Jack A. Schalken, Department of Experimental Urology, University Medical Center, Nijmegen, the Netherlands. e-mail: J.Schalken@uro.umcn.nl.
As prostate cancer is not a single disease, it is important to identify the pivotal pathway in the patient being treated. The molecular environment is the site of current oncological research to define new therapeutic targets for hormone-refractory disease, offering the potential to eventually individualize treatment through stratification of pathways. Targets may be validated either phenotypically (e.g. androgen receptor, cadherin) or functionally (e.g. prostate cancer-specific genes). In addition, several other candidates are potentially suitable, while others await discovery. Important initial steps have been made in the search for prostate cancer stem cells; identifying stem cells and the stromal, hormonal, and other signalling molecules that influence their behaviour would have important implications for managing prostate cancer. Although individual therapeutic pathways might be ineffective in a particular molecular environment, combinations of approaches might be capable of producing synergistic effects. A multimodal approach thus might be the best solution. Determining where best to search for a molecular target, and validating whether the target is associated with a sufficiently aggressive malignant process to justify further study is difficult, but the potential benefits are enormous.
The challenges posed by macroscopic, clinically advanced prostate cancer are no less daunting at the molecular level. Laden with biochemical pathways and complex cellular signalling, the molecular environment is the locale where current oncological research attempts to define new therapeutic targets for hormone-refractory disease.
Determining where best to search for a molecular target, and establishing whether the target is associated with a sufficiently aggressive malignant process to justify further study, pose difficulties that require a careful review of accumulating evidence and an occasional educated guess, but the potential benefits are enormous. Once identified and verified, and once their molecular ecology is elucidated, these sentinel targets of malignant behaviour will provide greater insight into the molecular mechanisms of prostate cancer, with important consequences for diagnosis, therapeutic index and tolerability of intervention strategies . Target validation, occupying the crucial region between characterization of a critical malignant process by the molecular oncologist and clinical trials in human subjects, is the key to the overall success of efforts to improve outcomes in prostate cancer .
Targets may be validated phenotypically, i.e. they are aberrantly expressed in the malignant cells, or functionally, by the extent to which there is a functional biological relation between the target and cancer cell survival or aggressiveness  (Table 1). By defining targets as comprehensively as possible, molecular medicine offers the tantalising possibility that patients may be characterized sufficiently by molecular profiling to support stratification by target profile, offering the possibility of individualized treatment [4,5]. At present, small molecules targeted to signalling and other pathways have resulted in several ‘smart drugs’ in various stages of preclinical and clinical investigation [6–8].
Table 1. Validated targets in prostate cancer
Endothelin receptor (overexpression of ETA)
EGF receptor (increased expression of TGF-α, decreased expression or alternatively spliced EGFR)
HER2 (overexpression of HER2)
IGF receptor (increased levels of serum IGF)
Gene: PCA3 (overexpression of noncoding mRNA) (increased and ectopic expression of neutrophin and increased expression of trkA)
Gene: PCGEM1 (overexpression of noncoding mRNA) (increased % of positive cells after endocrine therapy/TP cell marker)
Prostate cancer-specific genes themselves represent potential targets; such genes might indeed represent the most fundamental sites of attack for therapeutic interventions. A gene that is truly specific for prostate cancer could potentially form a basis for extremely precise and effective gene therapeutic approaches directed preferentially to diseased cells .
However attractive the theoretical benefits, the limitations and difficulties associated with this gene therapy approach are substantial. Currently known prostate cancer-specific genes are listed in Table 2[9–11]; it is immediately apparent that the catalogue of genes known with certainty to be expressed or strongly overexpressed in prostate cancer is quite small. Apart from these candidates, the remaining genes are expressed at levels similar to, or even lower than, those associated with benign tissue, which indicates that they are not particularly suitable targets for gene therapeutic purposes .
