Molecular biomarkers in urothelial bladder cancer

Abstract

Bladder cancers are a mixture of heterogeneous cell populations, and numerous factors are likely to be involved in dictating their recurrence, progression, and the patient's survival. For any candidate prognostic marker to have considerable clinical relevance, it must add some predictive capacity beyond that offered by the conventional clinical and pathological parameters. None of the biomarkers reported to date have shown sufficient sensitivity and specificity for detecting the whole spectrum of bladder cancer diseases in routine clinical practice. The limitations of established prognostic markers requires us to identify better molecular parameters that could be of interest in predicting the prognosis of bladder cancer patients, in particular, the high‐risk patient groups that are at risk of progression and recurrence. Recent progress in epigenetic modification and gene silencing opened a new avenue for the identification of epigenetic markers, which appears to be more useful for cancer diagnosis and prognosis. Although epigenetic markers also have limitations, the combined epigenetic marker approach may increase sensitivity and reliability. The epigenetic silencing of tumor‐suppressor genes is interesting from a clinical standpoint because of the possibility of reversing epigenetic changes and restoring gene function in a cell. In addition, microarray technology provides us with additional tools for the analysis of global gene‐expression analysis of tumor samples. Future microarray analyses are likely to reveal particular gene‐expression signatures that predict the likelihood of bladder cancer progression and recurrence, as well as a patient's survival and responsiveness to different anticancer therapies, with great specificity and sensitivity. (Cancer Sci 2008; 99: 646–652)

More than 90% of bladder cancers are transitional‐cell carcinomas and approximately 60% of these are low‐grade, non‐muscle‐invasive transitional‐cell carcinomas. After endoscopic resection, the majority of bladder cancer patients develop cancer recurrences, 16–25% of which are high‐grade cancers. Approximately 10% of patients with non‐muscle invasive bladder cancers subsequently develop invasive or metastatic cancer. Almost 25% of patients with newly diagnosed bladder cancer have muscle‐invasive disease, the vast majority being cancers of high histological grade. Almost 50% of patients with muscle‐invasive bladder cancer already have occult distant metastases.^{( 1 )} The frequent recurrence of non‐muscle invasive bladder cancer and subsequent cancer progression after transurethral resection are therefore a problem for both patients and urologists.

Currently, patients with bladder cancer are monitored for cancer recurrence or progression by periodic cystoscopy and urine cytology, the frequency of which varies depending on the risk factors associated with the disease. Cystoscopic examination is associated with high cost, substantial patient discomfort, and variable sensitivity. In addition, urine cytology has poor sensitivity in detecting low‐grade disease, and its accuracy is dependent on the pathologist's experience. More sensitive and non‐invasive methods are therefore required. Many urine‐based tumor markers have been developed for the detection and monitoring of bladder cancer; indeed, the Food and Drug Administration (FDA) has already accepted some of these tumor‐marker tests for use in routine patient care.^{( 2} , ³ , ⁴ , ⁵ , ^{6 )} Although initial studies with the new markers are mostly promising, successive reports often fail to show comparable results.^{( 7 )}

One challenge for the urologist is to develop rational surveillance protocols that provide cost‐effective, non‐invasive monitoring for low‐risk patients, while using a more active approach to identify high‐risk refractory cancers before they metastasize. To date, numerous potential markers from patient's serum, bladder washes, urinary specimens, and cancer tissues have been identified by a variety of molecular biology and genetic studies. Molecular markers such as Ki‐67 and TP53 do appear to have some promising correlations with bladder cancer development, but their predictive value remains to be verified conclusively. The present review focuses on the recent advances in transitional bladder cancer research with respect to the identification of suitable molecular prognostic markers.

Conventional biomarkers in urine

In cases of bladder cancer, life‐long surveillance is required to detect subsequent tumor recurrence. Current patient‐monitoring protocols generally consist of cystoscopic evaluations and urine cytology every 3–4 months for the first 2 years and at longer intervals in subsequent years. Voided urine cytology is a highly specific, non‐invasive adjunct to cystoscopy. It has good sensitivity for the detection of high‐grade bladder cancers but poor sensitivity for low‐grade disease. Thus, non‐invasive, objective, and accurate biomarkers are needed not only for the primary detection of bladder cancer, but also for the surveillance of the disease. The recent emergence of sensitive markers for bladder cancer has provided new opportunities for early bladder cancer detection. The FDA in the USA has already accepted urinary tests for monitoring patients with bladder cancer, including the bladder tumor antigen stat test, bladder tumor antigen TRAK test, the fibrinogen–fibrin degradation products test, UroVysion, ImmunoCyt, and the nuclear matrix protein 22 assay (Table 1).^{( 2} , ³ , ⁴ , ⁵ , ^{6 )} Unfortunately, none of the biomarkers reported to date have shown sufficient sensitivity and specificity for the detection of the whole spectrum of bladder cancer diseases in routine clinical practice.^{( 7 )}

Table 1.

