This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ruijter, E. Articles by Schalken, J. Search for Related Content PubMed PubMed Citation Articles by Ruijter, E. Articles by Schalken, J.

Molecular Genetics and Epidemiology of Prostate Carcinoma

Departments of Urology (E.R., F.D., J.S.) and Pathology (E.R., C.v.d.K., D.R.), University Hospital Nijmegen, 6500 HB, Nijmegen, The Netherlands; Department of Pathology (G.M.), University of Colorado, Health Sciences Center, Denver, Colorado 80260, U.S.A.; and Department of Pathology and Companion Animals (J.S.), Veterinarian Faculty, University of Utrecht, 3508 TD, Utrecht, The Netherlands

	Abstract

Top Abstract I. Introduction II. Somatic Genetic Alterations... III. Etiology of Familial... IV. Human Papillomavirus and... V. Morphological Diversity of... VI. Concluding Remarks References

 

I. Introduction

A. Prostate carcinoma epidemiology: mortality rates, incidence, and prevalence

B. Familial clustering of prostate carcinoma

C. Scope of the review

II. Somatic Genetic Alterations in Prostate Carcinoma

A. Alterations in DNA methylation

B. The androgen signaling cascade

C. Vitamin D and prostate carcinoma

D. Oncogenes

E. TSGs

F. Metastasis suppressor genes (MSGs)

G. Telomerase activity

III. Etiology of Familial Clustering of Prostate Carcinoma

IV. Human Papillomavirus and Prostate Carcinoma

V. Morphological Diversity of Prostate Carcinoma

VI. Concluding Remarks

	I. Introduction

Top Abstract I. Introduction II. Somatic Genetic Alterations... III. Etiology of Familial... IV. Human Papillomavirus and... V. Morphological Diversity of... VI. Concluding Remarks References

 
A. Prostate carcinoma epidemiology: mortality rates, incidence, and prevalence
AFTER lung cancer, prostatic adenocarcinoma (PCa) is the second leading cause of male cancer deaths in the United States. A total of 41,800 men died of PCa in 1997, and in men over the age of 55 yr, PCa is responsible for nearly 4% of all deaths (1). The age-specific PCa mortality rate has increased approximately 14% over the past six decades (1). However, the PCa-related mortality is relatively low when compared to the total number of patients diagnosed each year with PCa, i.e., clinically diagnosed PCa. In 1997 in the United States, 334,500 new cases of PCa were estimated to be diagnosed clinically, 3 times more than those from the respiratory tract (1). Thus, the vast majority of the PCa patients will die with this disease rather than from it. This might be explained by the advanced age of men at the time of diagnosis in combination with relatively slow tumor growth, implying that a large number of men may not live long enough to die of PCa (1, 2). The number of men diagnosed each year with PCa has shown a 30% increase over the last 25 yr, and for the next decade, a similar trend of rising incidence is expected (3). Moreover, PCa incidence is estimated to double by the year 2030 (4). This is due to the strong association of PCa with age in combination with the rising average age of American men, improvements in detection techniques, and programs for early detection of PCa (5)

The number of males diagnosed each year with PCa is not similar among different racial populations. Bernstein and Ross (6) observed that PCa is most commonly diagnosed in African-Americans (116/100,000 persons per yr). Intermediate incidence rates are found in Caucasians (71/100,000) and lowest rates among Asians (Japanese, 39/100,000; Chinese, 28/100,000). Thus, this particular registration found a more than 4-fold difference in PCa incidence worldwide. This difference is a conservative estimate when compared to cancer registration studies that found a 30-fold difference or more in PCa incidence between African-Americans and Asians (7).

Interestingly, autopsy cases and cystoprostatectomy specimens removed for bladder cancer reveal a high prevalence of PCa lesions that have not been diagnosed clinically (8, 9, 10). It has been estimated that 15–30% of males over the age of 50 and as many as 80% of the males over the age of 80 harbor such clinically undetected foci of PCa (9, 10). Consequently, 8 million US males older than 50 yr are estimated to have PCa, which makes PCa the most prevalent cancer in man (3, 5). Comparing the incidence and prevalence rates, the vast majority of men who harbor PCa lesions will not be diagnosed with this disease. Geographically, the prevalence of PCa in autopsy cases without a prior clinical diagnosis of this disease is roughly the same worldwide (10, 11, 12). Thus, although men from different races harbor foci of carcinoma in their prostates at a similar frequency, the number of men whose tumors will become clinically apparent is much higher among men from African-American populations compared to Caucasians or Asians. Interestingly, Whitmore (2) found that African-American men appear to have a larger volume of PCa at autopsy compared to Caucasians. These relatively large carcinomas are likely to progress to clinically apparent disease at a faster rate. Consistent with this hypothesis is the observation that African-American men typically present with PCa at younger ages with larger volume, higher stage, and often higher grade tumors than Caucasian men, with consequently higher mortality rates (6, 13). These studies also suggest that the events that account for racial/ethnic differences in PCa incidence had occurred very early in the process of carcinogenesis, possibly already at the stage of (pre)malignant transformation. Interestingly, some studies have focused on racial/ethnic differences in the characteristics of prostatic intraepithelial neoplasia (PIN), the lesion considered to be the precursor of PCa (11, 14). Sakr et al. (11) revealed that the overall prevalence of high-grade PIN lesions found during autopsy in African-American men (n = 314) exceeded that in their Caucasian counterparts (n = 211). High-grade PIN lesions in African-American men also tended to be more diffuse with multifocal or extensive involvement of the prostate compared to Caucasians (11, 14). In addition, extensive high-grade PIN appeared approximately a decade earlier in African-American males compared to Caucasians (11).

The finding of differences in the prevalence of high-grade PIN lesions in African-Americans vs. Caucasians seems to contradict similar worldwide rates of PCa found at autopsy. One likely explanation is that high-grade PIN is not a precursor lesion to carcinomas that remain ’clinically silent’ but to those PCa lesions that will progress to be detected clinically. The above-mentioned observations of PCa epidemiology make this disease unique among other types of malignancies (Fig. 1).

View larger version (26K): [in this window] [in a new window]   Figure 1. Frequency of PCas among the general population. If the prevalence of US citizens among men more than 50 yr of age is 30%, then approximately 8,000,000 Americans harbor foci of carcinoma in their prostate. Yet, in 1997 only 334,500 men were estimated to be clinically diagnosed with this malignancy, and the vast majority of these males died from causes other than prostate carcinoma.

Current understanding suggests that accumulation of multiple genetic alterations is important for cancer to occur. Persons who inherit one of these genetic alterations are thought to be predisposed to cancer development and are at higher risk for cancer at an early age than individuals who acquire these alterations later in life. Therefore, one feature of an inherited form of cancer is an increased clustering of cancers in the families of cases with early onset. Indeed, Carter et al. (22) reported cumulative risks of PCa of 14%, 25%, and 40% for first-degree relatives of PCa cases with an age at diagnosis >65, 53–65, and <53, respectively (22). The consistent finding of a younger age at diagnosis in hereditary PCa case patients has raised the issue as to whether or not to screen in men with a strong family history of PCa, starting at age 40 yr (23).

