Early effects of pharmacological androgen deprivation in human prostate cancer

Authors


Eugene D. Kwon, Departments of Urology/Immunology, Mayo Clinic, 200 First Street SW, Guggenheim 4–11 A, Rochester, Minnesota 55905, USA. e-mail: kwon.eugene@mayo.edu

Abstract

OBJECTIVES

To assess the early histological effects of pharmacological androgen deprivation (AD), which have been assessed only over longer periods, as surgical castration leads rapidly to diminished cell proliferation and enhanced cell death within the prostate.

PATIENTS AND METHODS

With Institutional Review Board approval, 35 patients were randomly assigned (seven in each group) to receive 0, 7, 14, 21 and 28 days of AD (flutamide, 250 mg orally three times/day, and one injection with leuprolide acetate 7.5 mg) before radical prostatectomy. The surgical specimens were assessed by conventional histology and immunohistochemistry, while macroarray analysis and quantitative real-time polymerase chain reaction (QRT-PCR) were used to examine gene expression.

RESULTS

There were morphological changes within the prostatic tissues as early as 7 days after initiating AD, similar to the response to castration. There was tumour cell vacuolization indicating cellular injury, glandular atrophy and mononuclear cell infiltration as prominent and progressive effects but, by contrast with castration studies, there were no changes in epithelial proliferation or apoptosis. Macroarray analysis, validated by QRT-PCR and immunohistochemistry, showed up-regulation of numerous and potentially counter-effective genes involved in the cell cycle and apoptosis.

CONCLUSIONS

Pharmacological AD induces significant involution within prostatic tissues over 7–28 days, but allows the persistence of some viable tumour cells capable of proliferation.

Abbreviations
AD

androgen deprivation

RP

radical prostatectomy

H&E

haematoxylin and eosin

HPF

high-power field

TUNEL

terminal dUTP nick-end labelling

LCM

laser capture microdissection

QRT

quantitative real-time (PCR).

INTRODUCTION

The dependence of untreated prostatic adenocarcinoma on male sex hormones underpins the use of androgen deprivation (AD) as a first-line treatment for metastatic prostate cancer [1]. However, despite an initial response in most patients, AD typically has a short-term palliative effect, and ultimately fails because of the emergence of androgen-resistant disease [2]. When administered in conjunction with radiotherapy [3] or as an adjuvant to surgical resection of advanced (pT3 or pN+) prostate cancer [4], AD appears to improve outcomes. However, in the neoadjuvant setting before radical prostatectomy (RP), AD leads to a reduction in cancer volume but fails to delay biochemical recurrence or extend disease-specific survival [5,6].

Characterizing the effects of AD on prostate cancer might help to explain these limitations of AD therapy, and perhaps suggest ways to improve its application. Although there are extensive reports relating to the effects of AD on normal and malignant prostatic tissue in animal models [7–12], there are relatively few human studies. Most of the limited data in humans are derived from studies of neoadjuvant AD by ≥ 2–3 months of pharmacological therapy before RP. The histomorphological effects on the prostate of AD in this setting have been well characterized [13–15], but the effects of AD on epithelial proliferation and cellular death, presumed to underlie the observed changes, have been examined in few studies, with conflicting findings [14–17].

In a recent study it was shown that surgical castration in humans leads rapidly to the enhancement of apoptosis and suppression of proliferation in the benign and malignant prostatic epithelium, but with a return to baseline levels after 7–10 days [18]. However, by contrast with surgical castration, which precipitously removes circulating androgen, pharmacological AD uses alternative mechanisms over longer intervals [19], and thus the time course of effects on prostate tissues might differ from those induced by castration. The aim of the present study was to document the histological and molecular changes within the prostate after 7–28 days of medical AD.

PATIENTS AND METHODS

With Institutional Review Board approval, 35 patients aged <70 years with biopsy confirmed clinical stage T1–2b prostate adenocarcinoma and no previous hormonal therapy were recruited for study at the Loyola University Medical Center, Maywood, IL, over a 1-year period. With informed consent, the patients were randomly assigned (seven in each group) to receive 0, 7, 14, 21 and 28 days of AD (consisting of flutamide, 250 mg orally three times/day, and one injection with leuprolide acetate 7.5 mg, administered with the first dose of flutamide) immediately before RP. Patients were monitored weekly for compliance. Two patients (both in the 21-day group) were subsequently excluded due to intervening medical conditions that precluded surgery. Serum PSA levels were measured (Hybritech Tandem-R, Beckman Coulter, Fullerton, CA, USA) before starting AD and before RP on the day of surgery.

