Expression of Epidermal Growth Factor and Androgen Receptors in Ovarian Cancer

GYNECOLOGIC ONCOLOGY ARTICLE NO.

66, 250–254 (1997)

GO974764

Expression of Epidermal Growth Factor and Androgen Receptors in Ovarian Cancer John V. Ilekis, Ph.D.,* Joseph P. Connor, M.D.,* Gail S. Prins, Ph.D.,† Karen Ferrer, M.D.,‡ Craig Niederberger, M.D.,† and Bert Scoccia, M.D.* *Department of Obstetrics and Gynecology, †Department of Urology, and ‡Department of Pathology, University of Illinois at Chicago, College of Medicine, 820 South Wood Street, Chicago, Illinois 60612 Received January 21, 1997

Ovarian cancer is the second most common malignancy of the female reproductive tract. Approximately 50% of ovarian cancers have elevated levels of epidermal growth factor receptor (EGFR). This overexpression is correlated with a poor prognosis for patient survival. Ovarian cancers also express a number of sex steroid receptors. The androgen receptor (AR) is the predominant sex steroid receptor and is expressed in over 80% of ovarian cancers. We investigated whether a relationship exists between EGFR and AR in ovarian cancer. Sixty serous cystadenocarcinomas were analyzed for their relative levels of EGFR and AR by Western blot analysis. Data were analyzed by Student’s t test and linear regression analysis for statistical significance. More than 98% of the tumors expressed detectable levels of EGFR, while 65% of the tumors expressed detectable levels of AR. The levels of EGFR (mean { SEM) were found to be significantly (P õ 0.01) higher in AR/ (516 { 15) than in AR0 (304 { 57) tumors. EGFR levels significantly correlated to AR levels (r Å 0.49, P õ 0.001). These results demonstrate an association between EGFR and AR levels in ovarian cancer. Whether this association represents a causal or a casual relationship remains to be determined. q 1997 Academic Press

INTRODUCTION

Epidermal growth factor receptor (EGFR) is a 170-kDa glycoprotein that is expressed on the cell surface of a variety of cells. It functions as a receptor for a family of ligands and is involved in regulating cellular growth and differentiation [1]. The receptor protein can be divided into three functional domains: an extracellular domain that contains the ligand-binding site, a transmembrane domain that anchors the receptor to the cell membrane, and a cytoplasmic domain that possesses tyrosine kinase activity. Particular attention has focused on the role of EGFR in the area of cancer research since the initial observation that a viral oncogene, v-erbB, is a truncated form of the receptor that encodes only the tyrosine kinase domain of the receptor [2]. Mounting evidence strongly suggests that aberrations in EGFR expression are involved in the etiology of ovarian cancer [3–5].

Gyn 4764

/

6d1f$$$$41

MATERIAL AND METHODS

Tumor specimens. A total of 60 serous cystadenocarcinomas was analyzed for EGFR and AR expression by Western blot analysis. These specimens included tissue obtained from 49 primary tumors and 11 from metastatic sites. Tissues

250

0090-8258/97 $25.00 Copyright q 1997 by Academic Press All rights of reproduction in any form reserved.

AID

Approximately 50% of ovarian cancers have elevated levels of EGFR, and this overexpression is correlated with a poor prognosis [3, 5]. Since the overexpression of EGFR is associated with the aggressiveness of the disease, this implies that increased levels of the receptor may play a role in tumor progression [3]. Little is known about the relationship between EGFR and steroid receptor expression in ovarian cancer. The androgen receptor (AR) is the predominant sex steroid receptor expressed in ovarian cancers. It has been reported that over 80% of tumors express AR, while only approximately 50% of tumors express either the estrogen or the progesterone receptor [6, 7]. In this regard, it is attractive to postulate that androgens through their interaction with AR may play an important role in the pathogenesis of the disease, since the incidence of ovarian cancer is highest after menopause, at which time androgens are the main steroids produced by the postmenopausal ovary [8]. In support of this concept, studies have shown that normal ovarian surface epithelium proliferation can be stimulated by androgens, and anti-androgens have been shown to inhibit the proliferation of several ovarian cancer cell lines [9, 10]. Moreover, androgens can increase EGFR levels in other biological systems. Androgen treatment has been shown to increase EGFR levels in the liver of mice and in a prostate cancer cell line [11, 12]. Since the biological effect of androgens is mediated via the androgen receptor, this may suggest that AR is also involved in regulating the level of EGFR in ovarian cancer. As an initial step in addressing this hypothesis, we investigated whether a relationship exists between EGFR and AR levels in ovarian cancer.

