Ivan Bieche^1,2,3, Beatrice Parfait¹, Sengül Tozlu², Rosette Lidereau2 and Michel Vidaud¹

¹Laboratoire de Génétique Moléculaire-UPRES JE 2195, Faculté des Sciences Pharmaceutiques et Biologiques, Universite Rene Descartes-Paris V, Paris and ²Laboratoire d’Oncogenetique-INSERM E0017, Centre Rene Huguenin, St-Cloud, France. ³To whom correspondence should be addressed at: Laboratoire de Genetique

Abbreviations: AR, androgen receptor; ER, estrogen receptor; PR, progesterone receptor; MPA, medroxyprogesterone acetate; SBR, Scarff-BloomRichardson.

Moleculaire-UPRES JE 2195, Faculte des Sciences Pharmaceutiques et Biologiques, Universite Rene Descartes-Paris V, 4 Avenue de l’Observatoire, F-75006 Paris, France; E-mail: ivan.bieche@pharmacie.univ-paris5.fr.

Summary

Little is known of the function and clinical significance of the androgen receptor (AR) in human breast cancer. Paradoxically, synthetic progestins, such as medroxyprogesterone acetate, are used for second line hormone therapy of breast cancer following tamoxifen failure. A sensitive and accurate assay for AR expression in breast tumors is thus required. Here we have developed and validated a real-time RT-PCR assay to quantify AR gene expression at the mRNA level in a series of 131 patients with unilateral invasive primary breast tumors. AR expression varied widely in tumor tissues (by at least 3 orders of magnitude), being underexpressed in 24/131 (18.3%) and overexpressed in 45/131 (34.4%) relative to normal breast tissues. We observed links (or trends) between AR status and age, menopausal status, Scarff-Bloom-Richardson histopathological grade, lymph node status and estrogen receptor a and progesterone receptor status. High AR mRNA levels were negatively linked to MYC gene overexpression (P = 8 X10^-6), confirming previous in vitro studies. Our results also suggest a role of the ARA70 gene (which encodes a major AR co-activator) in the AR pathway dysregulation observed in breast cancer. This simple, rapid and semi-automated method will be useful for screening cancer patients for altered AR expression and for predicting the response to androgen therapy in AR-related cancer patients.

---

Introduction

The role of estrogen receptor (ER) α and the progesterone receptor (PR) in human breast cancer is well established. Considerably less is known about the functional role and clinical significance of androgen receptor (AR) expression in this setting. Biochemical and immunohistochemical studies show that AR-positive tumors are more frequent (70-90%) than ERα-positive and PR-positive tumors (60-80 and 5070%, respectively) (1-4). Although ERα, PR and AR are frequently co-expressed in breast tumors, ~10% of AR-positive tumors and, perhaps more importantly, 25% of AR-positive tumor metastases can be negative for ERα and PR (1,5).

Androgens have been shown to regulate the proliferation of AR-positive breast cancer cell lines in culture (6). Synthetic progestins, such as medroxyprogesterone acetate (MPA), are used as second line hormone therapy for breast cancer following tamoxifen failure (7). Birrell et al. (8) suggested that the antiproliferative effect of MPA in advanced cancer is mediated by AR. In vitro studies confirmed that MPA inhibits the proliferation of ERα-negative and PR-negative cell lines via AR (9).

Taken together, these findings suggest that AR determination may give additional predictive information on the response to endocrine treatments in breast cancer. AR expression has mainly been studied by means of a cytosol steroid-binding assay and immunohistochemistry. Although the former measures the status and functionality of the protein, it has several methodological shortcomings (1) and is time consuming. Furthermore, it requires the use of radioactive reagents and large amounts of tumor tissue, so that it is rarely used routinely in clinical laboratories. Immunohistochemical methods suffer from a lack of inter-laboratory standardization and cannot quantify the full range of alterations. However, this method also gives information concerning the status of the protein, but above all measures alterations on an individual cell basis.

We quantified AR mRNA expression in a series of 131 patients with unilateral invasive primary breast tumors, using real-time quantitative RT-PCR assay. This recent method of nucleic acid quantification in homogeneous solutions has the potential to become a standard in terms of its performance, accuracy, sensitivity, wide dynamic range, high throughput capacity and inter-laboratory agreement, and also yields statistical confidence values (10).