Prostatectomy specimens were obtained from the operating room, transported on ice and sliced perpendicular to the long axis of the gland at 3-mm intervals. Alternate slices were fixed in 10% buffered formalin, embedded in paraffin wax, sectioned (5 µm) and stained with haematoxylin and eosin (H&E). Two pathologists (E.M.W. and E.F.C.) and one urologist (R.G.M.) evaluated the specific histological changes while unaware of the sample origin. For each specimen, 10 separate high-power (× 400) fields (HPF) of 15–20 slides were examined and scored.

Benign tissues were scored for glandular atrophy, epithelial cell vacuolization and basal cell hyperplasia as 0%, 25%, 50% or 75% of cells or acini affected. Apoptotic bodies were scored as 0 (≤10/HPF in all HPFs examined) 1 (≥10/HPF in ≤ 25% of HPFs), 2 (≥10/HPF in 25–50% of HPFs) and 3 (≥10/HPF in 50–100% of HPFs). Mononuclear inflammatory cell infiltration was scored as 0 (no foci/HPF), 1 (one focus/HPF), 2 (multiple foci/HPF) and 3 (complete infiltration). In areas of prostate cancer, vacuolization of tumour cells, loss of nucleoli, apoptotic bodies and mononuclear cell infiltration were scored as for benign tissues.

Remaining prostate tissues were immediately placed in tissue freezing medium (Tissue-Tek® OCT Compound, VWR, West Chester, PA, USA) and snap-frozen in liquid nitrogen. Prostate cryosections (5 µm, − 20 °C) were mounted on Superfrost Plus slides (Fisher Scientific, Springfield, MA, USA) and stained for apoptotic cells by terminal dUTP nick-end labelling (TUNEL), according to the manufacturer’s protocol (ApopTag kit, Chemicon International, Temecula, CA, USA). Ten separate HPFs were scored from each specimen for the proportion of TUNEL-positive cells as 100 × the number of TUNEL-positive cells/the total number of tumour or epithelial cells. Cytospin preparations of Jurkat cells, cultured with or without 0.02% anti-Fas antibody, were used as positive and negative controls, respectively.

Laser capture microdissection (LCM) was used according to the manufacturer’s protocol (Arcturus PixCell II LCM system, Arcturus Engineering, Mountain View, CA, USA), with guidance from the study pathologists, to separately isolate benign and malignant glands. Ten 5 µm thick cryostat H&E-stained prostate tissue sections were dissected from each of six patients undergoing AD and six untreated controls. RNA was extracted from LCM-captured cells using the RNeasy Mini kit (Qiagen, Valencia, CA, USA). RNA was subsequently amplified by 1000–100 000 times using high-fidelity mRNA amplification, as previously described [20]. RNA purity was confirmed by a 260/280-nm spectrophotometric ratio of ≥1.9.

The macroarray filters (Human 1 Membrane RiboScreen, Pharmingen BD Biosciences, San Diego, CA, USA) used to assess the effects of AD on gene expression in prostatic tissues consisted of 288 cDNA test oligonucleotides in duplicate and 192 control oligonucleotides. The test elements included a variety of cell-cycle regulators, angiogenic factors, apoptosis regulators, cytokines, chemokines and receptors (listing available). cDNA macroarrays were labelled and hybridized according to manufacturer’s protocols. Macroarrays were subsequently exposed by phosphor-imaging and scanned using a Cyclone Phosphor Imager (600-dpi), with the OptiQuant software (Packard Instrument Co, Inc., Meriden, CT, USA) used for digital image acquisition and analysis

Quantitative real-time (QRT)-PCR was used for selected genes of interest to confirm the findings of the macroarray analysis. Using the Beacon Designer software (Bio-Rad, Hercules, CA, USA), the primer and probe sequences (available on request) for specific amplification and detection of the targets (p53, c-IAP1 and TRAF5) as well as the reference gene (nuclear DNA-encoded 28S rRNA) were selected. Probes were labelled at the 5′ end with the fluorescent reporter dye 6′carboxyfluorescein and at the 3′ end with the quencher dye Black Hole Quencher-1. The single-stranded oligonucleotides were synthesized, purified using PAGE and quantified spectrophotometrically (Integrated DNA Technologies, Inc., Coralville, IA, USA).