The best fitting model that explained the familial aggregation and age at diagnosis is a rare autosomal dominant susceptibility gene, and this model fitted best when probands were diagnosed at 60 yr of age or less (24). The model predicted that the frequency of the susceptibility gene in the population is 0.006 and that the risk of PCa by age 85 yr is 89% among carriers of the gene and 3% among noncarriers (24). Of all PCa occurrences, approximately 9% were estimated to result from such a gene (22). Based on the results of a segregation analysis, three criteria for hereditary PCa were defined: 1) a cluster of three or more first-degree relatives with PCa, or 2) PCa in each of three generations in the paternal or maternal lineage, or 3) two or more first- or second-degree relatives with PCa under the age of 55 (22).

C. Scope of the review
The epidemiological characteristics of PCa have been recognized for several decades (3). It is of great importance to understand the factors responsible for prostate carcinogenesis: why some carcinomas remain ’clinically silent’ during life whereas other tumors progress to present clinically and may lead to PCa-related death. A better understanding of these mechanisms in molecular genetic terms will likely point to more rational approaches to disease prevention, intervention, and treatment. A number of reviews have focused on molecular genetic changes associated with prostate carcinogenesis, such as viral oncogenesis (25) and the role of oncogenes (26) or tumor suppressor genes (TSGs) (27), whereas others have provided reviews with a wider scope (28, 29). This review seeks to provide a comprehensive overview of the current state of our knowledge regarding the etiology and molecular pathogenesis of PCa in relation to the epidemiological characteristics of the disease.

	II. Somatic Genetic Alterations in Prostate Carcinoma

Top Abstract I. Introduction II. Somatic Genetic Alterations... III. Etiology of Familial... IV. Human Papillomavirus and... V. Morphological Diversity of... VI. Concluding Remarks References

 
A. Alterations in DNA methylation
In human cancer cells, some genomic alterations are characterized by abnormal methylation. The patterns of abnormal methylation include hypermethylation, redistribution of methylation, and demethylation of normally methylated regions. Loss of 5'-methylcytosines, or hypomethylation, has been reported to occur in human PCa, but its significance is not entirely clear (30). Of more biological importance are genomic regions of hypermethylation. The most important site of abnormal methylation occurs in regions of high-density C–G dinucleotide sequences, referred to as CpG islands. These CpG islands are generally found in or near the 5'-region of genes, which may contain the promotor and one or more exons of its associated gene.

Methylation of CpG-rich islands may be associated with transcriptional inactivation of the associated gene, such as the CDKN2 (p16) gene (31). This gene, located at 9p21, encodes a cyclin-dependent kinase-inhibitory protein that controls passage through the G₁ phase of the cell cycle (32). Inactivation of the CDKN2 gene by homozygous deletion, point mutation, or aberrant methylation in the 5'-promotor region may induce progression through the cell cycle. It has been shown that homozygous deletion of this gene occurs in approximately 20% of the PCa samples (33). Methylation analysis of a CpG-rich promotor region in exon 1 of the CDKN2 gene revealed dense methylation in three PCa cell lines (TSU-PR1, PPC-1, PC-3), and this was found to correlate with a lack of mRNA expression by RT-PCR (32). Importantly, in vitro treatment of these PCa cell lines with the demethylating agent 5-aza-2'-deoxycytidine induced reexpression of CDKN2 transcripts (34). In human PCa samples, Jarrard et al. (31) showed that 3 (13%) of 24 primary and 1 (8%) of 12 metastatic tumor samples demonstrated promotor methylation. In the same study, deletions near the CDKN2 gene were detected in 12 (20%) of 60 primary tumors and in 13 (46%) of 28 metastases.

In early stage PCa, Gu et al. (35) could not find any intragenic alteration of the methylation pattern (35). However, two samples did have deletions proximal to or within the p16 gene. These results indicate that mutations in p16 or that inactivation of p16 by DNA methylation may not be a dominant pathway for the transformation into PCa, but may be involved in progression of disease.

Recently, Lee et al. examined the methylation status of the glutathione S-transferase (GST) gene promotor in prostate tissue (36). The GST gene, located on chromosome 11q13, is an essential part of an important cellular pathway to prevent damage from a wide range of carcinogens. Analysis of a CpG island within the promotor region revealed hypermethylation of the GST gene in all 20 PCa samples. A striking decrease of GSTP1 expression was found to accompany prostate carcinogenesis. Further analysis of 91 PCa samples by immunohistochemical staining with an anti-GSTP1 antibody revealed that no GSTP1 expression could be detected in 88 of these cancer specimens (36). Methylation alterations at the GST gene were not detected in DNA samples from 12 nonmalignant tissues, including 5 samples from benign prostatic hyperplasia (36). However, hypermethylation has been found in nonmalignant tissues, e.g., at the ‘hypermethylated in cancer’ (Hic1) gene. Hypermethylation of Hic1 was also found in 25 of 26 PCa specimens (37). Methylation of the GST gene promotor and the Hic1 gene appears to be a common phenomenon in PCa. This finding warrants further study to understand its biological implications. Clinical studies are currently undertaken to design a GSTP1 CpG-island methylation assay that may serve as a new molecular diagnostic tool for PCa detection. The finding of hypermethylation of the Hic1 gene in nonmalignant tissues may provide further insight into the marked tendency of the prostate toward malignant transformation.

B. The androgen signaling cascade
Testosterone (T) is the principal male sex hormone secreted by the testis that circulates in the blood bound to albumin and steroid-binding globulins (38). Only free T is able to enter the cell, after which more than 90% is irreversibly converted into the main prostatic androgen, dihydrotestosterone (DHT) through the action of the enzyme 5- reductase (Fig. 2) (38). The binding affinity to the prostatic androgen receptor (AR) of DHT is 5 times higher than that of T (39). The AR is normally associated with heat shock proteins and in this state cannot bind to androgen-responsive elements (40). Androgen binding induces dissociation from heat shock proteins, hyperphosphorylation, conformational changes, and dimerization of the receptor. This allows binding of the AR to specific DNA sequences called androgen-responsive elements located within the promoters of androgen-responsive genes. In conjunction with cofactor proteins and other transcriptional factors, the AR is then able to up- or down-regulate the transcription of genes (Fig. 2) (41).

View larger version (29K): [in this window] [in a new window]   Figure 2. Concepts of androgen activation through receptor binding. HSP, Heat-shock protein; ARE, androgen-responsive elements.

Many different terms, such as hormone-escaped, therapy-resistant, androgen-refractory, or androgen-independent disease, have been used to describe the process of therapy failure; the latter term will be used in this review. Androgen-independent growth of PCa may portend a poor prognosis as it reflects the acquisition of new growth advantages by the cancer cells. Therefore, it is of major clinical importance to elucidate the qualitative and quantitative alterations of AR gene alterations in PCa specimens.

It has been suggested previously that the AR is the cause of androgen independence of PCa. Therefore, it was not anticipated that immunohistochemical and RT-PCR methods proved expression of the AR in almost all primary and advanced tumor lesions, irrespective of the sensitivity of the tumors to androgen ablation therapy (45, 46). Thus, mechanisms other than loss of AR expression are likely to be involved in progression of PCa to an androgen-independent state. These might involve both androgen-independent mechanisms, such as activation of growth factor pathways, and androgen-dependent mechanisms, such as rate of metabolism, hypersensitivity to residual nontesticular androgens, increased androgen biosynthesis from adrenal precursor steroids, or alterations of the AR gene.

