Androgen Deprivation Boosts Prostatic Infiltration of Cytotoxic and Regulatory T Lymphocytes and Has No Effect on Disease-Free Survival in Prostate Cancer Patients

This article is featured in Highlights of This Issue, p. 1213

Because prostate cancer (PCa) is driven, in part, by androgens, hormone therapy has been used so far in its management (1). The value of neoadjuvant hormone therapy (NHT) prior to radical prostatectomy (RP), however, has recently been questioned by urologists and oncologists, because its effectiveness is uncertain. Its aim stems from its ability to shrink the tumor and reduce margin positivity so as to allow adequate surgical coverage of the cancerous area. Several studies have shown an improvement in clinical and pathologic endpoints, but not a constant improvement in overall survival (2). On the other hand, some experimental data suggest that androgen deprivation improves the efficacy of PCa immunotherapy (3, 4), and could thus be an attractive alternative to RP in selected patients. Identification of the immunologic pathways/mechanisms regulated by androgen withdrawal in PCa may help treatment decision making, and the development of modern, less aggressive therapeutic approaches to localized disease.

We have recently shown that the lack of constitutive interleukin-7 (IL-7) gene expression in PCa, and related lymphocyte depletion, is a mechanism whereby PCa evades immunosurveillance (5). This article assesses the possibility that androgen deprivation may restore this pathway and interfere with other aberrant immunologic mechanisms regulating PCa and draining lymph node microenvironments and also have an impact on patient outcome.

We collected biological samples (cancer and normal prostate samples, and draining lymph nodes), clinical data, and pathologic data of 126 patients treated by RP for PCa between 2002 and 2009 at the “S.S. Annunziata” Hospital, Chieti, Italy. Seventy-six had received preoperative NHT for 3 months consisting of an androgen receptor blocker (flutamide, 250 mg orally 3 times/day), and one i.m. injection of a gonadotropin-releasing hormone (GnRH) agonist (leuprolide acetate, 7.5 mg depot, administered with the first dose of flutamide). The other 50 patients were selected by matching for age, Gleason score (Gs), pathologic stage, and prostate specific antigen (PSA) level at diagnosis, as shown in Table 1. In addition, we obtained normal prostates (histologically negative for PCa or benign prostatic hyperplasia) from 12 untreated patients, ages 57 to 63, prostatectomized for bladder cancer (control patients) and pelvic lymph nodes (control lymph nodes) from autopsies of 5 men, ages 59 to 65, who died for reasons other than cancer and were histologically free from PCa.

Details of assessment of the clinicopathologic profiles and tissue sample processing are reported in Supplementary Methods.

Written informed consent was obtained from patients. The study was approved by the Hospital's Ethical Committee, and conducted in accordance with the principles outlined in the Declaration of Helsinki.

Single immunohistochemistry on formalin-fixed, paraffin-embedded samples or on frozen samples sections was performed as described in Supplementary Methods. Details of double and triple immunohistochemistry and double immunofluorescent stainings are also provided in Supplementary Methods.

Automated cell count was done by light microscopy using a Leica Imaging Workstation (Leica) by applying a dedicated algorithm in Qwin image analysis software (version 2.7). CD4⁺, CD8⁺, and Foxp3⁺/CD25⁺ cells were counted by adding together the intraepithelial and stromal positive cells scattered in randomly chosen fields for the normal prostate samples (of both control and PCa patients), and in fields randomly chosen within neoplastic areas for the PCa samples. Values are represented as the mean ± SD of positive cells/field on single or double immunostained, formalin fixed, paraffin-embedded sections at ×400 in an 85,431.59 μm² field. Eight to 12 high-power fields were examined for each section and 2 sections per sample were evaluated.

We used the P.A.L.M. Micro Beam System (P.A.L.M. Microlaser Technologies) for laser capture microdissection (LCM) of two 10-μm frozen sections from each normal (of both control and PCa patients) and neoplastic prostate sample to obtain its epithelial and stromal components (details in Supplementary Methods).

The real-time reverse transcriptase (RT)-PCR was carried out, on the RNA extracted from microdissected cells, using the MiniOpticon System (Bio-Rad) with SYBR Green fluorophore (details in Supplementary Methods).