We examined the relationship between AR expression status and classical clinical and pathological parameters, including patient outcome. AR mRNA levels were interpreted according to ERα, ERft and PR transcript levels measured using the same methodology and on the same homogeneous total RNA solutions.

We also sought relationships between AR expression and that of genes known to be altered in breast cancer (RBI, CCND1, MYC and ERBB2), as well as several major genes involved in different steps of the AR pathway dysregulation observed in prostate cancer, i.e. the ARA70 gene (which codes for a major AR co-activator) (11), two well-known AR-responsive genes in prostate cancer (PAP, coding for prostatic acid phosphatase, and PSA, coding for prostatespecific antigen) (12) and DNMT1, a DNA methyltransferase gene that is altered in tumors (13), because loss of AR expression is associated with methylation of the AR promoter in prostate cancer cells (14).

Materials and methods

Patients and samples

We analyzed tissue from primary breast tumors excised from 131 women treated at the Centre Rene Huguenin from 1977 to 1989. Tumor tissue samples of the 131 patients were collected in accordance with French regulations.

The samples were examined histologically for the presence of tumor cells. A tumor sample was considered suitable for this study if the proportion of tumor cells was >60%. Immediately following surgery the tumor samples were stored in liquid nitrogen until RNA extraction.

The patients (mean age 58.2 years, range 34-91) met the following criteria: primary unilateral non-metastatic breast carcinoma on which complete clinical, histological and biological data were available; no radiotherapy or chemotherapy before surgery. The main prognostic factors are presented in Table I. The median follow-up was 8.1 years (range 1.0-15.9). Forty-seven patients relapsed (the distribution of first relapse events was as follows: 13 local and/ or regional recurrences, 30 metastases and 4 both).

Table I. Characteristics of the 131 patients and relation to disease-free survival

Specimens of adjacent normal breast tissue from nine of the breast cancer patients and normal breast tissue from three women undergoing cosmetic breast surgery were used as sources of normal RNA.

Real time RT-PCR

Theoretical basis. Quantitative values were obtained from the threshold cycle number at which the increase in the signal associated with exponential growth of PCR products begins to be detected using PE Biosystems analysis software, according to the manufacturer’s manuals.

The precise amount of total RNA added to each reaction mix (based on optical density) and its quality (i.e. lack of extensive degradation) are both difficult to assess. We therefore also quantified transcripts of the RPLP0 gene (also known as 36B4) encoding human acidic ribosomal phosphoprotein P0 as an endogenous RNA control and each sample was normalized on the basis of its RPLP0 content. The relative AR gene expression level was also normalized to a calibrator, or 1X sample, consisting of a pool of normal breast tissue specimens. Final results, expressed as «-fold differences in AR gene expression relative to the RPLP0 gene and normal breast tissues (the calibrator), termed n_AR, were determined in exponent as follows:

n_AR = 2(ΔCt_sample — ΔCt_calibrator)

where ΔCt values of the sample and calibrator are determined by subtracting the average Ct value of the AR gene from the average Ct value of the RPLP0 gene.

Primers and PCR consumables

Primers for the RPLP0 and target genes were chosen with the assistance of the computer programs Oligo 4.0 (National Biosciences, Plymouth, MN) and Primer Express (Perkin-Elmer Applied Biosystems, Foster City, CA). We conducted BLASTN searches against dbEST, htgs and nr (the non-redundant set of the GenBank, EMBL and DDBJ database sequences) to confirm the total gene specificity of the nucleotide sequences chosen for the primers and the absence of DNA polymorphisms. The nucleotide sequences of the primers are shown in Table II. To avoid amplification of contaminating genomic DNA, one of the two primers was placed in a different exon.

Table II. Oligonucleotide primer sequences used

RNA extraction

Total RNA was extracted from breast specimens using the acid phenol/ guanidium method. The quality of the RNA samples was determined by electrophoresis through agarose gels and staining with ethidium bromide and the 18S and 28S RNA bands were visualized under UV light.

cDNA Synthesis

RNA was reverse transcribed in a final volume of 20 μl containing 1X RT buffer (500 mM each dNTP, 3 mM MgCl₂, 75 mM KCl, 50 mM Tris-HCl, pH 8.3), 10 U RNasin RNase inhibitor (Promega, Madison, WI), 10 mM dithiothreitol, 50 U Superscript II RNase H^- reverse transcriptase (Gibco BRL, Gaithersburg, MD), 1.5 mM random hexamers (Pharmacia, Uppsala, Sweden) and 1 pg total RNA. The samples were incubated at 20°C for 10 min and 42°C for 30 min and reverse transcriptase was inactivated by heating at 99°C for 5 min and cooling to 5°C for 5 min.