To date, 39 AR gene mutations in PCa samples have been collected on an Internet site (http://www.mcgill.ca/androgendb) (47). Twenty additional mutations have been described in two other reports (48, 49). A total of 12 PCa specimens contained a mutation in the amino-terminal domain (exon 1), which encodes more than half of the AR protein. Other mutations have been found in the DNA-binding domain (exons 2–3, 3 PCa specimens) and in the hinge region (exon 4, 2 PCa specimens) of the AR gene. Mutations in the steroid-binding domain (exons 4–8, 42 PCa specimens) of the AR gene are of particular clinical importance because they could render the mutant receptor constitutively active due to loss of the normal repression of AR transactivation in the absence of androgen. Another effect of AR gene mutations is that the AR can be activated not just by DHT but also by other steroid metabolites that otherwise have a low affinity for the AR (43, 50, 51).

For example, Culig et al. (51) reported an AR gene mutation in the steroid-binding domain of a metastatic PCa sample (51). The transcriptional activity of the mutant AR was similar to that of the normal AR in the presence of DHT but was increased in the presence of progesterone and adrenal androgens. Growth stimulation of tumors with such receptor mutations would be expected to occur after surgical castration alone (due to adrenal androgens) or estrogen or progesterone therapy (cyproterone acetate, chlormadinone acetate) (Fig. 3). Of similar clinical importance is the finding that one of the currently available antiandrogens, flutamide, has an increased affinity for the mutant AR (52). Flutamide is able to induce dissociation of heat shock protein on binding to the 868 (Thr Ala) mutant, which subsequently enables the AR to bind to androgen-responsive elements, resulting in the expression of agonist activity (Fig. 3) (40). The benefits of discontinuation of flutamide in patients whose metastatic PCa becomes androgen independent has been referred to as the ’flutamide withdrawal syndrome’ (53). There is also preliminary evidence that withdrawal responses to the antiandrogen, chlormadinone acetate, are associated with the presence of a mutated AR (54).

View larger version (17K): [in this window] [in a new window]   Figure 3. The mutant androgen receptor (AR) can be activated by a variety of steroid hormones, antiandrogens, and growth factors.

The differences in the prevalence of AR gene mutations could be attributed to the examination of small fragments of the AR gene rather than the complete AR coding region. Different AR gene mutation rates may also be due to the variability in the exons examined and to techniques used for mutation analysis, including single-stranded conformation polymorphism and denaturating and temperature gradient gel electrophoresis. In addition, mutations may not be detected if a considerable amount of somatic DNA is present in the sample analyzed. Such contamination is especially a problem with PCa because these carcinomas characteristically expand through diffuse penetration of the adjacent stroma by small abortive glands and single cells. More reliable estimates of AR mutation rates are anticipated when nonmalignant tissue is carefully removed from the sample to be analyzed, e.g., by (laser) microdissection.

AR gene amplification may also contribute to the development of androgen-independent growth of PCa (63, 64). Koivisto et al. (63) examined tumor specimens from 26 PCa patients at the time of, and before, androgen-independent disease. Fifteen (28%) of the recurrent androgen-independent tumors, but none of the untreated primary tumors, contained AR gene amplification as determined by fluorescence in situ hybridization. This group also demonstrated that AR gene amplification does not commonly occur during the clinical progression of cancer in patients who had not received androgen ablation therapy. These findings strongly suggest that it had occurred as a result of selection during androgen ablation therapy. Of note, the median survival time after recurrence of androgen ablation therapy was significantly 2 times longer for patients with AR amplification in comparison to those with no amplification (63). This observation was not anticipated because DNA amplification is usually associated with genetically highly unstable (65) and aggressive tumors (66). Koivisto et al. (63) speculated that the AR-amplified tumor cells, although being genetically unstable, are still subject to rigorously regulated androgen-dependent growth and therefore show a lower level of malignant potential.

The realization that the AR receptor is expressed and obviously active during all stages of prostate carcinogenesis has changed our concept of the androgen-independent cancer. The androgen-signaling cascade also is apparently still a reasonable target for therapy in advanced stages of PCa. However, the currently available therapeutic means are not sufficient to block androgen stimuli through a hyperactive AR during these later stages of the disease. Consequently, it will be important to determine in clinical trials whether complete androgen ablation therapy is effective as a second-line therapy for recurrent androgen-independent tumors with AR amplification. In addition, it is important to establish whether complete androgen ablation as a primary treatment also leads to recurrence through AR gene mutations or through AR amplification.

There is increasing evidence that peptide growth factors can also activate the AR through a ligand-independent mechanism, such as the cAMP, protein kinase A, and protein kinase C pathway (67). For example, Culig et al. (68) demonstrated that insulin-like growth factor-1 was able to induce AR-mediated transcription in the DU-145 cancer cell line. Less potent stimulatory effects were observed with keratinocyte growth factor and epidermal growth factor. In the PCa cell line LNCaP, insulin-like growth factor-1 increased the level of prostate-specific antigen (68). Interestingly, the paracrine mode of action of growth factors in the nonmalignant prostate and during early stages of PCa is altered toward autocrine stimulation of epithelial cells during advanced stages of disease. Thus, in androgen-independent PCa, the tumor cells might have gained the ability to produce growth factors by themselves, resulting in proliferation through action of the AR (Fig. 4) (69). Moreover, some of the mutant AR may be activated more efficiently by polypeptide growth factors than the wild-type receptor (Fig. 3). These observations may be another explanation why some tumors continue to grow after androgen ablation therapy. There are at least two therapeutic implications related to the above mentioned observations.

View larger version (26K): [in this window] [in a new window]   Figure 4. Site and action of polypeptide growth factors in nonmalignant and malignant prostate. In nonmalignant prostate tissue and during early stages of prostate carcinoma, some growth factors are produced by stromal cells with resultant effects on the population of epithelial cells. However, during advanced stages of prostate carcinoma, the cancer cells gain the ability to produce some of the growth factors by themselves.

A lifelong enhanced activity of the AR due to an inherited polymorphism in the AR gene might also alter the risk of PCa. Two highly polymorphic CAG (73) and GGN (74, 75, 76) (where N is any of the four nucleotides) microsatellite repeats are present in exon 1 of the AR gene, residing the transactivation domain. With respect to the GGN repeat, the most frequently occurring alleles among 73 white men were 22 (51%) and 23 (42%) repeats, counting only GGC and GGT triplets (76). Limiting their counts to the number of GGN repeats to those contiguous triplets in which the third nucleotide was cytosine, Irvine et al. (77) reported that CAG and GGC repeats were in linkage disequilibrium among PCa cases, but not among controls.

The odds ratio for PCa comparing any GGC repeat length to the control group mean of 16 was 1.18 (P = 0.08) (77), and in another study the odds ratio for PCa was 1.60 among men with 16 or fewer repeats compared with those with more than 16 repeats (78). In a recent study by Platz et al. (74), 582 PCa cases and 794 controls were analyzed for GGN repeats. No statistically significant difference in the mean GGN repeat length was observed between cases and controls. However, cases had a narrower spread of repeat lengths than controls (P = 0.03), with fewer extreme lengths in either direction. For every one repeat deviation in either direction from 23, the risk of PCa decreased by 8% (P = 0.04) (74). Reasons for the disparity in the direction and/or magnitude of the effect between the three studies are unclear but possibly include instability in the estimates of effect due to small numbers, especially at the extreme of the GGN repeat length distribution; genetic variability in the underlying populations from which the case-controls were drawn; different proportions of cases who were older at diagnosis; and etiological differences due to an increased proportion of early lesions resulting from screening for prostate-specific antigen in recent years (74).

