EPIDEMIOLOGY OF BLADDER CANCER

Bladder cancer is the fourth most incident cancer in males and ninth most incident in females. In the United States, over 67 000 new cases are diagnosed per year,1 and over 350 000 cases diagnosed worldwide.2 In the United States, the annual age-adjusted incidence rate for men is approximately 32 per 100 000.3 Men have a higher risk of bladder cancer than women, by a rate ratio of at least 3:1. Approximately 66% of bladder cancers are diagnosed among individuals age 65 years or older.

Bladder cancer arises primarily from the transitional cells of the bladder mucosal epithelium (90% of cases) and may present as a noninvasive, papillary tumor protruding from the mucosal surface that is readily resectable. Although these superficial tumors frequently recur after cystoscopic resection, they usually do not invade the bladder wall or metastasize. However, about one-third of incident bladder cancers present as solid, nonpapillary tumors, which originate from in situ dysplasia and carcinoma in situ. These tumors invade the bladder wall and have a high propensity for metastasis.4, 5, 6 For these invasive tumors, 5-year survival rate is only 30–50%.7, 8 This stark difference in morphology and survival implicates separate oncogenic pathways for noninvasive vs muscle-invasive cancer.9

The incidence of bladder cancer was strongly associated with occupational exposure to aromatic amines used in the dye industry, before their potent carcinogenicity to the bladder was demonstrated.10 With reduction of such workplace exposure, active smoking is now the strongest environmental risk for bladder cancer, contributing to more than 50% of cases.2 Recent modest reductions in the incidence of bladder cancer are attributed to decreasing exposure to tobacco smoking and to occupational carcinogens. Although diet might also influence bladder carcinogenesis, owing to the many potential carcinogens or chemopreventive nutrients therein,11 no consistent association between intake of selected nutrients or micronutrients and bladder cancer has emerged. Other lifestyle habits have been widely investigated, such as intake of coffee, artificial sweeteners, hair dyes, and have not been materially associated with bladder cancer.2 Any contribution from the chronic irritation of the bladder mucosa from recurrent urinary tract infections or stones is modest.12, 13 Hence, at this time, the primary environmental risk for bladder cancer is smoking. Active smoking and longer smoking duration are associated with increased risk for bladder cancer, when compared with never- or former-smokers.14, 15, 16 Smoking intensity (cigarettes smoked per day or pack-years smoked) is not a separate independent risk variable.17

BLADDER CANCER CARCINOGENESIS

In the most general sense, conventional models of carcinogenesis suppose the existence of successive mutation events within a specific cell clone, enabling its eventual escape from regulatory mechanisms for cell division and maintenance of genomic integrity.10 Environmental carcinogens are thought to interact with the genome at an early stage, for example, by forming DNA adducts that induce DNA mutation at their binding sites. Exposure to carcinogens at later stages of carcinogenesis may involve completion of previous carcinogenetic steps. Both time lines appear to be operative in the effect of smoking on tissues of the lung, bladder, and upper digestive tract.

Aromatic polycyclic hydrocarbons, nitrosamines, and aromatic amines generated by smoking have well-established carcinogenic influence. Smoking imparts risk for destabilization of cell-cycle control pathways and susceptibility to bladder cancer.18 However, the human population exhibits extraordinary variability in susceptibility to the risk of smoking. The chief genetic polymorphisms studied to date are xenobiotic-metabolizing enzymes.19 Polymorphisms in the enzymes CYP1A1 and GSTM1 contribute to the host ability to detoxify these carcinogens, and hence host susceptibility to their carcinogenic influence. To date, however, genetic predispositions in susceptibility to the direct carcinogenic effect of smoking in bladder carcinogenesis have not been elucidated.