Immune cell counts are reported as mean and SD. Between-group differences in immune cell count (in the prostate or lymph nodes), or the relative expression of cytokines/chemokines by real-time RT-PCR, were assessed by 1-way ANOVA. The difference between each pair of means was evaluated with the Tukey pairwise multiple comparisons test. Differences between groups of patients with different Gleason scores for the relative expression of cytokines/chemokines by real-time RT-PCR were assessed by ANOVA. The χ² test and the Mann–Whitney U probability test were used to examine the association between cytokine/chemokine expression levels, or cell counts in prostate samples, and the clinical and pathologic characteristics. The 3-year disease-free survival curves were constructed by the Kaplan–Meier method, and differences were analyzed with the log-rank test. The mean follow-up time was 43 months (range, 3–96 months). The Kruskal–Wallis test was used to determine whether there was a significant delay in disease recurrence between treated and untreated patients. The SPSS software, version 11.0 (SPSS Inc.) was employed, with P < 0.05 as the significance cutoff.

To find out whether androgen blockade affects the lack of IL-7 production in PCa, and thus the related intraprostatic lymphocyte depletion (5, 6), we first used LCM and real-time RT-PCR to determine IL-7 expression on cancer and normal prostate samples from untreated and NHT-treated patients.

As both the epithelium and the fibromuscular stroma account for prostate gland responsiveness to androgens (7), we isolated and analyzed both components from these samples.

Real-time RT-PCR corroborated our previous observation of a significant reduction (∼53 times) of IL-7 mRNA expression level in neoplastic versus normal prostatic epithelium (ref. 5; Fig. 1A). Both the epithelial and the stromal components of PCa samples from treated patients expressed IL-7 mRNA levels fully comparable to those in the untreated patients. By contrast, the histologically normal samples from treated patients displayed a considerable increase of IL-7 mRNA expression level in the stroma (∼6 times) compared with those from the untreated patients (Fig. 1A). In addition, immunohistochemical examination of subsequent serial sections showed that IL-7 protein production was distinct in the stroma and bright in the glandular epithelium of normal samples from untreated patients, and scanty in PCa samples from both untreated and treated patients (Fig. 1B, a–d and g, h). IL-7 was moderately to strongly produced in the stroma and glandular epithelium of normal samples from treated patients (Fig. 1B, e and f). Double immunofluorescence and confocal analyses of these samples revealed that increased IL-7 production in the stroma was mainly attributable to vimentin⁺ fibroblasts and desmin⁺ αsma⁺ smooth muscle cells forming normal prostatic stroma (Fig. 1C, a–i).

Because androgen blockade fostered IL-7 production in the stroma of normal tissue surrounding PCa from the treated patients, we next investigated whether the lymphocyte content of their prostates was altered.

CD4⁺ T cells dramatically increased in both normal (∼6 times) and neoplastic (∼4 times) prostate tissues from treated versus untreated patients, whereas the mean number of CD8⁺ cells significantly increased in neoplastic tissue following NHT (Table 2; Fig. 2A, a–h). Thus, the increased IL-7 production was accompanied by intraprostatic T lymphocyte accumulation (8, 9).

The expression of T-cell intracellular antigen-1 (TIA-1), which lies in the granules of cytotoxic lymphocytes (10), was more frequent and stronger in prostate samples from the treated patients, and colocalized with CD8⁺ cells and CD4⁺ cells to a lesser extent (Fig. 2B, a–d). Other molecules typically related to activated cytotoxic T effector (Teff) lymphocytes (11, 12), such as granzyme-B and perforin, were only detectable in prostate samples from treated patients and mostly colocalized with CD8⁺ cells (Fig. 2B, e–h). The high density of intraprostatic Teff lymphocytes promoted by androgen blockade prompted us to search for the production of specific Th1-type chemoattractants, such as monokines induced by IFN-γ (Mig)/CXCL9 and IFNγ-inducible protein (IP)-10/CXCL10, which have been associated with a significant Teff lymphocyte recruitment (13, 14). As shown in Figure 2C and D, the relative expressions of IP-10/CXCL10 mRNA, but not that of Mig/CXCL9, was significantly (∼7 times) increased following androgen depletion in the normal prostate epithelium, whereas no significant alterations occurred in the epithelium and stroma from PCa samples. In keeping with these data, immunohistochemistry showed a stronger IP-10/CXCL10 production in the normal glandular epithelium from the treated patients (Fig. 2E, a and b).

