Pathway Activation Markers

Signal trafficking regulation through major cell hubs is central in the processes of tumor initiation and progression, and represents a major target for tumor-specific therapies. In prostate adenocarcinoma, the phosphatidylinositol 3-kinase (PI3K) is the most frequently activated signal transduction pathway. In contrast, aberrant activation of the MAPK pathway seems less frequently involved, possibly due to absence of Ras and Raf mutations. In PCa the most common mechanism of PI3K pathway activation is the deletion of the gene encoding the phosphatase and tensin homologue (PTEN) protein, whose inactivation in turns leads to phosphorylations of AKT and of the S6 ribosomal subunit, which are therefore powerful biomarkers of PI3K activation. Upregulation of the PI3K/AKT/mTOR pathway is estimated to occur in about 50% of PCa, but not only through loss of PTEN. In addition, PTEN deletions seem to be more frequent in metastatic than in organ-confined disease. Immunohistochemistry can easily detect the activation of the major components of the PI3K pathway using antibodies against specific phosphorylation sites of the AKT, mTOR, and S6 proteins. Decreased expression of PTEN can be identified by immunohistochemistry as well, while its genetic inactivation may be determined by fluorescence in situ hybridization (FISH) or by genome-wide analyses such as array CGH (aCGH). In humans, down-regulation of PTEN and consequent activation of the PI3K pathway members in PCa tissue samples has been correlated to higher Gleason grade, advanced stage, and development of androgen resistance. Immunohistochemical detection of phospho-AKT in PCa tissue also predicts early biochemical recurrence and poor outcome. Of note, in mice inhibition of mTOR with rapamycin analogues results in complete reversal of the neoplastic prostate phenotype driven by activated AKT1. Several phase 1 and 2 clinical trials with specific agents targeting key activated proteins in the PI3K pathway are ongoing in PCa patients, including selective inhibitors of p110, AKT1, and mTOR. The TMPRSS2-ERG translocation (see later discussion) is found in upwards of 60% of prostate cancers and is androgen driven. Because this translocation results in similar downstream effects to MAPK activation, PCa displaying the fusion may be considered MAPK active. The combination of androgen blockade with tyrosine kinase inhibitors targeting the PI3K or the MAPK pathway may represent alternative therapeutic choices in PCa. Selection of patients for these therapeutic regimens therefore becomes very important. Such selection can only occur through the use of selective biomarkers.

Chromosome 8 Aberrations and MYC

Losses of 8p and gains of 8q represent 2 of the most common chromosome aberrations in PCa. In particular, the 8p locus seems to be the most commonly deleted region in PCa, occurring in 30% of organ-confined tumors and in about half of advanced cases. Single genomic losses at 8p represent early events in prostate carcinogenesis, as demonstrated by the occurrence of frequent 8p deletions in experimental prostate intraepithelial neoplasia (PIN). In humans, most PIN lesions are also associated with loss of heterozygosity at chromosome 8p21, where the NKX3.1 gene, encoding a validated tumor suppressor gene, resides. NKX3.1 is an androgen-regulated homeodomain transcription factor that regulates the proliferation rate of prostate luminal epithelial cells. The development of PIN lesions in NKX3.1-deficient mice has been recently associated with alterations of the PTEN/AKT axis. Haploinsufficiency of NKX3.1 might extend the proliferative stage of regenerating luminal cells, leading to epithelial hyperplasia and dysplasia. NKX3.1 has been used as a biomarker, with conflicting results to date.

On the other hand, 8q is the most commonly gained region in advanced PCa. Genetic polymorphisms at 8q are consistently associated with prostate cancer risk across multiple ethnic groups, with the highest susceptibility at the 8q24 region. Specific genetic variants at 8q24 have also been correlated with higher Gleason grade and more aggressive prostate cancer behavior. The coding region closest to 8q24 is the well-known oncogene MYC. Although many attempts have been made to correlate polymorphism at 8q24 with MYC upregulation, none to date has clearly demonstrated a relationship with MYC increased transcription. MYC is a well-known regulator of proliferation and biologic activity in prostate cancer cells, and its amplification is associated with the presence of PIN and poor clinical outcome of PCa. In addition, transgenic expression of MYC results in PIN as well as invasive cancers in the prostate. Nuclear overexpression of the MYC protein has been shown in PIN, suggesting a role for MYC in early prostate carcinogenesis.

