Available online at www.sciencedirect.com

Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167

Estrone sulfate (E1S), a prognosis marker for tumor aggressiveness in prostate cancer (PCa)ଝ Frank Giton a,∗, Alexandre de la Taille b, Yves Allory b, Herve´ Galons c, Francis Vacherot b, Pascale Soyeux b, Claude Clement´ Abbou b, Sylvain Loric b, Olivier Cussenot b, Jean-Pierre Raynaud d, Jean Fiet b a AP-HP CIB INSERM IMRB U841eq07, Henri Mondor, Facult´edeM´edecine, 94010 Cr´eteil, France b INSERM IMRB U841 eq07, CHU Henri Mondor, Facult´edeM´edecine, 94010 Cr´eteil, France c Service de Chimie organique, Facult´e de Pharmacie Paris V, 75006 Paris, France d Universit´e Pierre et Marie Curie, 75252 Paris, France Received 26 December 2006; accepted 26 October 2007

Abstract

Seeking insight into the possible role of estrogens in prostate cancer (PCa) evolution, we assayed serum E2, estrone (E1), and estrone sulfate (E1S) in 349 PCa and 100 benign prostatic hyperplasia (BPH) patients, and in 208 control subjects in the same age range (50–74 years). E1 (pmol/L ± S.D.) and E1S (nmol/L ± S.D.) in the PCa and BPH patients (respectively 126.1 ± 66.1 and 2.82 ± 1.78, and 127.8 ± 56.4 and 2.78 ± 2.12) were significantly higher than in the controls (113.8 ± 47.6 and 2.11 ± 0.96). E2 was not significantly different among the PCa, BPH, and control groups. These assays were also carried out in PCa patients after partition by prognosis (PSA, Gleason score (GS), histological stage, and surgical margins (SM)). Significantly higher E1S levels were found in PCa with: PSA > 10 ng/L (3.05 ± 1.92) versus PSA ≤ 10 ng/mL (2.60 ± 1.55), stage pT3-T4 (2.99 ± 1.80) versus pT2 (2.58 ± 1.58), and positive (3.26 ± 1.95) versus negative margins (2.52 ± 1.48). E1 was higher in poor- than in better-prognosis PCa. E2 was significantly higher in PCa with GS ≥ 4 + 3 (109.5 ± 43.8) versus GS ≤ 3 + 4 (100.6 ± 36.5) and increased significantly when GS increased from 3 + 3 to 4 + 4. Estrogens, especially E1S appeared to be possible markers of PCa progression. Attempting to identify potential sources of E2 in PCa according to prognosis, as well as in BPH, we found a significant correla- tion coefficient between E1S and E2 (0.266–0.347) in poor-prognosis PCa and no correlation in BPH (0.026) and better-prognosis PCa (0.013–0.104). It is as though during progression of PCa from good to poor prognosis there were a shift in the E1 to E2 metabolic pathway from predominantly oxidative to predominantly reductive. © 2008 Elsevier Ltd. All rights reserved.

Keywords: Estrone sulfate; Estrogens; Prostate cancer; Tumor aggressiveness

1. Introduction

Although the prostate gland is the site of intense steroid metabolism [1], normal and pathological development of the Abbreviations: PCa, prostate cancer; BPH, benign prostatic hyperplasia; prostate depends on the blood hormone environment, i.e., on E1, estrone; E2, estradiol; E1S, estrone sulfate; TT, testosterone; A, 4- androstenedione; GS, Gleason score; HS, histological stage; SM, surgical the concentration of androgens and estrogens synthesized by margins. the gonads and adrenals, as well as on those that derive from ଝ This work has been presented in part at the 17th International Symposium peripheral metabolism in all tissues. Sensitive and specific of the Journal of Steroid Biochemistry and Molecular Biology, Seefeld, Austria, assays of blood androgen and estrogen concentrations could 31 May–3 June 2006. ∗ Corresponding author. Tel.: +33 1 49 81 35 58; fax: +33 1 49 81 35 52. therefore shed light on both normal and pathological prostate E-mail address: [email protected] (F. Giton). function.

0960-0760/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jsbmb.2007.10.005 F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167 159

