The Prostate 2926 1-267 (I 996)

Differential Expression of 5a-Reductase lsoenzymes in the Human Prostate and

Prostatic Carcinomas

Helmut Bonkhoff, Ute Stein, Gerhard Aumuller, and Klaus Remberger Institute of Pathology, University of the Saarland, Homburg (Saar), Germany, Department

of Anatomy and Cell Biology, Phillips-University Marburg, Marburg, Germany

BACKGROUND. Steroid 5a-reductase is essential for the intracellular accumulation of di- hydrotestosterone (DHT), which mediates androgen effects on target tissue. METHODS. In the present study, we describe the differential expression and cellular lo- calization of 5a-reductase 1 and 2 isoenzymes in the human prostate, and untreated and hormone-resistant prostatic carcinomas. The secretory epithelium of normal and hyperplas- tic glands showed strong nuclear 5a-reductase 1 reactivity. Accordingly, the DHT forming 5a-reductase process in secretory luminal cell types may be mediated predominantly by the type 1 isoenzyme. The androgen-independent basal cell layer variably expressed type 1 and 2 isoenzymes in nuclear and cytoplasmatic compartments. This suggests that circulating androgens are involved to control the basal cell layer, which represents the proliferative compartment of the human prostate. RESULTS. When compared with benign prostate tissue, increased 5a-reductase reactivity was detected in prostate cancer, particularly in high-grade tumors and androgen-insensitive states of the disease. In cancerous lesions, the type 1 isoenzyme tended to shift to the cytoplasm, while the nuclear staining remained unchanged or slightly increased. Referring to the type 2 isoenzyme, increased cytoplasmatic and nuclear enzyme activity was detected in malignant cells when compared with adjacent benign prostate tissue. Even endocrine differentiated tumor cells that consistently lacked the nuclear androgen receptor variably expressed 5a-reductase immunoreactivity. CONCLUSIONS. Although the functional sigruficance of the differential subcellular local- ization of type 1 and 2 isoenzymes is currently unknown, the present data suggest that prostate cancer retains the DHT forming 5a-reductase process in high-grade lesions and recurrent disease. Accordingly, circulating androgens may be still significant in these hor- mone-refractory malignancies.

KEY WORDS:

0 1996 Wiley-Liss, h c .

5a-reductase, prostate, prostate cancer

INTRODUCTION carcinomas compared to surrounding benign prostate

Androgens play an pivotal role in normal and ab- normal growth of the human prostate [l]. While tes- tosterone is the major testicular androgen circulating in blood, the androgen effects on prostatic target cells is mediated by dihydrotestosterone (DHT), which has a higher affinity than testosterone for the nuclear androgen receptor (AR) [2]. The conversion of circu-

tissue [4-61. In addition, the loss of enzyme activity seems to be quantitatively correlated with the histo- logical grade of malignancy [4-61. It has been con- cluded that the reduced DHT forming 5a-reductase pathway may be of pathogenetic significance for the development of androgen-insensitive growth in pros- tate cancer [6].

lating testosterone into DHT by steroid 5a-reductase is therefore a key reaction in the androgen ~ ~ ~ d i a t e d Prostate growth Recent biod'lernical data have shown decreased 5a-reductase activity in prostatic

Received for publication ~ p r i l 3 , 1995; accepted August 16, 1995. Address reprint requests to Dr. H. Bonkhoff, Institute of Pathol- ogy, University of the Saarland, 66421 Homburg (Saar), Germany.

0 1996 Wiley-Us, Inc.