PCR amplification

All PCR reactions were performed using an ABI Prism 7700 Sequence Detection System (Perkin-Elmer Applied Biosystems). PCR was performed using the SYBR Green PCR Core Reagents kit (Perkin-Elmer Applied Biosystems). The thermal cycling conditions comprised an initial denaturation step at 95°C for 10 min and 50 cycles at 95°C for 15 s and 65°C for 1 min. Experiments were performed in duplicate for each data point.

Statistical analysis

Relapse-free survival was determined as the interval between diagnosis and detection of the first relapses (local and/or regional recurrences and/or metastases).

Clinical, histological and biological parameters were compared using the x² test, with Yates’ correction for adjustment of the continuity of the x² distribution where appropriate. Differences between the two populations were judged significant at confidence levels >95% (P < 0.05). Survival distributions were estimated by the Kaplan-Meier method (15) and the significance of differences between survival rates was ascertained using the log rank test (16).

Results

AR mRNA expression in normal breast tissues

To determine the cut-off point for altered AR expression in breast cancer tissue, the n_AR value, calculated as described in Materials and methods, was determined for 12 normal breast RNA samples. As this value consistently fell between 0.70 and 1.61 (1.15 ± 0.27, mean ± SD), values of 2 (mean + 3 SD) or more were considered to represent overexpression and values of 0.35 (mean 3 SD) or less were considered to represent underexpression of AR mRNA.

AR mRNA expression in tumor breast tissues

The 131 breast tumor RNA samples tested had a wide range of n_AR values (0.008-10.3, i.e. at least 3 orders of magnitude). Compared with normal breast tissues, 69 (52.7%) tumors showed altered AR mRNA expression. Twenty-four tumors (18.3%) showed AR mRNA underexpression (n_AR 0.008-0.31) and 45 (34.4%) showed overexpression (n_AR 2.04-10.3). AR mRNA levels were similar to those observed in prostate tumor tissues (data not shown).

Correlation between AR mRNA levels and clinical and pathological parameters

We sought links between AR mRNA expression status and standard clinical and pathological factors in breast cancer (Table III). Links (or trends) were found between AR gene status and age (P = 0.063), menopausal status (P = 0.070), Scarff-Bloom-Richardson (SBR) histopathological grade status (P = 0.00083) and lymph node status (P = 0.049). Patients with tumors overexpressing and/or underexpressing AR did not relapse more frequently (Table III) and did not have significantly shorter relapse-free survival after surgery (log rank test) compared with patients with tumors normally expressing AR.

Table III. Relationship between AR mRNA level and the standard clinical and pathological factors

Relationship between AR mRNA levels and ERα, PR and ERβ expression status

Patients were subdivided into three equal groups with low (n = 43), intermediate (n = 44) and high (n = 44) ERα, PR and ERβ mRNA levels. As shown in Table IV, we found a strong positive association between AR gene status and ERα (P < 10^-7) and PR gene (P = 3X10^-7) status and a negative association with ERβ gene status (P = 0.0026). Seven (5.3%) ‘ERα-negative’ (low ERα mRNA expressed) tumors overexpressed AR and one (0.8%) AR-underexpressing tumor had a high ERα mRNA level. The AR and ERα mRNA status of these tumors was confirmed by repeat RT-PCR.

Table IV. Relationship between AR mRNA levels and ERα, PR and ERβ mRNA levels

Relationship between AR mRNA levels and RB1, CCND1, MYC and ERBB2 expression status

The 131 tumors studied for AR expression had previously been tested for RB1, CCND1, MYC and ERBB2 mRNA expression (17-19; manuscript in preparation). We found a significant positive link between AR underexpression and RB1 underexpression (P = 0.0046) and a significant negative link between AR overexpression and MYC overexpression (P = 8X10^-6), but no link between AR and CCND1 or ERBB2 mRNA status (Table V).