07-14-97 22:36:06

goa

AP: GYN

EGF AND ANDROGEN RECEPTORS IN OVARIAN CANCER

251

were obtained from the ovarian cancer tissue bank of the Cooperative Human Tissue Network. The tissue specimens were snap-frozen within 1 hr of collection, stored at 0707C, shipped to the investigators on dry ice, and stored at 0707C until processing. The median age of the patients was 58 years (range 26–85). Eighty-six percent of the tumors were stages III and IV and 90% of the tumors were grades 2 and 3. Histological classification was based on the World Health Organization criteria and tumor staging according to the International Federation of Gynecology and Obstetrics classification [13, 14]. Western blot analysis. Western blot analysis was performed as previously described [15]. Briefly, frozen tissue was pulverized on dry ice and immediately homogenized in a buffer containing a cocktail of protease inhibitors. The tissue homogenate was centrifuged at 1500g to remove unbroken cells and large particulates. An aliquot from the resulting supernatant fraction was used for protein determination and the remainder of the supernatant was stored at 0707C until Western blot analysis. A portion of the supernatant was diluted in a gel loading buffer to a final concentration of 1 mg/ml, and 25 ml (25 mg protein) was used for Western blot analysis. Proteins were fractionated by SDS– PAGE. A polyacrylamide concentration of 8% was used. Following SDS–PAGE, the proteins were electroblotted onto a nitrocellulose membrane. A mouse monoclonal antiEGFR antibody directed against the ligand-binding site epitope (Upstate Biotechnology, Inc., Lake Placid, NY [16]) and a rabbit polyclonal anti-AR antibody directed against the first 21 amino acids of AR (provided by co-investigator G.S.P. [17]) were used for immunodetection. The antibody concentrations used for immunodetection of EGFR and AR were 10 and 1 mg/ml, respectively. Blots were incubated overnight at 47C with the appropriate primary antibody. Following three washes of the blots in buffer, the blots were incubated for 2 hr with the appropriate secondary antibody complexed to horseradish peroxidase. (Anti-mouse IgG goat and anti-rabbit IgG goat secondary antibodies were used for EGFR and AR, respectively.) Following the incubation of the secondary antibody, the blots were washed and immunoreactivity was detected by chemiluminescence using a commercial kit (ECL; Amersham Corp., Arlington Heights, IL) according to the instructions provided by the supplier. To reduce methodological variation, all blots were batched together for immunodetection using each of the primary antibodies. Moreover, to correct for gel loading, following immunodetection for EGFR and AR, the blots were stripped of the primary and secondary antibodies, batched all together, and immunodetected for a-tubulin. Stripping of the blots was performed by incubating the membranes at 507C for 30 min in a solution consisting of 66 mM Tris–HCl (pH 6.8), 0.7% b-mercaptoethanol, and 0.2% SDS. The monoclonal anti-a-tubulin antibody was obtained from Sigma

AID

Gyn 4764

/

6d1f$$$$42

07-14-97 22:36:06

FIG. 1. Representative Western blots for EGFR and AR. Shown on the left are Western blots immunodetected with either the EGFR antibody (A) or the AR antibody (B). The bands corresponding to EGFR at 170 kDa and AR at 110 kDa are marked. Shown on the right are the levels of EGFR, AR, and a-tubulin in three serous cystadenocarcinomas (1–3) obtained by Western blot analysis. Positions of molecular weight markers in kilodaltons are shown.