262 Bonkhoff et al.

Recently, two functional 5a-reductase genes have been cloned in men [7,8]. Messenger RNA encoding type 1 and type 2 isoenzymes are differentially ex- pressed in male reproductive tissue and in non-an- drogen target organs [9]. Based on differential tissue distribution, it has been suggested that the type 2 isoenzyme plays an anabolic role, and the type 1 a more catabolic role in the metabolism of androgens [9]. The putative anabolic function of the type 2 iso- enzyme is supported by recent data indicating that mutations in the type 2 gene, but not in the type 1 gene, cause developmental abnormalities in DHT-de- pendent tissue in male pseudohermaphroditism [8]. However, recent immunohistochemical data have shown that the 5a-reductase 2 isoenzyme is not re- stricted to androgen target organs in the male but is present in a large number of nontarget tissue in both males and females [lo]. Although the respective role of 5a-reductase 1 and 2 isoenzymes in the DHT-form- ing process is not yet fully understood, the availabil- ity of specific 5a-reductase inhibitors has recently attracted mounting attention in fields of medical treatment of androgen-mediated growth in the hu- man prostate [ll].

In the present study, we have used polyclonal an- tibodies against 5a-reductase 1 and 2 to describe the differential expression and cellular localization of these isoenzymes in untreated prostate cancer and clinically androgen-independent disease, in compar- ison with benign prostate tissue. Double-label tech- niques for the panendocrine marker Chromogranin A (ChrA) were used to differentiate between immuno- reactive exocrine and endocrine cell types in prostate cancer.

MATERIALS AND METHODS

The material comprised 10 transurethral resections (TUR) from patients with benign prostate hyperplasia (BPH), 23 total prostatectomy specimens resected for prostate cancer, and 30 palliative TUR from patients with recurrent disease after orchiectomy. In TUR specimens from patients with BPH, histological ex- amination revealed normal glands, prostatic ducts, and glandular hyperplasia; fibromuscular nodules; and atrophic glands. The total prostatectomy speci- mens from patients with locally extensive disease (pT3) were examined in whole-mounted sections. The immunohistochemical investigations were per- formed on 23 grade 111, 14 grade 11, 4 grade I adeno- carcinomas, 13 high-grade PIN lesions, and adjacent benign prostate tissue (n = 19) present in the same tissue. The cancerous lesions were graded according to Bocking et al. 1121. Grading criteria for PIN lesions were based on the degree of nuclear atypias [13]. The

palliative TUR material from patients with clinically androgen-independent carcinomas revealed 30 poorly differentiated adenocarcinomas (grade 111). The immunohistochemical stainings were performed in formalin-fixed, paraffin-embedded material.

The polyclonal rabbit antibody L5-4 directed against 12 amino acids (amino acids 234-246 of the hu- man prostate 5a-reductase 2), as deduced from the cDNA sequence [q was used to demonstrated the 5a-reductase 2 isoenzyme [lo]. The 5a-1 isoenzyme was detected with the polyclonal antibody G-1 raised in rabbits against a native peptide (C-terminal amino acids 234-259) of the human 5a-reductase 1 isoen- zyme [14,15]. Monospecificity and sensitivity of L5-4 and G-1 antibodies were recently characterized by Western blotting and enzyme-linked immunoabsor- bent assays (ELISA) [10,14,15]. In antibody-capture ELISA tests, no sigruficant crossreactivity of the L5-4 antibody with the C-terminal peptide of the human 5a-reductase 1 could be demonstrated [ 101. Further- more, Western blotting showed that the L5-4 antibody interacts with a protein of 26- kD from human pros- tate, whereas the G-1 antibody recognizes a slightly smaller human prostate protein of 23 kD [10,15].

Prior to the application of primary antisera, the endogenous peroxidase was blocked by H202 (3%), and the slides were incubated with nonimmune swine serum for 20 min. Primary antibodies G-1 and L5-4 were applied in a dilution of 1:300. The slides were incubated overnight at 4°C. Detection was achieved by a swine antirabbit antibody (Dako, Ham- burg, Germany) and by a peroxidase antiperoxidase (PAP) complex from the rabbit (Dako). To intenslfy the immunoreaction, the linking antibody and PAP complex were applied once again for 10 min each. Antigenantibody-binding complexes were visual- ized using 3,3-diaminobenzidin (DAB) (Sigma, Dei- senhofen, Germany) or the DAPnickel complex, as recently described (161. The nuclei were faintly coun- terstained by hematoxylin. Negative controls were performed by omitting the primary antibodies, which showed no immunostaining.