Table V. Relationship between AR mRNA levels and RB1, CCND1, MYC and ERBB2 mRNA levels

Relationship between AR mRNA levels and ARA70, DNMT1, PAP and PSA expression status

ARA70, DNMT1, PAP and PSA mRNA levels were analyzed in 10 AR-underexpressing and 10 AR-overexpressing breast tumors (Table VI). For the ARA70 and DNMT1 genes patients were subdivided into two equal groups of tumors with low (n = 10) and high (n = 10) mRNA levels. For the PAP and PSA genes, which were very weakly expressed, patients were subdivided into tumors with detectable and no detectable mRNA molecules. We found a significant positive association between AR and ARA70 expression (P = 0.0073), but no link between AR status and DNMT1, PAP or PSA mRNA levels. Moreover, the highest levels of PAP and PSA gene expression in this breast tumor series were far lower that those observed in prostate tumor tissues (data not shown).

Table VI. Relationship between AR mRNA levels and ARA70, DNMT1, PAP and PSA mRNA levels

Discussion

In this study we applied a recent RT-PCR method (10) to the quantification of AR gene expression. We tested 12 normal breast tissue and 131 unilateral invasive primary breast tumor RNAs. AR mRNA was detected in all breast tumor samples and also in all normal breast tissues. These results confirm the higher sensitivity of RT-PCR compared with steroid-binding and immunohistochemical assays. Another major advantage of real-time RT-PCR is the large linear dynamic range, suited to analyzing genes, such as AR, associated with wide ranges of mRNA expression in tumor tissues (0.008-10.3 times normal in this series). It is noteworthy that this range (~3 orders of magnitude) is smaller than those of ERα and PR (at least 4 orders of magnitude; data not shown), suggesting that AR levels are more tightly controlled than those of other sex hormone receptors.

We observed both underexpression (18% of samples) and overexpression (34%) of AR mRNA in this breast tumor series. The 24 AR-underexpressing tumors had very low levels of AR mRNA (mean of the n_AR values 0.07 ± 0.06) compared with the 62 tumors with normal AR expression (1.12 ± 0.49), suggesting a bimodal distribution of AR expression and allowing us to use an unequivocal cut-off (n_AR = 0.35) to distinguish the two tumor groups. As a strong correlation has been reported between AR mRNA copy number and AR protein abundance (20), the 24 AR-underexpressing tumors would correspond to ‘AR-negative’ tumors in steroid-binding and immunohistochemical assays.

Overall, the results for AR-negative tumors in this study agree with those reported in the literature. The frequency (18%) of AR-negative tumors in our breast tumor series is similar to that obtained with steroid-binding and immunohistochemical assays (1,4). AR, PR and ERα expression were strongly intercorrelated, but we observed one AR-negative tumor that contained ERα (0.8% of our tumor series) and several AR-overexpressing tumours that did not contain ERα (5.3%), in keeping with others reports (1,4,21). The negative association between AR and ERβ was probably due to the negative link between ERβ and ERα (data not shown). In addition to PR and ERα negativity, we found that AR gene underexpression was associated with SBR histopathological grade III but not with a poor prognosis, confirming that AR is more a marker of tumor aggressiveness (poorly differentiated tumors) than a predictor of patient outcome in breast cancer. As expected, we also found a correlation between AR underexpression and RB1 underexpression; indeed, in the same tumor series RB1 underexpression was also associated with poorly differentiated tumors (correlation with SBR histopathological grade III and PR and ERα negativity) (17).

The 107 AR-positive tumors fell into two groups: those with normal AR expression (n = 62) and those with AR overexpression (n = 45). The amount of AR mRNA increased in tumors from both elderly and post-menopausal patients (Table III), in agreement with Lea et al. (1). This may be due to AR up-regulation to compensate for the decline in circulating sex steroids.

We observed a strong negative link between AR overexpression and MYC gene overexpression. No such link was observed between ERα (or ERP) and MYC expression (data not shown). This study confirms the direct regulation of AR transcription by the c-myc transcriptor factor via a myc consensus site in an AR exonic region (22) and the downregulation of MYC mRNA associated with androgen-induced suppression of the transformed phenotype in the human prostate carcinoma cell line LNCaP (23). No correlation was observed between AR overexpression and altered expression of the RBI, CCND1 and ERBB2 genes. This is in disagreement with previous data (24-26), indicating that retinoblastoma protein, cyclin D1 and c-erbB2 control the transcriptional activity of AR, the latter regulating its own trancription.

