Molecular and Cellular Pathogenesis of Bladder Cancer: The Road to Translation
Clinicopathological phenotyping of bladder cancer has yielded a fascinating Janus-like molecular portrait of UC.[33,34] The vast majority (80%) of UCs are genetically stable low-grade papillary tumors (Ta) with a propensity for multifocality and post-resection recurrence, very limited invasion potential (≤ 5% risk of progression to muscle invasion), and high (90%) long-term disease-specific survival. Conversely, the remaining lesions represent genetically unstable tumors that have arisen de novo or from high-grade CIS lesions and that carry lethal potential, with a 50% risk of progression and 50% disease-specific survival following surgery for muscle-invasive disease.
Among noninvasive, low-grade papillary tumors, the most common genetic lesions are deletions of chromosome 9 and point mutations in the fibroblast growth factor receptor 3 gene (FGFR3) and in the alpha catalytic subunit of phosphatidylinositol 3-kinase (PIK3CA). More than half of bladder tumors of all stages and grades show chromosome 9 alterations and loss of heterozygosity (LOH). Several candidate tumor suppressors on 9p (eg, CDKN2A encoding p16 and p14) and 9q (eg, PTCH1, DBC1, TSC1) have been identified, but given the observed complexities, a multigenic model of inactivation may be required in order to link genotype to phenotype.
Activating FGFR3 mutations are found in 80% of low-grade Ta lesions and are the most common genetic alteration in bladder cancer. FGFR3 mutations are also found in 75% of benign nondysplastic urothelial papillomas and are rarely associated with CIS and TP53 mutations. The most common FGFR3 mutations—S249C in exon 7 (67%) and S375C in exon 10 (20%)—result in ligand-independent receptor dimerization, while the exon 15 mutation (3%) in the kinase domain predicts constitutive activation via altered protein conformation. Papillary tumors with concomitant CIS are generally FGFR3 wild-type, with patterns of chromosomal changes and gene expression signatures that are different from those seen in FGFR3-mutated tumors. While it appears that for the most part FGFR3 mutations do not play a significant role in invasive progression, invasive and metastatic tumors that carry activating mutations paradoxically behave in a more aggressive fashion. Current molecular studies have also suggested that NMIBC lesions might be better categorized on the basis of their FGFR3 status than by traditional grade and stage, and that such a categorization might improve current prognostic models.
The early identification of HRAS mutations in bladder cancer is a historic landmark in cancer research. Mutations in RAS genes (HRAS, NRAS, and KRAS2) have been identified in UC with more variable and lesser frequency than FGFR3 mutations, and with no association with tumor grade or stage. RAS and FGFR3 mutations appear to be mutually exclusive in Ta lesions, a finding that supports observations that oncogenic effects of mutationally activated FGFR3 are mediated by the Ras signaling pathway and that constitutional activation of the receptor tyrosine kinase–Ras signaling pathway is responsible for the genesis of an overwhelming majority of Ta lesions. RAS mutations are equally frequent in Ta and invasive subgroups, however, raising the possibility that a subset of invasive cancers may have evolved from a low-grade papillary lesion. Synchronous and metachronous multifocal Ta tumors share genetic alterations, indicating genomic stability and a clonal origin. Interestingly, Ta tumors with more complex genetic alterations can appear earlier than their clonal counterparts with shared but less complex alterations, a phenomenon that indicates growth-advantaged evolution in the former lesions.
To date, the most compelling differences between low-grade noninvasive lesions and invasive, high-grade tumors are alterations in the p53 and Rb tumor suppressor pathways. Multiple TP53 mutations, overexpression of HDM2, loss of p21 expression, and stabilized p53 expression, all of which carry an adverse prognosis in non-mutated tumors, have been identified. LOH in RB1 has been identified in over 50% of invasive tumors, and together with p53 status, Rb expression has been demonstrated to have prognostic significance. Additionally, E2F3 transcription factor overexpression associated with a 6p22 amplicon occurs exclusively in about 10% of invasive tumors, along with a high proliferation index and loss of Rb or p16 expression.