Interestingly, one subset of mostly CD4⁺ cells expressing surface CD25 and high levels of the transcription factor Forkhead box P3 (Foxp3), but lacking CD127 (as assessed by double immunostainings CD25/Foxp3 followed by single CD4 and CD127 staining of consecutive 3-μm serial sections), and thus identifiable as T regulatory (Treg) lymphocytes (15, 16), was also clearly increased in both normal and neoplastic prostate tissues from the treated patients (Table 2; Fig. 2A, i–l). Monitoring of the expression of the cell-cycle–associated marker Ki67 to assess whether this increase was induced by stimulation of their proliferation revealed that most of them were not dividing inside the prostate (data not shown), as observed for other prostate infiltrating lymphocyte subsets (5). We thus explored the possibility that selective chemoattractants released in the androgen-depleted prostate microenvironment were involved in Treg cell accumulation.

As most intraprostatic Treg specifically express the chemokine receptor CCR4 (Fig. 3A; ref. 17), they should be responsive to macrophage-derived chemokine (MDC)/CCL22, and to thymus and activation-regulated chemokine (TARC)/CCL17 (18).

LCM followed by real-time RT-PCR analyses of prostate samples from the treated patients showed that in both the epithelial and the stromal component of their normal and neoplastic tissues MDC/CCL22 mRNA expression was low to undetectable, and not dissimilar to that in the untreated patients (Fig. 3B).

TARC/CCL17 mRNA expression (Fig. 3C) was undetectable in both the normal samples and the PCa epithelium of the untreated patients, whereas it was found in the neoplastic stroma. In the treated group, it was basically unchanged in PCa samples, whereas it was clearly evident in both the epithelium and particularly the stroma of their normal samples.

Immunohistochemistry, too, showed that TARC/CCL17 production was absent in the normal samples of the untreated patients and present in their PCa stroma (Fig. 3D, a and b), where it colocalized with CD68⁺ monocytes/macrophages, and with less differentiated CD11b⁺CD33⁺ myeloid cells (Fig. 3E, a; ref. 19). TARC/CCL17 production was strong in the normal samples and scanty to moderate in the PCa samples from the treated patients (Fig. 3D, c and d). Double immunohistochemistry revealed that CD68⁺ monocytes/macrophages were the major contributor to the substantial TARC/CCL17 production found in the normal prostatic stroma of the treated patients (Fig. 3E, b).

No significant differences of TARC/CCL17, IP-10/CXCL10, or IL-7 expression emerged between groups of patients with different Gleason scores (for both treated and untreated). Data about cytokine or chemokine production in normal samples from the untreated patients were not significantly different to those obtained in normal samples from controls. No significant association was disclosed between cytokine or chemokine expression levels and the clinical and pathologic characteristics.

Because of its wide effects on the immune system (20), we next asked whether androgen blockade also regulates TARC/CCL17 production by macrophages homing to lymphoid tissues.

Immunohistochemistry was thus used to determine the production of TARC/CCL17 in pelvic lymph nodes removed during RP from both groups of patients.

The production of TARC/CCL17 (Fig. 3F, a–c), which colocalized with CD68⁺ monocytes/macrophages, and to a lesser extent, with CD1a⁺ dendritic cells (Fig. 3F, g and h), was well represented in lymph nodes from the treated patients (Fig. 3F, c), whereas it was clearly restricted to less number of immune cells in those from the untreated patients (Fig. 3F, b), and almost absent in the control nodes (Fig. 3F, a).

