Gonadal androgen suppression (castration via orchiectomy or gonadotropin-releasing hormone analogues) suppresses circulating testosterone levels but does not achieve adequate androgen ablation within the prostate cancer microenvironment because it does not address adrenal and intratumoral steroid contributions. These residual extragonadal sources of androgens allow prostate cancer cells to survive, adapt, and evolve into castration-resistant prostate cancer (CRPC). The persistent significance of the androgen receptor (AR) axis in CRPC was recently validated by the clinical efficacy of androgen synthesis inhibitors (abiraterone) and novel, second-generation AR antagonists (enzalutamide). The appreciation that conventional therapeutic approaches achieve a suboptimal ablation of intratumoral androgens and AR axis signaling output opens transformative therapeutic opportunities. A treatment paradigm of comprehensive AR axis targeting at multiple levels (androgen synthesis, metabolism, and action) and at all relevant sites (gonadal, adrenal, intratumoral) simultaneously at the time of initiation of endocrine therapy (instead of the current approach of sequentially adding one agent at a time and only after disease progression) deserves examination in clinical trials to explore whether maximal first-line AR axis suppression via combination therapy can achieve maximal induction of cancer cell apoptosis (before they have the chance to adapt and evolve into CRPC) and thus, improve patient outcomes. Cancer Res; 73(15); 4599–605. ©2013 AACR.
The androgen receptor (AR) is a transcription factor that plays critical roles in prostate adenocarcinoma pathophysiology (1). Gonadal androgen suppression [surgical or chemical castration via orchiectomy or gonadotropin-releasing hormone (GnRH) analogues, respectively] is an effective systemic treatment for advanced prostate cancer in use for the past 7 decades (2). However, it is not curative, as, almost universally, resistant disease eventually emerges, which had been described in the past as “hormone-refractory” and “androgen-independent” prostate cancer (1). These terms have now been confirmed to be inaccurate, as AR and its transcriptional output most frequently remain expressed and critically important in prostate cancer cells even in this state (1), leading to the adoption of the more appropriate term, “castration-resistant” prostate cancer (CRPC). The latter indicates a clinical state in which, despite suppressed circulating testosterone levels (<50 ng/dL), the AR axis has been reactivated because of a plethora of signaling mechanisms that operate within the prostate cancer cell and its local milieu (1). This revised view of the persistent role of AR signaling in CRPC led to the development of novel therapeutic agents such as the androgen synthesis inhibitor abiraterone (3) and the second-generation AR antagonist enzalutamide (MDV3100; ref. 4), and, in turn, was validated by their clinical efficacy.
Mechanisms Promoting Persistent AR Axis Output in CRPC and Clinical Validation of Its Significance
Several mechanisms that promote persistent AR axis activation in CRPC cells, despite castrate levels of peripheral testosterone, have been reported (Fig. 1), including the following:
Persistence of intratumoral androgens via in situ synthesis and metabolism (refs. 5–11; and discussed in more detail below);
AR overexpression (frequently due to AR gene amplification) and missense mutations in the AR ligand-binding domain (LBD) that sensitize the receptor to even low androgen concentrations and/or broaden ligand specificity, leading to promiscuous interactions with alternative ligands (e.g., progesterone or the first-generation AR antagonists, flutamide, bicalutamide, and nilutamide, that can be converted into agonists under these conditions; refs. 1, 12–16);
Expression of AR variants that lack the LBD and can signal in a ligand-independent, constitutively active manner (17–18);
Stoichiometric and qualitative changes in the coregulatory components of the AR complex, including the AR coactivators and corepressors that modulate the transcriptional response. For example, all three steroid receptor coactivators (SRC) of the p160 family (SRC-1, SRC-2, and SRC-3) have been reported to be overexpressed in prostate cancer and this is linked to inferior clinical outcomes (1, 19–21). In particular for SRC-2, encoded by NCOA2, gene amplifications and point mutations have been detected in prostate cancer and are associated with increased AR transcriptional activity (22); and
Activation of the AR complex via cross-talk with other signaling pathways, such as HER2, insulin-like growth factor-1 receptor, Src, and Akt pathways that phosphorylate AR or its coactivators (1, 21, 23–26).
These mechanisms are not mutually exclusive, and several of them may operate simultaneously in CRPC cells and synergistically enhance AR transcriptional output (1).
