Molecular Pathology of Urothelial Carcinoma

Abstract

Urothelial carcinoma is characterized by the presence of a wide spectrum of histopathologic features and molecular alterations that contribute to its morphologic and genomic heterogeneity. It typically harbors high rates of somatic mutations with considerable genomic and transcriptional complexity and heterogeneity that is reflective of its varied histomorphologic and clinical features. This review provides an update on the recent advances in the molecular characterization and novel molecular taxonomy of urothelial carcinoma and variant histologies.

INTRODUCTION

Recent advances in molecular biology and next generation sequencing technologies have greatly enhanced our knowledge and understanding of the genomic landscape of UC and provided insights into the molecular biology and pathogenesis of UC and some of its variants.[1] One of the main challenges in urothelial carcinoma is the presence of considerable intra- and inter-tumoral heterogeneity that can be appreciated at the clinical, morphological, genomic and transcriptomic levels.[2] Many UC variant histologies are recognized and can be pure or, more frequently, mixed with UC-NOS. These variants have historically been based on defined morphological features and more recently new molecular studies provided valuable insights into them.[1, 3] The aim of this review is to provide an update on recent advances in molecular features of UC and a subset of its variants and potential clinical applications.

MOLECULAR CHARACTERIZATION OF INVASIVE UROTHELIAL CARCINOMA

Multiple somatic mutation clones can be detected in the normal appearing urothelial lining of the bladder in both healthy and diseased bladders.[4, 5] This mutational background provides a permissive environment “field effect” for the development of urothelial carcinoma through a process of clonal expansion such that the ultimate invasive disease is comprised of multiple clones sharing ancestral mutations but also harboring many novel and unique mutations, which was revealed by multi-regions sequencing from the same bladder.[4, 6, 7]

Recent studies with focus on the genomic landscape of UC have consistently identified high rates of somatic mutations (>7 mutations per Mb), only surpassed by lung carcinomas and melanoma, which is characteristic of a carcinogen-induced malignancy.[8–10] By mutation signature analysis from whole exome sequencing (WES), APOBEC mutagenesis was the main mutational signature, which is attributed to a family of cytidine deaminase enzymes responsible for innate immunity that restricts the propagation of retroviruses and retrotransposons.[8] In muscle-invasive bladder cancer (MIBC) cohort reported by the Cancer Genome Atlas (TCGA), two APOBEC-mediated mutation signatures were associated with 66% of the single nucleotide variants (SNVs), and the presence of APOBEC3-associated mutation signature was associated with better prognosis and improved 5-year overall survival.[10–12] Another mutational signature associated with ERCC2 mutation corresponded to approximately 20% of all SNVs in MIBC. This mutational signature was additionally shown to be associated with smoking independent of ERCC2 mutation status.[13] ERCC2 encodes a DNA helicase with a central role in a highly conserved DNA repair pathway through nucleotide-excision repair mechanism. Mutations in ERCC2, as well as other genes involved in DNA damage response and repair (DDR), were recently shown to be associated with improved response to cisplatin based chemotherapy as well as immune checkpoint blockade and radiation therapy for advanced UC.[14–18] Moreover, the presence of putative deleterious DDR gene alterations in pretreatment tumor tissue was strongly predictive of chemosensitivity, durable response, and superior long-term survival in MIBC patients treated with neoadjuvant dose-dense gemcitabine and cisplatin (ddGC).[19] Based on these results, there are now clinical trials designed to prospectively profile transurethral resection tumor samples by NGS for the presence of alterations in DDR genes and correlation with response to neoadjuvant ddGC, with the possibility of bladder preservation in patients demonstrating response to this treatment (clinical trial NCT03609216).