Over the years, numerous studies have attempted to evalu- Twelve-hour fasting blood samples were drawn between 8 ate the role of androgens, particularly testosterone, in PCa and and 10 am. Serum was separated from the plasma and divided BPH [2–19]. However, few studies have addressed the role of into several fractions. Each volunteer was subjected to clini- estrogens (estradiol and estrone) in these pathologies, due to dif- cal chemistry testing that included the following parameters: ficulties encountered in measuring their low circulating levels in blood glucose, creatinine, triglycerides, ␥-glutamyltransferase males. Some studies [8,9,17,19] suggested a reduction in the risk (␥-GT), alanine l-aminotransferase (ALAT), and l-aspartate of PCa with higher levels of estradiol (E2), whereas others did aminotransferase (ASAT). Volunteers with abnormal clin- not [6,7,12]. ical chemistry values were excluded from the study. Recent evidence suggests that estrogens significantly con- Serum samples were kept frozen until hormone and PSA tribute to the genesis of prostate cancer (PCa) [20].Inthe assays. presence of androgen, E2 has been shown to stimulate carci- Of the 539 selected healthy men aged 20–74, 268 were over noma in situ and adenocarcinoma in Noble rats [21,22]. The 50 years of age. Among these 268 subjects, 60 were excluded administration of E2 in conjunction with testosterone to Noble for having moderate symptoms of BPH or a PSA > 3 ng/mL, rats results in a higher incidence of PCa than does adminis- abnormal gonadotrophins, low testosterone (<8.67 nmol/L), or ® tration of testosterone alone [23,24]. Toremifene , a selective E1S < 5th or >95th percentiles of the overall population included estrogen receptor modulator (SERM), suppressed the develop- in the study. ment of high-grade of prostatic intraepithelial neoplasia (PIN), In addition, 51 healthy volunteers from among the laboratory and decreased the incidence of adenocarcinoma in the transgenic staff, aged 40–73, provided two blood samples during the same mouse prostate model [25]. day, the first one between 8 and 10 am in the fasting state, and the Estrogen metabolites have been proposed as causative factors second one between 4 and 6 pm, to assess a potential circadian in neoplastic transformation of prostatic epithelium [26,27].To rhythm of serum androgens and estrogens. obtain insight into the possible role of estrogens in the evolu- tion of PCa, we assayed unconjugated serum estrogens (E2,E1) 2.2. Patients and estrone sulfate (E1S) – an estrogen which is not directly active but whose plasma level is the highest – in all consecutive Patients who underwent radical prostatectomy had untreated untreated and newly diagnosed PCa patients who underwent rad- prostate cancer (PCa) diagnosed on the basis of prostate biopsy, ical prostatectomy in our urology service, starting in July 2004 negative computerized tomography (CT-scan) results, and neg- (and still ongoing). The same estrogens were simultaneously ative bone scan. All were treated for clinically localized prostate assayed in patients with BPH (and who underwent surgery for cancer. it), and in controls recruited during the same period. Androgens BPH patients underwent prostatectomy or transurethral pro- such as testosterone (TT) and 4-androstenedione (A) were also static resection for acute urinary retention or severe or moderate measured in PCa and BPH, since they are possible precursors urinary symptoms. None was treated with Finasteride® or for estrogens. Dunasteride®. The patients recruited for this prospective study were con- secutive, but only those with a BMI ≤ 29 kg/m2 (nPCa = 349, 2. Populations studied nBPH = 100) were retained. Blood samples were drawn on the day before surgery at 4 and 6 pm, as per the practice at our 2.1. Healthy volunteers hospital. PCa patients were classified as having a good or a poor prog- Volunteer men aged 20–74 were recruited in a health nosis according to histological criteria: Gleason score (GS), care center (Institut Inter-regional´ pour la Sante´ (IRSA), histological stage (HS), surgical margins (SM), and serum PSA Tours, France). All volunteers with a body mass index (BMI) levels. All participants signed a written informed consent docu- >29 kg/m2 or suffering from chronic disease (diabetes, can- ment. cer, renal disease, liver disease, chronic inflammatory disease, heart or pulmonary disease, and myocardial infarction dur- 3. Methods ing the preceding 3 months) were excluded from the study, as were men under treatment with one or more drugs known 3.1. Histological methods to influence the hypothalamic/pituitary/gonadal axis, such as androgens, gonadotropin-releasing hormone (GnRH) agonists, All specimens obtained from patients who underwent radi- gonadotropins, antiandrogens, corticoids, Synacthen®, thyroid cal prostatectomy were analyzed in our Pathology Department drugs, and pharmacological agents that could elicit hyperpro- laboratory. Each of the 349 surgically removed prostates was lactinemia. weighed, stained, and fixed in 10% formalin. After fixation, a 1- These subjects underwent digital rectal examination for cm piece of tissue was cut from the proximal and distal surgical prostate tumor detection and responded to questions concern- margins (base and apex) and transected into multiple sections. ing their urinary symptoms. None of them was under treatment The remaining prostate gland was cut into sequential sections for benign prostate hyperplasia (BPH) or prostate disease. They from the right and left lobes, from apex to base. The remainder all had no or only moderate symptoms of BPH (n = 10). of the prostate was then embedded in paraffin. The seminal vesi- 160 F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167 cles were cut and separately embedded at the junction of entry for evaporation. The dried residues were re-dissolved in 0.5 mL to the prostate gland. of methanol/water, 25/75 (v/v), for chromatography. Extracapsular extension was diagnosed if tumor was found The second, chromatographic step was carried out using the in the periprostatic soft tissues or was seen penetrating through a VisiprepTM VacuumManifold (Supelco®, Bellefonte, PA16823, fibromuscular capsule and exiting the other side. Seminal vesicle USA), in order to facilitate entry of the solvents through the invasion was diagnosed if tumor was seen within the muscu- hypersil phase minicolumn. The minicolumns were regenerated lar wall of the seminal vessel. A positive surgical margin was by successive addition of 6 mL H2O and 12 mL CH3OH/H2O denoted by the presence of carcinoma abutting on India ink. All 25/75 (v/v). After introducing the 0.5 mL of the extract con- pelvic lymph nodes were evaluated for the presence of metastatic taining the conjugated serum steroids into the minicolumns, we disease. All cases were assigned a GS. Each tumor was staged added 2.5 mL CH3OH/H2O, 25/75, to the minicolumns, but the according to the 2002 TNM system. eluates were not retained. We then added another 2.5 mL of CH3OH/H2O, 25/75, kept the eluate, and then evaporated it to 3.2. Steroid assays dryness. After re-dissolving this extract containing E1S in phosphate All hormones were determined in the samples by immunoas- gelatin buffer, we withdrew an aliquot of the extract for recovery say, except for dehydroepiandrosterone (DHEA) and E1, which measurement. were measured by gas chromatography/mass spectrometry Radioimmunoassay was carried out in the third and final step. (GC/MS). Interassay precision was expressed as the coefficient of variation between 4.3 and 5.1%, according to the E1S control serum level 3.3. Steroid immunoassays (Table 1). Specificity depended on the cross-reactivities of the 3.3.1. 4-Androstenedione (A), dihydrotestosterone anti-E1S antibody (E1 < 0.6%, E2 < 0.001%, E2S < 0.014%, (DHT), and testosterone (TT) E2-glucuronide <0.001%, DHEAS < 0.002%, androstenediol As previously published [28–30], 4-androstenedione (A), sulfate <0.001%, androsterone sulfate <0.0001%), and on the dihydrotestosterone (DHT), and testosterone (TT) were assayed extraction and chromatographic steps. The chromatographic in 0.5 mL of serum, using a three-step procedure including: (1) step separates E1S from DHEAS, whose serum levels are ® extraction, (2) Celite chromatography, and (3) time-resolved 3000–6000-fold higher than E1S. The least detectable dose fluoroimmunoassay (TR-FIA) for A and TT, or radioimmunoas- of E1S in assayed 0.5 mL of plasma was determined to be say for DHT. The Celite® chromatography step monitored by 0.082 nmol/L. minute doses of 3H steroids separated the three steroids without Sixty-six sera were assayed both by our RIA method and overlap [28]. by liquid chromatography/mass spectrometry/mass spectrome- try (LC/MS/MS.) Statistical analysis using the non-parametric 3.3.2. Bioavailable testosterone (BT) paired Wilcoxon’s test did not show a significant difference BT, or SHBG-unbound testosterone, was analyzed accord- between results obtained using these two E1S assay methods. ing to the Tremblay and Dube method [31,32]. Briefly, after ◦ 3 incubation of serum at 37 C with spiked H testosterone, 3.3.4. Estradiol assay (E2) SHBG-unbound testosterone was separated by precipitation E2 was measured using a sensitive and specific kit (Clini- from SHBG-bound testosterone by the addition of a satu- cal AssaysTM Estradiol-2 Code P2210 DiaSorin, 92160 Antony, rated solution of ammonium sulfate, followed by centrifugation. France). Radioactivity was counted in the supernatant and the %BT was The serum levels of E2 in controls were compared two by two calculated. BT was then converted to nmol/L by multiplying by (with the non-parametric paired test of Wilcoxon) using E2 RIA the corresponding TT value. and GC/MS methods. In the following E2 serum level ranges: (1) <73.5 pmol/L (n = 34); (2) 73.5 pmol/L ≤ E2 < 110.3 pmol/L 3.3.3. E1S assay [33] (n = 263); (3) 110.3 pmol/L ≤ E2 < 147 pmol/L (n = 156); (4) E1S was measured by competitive radioimmunoassay, using E2 ≥ 147 pmol/L (n = 50), we did not find any statistical differ- 3 an anti-E1-S-6-carboxymethyloxime/BSA antibody [33] and H ences between the results obtained using the two methods. This E1S, after organic solvent extraction and C18-hypersil phase was also done for the same interval E2 values, using the Kendall 7 ␮M partition chromatography. The assay procedure involved non-parametric test, which yielded very significant correlations. three steps: The first step began with the addition of 4000 dpm of The least detectable doses of E2 measured by RIA and by E1S, ammonium salt, [6,7-3H(N)], 60 Ci/mmol, NET 203 (NEN GC/MS were respectively 7.34 and 11.01 pmol/L. Life Science Products, B-1930 Zaventem Belgium) to 0.5 mL of serum as an internal standard to assess procedural losses, and 3.3.5. GC/MS assay of E1 and DHEA incubated 20 min at 20 ◦C. The extraction was carried out with These assays were adapted from the method described by 6 mL cyclohexane/ethylacetate, 50/50 (v/v) (to eliminate uncon- Labrie et al. [34]. jugated steroids). The aqueous phase was re-extracted with 6 mL Calibration standards ranging from 0.69 to 56.17 nmol/L for pure ethanol, and then centrifuged at 3300 rpm for 10 min at DHEA and from 29.6 to 2396.5 pmol/L for E1 were prepared +4 ◦C. The alcoholic supernatants were transferred to glass tubes from charcoal-treated human serum. F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167 161