A more fruitful area of investigation for bladder carcinogenesis has been evaluating the molecular basis for aggressive bladder cancer. Particular focus has been given to epigenetic alteration,20 with DNA hypermethylation and inactivation of tumor suppressor genes leading to poor patient survival.21, 22, 23 Epigenetic gene methylation has been linked to exposure to tobacco smoke and exposure to drinking water arsenic.24 Promoter hypermethylation may affect an increasing number of genes with increasing disease stage in bladder cancer.21, 25, 26

In patients with bladder cancer, increasing age, male gender, and current cigarette smoking are positively identified with a methylation trait.17 Occult DNA promoter hypermethylation as a latent trait can drive a methylator phenotype.27 Analysis of cumulative epigenetic events permits assessment of the underlying propensity for promoter hypermethylation.28 Both smoking and inorganic arsenic exposure are associated with gene-specific promoter hypermethylation in bladder cancer.21, 25 The genetic polymorphisms and sites of hypermethylation are numerous, involving genes that span the range of cellular processes, including intracellular regulatory networks driving development growth.17 Accumulation of hypermethylation events may be related to the aging process and increased duration of carcinogen exposure, with increased DNA damage and genomic instability. Epigenetic predictive models using artificial intelligence also support the premise that promoter hypermethylation is a reliable predictor of tumor progression in bladder cancer.29 Collectively, these data support the concept that cigarette smoke is not only a direct mutagenic carcinogen, but also associated with the genesis of epigenetic alterations in bladder cancer.30 Continuous tobacco carcinogen exposure, in combination with increasing age and male gender, drives and enhances the selection and clonal expansion of epigenetically altered cells.17

Previously published clinical reports have indicated a possible familial component to bladder cancer.31, 32, 33, 34 There is a significantly increased risk of bladder cancer among first-degree relatives of individuals who have bladder cancer, with an earlier age of disease onset.35, 36, 37 In a large twin study, environmental factors were estimated to contribute a significant 69% to bladder cancer risk, whereas inherited genetic factors contributed a nonsignificant 31% of risk.38 Attention has been given to the NAT2 slow acetylator and GSTM1-null genotypes as potential familial risk factors.39 Such data support the concept of low-penetrance cancer-predisposing polymorphisms acting together and/or interacting with environmental factors, to increase the familial risk of bladder cancer.40

However, the familial risk of bladder cancer is not well understood and, to date, little attention has been paid to the joint effect of smoking and family history in modifying the risk of bladder cancer, in part because of the limitations of case control studies. This limitation was overcome in a family-based population analysis of incident bladder cancer, in which probands who had smoked and who also had a positive family history were at 5.31-fold increased risk of bladder cancer.41 For never-smoking individuals with bladder cancer, the disease was diagnosed at a later age (>65 years) than smokers, in which the cancer presented at ages 40–65 years. Smoking imparted a 6.89-fold increased risk (95% confidence interval 2.25-fold to 21.21-fold). This study indicated that a positive family history of bladder cancer interacted with smoking habits to increase the risk of bladder cancer; the P-value for the interaction of smoking and family history was P=0.04. Stated differently, a family history of bladder cancer confers increased risk only among active smokers.

The chief questions pertaining to the pathogenesis of bladder cancer are summarized in Table 1. It is against this backdrop that a most remarkable study of bladder carcinogenesis has been conducted, which provides substantive new insights into the first five questions.

Table 1 Chief questions pertaining to the pathogenesis of bladder cancer
Full size table

WHOLE-ORGAN MAPPING OF THE BLADDER CANCER GENOME

A bold new approach has now been taken to examine bladder carcinogenesis. Using a whole-organ mapping strategy, Czerniak and coworkers42, 43, 44 have previously examined genetic hits across the entire mucosa of bladders affected by cancer. Their construction of a genome-wide map over the entire mucosa has enabled identification not only of genetic alterations in mucosal regions of evident dysplasia, but also in morphologically normal mucosa. Rectangular tissue samples, 1 × 2 cm in size, were obtained from the entirety of resected human bladders with invasive cancer and analyzed for morphological features of cancer or its precursor lesions, and at the molecular level. The comprehensive completion of this extraordinary body of work is presented in this current issue of Laboratory Investigation, as summarized in Box 1.45