Chemical (St. Louis, MO), and immunodetection was performed essentially as described above using a dilution of 1:4000 of the primary antibody. Immunoreactivities, corresponding to bands at 170 kDa for EGFR, 110 kDa for AR, and 50 kDa for a-tubulin, were quantified by densitometry relative to a standard reference preparation (25 mg protein) contained on each blot. The standard reference preparation consisted of a serous cystadenocarcinoma specimen expressing high levels of both EGFR and AR. Exposure times were optimized for each of the proteins (EGFR, AR, and a-tubulin) so that the densitometric quantification was within the linear range. A threshold sensitivity of 60 densitometry units was chosen as a negative cutoff for both EGFR and AR. Below this threshold value, bands were either barely discernable or undetectable under standard exposure conditions. Results are expressed as relative densitometry units normalized to the expression of atubulin. Statistical analysis. Statistical significance by Student’s t test and linear regression analysis was determined using the statistical computer program Kwikstat (Texasoft, TX). A P value of £0.05 was considered significant. RESULTS

Representative Western blots for EGFR and AR are shown in Fig. 1 (left). Also shown in Fig. 1 (right) are the bands obtained by Western blot corresponding to EGFR, AR, and a-tubulin from three ovarian cancers. As documented in the literature, EGFR is detected at 170 kDa and AR is detected

goa

AP: GYN

252

ILEKIS ET AL.

at 110 kDa [1, 18]. A number of other minor bands, however, are also detected upon overexposure. Except for the lower band detected in the EGFR blot at 80 kDa and the AR blot at 50 kDa, these bands represent high-abundance proteins as determined by amido black staining of the Western blots. Their immunoreactivity is nonspecific since substituting the appropriate control IgG for each of the primary antibodies results in a similar intensity of immunoreactivity, while the EGFR and AR bands are no longer detected (data not shown). The 80-kDa band detected in the EGFR blot is an EGFR-like protein first reported in placenta [15]. The 50kDa band detected in the AR blot is a proteolytic cleavage product of AR [17]. Detectable levels of EGFR were found in 48 of the 49 primary and in all of the 11 metastatic specimens. The percentage (ú98%) of EGFR/ specimens obtained is consistent with the expected range of 70–100%, as reported in the literature using radioligand-binding assays [3, 19]. Although not statistically significant, a trend for increased AR/ tumors was noted in specimens obtained from primary tumors compared to metastatic sites. Detectable levels of AR were found in 69% of the primary tumors and 45% of the metastatic specimens. The percentage (65%) of AR/ specimens obtained is lower than the percentage (80–90%) reported in the literature by radioligand-binding assays [6, 7], and is probably due to the somewhat lower sensitivity of Western blot analysis compared to the radioligand-binding method. No attempt was made to correlate the stage or grade of the tumor with the level of either EGFR or AR because of the lack of sufficient number of specimens available in earlier stages of the disease. To determine if there was an association between EGFR and AR expression, the data were grouped according to AR status and the level of EGFR expression (Fig. 2). A higher number of tumors (12 of 40) in the AR/ group contained

In this study only serous cystadenocarcinomas were used for analysis. The rationale for using only serous tumors was

FIG. 2. EGFR levels in individual tumors. EGFR levels are grouped according to AR status.

FIG. 4. Relationship between EGFR and AR levels in ovarian cancer. Linear regression analysis of 60 tumors indicates a significant correlation (P õ 0.001, r Å 0.49) between receptor levels.