At medium power ( x loo), the nuclearkytoplasmic 5a-reductase immunoreactivity was graded from 0 (negative) to 3+ (intense reaction). The results ob- tained for each histological grade were correlated with the staining pattern of the surrounding benign prostate tissue present in the same section (Tables I, 11). On adjacent sections, the endocrine phenotype of immunoreactive cells was assessed by double-label techniques. In the first staining sequence, the 5a- isoenzymes were demonstrated as described above using the DAB-nickel complex as chromagen. The pan-endocrine marker Chromogranin A (ChrA) was detected in the second staining sequence by a mono-

Expression of 5a-Reductase I and 2 in Prostate Tissue 263

clonal mouse antibody (Dako), a biotinylated rabbit to mouse antibody (Dako), and DAB as chromagen.

Furthermore, the androgen receptor (AR) status of endocrine-paracrine cell types was investigated by double-label techniques, recently described in greater detail [ l q . Briefly, AR was detected with the mono- clonal mouse antibody F.39.4.1 (Sanbio, Uden, the Netherlands), a rabbit antimouse antibody (Dako) and the PAP complex from the mouse (Dako). In the second staining sequence, ChrA was demonstrated as described above.

RESULTS

5a-Reductase immunoreactivity revealed a wide- spread distribution of both type 1 and 2 isoenzymes in stromal and epithelial compartments of benign and neoplastic prostate tissue. In normal and hyperplastic conditions, type 1 isoenzyme was predominantly lo- calized in nuclei of secretory luminal cell types (Fig. la). The basal cell layer was less consistently positive, although prostatic ducts and fibroglandular nodules showed moderate nuclear and cytoplasmatic positiv- ity in the basal cell compartment (Fig. la).

The type 2 isoenzyme was confined predomi- nantly to the basal cell layer, which showed weak cytoplasmic staining (Fig. lb). Increased nuclear and cytoplasmic reactivity was detected in the basal cells of prostatic ducts and fibroglandular nodules (Fig. lb). Endocrine-paracrine cell types that consistently lacked the nuclear androgen receptor on adjacent sec- tions variably expressed type 1 and 2 isoenzymes, preferentially in nuclei (Fig. 3a).

The prostatic stroma was immunoreactive for both isoenzymes, especially in penglandular fibrocytes, stromal nodules, and smooth muscle cells, whereas endothelial cells and outer root sheat cells were weakly positive. In general, the type 1 isoenzyme was more abundant than the type 2 isoenzyme in stromal and epithelial compartments.

Sigruficant differences were found in high-grade PIN lesions and prostatic carcinomas, as compared with the benign prostatic epithelium present in the same section (Tables I, 11). High-grade PIN and can- cerous lesions tended to be more reactive for 5a-re- ductase isoenzymes than surrounding benign tissue. In high-grade PIN and cancerous lesions, the type 1 isoenzyme appeared more abundant in the cyto- plasma but remained unchanged or slightly increased in the nucleus, especially in hormone-treated tumors (Table I, Fig. 2a). Referring to the type 2 isoenzyme, the cytoplasmatic reactivity progressively increased in PIN and neoplastic lesions, and the nuclear stain- ing significantly increased in relapsed tumors after hormonal therapy (Table 11, Fig. 2b). The nuclear

Fig. 1. Glandular hyperplasia a: In secretory luminal cell types, the 5a-reductase I isoenzyme is expressed predominantly in nu- clei. Nuclear and cytoplasmic immunoreactivity is detected focally in the basal cell layer (arrowhead). b: The type 2 isoenqme is confined to basal cell types, which show weak cytoplasmatic and nuclear membrane positivity (arrowhead). Slight hematoxylin coun- terstain. ~400.

staining of the type 2 isoenzyme was localized pref- erentially in the nuclear membrane, whereas the type 1 isoenzyme was more reactive in the nuclear matrix (Fig. 2b).