Table 1 CVs of three quality control serums, assayed in 45 runs

Quality controls E1S (nmol/L) E1 (pmol/L) E2 (pmol/L) TT (nmol/L) BT (%) DHEA (nmol/L) A (nmol/L) DHT (nmol/L) Level 1 Mean assayed value 1.43 30.3 24.8 8.67 49.7 0.94 3.56 1.76 CV (%) 4.5 11.2 11.9 3.6 3.9 9.3 7.3 8.6 Level 2 Mean assayed value 2.75 95.7 139.3 13.94 30.9 3.88 7.12 3.48 CV (%) 4.3 5.0 4.4 3.3 4.1 4.9 4.9 7.8 Level 3 Mean assayed value 5.47 294.2 285.4 20.94 14.3 7.28 14.11 6.89 CV (%) 5.1 4.3 3.9 2.3 9.2 4.3 3.6 5.2

After extraction and derivatization of steroids from sera, Table 2 Histological results and PSA values of prostate specimens DHEA and E1 were chromatographed using a GC system (Ref. 6890N, Agilent Technologies) that employs a 50% phenyl- Gleason score pT stages methyl polysiloxane (Ref. DB-17HT) capillary column (internal × ␮ 3 + 3 23.5% pT2a 10.6% diameter 30 mm 0.25 mm, film thickness 0.15 m) and helium 3 + 4 36.1% pT2b 3.2% as the carrier gas. Analytes were detected using an HP5973 (Agi- 4 + 3 28.1% pT2c 44.1% lent Technologies) quadrupole mass spectrometer equipped with 4 + 4 9.2% pT3a 25.4% a chemical ionization source. The lower limits of quantification 4 + 5 2.5% pT3b 7.1% (LLOQ) were 0.69 and 29.6 pmol/L, respectively, for DHEA and 5 + 5 0.6% pT4 9.6% E1. Surgical margins Positive 33.1% Negative 66.9% 3.3.6. PSA immunoassays ® PSA values PSA was measured using the Hybritech PSA test ≤10 ng/ml 71.1% (Beckman-Coulter). >10 ng/ml 28.9%

3.3.7. Statistical methods 4.2. Prostate specimens The statistical tests used were the non-parametric Mann & Whitney, Wilcoxon, Kendall correlation, and Spearman corre- The histological results (GS, HS, SM), and PSA levels from lation tests, and the simple regression test. 349 patients’ prostate specimens appear in Table 2.

4. Results 4.3. BMI

4.1. Precision of steroid assays On one hand, the BMI of the control subjects, BPH patients, and all PCa patients (Table 3) of whatever prognosis, and on the Interassay precision (CV) (n = 45 runs) for three increasing other, after PCa classification (Table 4) according to serum PSA serum control levels is reported in Table 1. level, GS, HS, and SM, were not significantly different.

Table 3 Comparison by age, BMI, TT, BT, A, E2,E1,E2/TT, and E1S serum levels among PCa, BPH, and male control subjects in the 50–74-year age range Age and hormone levels (mean ± S.D.) PCa, n = 349 BPH, n = 100 Control (C), n = 208 PCa vs. C PCa vs. BPH BPH vs. C

Age (years) 62.3 ± 6.2 63.6 ± 5.9 62.1 ± 5.5 ns ns ns BMI (kg/m2) 25.8 ± 2.8 26.6 ± 2.8 25.4 ± 2.4 ns ns ns TT (nmol/L) 13.31 ± 5.83 15.77 ± 6.45 16.39 ± 4.92 na p = 0.0002 na BT (nmol/L) 3.14 ± 1.85 3.17 ± 1.39 4.53 ± 1.59 na ns na A (nmol/L) 2.42 ± 1.21 2.39 ± 1.06 3.24 ± 1.36 na ns na E2 (pmol/L) 109.6 ± 39.3 113.0 ± 37.1 115.3 ± 27.1 ns ns ns E1 (pmol/L) 126.1 ± 66.1 127.8 ± 56.4 113.8 ± 47.6 p = 0.0209 ns p = 0.0467 E2/TT 10.3 ± 15.1 7.4 ± 3.3 7.2 ± 2.7 na p = 0.0208 na E1S (nmol/L) 2.82 ± 1.78 2.78 ± 2.12 2.11 ± 0.96 p < 0.0001 ns p = 0.0008 ns: not significant; nd: not applicable. 162 F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167