Lymph nodes from treated patients showed a distinct accumulation of Foxp3⁺ cells (42 ± 13; mean ± SD of positive cells/×400 field; Fig. 3F, f), the vast majority expressing CCR4 (Fig. 3F, i), and crowding TARC/CCL17 producing immune cell foci. This Foxp3⁺ cell density was not significantly different from that in lymph node from untreated patients (24 ± 9; Fig. 3F, e), although it was considerable higher (P < 0.05) than that in control lymph nodes (10 ± 6; Fig. 3F, d).

We asked whether somewhat opposite immunologic events, namely recruitment of both Teff and Treg cells, elicited by androgen depletion in the PCa microenvironment, influence patient outcome and have clinical significance. As most Teff were CD8⁺ lymphocytes, we first assessed the combined influence of low versus high number of intratumoral CD8⁺ and Treg on disease-free survival, and then used the χ² test to assess the association between NHT and the intratumoral content of these subsets.

Using the median values as cutoff, patients were divided into groups with low and high intratumoral CD8⁺, Treg, or CD8⁺/Treg ratio. Each group was represented by Kaplan–Meier biochemical disease-free survival curves constructed on the basis of patient's PSA failure during the first 3-year follow-up after surgery.

Patients with higher frequencies of intratumoral CD8⁺ showed a significant improvement of disease-free survival compared with patients with lower frequencies (Fig. 4A; median = 22.5, log-rank P = 0.002). By contrast, patients with high versus low frequency of Treg showed a reduced disease-free survival (Fig. 4B; median = 2.3, log-rank P = 0.019). As it has been reported that the balance between CD8⁺ and Treg has a greater impact on the outcome (21–23), than their number alone, we next performed survival curves of patients with high versus low CD8⁺/Treg cell ratio (Fig. 4C). Using the median value (10.65) as the cutoff, survival curves constructed with 3-year follow-up data revealed that the group with a high ratio—63 of 126 patients, 28 untreated (1 censored enclosed) and 35 treated (7 censored enclosed)—had a significant improvement of disease-free survival (log-rank P = 0.003) compared with the low ratio group—63 patients, 22 untreated (1 censored enclosed) and 41 treated (4 censored enclosed). Two patients in the first group displayed biochemical recurrence (the first occurred 12 months after surgery) compared with 13 in the second group (the first occurred 3 months after surgery). The χ² test showed a significant (P < 0.01) association between NHT and the high intratumoral frequency of both CD8⁺ (OR = 27.00; 95% CI, 9.36–77.92) and Treg (OR = 11.93; 95% CI, 4.95–28.74), and the absence of any association between NHT and the high or low intratumoral CD8⁺/Treg ratio (P = 0.27; OR = 0.67; 95% CI, 0.32–1.37).

Finally, though our cohort size is a limiting factor, we try to assess whether NHT had an impact on survival through mechanisms other than its own immunologic effects. We thus compared the disease-free survival of treated versus untreated patients (Fig. 4D). At the end of the study, 80% (40 of 50) of the untreated patients were still disease-free (8 subjects displayed recurrence, 2 were censored) compared with 76% (58 of 76) of the treated patients (7 subjects displayed recurrence, 11 were censored). Thus, treated patients had no significant decrease in the frequency of recurrence (log-rank P = 0.291), nor a significant delay in its occurrence (P = 0.344, as assessed by the Kruskal–Wallis test), in line with data from most of the previous clinical studies (1).

Androgen receptor (AR) signaling plays a crucial role in all steps of prostate carcinogenesis, as in normal prostate development and function (24), through both the epithelial and the mesenchyme/stroma compartments (7). These, in turn, cross-communicate and interact with prostate homing immune cells, and thus give rise to complex relationships that may be subverted by androgen ablation through the following pathways.

Histologically normal prostate tissue surviving NHT-dependent apoptosis (24), probably represented by AR⁻ basal and a mix of AR⁻ and AR⁺ basal-intermediate epithelial cells, thus seems to preserve important biological functions in (presumably indirect) response to androgen ablation, such as expression of IP-10/CXCL10 and TARC/CCL17 by the glandular epithelia and that of IL-7 by stromal fibroblasts and smooth muscle cells. By contrast, PCa tissue remnants of the same apoptotic event, mostly composed of AR⁻ cancer cells (24) equipped with fibroblasts/myofibroblasts (namely cancer-associated fibroblasts or CAF; ref. 25), are almost unresponsive.

