As mentioned previously, there have been very few reported cases of inherited bladder cancer. Families have been described without any evidence of cytogenetic abnormalities.2 Recently, a family was identified with a translocation of 20q and 5p.30 In addition to bladder cancer, this family also demonstrated metastatic melanoma. In this family, however, a putative oncogene or tumor-suppressor gene has not been identified.
Despite the initial discovery of a ras mutation in a bladder carcinoma cell line, very few ras gene mutations have been detected in primary TCC of the bladder.31 It now appears that less than 10 percent of primary bladder tumors contain ras gene mutations.32,33 Some investigators have observed increased expression of ras protein, but this result remains unproven because of questions about the specificity of the antibodies used.34
Growth factors are a class of proteins that bind to specific cell surface receptors, inducing a variety of responses including mitoses in susceptible target cells. Several laboratories have independently studied the expression of urothelial epidermal growth factor receptors (EGFR) using either immunohistochemistry to detect message with antibodies to various portions of the EGFR or autoradiography with isotope-labeled ligand.18 Most groups have found a higher density of receptors on malignant cells compared with normal epithelium.18,35 Moreover, epidermal growth factors are excreted in high concentrations in human urine, allowing incubation continually with normal premalignant and malignant urothelial cells. EGFRs are normally found only on the basal cell layer of the bladder epithelium. They also can be richly expressed on the superficial layers of malignant tissue. This abnormal distribution of receptors presumably allows greater access of malignant transitional cells to urinary epidermal growth factor (EGF) and has led investigators to suspect that EGF plays a role in the development and growth of bladder cancer.18 Others have reported that the EGFR gene is expressed mostly in high-grade invasive tumors, and increased staining has been correlated with increased stage and death from disease in patients with Ta or T1 lesions.36,37 However, other investigators have found no gross abnormalities at the DNA level; thus overexpression may be owing to an increase in mRNA transcription alone.38 The c-erbB1 proto-oncogene that maps to chromosome 7 encodes the EGFR gene. Interestingly, trisomy of chromosome 7 is a frequent genetic observation in bladder tumors and could lead to an increase in EGFR expression in tumor cells.21,22 Another gene on chromosome 7, MET, is a candidate proto-oncogene in bladder cancer because mutations of MET have been found in hereditary and sporadic papillary renal tumors with trisomy 7.39 We have sequenced 48 primary bladder tumors and found no mutation of the critical exons of the tyrosine kinase domain.
DNA amplification and increased levels of expression of c-erbB2 have been reported as well as alterations of c-myc and c-src.40-43 However, gross alterations or amplifications of these genes have been described rarely in primary bladder tumors. Although many proto-oncogenes have been identified in human tumors, very few have been found to be consistently altered in bladder cancer. In contrast, chromosomal loss and inactivation of tumor-suppressor genes have been found to play a significant role in progression of bladder tumors.
Chromosomal Deletions and Tumor-Suppressor Genes.
Southern analysis with RFLP markers followed by the recent availability of highly polymorphic small repeat sequences known as microsatellites has allowed genome-wide assessment of chromosomal loss in bladder cancer.44-48 The most common loss is the genetic event identified as loss of chromosome 9.49 Deletion of chromosome 9 appears to be just as common in superficial tumors and invasive tumors. Inactivation of a putative tumor-suppressor gene on chromosome 9 is therefore a key candidate for the initiating event in bladder carcinoma. Careful mapping of chromosome 9 with microsatellite markers has revealed that there are at least two distinct regions of loss: one on chromosome 9p21 and the second on chromosome 9q.50-52 Southern blot analysis, comparative multiplex PCR, and FISH analysis have revealed the presence of small homozygous deletions of 9p21 in primary bladder tumors.53,54 These deletions also have been seen in cell lines and have been implicated in the genesis of a variety of tumor types.55,56
Another common area of allelic loss is on chromosome 17p. Losses of 17p have correlated with mutations of p53 and occur predominantly in invasive bladder tumors.57,58 However, a subset of superficial tumors, especially flat lesions, has been found to contain a higher rate of 17p loss.19,59
Southern blot analysis of primary bladder tumors with polymorphic markers has revealed frequent loss of chromosome 13q at the Rb locus.45 Loss of 13q also has been associated with tumors of high stage, and immunohistochemical studies recently have confirmed that Rb is the major target of 13q deletions in bladder tumors.60
A number of other areas of allelic loss have been identified in bladder cancer. Chromosome 11p loss originally was described by Southern blot analysis and has been confirmed by microsatellite analysis.44,47,48 Although both Wilms' tumor loci are candidate targets for the deletion of 11p observed in bladder cancer, bladder tumors are not seen in the spectrum of urogenital abnormalities or as second primary malignancies in Wilms' tumors.61