Collectively, these findings support the role of the AR axis as an important therapeutic target in CRPC and led to the development of novel second-line endocrine therapies for prostate cancer. The CYP17 enzymatic inhibitor abiraterone prolonged median overall survival in men with chemotherapy-refractory CRPC by 3.9 months (3) and also showed clinical activity in chemotherapy-naïve CRPC (27), thus validating the importance of residual intratumoral androgens in CRPC pathophysiology. The second-generation AR antagonist enzalutamide (MDV3100; ref. 4), which was rationally designed to overcome the antagonist-to-agonist conversion of first-generation AR antagonists (28), prolonged median overall survival in men with chemotherapy-refractory CRPC by 4.8 months (4) and is currently being tested in chemotherapy-naïve CRPC.
In Situ Steroid Synthesis and Metabolism in CRPC
Several groups have reported that, compared with primary prostate cancers or normal prostate tissue, CRPC exhibits increased expression of enzymes involved in androgen synthesis and substantial tissue androgen levels that should be sufficient to stimulate AR (and may be even higher than levels present within primary prostate cancers from untreated eugonadal men; refs. 5–11). Collectively, these findings raise the hypothesis that, despite peripheral castration, prostate cancer cells may never (yet) encounter a completely androgen-free local milieu, and that the term “androgen deprivation therapy” that is frequently used to describe first-line therapy with GnRH analogues or orchiectomy may actually be a misnomer (at least at the level of the tumor microenvironment). The clinical efficacy of the CYP17 inhibitor abiraterone validates this hypothesis.
In a recent comprehensive integrated oncogenomic analysis of primary and metastatic prostate cancer specimens, we documented aberrant expression of enzymes involved in androgen synthesis and metabolism. As a group, metastatic prostate cancers expressed higher average transcript levels for AR and several steroidogenic enzymes, including SRD5A1, SRD5A3, and AKR1C3, whereas expression of SRD5A2, CYP3A4, CYP3A5, and CYP3A7 was decreased (11). The latter three enzymes are involved in phase I of dihydrotestosterone (DHT) inactivation (oxidation), and their decreased expression is predicted to increase in situ androgen levels and enhance AR activation. More importantly, we detected high intertumor variability of expression of individual enzyme transcripts in primary and metastatic prostate cancers, raising the hypothesis that within each individual tumor, enhanced activation of the androgen axis may occur at the level of a different enzyme, but with a common end result, i.e., increased tissue androgen levels and stimulation of AR (11).
These data thus highlight the androgen synthesis pathway en block as a mechanism of CRPC resistance to androgen deprivation therapy and can help explain the significant variability between lists of upregulated enzymes found in prior studies (5, 9, 29). One related question has been whether CRPC metastatic sites express the complete panel of enzymes necessary for de novo steroidogenesis using cholesterol as a precursor. Although it is generally accepted that prostate cancers can locally convert adrenal precursors to more active androgens (testosterone and DHT) via their own AKR1C3 and 5α-reductase (SRD5A1; 29), it has been proposed that they lack adequate CYP17A1 expression, and as a result, they remain dependent on the contribution of adrenal precursors (as a two-site, “adrenal-prostate cancer” steroidogenic unit; ref. 29). In our study, we found that this is the case for the majority, but not all, prostate cancers. A small subset of prostate cancers overexpresses CYP11A1, CYP17A1, 3β-hydroxysteroid dehydrogenase (HSD3B1), HSD3B2, and STAR (which are necessary for the conversion of cholesterol to androstenedione), and thus may be self-sufficient for de novo steroidogenesis (11).
Mechanisms of Dysregulated Androgen Metabolism in Prostate Cancer: Does Acute Adaptation Precede (and Allow for) Clonal Selection?
The mechanism(s) underlying this aberrant expression in CRPC of transcripts involved in androgen metabolism remain(s) to be fully characterized. Obviously, the full elucidation of these mechanisms could reveal novel therapeutic targets for inhibition of the AR axis in CRPC. In our study of integrated gene expression and comparative genomic hybridization datasets, these aberrant expression patterns were only rarely associated with respective copy-number alterations (CNA; ref. 11). On the contrary, AR overexpression was, in agreement with previous studies (13), frequently associated with AR gene amplification, an event that possibly would require a process of clonal selection. In the absence of frequent CNAs, we examined whether the dysregulation of androgen metabolism enzymes occurs at the mRNA level (11). Expression of several enzyme transcripts, in particular of the AKR1C family, is induced by androgen deprivation (11, 30) within a timeframe (24–48 hours) that is too fast for clonal selection. Conversely, androgen treatment can suppress the expression of steroidogenic enzymes. The finding that androgen deprivation rapidly upregulates the mRNA levels of AKR1C3, an enzyme that can convert androstenedione to testosterone, raises the hypothesis that androgen deprivation triggers an acute adaptation feedback loop that enhances the ability of the prostate cancer cell to metabolize adrenal precursors into testosterone and DHT, thus sustaining tissue androgen levels and AR stimulation.