The second polymorphic AR gene microsatellite is the CAG repeat (73). An expansion of the microsatellite to 40–62 CAG repeats is associated with X-linked spinal and bulbar muscular atrophy (Kennedy’s disease) (79). The average length of the CAG microsatellite in the normal population is 21 ± 2 (range 11–35) (73). Giovannuci et al. (80) found an association between CAG repeat length less than 19 and higher risk of PCa relative to more than 25 CAG repeats (relative risk, 1.52; 95% confidence interval, 0.92–2.49; P = 0.04) (80). This group also demonstrated that men with shorter repeats were at particularly high risk for metastatic PCa. Similarly, Stanford et al. observed a 3% decrease of PCa risk for each additional CAG repeat (78). The above mentioned data have been confirmed by other studies (77, 81, 82).

Several investigators have tried to correlate the expansion of the CAG repeats with a particular alteration in AR action, but the results remain ambiguous. A reduced amount of specific androgen binding in tissue of patients with Kennedy’s disease has been reported as a result of the expanded CAG repeats (83, 84), although this finding could not be confirmed by Chamberlain et al. (85). Others have indicated that expansion of the Gln repeat in the AR N-terminal region results in a structurally altered protein with reduced transcriptional capacity (85, 86, 87, 88). However, three studies demonstrated that an AR lacking the CAG repeats transactivates normally (89, 90, 91). Also, functional analysis of a CAG repeat expansion in two brothers with Kennedy’s disease revealed that hormone binding, transactivation, and transrepression potentials were identical to that of a wild-type receptor (92). An alternative molecular mechanism was raised by Choong et al. (91) who observed that CAG repeat expansion in the region of the first exon of the AR gene reduced AR mRNA and protein levels. It is obvious from these studies that definitive conclusions regarding the mode of action of CAG repeats on AR function still remain to be determined.

Interestingly, the frequency distribution of the AR gene CAG repeat length varies among different racial/ethnic groups. Shorter alleles, associated with increased risk of PCa, are found more frequently in African-American men who have a higher incidence of PCa. Conversely, populations at lower risk of PCa (Asians and whites) exhibit a relative high number of CAG repeats (73, 77) (Fig. 5). The resultant effect of any inherited polymorphism of a given gene encoding hormone receptors is present throughout life. Therefore, even polymorphisms associated with modest fluctuations in risk of PCa could explain a large proportion of the racial/ethnic differences in PCa incidence. In contrast, rare mutations associated with substantially increased risk are likely to account for a smaller fraction of these differences. To date, only one group has evaluated AR gene mutations among various ethnic populations (49). No AR gene mutations were found in clinically detected PCa samples from 26 American and 38 Japanese men.

View larger version (16K): [in this window] [in a new window]   Figure 5. Frequency distribution of the AR gene CAG repeat length in the African-American and African-Asian male population. [Data based on A. Edwards et al.: Genomics 12:241–253, 1992 (73 ).]

Other genes involved in the androgen-signaling cascade may also play a role in racial/ethnic differences of PCa risk, such as genes encoding 3ß-hydroxysteroid dehydrogenase (3ß-HD) type 2 and 5R type 2. Devgan et al. (93) reported the existence of a complex (TG)_n (TA)_n (CA)_n dinucleotide repeat in 3ß-hydroxysteroid dehydrogenase type 2 gene (HSD3B2) (93). A total of 25 different alleles were identified of which the 289-bp allele was shown to be the most common allele in all populations studied. Interestingly, a 275-bp allele was significantly more common among African-Americans, a 340-bp allele among European-Americans, and 281-bp and 302- to 334-bp alleles were found to be significantly more frequent among Asian men (93). Because the 3ß-hydroxy-steroid dehydrogenase type 2 is one of the two enzymes involved in catabolism of DHT, polymorphisms in the HSD3B2 gene may alter DHT degradation and consequently risk of PCa (93). Unfortunately, functional effects of the different allelotypes have not yet been reported.

There are two distinct 5-R genes in man, each coding a biochemically distinct enzyme: the SRD5A1 gene encodes type 1 enzyme (expressed mostly in newborn scalp, in skin, and in liver), and the SRD5A2 gene encodes type 2 enzyme (primarily expressed in genital skin and the prostate). The latter gene was cloned independently by two groups and maps to a single band on the short arm of chromosome 2 (2p23) spanning more than 40 kb of genomic DNA (94, 95).

Ross et al. (96) demonstrated that young Japanese men have lower 5-R activity than young Caucasian-American and African-American men. Similarly, Wu et al. (97) reported that the DHT-to-T ratio was highest in African-Americans, intermediate in whites, and lowest in Asian-Americans, corresponding to the respective risk of developing PCa in these groups. These studies have provided part of the rationale for using finasteride, a 5-R inhibitor, as a chemopreventive agent for PCa (98). A 5-R inhibitor acts selectively inside its target cells, and for this reason it does not lower the serum concentration of T and causes almost no adverse reactions, such as reduction of libido and potency (99).

Davis and Russell (100) reported a polymorphic three-allele system in the 3'-untranslated region (UTR) of the SRD5A2 gene. This polymorphism consists of a variable numbers of TA dinucleotide repeats. The most frequent allele found in 87% of men without PCa is (TA)₀, i.e., a chromosome lacking TA dinucleotide repeats in the amplified region (101). The remaining two alleles, (TA)₉ and (TA)₁₈, are found in a minority of the men (101). Other groups have confirmed the SRD5A2 gene polymorphisms, although the frequency of the three alleles differed slightly between the studies (100, 102, 103). The functional effects of polymorphisms in the SRD5A2 gene are not completely understood. Because the TA repeat is located in the 3'-UTR of the SRD5A2 gene, it should not affect the structure or function of the resulting protein. However, it has been proposed that certain SRD5A2 polymorphisms may encode for 5-R enzyme variants with different activities (102) that influence regulation of the 5-R enzyme production (101), probably by altering mRNA stability (104). As a result, these polymorphisms might alter DHT levels and consequently risk of PCa (102). There are currently no studies available that have addressed length of the TA repeats in parallel with intraprostatic levels of DHT. Such evaluations would more directly assess the biological implications of this polymorphism.

Reichardt et al. showed that 18% of healthy African-Americans have at least one (TA)₁₈ allele, while no such alleles have been identified in any healthy Asian-Americans or non-Hispanic whites (102). This difference in prevalence of the (TA)₁₈ alleles by ethnicity proved to be statistically significant in this study. The hypothesis that longer TA alleles are associated with an increased risk of PCa could not be confirmed in a predominantly Caucasian population (101). However, the at-risk functional allele may be in linkage disequilibrium with different marker alleles in different ethnic groups, with short alleles increasing the risk of PCa in Caucasians (101). The resultant effect of different alleles in an African-American population still needs to be determined.

The molecular rationale for racial/ethnic differences in PCa risk linked to 5-R metabolism has been investigated further by Makridakis et al. (103). This group demonstrated that a missense substitution in the SRD5A2 gene, which replaces valine at codon 89 with leucine (V89L), is distributed differentially among races. This substitution is associated with a reduced 5-R activity in vivo (103). Interestingly, African-Americans who are at particular high risk for developing PCa also have the highest frequency for valine 89 allele of the V89L amino acid substitution. Asians, the lower risk population, exhibit the lowest frequency of the valine variant and the highest frequency of the leucine variant (103). Latinos, at intermediate risk of PCa, also have intermediate frequencies of the V89L substitution. Thus, a gradient of the V89L substitution exists among racial/ethnic groups that parallels PCa risk.