Table 4 Comparison among of E1S, E1,E2, and E2/TT serum estrogen levels; BMI, and prostate weight, according to prognosis parameters 2 n E1S (nmol/L) E1 (pmol/L) E2 (pmol/L) E2/TT BMI (kg/m ) Prostate weight (g) PSA (nmol/L) ≤10 248 2.60 ± 1.55 122.7 ± 65.6 103.1 ± 39.2 9.5 ± 9.3 25.78 ± 2.89 52.5 ± 22.2 p 0.0287 (*) 0.4976 0.9245 0.1009 0.9333 0.3762 >10 101 3.05 ± 1.92 128.5 ± 56.6 103.6 ± 38.0 12.6 ± 24.7 25.81 ± 2.61 54.9 ± 19.2 Gleason ≤3 + 4 208 2.66 ± 1.71 126.2 ± 69.5 100.6 ± 36.5 8.9 ± 8.1 25.78 ± 2.70 54.7 ± 23.4 p 0.1681 0.9173 0.0496 (*) 0.0375 (*) 0.9311 0.0931 ≥4 + 3 141 2.87 ± 1.64 127.0 ± 61.8 109.5 ± 43.8 12.5 ± 22.1 25.74 ± 2.94 50.5 ± 18.1 Stage pT2 199 2.58 ± 1.58 122.1 ± 68.6 102.4 ± 35.8 9.2 ± 8.6 25.67 ± 2.60 54.0 ± 22.5 p 0.0298 (*) 0.2522 0.4078 0.2549 0.5839 0.2429 pT3-pT4 150 2.99 ± 1.80 131.4 ± 62.3 106.1 ± 45.1 10.8 ± 16.2 25.88 ± 3.05 51.2 ± 17.6 Margins Negative 233 2.52 ± 1.48 118.5 ± 55.9 100.6 ± 38.4 9.1 ± 8.9 25.78 ± 2.79 51.9 ± 21.9 p 0.0002 (*) 0.0026 (*) 0.0373 (*) 0.1199 0.9171 0.5187 Positive 116 3.26 ± 1.95 144.3 ± 84.1 110.5 ± 42.4 11.4 ± 17.5 25.74 ± 2.81 53.6 ± 21.3

* Significant.

4.4. Circadian variations in serum steroids in 51 healthy E1 in the controls (r = 0.389 and 0.585, respectively), in the men PCa patients (r = 0.125 and r = 0.392), and in the BPH patients between E1S versus E1 (r = 0.469) were observed, but no cor- We found a significant decrease in serum TT, BT, DHT, 4- relation between E1S versus E2 in the BPH patients was found androstenedione, and DHEA in blood samples from men aged (r = 0.026). It was as if E1S could not lead to the formation of 40 to 73 years drawn between 4 and 6 pm, compared with serum E2 in BPH. obtained from blood samples from the same subjects drawn between 8 and 10 am. This significant variation rendered the 4.7. Comparison of serum estrogen levels by PCa comparison of serum androgens between control subjects and prognosis (Table 4) patients inappropriate. However, no significant variation in E1, E S, and E serum levels was observed between morning and 1 2 PCa patients were divided into two groups, according afternoon blood samples drawn from same patients. to the results of each prognosis factor: 248 patients with PSA ≤ 10 ng/ml and 101 with PSA > 10 ng/ml; 208 with 4.5. Estrogen variation by age among healthy control GS ≤ 3 + 4, and 141 with GS ≥ 4 + 3; 199 in HS pT2, and 150 subjects in HS pT3-T4; and 233 patients with negative SM and 116 with positive SM. Estrogen levels and BMI values in each group We did not find any age variation in E2,E1, and E1S serum are reported in Table 4. Mean E1S levels were significantly levels in the 50–74-year range or in the full control population increased in patients with PSA > 10 ng/ml (3.05 nmol/L ± 1.92) (20–74 years), as reported in Fig. 1(a–c). compared with E1S levels in patients with PSA ≤ 10 ng/mL (2.60 nmol/L ± 1.55). E1S was found to be significantly higher 4.6. Comparison of mean serum estrogen concentrations in in patients in HS pT3-T4 (2.99 nmol/L ± 1.80) compared with the PCa and BPH patients and healthy control subjects patients in HS pT2 (2.58 nmol/L ± 1.58), and in patients with (Table 3) positive SM (3.26 nmol/L ± 1.95) compared with patients with negative SM (2.52 nmol/L ± 1.48). Although E1S was higher in We observed a significant increase in mean E1S and patients with a higher GS than in patients with a lower GS, this E1 levels (±S.D.) in PCa patients (2.82 nmol/L ± 1.78, difference was not significant (p = 0.1681). However, we found and 126.1 pmol/L ± 66.1, respectively), and in BPH patients a significant increase in E1S when GS increased from 3 + 3 to (2.78 nmol/L ± 2.12, and 127.8 pmol/L ± 56.4, respectively), 4+4(Fig. 2a). E1 levels were higher in patient groups with poor compared with healthy normal men (2.11 nmol/L ± 0.96, and prognosis factors, but were not significantly different among 113.8 pmol/L ± 47.6, respectively). patients with good prognosis factors, except for patients with Although the serum E2 level in PCa patients was lower than positive SM (p = 0.0026). in the control subjects, it was not significantly different among The mean E2 level was found to be significantly higher in the PCa, BPH, and control groups. patient groups with high GS and positive SM compared to In order to investigate E1S as possible metabolic source of patients with a low GS and negative SM. In addition, we found estrogen, we determined the correlation between E1S versus significantly increased serum E2 levels when the GS increased E2 and E1. Significant correlations between E1S versus E2 and p = 0.0023 (Fig. 2b). F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167 163

Fig. 1. (a–b) Regression curve variation of the three estrogens with age in the full control population.

4.8. Serum androgen levels in PCa and BPH (Table 3) E2 was significantly correlated with TT (r = 0.283) (Table 5, Fig. 3c). The mean TT serum level was significantly higher In PCa, whatever the prognosis factor, E1 was always sig- (15.77 nmol/L ± 6.45) in BPH patients than in PCa patients nificantly correlated with E1S. TT was also always significantly (13.31 nmol/L ± 5.83) (p = 0.0002), with a resulting signifi- correlated to E2 in PCa, but correlations were higher in groups of cantly higher E2/T ratio in PCa than in BPH. patients with PSA > 10 (r = 0.350), with GS ≥ 4+3(r = 0.366), The comparison of TT, BT, DHEA, A, and DHT in PCA and with HS pT3-T4 (r = 0.372) than in patient groups with patients according to PSA, GS, HS pT2/pT3-T4, and SM did PSA < 10, with GS ≤ 3 + 4, and with HS pT2. not reveal significant differences, except for BT, which was sig- TT was significantly correlated to E1 in groups of PCa nificantly higher in patients with PSA > 10 ng/mL, and for A, patients with a poor prognosis, except in those with positive which was significantly higher in positive SM. SM, and not correlated to E1 in those with good prognosis. A was significantly correlated to E1, except in the PSA > 10 ng/ml group, and not correlated to E2, except in the pT3-T4 group. 4.9. Correlations between androgens and estrogens in PCa patients according to prognosis factors (PSA, GS, HS, and SM) and in the BPH population (Table 5) 5. Discussion