Currently available urinary markers for bladder cancer

Test	Marker	Sensitivity (%)	Specificity (%)
BTA stat ( 2 , 3 )	Human complement factor H‐related protein	60–70	50–75
BTA TRAK( 2 , 3 )	Human complement factor H‐related protein	60–70	50–75
FDP( 4 )	FDP	78–91	75–90
UroVysion( 5 )	Chromosomal probes	70–100	90
ImmunoCyt( 6 )	High molecular weight CEA and mucins	70–95	70–85
NMP22( 3 , 4 )	Nuclear mitotic apparatus protein	60–75	70–85

BTA, bladder tumor antigen; CEA, carcino embryonic antigen; FDP, fibrin degradation product; NMP, nuclear matrix proteins.

Chromosomal markers

Cytogenetic studies have identified many changes in the structure and copy number of chromosomes in transitional‐cell carcinomas of the urinary bladder. For example, loss of heterozygosity (LOH) studies have shown that the loss of 17p, 3p, 13q, 18q, or 10q is found more frequently in high‐grade, high‐stage bladder cancer.^{( 8 )} Moreover, loss of 9q was observed in low‐ as well as high‐grade bladder cancer, which suggests that the loss of 9q may be a primary event in the genesis of bladder cancer.^{( 9 )} In non‐muscle invasive bladder cancers, early changes include deletions of 11p and 8p, and gains of 8q and 1q.^{( 10 )} These non‐muscle invasive papillary lesions do not exhibit a LOH of 17p, which is detected in 60% of invasive cancers.^{( 11} , ^{12 )} Thus, the LOH of 17p may participate in the progression of bladder cancer. The prognostic value of these changes in chromosome copy number requires further large prospective studies on patients with bladder cancer. Indeed, it has been suggested that although copy number alterations have been associated with progression‐free survival, they are not independent prognostic factors for disease progression.^{( 13 )} Many tumors also bear DNA replication errors (RER), where bases in simple mononucleotide or dinucleotide repeat sequences (microsatellites) have been added or deleted, when compared to the matching DNA from normal tissue specimens.^{( 14 )} These RER, which result in so‐called microsatellite instability, arise from the dysfunction of the DNA mismatch repair genes (hMSH2, hMLH1, hMSH6, PMS1, and PMS2). One study has shown that microsatellite alterations in bladder cancer are associated with invasive cancer;^{( 15 )} however, this finding requires further verification. Thus, the search for microsatellite instability can only complement current diagnostic methods.

Genetic markers

Many of the genetic markers that are associated with bladder cancer have been subjected to extensive studies examining their biological roles in bladder cancer development and progression. We focus on the more promising prognostic markers, including protooncogenes and oncogenes, tumor‐suppressor genes, cell cycle regulators, and cell‐adhesion molecules. The potential prognostic values of genetic markers are shown in Table 2.^{( 16} , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ^{39 )} In summary, the observations described above reveal a number of significant correlations between various molecular markers and tumor progression. In particular, p53, Ki‐67, retinoblastoma (Rb), epidermal growth factor receptor (EGFR), E‐cadherin, and several cyclins appear to be of prognostic value with regard to bladder cancer metastasis, recurrence, and overall and cancer‐specific survival. None of these markers are currently being utilized in clinical practice and their usefulness as independent prognosticators remains to be confirmed by large prospective comparative studies. More reliable and accurate methods to detect these markers should also be developed to improve their utility in the clinic.

Table 2.