Table 2. Spectrum of known, prostate cancer-specific genes*
Expression in prostate cancer
*Only the indicated genes are overexpressed in the setting of malignancy, thus offering a potential mechanism for precise diagnosis and therapeutic gene delivery. Data from Hessels et al.; Verhaegh et al.; Bussemakers et al..
Dm neurokinin-3 related TF 1
homeobox transcription factor
acid prostate phosphatase
prostate-specific noncoding gene
prostate-derived Ets factor
Ets transcription factor
prostate stem cell antigen
GPI-anchored cell-surface protein
prostate-specific membrane antigen
protein cross-linking enzyme
transmembrane protease Ser 2
prostate-specific noncoding gene
More suitable are the notable exceptions. PCGEM1, which encodes a structural RNA fragment rather than a protein (noncoding RNA), is substantially overexpressed in prostate cancer, shows androgen dependence, and is associated with a high-risk patient group [12,13]. However, this gene is only up-regulated in African-American men with prostate cancer, suggesting that it would be most useful in a specific subpopulation of patients rather than a large group of unselected individuals [13,14]. Prostate-specific membrane antigen  and prostate stem cell antigen [15,16] are also often overexpressed, but the most consistently overexpressed gene in prostate cancer is prostate cell antigen PCA3 (also known as DD3).
PCA3 has been identified and extensively characterized by members of our group [10,11,17]. The promoter region of this gene has unique expression characteristics, and further study of this molecular target has resulted in a PCA3-based assay migrating successfully from the ‘omics toolkit’ of a specialized technology platform to imminent clinical deployment as a diagnostic application [10,11,18]. PCA3 is one of the most prostate cancer-specific genes described to date [10,11], and is intensively up-regulated in malignant cells. In an initial investigation, > 95% of clinical prostate cancer specimens (53 of 56) overexpressed PCA3. It was shown conclusively that this gene is not expressed to any extent in other tissues. In nonmalignant prostate tissue (including BPH specimens obtained from the same patient), the gene is expressed, albeit at an almost negligible level when compared with the cancer processes (Fig. 1) .
Reverse transcription-polymerase chain reaction (RT-PCR) showed that clinical samples from a range of urological, gynaecological and breast malignancies, as well as normal nonprostate tissues from various organ systems, were utterly devoid of PCA3 activity (Fig. 2). However, in prostate tumours, expression of PCA3 mRNA product is ≈ 10- to 100-fold higher than in adjacent non-neoplastic tissue; the average up-regulation is between 70 and 80-fold [11,18]. Thus PCA3 is a good candidate for a diagnostic and prognostic marker for prostate cancer [9,11,18].
The promoter region of PCA3 has intriguing attributes. This region was characterized by cloning several different promoter-human growth hormone (hGH) reporter constructs, which identified the presence of activator and repressor sites within a 500-bp segment. A minimal fragment extending only 152 nucleotides (Fig. 3) has all the elements required to drive the promoter . As shown in Fig. 3, the promoter-hGH construct had high activity in LNCaP cell lines that were PCA3-positive . By contrast, transcription of the reporter gene was negligible in PCA3-negative prostatic cell lines and marginal in malignant cells from breast, vulva, renal, colon and bladder neoplasms (Fig. 4). This provides emphatic evidence that the PCA3 promoter is tissue- and cell-type specific, a truly prostate cancer-specific entity [10,19].
As even a relatively minute number of PCA3 transcripts can be identified with RT-PCR, a quantitative assay would be a valuable advance for detecting neoplastic prostate cells in blood or urine, facilitating the diagnosis and molecular staging of cancer. The diagnostic potential of a PCA3-based assay has been evaluated in many studies [18,20,21]. A test measuring PCA3 was shown to be useful as an adjunct to PSA testing . In one study, transcripts were determined quantitatively from malignant and nonmalignant biopsy specimens from 108 men with elevated PSA levels (>3 ng/mL). Another comparative assay was performed on urine sediment collected after digital prostatic massage. PCA3 was markedly up-regulated in prostate tumours compared with benign tissue (median 158.4 × 105 vs 2.4 × 105 copies/µg); > 95% of cancer specimens showed up-regulation, and specimens with < 10% of cancer cells could be discriminated from completely benign glands. Of the 108 men, 24 had biopsy-confirmed prostate cancer. In these 24 men, quantitative assay of urinary sediment was positive for PCA3 in 16, for a sensitivity of 67% and a negative predictive value (NPV) of 90%.