AID

Gyn 4764

/

6d1f$$$$42

07-14-97 22:36:06

FIG. 3. The level of EGFR in AR0 and AR/ tumor groups. The mean levels are significantly different from one another (P õ 0.01).

elevated values of EGFR greater than 600 units compared to the AR0 group (1 of 20). As shown in Fig. 3, the level (mean { SEM) of EGFR was found to be significantly higher (P õ 0.01) in AR/ (516 { 15) compared to AR0 tumors (304 { 57). To determine whether EGFR levels directly correlated to AR levels, linear regression analysis was performed. A significant correlation (r Å 0.49, P õ 0.001) was found comparing the levels of EGFR to AR within individual specimens (Fig. 4). Thus, higher EGFR levels were associated with AR/ tumors and the level of EGFR correlated with the level of AR. DISCUSSION

goa

AP: GYN

EGF AND ANDROGEN RECEPTORS IN OVARIAN CANCER

that these represent the predominant form of ovarian cancer and to exclude possible differences in the expression of EGFR and AR due to the histological type of the ovarian cancer. Moreover, most specimens were obtained from the primary tumor rather than from metastatic sites, since sampling from the primary site would be potentially more useful for patient prognosis. Based on a limited sample size in which both primary and metastatic tumors were available from the same individual, an analysis of EGFR and AR levels in tumor tissue obtained from the primary tumor or from a metastatic site (generally omentum) revealed them not to be significantly different from one another (data not shown), although a trend for increased EGFR levels was noted in tissue obtained from the metastatic site. Our result is consistent with a previous study that also reported a similar trend [3]. Western blot analysis was chosen to quantify the expression of both EGFR and AR. Western blot analysis in conjunction with chemiluminescence immunodetection provides a more sensitive and quantitative measurement than immunohistochemical methods. In addition, each specimen can be assessed for abnormal gross structural changes in either EGFR or AR, which is not possible with either immunohistochemical or radioligand methods. Although this method is not as sensitive as radioligand-binding assays, it is capable of attaining detection limits approaching the low femtomole per milligram protein range. This level of detection is near the negative cutoff values established for AR and EGFR expression in ovarian cancer by several investigators using radioligand-binding assays [6, 7]. In several specimens, gross abnormalities based on aberrant molecular weights were detected for EGFR and AR. One specimen expressed a lower molecular size of EGFR at 120 kDa in addition to the normal receptor at 170 kDa. Structural abnormalities in EGFR have been reported in other types of cancers and are quite common in gliomas [20]. Based on our study, however, gross structural changes in EGFR are not a common occurrence in serous cystadenocarcinoma. Three specimens expressed, in addition to the AR band at 110 kDa, a slightly lower migrating band at 100 kDa. Only two specimens expressed the 100-kDa form. The significance of the 100-kDa AR form is not clear; a similar molecular weight species has also been reported in the fetal adrenal [18]. Unfortunately, the number of tumors that expressed the lower molecular weight form of AR was too low to evaluate its influence on the level of EGFR. AR expression has been previously shown to be associated with lower stage and grade ovarian tumors [21]. Since it is well documented that both the stage and the grade of the tumor are important prognostic factors in assessing patient survival in ovarian cancer, it is not unexpected that AR/ status is also associated with better patient survival [22]. In contrast, the level of EGFR has been shown to be an important prognostic factor independent of tumor stage or grade