The fundamental concept is that critical regions of chromosomes, both for noncoding DNA sequences and for novel genes, may be involved in the earliest phases of cancer development. Based on this work with bladder cancer, Czerniak and coworkers term the target genes they have identified in the key six chromosomal regions as ‘forerunner genes’ (FR genes). This is based on the concept that these genes are relevant for the development of the initial clonal expansion of in situ urothelial neoplasia.44

Three waves of genetic ‘hits’ in the bladder mucosa were identified. The first wave constituted clonal expansion of phenotypically normal-appearing urothelial cells over large portions of the bladder mucosa. The second wave was associated with subregions of clonally expanded cells showing the first microscopically recognizable features of dysplasia. The third and last wave was associated with the fully transformed phenotype of severe dysplasia/carcinoma in situ. Notably, further advancement to invasive cancer was not associated with additional major loss of genetic material. During the first wave, loss of a genomic region around RB1 in one chromosome was not uniformly associated with inactivation of the second RB1 allele. Inactivation of the second allele, as by point mutation, with ultimate inactivation of the RB1 gene, was a later event corresponding to the third wave. Hence, loss of FR genes in the region of RB1 provide the initial growth advangage during the first wave, to enable expansion of a clonal geographic region of morphologically normal urothelial epithelium. Subsequent loss of a tumor suppressor gene, such as RB1, was a transforming event, associated with development of high-grade dysplasia/carcinoma in situ, progessing to invasive cancer.

These data also provide evidence that a limited number of critical steps, in this case at least six, are required for tumorigenesis of invasive bladder cancer. Although the constellation of genomic-wide hits is highly variable, they ultimately act in concert.

ALTERATIONS IN FORERUNNER GENES IN PATIENTS WITH BLADDER CANCER

The DNA sequences, expression patterns, and methylation status of RB1 and candidate FR genes in the 13q14 chromosomal region around RB1 were then analyzed, as summarized in Box 2.44, 45 The detailed analysis of P2RY5 polymorphisms, including germ-line mutations, provides remarkable data in support of potential inherited host susceptibility to the carcinogenic effects of smoking. Mapping of the P2RY5 polymorphisms in normal bladder mucosa demonstrated that this gene, which resides within intron 17 of the RB1 gene, is affected before loss of RB expression during bladder carcinogenesis. Third, pedigree analysis revealed one family with an inherited risk of cancer, which carried a germ-line mutation in P2RY5. In addition, the expression pattern of several additional candidate FR genes indicates that GPR38, CAB39L, RCBTB1, and ARL11 had more than 50% reduction of their expression in several bladder cancer cell lines and should be further investigated as potential candidate FR genes. Lastly, downregulation of FR genes was identified in 2/3 of a series of cancer cell lines from a variety of other human malignancies.

The body of work presented by Majewski et al45 constitutes an extraordinarily comprehensive exploration of bladder carcinogenesis in specific and principles of carcinogenesis in general. The data provide substantive new insight into the following: familial risk of bladder cancer; host susceptibility to the effects of smoking; and molecular changes in FR genes, which precede both morphological changes in the bladder mucosa and genetic/epigenetic alterations in more conventional tumor suppressor genes. The clonal expansion, over vast expanses of the bladder mucosa, of urothelial cell populations containing losses of FR genes may represent the earliest molecular change in bladder carcinogenesis. At the least, these genetic alterations are earlier than previously reported preneoplastic changes. In the case of chromosome 13, the fact that the FR genes are in the immediate vicinity of the known oncogene RB1, and that their loss precedes loss of RB1function, significantly revises the concept of chromosomal regions of susceptibility. Lastly, initial evidence is given that losses of genetic material in the FR gene-containing regions can be detected in exfoliated urine DNA from virtually all patients with bladder cancer, thereby constituting a potential diagnostic tool for detecting bladder cancer at its earliest stages.