Besides the genes involved in DDR, many other genes are significantly mutated in UC that are involved in important cellular signaling pathways and canonical functions (Figure 1). Most alterations were identified in genes involved in regulating the cell cycle, chromatin modification, receptor tyrosine kinase signaling, gene transcription, and other functions.[9, 10, 20] For example, one of the most commonly mutated genes is the cell cycle regulator TP53 (mutated in nearly half of MIBC), which is generally mutually exclusive with MDM2 amplifications (present in approximately 7% of MIBC). Other relatively common inactivating alterations occur in RB1 (mutation, deletion) and CDKN2A (deletion). Oncogenic alterations in several genes involved in cell signaling are also present, including activating hotspot mutations in receptor tyrosine kinases FGFR3 (and less frequently FGFR3 fusion and amplification), ERBB2 (including ERBB2 amplification) and ERBB3 and PIK3CA. Such alterations are important therapeutic targets in many cancer types but have not shown similar efficacy in bladder cancer, which could be related to mutation heterogeneity and clonality within the tumor or co-mutation pattern, as was reported by the neratinib basket trial for tumors with oncogenic ERBB2 and ERBB3 mutations.[21] One exception that is showing promise in the application of targeted therapy in UC is the presence of oncogenic alterations in FGFR3 (mutation, amplification and fusion). Recent reports showed that FGFR3-targeted agents erdafitinib and BGJ398 (infigratinib) demonstrated efficacy in patients with advanced and metastatic UC that harbored FGFR3 mutations and overexpression.[22, 23] Supported by these results, erdafitinib is now approved by the Food and Drug Administration (FDA) for the treatment of patients with locally advanced or metastatic UC with FGFR3 or FGFR2 alterations who have disease progression following chemotherapy.[24] Interestingly, the most frequent mutation in UC that was not reported in large scale exome-based sequencing studies is the recurrent point mutations in TERT promoter region. In many studies, TERT promoter mutations were identified in UC in different grades and stages and spectrum of morphologic variants, but not in benign proliferative conditions such as cystitis cystica et glandularis and florid Brunn nests,[25–28] or benign urothelial neoplasms such as urothelial papilloma and inverted papilloma.[29–32] These findings have important clinical implications in facilitating the distinction between UC with deceptively bland morphologic features, such as nested variant of UC, from its benign mimics (namely, cystitis cystica, florid Brunn nests and inverted urothelial papilloma). The detection of TERT promoter mutation in early stage and low-grade disease suggests that it is an early genetic event in the development of UC.[29] More importantly, this alteration can also be detected from urine samples, making it a potential biomarker for urothelial neoplasia that can be detected by noninvasive procedures.[33, 34] The genetic landscape of early stage UC is overall similar to that reported in MIBC but differs in the frequency of certain mutations. For example, mutations in FGFR3 and chromatin modifier KDM6A are significantly more frequent in early stage and low grade UC compared to MIBC; whereas mutations in TP53, RB1 and chromatin modifier ARID1A show the reverse pattern.[35]

Figure 1. The genomic landscape of MIBC.

as reported by the TCGA cohort [10] with co-occurring alterations in canonical cellular functions and signaling pathways [89, 90] From Cerami, E., et al., The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov, 2012. 2(5): p. 401–4, Gao, J., et al., Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal, 2013. 6(269): p. pl1.