We investigated correlations of the non-conjugated estrogens Assay methods were carefully validated [28,29]. Immunoas- E2 and E1 with steroids from which they could have origi- say results for E2 were compared with those obtained using the nated (E1S, TT, and A) within each PCa group according to GC/MS method, and those for E1S using LC/MS/MS [33].We stage of tumor evolution and in the BPH population. We looked determined E1 by GC/MS [34], which was not the case for some for these correlations in seeking possible differences in steroid previously published plasma E1 assays involving PCa patients metabolic pathways among the PCa patients, according whether [2–4,7,12,35]. their prognosis was poor or more favorable, as well as in the BPH We verified the absence of age-dependent variations in E2, patients. E1, and E1S serum levels in normal men between 20 and 74 We in fact found a significant correlation (p < 0.0001) years old and finding no statistically significant difference in between E2 versus E1S in groups of PCa patients with serum concentrations of these estrogens in 51 samples obtained PSA > 10 ng/mL (r = 0.266), with GS ≥ 4+3 (r = 0.284), with the same day between 8–10 am and 4–6 pm. HS pT3-T4 (r = 0.294), and with positive SM (r = 0.347). We For the first time in a rather consistent group of PCa patients, found no significant correlation between E2 versus E1S in PCa we report significantly higher E1S serum levels in PCa and patient groups with good prognosis (r = 0.107, 0.018, 0.013, and BPH patients than in age-matched, healthy control subjects 0.023) (Table 5, Fig. 3a and b). In BPH patients, there was [36]. There was no significant difference in the BMI of these no correlation whatsoever between E2 and E1S(r = 0.026), but three groups, nor did there appear to be significant differences

Fig. 2. (a and b) Regression curve variation of E1S and E2 levels with Gleason score. 164 F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167

Fig. 3. (a–c) Regression curve of the variation in E2 vs. E1S with Gleason score (≤3+4,≥4 + 3) in the PCa and BPH populations. in serum E1S levels between the PCa and BPH patients stud- breast carcinoma in rat studies [39]. Administration of E1S ied. using subcutaneous Silastic® implants in male rats led to sig- E1S has mainly been studied in breast cancer in women. It nificantly high serum levels of E1 and E2 (111–184 pmol/L) has been hypothesized that plasma E1S could serve as a major [40]. source of E2 in hormone-dependent breast cancer [37,38], and Serum E1S levels in adult men have been found to be 15–100- it has been shown that E1S stimulated the growth of induced fold the E2 levels [33,41], the same concentration range found in premenopausal women. Like in females, in males [42], this conjugated estrogen, which has a long half-life, may be a Table 5 major source of estrogens resulting from the hydrolysis of E1 Correlations between potential sources of estrogens (E1S, TT, A) vs. E2 and E1, by sulfatase followed by reduction to E , principally by 17␤- according to prognosis factors in a population of 349 PCa and 100 BPH patients 2 hydroxysteroid dehydrogenase type 1. Indeed, we previously E STTA 1 reported a significant correlation between serum E1S, E2, and E1 PSA ≤ 10 (n = 248), PSA > 10 (n = 101) levels [33] in 50 healthy men. Similarly, serum E1S and E1 levels E2 were significantly higher in PCa and in BPH than in controls. PSA ≤ 10 0.107 0.318* 0.014 However in the present study, E2 serum levels were found not PSA > 10 0.266* 0.350* 0.014 to be significantly different among PCa patients, BPH patients, E1 and control subjects. ≤ PSA 10 0.416* 0.064 0.426* Nakamura et al. [43] recently demonstrated the production of PSA > 10 0.402* 0.373* 0.177* E2 and E1 from E1S and A in several human prostate cell lines, GS ≤ 3+4(n = 208), GS ≥ 4+3(n = 141) including LNCaP, PC3, and DU145. In physiological concen- E2 trations of the substrates, 10-fold more E1 was produced from ≤ Gleason 3 + 4 0.018 0.265* 0.086 E1S than from A, indicating the presence of active sulfatase and ≥ Gleason 4 + 3 0.284* 0.366 * 0.086 aromatase in these cell lines. As has been reported [42,44] in E1 both human breast and prostate cancer, E2 and E1 deriving from Gleason ≤3 + 4 0.380* 0.076 0.315* the action of sulfatase on E1S tended to be higher than E2 and Gleason ≥4 + 3 0.416* 0.364* 0.371* E1 deriving from the action of aromatase on A. In immunohisto- HS pT2 (n = 199), HS pT3-pT4 (n = 150) chemistry studies, Nakamura et al. [43] also showed the presence E2 of steroid sulfatase in the vast majority of human prostate tissues pT2 0.013 0.261* 0.151* and in BPH tissue. Thus, in human PCa, steroid sulfatase must pT3-pT4 0.294* 0.372* 0.186* be considered to play an important role in estrogen production. E1 The higher serum E1S levels in men with prostate diseases than pT2 0.373* 0.041 0.276* in healthy males could be considered a substrate for sulfatase, pT3-pT4 0.468* 0.344* 0.461* which would lead to the production of higher estrogen levels in Negative SM (n = 233), positive SM (n = 116) cells, which could be deleterious for the prostate gland. E2 To our knowledge, only a few studies that include assays of Negative 0.023 0.324* 0.058 circulating E2 have been conducted on the role of estrogens in Positive 0.347* 0.321* 0.090 prostate cancer. Some authors have considered low levels of cir-