Interestingly, monocyte/macrophages homing to normal prostate tissue and lymph nodes, not those infiltrating PCa, account for most of the TARC/CCL17 produced following androgen deprivation. Whether TARC/CCL17 production is a direct (26) or indirect macrophage response remains to be investigated. It is only clear that androgen deprivation is unable to overcome macrophage conditioning by PCa.

IL-7 is a nonredundant trophic factor for T lymphocyte development that primarily acts by promoting lymphocyte survival and/or proliferation (27), and also displays T lymphocyte–mediated antitumor properties (28, 29). It has been identified from different cell types (30, 31), including smooth muscle cells (31) and fibroblasts (32). Using prostatic fibroblasts and smooth muscle cells, regularly endowed with AR (24), androgen deprivation supplies the IL-7 lost in the prostate microenvironment at tumor onset (5), and likely offsets T-cell depletion in PCa, because both CD8⁺ and CD4⁺ lymphocytes are greatly increased in the prostate from treated patients.

IP-10/CXCL10 directs the trafficking of activated effector CD8⁺ and CD4⁺ T lymphocytes by binding to its high-affinity receptor CXCR3 (13, 14), and displays antitumor and antimetastatic properties (33, 34). Its expression may be induced from a variety of cells (35), including epithelial cells (35, 36). Here, we show that androgen deprivation boosts IP-10/CXCL10 gene expression in normal epithelium in association with intraprostatic infiltration of cytotoxic Teff lymphocytes endowed with antitumor activity (37).

TARC/CCL17 induces Treg cell recruitment by binding to the receptor CCR4 (17, 18). This chemokine is expressed by mature dendritic cells (18, 38), monocytes and activated macrophages (39), and epithelial cells (40, 41). This study reveals that androgen deprivation not only induces TARC/CCL17 gene expression in normal prostatic epithelium, but also boosts its expression by macrophages in normal prostate and lymph nodes, with a related influx of CCR4⁺Foxp3⁺ naturally occurring Treg cells (17, 42) able to suppress antitumor Teff lymphocytes (43, 44).

The inflammatory environment and associated immune cell recruitment promoted by androgen withdrawal is consistent with emerging evidence of the reciprocal negative cross-talk between AR and transcription factor NF-κB (45), which play a key role in the control of inflammatory and immune response genes (46), as also with the mutual antagonism documented between androgen and IFN signaling pathways (47).

Our data indicate that cytokine/chemokine gene upregulation following androgen deprivation only involves normal prostate, whereas lymphocyte recruitment also involves cancerous tissues. This may be because of the fact that PCa consists of different foci (48) mingled with histologically normal areas. Thus, the effects of androgen deprivation on normal tissue may easily reflect on the nearby cancer foci. They, indeed, are invaded by both cytotoxic-effector CD8⁺ and immune-suppressive Treg cell populations. Thus, their intratumoral balance remains substantially unchanged in samples from treated versus those of untreated patients. This “double face” of the NHT-dependent immunologic effects results in the lack of a significant clinical benefit, as revealed by the Kaplan–Meier disease-free survival curves, and the lack of association between NHT and a high CD8⁺/Treg ratio, which, as observed in other tumor types (21–23), is proving a good prognostic marker for PCa patients.

Taken as a whole, our findings provide mechanistic insights into some immunologic pathways elicited by NHT inside the prostate and lymph node microenvironments, and offer new tools with which to assess its potential in the management of PCa patients.

The full exploitation of the immunostimulatory effects exerted by androgen blockade, probably via IL-7 and IP-10/CXCL10 induction and associated intratumoral cytotoxic-effector lymphocyte infiltration, requires to counteract the concomitant immune-suppressive effects likely mediated by TARC/CCL17 induction and associated intratumoral Treg lymphocyte accumulation.

No potential conflicts of interest were disclosed.

This work was supported by a grant awarded by Fondazione Cassa di Risparmio della Provincia di Chieti (CariChieti), Italy.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.