Potential genetic markers in bladder cancer

Class	Marker	Locus	Method	Potential prognostic value
Protooncogenes and oncogenes	EGFR( 16 , 17 )	7p12	IHC	High grade, high stage
	HER‐2/neu (c‐erb‐B2)( 18 , 19 )	17q21.1	IHC	High grade, high stage, poor survival, metastasis
	FGFR3( 20 , 21 )	4p16.3	PCR	Low grade, low stage, prognosis (recurrence, progression, survival)
	c‐myc( 22 )	8q24	IHC	No association with recurrence, progression, or survival
Tumor‐suppressor genes	p53( 23 , 24 )	17p13.1	IHC	High stage, prognosis (recurrence, progression, survival), resistance to chemotherapy
	Rb( 25 , 26 , 27 )	13q14.2	IHC	High stage, prognosis (recurrence, progression, survival)
Cell cycle regulators	p21( 28 )	6p21.2	IHC	High stage, prognosis (recurrence, survival)
	p27( 29 )	12p13.1	IHC	High grade, survival
	Ki‐67( 30 )	10q25	IHC	Progression, recurrence
	Cyclin D1( 31 , 32 )	11q13	IHC	Low grade, low stage, recurrence
	Cyclin E( 33 , 34 )	19q12	IHC	Low stage, survival
Cell‐adhesion molecules	MMP‐2( 35 )	16q13	PCR	High stage, survival
	E‐cadherin( 36 , 37 )	16q22.1	IHC	High grade, high stage, progression, survival
	CD44( 38 , 39 )	11p13	PCR	High stage, survival

EGFR, epidermal growth factor receptor; FGFR, fibroblast growth factor receptor; HER, human epidermal growth factor receptor; IHC, immunohistochemistry; MMP, matrix metalloproteinase; PCR, polymerase chain reaction; Rb, retinoblastoma.

Genetic polymorphism

Although the identification of major cancer susceptibility genes such as VHL and WT1 has provided important insights into the molecular basis of cancer, germline mutations in such genes are rare and account for an increased cancer risk in only a small segment of the population. In contrast, common polymorphic variants, which alone do not cause cancer, can in certain environments or in concert with other genetic alterations influence cancer development or progression. These minor susceptibility genes are expected to be prevalent in the population at large and therefore have the potential to influence disease progression in a large percentage of the population. This concept has been substantiated in a variety of cancer settings, particularly by the genetic epidemiology of bladder cancer. Various enzymes, cytokines, and the gene repair system may be involved in the carcinogenesis, recurrence, and progression of bladder cancer. Only a few genetic polymorphisms give us marginal information about a patient's prognosis, in spite of bladder cancer studies on numerous single nucleotide polymorphisms, including N‐acetyltransferase 2, tumor necrosis factor‐α, vascular endothelial growth factor, glutathione S‐transferase‐µ, glutathione S‐transferase‐θ, and human 8‐oxoguanine DNA glycosylase 1 (Table 3).^{( 40} , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ^{48 )} Although low‐penetrance predisposing polymorphisms have been reported, very few consistent gene–disease associations have been identified. It is hoped that the ongoing worldwide efforts in obtaining large and informative DNA collections, combined with the rapid development of high‐throughput genotyping technologies, will provide useful prognostic markers for clinicians that are applicable in clinical settings.

Table 3.

Genetic polymorphisms and the risk of bladder cancer development

Gene	Polymorphism	Odds ratio
p53 ( 40 )	Arg72Pro	4.7
CCND1 ( 41 )	241 A/G	1.8
NQO1 ( 42 )	609 C/T	1.6
CYP2E1 ( 42 )	C1019 (c1/c1)	1.8
GPX1 ( 43 )	Pro198Leu	2.6
CDH1 ( 44 )	–160 C/A	3.5
GSTM1 ( 45, 46, 47, 48 )	Null‐genotype	1.6–2.8
GSTT1 ( 45, 46 )	Null‐genotype	0.6–2.6
GSTP1 ( 48 )	313 A/G	2.4
hOGG1 ( 46 )	Ser326Cys	2.0

Epigenetic markers

Tumorigenesis is a multistep process that results from the accumulation and interplay of genetic and epigenetic mutations. Whereas genetics refers to the study of information inherited on the basis of gene sequence, epigenetics is the study of inherited reversible changes in gene function or other cell phenotypes that occur without any change in DNA sequence. DNA methylation is an epigenetic mechanism used for the long‐term silencing of gene expression. It can maintain differential gene‐expression patterns in a tissue‐specific and developmental‐stage‐specific manner. DNA methylation occurs throughout the genome and involves the addition of a methyl group to the cytosine ring of the CpG dinucleotide.^{( 49 )} Transcriptional repression by DNA methylation is mediated by changes in chromatin structure. Specific proteins bound to methylated DNA recruit a complex containing transcriptional corepressors and histone deacetylases.^{( 50 )} Histone deacetylation results in chromatin compaction and hence transcriptional inhibition.