E1 culating E2 to represent an additional risk factor [5,8,9,17,19]. Negative 0.274* 0.170* 0.225* These results support the idea that estrogens protect against can- Positive 0.532* 0.017 0.448* cer via the inhibition of the growth of prostate epithelial cells. Others found no pattern of risk association with increasing lev- BPH E1STTA els of E2 [6,7,12]. However, these authors did not carry out E1S E2 0.026 0.283* 0.051 assay in the subjects studied. E1 0.479* 0.288* 0.466* The wide heterogeneity observed in tumor aggressiveness in * Significant. terms of the values of the PCa parameters studied (PSA, GS, HS, F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167 165 and SM) led us to partition the full population of PCa patients It is as if E1S were a source of E2 that is negligible in good- into groups with poor and better prognoses. We then studied the prognosis PCa, but that becomes significant in poor-prognosis serum hormone status of each patient so as to reveal possible PCa. repercussions on prostate intracrinology. The hypothesis of an increase in sulfatase activity during the Relationships between estrogen serum levels and PCa prog- evolution of good- to poor-prognosis PCa is doubtful, because nosis factors revealed significantly higher E1S levels in patient E1 produced directly from E1S by the action of sulfatase is groups with poor prognoses versus those with better ones well-correlated to E1S in both good- and poor-prognosis PCa. (Table 4). We found an increase in E1S when the GS increased It is more likely as though E1 was not reduced to E2 in good- (p = 0.0616) (Fig. 2a). prognosis PCa, but was easily reduced into E2 in poor-prognosis If E1S is a reserve pool for more active estrogens, then it PCa. Such a predominantly reductive action of several 17␤- is not surprising that we found an increase in E1 and E2 in hydroxysteroid dehydrogenases, which convert E1 into E2, has groups of patients with poorer prognoses. This increase attained been reported in transformed LNCaP [49]. significance (p < 0.05) for serum E2 levels between the patient In BPH, the absence of correlation between E1S versus E2 groups with GS ≥ 4 + 3 and GS ≤ 3 + 4. In addition, as for E1S, (r = 0.026) and between A versus E2 (r = 0.051), while good we found a regular increase in E2 serum levels when the GS correlations between E1S versus E1 (r = 0.479) and between A increased from 3 + 3 to 4 + 4, and this increase was significant versus E1 (r = 0.466) are found, suggesting that reduction of E1 (p = 0.0023) (Fig. 2b). We thus show that in PCa patients with into E2 is impaired. Like in good-prognosis PCa, the hypothesis more aggressive tumors, serum estrogen levels, particularly E1S, of a predominantly oxidative pathway in BPH, leading to E1 and E2, increase further. formation – to the detriment of E2 – may be advanced. With respect to androgens (TT, BT, DHEA, A, and DHT), we did not find significant concentration differences among the 6. Conclusion PCa prognosis groups, except for higher BT levels in the group with PSA > 10 ng/ml and higher A in the group with positive We have shown higher concentrations of serum estrogens SM. in a population of PCa patients, especially of E1S in advanced Mean serum TT was significantly higher (p = 0.0002) in PCa, compared with good-prognosis PCa. In advanced PCa, E1S the population of BPH than in the full PCa population is easily converted into E and could be considered a possible ± ± 2 (15.77 6.45 nmol/L versus 13.31 5.83 nmol/L), and E2/TT marker for tumor aggressiveness in prostate cancer evolution. was significantly higher in PCa patients. Thus, although E1S However, it would be of interest to confirm these new findings levels were not statistically different in these two pathologies, among a still larger population of patients and control subjects. the consequences of this similarity are probably not the same. In contrast, in BPH, high serum E1S levels are not readily The two major pathways considered to be involved in pro- converted into E2. viding estrogens to human tissues are aromatization of the It would also be of interest to confirm these hypothesized androgens TT and A by aromatase [45–47] into E2 and E1 and estrogen metabolic pathways by measuring estrogen concen- conversion of E1S into E1 by a sulfatase [48–50]. Moreover, trations, enzyme activities and enzyme expression (aromatase, E1 and E2 are in reversible equilibrium, due to the presence of sulfatase, 17-hydroxy steroid dehydrogenases, particularly) in various 17␤-hydroxysteroid dehydrogenases [51–54]. the prostate tissues of PCa and of BPH patients. Seeking variation in the source of E2 in BPH and PCa (Table 5), we calculated correlations between E versus E S 2 1 Acknowledgements and TT in these pathologies. These results imply that E1Swas an important source of E (r = 0.266–0.347) in the group of PCa 2 We thank the “Association pour la recherche´ sur les patients, unlike the BPH group (r = 0.026), in whom TT appeared tumeurs prostatiques” (ARTP), Dr. Noah Hardy for editing the to be a greater source of E (r = 0.283). 2 manuscript, Dr. Alain Belanger (Molecular Endocrinology Lab- Indeed, the correlation between E S versus E in PCa was 1 2 oratory, CHUL Sainte Foy, Quebec, Canada) for comparison dependent on the state of advancement of the PCa, since in good- with LC/MS/MS, and ReneB´ erub´ e´ for his invaluable technical prognosis PCa, no correlation was found between E S versus 1 expertise. E2, like in BPH, whereas E2 was correlated to E1S in advanced PCa. Thus, although high serum E1S levels were found not to be statistically different in BPH and PCa, the metabolic pathways References leading to E2 do not appear to be similar in these two pathologies. [1] F. Labrie, A. Dupont, J. Simard, V. Luu-The, A. Belanger, Intracrinology. Thus, according to PCa prognosis factors, we observed a The basis for the rational design of endocrine therapy at all stages of prostate major change in the steroid metabolic pathway, since in good- cancer, Eur. Urol. 24 (Suppl. 2) (1993) 94–105. prognosis PCa, low serum estrogen levels are associated with an [2] W. Bartsch, H.-J. Horst, H. Becker, G. Nehse, Sex hormone binding glob- absence of correlation between E1S versus E2 (r = 0.013–0.107), ulin binding capacity, testosterone, 5alpha-dihydrotestosterone, oestradiol whereas in advanced PCa, higher serum estrogen levels were and prolactin in plasma of patients with prostatic carcinoma under var- ious types of hormonal treatment, Acta Endocrinol. 85 (3) (1977) 650– associated with a very significant correlation between E1S versus 664. E2 (r = 0.266–0.347). Therefore E1S becomes a good substrate [3] S. Rannikko, A.-L. Kairento, A.-L. Karonen, H. Adlercreutz, Hormonal for E2 only in advanced PCa. pattern in prostatic cancer. I. Correlation with local extent of tumour, pres- 166 F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167