This hypothesis is also consistent with studies of neoadjuvant medical castration therapy in men with localized prostate cancer that exhibited suboptimal suppression of intratumoral androgens (by only about 70%–80%, in contrast with the >90% concomitant reduction in serum androgens) and AR-dependent gene expression (31–32), suboptimal induction of prostate cancer apoptosis (33), and disappointingly low rates of pathologic complete response (<3%; ref. 34). Once again, the term “androgen-deprivation therapy” may be misleading, as it overestimates the impact of the systemic treatment on androgen levels and the AR axis within the tumor microenvironment (5–10). Instead, a noncommittal term such as “medical castration therapy” would be more appropriate.
Combination-Based Endocrine Therapy for Prostate Cancer: Finally Moving Beyond Proof-of-Concept
The appreciation of the importance of extragonadal contributions to AR signaling in prostate cancer is not recent. Second- and third-line hormonal manipulations have previously been used in CRPC, yielding small successes that provided proof-of-principle. Efforts to suppress adrenal steroids by surgical adrenalectomy and hypophysectomy date back 5 decades with anecdotal responses reported. Glucocorticoids, via suppression of adrenocorticotropic hormone and adrenal androgen synthesis, also have documented activity in CRPC with reported prostate-specific antigen (PSA) responses as high as approximately 60% in small phase II studies. Chemical adrenalectomy with aminoglutethimide or ketoconazole has provided PSA responses, but a survival benefit was never formally shown. The concept of adding an antiandrogen to gonadal suppression for “combined androgen blockade” or “maximal androgen blockade” has been proposed before (35). However, clinical trials conducted in the pre-abiraterone/pre-enzalutamide era had failed to show a consistent, clinically meaningful benefit from the addition of a first-generation antiandrogen (or any other second-line hormonal agent) to gonadal suppression (a meta-analysis of 27 randomized trials including a total of 8,275 patients revealed, at best, a nonsignificant gain of 1.8% in 5-year survival; ref. 36). Similarly, despite a very strong preclinical rationale, a use for 5α-reductase inhibitors has not yet been found in CRPC (although the combination of dutasteride with ketoconazole and hydrocortisone provided promising PSA responses in a recent study; ref. 37).
So why did abiraterone and enzalutamide achieve a survival benefit, whereas earlier second-line hormonal approaches did not? This could be because of more effective targeting of androgen synthesis and AR LBD, respectively, with fewer side effects. Abiraterone is a more specific and better-tolerated inhibitor of steroidogenesis than aminoglutethimide and ketoconazole. Enzalutamide seems to lack the AR agonistic effects of first-generation antiandrogens. With these advanced therapeutic agents in our armamentarium now, and with a galvanized interest in the AR axis as a therapeutic target in prostate cancer, the next steps will be to optimize their use for maximal therapeutic benefit. For example, the best timing, setting, and sequence for these novel agents to maximize clinical benefit and minimize the risk of cross-resistance remain to be defined.
The “Androgenic Set Point” of Prostate Cancer Cells and Maintenance of Cell Survival
In studies of neoadjuvant medical castration therapy, the probability of early biochemical recurrence was inversely correlated with the degree of pathologic effect (38) and positively correlated with the residual expression of AR-dependent genes, including PSA, identified as early as 3 months after the initiation of hormonal therapy (39). Therefore, failure to adequately suppress the AR axis (or early reactivation of AR signaling) is associated with inferior clinical outcomes.