MOLECULAR TAXONOMY OF UROTHELIAL CARCINOMA

Expression profiling of urothelial carcinoma by RNA and/or immunohistochemistry revealed several molecular subtypes and several classification schemes, some of which were linked to prognosis or response to a variety of therapies. Different classifications used different names for different subtypes, but there is significant overlap among them.[10, 36–39] The most comprehensive and widely used classifications are those proposed by the Lund University group and that of the TCGA.[10, 39, 40] To further provide uniformity to the terminology applied to RNA-based classification, a recent international effort led by the Bladder Cancer Molecular Taxonomy Group provided a consensus molecular classification of MIBC that was based on re-analyzing 1750 MIBC transcriptomic profiles from 18 datasets, 16 of which were already published (Figure 2, Figure 3).[41, 42] Central to these various classification systems is the presence of two major categories, luminal and basal, with additional subtypes into these two major categories. This basal-luminal division of MIBC was initially adopted following perceived similarities to the expression profiles of breast cancer. As such, basal bladder tumors typically express basal type keratins (e.g., KRT5, KRT6 and KRT14); whereas luminal tumors typically have high expression levels of FGFR3, transcription factors PPARG, GATA3, FOXA1, and ELF3, uroplakin genes found in umbrella cells, and KRT20.[36, 37, 43] The TCGA analysis further expanded and refined this classification and identified three subgroups within the luminal subtype along with a basal-squamous subtype as well a distinct neuronal cluster (total of five subtypes) (Figure 4). This identified subgroups within the luminal subtype; luminal-papillary, luminal-infiltrated and luminal, roughly corresponded to the luminal subtypes by consensus molecular classification, namely luminal papillary (LumP), luminal nonspecified (LumNS) and luminal unstable (LumU), and stroma-rich subtype. The TCGA luminal-papillary subtype is enriched with papillary tumors, lower stage and FGFR3 overexpression as a result of FGFR3 mutations, amplification, and FGFR3-TACC3 fusions. The TCGA luminal-infiltrated subtype is characterized by the presence of myofibroblast and smooth muscle gene signatures, lymphocytic infiltrate and increased expression of immune markers such as CD274 (PD-L1) and PDCD1 (PD-1). Similar expression patterns were linked to a wild-type p53 signature and chemoresistance in one study [36] but were reported to derive most benefit from anti-PD-L1 treatment in another study.[44] The basal subtype is enriched with tumors exhibiting squamous morphology by histopathologic review and was thus called basal-squamous but also included tumors that were devoid of unequivocal squamous differentiation.[10, 41] This subtype is associated with high expression of basal and stem-like markers such as CD44, KRT5, KRT6A, and KRT14 (Figure 5), and squamous differentiation markers such as desmocollins (DSC1–3) and desmogleins (DSG1–4), TGM1 (transglutaminase 1) and PI3 (elafin). The basal-squamous subtype is also enriched in TP53 mutations, more common in females, shows increased lymphocytic infiltrates and a strong immune gene signature expression. The TCGA neuronal subtype corresponded to the neuroendocrine-like consensus subtype (and small cell/neuroendocrine-like subtype by Lund University classification) and is enriched with tumors with neuroendocrine morphology and consistently showed relatively high expression levels of genes involved in neuronal differentiation and development, as well as typical neuroendocrine and neural differentiation markers such as chromogranin, PEG10, PLEKHG4 and TUBB2B.[10, 40, 41] Although the majority of tumors in this cluster (85%) had alterations in genes in the p53/cell-cycle pathway, only a subset was associated with mutations in both TP53 and RB1, which is the hallmark alteration in small cell/neuroendocrine carcinoma.[45] Nonetheless, this subtype was associated with the worst clinical outcome.[10, 46]

Figure 2. Consensus molecular classification of muscle invasive urothelial carcinoma.

(A) Clustered network of consensus classification based on 1750 transcriptomes originally classified according to six different input molecular subtypes. The circles inside each clique represent subtype contribution from each input subtype matched by color. Circle size is proportional to the number of samples assigned to the subtype. (B) Input subtypes repartitioned among each consensus class. (C) Relationship between subtyping results from the six input classification schemes. Samples are ordered by predicted consensus classes. Ba/Sq = basal/squamous; LumNS = luminal nonspecified; LumP = luminal papillary; LumU = luminal unstable; MDA = MD Anderson Cancer Center; TCGA = the Cancer Genome Atlas; UNC = University of North Carolina. [41] From Kamoun, A., et al., A Consensus Molecular Classification of Muscle-invasive Bladder Cancer. Eur Urol, 2020. 77(4): p. 420–433.