Duplicate samples for each gene were analysed using the iCycler (Bio-Rad). Standard curves were created by performing QRT-PCR on 102−108 copies of DNA templates encoding the genes of interest. The standards correspond directly to the amplicon for each gene, being 88 nucleotides long for p53, 83 for c-IAP1, 76 for TRAF5 and 84 for 28S (sequences available on request). The copy number for each gene was estimated by comparing the observed cycle threshold to the standard curve, and subsequently normalized to the 28S copy number.

Immunohistochemical staining was carried out using 5 µm cryosections for Ki67 (as a marker of proliferation), tissue transglutaminase (as a marker of apoptosis), p53, c-IAP1 and TRAF5 (to confirm macroarray and QRT-PCR findings). Sections were blocked for 30 min using 3% horse serum (VectaStain, ABC method; Vector Laboratories, Burlingame, CA, USA) and 0.1% BSA in PBS before being incubated for 1 h with monoclonal mouse anti-human antibody against tissue transglutaminase (TGase II, NeoMarkers, Freemont, CA, USA), Ki67 (Vector Laboratories, Newcastle, UK), p53, c-IAP1, TRAF5 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) or irrelevant isotype-matched control antibody (Southern Biotechnology, Birmingham, AB, USA). After washing, sections were incubated for 30 min with corresponding biotin-conjugated secondary antibody, washed, and then incubated for another 30 min with horseradish peroxidase-streptavidin complex. After a final wash, sections were developed in aminoethylcarbazole (Sigma, St. Louis, MO, USA) and counterstained with haematoxylin. Slides were then scored as previously detailed [21].

Histological scores for prostate tissues are summarized as the mean (sem) percentage; scores for prostate tissues treated for 7, 14, 21 or 28 days with AD were each compared with untreated (0-day) control prostates using paired t-tests. A two-way anova was used to compare the gene expression levels among prostate tissue specimens, with treatment (AD vs none) and tissue type (cancer vs benign) as the two factors, and including an interaction term. A Wilcoxon rank-sum test was used to determine whether there was a significant difference between treated and untreated tissue. No corrections were made for multiple comparisons, given the exploratory nature of the macroarray experiments. For all statistical tests, P < 0.05 was considered to indicate statistical significance.

RESULTS

Serum PSA levels declined rapidly after AD, by ≈ 40% by 7 days of treatment and 75% by 28 days, relative to pretreatment values (P < 0.05 for trend). Histological changes in both benign and malignant tissue were evident as early as 7 days after AD (Table 1). Thus, benign prostatic tissue had glandular atrophy, basal cell hyperplasia and mononuclear cell infiltration in response to AD (Fig. 1A,B). These effects were more pronounced with a longer duration of AD. Similarly, malignant cells showed increasing vacuolization and loss of nucleoli, indicating cellular injury, as well as mononuclear cell infiltration, indicating an anti-tumour immune response (Fig. 1C,D).

Table 1. Histological effects of AD on benign and malignant prostatic tissues
SampleDuration of AD, days
07142128
  • *

    P < 0.05 vs 0 days.