Losses of 3p, 4, 5q, 8, 14q, and 18q also have been reported.19,47,48,62 Two distinct regions of loss have been identified on chromosome 4, one on 4p and one on 4q.63,64 There also has been a report of two distinct regions of loss on chromosome 14, one on proximal and one on distal 14q. These losses of chromosome 14q correlated closely with increasing grade and stage.65
The Clonal Origin of Bladder Cancer
Many bladder tumors present as multifocal disease at diagnosis. The concept of field cancerization was described originally by Slaughter to explain the occurrence of multiple skip lesions and second primary tumors in patients with aerodigestive tract tumors.66 This hypothesis also was extended to bladder tumorigenesis to describe the possible presence of a field defect secondary to continued exposure of exogenous and endogenous compounds excreted in urine.11
We have examined the hypothesis of field cancerization in bladder cancer using molecular genetic techniques.67 We tested tumors from four female patients with a method that analyzes X chromosome inactivation, which can determine whether tumors were derived from the same precursor cell. This technique was complimented by analysis of allelic loss on various chromosomes, as described previously. In each patient examined, all tumors had the same X chromosome inactivation, whereas normal bladder retained the same polyclonal X chromosome inactivation pattern as expected. Moreover, each of the evaluable tumors from a given patient had lost the same chromosome 9 allele, commonly found early in progression. Later events in progression, such as 17p and 18q loss, were not shared by different tumors from the same patient, implying that multiple tumors in the same patient arose from the uncontrolled early spread of a single transformed cell. These tumors then proceeded through independent and variable genetic events during progression. If a field defect existed, one would expect multiple independent transforming events in each tumor, implying a multiclonal origin for these lesions.
Since this study, other investigators have confirmed the hypothesis that most multiple tumors in the bladder arise from a single progenitor cell. In another study, multiple tumors from 28 of 30 patients were found to contain the same X chromosome inactivation pattern, implying evolution from the same progenitor cell.68 This understanding about bladder cancer genesis has implications for our understanding of tumor progression and may be useful for cancer diagnosis (see below). It also has allowed the designation of a preliminary progression model for bladder cancer.
Molecular Progression Model
As mentioned previously, careful characterization of genetic alterations within histopathologic lesions at various stages of progression allows the delineation of a molecular progression model. We have previously defined a simple progression model for bladder cancer.18 In this model, critical allelic losses have been placed in various steps of progression, but oncogenes are not demonstrated because they are involved, so far, in only a minority of primary bladder tumors at the genetic level. Critical steps in this model include initiation of bladder cancer owing to the inactivation of a putative tumor-suppressor gene on chromosome 9, loss of p53 function from the preinvasive to invasive state, and a variety of other genetic alterations associated with invasion and metastasis.
One interesting aspect of this progression model is the distinct differences between the progression of flat and papillary superficial lesions. Both these lesions share a high frequency of chromosome 9 loss that remains almost unchanged during the progression to invasive tumors. However, loss of chromosome 17p is far more frequent in flat lesions and has been associated with inactivation of the p53 gene.19,59,69 This is intriguing because inactivation of p53 may lead to accumulation of further genetic changes and the propensity of these lesions to acquire a more invasive phenotype. Another distinct change is loss of chromosome 14q. 14q deletion is almost exclusively seen in flat lesions or invasive tumors and virtually absent in papillary lesions.69 Interestingly, the frequency of 14q loss is even higher in flat lesions than in invasive tumors, suggesting that not all invasive tumors arise from flat lesions. 14q loss thus may lead to the initiation of flat lesions, from which only a fraction may continue to progress to invasive tumors. In this way, invasive tumors may be the final progression pathway for some papillary lesions and many flat lesions. Further characterization of the critical gene on chromosome 14q may lead to a better understanding of the events that lead to the development of flat lesions and their propensity for invasion.65
Although there is no common or defined syndrome for familial bladder cancer, familial uroepithelial tumors have been reported. Often these cases appear as a manifestation of the cancer family syndrome, known as Lynch syndrome (HNPCC).4 Of the many neoplasms that occur in these families, TCC is the fourth most common, affecting individuals who manifest TCC alone, TCC and colon cancer, or TCC and other carcinomas.70,71 Interestingly, in Lynch syndrome, TCC is predominant in the upper tract in contrast to sporadic TCC.