Other groups have investigated whether racial/ethnic variations in hormone levels exist. Ross et al. (105) demonstrated that serum T levels were approximately 15% higher in young African-American males compared with their white counterparts. In a study by Ellis and Nyborg (106), however, serum T levels are only 3.3% higher in 525 black individuals compared with 3,654 non-Hispanic whites. Inconsistency in the magnitude of the effect between those two studies might be related to age. Ross et al. studied only college-aged men, while in the study of Ellis and Nyborg, men aged 31–50 yr were enrolled.

The study by Ellis and Nyborg (106) found that the racial/ethnic difference in serum T levels was age related. In the youngest age group (31–34.9 yr), black men had a mean T level that was 6.7% higher than white men, but it declined to 3.7% by age 35–39 and to 0.5% by ages 40–50. These observations suggest that at the time PCa evolves, T levels are higher in blacks than in whites. Higher levels of sex hormone-binding globulin (SHBG) have also been associated with a higher risk of PCa (107). Interestingly, American males were found to have higher levels of SHBG than did their Japanese counterparts, consistent with the risk of PCa among these populations (96).

C. Vitamin D and prostate carcinoma
In addition to androgenic steroids, the secosteroid, 1,25-dihydroxyvitamin D₃ [1,25-(OH)₂D₃], also plays an important role in the growth and function of the normal prostate, as well as in prostate carcinogenesis. Vitamin D, which is supplied physiologically by UV irradiation through a precursor in the skin or obtained from the diet, undergoes sequential cytochrome P450-dependent hydroxylations to form 25-hydroxyvitamin D₃ [25(OH)D₃] in the liver and 1,25-(OH)₂D₃ and 24,25-dihydroxyvitamin D₃ [24,25(OH)₂D₃] in the kidney (108). The metabolites 25(OH)D₃ (25–100 nM) and 24,25(OH)₂D₃ (2–10 nM) are both present in the plasma at much higher concentrations than the hormonally active metabolite 1,25-dihydroxyvitamin [1,25-(OH)₂D₃; 100 pM], but they have limited biological activity with respect to calcium homeostasis (108). This is because most of the actions of vitamin D are mediated through an intracellular receptor that has much higher affinity for 1,25-(OH)₂D₃ than for the other metabolites. 1,25-(OH)₂D₃ is capable of acting through both nongenomic signaling pathways involving a membrane-associated receptor and genomic pathways involving the nuclear vitamin D receptor (VDR) (109). Hormone bound to the VDR induces changes in gene expression much like the steroid receptors. In the absence of hormone, the VDR binds to vitamin D-response element in the DNA as a heterodimeric complex with retinoid X receptor. Recent studies have suggested that ligand binding promotes the dissociation of corepressors such as NCoR (nuclear receptor corepressor) and binding of coactivators such as steroid receptor coactivator-1 causing transactivation (110).

The initial hypothesis that vitamin D may have a protective effect against developing prostate carcinoma was raised by Schwartz and Hulka (111). They showed that PCa incidence increases with age, and the elderly are known to have lower levels of vitamin D. There is a high incidence of PCa in blacks, also known to have relative low levels of vitamin D (112). Also, there is an increased incidence of PCa in migrant Asians as they adopt a Western diet, i.e., lacking the large amount of vitamin D-rich fish oil (111). Lastly, US mortality rates from PCa demonstrate a negative correlation with UV radiation exposure (113). In addition to these epidemiological observations, several studies have shown that vitamin D metabolites have antiproliferative and prodifferentiating effects on PCa cell lines in vitro and in vivo (109, 114, 115, 116, 117, 118). Moreover, physiological levels of 1,25-(OH)₂D₃ significantly reduce the invasive potential of DU-145 cells (119) and reveal impressive antimetastatic effects on PCa cells in vivo (120). It should be noted, however, that different PCa cell lines may exhibit varying sensitivity to growth inhibition by 1,25-(OH)₂D₃ (117). If such variability also exists in human PCa, the effectiveness of 1,25-(OH)₂D₃ and analogs as therapeutic agents may be limited. Thus, it is critical to understand which factors influence the actions of 1,25-(OH)₂D₃ on PCa cell proliferation. Relative insensitivity to 1,25-(OH)₂D₃-mediated growth inhibition in the PCa cell lines PC-3 and DU-145 was associated with low functional VDR levels (117). Transfection of these cell lines with a VDR cDNA-expression vector induces growth sensitivity to 1,25-(OH)₂D₃. Similarly, the cell line JCA-1 fails to express a detectable number of VDRs and is not measurably affected by 1,25-(OH)₂D₃ in growth studies (109). Transfection with a wild-type VDR cDNA construct in JCA-1 cells resulted in expression of high affinity nuclear VDRs and a dose-dependent inhibition of growth by 1,25-(OH)₂D₃ (109).

These studies suggest that low VDR levels are responsible for the failure of these cell lines to respond to vitamin D. LNCaP cells are highly sensitive to growth inhibition by 1,25-(OH)₂D₃ and contain approximately 2- to 3-fold more VDR than the relatively 1,25-(OH)₂D₃-insensitive PC-3 and DU-145 cell lines. However, the PCa cell line ALVA-31 displays less than 20% growth inhibition to 1,25-(OH)₂D₃, yet it contains the highest levels of VDRs of these four cell lines (117). Thus, sensitivity of growth inhibition to 1,25-(OH)₂D₃ does not correlate with VDR content in ALVA-31 and LNCaP cells. Because VDR is capable of transactivating reporter gene expression in ALVA-31 and LNCaP cell lines, differential growth sensitivity to 1,25-(OH)₂D₃ is not likely to result from VDR or retinoid X receptor mutations or from differences in 1,25-(OH)₂D₃ metabolism (117). From the above-mentioned studies it was suggested that expression of functional VDR is necessary, but not sufficient, for maximal growth inhibition by 1,25-(OH)₂D₃ (121). Differential expression and/or regulation of the retinoblastoma pathway, the cyclin-dependent kinase inhibitor (CDKI), p21, or 1,25-(OH)₂D₃ regulation of cyclin-dependent kinase may also explain some of the differences in sensitivity to the antiproliferative effects of 1,25-(OH)₂D₃ among PCa cell lines (121, 122).

The finding that 1,25-(OH)₂D₃ demonstrates antiproliferative, prodifferentiation and antimetastatic effects could open new avenues to the development and targeting of prophylactic interventions. However, 1,25-(OH)₂D₃ may not be the suitable agent for use as a chemopreventive agent because of the risk of hypercalcemia (123). Interestingly, Schwartz et al. (124) recently demonstrated that the PCa cell lines DU-145 and PC-3 could synthesize 1,25-(OH)₂D₃ from its precursor 25(OH)D₃. Therefore, supplementation of men with 25(OH)D₃ could promote the local synthesis of 1,25-(OH)₂D₃ by prostate cells and thereby inhibit invasiveness of PCa cells (124). If the prostate cells synthesize 1,25-(OH)₂D₃ in vivo, then systemic levels of 1,25-(OH)₂D₃ measured in the serum may not reflect levels of 1,25-(OH)₂D₃ at the site of the target cell. Thus, the risk of PCa may be influenced by intraprostatic and systemic levels of vitamin D₃ (124).