Phenotypic characterization of 5a-reductase im- munoreactivity revealed that endocrine differentiated tumor cells widely expressed both type 1 and 2 isoen- zymes, which were more abundant in the nucleus (Fig. 3b). On adjacent sections, these endocrine cell types consistently lacked nuclear AR immunoreactiv- ity, while exocrine (ChrA-negative) tumor cells van- ably expressed the receptor protein. These results have been obtained in four hormone-treated adeno- carcinomas with extensive neuroendocrine differen-

264 Bonkhoff et al.

3.

2-

1-

0-

TABLE 1. Compiarative Evaluation of 5a-Reductase I l m m u n d v i t y in Benign Prostate Tissue (BPT),

High-Grade PIN (PIN), Low-Grade (GI, GII), High-grade (Glll), and Recurrent Prostate Cancer After

Hormonal Therapy (rPC)*

TABLE 2. Comparative Evaluation of 5a-Reductase 2 lmmunoreactivity in Benign Prostate Tissue (BPT),

High-Grade PIN (PIN), Low-Grade (GI, GII), High-Grade (GIII), and Recurrent Prostate Cancer After

Hormonal Therapy (rPC)*

:: x x tt x x

e* x x

x x x x

x x

n=l9 n-13 n=l8 11-23 n=30

'score: Neg (no immunoreactivity detectable); + 1, + 2, + 3 (weak, moderate, strong staining). Solid circles, nuclear staining; crosses, cytoplasmatic staining.

tiation (more than 50% of ChrA-positive cells). Iden- tical findings were observed in untreated carcinomas, which showed focal endocrine differentiation as de- tected by ChrA.

DISCUSSION

In the present study we used polyclonal antibodies against 5a-reductase 1 and 2 to describe the distribu- tion of these isoenzymes in benign and malignant prostate tissue. The specificity and sensitivity of these antibodies for the human 5-a-reductase 1 and 2 has been reported recently in greater detail [10,14,15]. Al- though the cellular distribution is important to un- derstand the regulatory impact of 5-01 reductase isoenzymes on prostatic growth, it should be empha- sized that functional aspects of substrate affinity can hardly be deduced from immunohistochemical data. Nevertheless, the results of the present study may provide more insight in the androgen regulation of the various cell types during normal and abnormal prostatic growth.

The normal and hyperplastic prostate epithelium is composed of secretory luminal, basal, and endo- crine-paracrine cells, which have different hormonal requirements [18]. The basal cell layer is androgen- independent in androgen-cycling experiments [18-

PIN G 1. II

n=l3 n-18 1-1323 n d 0

*Score: Neg (no immunoreadivity detectable); + 1, +2, + 3 (weak, moderate, strong staining). Solid circles, nuclear staining; crosses, cytoplasmatic staining.

201 and represents the proliferative compartment of the normal and hyperplastic prostate epithelium [21].

Although androgen-independent, basal cells may be androgen-responsive, as documented by the pres- ence of the nuclear AR [22]. In the present study, we have shown that basal cells variably express 5a-re- ductase isoenzymes, which are considered to play a pivotal role in the androgen regulation of prostatic target cells. The type 2 isoenzyme is predominantly localized in the cytoplasm of basal cells, whereas the type 1 isoenzyme is expressed in nuclear and cyto- plasmic compartments. The presence of 5a-reductase reactivity and nuclear AR [22] in basal cell types clearly indicates that androgens are involved to con- trol the basal cell layer, which presumably houses the stem cell population of the prostatic epithelium [23,24]. Secretory luminal cells represent the major phenotype in the normal and hyperplastic epithelium and require continuous support by circulating andro- gens for its maintenance [18-201. The type 1 isoen- zyme is strongly expressed in the nuclear matrix of secretory luminal cells, while the type 2 isoenzyme is unreactive or weakly positive in this cell type. This may indicate that the DHT-forming process in the secretory epithelium is mediated predominantly by the 5a-reductase 1 isoenzyme. In contrast to secretory luminal and basal cells, prostatic endocrine-paracrine