Figure 3. Summary of the main characteristics of the consensus molecular classes.

From top to bottom, proportion of consensus classes in the 1750 tumor samples; consensus class names; schematic graphical representation of tumor cells and their microenvironments (immune cells, fibroblasts, and smooth muscle cells); and a table listing the main characteristics such as oncogenic mechanisms, mutations, stromal infiltrate, immune infiltrate, histology, clinical characteristics, and median overall survival. Ba/Sq: basal/squamous; LumNS: luminal nonspecified; LumP: luminal papillary; LumU: luminal unstable; MIBC: muscle-invasive bladder cancer; NE: neuroendocrine; NK: natural killer. [41]. From Kamoun, A., et al., A Consensus Molecular Classification of Muscle-invasive Bladder Cancer. Eur Urol, 2020. 77(4): p. 420–433

Figure 4. TCGA molecular subtypes of muscle invasive urothelial carcinoma by RNA expression profiling.

Top to bottom; 5 mRNA expression subtypes (luminal-papillary, luminal-infiltrated, luminal, basal-squamous and neuronal); 4 previously reported TCGA subtypes; selected clinical covariates and key genetic alterations; normalized expression for miRNAs and proteins for selected genes. Samples within the three luminal subtypes, the basal-squamous subtype, and the neuronal subtype are ordered by luminal, basal, and neuroendocrine signature scores, respectively. [10] From Robertson, A.G., et al., Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer. Cell, 2017. 171(3): p. 540–556 e25.

Figure 5. Basal urothelial carcinoma.

In this example of invasive UC without morphologic evidence of squamous differentiation, there is near absence of luminal marker GATA3 expression and overexpression of basal markers CK5/6 and CK14, consistent with basal subtype.

RNA-based molecular classification has been applied in many studies as a means for patient risk stratification and association with outcome [10, 41, 46] and reported correlation with responses to chemotherapy or immunotherapy in advanced disease.[36, 44, 47] Although these molecular classifications provide important biologic insights of MIBC, molecular subtyping continues to evolve and as such may not be ready yet for routine clinical applications as several questions remain unanswered and prospective validation is still lacking. Further work and investigation are needed to delineate the exact roles of these different classification schema and the effect of intratumoral heterogeneity and therapeutic intervention on the stability of such molecular subtypes. This is highlighted by studies showing intratumoral heterogeneity of molecular subtype assignment to different parts of the same tumor[48, 49], to primary vs. metastatic tumors from the same patient [50], or to initial vs. recurrent or progressed tumors from the same patient.[51] There is also pre-clinical evidence of switch between luminal and basal subtypes in organoid generated from luminal bladder cancers,[52] further underscoring the need for more investigations into the biology and stability of these molecular subtypes.

MOLECULAR UPDATES ON HISTOLOGIC VARIANTS OF UROTHELIAL CARCINOMA

Squamous differentiation

Squamous differentiation (SqD) is the most common divergent differentiation in UC, identified in up to 30% of high- grade and/or stage disease.[3, 53] Expression profiling studies that included UC with SqD consistently reported that this tumor type nearly always clustered with the basal/squamous subtypes in virtually all classification schema.[10, 36–38, 54] These tumors characteristically show overexpression of high molecular weight keratins (CK5, CK6, and CK14) and epidermal growth factor receptor (EFGR) as well as downregulation of markers of urothelial differentiation such as uroplakins, GATA3, FOXA1, PPARG and thrombomodulin. As these studies were based on bulk samples with mixed squamous and urothelial components, they were unable to provide insights into the exact mechanisms involved in the development of squamous morphology in this setting.

Two studies investigated intratumoral heterogeneity within UC with SqD by analyzing separate areas of “urothelial” and “squamous” morphology from the same tumor.[48, 49] They reported that in a subset of cases the expression profiles of the urothelial area were classified as luminal whereas the squamous areas were mostly classified as basal/squamous.