This could explain, at least in part, inconsistent results reported from several prospective epidemiological studies of PCa that assessed vitamin D metabolite levels in blood samples (125, 126).

As with the androgenic steroids, the vitamin D signaling cascade might be altered by genetic changes. Although studies on VDR gene mutations in PCa have to date not been reported, a series of common polymorphisms in this gene have been identified. The alleles of human VDR gene can be distinguished by restriction fragment length polymorphisms (RFLPs) found for BsmI and ApaI (intron 8) and TaqI (exon 9) (127). The presence (b, a, t) or absence (B, A, T) of a restriction site defines the specific allele. The RFLPs are highly linked in the population. The BsmI cleavage site (b) is associated in 97% of individuals with the absence of a TaqI site (T) (128). This strong linkage disequilibrium results in two common haplotypes BAt and baT. A fourth polymorphism, a poly (A) microsatellite, is in the 3'-UTR (81, 128). In a study of non-Hispanic whites, the S (short) and L (long) alleles of the poly(A) microsatellite were 97% concordant with the BsmI alleles, with B corresponding to S and b corresponding to L (129).

Several groups have evaluated whether VDR gene polymorphisms could alter the risk of PCa (81, 129, 130). Two recent case-control studies [117 (128) and 591 controls (131)] reported significantly higher serum 1,25-(OH)₂D₃ levels among individuals who were homozygous for the BAt haplotype compared with individuals who were heterozygous or homozygous for the baT haplotype. Therefore, the BAt haplotype may have a protective effect on developing PCa. Indeed, Taylor et al. (130) estimated that men who were homozygous for the presence of a TaqI restriction site (genotype tt) have one-third the risk of developing PCa requiring prostatectomy compared with men who are heterozygotes or homozygotes for the T allele (odds ratio = 0.34; 95% confidence interval, 0.16–0.76; P < 0.01). Similarly, Ingles et al. (81) reported an odds ratio of 0.22 (95% confidence interval, 0.06–0.75) for the risk of PCa among men with the SS genotype compared with men with the SL or LL genotypes.

Ma et al. (131) could demonstrate only a 57% reduction in risk of PCa related to the BB genotype in older men with plasma 25(OH)D₃ levels below the median. Because the elderly are more likely to have restricted sunlight exposure, low vitamin D intake, and reduced capacity of the skin to produce vitamin D, any genetic influence on vitamin D levels may be stronger among older than younger men (131). Similarly, low levels of 25(OH)D₃ may result from active conversion of this metabolite into 1,25-(OH)₂D₃ by 1-hydroxylase activity. If higher 1,25-(OH)₂D₃ in the presence of lower 25(OH)D₃ is beneficial with respect to PCa, this could explain the reduced risk observed for the BB genotype among men with lower 25(OH)D₃ (131). However, in their entire study population of 372 cases and 591 controls, no significant associations between VDR gene polymorphisms and PCa risk was found (131). Differences in the control selection were probably not the reason for the divergent results, as the prevalence of the polymorphisms was similar among the three above-mentioned studies (81, 130, 131). No association between VDR polymorphisms with tumor aggressiveness was found, reducing the probability of this form of bias by case selection among the different reports (130, 131). Thus, studies that did find a difference between VDR polymorphisms and risk of PCa may have resulted from selection of men that had lower 25(OH)D₃ levels.

A major difficulty in accepting the hypothesis that VDR gene polymorphisms may alter risk of PCa is that, unlike the AR polymorphism, none of the BsmI, ApaI, and TaqI polymorphisms alter the predicted amino acid sequence of the translated VDR protein. The BsmI and ApaI polymorphisms are located in intron 8 of the VDR gene (132). Only the TaqI polymorphism is located in the coding region but leads to a silent codon change in exon 9, with ATT and ATC both coding for isoleucine (133). Therefore, it has been proposed that expression of these receptor alleles or the functional activities of their products are potential discriminating factors (81, 134). Morrison et al. (128) found that in minigene reporter constructs, in which the 3'-UTR of alleles BAt and baT were fused 3' to a luciferase reporter gene, luciferase activity was 30–40% lower when the baT construct was transfected in COS cells.

Therefore, this group proposed that T mRNA was less stable than t mRNA. Verbeek et al. (134) kept all influences equal for both VDR mRNAs transcribed either from the T or t allele by choosing a heterozygous system (Tt) as opposed to the comparison of two (TT vs. tt) homozygous cell lines. This group consistently found approximately 30% less RT-PCR product derived from the t allele than the T allele in normal lymphocytes, the hematopoietic cell line NB4, and the PCa cell lines, PC-3 and DU-145 (128). They could not demonstrate any difference in VDR mRNA stability between the two alleles in either the cell lines or the normal lymphocytes. These results indicate the existence of a variation in transcriptional regulation rather than mRNA stability between the alleles. Therefore, it has been suggested that unknown gene(s) in linkage with the polymorphisms might be responsible for the relationship between risk of PCa and VDR gene polymorphisms, and these factors merit further study.

Baker et al. (132) have cloned a second group of polymorphisms in the VDR gene. The VDR gene has two ATG translational initiation sites that are separated by three codons. A polymorphism located in the first ATG presumably determines the site of translation initiation (135). Consequently, the variants in this polymorphic site would differ in length by three amino acids. Two studies have shown that the elongated VDR allele is associated with an increased risk of osteoporosis (136, 137). However, it has not yet been proven whether this start codon polymorphism results in a change in VDR function. It will be of interest to determine whether this VDR polymorphism, which has no known relationship to the poly(A)-linked polymorphisms, is also associated with altered risk for PCa.

As for the AR gene, polymorphisms of the VDR gene have also been evaluated for their potential role in racial/ethnic differences in risk of PCa. In Caucasians, the at-risk allelic variant is associated with the bAT-poly(A)L haplotype. In a study restricted to African-American individuals, however, the poly(A)L allele genotype was not associated with risk of PCa (138).

Moreover, BsmI b alleles were associated with a 2-fold decrease in risk of advanced PCa in African-American men, in contrast to a 4-fold increase risk of PCa in a Caucasian population (81, 138). Although these results might seem inconsistent, the at-risk functional allele may be in linkage disequilibrium with different marker alleles in different ethnic groups, e.g., BsmI b in Caucasians and B in African-Americans (113).

D. Oncogenes
To date, more than 100 cellular oncogenes have been identified. Only oncogenes that have been studied more extensively for their possible role in the development of PCa will be discussed.

1. ras Oncogene. ras Oncogenes are activated by point mutations in the protooncogene. Mutations most commonly occur at codons 12, 13, or 61, thereby altering the ability of the ras protein to affect signal transduction, leading to unregulated cellular growth (139). The low frequency of ras mutations in PCa samples obtained from Western study populations (4/146; 2.7%) indicates that this oncogene is apparently not commonly involved in PCa initiation or progression (140, 141, 142, 143, 144, 145, 146). However, when PCa samples from only Japanese cases are considered, mutations in the ras gene occur in as many as 18% of the cases evaluated (25/138) (140, 141, 142, 143, 144, 145, 146). This difference in ras mutation rate may reflect a different etiology causing PCa or affecting its progression between the two populations.