Inactivation of gene expression by the abnormal methylation of CpG islands can act as a ‘hit’ for the generation of cancer.^{( 49} , ^{51 )} Thus, the alteration of DNA methylation in CpG islands is emerging as a key event in the inheritance of transcriptionally repressed regions of the genome. Many tumor‐suppressor genes contain CpG islands and show evidence of methylation‐specific silencing. Recent studies have reported that DNA hypermethylation exists in various other cancers, including gastric cancer, hepatocellular carcinoma, and breast cancer.^{( 52} , ⁵³ , ^{54 )} The hypermethylation of CpG islands around the promoter region and decreased expression of the tumor suppressor genes RUNX3, RASSF1A, p16, RARβ, and E‐cadherin have been reported in transitional‐cell carcinomas of the bladder.^{( 55} , ⁵⁶ , ⁵⁷ , ⁵⁸ , ^{59 )} In some of these tumors hypermethylation is associated with a LOH, whereas in others it affects both alleles. It has been recently recognized that aberrant hypermethylation events can occur early in tumorigenesis, predisposing cells to malignant transformation. Furthermore, promoter hypermethylation of CpG islands is strongly associated with tumor development, stage, recurrence, progression, and survival in transitional‐cell carcinomas of the urinary bladder.^{( 55} , ^{59 )} The presence of methylation has been significantly associated with advanced‐stage carcinomas, high tumor progression rates, and increased mortality rates compared to tumors without methylation. These findings strongly suggest that specific patterns of promoter hypermethylation are associated with bladder cancer. The methylation status could therefore be useful as a diagnostic and prognostic marker for bladder cancer and could also be used as a therapeutic target for bladder cancer in the clinical setting (Table 4).^{( 55} , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ , ^{61 )}

Table 4.

Potential epigenetic markers in bladder cancer

Study	n	Method	Methylation marker	Methylation rate (%)		Potential prognostic value
				Tissue	Urine
Catto et al.( 55 )	280	MSP	RASSF1A	59		High grade, high stage, progression, survival
			DAPK	6
Chan et al.( 56 )	98	MSP	DAPK	58.2	45.5	No association with grade and stage
			RARβ	87.8	68.2	Detection of methylation in urine is more sensitive than urine cytology
			E‐cadherin	63.3	59.1
			p16	26.5	13.6
Dulaimi et al.( 57 )	45	MSP	APC		69	No association with grade and stage
			RASSF1A		51	Detection of methylation in urine is more sensitive than urine cytology
			p14ARF		35
Friedrich et al.( 58 )	125	QMSP	DAPK	NA	21.6	High grade, high stage
						Promising tools for non‐invasive detection of bladder cancer
			BCL2	52	64.9
			TERT	25.2	51.4
Kim et al.( 59 )	124	MSP	RUNX3	73		Tumor development, recurrence, progression
Maruyama et al.( 60 )	98	MSP	CDH1	36		Poor prognosis, survival
			RASSF1A	35
			APC	35
			CDH13	29
			FHIT	16
Yates et al.( 61 )	96	QMSP	RASSF1A	54.2		Reliable predictor of tumor progression
			E‐cadherin	40.6
			TNFRSF25	75.0
			EDNRB	66.7
			APC	31.3

MSP, methylation‐specific polymerase chain reaction; NA, not available; QMSP, quantitative methylation‐specific polymerase chain reaction.