Mean (sem) %
Benign tissues
 Glandular atrophy13 (2.6)30 (3.8)*40 (1.9)*50 (4.9)*65 (6.0)*
 Basal cell hyperplasia15 (41.5)30 (3.8)*48 (3.8)*50 (8.0)*65 (1.9)*
 Mononuclear cell infiltration 0.6 (0.15) 1.4 (0.1)* 1.8 (0.1)* 2.0 (0.2)* 1.6 (0.2)*
 Epithelial vacuolization30 (6.8)10 (3.8)20 (6.8)10 (4.5)10 (4.5)
 Apoptotic bodies 0.2 (0.1) 1.4 (0.2)* 1.4 (0.2)* 1.2 (0.1)* 0.8 (0.1)*
 TUNEL-positive cells 0.39 (0.16) 0.21 (0.09) 0.23 (0.68) 0.11 (0.02) 0.41 (0.12)
Tumours
 Cell vacuolization 0 (0)10 (2.3)*43 (5.7)*65 (2.7)*70 (3.8)*
 Loss of nucleoli 8 (1.9)10 (4.5)40 (3.0)*30 (4.5)*40 (3.8)*
 Mononuclear cell infiltration 0.3 (0.1) 1.2 (0.1)* 1.2 (0.1)* 2.2 (0.2)* 1.7 (0.1)*
 Apoptotic bodies 0.8 (0.1) 1.6 (0.1)* 1.4 (0.2) 1.6 (0.2)* 1.5 (0.2)
 TUNEL-positive cells 0.34 (0.11) 0.29 (0.14) 0.43 (0.21) 0.17 (0.07) 0.15 (0.02)
Figure 1.

Histological changes in the prostate after 7–28 days of AD: Benign tissues show (A) extensive basal cell hyperplasia and glandular atrophy (H&E × 100) and (B) mononuclear cell infiltration and epithelial cell vacuolization (H&E × 400). Tumours showed (C) marked tumour cell vacuolization and loss of nucleoli (H&E × 600) and (D) profuse mononuclear cell infiltration (H&E × 100). Immunostaining for the proliferation marker Ki67 detected (E) low levels of expression in glandular tissue, but (F) prominent staining in stromal cells (×400). TUNEL staining showed few apoptotic cells in (G) tumour or (H) benign glands; (I) Jurkat cells treated with anti-Fas antibody is shown as a positive control (×400) Tissue transglutaminase staining supported the TUNEL results, in that there was no detectable apoptosis in (J) tumour or (K) benign glands; (L) breast cancer is shown as a positive control (×400).

Proliferation, determined by Ki67 immunostaining, was low in benign and malignant prostatic tissues (Fig. 1E), but not substantially different with or without AD. Conversely, Ki67 expression was apparent in stromal cells of all 15 treated prostates (Fig. 1F), compared to one of seven untreated prostates. As we previously reported [21], infiltrating mononuclear cells also expressed high levels of Ki67. More apoptotic bodies were apparent in benign and malignant epithelium after AD, but did not increase over time. However, both TUNEL and tissue transglutaminase immunostaining failed to show significant levels of apoptosis (Fig. 1G–L).

On average, 100–200 cells per specimen, representing about 30 ng of total RNA, were recovered using LCM (Fig. 2). Using T7-based linear RNA amplification, a mean of 3 µg of RNA was typically generated for use in macroarray analyses. Comparison of macroarray gene expression profiles generated from treated and untreated specimens showed small but significant differences in the levels of 10% (30/288) of the genes examined (Tables 2 and3) in benign and malignant prostatic epithelium. Seven (2%) of these genes (Table 2) showed changes in expression that depended on tissue type examined, with the expression of FGF2, FGF11, interferon-γ and orphan R WPFGNT increased in tumours but unchanged in benign glands, interleukin-3Rα increased in tumours but decreased in benign glands, and c-fos and Ltn increased in benign glands but decreased in tumours.

Figure 2.

Isolation of benign and tumour tissues using LCM. Top photomicrographs (A and B) show a phenotypical benign gland. Bottom photomicrographs (C and D) show prostate tumour cells. From left to right A and C show pre-captured, and B and D show captured, cell compartments (×600).

Table 2. Genes showing dissimilar changes in benign vs tumour tissues in response to AD
GeneBenignTumourP
No ADADMeanNo ADADMean
  • *

    P < 0.05 between treatment groups.