2. myc Oncogene. c-myc Is a cellular protooncogene that encodes a nuclear phosphoprotein. Potential functions of c-myc include the promotion of DNA replication, regulation of the G₀/G₁ cell-cycle transition, and control of cellular differentiation (147). Few studies have examined the frequency of myc alterations in PCa (148, 149, 150, 151, 152, 153). Using Northern blot analysis, these studies have shown a significantly higher level of c-myc RNA transcripts in PCa specimens of varying grades compared with samples of benign prostatic hyperplasia (BPH) (148, 150, 151, 152, 153).

However, the Northern blot technique makes it impossible to establish which cell type gives rise to higher mRNA levels. Interestingly, such a correlation is possible with in situ hybridization, and the two studies that have performed such an analysis did not confirm a relationship between c-myc and PCa (149, 151). Thus, the role of c-myc in PCa needs to be extended by additional studies in which it is also possible to evaluate the origin of increased levels of c-myc mRNA and to determine whether the c-myc protein is simultaneously elevated.

3. erbB2 Oncogene. Localized on chromosome 17p21, the HER-2/neu oncogene encodes a 185-kDa transmembrane tyrosine kinase growth factor receptor with substantial homology to the EGFr (154). More than 100 previous reports have studied the HER-2/neu gene and its protein product in a variety of malignant tumors, indicating the significance of this gene in carcinogenesis (154). Given the significant recent interest in the use of anti-p185neu antibodies for the treatment of advanced stage refractory ovarian and breast carcinoma, the documentation of a subset of PCa patients with HER-2/neu overexpression that have a significant risk for disease progression might be of potential therapeutic interest (155). Some studies have revealed a higher protein expression or gene amplification in patients with clinically detected PCa compared with nonmalignant specimens (156, 157, 158, 159, 160), whereas others did not find such a correlation (161) or even found overexpression of the oncoprotein in BPH lesions (160). Similarly, a consensus as to the predictive value of HER-2/neu gene amplification and p185neu protein expression in PCa patients has not been reached (156, 157, 158, 159, 160, 161, 162, 163, 164). Possible explanations for the disparity in the direction and magnitude of these results include a variety of technical factors. For example, the use of various antibodies with differing sensitivities and specificities may produce either cytoplasmic or membranous staining, be ineffective when certain fixatives are used, or be impacted by the temperature of the immunohistochemical reaction (156, 165, 166).

In addition, fixation and processing protocols significantly affect the reactivity of the antigenic determinants detected by HER-2/neu antibodies, such as MAB-1 and pAB-60 (156, 165, 167). Antigen retrieval techniques featuring either enzymatic digestion or microwave irradiation contribute additional variables that may affect staining levels (156). Finally, interobserver variability of staining interpretation has also been reported (168). It has recently been demonstrated that fluorescence in situ hybridization (FISH) is more sensitive than immunohistochemistry for detection of abnormalities in the HER-2/neu gene, and future studies should be undertaken to determine whether a FISH-based HER-2/neu detection method may prove to be important in the prediction of prognosis in PCa patients (169).

E. TSGs
Malignant progression not only involves the enhancement of oncogene-derived transforming functions, but also requires the loss of regulatory functions of TSGs. Mutations of TSGs generally are considered recessive, and both copies of these genes must be inactivated before the cell is at risk for transformation. The first mutation of the gene is a somatic event or is inherited in the germline from one of the parents. The second hit is likely to inactivate the normal copy or allele of the gene that functioned as a ’repressor’ of the mutant allele.

1. Loss of heterozygosity (LOH) studies. Sites of consistent chromosomal deletion resulting from LOH studies may uncover the location of TSG. Several groups have investigated the frequency of LOH at different chromosomal regions, and their results are summarized in Fig. 6 (170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199). Interestingly, chromosome 8p shows frequent LOH in a variety of other common human tumors including colorectal and lung cancer. Whether the same gene is a target of these LOH events in prostate, colon, and lung cancer is unclear, but it is interesting to speculate that a gene that is inactivated in a large fraction of human cancers may reside in this region of the genome (28).

View larger version (14K): [in this window] [in a new window]   Figure 6. The average reported rate of LOH on several chromosomal arms in localized and advanced prostate carcinoma (after failure of androgen ablation therapy or metastatic disease).

Molecular-based studies of Rb alterations in PCa have been limited. Performing single-stranded conformation polymorphism analysis of RNA, Kubota et al. (206) revealed that 4 (16%) of 25 human primary PCa samples had Rb alterations. Several other groups have evaluated genetic alterations in the region of the Rb gene using polymorphic markers; reported LOH rates in primary PCa varied between 27% and 60% (191, 192, 193, 207). Interestingly, LOH at Rb have also been found in BPH samples (191, 208). It is unclear from those latter results whether Rb alterations also have a role in the genesis or progression of BPH.

Absence of Rb protein (pRb) by immunohistochemistry has been reported to vary between 17% and 78% (191, 192, 193, 205, 209). Several difficulties remain regarding the value of pRb alterations in PCa, such as the lack of complete correlation between LOH at Rb and absent pRb expression (192, 193). For example, one study demonstrated that three PCa samples without LOH at Rb, using an intragenic microsatellite marker, did not express pRb (193). Discordances between LOH at Rb and pRb expression might be due to the large size of the Rb gene (27 exons spread over 200 kb of genomic DNA), which poses problems for the detection of small deletions. It is also difficult to determine in some case whether the low rate of loss of pRb staining in tumors with LOH on 13q is caused by the presence of stable mutant pRb or the involvement of a second TSG closely linked to Rb. In addition, the evaluation of pRb expression may be confounded by fixation artifacts and by variables in the immunohistochemical reaction, such as sensitivity and specificity of the antibody and interpretation of the staining results (192, 193). Whether the reduction of pRb staining has any significance as to the prediction of the malignant potential of a given carcinoma lesion is a topic that requires further study in a well characterized population of PCa patients.

3. p53 Gene. Located at 17p13.1, the p53 gene encodes a 53-kDa phosphoprotein that negatively regulates cell growth (210). When DNA damage has occurred, the p53 gene will either induce growth arrest via activation of a downstream inhibitor pathway or commit the cell to a programmed death pathway depending on the extent of damage. Inactivation of this gene contributes to the genesis and progression of numerous human cancers, including these of the breast, colon, and lung (211). Several methods can be used to screen for p53 alterations, such as single-stranded conformation polymorphism of exons 5–8 (where the majority of mutations within the p53 gene are known to occur) and immunohistochemistry. The latter method exploits the increase in biological half-life of many mutations of the p53 protein (TP53) (212).

Such an increase in half-life results in the intranuclear accumulation of detectable quantities of the TP53 by a number of commercially available antibodies. While TP53 accumulations as determined by immunohistochemical detectable intranuclear staining and p53 gene mutations are not synonymous terms, most studies support a close relationship between them (213, 214).

The frequency of p53 mutations in primary PCa ranges from 1–42% (48, 143, 196, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222). LOH studies have shown that deletions in the region of the p53 gene occur in approximately 16% of primary, nonmetastatic PCas (range 0–50%) (215, 169, 170, 171, 178, 183, 186, 196). Accumulation of TP53 by immunohistochemistry has been demonstrated in between 2% and 22% of the primary PCa samples (143, 196, 213, 214, 220, 223, 224, 225, 226). The only exception is a study by Veldhuizen et al. (227) in which TP53 accumulation was detected in as many as 79% of the cases. This study did not differ from other reports in the grade and stage composition of the specimens analyzed. Unfortunately, mutation analysis to confirm these results was not performed. This might be important to know since the predominant cytoplasmic pattern of TP53 staining identified in this latter study is not necessarily associated with p53 gene mutations. For example, immunohistochemistry may detect a wild-type TP53 protein that has been stabilized as a result of DNA damage induced by exogenous factors (27).

