu CY, Bristow R, Cha MS, et al. Characterization of

active mitogen-activated protein kinase in ovarian serous

carcinomas. Clin Cancer Res. 2004;10:6432–6436.

15. Fabra A, Parada C, Vinyals A, et al. Intravascular injections

of a conditional replicative adenovirus (adl118) prevent

metastatic disease in human breast carcinoma xenografts.

Gene Ther. 2001;8:1627–1634.

16. Viniegra JG, Martinez N, Modirassari P, et al. Full activation

of PKB/Akt in response to insulin or ionizing radiation is

mediated through ATM. J Biol Chem. 2005;280:4029–4036.

17. Parada C, Losa JH, Guinea J, et al. Adenovirus E1a protein

enhances the cytotoxic effects of the herpes thymidine ki-

nase-ganciclovir system. Cancer Gene Ther. 2003;10:152–

160.

18. Ruggero D, Pandolfi PP. Does the ribosome translate can-

cer? Nat Rev Cancer. 2003;3:179–192.

19. Tee AR, Blenis J. mTOR, translational control and human

disease. Semin Cell Dev Biol. 2005;16:29–37.

20. Vogt PK. PI 3-kinase, mTOR, protein synthesis and cancer.

Trends Mol Med. 2001;7:482–484.

21. Hogdall EVS, Christensen L, Kjaer SK, et al. Distribution of

HER-2 overexpression in ovarian carcinoma tissue and its

prognostic value in patients with ovarian carcinoma—from

the Danish ‘‘MALOVA’’ ovarian cancer study. Cancer. 2003;

98:66–73.

22. Berchuck A, Rodriguez G, Kinney RB, et al. Overexpression of

Her-2 neu in endometrial cancer is associated with advanced

stage disease. Am J Obstet Gynecol. 1991;164:15–21.

23. Kupryjanczyk J, Madry R, Plisiecka-Halasa J, et al. TP53

status determines clinical significance of ERBB2 expression

in ovarian cancer. Br J Cancer. 2004;91:1916–1923.

24. Tanabe H, Nishii H, Sakata A, et al. Overexpression of

HER-2/neu is not a risk factor in ovarian clear cell adeno-

carcinoma. Gynecol Oncol. 2004;94:735–739.

25. Lee CH, Huntsman DG, Cheang MC, et al. Assessment of

Her-1, Her-2, and Her-3 expression and Her-2 amplifica-

tion in advanced stage ovarian carcinoma. Int J Gynecol

Pathol. 2005;24:147–152.

26. Elie C, Geay JF, Morcos M, et al. Lack of relationship

between EGFR-1 immunohistochemical expression and

prognosis in a multicentre clinical trial of 93 patients with

advanced primary ovarian epithelial cancer (GINECO

group). Br J Cancer. 2004;91:470–475.

27. Nielsen JS, Jakobsen E, Holund B, Bertelsen K, Jakobsen A.

Prognostic significance of p53, Her-2, and EGFR overex-

pression in borderline and epithelial ovarian cancer. Int J

Gynecol Cancer. 2004;14:1086–1096.

28. Yuan ZQ, Sun M, Feldman RI, et al. Frequent activation of

AKT2 and induction of apoptosis by inhibition of phospho-

inositide-3-OH kinase/Akt pathway in human ovarian can-

cer. Oncogene. 2000;19:2324–2330.

29. Dai DL, Martinka M, Li G. Prognostic significance of acti-

vated Akt expression in melanoma: a clinicopathologic

study of 292 cases. J Clin Oncol. 2005;23:1473–1482.

1810 CANCER October 15, 2006 / Volume 107 / Number 8

30. David O, Jett J, LeBeau H, et al. Phospho-Akt overexpres-

sion in non-small cell lung cancer confers significant

stage-independent survival disadvantage. Clin Cancer Res.

2004;10:6865–6871.

31. Kreisberg JI, Malik SN, Prihoda TJ, et al. Phosphorylation

of Akt (Ser473) is an excellent predictor of poor clinical

outcome in prostate cancer. Cancer Res. 2004;64:5232–5236.

32. Panigrahi AR, Pinder SE, Chan SY, Paish EC, Robertson

JFR, Ellis IO. The role of PTEN and its signalling pathways,

including AKT, in breast cancer; an assessment of relation-

ships with other prognostic factors and with outcome.

J Pathol. 2004;204:93–100.

33. Givant-Horwitz V, Davidson B, Lazarovici P, et al. Mitogen-

activated protein kinases (MAPK) as predictors of clinical

outcome in serous ovarian carcinoma in effusions. Gynecol

Oncol. 2003;91:160–172.

34. Singer G, Shih I, Truskinovsky A, Umudum H, Kurman RJ.

Mutational analysis of K-ras segregates ovarian serous car-

cinomas into two types: invasive MPSC (low-grade tumor)

and conventional serous carcinoma (high-grade tumor).

Int J Gynecol Pathol. 2003;22:37–41.

35. Singer G, Kurman RJ, Chang HW, Cho SK, Shih I. Diverse

tumorigenic pathways in ovarian serous carcinoma. Am J

Pathol. 2002;160:1223–1228.

36. Gemignani ML, Schlaerth AC, Bogomolniy F, et al. Role of

KRAS and BRAF gene mutations in mucinous ovarian car-

cinoma. Gynecol Oncol. 2003;90:378–381.

37. Shi YY, Hsu JH, Hu LP, Gera J, Lichtenstein A. Signal path-

ways involved in activation of p70(S6K) and phosphoryla-

tion of 4E-BP1 following exposure of multiple myeloma

tumor cells to interleukin-6. J Biol Chem. 2002;277:15712–

15720.

38. Gingras AC, Raught B, Sonenberg N. Regulation of transla-

tion initiation by FRAP/mTOR. Genes Dev. 2001;15:807–826.

39. Herbert TP, Tee AR, Proud CG. The extracellular signal-regu-

lated kinase pathway regulates the phosphorylation of 4E-

BP1 at multiple sites. J Biol Chem. 2002;277:11591–11596.

40. Molhoek KR, Brautigan DL, Slingluff CL. Synergistic inhibi-

tion of human melanoma proliferation by combination

treatment with B-Raf inhibitor BAY43-9006 and mTOR in-

hibitor rapamycin. J Transl Med. 2005;3:1–11.

41. Bjornsti MA, Houghton PJ. The TOR pathway: a target for

cancer therapy. Nat Rev Cancer. 2004;4:335–348.

42. Ruggero D, Montanaro L, Ma L, et al. The translation factor

eIF-4E promotes tumor formation and cooperates with

c-Myc in lymphomagenesis. Nat Med. 2004;10:484–486.

43. Montanaro L, Pandolfi PP. Initiation of mRNA translation in

oncogenesis—the role of eIF4E. Cell Cycle. 2004;3:1387–1389.

44. De Benedetti A, Graff JR. eIF-4E expression and its role in

malignancies andmetastases. Oncogene. 2004;23:3189–3199.

45. Avdulov S, Li S, Michalek V, et al. Activation of translation

complex eIF4F is essential for the genesis and mainte-

nance of the malignant phenotype in human mammary

epithelial cells. Cancer Cell. 2004;5:553–563.

46. Chen YH, Rodrik V, Foster DA. Alternative phospholipase

D/mTOR survival signal in human breast cancer cells.

Oncogene. 2005;24:672–679.

47. Richter JD, Sonenberg N. Regulation of cap-dependent

translation by eIF4E inhibitory proteins. Nature. 2005;433:

477–480.

Phospo-4EBP1 in Epithelial Ovarian Neoplasms/Castellvi et al. 1811