ence of metastases and grade of differentiation, Acta Endocrinol. 98 (4) dorsolateral prostates of Noble rats, J. Natl. Cancer Inst. 80 (13) (1988) (1981) 625–633. 1045–1053. [4] S. Rannikko, H. Adlercreutz, Plasma estradiol, free testosterone, sex [22] I. Leav, F.B. Merk, P.W. Kwan, S.M. Ho, Androgen-supported estrogen- hormone binding globulin binding capacity, and prolactin in benign enhanced epithelial proliferation in the prostates of intact Noble rats, prostatic hyperplasia and prostatic cancer, Prostate 4 (3) (1983) 223– Prostate 15 (1) (1989) 23–40. 239. [23] M.C. Bosland, H. Ford, L. Horton, Induction at high incidence of duc- [5] R. Haapiainen, S. Rannikko, H. Adlercreutz, O. Alfthan, Correlation of tal prostate adenocarcinomas in NBL/Cr and Sprague-Dawley Hsd:SD pretreatment plasma levels of estradiol and sex-hormone-binding globulin- rats treated with a combination of testosterone and estradiol-17 beta or binding capacity with clinical stage and survival of patients with prostatic diethylstilbestrol, Carcinogenesis 16 (6) (1995) 1311–1317. cancer, Prostate 8 (2) (1986) 127–137. [24] M.C. Bosland, Sex steroids and prostate carcinogenesis: integrated, multi- [6] E. Barrett-Connor, C. Garland, J.B. McPhillips, K.T. Khaw, D.L. Wingard, factorial working hypothesis, Ann. NY Acad. Sci. 1089 (2006) 168–176. A prospective, population-based study of androstenedione, estrogens, and [25] S. Raghow, M.Z. Hooshdaran, S. Katiyar, M.S. Steiner, Toremifene pre- prostatic cancer, Cancer Res. 50 (1) (1990) 169–173. vents prostate cancer in the transgenic adenocarcinoma of mouse prostate [7] A.W. Hsing, G.W. Comstock, Serological precursors of cancer: serum model, Cancer Res. 62 (5) (2002) 1370–1376. hormones and risk of subsequent prostate cancer, Cancer Epidemiol. [26] E.L. Cavalieri, P. Devanesan, M.C. Bosland, A.F. Badawi, E.G. Rogan, Biomarkers Prev. 2 (1) (1993) 27–32. Catechol estrogen metabolites and conjugates in different regions of the [8] P.H. Gann, C.H. Hennekens, J. Ma, C. Longcope, M.J. Stampfer, Prospec- prostate of Noble rats treated with 4-hydroxyestradiol: implications for tive study of sex hormone levels and risk of prostate cancer, J. Natl. Cancer. estrogen-induced initiation of prostate cancer, Carcinogenesis 23 (2) (2002) Inst. 88 (16) (1996) 1118–1126. 329–333. [9] A. Nomura, L.K. Heilbrun, G.N. Stemmermann, H.L. Judd, Prediagnostic [27] Y.Markushin, N. Gaikwad, H. Zhang, P.Kapke, E.G. Rogan, E.L. Cavalieri, serum hormones and the risk of prostate cancer, Cancer Res. 48 (12) (1988) B.J. Trock, C. Pavlovich, R. Jankowiak, Potential biomarker for early risk 3515–3517. assessment of prostate cancer, Prostate 66 (14) (2006) 1565–1571. [10] H.A. Guess, G.D. Friedman, M.C. Sadler, F.Z. Stanczyk, J.H. Vogelman, [28] J. Fiet, F. Giton, F. Ibrahim, A. Valleix, H. Galons, J.-P. Raynaud, Devel- J. Imperato-McGinley, R.A. Lobo, N. Orentreich, 5 Alpha-reductase activ- opment of a highly sensitive and specific new testosterone time-resolved ity and prostate cancer: a case-control study using stored sera, Cancer fluoroimmunoassay in human serum, Steroids 69 (7) (2004) 461–471. Epidemiol. Biomarkers Prev. 6 (1) (1997) 21–24. [29] J. Fiet, F. Giton, A. Boudi, P.Boudou, H. Soliman, J.-M. Villette, H. Galons, [11] L.J. Vatten, G. Ursin, R.K. Ross, F.Z. Stanczyk, R.A. Lobo, S. Harvei, E. Development of a sensitive and specific new plasma 4-androsten-3,17- Jellum, Androgens in serum and the risk of prostate cancer: a nested case- dione time-resolved fluroimmunoassay (TR-FIA), Steroids 66 (8) (2001) control study from the Janus serum bank in Norway, Cancer Epidemiol. 609–614. Biomarkers Prev. 6 (11) (1997) 967–969. [30] J. Fiet, R. Morville, D. Chemama, J.-M. Villette, B. Gourmel, J.-L. Brerault, [12] J.F. Dorgan, D. Albanes, J. Virtamo, O.P. Heinonen, D.W. Chandler, M. C. Dreux, Plasma androgen and gonadotrophin levels in normal adult men Galmarini, L.M. McShane, M.J. Barrett, J. Tangrea, P.R. Taylor, Relation- after percutaneous administration of 5␣-dihydrotestosterone, Int. J. Androl. ships of serum androgens and estrogens to prostate cancer risk: results from (5) (1982) 586–594. a prospective study in Finland, Cancer Epidemiol. Biomarkers Prev. 7 (12) [31] R. Tremblay, J.Y. Dube, Plasma concentrations of free and non-TeBG (1998) 1069–1074. bound testosterone in women on oral contraception, Contraception 10 (6) [13] R. Heikkila, K. Aho, M. Heliovaara, M. Hakama, J. Marniemi, A. Reuna- (1974) 599–605. nen, P. Knekt, Serum testosterone and sex hormone-binding globulin [32] F. Giton, J. Fiet, J. Guechot,´ F. Ibrahim, F. Bronsard, D. Chopin, J.-P. concentrations and the risk of prostate carcinoma: a longitudinal study, Raynaud, Serum bioavailable testosterone: assayed or calculated? Clin. Cancer 86 (2) (1999) 312–315. Chem. 52 (3) (2006) 474–481. [14] B.A. Mohr, H.A. Feldman, L.A. Kalish, C. Longcope, J.B. McKinlay, Are [33] F. Giton, A. Valleix, P. Boudou, J.-M. Villette, A. Belanger, H. Galons, J. serum hormones associated with the risk of prostate cancer? Prospective Fiet, Specific radioimmunoassay of estrone sulfate. Application to mea- results from the Massachusetts Male Aging Study, Urology 57 (5) (2001) surement in male plasma, J. Steroid Biochem. Mol. Biol. 81 (1) (2002) 930–935. 85–94. [15] H.B. Carter, J.D. Pearson, E.J. Metter, D.W. Chan, R. Andres, J.L. Fozard, [34] F. Labrie, A. Belanger,´ P. Belanger,´ R. Berub´ e,´ C. Martel, L. Cusan, J. W. Rosner, P.C. Walsh, Longitudinal evaluation of serum androgen levels Gomez, B. Candas, I. Castiel, V.Chaussade, C. Deloche, J. Leclaire, Andro- in men with and without prostate cancer, Prostate 27 (1) (1995) 25–31. gen glucuronides, instead of testosterone, as the new markers of androgenic [16] P. Stattin, S. Lumme, L. Tenkanen, H. Alfthan, E. Jellum, G. Hallmans, activity in women, J. Steroid Biochem. Mol. Biol. 99 (4–5) (2006) 82– S. Thoresen, T. Hakulinen, T. Luostarinen, M. Lehtinen, J. Dillner, U.H. 188. Stenman, M. Hakama, High levels of circulating testosterone are not asso- [35] C. Nagata, N. Takatsuka, H. Shimizu, H. Hayashi, T. Akamatsu, K. Murase, ciated with increased prostate cancer risk: a pooled prospective study, Int. Effect of soymilk consumption on serum estrogen and androgen concentra- J. Cancer 108 (3) (2004) 418–424. tions in Japanese men, Cancer Epidemiol. Biomarkers Prev. 10 (3) (2001) [17] C. Chen, N.S. Weiss, F.Z. Stanczyk, S.K. Lewis, D. DiTommaso, R. 179–184. Etzioni, M.J. Barnett, G.E. Goodman, Endogenous sex hormones and [36] F. Giton, A. De la Taille, Y. Allory, J. Tichet, C. Born, C.C. Abbou, J.-P. prostate cancer risk: a case-control study nested within the Carotene and Raynaud, Plasma E1S and hormone status in prostate adenocarcinoma, in: Retinol Efficacy Trial, Cancer Epidemiol. Biomarkers Prev. 12 (12) (2003) Proceedings of the 17th International Symposium of the Journal of Steroid 1410–1416. Biochemistry and Molecular Biology, Seefeld, Austria, 31 May–3 June, [18] J.K. Parsons, H.B. Carter, E.A. Platz, E.J. Wright, P. Landis, E.J. Metter, 2006. Serum testosterone and the risk of prostate cancer: potential implications [37] F. Vignon, M. Terqui, B. Westley, D. Derocq, H. Rochefort, Effects of for testosterone therapy, Cancer Epidemiol. Biomarkers Prev. 14 (9) (2005) plasma estrogen sulfates in mammary cancer cells, Endocrinology 106 (4) 2257–2260. (1980) 1079–1086. [19] G. Severi, H.A. Morris, R.J. MacInnis, D.R. English, W.Tilley, J.L. Hopper, [38] J.R. Pasqualini, C. Gelly, B.L. Nguyen, C. Vella, Importance of estrogen P. Boyle, G.G. Giles, Circulating steroid hormones and the risk of prostate sulfates in breast cancer, J. Steroid Biochem. 34 (1–6) (1989) 155–163. cancer, Cancer Epidemiol. Biomarkers Prev. 15 (1) (2006) 86–91. [39] S.J. Santner, M.C. Levin, R.J. Santen, Estrone sulfate stimulates growth [20] S.M. Ho, Estrogens and anti-estrogens: key mediators of prostate carcino- of nitrosomethylurea-induced breast carcinoma in vivo in the rat, Int. J. genesis and new therapeutic candidates, J. Cell. Biochem. 91 (3) (2004) Cancer 46 (1) (1990) 73–78. 491–503. [40] F. Giton, Memoire´ pour l’obtention du diplomeˆ de l’Ecole Pratique des [21] I. Leav, S.M. Ho, P. Ofner, F.B. Merk, P.W. Kwan, D. Damassa, Biochem- Hautes Etudes, Paris, Œstrogenes` et Prostate: evaluation´ de l’Inter´ etˆ du ical alterations in sex hormone-induced hyperplasia and dysplasia of the sulfate d’œstrone en pathologie prostatique, September 21, 2004. F. Giton et al. / Journal of Steroid Biochemistry & Molecular Biology 109 (2008) 158–167 167