Mutations of mismatch repair genes, including MSH2, MLH1, PMS1, and PMS2, have been found to be responsible for the majority of cases.72-75 These genes are involved in DNA mismatch repair and belong to a highly conserved group of repair proteins. As in other sporadic tumors from this syndrome, bladder tumors display characteristic genetic instability manifested by shifts or changes in the repeat size of microsatellite markers.76 These shifts are actually expansions and contractions of small DNA repeat elements. Approximately 2 percent of all sporadic bladder tumors display characteristic microsatellite instability associated with Lynch syndrome and mismatch repair.76
A candidate gene on chromosome 9p21, p16 (CDKN2/MTS-1), is the most common inactivated gene in bladder cancer.56,77 This gene has been found to be mutated in familial melanoma and pancreatic cancer.78,79 Although a few point mutations are observed in bladder cancer cell lines, the vast majority of primary tumors with loss of heterozygosity (LOH) of 9p21 do not contain obvious mutations of p16.80,81 This finding pointed to alternative mechanisms for gene inactivation or, potentially, that a second tumor-suppressor gene resided nearby. Recently, we have shown that homozygous deletions of chromosome 9p21 stretching into the p16 locus are quite common in primary bladder tumors.54 Much of the controversy surrounding this locus stems from difficulty in identifying homozygous deletions in primary tumors because of contaminating nonneoplastic cells. However, using the strategy of fine microsatellite mapping, homozygous deletions can be identified by the apparent retention of one or two closely spaced markers among a large region demonstrating LOH.54 These results were confirmed in a number of cases by Southern blot and FISH analyses demonstrating the specificity of the technique. It is now clear that at least 50 percent of all bladder tumors contain a homozygous deletion that includes p16. Moreover, we have demonstrated other alternative mechanisms of inactivation including methylation of the p16 promoter leading to transcriptional block and inactivation of p16.82 Although methylation is common in many other tumor types, inactivation of p16 by methylation is still uncommon in bladder cancer. The p16 locus is quite complex and codes for a second transcript called ARF. This distinct protein appears to be involved in a separate p53 pathway. Although its exact role in tumor progression is not clear, deletion of the p16 locus may inactivate genes involved in two critical tumor-suppressor gene pathways, Rb and p53.83 Further analysis of a putative second tumor-suppressor locus on chromosome 9 is hampered in bladder cancer by the wide occurrence of monosomy, perhaps indicating inactivation of a gene on chromosome 9q. Although the patched gene, inactivated in Gorlin's syndrome and sporadic basal cell carcinoma, has been found on chromosome 9q, an important role in bladder cancer has not been defined.84,85, 85a, 85b Mutations of p53 are ubiquitous in human cancer, and bladder cancer is no exception.86 Losses of 17p have correlated well with sequence analysis of p53 mutations and occur predominantly in invasive bladder tumors.57 However, a subset of superficial tumors, especially flat lesions, has been found to contain these mutations. Importantly, a large study based on immunohistochemical analysis of p53 demonstrated a significant decrease in overall survival for p53 positive patients versus those with tumors that were p53 negative.87 It was implied that mutation of p53 was an independently poor prognostic factor regardless of stage or therapy. Inactivation of p53 is critical in many tumor types.86 It has been postulated that p53 is a critical regulator of response to DNA damage.88 The appropriate presence of wild-type p53 leads to growth arrest in the presence of damage and perhaps to apoptosis with excessive damage. Cells that lack p53 protein are unable to undergo a normal G1/S arrest and perhaps propagate further accumulated genetic damage.
A number of lines of evidence also point to a role for the Rb gene in bladder carcinogenesis. Immunohistochemical studies have confirmed that Rb is the target of 13q deletions in most bladder cancers.60 A substantially worse prognosis for those tumors with negative standing also has been reported.89,90 Moreover, reintroduction of the Rb gene leads to slowing of cell growth and tumorigenicity in bladder carcinoma cells.91 The regulatory function of the Rb protein appears to be controlled by phosphorylation during the G1S phase of the cell cycle.92 p16, one of a number of CDK inhibitors, is also critical for this pathway. In many ways, inactivation of both p16 and Rb may be redundant, and in fact, tumors have demonstrated inactivation of one or the other of these genes but generally not both.88 Analysis of Rb and p16 status directly in bladder cancer has not been done on the same tumors. It is tempting to speculate, however, that the cyclin D1/p16/Rb pathway is vital in bladder cancer as in many other tumor types.
Bladder tumors are occasionally seen in patients with Cowden syndrome, 93 which arises from mutation of the PTEN/MMAC1 tumor-suppressor gene located at chromosome 10q23.94 Point mutations and homozygous deletions of PTEN/MMAC1 have been found in a subset of bladder tumors with chromosome 10q LOH and are associated with advanced disease.95