The disparity in the above-mentioned data with respect to the rate of p53 alterations may result from various factors. Interpretation of TP53 staining is prone to interobserver variability. Therefore, quantitative measurements are important to more objectively relate the number of nuclei with protein accumulation to variables such as grade, stage, and follow-up data of the patient. Another factor is that commercially available antibodies may differ in sensitivities and specificities. Similarly, variation in p53 gene mutation rates may result from using different techniques, such as temperature- or denaturating gradient gel electrophoresis or single-stranded conformation polymorphisms.

In addition to these technical factors, variations in reported incidence of p53 alterations also result from heterogeneity in the topographical distribution of the alterations (222, 226). This implicates that the outcome of incidence of any p53 alteration would depend not only on which tumors but also which regions of the tumors are being analyzed.

A consistent observation from the vast majority of the studies is that the predictive value of p53 depends on its association with high-grade and advanced stage PCa (27, 214, 217, 224, 225, 228). Consequently, it is anticipated that p53 is not useful as a prognostic marker in most patients with PCa (196, 229). However, it is important to consider that alterations in the p53 gene may play a role in the development of radiation resistance (230, 231, 232, 233). Several regulatory genes influencing the cellular commitment to apoptosis have been identified in various tumors, including bcl-2, p53, c-myc, and ras. It has been speculated that defects in such regulatory genes may limit the ability of the cell to induce programmed cell death, resulting in resistance to agents that exert cytotoxicity via induction of apoptosis (230). A number of studies have shown that the lethal effects of ionizing radiation are attenuated in certain cells harboring mutations in the p53 gene (231, 232, 233). Although not all studies are supportive (234), genetic alteration of the p53 gene may thus be used as a pretreatment marker to predict local treatment failure with ionizing radiation (230, 234). Similarly, p53 gene mutations have been shown to adversely predict overall survival for patients treated with androgen ablation therapy in locally advanced cancer of the prostate after radiotherapy (235). This suggests that induction of apoptosis after androgen ablation may also be blocked in patients whose tumors have p53 gene mutations.

F. Metastasis suppressor genes (MSGs)
Acquisition of metastatic ability is the ultimate biological hallmark for the aggressiveness of PCa. Genetic control of metastasis may be exerted by the increased expression of specific genes involved in transformation to the metastatic phenotype. On the other hand, deactivation of specific MSGs may promote the evolution of metastasis in a manner analogous to the action of TSGs on tumor initiation. At present, there is evidence suggesting several MSGs may play a role in PCa, of which E-cadherin, KAI1 and CD44, and PTEN will be reviewed.

1. E-cadherin. The frequent finding of LOH at chromosome 16q in PCa (Fig. 6) led to an examination of putative candidate genes in this region. Although 16q reveals three commonly deleted regions (195), the E-cadherin gene located on 16q22.1 is of particular interest. E-cadherin is involved in developmental morphogenesis and maintenance of the epithelial phenotype by mediating epithelial cell-cell recognition and adhesion processes (236). The role of this Ca²⁺-dependent cellular adhesion molecule in maintaining various aspects of epithelial cell differentiation has led a number of investigators to examine E-cadherin expression in carcinomas. The functional importance of E-cadherin in preserving epithelial tissue integrity was demonstrated by using antibodies against E-cadherin, leading to dissociation of epithelial cell layers in cell culture and to an increased invasive potential of cells (237). Increased invasive behavior can also be induced by introducing E-cadherin antisense transcripts into cells (238). Conversely, transfection of E-cadherin into E-cadherin-negative cell lines can result in reversion to a noninvasive phenotype (238, 239, 240). These data have provided the rationale to further investigate the role of E-cadherin in the development and progression of PCa. Relevant approaches include immunohistochemistry, mutation analysis, and LOH studies on tumor samples. Unfortunately, E-cadherin immunohistochemistry on archival tissue is often unreliable due to a combination of formalin fixation and autolytic artifacts (241). Therefore, the studies that have evaluated E-cadherin immunoreactivity on paraffin material should be interpreted with care.

A recently described alternative approach for fixing prostatectomy specimens combines injection of formalin into the prostate at multiple sites and microwave-stimulated heating of the specimen to 50 C (241). With the use of this modified fixation technique, immunohistochemical stains on paraffin tissue equals those on frozen sections. Moreover, much longer fragments of DNA can be extracted from the tissue blocks, which expands the possibilities for molecular analysis (241).

Retrospective studies using frozen PCa samples have revealed a strong correlation between decreased E-cadherin staining and increased histological grade (242, 243, 244), advanced clinical stage (242, 243, 244), and the presence of metastasis at diagnosis (244, 245). Of clinical importance is the finding that aberrant E-cadherin expression proved to be a powerful predictor of adverse prognosis (243, 245). The finding of two PCa metastases with normal E-cadherin expression (242) does not necessarily contradict the role of E-cadherin as a MSG because it is possible that E-cadherin is only transiently down-regulated (246, 247).

It is important to clarify the mechanisms by which E-cadherin function is regulated. Interfering with biochemical pathways governing epithelial differentiation and maintenance of epithelial integrity might provide new therapeutic approaches. So far, E-cadherin mutations in PCa samples have been reported in two studies (195, 248). Schalken et al. (248) reported one (4%) mutation in 28 human prostate and bladder carcinoma samples that would lead to a change in the E-cadherin polypeptide sequence. Suzuki et al. (195) found an equal frequency of E-cadherin gene mutations in 48 primary PCa samples that were all found to be silent mutations by sequencing.

Considering the infrequent mutational inactivation, defective E-cadherin function is more likely to result from posttransciptional alteration or due to loss of or reduced contact with the cytoskeleton. Interactions between the cytoskeleton and adhesion molecules have been shown to be essential for a variety of cellular functions, including cell-cell and cell-matrix interactions and cell motility. Normally, the highly conserved cytoplasmic domain of E-cadherin associates with three independent proteins, called -, ß-, and -catenin (249).

Significant changes in the cadherin-catenin complex have been shown to disturb the junctional complex, thereby affecting the greater motility of invasive cells. For example, the PCa cell line PC-3 is characterized by the presence of E-cadherin, but the complete lack of -catenin both at the RNA and protein level due to a homozygous deletion in the -catenin gene (250). Transfection of these cells with an -catenin expressing vector has been shown to induce the reversion to a typical epithelial morphology in PC-3 cells (251). The -catenin gene is mapped to chromosome 5q22 (252). LOH in this region occurs in approximately 20% of the PCa specimens (Fig. 6). Two studies have performed an immunohistochemical evaluation of -catenin in PCa specimens (243, 253), showing that in a small number of specimens immunohistochemistry predicted invasiveness more accurately than E-cadherin expression. To date, ß- and -catenin immunohistochemistry have not been studied in PCa. Only one study has evaluated the rate of alterations in the ß-catenin (CTNNB1) gene in 104 clinically localized PCa samples, finding only five mutations (254). The recent observation that adenomatous polyposis coli gene products interact with ß-catenin suggests a potential novel mechanism through which a TSG may alter regulation of cell-cell adhesion (255).

A third mechanism resulting in defective E-cadherin function is DNA m