[41] O. Prost, Y. Ottignon, A. Remy-Martin,´ D. Vuitton, J.-P. Miguet, G.L. [49] M.J. Reed, A. Purohit, L.W. Woo, S.P. Newman, B.V. Potter, Steroid sul- Adessi, Steroid sulfatase activities in normal and cirrhotic livers and plasma fatase: molecular biology, regulation and inhibition, Endocr. Rev. 26 (2) levels of estrone sulfate, estrone and estradiol-17 beta in men, Steroids 43 (2005) 171–202. (2) (1984) 189–199. [50] J.R. Pasqualini, G. Chetrite, B.L. Nguyen, C. Maloche, L. Delalonde, [42] J.R. Pasqualini, G. Chetrite, C. Blacker, M.C. Feinstein, L. Delalonde, M. Talbi, M.C. Feinstein, C. Blacker, J. Botella, J. Paris, Estrone M. Talbi, C. Maloche, Concentrations of estrone, estradiol, and estrone sulfate-sulfatase and 17 beta-hydroxysteroid dehydrogenase activities: a sulfate and evaluation of sulfatase and aromatase activities in pre- and hypothesis for their role in the evolution of human breast cancer from postmenopausal breast cancer patients, J. Clin. Endocrinol. Metab. 81 (4) hormone-dependence to hormone-independence, J. Steroid Biochem. Mol. (1996) 1460–1464. Biol. 53 (1–6) (1995) 407–412. [43] Y.Nakamura, T. Suzuki, T. Fukuda, A. Ito, M. Endo, T. Moriya, Y.Arai, H. [51] P. Soronen, M. Laiti, S. Torn,¨ P. Hark¨ onen,¨ L. Patrikainen, Y.Li, A. Pulkka, Sasano, Steroid sulfatase and estrogen sulfotransferase in human prostate R. Kurkela, A. Herrala, H. Kaija, V. Isomaa, P. Vihko, Sex steroid hormone cancer, Prostate 66 (9) (2006) 1005–1012. metabolism and prostate cancer, J. Steroid Biochem. Mol. Biol. 92 (4) [44] S.J. Santner, P.D. Feil, R.J. Santen, In situ estrogen production via the (2004) 281–286. estrone sulfatase pathway in breast tumors: relative importance versus the [52] P. Vihko, A. Herrala, P. Hark¨ onen,¨ V. Isomaa, H. Kaija, R. Kurkela, Y. Li, aromatase pathway, J. Clin. Endocrinol. Metab. 59 (1) (1984) 29–33. L. Patrikainen, A. Pulkka, P. Soronen, S. Torn,¨ Enzymes as modulators in [45] G.P. Risbridger, J.J. Bianco, S.J. Ellem, S.J. McPherson, Estrogens and malignant transformation, J. Steroid Biochem. Mol. Biol. 93 (2–5) (2005) prostate cancer, Endocr. Relat. Cancer 10 (2) (2003) 187–191. 277–283. [46] A. Purohit, L.J. Duncan, D.Y. Wang, N.G. Coldham, M.W. Ghilchik, M.J. [53] P. Vihko, A. Herrala, P. Hark¨ onen,¨ V. Isomaa, H. Kaija, R. Kurkela, A. Reed, Paracrine control of estrogen production in breast cancer, Endocr. Pulkka, Control of cell proliferation by steroids: the role of 17HSDs, Mol. Relat. Cancer 4 (1997) 323–330. Cell Endocrinol. 248 (1–2) (2006) 141–148. [47] A. Purohit, S.P. Newman, M.J. Reed, The role of cytokines in regulating [54] J.R. Pasqualini, G. Chetrite,´ Activity, regulation and expression of sulfatase, estrogen synthesis: implications for the etiology of breast cancer, Breast sulfotransferase, and 17␤-hydroxysteroid dehydrogenase in breast can- Cancer Res. 4 (2) (2002) 65–69. cer, in: R. Pasqualini, B.S. Katzenellenbogen (Eds.), Hormone-dependent [48] A. Purohit, O. Chapman, L. Duncan, M.J. Reed, Modulation of estrone Cancer, Marcel Dekker, Inc., New York/Basel/Hong Kong, 1996, sulphatase activity in breast cell lines by growth factors, J. Steroid Biochem. pp. 25–80. Mol. Biol. 41 (3–8) (1992) 563–566.