LOH at the PTEN locus on chromosome 10q, observed exclusively in invasive tumors, has implicated the PI3-kinase pathway in the progressive phenotype of UC. Other genes in the PI3-kinase pathway, including TSC1 and PIK3CA, are found in bladder tumors of all stages and grades, and the different roles of this pathway may therefore contribute to different phenotypes of disease.
ERBB2 amplification in seen in 10% to 14% of UC and is associated with higher grade and stage of disease. The association of more frequently observed ErbB2 overexpression with clinicopathological features is less certain. Additional insights into the contribution of ErbB2, along with other ErbB family members—EGFR, ErbB3, and ErbB4 (also overexpressed in UC)—to the two pathways of urothelial tumorigenesis are required. To date, amplification or point mutations of EGFR have not been reported, and this likely explains the lack of clinical activity of EGFR inhibitors in UC.
Recently, epigenetic silencing by histone methylation and hypoacetylation within seven stretches of contiguous genes, referred to as multiple regional epigenetic silencing (MRES), was reported in 26 of 57 bladder cancers (Ta, T1-T4). The MRES phenotype was tightly associated with a CIS gene expression signature, with absence of FGFR3 mutations, and with muscle-invasive disease, thus extending the two-pathway model of bladder cancer pathogenesis.
Tissue-specific transgenic murine models of bladder cancer have been generated, and the models generated thus far indicate that oncogene cooperativity is necessary to generate invasive and metastatic tumor—eg, p53 and Rb family gene inactivation by SV40 T-antigen or PTEN and p53 disruption. By contrast, mutant HRAS transgenic models generate mainly hyperplastic lesions with very slow development of non-invasive phenotypes resembling their human counterparts. FGFR3-mutant models have not been reported.
Unfortunately, we currently have inadequate insight into the genetic or epigenetic changes that may direct transformation of low-grade noninvasive tumors into high-grade invasive ones. Identification and clinical validation of predictive markers that reliably distinguish divergent pathways of behavior are required to advance real-world personalization of surgical and medical management of these tumors. Examples of existing challenges include inconsistencies in the data with respect to the prognostic value of p53 and Rb status, the relevance of low-frequency FGFR3 mutations in invasive cancers, and identification of alternative disease pathways for the 50% of high-grade invasive tumors that do not possess p53 or Rb defects.
Recent insights into the spatially restricted organization of the bladder epithelium and the cellular origins of transitional carcinomas promise to enrich the early genetic insights offered by the two-pathway carcinogenesis model. A p53 homologue—p63—has been shown to be critical for normal transitional epithelium development. Basal and intermediate layers of epithelium express p63, several high-molecular weight cytokeratins, and mature A/B blood group antigens—whereas apical umbrella cells express specific low–molecular weight cytokeratins and Lewis X determinant and are p63-negative. The bladder epithelium in TP63-null mice comprises a single layer of cells that resembles an umbrella-cell phenotype, suggesting the possibility of independent derivation from basal/intermediate cells. The prognosis of muscle-invasive tumors with a basal-cell phenotype appears to be inferior to that of tumors with a luminal phenotype. In xenograft studies, bladder cancer cells with basal-cell phenotypes account for tumor repopulation capacity, possess distinctive gene expression signatures that regulate invasion and metastasis, and define a population of NMIBC with adverse prognosis.[42,43]
The integration of cellular signatures with the genetic data from the two-pathway model is likely to further strengthen our understanding of the significant heterogeneity of UC, develop prognostic and predictive biomarkers for prospective validation, and identify high-value therapeutic targets. A molecular distinction between carcinogen-related and non–carcinogen-related cancers is required for a deeper understanding of pathogenesis. Extension of molecular (genetic and epigenetic) and cellular studies to other histological variants of bladder cancer, including squamous carcinomas, adenocarcinomas, small-cell carcinomas, and carcinosarcomas, will be critical to a fuller explanation of bladder neoplasia.