Glandular differentiation

Glandular differentiation is reported in variable incidences in different studies, ranging from 8% to 18%.[53, 55–57] There is limited literature on the molecular characteristics of glandular differentiation in UC but the available evidence suggests that there is overlap with those of UC, particularly the presence of high rates of mutations in TERT promoter region and in chromatin remodeling genes.[27, 58]

Plasmacytoid urothelial carcinoma (PUC)

Plasmacytoid urothelial carcinoma (PUC) is a rare and aggressive variant of UC with a diffuse and infiltrating pattern of discohesive, individual or small clusters of tumor cells, generally with minimal stromal reaction. Tumor cell nuclei are typically eccentrically located, giving a superficial resemblance to plasma cells and, in most cases, tumor cells contain intracytoplasmic vacuoles that push and compress the nucleus resulting in signet ring cell morphology.[59–61] PUC usually presents at advanced stage and is associated with high mortality rate, high propensity for relapse and frequent peritoneal carcinomatosis despite sometimes the apparent initial response to chemotherapy.[59–63] Next generation sequencing and functional studies identified the presence of CDH1 truncating mutations, and less frequently CDH1 promoter hypermethylationas the defining feature of PUC and were specific to this histologic variant.[59] These mutations results in loss of E-cadherin expression in the majority of cases (Figure 6), Beyond CDH1 alterations, the overall genomic landscape of PUC is generally similar to that of UC-NOS, with frequent mutations in chromatin modifiers, cell cycle regulators and TERT promoter.[59, 64]

Figure 6. Plasmacytoid urothelial carcinoma.

The invasive tumor consists of diffusely infiltrating and discohesive tumor cells. Typical of this tumor, there is complete loss of E-cadherin expression in tumor cells. Next generation sequencing identified a CDH1 (X337_splice) mutation.

Of note, in contrast to the presence of germline CDH1 mutations in patients with hereditary diffuse gastric cancers and a subset of mammary lobular carcinoma, no germline CDH1 mutations were identified in PUC, including in a patient with two primary tumors driven by CDH1 loss, a primary breast lobular carcinoma and a primary bladder PUC.[59, 65].

Micropapillary urothelial carcinoma (MPUC)

Micropapillary urothelial carcinoma (MPUC) is a rare variant of UC with reported aggressive clinical behavior. By molecular analysis, MPUC consistently harbors higher rates of ERBB2 amplification than classic UC, which in some reports was associated with worse outcome following radical cystectomy.[66–68] The distribution of ERBB2 amplification within MPUC tumors is variable, but it is preferentially present in the micropapillary component in tumors containing both MP and UC areas[69] (Figure 7), despite the higher rate of ERBB2 amplification in the UC areas in these mixed MPUC-UC tumors than the reported rates in pure UC or those not mixed with MP components.[10, 20, 70] Recurrent hotspots ERBB2 mutations have been reported in bladder MPUC[71], but it is likely that their frequency is similar to that reported in classic UC. By expression profiling, MPUC is generally luminal and display enrichment of PPARG and suppression of p63 target genes.[41, 72]

Figure 7. Micropapillary urothelial carcinoma.

In this mixed micropapillary (MPUC) and urothelial (UC) tumor, there is HER2 overexpression (+3) in the micropapillary component only which was associated with 8-fold ERBB2 amplification by next generation sequencing. Consistent with a luminal phenotype, this tumor expresses luminal marker GATA3 in both components.