We observed a positive correlation between AR and ARA70 expression, confirming the specificity of ARA70 in controlling transcription activity of AR in breast cancer, as in prostate cancer (27). We did not observe a correlation between AR and DNMT1 expression, suggesting that the loss of AR expression in breast tumors is not due to up-regulation of DMNT1 via hypermethylation of the AR promoter CpG island. It is noteworthy that this finding does not exclude AR promoter methylation as a possible cause of AR down-regulation, due to modified expression of a DNA methyltransferase gene other than DMNT1. Finally, we found no link between AR status and PAP or PSA status. These two genes showed far lower expression levels than in prostate tissue, confirming the high specificity of PAP and PSA expression for prostate tissue.

In conclusion, our data suggest the involvement of several AR-mediated pathways in the regulation of breast tumor growth. Further characterization of these pathways may lead to new androgenic therapies for breast cancer.

Accurate determination of AR status, combined with ERα status, could help to select optimal endocrine therapies for breast cancer. The rapid, cost-effective, highly sensitive high throughput RT-PCR assay used here to determine AR status should be useful as a routine tool in AR-based clinical applications in breast cancer and other AR-related cancers.

Acknowledgements

We thank the Centre René Huguenin staff for assistance in specimen collection and patient care. This work was supported by the Association pour la Recherche sur le Cancer and the Ministere de l’Enseignement Superieur et de la Recherche.

---

References

Lea,O.A., Kvinnsland,S. and Thorsen,T. (1989) Improved measurement of androgen receptors in human breast cancer. Cancer Res., 49, 7162-7167. 

Kuenen-Boumeester,V., Van Der Kwast,H., Van Putten,W.L.J., Claassen,C., Van Ooijen,B. and Henzen-Logmans,S.C. (1992) Immunohistochemical determination of androgen receptors in relation to oestrogen and progesterone receptors in female breast cancer. Int. J. Cancer, 52, 581-584. 

Kimura,N., Mizokami,A., Oonuma,T., Sasano,H. and Nagura,H. (1993) Immunocytochemical localization of androgen receptor with polyclonal antibody in paraffin-embedded human tissues. J. Histochem. Cytochem., 41, 671-678. 

Soreide,J.A., Lea,O.A., Varhaug,J.E., Skarstein,A. and Kvinnsland,S. (1992) Androgen receptors in operable breast cancer: relation to other steroid hormone receptors, correlations to prognostic factors and predictive value for effect of adjuvant tamoxifen treatment. Eur. J. Surg. Oncol., 18, 112-118. 

Kuenen-Boumeester,V., Van Der Kwast,T.H., Claassen,C.C., Look,M.P., Liem,G.S., Klijn,J.G.M. and Henzen-Logmans,S.C. (1996) The clinical significance of androgen receptors in breast cancer and their relation to histological and cell biological parameters. Eur. J. Cancer, 32A, 1560-1565. 

Hackenberg,R. and Schulz,K.-D. (1996) Androgen receptor mediated growth control of breast cancer and endometrial cancer moduled by antiandrogenand androgen-like steroids. J. Steroid Biochem. Mol. Biol., 56, 113-117. 

Goldhirsch,A. and Gelber,R.D. (1996) Endocrine therapies of breast cancer. Semin. Oncol., 23, 494-505. 

Birrell,S.N., Roder,D.M., Horsfall,D.J., Bentel,J.M. and Tilley,W.D. (1995) Medroxyprogesterone acetate therapy in advanced breast cancer: the predictive value of androgen receptor expression. J. Clin. Oncol., 13, 1572-1577. 

Hackenberg,R., Hawighorst,T., Filmer,A., Huschmand Nia,A. and Schulz,K.-D. (1993) Medroxyprogesterone acetate inhibits the proliferation of estrogenand progesterone-receptor negative MFM-223 human mammary cancer cells via the androgen receptor. Breast Cancer Res. Treat., 25, 217-224. 

Gibson,U.E.M., Heid,C.A. and Williams,P.M. (1996) A novel method for real time quantitative RT-PCR. Genome Res., 6, 995-1001. 

Yeh,S. and Chang,C. (1996) Cloning and characterization of a specific coactivator, ARA70, for the androgen receptor in human prostate cells. Proc. Natl Acad. Sci. USA, 93, 5517-5521. 