THE CONTEXT OF CURRENT LITERATURE

In identifying genomic imbalances, it is difficult to discern which imbalances are primary causal events and which are redundant hits dragged through clonal progression simply by cosegregation. Although analysis of families with a high frequency of cancer can help to identify genomic regions of interest, the vast majority of human cancers are sporadic in nature and are thus not amenable to family analysis. The alternative strategy of identifying the geographic distributions of genes, in human tissue containing cancer, has been used to advantage in cancers such as colorectal and esophageal cancers.46, 47 Identification of clonal plaque-like regions is of interest, regardless of whether morphological features of pre-neoplasia (dysplasia) are evident. Indeed, clonal expansion of normal mucosal tissue is of immense interest and underpins the concept of ‘forerunner genes’. At the opposite end of the genomic spectrum are hits restricted to regions of severe dysplasia or carcinoma in situ, which represent late events associated with progression to invasive cancer. The current study by Majewski et al45 spans the entirety of both geographic and familial cosegregation analyses.

These findings are central to the hypothesis that bladder carcinogenesis is initiated by genetically altered but histologically normal cells that produce fields of altered cells by intraepithelial displacement.48 Upon accumulation of further genetic changes, the fields of altered urothelium reach a state of criticality and there is production of frank tumors. The current work extends the published data set from conventional metaphase comparative genomic hybridization, which has detected recurrent gain and amplification at 6p22 in up to 31% of bladder tumors and bladder cancer cell lines.49, 50

The work of Majewski et al,45 and that of the preceding papers from this same group,42, 43, 44 also puts further emphasis on the role of the chromosomal region containing RB1. Invasive urothelial cancers are characterized by severe disturbances in proximate cell-cycle regulators, of which RB1 is key.51 These disturbances are exacerbated by chromosomal instability and, in turn, permit and promote such instability. This leads to concomitant and interacting defects in cell-cycle regulation and the control of genomic stability, enabling the transition toward invasive cancer. In a previous study relevant to the RB1 story, a 2.5 Mb region of 6p22 was found to play a role in bladder carcinogenesis.52 Specifically, among the several genes in this region, E2F3 belongs to the E2F family of transcription factors, and encodes two protein products (E2F3a and E2F3b) through the use of alternative promoters and different 5′-coding exons.53 E2F3a is a transcriptional activator, is expressed in proliferating cells, reaching peak levels in late G1, and is linked to transactivation of genes associated with DNA synthesis and cell-cycle progression.53, 54 Notable from the current study of Majewski et al45 is the fact that activation of E2F3a target genes is repressed during G1 when E2F3a is bound to the retinoblastoma protein, RB.55 During G1/S, RB becomes hyperphosphorylated, leading to the release of E2F3a and subsequent target gene activation.56 E2F3a is the predominant binding partner for RB in G0. Overexpression of E2F3 and a second gene from the 6p22 region, CKDAL1, has been documented in bladder tumor tissue microarrays 57 and in bladder cancer cell lines.58 Hurst59 has recently demonstrated that there is a functional link between E2F3 overexpression and proliferative advantage in bladder tumor cells with 6p22 amplification; knockdown of CDKAL1 had no effect on cell proliferation. Amplification of 6p22.3 was identified in 8 out of 91 invasive bladder tumors. In five tumors studied in detail, four showed 13q14.2 loss (including RB1) and expressed no RB protein. The fifth had an alternative mechanism for RB inactivation, through hyperphosphorylation. These authors conclude that inactivation of the RB pathway is required in addition to E2F3 overexpression in this subset of bladder tumors.

Analysis of the methylation status of CpG islands in the promoter regions of genes associated with primary papillary bladder cancer (GSTP1, DAPK, MDR1, TPEF, PAX6, and TSLC1) has been performed.60 Only TPEF and PAX6 had substantial CpG island methylation percentages, which were higher in tumor tissue when compared with normal mucosa adjacent to the tumor. However, even normal mucosa had elevated levels of TPEF and PAX6 hypermethylation, thereby possibly representing biomarkers for this form of bladder cancer. The latter study also points out the potential differences in molecular carcinogenesis between papillary and nonpapillary bladder cancer. However, it is remarkable that losses of genetic material in six chromosomal regions are relevant to both papillary and nonpapillary cancer as well.