Expression of 5ar-Reductase I and 2 in Prostate Tissue 265

Fig. 2. Poorly differentiated adenocarcinoma relapsed after hor- monal therapy. a: Intense Sa-reductase I is detected in nuclei, while the cytoplasm of tumor cells is less immunoreactive. b The 5a-reductase 2 isoenzyme is expressed predominantly in the cy- toplasm and nuclear membranes (armwhead). Slight hematoxylin counterstain. x 200.

cell types are considered androgen-unresponsive be- cause this cell type consistently lacks the nuclear AR [lq. However, the presence of 5a-reductase reactiv- ity indicates that circulating androgens or other sub- strates for this enzyme may be involved in the regu- lation of prostatic endocrine cell types.

Most prostatic cancers initially retain an androgen- responsiveness for stimulation of their growth but invariably relapse to an androgen-insensitive disease after androgen withdrawal therapy [25]. The under- lying mechanisms responsible for the progression to an androgen-insensitive growth are poorly under- stood. Experimental studies on prostatic cancer cell lines have indicated that the transition to an andro-

Fig. 3. Glandular hyperplasia (a) and recurrent prostatic adeno- carcinoma with extensive neuroendocrine differentiation (b). Si- multaneous demonstration of 5-a-reductase I immunoreactivity and Chr A. Strong nuclear 5-a-reductase I reactivity is detected in endocrine-paracrine cell types of hyperplastic glands (a) and endo- crine differentiated (Chr A positive) tumor cells (b). The arrow- heads indicate Chr A positive cell types. X I OOO (a), x 200 (b).

gen-insensitive state may be associated with the se- lective outgrowth of androgen-unresponsive tumor cells [26-291. However, recent data obtained in hu- man tissue have shown that most prostatic carcino- mas, even when progressed to a state of hormone insensitivity, contain a structural intact AR gene and widely express receptor proteins with retained andro- gen-binding capacity [30,31]. The present data about the distribution of 5a-reductase isoenzymes in pros- tate cancer lend credence to the concept that andro- gens are still involved to control tumor growth in high-grade and clinical androgen-independent dis- ease. Most important, a loss of 5a-reductase reactivity

266 Bonkhoff et al.

in high-grade or hormone-refractory tumors could not be demonstrated in the present study. On the contrary, these neoplastic lesions tend to increase their 5a-redudase activities when compared to the adjacent benign prostate tissue. These findings are in contrast with recent biochemical data that have shown decreasing enzyme activity correlates with the histological grade of malignancy [5,6]. Discrepancies between biochemical and immunohistochemical anal- ysis in this field are most probably related to stromal derived enzyme activity that interfers with the results of the biochemical assay.

Sigruficant differences between 5a-reductase reac- tivity in cancerous lesions and adjacent benign tissue concern the staining intensity and subcellular local- ization of the two isoenzymes. The type 1 isoenzyme tends to shift to the cytoplasrna in PIN and neoplastic lesions, whereas the nuclear staining remains un- changed or slightly increased, when compared with benign prostate glands. Conversely, increased nu- clear and cytoplasmic staining is detected, especially in recurrent carcinomas when compared with the type 2 isoenzyme reactivity in benign tissue. The sig- nificance of the differential subcellular localization of 5a-isoenzymes for the DHT forming and desactivat- ing pathway remains currently unknown and war- rants further investigations. Nevertheless, the pres- ence of both 5a-reductase 1 and 2 isoenzymes at high levels in poorly differentiated and hormone-refrac- tory lesions may indicate that these tumors retain reg- ulatory pathways to accumulate the biological active testosterone.