Hakalahti,L., Vihko,P., Henttu,P., Autio-Harmainen,H., Soini,Y. and Vihko,R. (1993) Evaluation of PAP and PSA gene expression in prostatic hyperplasia and prostatic carcinoma using northern-blot analyses, in situ hybridization and immunohistochemical stainings with monoclonal and bispecific antibodies. Int. J. Cancer, 55, 590-597. 

Robertson,K.D., Keyomarsi,K., Gonzales,F.A., Velicescu,M. and Jones,P.A. (2000) Differential mRNA expression of the human DNA methyltransferases (DNMTs) 1, 3a and 3b during the G(0)/G(1) to S phase transition in normal and tumor cells. Nucleic Acids Res., 28, 2108-2113. 

Jarrard,D.F., Kinoshita,H., Shi,Y., Sandefur,C., Hoff,D., Meisner,L.F., Chang,C., Herman,J.G., Isaacs,W.B. and Nassif,N. (1998) Methylation of the androgen receptor promoter CpG island is associated with loss of androgen receptor expression in prostate cancer cells. Cancer Res., 58, 5310-5314. 

Kaplan,E.L. and Meier,P. (1958) Nonparametric estimation from incomplete observations. J. Am. Stat. Assoc., 53, 457-481. 

Peto,R., Pike,M.C. and Armitage,P. (1977) Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. Analysis and examples. Br. J. Cancer, 35, 1-39. 

Bièche,I. and Lidereau,R. (2000) Loss of heterozygosity at 13q14 correlates with RB1 gene underexpression in human breast cancer. Mol. Carcinog., 29, 151-158. 

Bieche,I., Laurendeau,I., Tozlu,S., Olivi,M., Vidaud,D., Lidereau,R. and Vidaud,M. (1999) Quantitation of myc gene expression in sporadic breast tumors with a real-time reverse transcription-PCR assay. Cancer Res., 59, 2759-2765. 

Bieche,I., Onody,P., Laurendeau,I., Olivi,M., Vidaud,D., Lidereau,R. and Vidaud,M. (1999) Real-time reverse transcription-PCR assay for future management of ERBB2-based clinical applications. Clin. Chem., 45, 1148-1156. 

Shan,L., Yang,Q., Nakamura,M., Nakamura,Y, Mori,I., Sakurai,T. and Kakudo,K. (2000) Active allele loss of the androgen receptor gene contributes to loss of androgen receptor expression in female breast cancers. Biochem. Biophys. Res. Commun., 275, 488-492. 

Isola,J.J. (1993) Immunohistochemical demonstration of androgen receptor in breast cancer and its relationship to other prognostic factors. J. Pathol., 170, 31-35. 

Grad,J.M., Dai,J.L., Wu,S. and Burnstein,K.L. (1999) Multiple androgen response elements and a myc consensus site in the androgen receptor (AR) coding region are involved in androgen-mediated up-regulation of AR messenger RNA. Mol. Endocrinol., 13, 1896-1911. 

Wolf,D.A., Kohlhuber,F., Schulz,P,, Fittler,F. and Eick,D. (1992) Transcriptional down-regulation of c-myc in human prostate carcinoma cells by the synthetic androgen mibolerone. Br. J. Cancer, 65, 376-382. 

Lu,J. and Danielsen,M. (1998) Differential regulation of androgen and glucocorticoid receptors by retinoblastoma protein. J. Biol. Chem., 273, 31528-31533. 

Knudsen,K.E., Cavenee,W.K. and Arden,K.C. (1999) D-type cyclins complex with the androgen receptor and inhibit its transcriptional transactivation ability. Cancer Res., 59, 2297-2301. 

Yeh,S., Lin,H.K., Kang,H.Y., Thin,T.H., Lin,M.F. and Chang,C. (1999) From HER2/Neu signal cascade to androgen receptor and its coactivators: a novel pathway by induction of androgen target genes through MAP kinase in prostate cancer cells. Proc. Natl Acad. Sci. USA, 96, 5458-5463. 

Mitchell,S.H., Zhu,W. and Young,C.Y. (1999) Resveratrol inhibits the expression and function of the androgen receptor in LNCaP prostate cancer cells. Cancer Res., 59, 5892-5895.