Another larger clinical study of a PCA3-based assay provided evidence supporting the role of molecular diagnosis for prostate cancer . In this trial, enrolling 443 men with PSA levels of ≥ 2.5 ng/mL, screening with a recently available diagnostic product (uPM3; first-generation PCA3 test; DiagnoCure, Inc., Quebec City, Quebec, Canada) had a positive predictive value (PPV) of 75%, and a negative predictive value (NPV) of 84%. Tinzl et al. reported similar results using the same assay in a prospective study of 158 evaluable patients (PSA, ≥2.5 ng/mL; PPV, 67%; NPV, 87%).
OTHER PERTINENT TARGETS
While some targets have been satisfactorily validated, several others are potentially suitable (Table 3) [22,23], while still others undoubtedly await discovery. Examples of intriguing potential targets include growth factor response inhibitors and other components of signalling pathways; inactivation of potential targets associated with generic malignant proliferative behaviour, e.g. telomerase; invasion/metastasis inhibition; the endothelin receptor axis; cadherin pathways; and stem cell modulation [7–9,23–26]. Whether any of these targets, attacked alone or in conjunction with an attack on another sentinel pathway, might be the Achilles’ heel of a given cancer cell is a question that remains to be answered [22,23].
Table 3. Candidate targets: among the molecular targets meriting further study may be one that proves to be the ‘Achilles heel’ of a prostate cancer
Data from Litvinov et al.; van Leenders et al..
Androgen receptor (AR)
Growth factor receptors
Invasion metastasis inhibitors
Prostate cancer-specific gene expression
Membrane antigens for antibody/vaccine targeting
It is important to find a relevant pathway in prostate cancer and to determine its pivotal role. Because prostate cancer cannot be seen as just one disease, it is important that the pathway that is the focus is relevant to the patient being treated [27,28]. Patients should be stratified according to the pathway being attacked . Prostate cancers are not alike, and probably have to be subdivided according to their molecular profile. Treatment may have to be individualized by patient group [28–30].
THE ANDROGEN RECEPTOR (AR)
One phenotypically validated target attracting renewed interest is the AR and its associated axis . The AR is present on the prostate cell as it makes the phenotypic transformation to malignancy; androgen stimulation promotes prostate cancer progression in the early phases of the disease, and so androgen ablation is a mainstay of treatment for advanced prostate cancer [32,33]. As has long been known, the AR is often preserved in end-stage disease, and can be present even after the malignant prostatic epithelium has entered a terminal, androgen-insensitive phase of illness [31,34,35]. Despite the ligand depletion characteristic of androgen deprivation in late disease, the AR itself remains intact; indeed, it may be amplified or mutated in 20–30% of androgen-independent tumours, and it remains capable of exerting downstream effects after activation by several growth factors, cytokines, specific AR modulators, and other molecules [33,36]. Whether this is an important target pathway will depend on the extent to which the AR axis in advanced disease can re-establish the salutary downstream effects of receptor stimulation (such as were associated with the antecedent stage of androgen responsiveness). The AR is still important in end-stage disease, but the specific target within the AR axis needs to be defined more precisely [5,31].
PROSTATE CANCER STEM CELL
The search for the prostate cancer stem cell (presumably the most undifferentiated and multipotent cell in the cancer) is pivotal to solving the dilemmas surrounding the identification of potential targets . Once the cancer stem cell, which is hierarchically related to all cancer cells, is satisfactorily identified, it should be considered as a tool to help validate new targets. The ideal target is one that is essential for the maintenance and renewal of the cancer stem cells. However, it is likely that the search for, and characterization of, prostate cancer stem cells will disclose fruitful new targets for therapy [38–41].