AID

Gyn 4764

/

6d1f$$$$42

07-14-97 22:36:06

253

[3, 4]. The data presented in this study indicate that in serous cystadenocarcinoma, higher EGFR levels are associated with AR-positive tumors. In view of this fact, it may appear paradoxical that higher EGFR levels are associated with ARpositive tumors. Our data may suggest that a subset of ARpositive patients may be identified, based on their elevated EGFR levels, who may have a poorer prognosis than ARpositive patients with lower levels of EGFR. Such a finding would be analogous to the prognostic value of the estrogen receptor and EGFR status found in breast cancer [23]. Alternatively, this finding may be pertinent only to serous cystadenocarcinomas, since the previously mentioned studies included a significant proportion of other histological types of ovarian cancers. Follow-up of the patients in this study to determine whether both EGFR and AR status would be a more valuable prognostic index than EGFR or AR status alone in predicting survival is ongoing. A role of AR in ovarian cancer growth is particularly attractive because of its preponderance relative to other sex steroid receptors and because of the increased androgen and reduced estrogen and progesterone production from the ovaries after menopause [8]. This notion is supported by the ability of androgens to modulate EGFR levels in another AR-expressing carcinoma such as prostate. Exposure of the LNCaP prostate cancer cell line to androgen results in an increase in EGFR levels with a subsequent increase in their proliferation following the addition of EGF [12]. Interestingly, no enhanced mitogenic response could be observed with androgen alone unless EGF was added to the medium. The mitogenic effect of EGFR ligands on normal ovarian surface epithelium and ovarian cancer cells has been well documented [4, 19]. The role of androgens in the proliferation of normal surface epithelium and ovarian cancer cell lines, however, is less clear and the role of androgens in modulating EGFR is unknown. Reports on the effect of androgens on the proliferation of normal surface epithelium are conflicting. In an early study, androgen treatment was shown to stimulate the growth of the ovarian surface epithelium, but no effect of androgen could be demonstrated in a more recent study [9, 24]. In addition, no direct effect of androgens on the stimulation of ovarian cancer growth has been documented, although anti-androgen treatment has been reported to inhibit ovarian cancer growth in cell culture [10]. Conceivably, the ability of androgens to stimulate ovarian cancer cell proliferation may be analogous to that of the LNCaP cell line, requiring not only androgen but also the presence of EGF. Further studies are under way in our laboratory to assess whether androgens can modulate EGFR levels in ovarian cancer cell lines and whether this up-regulation is capable of an increased proliferative response to EGF. In conclusion, our study is the first report to document an association between the levels of EGFR and AR in ovarian cancer. It is intriguing to speculate whether the levels of EGFR in a subset of AR/ tumors are dependent on andro-

goa

AP: GYN

254

ILEKIS ET AL.

gens. If so, this information may provide a rationale for the treatment of these tumors with anti-androgen therapy. In this regard, a recent clinical study reported that the use of the anti-androgen flutamide in the treatment of end-stage ovarian cancer was ineffective [25]. It should be noted, however, that this study was undertaken in a limited number of patients without regard for their AR status or histological type of the ovarian cancer. Clearly, more information is needed to determine the relationship and biological significance between EGFR and AR expression in ovarian cancer. ACKNOWLEDGMENTS The authors thank the technical staff (Jillin Shi, Anne Baldeshwiler, and Lynn Birch) and summer research students (Eugenia Hahn and Anita Visweswaran) whose efforts made this work possible.

REFERENCES 1. Khazaie K, Schirrmacher V, Lichtner RB: EGF receptor in neoplasia and metastasis. Cancer Metastasis Rev 12:255–274, 1993 2. Downward J, Yarden Y, Mayes E, et al.: Close similarity of epidermal growth factor receptor and v-erb B oncogene protein sequences. Nature 307:521–527, 1984 3. Scambia J, Benedetti-Panici P, Battaglia F, et al.: Significance of epidermal growth factor receptor in advanced ovarian cancer. J Clin Oncol 10:529–535, 1992 4. Morishige K, Kurachi H, Amemiya K, et al.: Evidence for the involvement of transforming growth factor alpha and epidermal growth factor receptor autocrine growth mechanism in primary human ovarian cancers in vitro. Cancer Res. 51:5322–5328, 1991. 5. Berchuck A, Rodriguez GC, Kamel A, et al.: Epidermal growth factor receptor expression in normal ovarian epithelium and ovarian cancer. I. Correlation of receptor expression with prognostic factors in patients with ovarian cancer. Am J Obstet Gynecol 164:669–674, 1991 6. Kuhnel R, De Graaff J, Rao B R, Stolk JG: Androgen receptor predominance in human ovarian carcinoma. J Steroid Biochem 26:393–397, 1987 7. Slotman BJ, Rao BR: Ovarian cancer: Etiology, diagnosis, prognosis, surgery, radiotherapy, chemotherapy and endocrine therapy. Anticancer Res 8:417–434, 1988 8. Adashi EY: The climacteric ovary as a functional gonadotropin-driven androgen-producing gland. Fertil Steril 62:20–27, 1994 9. Hamilton TC, Davies P, Criffiths K: Steroid-hormone receptor status of the normal and neoplastic ovarian surface germinal epithelium, in (Greenwald GS, Terranova PF (eds.): Factors Regulating Ovarian Function. New York, Raven Press, 1983, pp 81–85