Amongst the known diagnostic and prognostic markers for bladder cancer, RUNX3 is particularly interesting. Kim et al. demonstrated that methylation of the RUNX3 promoter sequence confers a 100‐fold increased risk of developing bladder cancer.^{( 59 )} RUNX3 methylation also appears to be positively associated with bladder cancer stage, recurrence, and progression, which suggests that RUNX3 not only inhibits cancer initiation but also suppresses the aggressiveness of primary bladder cancers. RUNX3 is a member of the RUNX family of genes, which play critical roles in cell specification during development and in neoplastic transformation. RUNX1 and RUNX2 are required for hematopoiesis and osteogenesis, respectively, and are genetically altered in leukemia and bone disease. RUNX3 functions as a tumor suppressor.^{( 59} , ^{62 )} The tumor‐suppressor activity of RUNX3 is associated with tumor growth factor (TGF)‐β signaling as primary gastric epithelial cells isolated from Runx3‐deficient (Runx3 ^−/–) mice were less sensitive to the growth‐inhibitory effect and apoptosis‐inducing activity of TGF‐β. Physical interaction between RUNX3 and small mother against decapentaplegic, the downstream mediators of TGF‐β signaling, is required for TGF‐β‐induced transcriptional upregulation of p21 during growth arrest and Bim during apoptosis. Inactivation of RUNX3 is also associated with various cancers, including lung, colon, pancreas, liver, prostate, bile duct, breast, larynx, esophagus, endometrium, uterine cervix, testicular yolk sac, as well as bladder cancer.^{( 59 )} Recently, RUNX3 was also identified as one of the five most informative genes for the CpG island methylator phenotype of colorectal cancer.^{( 63 )}

Detection of epigenetic alterations from urine

Some epigenetic events occur early in the disease process; therefore, molecular diagnosis may allow their detection before symptomatic or overt radiographic manifestations. Thus, from a clinical point of view the most promising application for methylation analysis is the detection of cancer or the utilization of methylation as a prognostic marker. As promoter hypermethylation occurs frequently in bladder cancer,^{( 55 )} several authors have investigated its presence in exfoliated urinary cells or tumor tissues.^{( 56} , ⁵⁷ , ⁵⁸ , ^{64 )} The screening of bodily fluids such as the urine may ultimately provide a truly non‐invasive diagnostic modality, thereby limiting the need for the current imaging techniques that provide anatomical details without a definitive pathological correlation. The hypermethylation of several gene promoters has been reported in DNA isolated from bladder cancer tissues and urine sediment (Table 4).^{( 55} , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ , ^{61 )} These studies revealed that the detection of aberrant promoter methylation in urine is feasible and appears to be more sensitive than conventional cytology. Recently, the feasibility of detecting DNA hypermethylation in voided urine and its potential role as tumor marker for bladder cancer was reported.^{( 65 )} In this study, a quantitative real‐time polymerase chain reaction (PCR) assay was introduced to examine the urine sediment DNA from 175 patients with bladder cancer and 94 age‐matched control subjects for the promoter hypermethylation of nine genes (APC, p14^ARF , CDH1, GSTP1, MGMT, CDKN2A, RARβ2, RASSF1A, and TIMP3). Compared to conventional methylation‐specific PCR, the quantitative analysis of the PCR products was critical for the reproducible interpretation of the results; this quantitative methylation‐specific PCR assay provided a highly sensitive automated approach for the detection of methylated alleles. The combined methylation analysis of four genes (CDKN2A, p14^ARF , MGMT, and GSTP1) displayed 69% sensitivity and 100% specificity. Similarly, another group has also examined the methylation status of tumor‐suppressor genes (APC, RASSF1A, and p14^ARF ) in matched sediment DNA from urine specimens obtained before surgery from 45 bladder cancer patients, as well as normal and benign controls.^{( 57 )} Hypermethylation of at least one of the three suppressor genes was found in the matched urine DNA from 39 of 45 patients (87% sensitivity; 100% specificity), including 16 cases that had negative cytology. Hypermethylation (91%) was a more common finding than positive cytology (50%) in urine. Thus, the detection of DNA hypermethylation in voided urine is promising in terms of the early detection and surveillance of bladder cancer. This study has raised questions about whether the combined methylation marker approach increases the sensitivity but decreases the specificity of the assay and increases the cost. Only an extension of the selected methylation marker panel might result in a higher sensitivity and specificity in the methylation analysis of the urine, therefore making this a promising non‐invasive diagnostic and monitoring tool for bladder cancer detection.

Epigenetic alteration and drug development

The potential reversibility of DNA methylation patterns suggests a promising target for cancer treatment and this forms the basis of epigenetic therapy. Treating cancer with DNA methylase inhibitors or histone deacetylase inhibitors has, in a number of cases, been effective in reactivating tumor‐suppressor genes, which in turn reduce cancer cell proliferation.^{( 66} , ^{67 )} Unlike many tumor suppressors such as p53, which are inactivated mainly by deletions and mutations, hypermethylation is unique in that it is inactivated primarily by epigenetic silencing. A hypermethylated gene can be reactivated and is therefore considered to be a good drug target.^{( 68 )} Epigenetic drugs such as DNA methylase inhibitors or histone deacetylase inhibitors can restore the activity of genes and target aberrantly heterochromatic regions, ultimately leading to the reactivation of tumor‐suppressor genes and other genes that are crucial for normal cell function. These drugs can be used therapeutically either alone or in combination with other therapeutic modalities such as chemotherapy, immunotherapy, or radiation therapy.