Fold change
FGF20.820.801.00.641.241.9*0.006
FGF110.840.790.90.711.301.8*0.037
Interferon-γ1.141.121.00.781.231.6*0.010
Orphan R WPFGNT1.060.990.90.811.101.4*0.016
c-fos0.921.311.4*1.180.960.8*0.019
Ltn0.991.221.2*1.00.760.7*0.018
Interleukin-3Rα1.120.930.8*0.871.121.3*0.010
Table 3. Genes showing similar changes in benign and tumour tissues in response to AD
GeneNo ADADMeanP
Fold change
TRAF50.871.742.00.028
TRIP0.931.611.70.046
Granzyme A0.911.431.60.019
c-IAP10.981.501.50.021
cas1.111.721.50.028
MCM30.711.091.50.040
FASTK1.161.641.40.021
Nip10.841.161.40.030
cdk30.971.251.30.014
p1070.921.161.30.018
bak0.791.041.30.030
p570.740.941.30.035
cyclin G21.081.311.20.012
p191.081.331.20.046
caspase7/Mch30.780.971.20.015
p530.971.191.20.017
TNFα0.811.011.20.024
cyclin B0.831.041.20.028
H963:PAF-R1.081.291.20.038
GPR9-60.740.911.20.017
caspase 10a0.901.081.20.049
Interferon-γ Rα1.011.151.10.049
TNFRp551.101.161.10.030

The expression of another 23 genes (Table 3) increased in both benign glands and tumours in response to AD. Among these were several apoptosis-related genes, including pro-apoptotic p53, Bak and caspase7 and anti-apoptotic c-IAP1 and TRAF5. As p53, c-IAP1 and TRAF5 are well-characterized genes, whose presence has been described in different type of tumours, they were further analysed by QRT-PCR and immunohistochemistry to validate the macroarray findings. QRT-PCR confirmed increased transcription of p53 (6.5-fold), c-IAP1 (4-fold) and TRAF5 (3.5-fold) after AD (Fig. 3A–C), as suggested by the macroarray experiments. Immunostaining showed increased expression of p53, c-IAP1 and TRAF5, localized to prostatic epithelial and tumour cells (Fig. 3D–I), after AD. The mean (sem) percentage of p53, c-IAP1 and TRAF5-positive cells increased from 10 (3.3), 12 (2.4) and 6 (2.0) in untreated specimens to 24 (5.3), 24 (3.3) and 16 (2.4) in treated specimens, respectively.

Figure 3.

Prostate tissues from AD-treated patients show increased p53, c-IAP1 and TRAF5 gene and protein expression. Quantification by QRT-PCR of the expression of (A) p53, (B) c-IAP1, and (C) TRAF-5 in LCM-isolated tumour tissue from six treated and six untreated patients. Values are the mean of duplicate QRT-PCR, which are shown as error bars. Photomicrographs of representative specimens show that, compared to untreated patients (D-F), those undergoing AD (G-I) have more prominent immunostaining for p53 (top), c-IAP1 (middle) and TRAF5 (bottom).

DISCUSSION

This is the first study to assess the effects of combined medical AD on the human prostate over 7–28 days, and shows that characteristic histological changes and a reduction in serum PSA level are evident as early as 7 days, and progress further over time. The observed histological effects included atrophy and involution of both benign and malignant acini, with evidence of cellular injury such as cytoplasmic vacuolization, nuclear condensation and surrounding stromal mononuclear cell infiltration. Cells in the prostatic stroma and the basal layer of benign glands became hyperplastic after AD. These findings are similar to those previously described a few days after surgical castration [18] and those after longer periods of neoadjuvant pharmacological AD before RP [13].

However, by contrast with previous reports of early changes in response to surgical castration [18,22,23], apoptosis and proliferation were unchanged in the present study, suggesting that cellular injury was the predominant mechanism underlying prostatic involution. The few patients in each group might have limited the power of this study to detect a significant change in rates of apoptosis and proliferation. However, the lack of even a statistically insignificant trend in these data is striking and somewhat unexpected, based on previous published studies. One possible explanation for these discrepancies might relate to the highly disparate nature of the cancers being treated. Thus, the present patients all had disease localized to the prostate, the biology of which is quite different from the predominantly advanced or metastatic disease treated by surgical castration [18,22,23].