Such genome-wide screening methodologies will no doubt be of value for other cancers. Over-and-above documented differences between papillary and nonpapillary bladder cancer, current data make clear that different genetic regions are likely to be operative in different tissue sites.61, 62, 63, 64, 65, 66, 67, 68

IMPLICATIONS FOR THE DETECTION AND PREVENTION OF BLADDER CANCER

In keeping with the most fundamental tenets of cancer care, early detection of bladder cancer is considered to improve patient prognosis and decrease the need for cystectomy, by identifying tumors before they become muscle invasive. Current detection methods involve the use of cystoscopy, radiographic imaging, and urine cytology. Because this detection paradigm is among the most expensive of cancer detection protocols for various human tissues, an intense search is underway to identify effective—and inexpensive—biomarkers of early bladder cancer.69 The ideal bladder tumor biomarker would be noninvasive, easy to use, reliable, and efficacious.70 Current urinary biomarkers under evaluation for detection of occult bladder cancer are given in Table 2.

Table 2 Current biomarkers for bladder cancer (Alvarez, 2007)
Full size table

Recent clinical reports include urine screening of individuals with ≥40 pack-years of smoking, using combined UroVysion, cytology, and urinary dipstick testing for hematuria, which found malignancy in 3.3% of these high-risk individuals.71 Likewise, microsatellite analysis of exfoliated urine to detect DNA LOH, combined with performance of UroVysion FISH and conventional urine cytology, enabled detection of over 93% of patients with recurrent bladder cancer.72 Analysis of urine for microsatellite alterations shows a high correlation with invasive tumor resident within the bladder73 and exhibits high sensitivity for patients who have invasive cancer.74

Surprisingly, the most important biomarker of bladder cancer is microscopic or gross hematuria, which occurs in 85% of patients with bladder cancer.75 Repeat home urine dipstick testing followed by medical evaluation of positive patients may achieve 90–95% sensitivity.76 With the home hematuria testing protocol, newly diagnosed patients with bladder cancer had muscle invasion in only 10% of cases, as opposed to 60% in newly diagnosed bladder cancer patients who had not been screened for hematuria.76 This translated into lower disease-specific and overall mortality than unscreened patients. Unfortunately, the positive-predictive value of hematuria is only 0.08%.77 Hence, there is desperate need to stratify hematuria-positive patients into low- and high-risk groups.

As discussed earlier, smoking is the most frequent risk factor associated with bladder cancer and is estimated to account for up to 50% of bladder cancers in men.78 Stopping smoking and avoiding exposure to second-hand smoke should be an effective method of decreasing the incidence of this disease. Such a strategy is an example of primary chemoprevention, namely preventing occurrence of disease in a healthy population. A prospective study to demonstrate this effect would be both expensive and time consuming because of the long lag between exposure to carcinogens and the development of bladder cancer.79 So-called ‘secondary prevention’ treats people with well-defined premalignant conditions to prevent the development of clinical cancer. As such premalignant disease states have not been generally recognized in bladder cancer, this approach has not been used in this disease.

The FR genes represent novel biomarkers for detection of the premalignant state and of outright cancer and also represent targets for therapeutic and preventive interventions. Examining urinary DNA for alterations in FR gene status might identify bladders at risk for urothelial neoplasia long before the preneoplastic pathway actually begins. Clinical monitoring and therapeutic management protocols, including recommendations about smoking, are readily envisaged.

Tertiary prevention, preventing recurrences in patients with a history of cancer who are currently clinically free of disease, is the approach that has been generally used to investigate chemoprevention of bladder cancer. Several randomized clinical trials have been completed, evaluating a variety of agents including high-dose vitamins, retinoids, and difluoromethylornithine for the chemoprevention of bladder cancer.80, 81, 82, 83, 84, 85, 86 These studies, although generally disappointing, have shown that drugs used for chemoprevention must have low toxicity and that study design may be critical to success.87 Again, monitoring of such patients on the basis of FR gene status is an attractive possibility.

One final comment pertains. Identification of FR genes creates extraordinary opportunity to examine the molecular biology of these novel modulators of cancer risk. It is not yet clear how many coding genes are present in the six chromosomal regions of interest. For those FR genes currently identified (eg, P2RY5), it is now imperative that their function be elucidated in the context of carcinogenesis.