Gene‐expression profiles as promising bladder cancer markers

New high‐throughput microarray technologies have made it possible to gain a comprehensive insight into the molecular basis of many human diseases.^{( 69 )} In particular, searching for the genes that are differentially expressed in human cancers has been greatly facilitated by DNA microarray technology. With this technology, the RNA expression levels of hundreds or even thousands of genes in a tumor can be surveyed simultaneously. Tissue microarray technology has also been developed to enable a large‐scale molecular histological analysis of hundreds of tumor samples. Oligonucleotide arrays, such as those manufactured by Affymetrix (http://www.affymetrix.com) or Illumina (http://www.illumina.com), represent some standardized approaches that are being widely accepted. These microarrays consist of grids bearing thousands of oligonucleotides that have been selected from known sequences by designing algorithms that choose probes that hybridize to their complements with high affinity and specificity.^{( 70 )} Various types of microarrays have been developed to analyze human gene expression. These microarrays represents genome‐wide transcriptional coverage of the well‐characterized National Center for Biotechnology Information (NCBI) Reference Sequence (RefSeq; Release 4) genes. These RefSeq genes are validated, annotated, and curated on an ongoing basis by NCBI staff and their collaborators. These microarray technologies are and will be of considerable value in the field of cancer diagnosis. In particular, microarray gene expression analysis is likely to greatly facilitate the identification of molecular prognostic markers that correlate with the outcome of particular bladder cancers.

Bladder cancer frequently occurs as a multifocal disease involving several simultaneous tumors scattered over the bladder. Modlich et al. have reported that multifocal non‐muscle‐invasive bladder cancers always cluster together, which suggests they share a common genetic background.^{( 71 )} The clonality of multifocal bladder cancer has also been reported by several other studies that are based on comprehensive LOH analyses.^{( 72} , ^{73 )} Tissue RNA expression microarrays of the multifocal tumors would be useful in confirming the genetic homogeneity of multifocal tumor cells.

Bladder cancers can also present as invasive cancers; these are histologically complex tissues that consist of a variety of cell types other than carcinoma cells. As with other methods, tissue RNA expression microarrays determine the expression levels of genes irrespective of the cell type. Despite this, tissue RNA expression microarray analyses of invasive bladder cancers may also offer some important clinical information. This is due to the fact that tumor development and progression are dependent on interactions between tumor cells and the host microenvironment. With regard to the microarray studies that have been carried out in the bladder cancer field to date, Sanchez‐Carbayo et al. have used cDNA microarrays to facilitate the hierarchical clustering of non‐muscle‐invasive and invasive bladder cancers.^{( 74 )} This technology allowed them to separate carcinoma in situ from papillary non‐muscle‐invasive lesions, and to identify subgroups within non‐muscle‐invasive and invasive cancers that differ in overall survival. The most extensive expression profiling study of bladder cancers reported to date is by Dyrskjot and coworkers, who identified clinically relevant subclasses of bladder cancer on the basis of their expression microarray profiles.^{( 75} , ^{76 )} Cluster analysis identified three major stages, namely, Ta, T1, and T2‐4, with the Ta tumors being further classified into subgroups. In particular, a 32‐gene molecular classifier set of genes could be used to classify benign and muscle‐invasive cancers with a close correlation to the pathological staging. This analysis also provided new predictive information on disease progression in Ta tumors compared with conventional staging. Furthermore, the gene‐expression profiles that characterized each stage and tumor subtype revealed their biological properties, thereby identifying new potential targets for therapy. Dyrskjot et al. then extended their work by examining the gene‐expression patterns of muscle‐invasive carcinomas and non‐muscle‐invasive transitional‐cell carcinomas with and without surrounding carcinoma in situ.^{( 77 )} The transitional‐cell carcinomas with surrounding carcinoma in situ and invasive carcinomas showed similar expression levels of a few gene clusters. As a result, a 16‐gene molecular classifier that represents the bladder carcinoma in situ gene‐expression signature was constructed. This classifier was suggested to be useful in the follow up of bladder cancer patients. Modlich et al. subjected tumor specimens from a cohort of uniformly treated patients with a well‐defined clinical outcome to a hierarchical cluster analysis.^{( 71 )} Non‐muscle‐invasive and invasive tumors were found to display distinct gene‐expression profiles. Moreover, distinct prognostic groups of invasive bladder cancer could be identified. These data thus provide additional insights into the molecular pathogenesis of bladder cancer and help detect novel prognostic markers for non‐muscle invasive, invasive, and metastasizing disease.