Conversely, previous studies of neoadjuvant AD before RP, in which the pathological spectrum is similar to that in the present study, have consistently shown suppression of tumour cell proliferation, but after ≥ 3 months of AD [14,16,17]. It is plausible that the effects of AD on the proliferation of localized prostate cancer require longer than the 4-week period assessed here. There is also some evidence that tumours can vary in the extent to which AD inhibits their proliferation, and that this might be of prognostic significance [24,25]. Certainly the persistence of a population of proliferating prostate cancer cells might be one mechanism whereby androgen resistance can emerge [2].

Similarly, the apoptotic response of prostate cancers after 3 months of pharmacological AD appears to be heterogeneous [15,17,26]. This is in keeping with findings after surgical castration [23,27,28], as well as xenograft studies, in which the apoptotic effect of AD depends on the cell line being used [10–12]. Again, there appears to be some prognostic advantage if significant apoptotic cell death is induced within a tumour after AD [25,29].

The intracellular molecular changes that underlie the prostatic response to AD have not previously been well characterized. Our macroarray gene expression experiments, validated by selected QRT-PCR and immunostaining, showed that AD leads to significant changes in the expression of some genes involved in cell-cycle regulation and apoptosis, as well as growth factors, their receptors, and signal transducers. Notably, both pro-apoptotic and anti-apoptotic genes are up-regulated after 7–28 days of AD.

The balance of apoptotic and anti-apoptotic gene expression would clearly affect the rate of apoptosis, and the expression of protective anti-apoptotic genes, e.g. c-IAP1 and TRAF5, after AD might explain the lack of change in apoptotic activity. It is also likely that phenotypic alterations in response to AD develop over time, thus correlating with changing patterns of proliferation and apoptosis. For instance, TGF and its receptors are up-regulated immediately after surgical castration, and might be one factor leading to enhanced apoptosis [28], while in the present study, TGF expression (on macroarray analysis) was unchanged after 7–28 days of AD.

The present findings have important clinical implications related to various applications of AD in the treatment of prostate cancer. Thus, the efficacy of AD combined with radiotherapy [3] might result from a synergism between the two methods in terms of eliminating cancer cells [30]. Conversely, outcomes after surgical treatment appear to be unchanged by up to 3 months of neoadjuvant AD [6], possibly because a biologically significant fraction of tumour cells remains viable. Recent studies showed that prolonging the duration of AD beyond 3 months leads to a further reduction in prostate and tumour volume, and positive surgical margin rate [5,31], but a longer follow-up is required to see whether this translates into improvements in recurrence risk or survival. Of concern, the re-emergence of high proliferative activity in some tumours suggests the possibility that androgen resistance might already be developing after 6 months of AD [31], which might negate the efficacy of such an approach.

One limitation of the present study is that it did not examine events within the first week or beyond 4 weeks after the initiation of AD. The period of the study was selected on the basis that suppression of androgens in response to LHRH agonist occurs over 3–4 weeks [19]. However, the antiandrogen flutamide is rapidly absorbed after oral administration, and acts to competitively block androgens at the receptor level, leading rapidly to intraprostatic androgen exposure similar to or even lower than castrate levels [32]. As morphological changes after castration peak within days and return to baseline levels at about a week [7,18], some of the present effects of AD might have been missed before the earliest sample time examined.

In summary, we show that AD by combining LHRH analogue and antiandrogen leads rapidly to the involution of benign and malignant prostatic glands, coupled with stromal proliferation and mononuclear cell infiltration. Changes in epithelial proliferation or apoptosis are not apparent over 7–28 days, but might occur outside this period. Ongoing cellular injury and epithelial involution are evident over the period studied, but even after 28 days a significant proportion of malignant cells are viable and maintaining basal levels of proliferation. The molecular changes brought about by AD at an intracellular level appear to be complex, and further study is warranted.

ACKNOWLEDGEMENTS

The authors thank Drs Donald J. Tindall, John C Cheville, R. Houston Thompson and Gregory S. Schenk for their review of the manuscript.

CONFLICT OF INTEREST

None declared. Supported in part by a grant from the AFUD/AUA Research Scholar Program (M.M., Scholar), by the National Institutes of Health/National Cancer Institute Grant CA 82185 (E.D.K.), by the Department of Defense Grant PC 991568 (E.D.K.), by CaPCURE, and by the Mayo Foundation, Rochester.

Ancillary