TMPRSS2-ERG Fusion

One of the most important recent findings in PCa is the discovery of the TMPRSS2-ERG fusion. The fusion of the 5′ untranslated region of the androgen-regulated TMPRSS2 gene with the transcription factor ERG leads to the aberrant expression of ERG in an androgen-dependent manner. TMPRSS2-ERG gene fusion is rare in normal prostate tissue whereas it is consistently detectable in PIN, in organ-confined as well as in metastatic hormone refractory PCa. The presence of the TMPRSS2-ERG fusion has been associated, in some studies, with high tumor stage, presence of lymph node metastases, and poor outcome. These reports, together with the finding that the fusion occurs in both PIN and adjacent invasive cancer cells apparently as an “all or none” phenomenon in individual invasive clones within the same prostate, led to the hypothesis that the TMPRSS2-ERG rearrangement might define, from the early steps of carcinogenesis, a subset of more aggressive PCa. Unfortunately, attempts to correlate the fusion status with high Gleason score or with outcome have given conflicting results in subsequent studies. This issue is further complicated by the presence of a growing number of different translocations involving the ETS transcription factor family. Further work is required to understand the role that this prevalent translocation plays in prostatic carcinogenesis and its relationship to the biologic behavior of prostate adenocarcinoma.

Biomarkers of Altered Lipid Metabolism in Prostate Cancer

More than 80 years ago the Nobel prize winner Otto Warburg proposed that tumorigenicity of cancer cells derived from their ability to switch from oxidative phosphorylation to glycolysis to satisfy their high energetic needs. The products of the glycolytic pathway and of the subverted metabolic pathways may provide the substrate for the structural need in the tumor cell enhanced proliferative state. Lipogenesis is therefore a distinctive feature of tumor cells. Cancer cells synthesize de novo large amounts of fatty acids and cholesterol irrespective of the circulating lipid levels, and benefit from this increased lipid synthesis in terms of growth advantage, self survival, and drug resistance. Numerous studies have shown that inactivation of most lipogenic enzymes, such as ATP citrate lyase (ACL), fatty acid synthase (FASN), and acetyl-CoA carboxylase, results in either cytostatic or cytotoxic effects in tumor cells. Cholesterol is critical for the composition and stabilization of cell membranes while fatty acids may also be required for post-transcriptional regulation of key signal transduction proteins through posttranslational modifications such as palmitoylation and myristoylation. Fatty acid synthase (FASN) is the only enzyme that is able to synthesize fatty acids de novo in normal and cancer cells, and its main enzymatic product palmitic acid is responsible for the acylation (palmitoylation) of key regulatory switches in most signal transduction pathways. FASN was proposed as the first bona fide candidate metabolic oncogene in the prostate. The mere forced expression of FASN is able to transform immortalized prostate epithelial cells, to form invasive tumors in immunodeficient mice while transgenic mice expressing FASN in the prostate develop PIN. The proposed mechanisms of FASN oncogenicity include structural needs of synthesized lipids, protection from endoplasmic reticulum stress, inhibition of the intrinsic (mitochondrial) pathway of apoptosis, and palmitoylation of Wnt-1 with subsequent stabilization of β-catenin and activation of the pathway. Natural and irreversible inhibitors of FASN are able to reduce cell proliferation and induce cell death in PCa cell lines, and to reduce the volume of PCa xenograft tumors in mice. FASN can be therefore considered an excellent and promising biomarker and therapeutic target in prostate cancer.

Caveolae are plasma membrane microdomains rich in cholesterol, similar to lipid rafts but characterized by the presence of a protein family named caveolins. Both lipid rafts and caveolae are cholesterol- and sphingolipid-enriched microdomains in cell membranes that regulate phosphorylation cascades originating from membrane-bound proteins. Altered cholesterol synthesis results in changes in membrane cholesterol and, in turn, in Akt signaling in both normal and malignant cells. Caveolins possess a key regulatory activity in cell molecular transport and cell trafficking. The most studied member of this family, Caveolin-1, has been involved in prostate cancer initiation and progression and identified as a marker of aggressive PCa. Caveolin-1 knock-out in the PCa TRAMP mouse model significantly reduces prostate tumor size and the development of metastases. This effect might be at least in part mediated by FASN, which is downstream of Caveolin-1. Caveolin 1 may therefore be an important biomarker of aggressive behavior in prostate cancer.

Stem Cell Markers

Identification and isolation of cancer stem cells (CSCs) in solid tumors, including PCa, has been an area of intense research effort in the last 10 years. A growing body of literature supports the CSCs hypothesis as a model to explain intratumoral heterogeneity and the development of resistance to therapy.

CSCs (also known as tumor-initiating cells, TICs) are characterized by unlimited renewal potential when injected into immunodeficient mice in which neoplasms recapitulating the heterogeneity of the original tumor form at high efficiency. CSC phenotype is usually defined by a panel of multiple rather than single surface markers. In PCa these cells have been identified as CD44 ⁺ /α2β1 ^high /CD133 ⁺ and androgen receptor negative. This phenotype was widely employed to select and isolate putative CSCs from prostate cell lines, xenografts, and biopsies of human tumors to test their invasive property and to identify a specific CSC genetic signature.