AID

Gyn 4764

/

6d1f$$$$43

07-14-97 22:36:06

10. Slotman BJ, Rao BR: Response to inhibition of androgen action of human ovarian cancer cells in vitro. Cancer Lett 45:213–220, 1989 11. Scoccia B, Kovar P, Benveniste R: Gonadal and adrenal effects on hepatic epidermal growth factor receptor expression in a murine model. Endocrinology 129:3240–3246, 1991 12. Mulder E, Van Loon D, De Boer W, et al.: Mechanism of androgen action: Recent observations on the domain structure of androgen receptors and the induction of EGF-receptors by androgens in prostate tumor cells. J Steroid Biochem 32:151–156, 1989 13. Scrov SF, Scully RE: Histological typing of ovarian tumors, in International Histological Classification of Tumors. Geneva, WHO, 1973, No. 9 14. International Federation of Gynecology and Obstetrics: Classification and staging of malignant tumors in the female pelvis. J Int Fed Gynecol 3:204, 1965 15. Ilekis JV, Stark BC, Scoccia B: Possible role of variant RNA transcripts in the regulation of epidermal growth factor receptor expression in human placenta. Mol Reprod Dev 41:149–154, 1995 16. Wu DG, Wang LH, Chi Y, Sato GH, Sato JD: Human epidermal growth factor receptor residue covalently cross-linked to epidermal growth factor. Proc Natl Acad Sci USA 87:3151–3155, 1990 17. Prins GS, Birch L, Greene GL: Androgen receptor localization in different cell types of the adult rat prostate. Endocrinology 129:3187–3199, 1991 18. Wilson CM, McPhaul MJ: A and B forms of the androgen receptor are expressed in a variety of human tissues. Mol Cell Endocrinol 120:51–57, 1996 19. Rodriguez GC, Berchuck A, Whitaker RS, Schlossman D, ClarkePearson D, Bast R Jr.: Epidermal growth factor receptor expression in normal ovarian epithelium and ovarian cancer. II. Relationship between receptor expression and response to epidermal growth factor. Am J Obstet Gynecol 164:745–750, 1991 20. Sugawa N, Ekstrand AJ, James CD, Collins VP: Identical splicing of aberrant epidermal growth factor receptor transcripts from amplified rearranged genes in human glioblastomas. Proc Natl Acad Sci USA 87:8602–8606, 1990 21. Slotman BJ, Baak JPA, Rao BR: Correlation between nuclear DNA content and steroid receptor status in ovarian cancer. Eur J Obstet Gynecol Reprod Biol 38:221–227, 1990 22. Slotman BJ, Nauta JJP, Rao BR: Survival of patients with ovarian cancer—Apart from stage and grade, tumor progesterone receptor content is a prognostic indicator. Cancer 66:740–744, 1990 23. Sainsbury JRC, Frandon JR, Sherbert GV, Harris AL: Epidermalgrowth-factor receptors and estrogen receptors in human breast cancer. Lancet i:364–366, 1985 24. Karlan BY, Jones J, Greenwald M, Lagasse LD: Steroid hormone effects on the proliferation of human ovarian surface epithelium in vitro. Am J Obstet Gynecol 173:97–104, 1995 25. Tumolo S, Rao R, van der Burg ME, Guastalla JP, Renard J, Vermorken JB: Phase II trial of flutamide in advanced ovarian cancer: An EORTC Gynaecological Cancer Cooperative Group study. Eur J Cancer 30A:911–914, 1994

goa

AP: GYN