A model for prostate epithelial stem cell biology was described previously [38,42]. Figure 5 schematically portrays a current model of stem cell differentiation. A stem cell undergoes asymmetrical division, resulting in self-renewal into a more committed progenitor, the transiently proliferating/amplifying ‘intermediate’ stem cell. This intermediate cell population (both early and late progenitors) gives rise to exocrine and neuroendocrine cell lineages .
To the extent that prostate cancer is a clonal expansion of malignantly transformed cancer stem cells, resembling the intermediate amplifying cell, differentiation from abnormal cancer stem cells would appear to recapitulate that behaviour. Identifying stem cells and the stromal, hormonal and other signalling molecules that influence their behaviour would have important implications for managing prostate cancer .
Important initial steps have been taken to identify this intermediate population of cells in prostate tissue [43,44]. Our group showed that a population of intermediate prostatic cancer stem cells commonly seen in androgen-refractory disease can be distinguished by a phenotype of keratin constituents (K5, K18), c-met, α2 integrin expression, and inducible laminin-γ2 [38,39,45]. During androgen deprivation or, similarly, during a period of androgen independence, this population increases [46,47]. It is thus reasonable to assert that any marker or any growth-signalling pathway that is pivotal for this intermediate stem cell population is an appropriate target . Although < 2% of cells in malignant epithelium are positive for c-met expression, it is a very important target for further study .
The cadherin pathway is a phenotypically well-validated target associated with cell adhesion, nuclear signalling, and cellular differentiation [49,50]. E-cadherin is reduced in aggressive prostate cancer, suggesting an association between loss of E-cadherin and tumour progression, and metastasis secondary to diminished cell-to-cell adhesion. Not surprisingly, loss of E-cadherin correlates with poor prognosis, not only in prostate cancer [25,49,50] but also in bladder cancer, breast cancer and others . We have shown that E-cadherin is lost with the advent of invasive prostate cancer, ‘switching’ to a mesenchymal type of cadherin (-2 and -11). An E-cadherin-associated protein β-catenin is also intimately involved in the wnt signalling cascade . Efforts directed at reversing the switch away from invasive cadherins and restoring E-cadherin suggest that this could be a very attractive type of therapy.
Important advances in molecular oncology continue to define many basic mechanisms associated with the development, progression, invasiveness, and intractability of prostate cancer. These findings may suggest alternatives to the generally unsatisfactory therapeutic approaches currently available for metastatic disease. Potential targets are suggested by multidisciplinary investigations undertaken to define or interfere with intrinsic cancer stem cell pathways or bidirectional host–cancer interactions [52,53]. A profound understanding of the pathway in which the potential target plays its role is needed, and the target should be clearly shown to have an essential role in a malignant cascade of molecular events.
The paramount issue confronting the molecular researcher is not where to look for potential targets, but rather how to confirm or reject the importance of the target under consideration. Determining whether a pivotal pathway has been identified requires meticulously assembling experimental evidence to support functional or phenotypic validation of the molecular process. Impelled, as researchers are, to confront the grim challenges of advanced prostate cancer, it must be realized that neither all processes nor all cells provide equally important opportunities to thwart disease progression.
Yet molecular research continually reaffirms that not all prostate cancers are alike. Thus, the dilemma of where to search for new targets will probably intensify as we approach an era when it might be possible to stratify prostate malignancies by their molecular profile, and design treatments for groups of patients sharing similar prostate cancer phenotypes [28–30,54]. It will be necessary to individualize treatment in some fashion, not idiosyncratically with one treatment per patient, but certainly by population, which is why the increasingly complex endeavours of molecular profiling and molecular therapy will have to forge an understanding with one another.
A multimodal approach might be the best solution to this dilemma; one should not reject or postpone the combination of several therapeutic pathways because some of them may, independent of one another, be ineffective in a particular molecular environment. However, combinations of approaches might produce synergistic effects.