As with prostate, lung, hematological, and other cancers, an arsenal of genomic strategies is being utilized to identify novel gene fusions and other cryptic drivers of bladder cancer progression that may be drug targets. Innovative strategies to discover compounds or targets for induction of synthetic lethality in p53, Rb, and RAS-mutated backgrounds are relevant to at least 50% of MIBC. A significant fraction (31%) of upper tract urothelial carcinomas demonstrate microsatellite instability reminiscent of Lynch type 2 lesions and may be biologically distinct from bladder urothelial carcinoma. Whether a subset of these upper tract tumors will be amenable to synthetic lethal approaches now under study for tumors with DNA mismatch–repair defects remains to be seen. Gene expression models have allowed prediction of chemotherapy sensitivity in the United States National Cancer Institute’s Developmental Therapeutics Program (NCI-DTP) NCI-60 Human Tumor Cell Line Screen (which does not include bladder cancer lines). Use of this paradigm has been explored in urothelial carcinoma, and adoption of these principles for drug discovery and identification of synthetic lethal targets in bladder cancer lines may pave the way for individualized prediction of drug sensitivity and therapy design in MBC. Further insights into epithelial-stromal interactions—eg, interactions mediated by c-met; the downregulation of immune response by infiltrating myeloid cells; the role of T-lymphocyte subsets in regulating UC; and the interactive role of VEGF, FGF, and other cytokines in driving bladder-specific angiogenesis and lymphangiogenesis—will broaden our understanding of how the UC tumor microenvironment is reshaped to confer lethal progression. The epithelial and stromal components of the UC microenvironment represent therapeutic targets, and the challenge for translation is to synthesize an integrated, feasible, and effective strategy.
The pre-operative setting of localized NMIBC and MIBC offers an experimental platform for translational explorations with novel agents—eg, the study of gene signatures predictive of therapy-induced complete pathological response, validation of therapy targeting, insights into drug resistance, reconciliation of preclinical observations, and development of surrogate predictive markers. The risk of adverse outcomes with novel agents in a potentially curable setting must be balanced with the imperative to improve outcomes for individual patients. Pre-operative experimental therapy offers a complementary view to that of parallel trials in metastatic disease in which distant organ colonization by MBC cells may generate biology distinct from that seen in the bladder environment. A cross-section of clinical trials with molecular therapeutics and related targets implicated in the biology of bladder cancer is shown in Table 1. To date, no single agent (including EGFR family inhibitors, angiogenesis inhibitors, signal transduction inhibitors) has demonstrated significant clinical activity. High-quality targets—or multiple pathway–based targets—for therapy are needed for major advances in the field.
The Imperatives for Personalized Medicine in Bladder Cancer
It has been rightly lamented that only a small minority of bladder cancer patients are enrolled in clinical trials. Robust collaborative clinical care and research partnerships among multidisciplinary groups within and among institutions are required to drive progress in the field. An enhanced portfolio of novel therapeutics tethered to well-annotated tissue repositories and underpinned by strong translational science has the potential to change the therapeutic landscape meaningfully and to enhance organ-preservation options in localized disease. A short list of research priorities in bladder cancer, reflecting the imperatives for individualized management of the disease, is shown in Table 2. In addition, the special needs of the elderly and the increasing pressures on resource-constrained environments within and outside the developed world must be confronted. The cost of cancer care in the United States is projected to rise by nearly 40% between 2010 and 2020, to an annual cost of $173 billion.
These are formidable challenges, and Benjamin Franklin’s adage, "an ounce of prevention is worth a pound of cure," cannot be more relevant given today’s research funding shortfalls. Wide variations in the care of early bladder cancer exist, and among high–treatment intensity urology providers, overall survival is unchanged while rates of transition to major surgery are actually increasing. It has been said that for bladder tumors, it is time for a paradigm shift. We believe that the time is overdue.
Financial Disclosure: Dr. Kamat has served as a consultant or advisor to Archimedes, Inc., Endo Pharmaceuticals, TetraLogic Pharmaceuticals, AstraZeneca, and Precision Therapeutics; has received research support from Adolor, Bioniche/Endo, Celgene, and Alere, Inc; and has received honoraria from GE Healthcare. Dr. Mathew has no significant financial interest or other relationship with the manufacturers of any products or providers of any service mentioned in this article.