Short-term (24–48 hours) exposure of prostate cancer cells to androgen-depleted medium in vitro stimulates the expression of steroidogenic enzymes (11, 30), AR itself (30, 40), and its coactivators (19, 41). In the case of the AR gene, it has been found that agonist-bound AR protein negatively regulates gene transcription via recruitment of the histone demethylase and transcriptional corepressor lysine-specific demethylase 1 (LSD1) to a highly conserved site in the second AR gene intron (30). Similar LSD1-dependent mechanisms have been reported for the rapid androgen-mediated downregulation of the steroidogenic enzyme AKR1C3 by agonist-bound AR (30). Collectively, these data suggest that agonist-bound AR directly mediates a physiologic intracellular feedback loop to negatively regulate AR axis activity. Conversely, androgen withdrawal is proposed to trigger an acute adaptive response as a preprogrammed attempt of the prostate cancer cells to retain AR transcriptional activity, restoring it toward a predetermined “androgenic set point,” which is critical for their survival. According to this hypothesis, these adaptive cellular responses would allow a sizeable pool of prostate cancer cells to maintain adequate intratumoral androgen levels and AR axis activity and survive, despite peripheral castrate androgen levels, eventually leading to the emergence of castration-resistant disease (possibly after later acquiring clonal genetic lesions, such as AR gene amplification).
Therefore, as current approaches achieve a suboptimal inhibition of AR signaling output, a more comprehensive AR axis targeting at multiple levels (androgen synthesis, metabolism, and action) and at all relevant sites (gonadal, adrenal, intratumoral) simultaneously at the time of initiation of endocrine therapy, deserves examination in clinical trials to explore whether it can improve patient outcomes over the current treatment paradigm of sequentially adding one agent at a time and only after disease progression.
Personal Opinion: What Will the Landscape of Advanced Prostate Cancer Treatment Be in the Next 5 Years?
The road map to comprehensive AR axis targeting
The above concepts lay the framework for new directions in the treatment of advanced prostate cancer. Both CYP17 inhibitors and second-generation AR antagonists are being tested in pre-CRPC disease states (hormone-naïve advanced prostate cancer and in the neoadjuvant setting), and it is reasonable to anticipate that they will be active there as well, by enhancing the efficacy of conventional medical castration therapy. In addition, the consecutive use of these agents, including identifying the optimal sequencing approach to augment clinical benefit and minimize the risk of cross-resistance, remains to be established by clinical evidence (so far, the phase III clinical trials of both abiraterone and enzalutamide have excluded patients previously treated with the other agent).
It is my personal opinion that the most promising approach would be the early application of combination systemic therapy at the initiation of medical castration rather than after the onset of CRPC. Combinations of both classes of agents with GnRH analogs are being explored in clinical trials, and I am optimistic that as part of a frontline comprehensive AR axis-targeting approach, they could move us closer to our goal of a completely androgen-free prostate cancer microenvironment. Further toward that goal, additional steroidogenic enzymes are being explored as therapeutic targets: inhibitors of AKR1C3 (42–45), HSD3B1 and HSD3B2, and SRD5A1 would also be interesting choices to be included in future clinical trials of multidrug combination regimens aiming at maximal frontline inhibition of the AR axis to augment prostate cancer cell apoptosis and deplete the pool of surviving prostate cancer cells that can later accumulate additional genetic events and resurge as CRPC.
Additional future directions
Several other CYP17 inhibitors (TAK-700/orteronel) and AR antagonists (ARN-509) are in clinical development. Galeterone (TOK-001 or VN/124-1) has been reported to be a CYP17 inhibitor, an AR antagonist, and also to promote AR degradation. Additional exciting opportunities have emerged in the field of combining AR axis–targeting agents with inhibitors of other oncogenic signaling pathways. For example, recent data suggest that AR inhibition derepresses phosphoinositide 3-kinase/Akt signaling (46), thus providing a strong rationale for combinations of inhibitors of both pathways.
Emerging technical developments can also address present needs in this field: Accurate, standardized, inexpensive assays with a reasonably rapid turnaround time for measurement of intratumoral steroids and AR signaling output (e.g., in biopsy material) will help select patients and guide therapy with these novel agents. Furthermore, interpatient variations in intracrine steroid signaling can be evaluated by noninvasive, real-time monitoring of the expression of AR (full-length and alternatively spliced constitutively active variants) and steroidogenic enzymes in circulating tumor cells (CTC; ref. 11) and serve as potential basis for individualized therapy.