CD44 ⁺ /CD24 ^– cells isolated from LNCaP cells possess increased clonogenic properties, form tumors in NOD/SCID mice, and show a gene signature of invasiveness originally identified in breast tumors. There is also experimental evidence that a subpopulation of CD44 ⁺ CSC-like cells is able to invade Matrigel, suggesting that basement membrane invasion is a characteristic trait of these putative prostate CSCs. It is intriguing that in human PCa tissues all the cells with neuroendocrine differentiation are CD44 ⁺ , suggesting a role for such cells in the resistance to androgen therapy and tumor recurrence. Additional studies are required to better identify putative prostate CSCs in PCa tissue samples. To the authors’ knowledge there are no conclusive studies that have characterized the molecular phenotype and the clinical significance of CSCs in PCa patients. Based on current knowledge, it seems most plausible that prostate CSCs originate from cells in different stages of prostate epithelial cell differentiation, providing an alternative explanation for the well-known morphologic and biologic heterogeneity of human PCa. Reliable recognition of prostate CSCs and development of novel markers for their identification might help the development of therapies specifically targeted at the CSCs compartment.

Single Nucleotide Polymorphism Analysis, Gene Expression Profiling, and Comparative Genomic Hybridization

SNP array technology allows genome-wide and high-throughput analysis of DNA polymorphisms with the intent of comparing different cancers, exploring tumor-normal differences, and assessing polymorphic alleles with predictive behavior. SNP is therefore the method of choice to assess genetic variants and allelic imbalances. Although in the prostate no susceptibility genes comparable to BRCA1 in breast or APC in colon have been discovered so far, multiple germ-line polymorphisms have been correlated to increased individual risk of developing PCa. In addition to susceptibility studies, SNP arrays have been used to distinguish genetic subsets of PCa with different clinical behavior. Progression of PCa toward metastasis and high Gleason grade in the primary tumor were recently correlated to gains of 8q, 1q, 3q, and 7q, whereas androgen ablation therapy was characterized by gains at 2p and 10q using SNP array analysis. In another large control study based on SNP arrays, germ-line deletion at 2p24.3 was strongly associated with aggressive PCa. Germline polymorphisms within genes encoding for androgen-metabolizing proteins have been also correlated to response rate in PCa patients treated with hormone therapy.

aCGH is the best technology to detect DNA copy number alterations in cancer compared with normal tissues. Deletions in 8p, 13q, 6q and gains in 8q and 7q, and specific losses at 10q24 (PTEN) and gains at Xq12 (AR) are the most common copy number alterations in PCa. Loss of 8p and gain of 11q13 have been associated with advanced stage and biochemical recurrence, respectively. Of note, intermediate risk PCa was characterized by copy number alterations previously associated with high-risk disease. The progressive optimization of the aCGH on DNA extracted from FFPE tissues will extend the applications of this technique with further expansion into clinical practice.

Gene expression profiling of up to 26K genes is now available also for RNA extracted from FFPE tissues. Expression-based models have been correlated to patient outcome and to biochemical recurrence in PCa. A case-control study comparing men who had just PSA failure with those who developed metastatic PCa after radical prostatectomy showed that the 2 groups could be separated by an expression model containing 17 genes with a specific enrichment in the 8q24 locus. Unfortunately, the high number of genes used in constructing these predictive models and the variable cutoffs employed to estimate the signal to noise ratio in each model, together with the complex bioinformatic algorithms required to deconvolute and interpret the data, have prevented the routine use of gene expression analyses in clinical practice, although these may become useful adjuncts to the commonly used Partin tables or Kattan nomograms. Once the RNA extraction techniques from FFPE tissues are optimized and the relevant gene signatures applied, this technology will likely become necessary for stratification, prognostication, and assessment of predictive behavior in PCa patients.

Free Plasma DNA

It has been known since the 1950s that free plasma DNA (FPDNA) circulates in the blood of normal individuals while increases in FPDNA may occur in pathologic conditions including autoimmune diseases, liver cirrhosis, major traumas, and malignant tumors. The detection of circulating free plasma DNA seems to be a promising and noninvasive test for early tumor detection, assessment of disease recurrence, and monitoring of therapy. Increased FPDNA levels have been observed in patients with several epithelial malignancies, including PCa, when compared with healthy subjects. In addition, FPDNA has been recently correlated with both tumor stage and the presence of CTCs, suggesting a role of FPDNA as candidate biomarker for the monitoring of PCa patients.