There is now increasing evidence that neuroendo- crine differentiation may have basic implications in the androgen-insensitive growth of common prostate cancer [32,33]. In fact, endocrine-differentiated tumor cells consistently lack the nuclear AR in both primary and relapsed prostatic carcinomas [ 17. Surprisingly, these presumably androgen-insensitive cell popula- tions variably express 5a-reductase reactivity. This in- dicates that s i e c a n t amounts of DHT may be pro- duced to act on these cells but without interference of the receptor. Alternatively, other known substrates for the 5a-reductase, including corticosterone and progesterone may be involved to control endocrine- differentiated tumor cells.

In summarizing the present data, we may con- clude that prostate cancer retains the DHT-forming 5a-reductase pathway, even in high-grade and hor- mone-refractory lesions. Because most hormone resistent prostate cancers obviously contain a func- tional intact AR [30,31], it is conceivable that andro- gens still play an active role in these clinically androgen-insensitive tumors. Residual amounts of adrenal-derived androgens may be of clinical impor-

tance in prostate malignancies treated by bilateral orchiectomy .

ACKNOWLEDGMENTS

This work was supported by grant Bo 1018/2-1 from the Deutsche Forschungsgemeinschaft.

REFERENCES

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

Davies P, Eaton CL: Regulation of prostate growth. Re- view. J Endoainol 131:5-17, 1991. Bruchovsky N, Wilson JD: The conversion of testoster- one to andorstan-17P-ol-3-one by rat prostate in vivo and vitro. J Biol Chem 243:2012-2021, 1968. Petrow V: The dihydrotestosterone (DHT) hypothesis of prostate cancer and its therapeutic implications. Prostate 9 3 3 , 1986. Voigt KO, Bartsch W: Intratissular androgens in be- nigne hyperplasia and prostatic cancer. J Steroid Bio- chem 25:749-757, 1986. Klein H, Bressel M, Kastendieck H, Voigt KD. Quanti- tive assessment of endogenous testicular and adrenal sex steroids and of steroid metabolizing enzymes in untreated human prostatic cancerous tissue. J Steroid Biochem 30:119-130, 1988. Klein H, Bressel M, Kastendieck H, Voigt KD Biochem- ical endocrinology of prostate cancer. In Voigt K-D, Knabbe C (eds): “Endocrine-Dependent Tumors.” New York: Raven Press, 1991, 131-163. Andersson S, Bishop RW, Russell DM Expression of cloning of alpha-reductase, an enzyme essential for male sexual differentiation. J Biol Chem 264.16249- 16255, 1991. Andersson S, Bermann DM, Jenkins FP, Russell DW: Deletion 5 alpha 2 gene in male pseudohermaphrodit- ism. Nature 354:159-161, 1991. Normington K, Russell DW: Tissue distribution and kinetic characteristics of rat steroid 5 alpha-reductase. Evidence for distinct physiological functions. J Biol Chem 26719548-19554, 1992. Eicheler W, Tuohimaa P, Vilja P, Adermann K, Forss- mann WE, Aumuller G: Imunohistochemical localiza- tion of human 5 alpha-reductase 2 with polyclonal an- tibodies in androgen target and non-target human tissues. J Histochem Cytochem 42567-675, 1994. Geller J: Nonsurgical treatment of prostatic hyperpla- sia. Cancer 7039-345, 1992. Bocking A, Krehn J, Heinzel-Wack M: Combined his- tological grading of prostatic cancer. Cancer 50:288- 294, 1982. Brawer MK: Prostatic intraepithelial neoplasia. A pre- malignant lesion. Hum Path01 23:242-248, 1992. Eicheler W, Bacher M, Seitz J, Adermann K, Forssmann WG, Aumiiller G: Characterization and tissue distribu- tion of 5 alpha-reductase using a polyclonal antibody. In Spera G, Fabrini A, Gnessi L, Bardin CW (eds): “Mo- lecular and Cellular Biology of Reproduction.” New York: Raven Press, 1992, p 257-265. Eicheler W, Seitz J, Forssmann WG, Adernmann K, Aumiiller G: Immunological characterization of rat liver 5 alpha-reductase. Eur J Cell Biol6O(suppl3):151, 1993. Bonkhoff H, Wernert N, Dhom G, Remberger K Rela- tion of endocrine-paracrine cells to cell proliferation in

Expression of 5cu-Reductase I and 2 in Prostate Tissue 267

normal, hyperplastic and neoplastic human prostate. Prostate 19:91-98, 1991.