The advances in the field of AR targeting in prostate cancer have opened new opportunities and, simultaneously, generated new questions. Despite their documented clinical activity, neither CYP17 inhibitors (abiraterone) nor second-generation AR antagonists (enzalutamide) are curative as single agents (at least in the clinical states in which they have been tested so far), and refractory disease eventually develops. Although the mechanisms of resistance (de novo or acquired) to these novel agents are not yet fully studied, it is of critical significance that these refractory tumors usually continue to produce PSA (at least to some degree). Because PSA (KLK3) gene expression requires AR transcriptional activity, this preliminary observation (which needs to be supported by more comprehensive gene expression profiling analyses in the near future) suggests that the AR axis remains active (at least to some degree) even after treatment with these novel AR-targeting agents. In the case of abiraterone, preclinical models of resistance have suggested that abiraterone treatment upregulates the expression of AR (both full-length and splice variants; ref. 47) and several steroidogenic enzymes, including CYP17A1 (abiraterone's own target) and AKR1C3 (47–48), and promotes in situ accumulation of pregnenolone and progesterone (that are synthesized upstream of CYP17; ref. 48). Pregnenolone and progesterone are putative AR agonists (especially in the case of AR harboring certain LBD mutations; ref. 48). These putative mechanisms of resistance to CYP17 inhibitors, which await confirmation in clinical specimens, suggest that despite treatment with these agents, some intratumoral steroids persist and may be available to drive AR activity. This lends further rationale to combinations of CYP17 inhibitors with other enzymatic inhibitors (e.g., of AKR1C3) or with potent second-generation AR antagonists (e.g., enzalutamide). The constitutively active AR splice variants have been proposed as mechanisms of resistance to both abiraterone (47) and enzalutamide (49) and will need to be addressed in the future with a new treatment paradigm.
Other important questions in this field include the following:
What are the mechanisms underlying the feedback loops that lead to the increased expression of steroidogenic enzymes in CRPC? What other transcription factors, beyond AR, control this process, and can they possibly serve as therapeutic targets in the future?
What is the role of the bone microenvironment in in situ steroidogenesis in prostate cancer metastasis? Steroidogenic enzyme expression has been reported in the epithelial compartment (carcinoma cells), via immunohistochemistry in tissues (1) and quantitative real-time PCR in CTCs (11), implying capacity for cell-autonomous steroidogenesis and intracrine signaling. However, steroidogenic enzyme expression has also been reported in the stroma, suggesting that (at least some) enzymatic reactions may occur outside the prostate cancer cell, in a two-compartment model involving paracrine signaling. Moreover, can the bone microenvironment promote steroidogenesis in prostate cancer cells at the metastatic site? Conversely, what is the impact of high local tissue androgen levels on bone remodeling in prostate cancer bone metastasis?
What is the role of other steroids (e.g., progesterone, mineralocorticoids, glucocorticoids) and their nuclear receptors (PR, MR, GR, respectively) in CRPC?
Although generally the importance of the overexpression of steroidogenic enzymes for CRPC is attributed to increased androgen synthesis, what other, nonendocrine roles may these enzymes play? For example, several of these enzymes may have roles in drug detoxification, prostaglandin metabolism, N-linked glycosylation, etc.
On the basis of the clinical activity of CYP17 inhibitors and enzalutamide, AR is now a validated therapeutic target in CRPC. Still, however, current treatment approaches do not achieve complete inhibition of AR signaling in prostate cancer because they do not address all aspects of this axis (androgen synthesis, metabolism, and action) and all relevant sites (gonadal, adrenal, intratumoral) early enough in the course of hormonal therapy. Failure to adequately and comprehensively inhibit the AR axis may allow prostate cancer cells to survive, adapt, and evolve into CRPC. There is a need and opportunity to explore in clinical trials a new paradigm in the management of advanced prostate cancer: first-line regimens based on combinations of new-generation inhibitors of all these AR axis components.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
The author acknowledges the joint participation by Adrienne Helis Malvin Medical Research Foundation through its direct engagement in the continuous active conduct of medical research in conjunction with Baylor College of Medicine.
This work was also supported by the Prostate Cancer Foundation, the Conquer Cancer Foundation of the American Society of Clinical Oncology Young Investigator and Career Development Awards and a Pilot/Feasibility Program of the Diabetes & Endocrinology Research Center (P30-DK079638) at Baylor College of Medicine. N. Mitsiades is a Dan L. Duncan Scholar, a Caroline Wiess Law Scholar, and a member of the Dan L. Duncan Cancer Center (supported by the NCI Cancer Center Support Grant P30CA125123) and the Center for Drug Discovery at Baylor College of Medicine.