An important concern for the application of FPDNA tests are false positives, particularly in patients with autoimmune or inflammatory diseases and a recent history of trauma or surgical procedures. To hone in on the tumor origin of circulating DNA, specific genetic aberrations (eg, loss of heterozygosity and microsatellite analysis) or epigenetic alterations (eg, gene promoter hypermethylation) should be identified in FPDNA. Such alterations have been successfully reported in blood and bone marrow samples. However, discrepancies between genetic aberrations in the primary tumor DNA and in FPDNA have been reported, and might be ascribed to tumor heterogeneity with various populations of cells with diverse genetic alterations undergoing lysis. Further clinical validation of these findings may unravel a role for FPDNA as a valuable new biomarker for monitoring the metastatic progression in PCa patients.

Circulating Tumor Cells and Disseminating Tumor Cells

The occurrence of a metastatic, castration-resistant tumor represents the major cause of PCa-related mortality. Metastatic spread has been typically considered a late process in malignant progression, but several studies have recently suggested that dissemination of primary cancer cells to distant sites might be an early event in tumorigenesis. In addition, tumor cells can bypass the lymph node filter and disseminate directly through the bloodstream to distant organs. These findings led to the development of different assays for the detection of disseminated tumor cells (DTCs) in bone marrow (BM) and CTCs in the peripheral blood. The 2 main techniques employed are the immunologic and the polymerase chain reaction (PCR)-based molecular approach. In the immunologic approach, immunochemistry using monoclonal antibodies against surface epithelial antigen is the most widely applied technique. This method is easy to perform and enables the evaluation of cell size and morphology, but sensitivity and specificity are antibody dependent. Real-time reverse transcription (RT)-PCR–based assays are extremely sensitive and able to detect aberrations at a single cell level. Prostate-specific transcripts such as PSA, prostate-specific membrane antigen, and prostate stem cell antigen are usually used as single or multiplexed surrogate markers in blood CTCs using RT-PCR or quantitative real-time PCR (qPCR). The major problem of the molecular approach is the false positive rate due to illegitimate transcripts and heterogeneous expression of target markers. The introduction of a cancer cell-enrichment step during the CTC isolation process and the establishment of a reliable cutoff value for analysis may overcome these problems. Most clinical reports on DTC focused on BM, the most common metastatic site in PCa. Some investigators reported significant correlations between the presence of DTCs and clinical-pathologic parameters such as high Gleason grade or metastatic disease. In addition, the presence of DTC in the BM at the time of diagnosis represents an independent negative prognostic parameter in patients with localized PCa. Because BM aspiration is invasive, uncomfortable for the patients, and not suitable for repeated analysis, recent efforts have focused on the detection of CTCs in the peripheral blood. CTCs can now be easily detected by PCR in the blood at the time of diagnosis before, during, and after therapy, and their increased number has been positively associated with high Gleason score and stage. In addition, the detection of PSA mRNA by qPCR has been significantly correlated to time to progression and overall survival. Technical limitations of the PCR technique and the need for a more standardized method for the detection of CTCs in the peripheral blood led to the development of new technologies. The CellSearch (Veridex) is a device recently approved by the Food and Drug Administration for the monitoring of metastatic breast, colon, and prostate cancer able to isolate single CTCs by immunomagnetic enrichment followed by fluorimetric count ( http://www.accessdata.fda.gov/cdrh_docs/reviews/K073338.pdf ). Data generated using this automated system showed that CTCs could be detected in 55% to 62% of patients with castration-resistant prostate cancer (CRCP). A baseline CTC count of 5 cells/7.5 mL or more of blood before therapy represents a powerful predictor of poor overall survival (OS). In addition, the study of CTC dynamics following therapy showed that the CTC count predicts clinical outcome better than the algorithms based on PSA. Patients whose CTC counts decreased from 5 cells or more at baseline to less than 5 cells after treatment had a better OS compared with those showing an increase during therapy. By contrast, in patients with organ-confined PCa the number of detectable CTCs appears low and does not correlate with known prognostic factors. Further molecular characterization of CTCs or DTCs in cancer patients could provide additional information on cancer biology and could improve the management of the disease by selecting effective targeted therapy in the individual patient context. Recent studies show that both high- and low-resolution techniques such as FISH or CGH could be performed on isolated cancer cells to obtain a genomic profile of CTCs/DTCs that could be related to prognosis and response to therapy. The ultimate goal of the research on DTC/CTC is their propagation in vitro after isolation from cancer patients. Such an approach could provide a tool to test personalized oncologic treatments directly on the cells responsible for tumor progression of each specific patient.