17. Bonkhoff H, Stein U, Remberger K: Androgen receptor status in endocrine-paracrine cell types of the normal, hyperplastic and neoplastic human prostate. Virchows Arch A Pathol Anat 423:291-294, 1993.

18. Aumiiller G: Morphologic and endocrine aspects of prostatic function. Prostate 4195, 1983.

19. English HF, Santen RJ, Isaacs JT: Response of glandular versus basal rat ventral prostatic epithelial cells to an- drogen withdrawal and replacement. Prostate 11:229- 242, 1987.

20. Isaacs JT, Coffey DS: Etiology and disease process of benign prostatic hyperplasia. Prostate Suppl 233-50, 1989.

21. Bonkhoff H, Stein U, Remberger K The proliferative function of basal cells in the normal and hyperplastic human prostate. Prostate 24114-118, 1994.

22. Bonkhoff H, Remberger K Widespread distribution of nuclear androgen receptors in the basal cell layer of the normal and hyperplastic human prostate. Virchows Arch A Pathol Anat 4223548,1993.

23. Bonkhoff H, Stein U, Remberger K: Multidirectional differentiation in the normal, hyperplastic and neoplas- tic human prostate. Simultaneous demonstration of cell specific epithelial markers. Hum Pathol25:42-46,1994.

24. Bonkhoff H, Remberger K Differentiation pathways and histogenetic aspects of normal and abnormal pro- static growth. A stem cell model. Prostate 28:98-106, 1996.

25. Gittes RF: Carcinoma of the prostate. N Engl J Med 324236-245, 1991.

26. Bruchovsky N, Brown EM, Coppin CM, Goldenberg SL, L E Riche JC, Murray NC, Rennie PS: The endocri- nology and treatment of prostate tumor progression. Prop Clin Biol Res 239347437, 1987.

27. Isaacs JT: New principles in the management of meta- static cancer. Progr Clin Biol Res 185:383-405, 1985.

28. Quarmby VE, Beckmann WC Jr, Cooke DB, Lubahn DB, Joseph DR, Wilson EM, French FS: Expression and localization of androgen receptor in the R-3327 Dun- ning rat prostatic adenocarcinoma. Cancer Res 50:735- 739, 1990.

29. Trapmann H, Ris-Stalpers C, Van der Korput JAGM, Kuiper GGJM, Faber PW, Romijn JC, Mulder E, Brink- mann AO: The androgen receptor functional structure and expression in transplanted human prostate tumor cell lines. J Steroid Biochem Mol Biol37837-842, 1990.

30. Van der Kwast TH, Schalken J, Ruizeveld de Winter JA, Van Vroonhoven CCJ, Mulder E, Boersma W, Trapman J: Androgen receptors in endocrine-therapy-resistant human prostate cancer. Int J Cancer 48:189-193, 1991.

31. Ruizeveld de Winter JA, Janssen PJA, Sleddens HMFE3, Verleum-Mooijman M f f , Trapmann J, Brinkmann AO, Samerse AB, Schroeder FH, Van der Kwast TH: Androgen receptor status in localized and locally pro- gressive, hormone refractory human prostate cancer. Am J Pathol 144:735-746, 1994.

32. Di Saint'Agnese PA: Neuroendocrine differentiation in carcinoma of the prostate. Cancer 70:254-268, 1992.

33. Di Saint'Agnese PA Neuroendocrine differentiation in the prostate carcinoma. Hum Path01 23287-296, 1992.