Small cell/neuroendocrine carcinoma of the bladder (SmCC)

Small cell/neuroendocrine carcinoma of the bladder (SmCC) is rare and is morphologically similar to that of the lung and other organs, and can be admixed with a urothelial or other divergent differentiation component.[53] By molecular analysis, SmCC is characterized by TP53 and RB1 co-mutations, which in one study, was identified in 90% and 87% of cases, respectively (80% of tumors displayed co-alterations of both genes).[45] Furthermore, even in tumors without RB1 deletion or loss of function mutations, there was loss of RB expression by immunohistochemistry, suggesting an alternative mechanism for RB loss, such as epigenetic silencing. Moreover, genes that were commonly mutated in UC were also found to be mutated in bladder SmCC, including TERT promoter mutations (95%) and truncating alterations in genes involved in chromatin modification such as CREBBP, EP300, ARID1A and KMT2D in approximately 75% of samples.[45, 73] This mutation pattern can be very helpful in determining a urothelial origin of a SmCC tumor from a metastatic location or when the site of origin is not obvious (for example, urothelial vs. prostatic vs. pulmonary).

SmCC is associated with a high level of chromosomal instability that is characterized by the presence of whole genome duplication in 72% of tumors, particularly those with TP53 missense mutations that resulted in biallelic silencing. Compared to that of the lung, bladder SmCCis enriched for APOBEC mutation signature in contrast to the dominant smoking signature that is associated with lung SmCC.[45, 54] These findings support the precursor urothelial origin of bladder SmCC. A recent study confirmed these observations and reported a urothelial-to-neural phenotypic switch with a dysregulated epithelial-to-mesenchymal transition network in bladder SmCC.[74]

Sarcomatoid urothelial carcinoma

Sarcomatoid urothelial carcinoma is a rare form of bladder cancer in which a component of the tumor exhibits mesenchymal phenotype that can in some tumors be a true heterologous elements in the form of cartilaginous, osseous, rhabdomyoblastic or other elements.[75–77] Like other variants of UC, it has been reported that the sarcomatous and urothelial components within the same tumor share common clonal origin and that sarcomatoid UC is enriched with mutations in TP53, RB1 and PIK3CA and is associated with dysregulation of epithelial-mesenchymal transition (EMT) network and overexpression of EMT markers.[76–79] By expression profiling they share features of the basal molecular subtype of UC, characterized by downregulation of luminal markers such as FOXA1 and GATA3, and upregulation of basal markers in a subset of tumors, such as keratins 5, 6 and 14.[41, 78]

Nested variant of urothelial carcinoma (NVUC)

Nested variant of urothelial carcinoma (NVUC) is another rare variant of UC characterized by deceptively bland morphologic features that can be associated with aggressive clinical course.[80, 81]. A “large nested” variant of UC is currently recognized where the invading tumor nests are noticeably large and irregular.[82, 83] Identifying this variant can be diagnostically challenging as it shares overlapping features with benign entities such as proliferative cystitis, von Brunn nest hyperplasia, nephrogenic adenoma, or inverted papilloma.[82, 84, 85] So far, a few molecular findings have been reported in this tumor type, the most common of which is the high rate of TERT promoter mutations that was not found in benign mimickers,[25, 26, 86] as well as occasional mutations in TP53, JAK3, and CTNNB1, suggesting that this tumor likely harbors molecular alterations similar to those of UC. A recent study on large nested UC reported FGFR3 mutations in the vast majority of cases.[87] Nested UC typically exhibits luminal expression pattern by studies that were based on IHC expression of luminal markers particularly FOXA1, GATA3 and CK20.[86–88]

Conclusion and future directions

Recent advances in the molecular biology and genomics of urothelial carcinoma provided great insights into the molecular mechanisms underlying the development, progression and heterogeneity of this disease. The recently proposed molecular classifications provide additional important insights into the biology of MIBC and a consensus on a molecular classification scheme that is applicable to routine clinical use is urgently needed. Ideally such classification will help to address several unanswered questions including how stable these molecular subtypes are within a given tumor (questions of intratumoral heterogeneity within the primary tumor, or primary vs. metastasis), and following therapy. It would also be helpful to determine whether molecular subtyping should be incorporated into clinical decision-making, and if so, in which disease states and/or treatment settings? Addressing these questions should hopefully enhance our ability to provide novel and precise biomarker-directed therapies.