Blaveri et al. have characterized the global gene‐expression patterns of 80 bladder cancers, nine bladder cancer cell lines, and three normal bladder samples using cDNA microarrays containing 10 368 human gene elements.^{( 78 )} Unsupervised hierarchical clustering successfully separated the samples into two subgroups that contained non‐muscle‐invasive (pTa and pT1) or muscle‐invasive (pT2‐pT4) tumors. Supervised classification based on a limited subset of genes had a 90.5% success rate in separating non‐muscle‐invasive from muscle‐invasive tumors. Tumors could also be classified into transitional versus squamous subtypes (89% success rate) and a good versus bad prognosis (78% success rate). It was concluded that the genes driving the separation between tumor subsets may be important biomarkers for bladder cancer development and progression. Furthermore, these biomarkers could be candidates for therapeutic targeting.

The full utility of microarray analyses in bladder cancer research, diagnosis, prognosis, and treatment remains to be determined by additional clinical trials. One particularly critical issue that should be addressed by microarray analyses is the identification of non‐muscle‐invasive disease subtypes and the patients who are more likely to develop positive lymph nodes or distant metastases. It is likely that in the near future, gene profiling will be an effective way of predicting the response to specific therapeutic regimes, as it will determine the molecular signatures of the tumors with respect to their chemosensitivity or resistance to anticancer drugs. Moreover, the discovery of prognostic markers in cancer progression, as well as the identification of molecule‐susceptible targets, will lead to the development of novel alternative therapies. Thus, the classical concept of the tumor marker is currently being expanded from an individual biological determinant to gene clusters that can act as predictive classifiers.

Conclusions

Transitional‐cell carcinoma of the urinary bladder has a diverse collection of biological and functional characteristics. Although the current pathological and clinical variables provide important prognostic information, these variables are still limited in their assessment of the true malignant potential of most bladder cancers. A better understanding of the molecular mechanisms involved in carcinogenesis and cancer progression has identified a large number of molecular markers of bladder cancer, each of which have a potential diagnostic and prognostic value. Cystoscopy is the mainstay for diagnosing bladder cancer, but it is associated with a high cost and patient discomfort. Cytology and many urine‐based tumor markers give us marginal information for detecting and predicting the prognosis of bladder cancer.

Numerous factors, including chromosomal markers, genetic polymorphisms, and genetic and epigenetic alterations may be involved in tumorigenesis, progression and the patient's survival. Despite numerous single nucleotide polymorphism studies relating to bladder cancer, only a few genetic polymorphisms have been discovered, offering marginal information about a patient's prognosis. In contrast, the promoter hypermethylation of CpG islands strongly associates with tumor development, stage, recurrence, progression, and survival in transitional‐cell carcinoma of the urinary bladder. Detection of DNA methylation in voided urine is feasible and appears to be more sensitive than conventional urine cytology. Ultimately, all types of urological cancer could be screened in urine using a larger panel of hypermethylated genes. The epigenetic silencing of tumor‐suppressor genes is interesting from a clinical standpoint because it is possible to reverse epigenetic changes and restore gene function in a cell. Treatment with DNA methylase inhibitors or histone deacetylase inhibitors can restore the activity of dormant genes and decrease the growth rate of cancer cells in a heritable fashion. It should therefore be possible to partially reverse the cancer phenotype using these inhibitors. Gene‐expression analysis using microarray technology represents a high‐throughput approach to determine the biological behavior of the tumor, including its ability to grow, recur, progress, and metastasize. Cancer cells are generated by the aberrant expression of genes; therefore, analyses using microarrays will provide insight into which molecules and processes contribute to the phenomenon of bladder cancer development and prognosis. As growing databases of tumor data become available for analyses there is the potential for the development of personalized target therapy that is tailored toward specific molecular defects, thereby significantly lowering the morbidity associated with bladder cancer.