67. Молекулярная биология рака мочевого пузыря

Введение

Рак мочевого пузыря давно подразделяется на два основных типа — немышечно-инвазивный рак мочевого пузыря (NMIBC) (75%) и мышечно-инвазивный рак мочевого пузыря (MIBC) (25%). NMIBC включают опухоли, которые не проникли через эпителиальную базальную мембрану (стадия Ta), высоко-дифференцированное опухоли, которые инвазируют субмукозу, но не прилегающую мышцу (стадия T1), и рак in situ (CIS), высоко-дифференцированное поражение с высоким риском прогрессии в MIBC. Та опухоли обычно рецидивируют, но инвазия в мышечный слой наблюдается редко (от 10% до 15%), и прогноз хороший. Напротив, пациенты с MIBC (стадия T2 или выше) имеют плохой прогноз (<50% выживаемости через 5 лет). Опухоли T1 стадии представляют собой клинически проблемную и молекулярно гетерогенную группу с чертами, ассоциированными как с NMIBC, так и с MIBC. Более 90% случаев рака мочевого пузыря — это уротелиальные карциномы (UC), которые разделяют черты нормальной эпителиальной выстилки мочевого пузыря, и именно эти виды рака были наиболее изучены. Редкие эпителиальные варианты включают плоскоклеточную карциному, аденокарциному и мелкоклеточную карциному. Вследствие только ограниченного молекулярного анализа этих типов рака, они здесь не обсуждаются.

В предгеномную эру молекулярные изменения были выявлены в ключевых кандидатных генах, с разной частотой альтераций в них в NMIBC и MIBC, поддерживая канцерогенную модель двух путей. Последние достижения в области высокопроизводительных общегеномных технологий позволили проводить анализ ДНК, РНК и протеинах с высоким разрешением, а их применение для изучения большого количества опухолей в рамках таких проектов, как Атлас Генома Рака (TCGA), значительно расширило наше понимание молекулярных характеристик этого гетерогенного заболевания и выявило перекрывающиеся черты, которые характерны для стадии и дифференцировки опухоли. В этой главе кратко излагаются современные знания о молекулярном ландшафте рака мочевого пузыря и перспективы для улучшения наблюдения за болезнями, перехода к более персонализированной медицине и разработки новых методов лечения.

Мутационный пейзаж

Исследования кандидатных генов идентифицировали ключевые гены, вовлеченные в рак мочевого пузыря — FGFR3, PIK3CA, CDKN2A, TP53, TSC1, RB1, STAG2 и семейство RAS генов. Исследования секвенирования следующего поколения (NGS) подтвердили частые изменения в этих генах, выявили дополнительные онкогенные факторы и гены-супрессоры опухолей (Таблица 67.1) и предоставили информацию о мутационных процессах, которые формируют геномы опухолей мочевого пузыря. В следующих разделах представлен обзор соматических мутаций и структурных изменений, которые в настоящее время определяют мутационные ландшафты NMIBC и MIBC.

Частота мутаций, мутационные сигнатуры и мутационные процессы

TCGA исследование, проведенное в 2014 году на 131 MIBC, не получавших химиотерапию, выявило высокую частоту соматических мутаций (средняя и медианная частота соматических мутаций 7,7 и 5,5 мутаций на megabase [Мб] соответственно), что сходно с данными, полученными для немелкоклеточного рака легких и меланомы. Это было подтверждено в расширенном TCGA исследовании 412 опухолей, где были зарегистрированы средняя и медианная частота несинонимичных мутаций 8,2 и 5,8 мутаций на 1 Мб, соответственно. В NMIBC частота мутаций значительно ниже, в исследованиях сообщается о 2,4 (среднее) и 1,64 (медиана) всех мутаций на 1 Мб в Та опухолях и 1,8 несинонимичных мутаций на 1 Мб (медиана) в когорте опухолей Ta и T1 стадий.

In addition to overall mutation burden, the pattern of base changes or “mutational signature” can provide insight into underlying mutational processes. To date, there are 30 such signatures held in the Catalogue of Somatic Mutations in Cancer (COSMIC) database (http://cancer.sanger.ac.uk/cosmic/signatures). Bladder cancer is strongly linked to smoking, but genome sequencing studies have not detected the more common smoking- related signature characterized by C > A transversions (COSMIC signature 4). Instead, studies suggest that APOBEC activity and deficiencies in the nucleotide excision repair (NER) pathway may contribute to the mutation spectra observed in bladder cancer.

A signature characterized by C > T and C > G mutations at TCW motifs, where W is A or T, has been attributed to the activity of the APOBEC family of cytosine deaminases.7 Several studies have revealed that mutational loads in both NMIBC and MIBC are mainly driven by APOBEC-mediated mutagenesis. Expression of APOBEC3B correlates with APOBEC mutagenesis in MIBC,2 and in NMIBC, expression of APOBEC3A and APOBEC3B was associated with a specific expression subtype.9 In a recent study in primary Ta tumors, expression of APOBEC3H was implicated and showed association with a copy number subtype.5

ERCC2 is a DNA helicase, which plays a core role in the NER pathway. Kim et al.12 reported that ERRC2 mutation status was significantly associated with a mutational signature characterized by a broad spectrum of base changes. Examination of three independent cohorts consisting mainly of MIBC revealed a strong association between somatic ERCC2 mutations and this signature, suggesting that this is driven by loss of normal NER function. The presence of the signature was associated with tobacco exposure, which creates DNA adducts that are typically repaired by the NER pathway.

In the 2017 TCGA study, non–negative matrix factorization (NMF) analysis of single nucleotide variants (SNVs) classified into 96 base substitution types within the trinucleotide context was used to identify the processes contributing to the high mutation rate.11 This revealed five mutation signatures including two variants of the APOBEC mutagenesis signature (APOBEC-a and APOBEC-b) that accounted for 67% of all SNVs. The other signatures included one characterized by C > T transitions at CpG motifs, likely resulting from spontaneous deamination of 5-methylcytosine, a POLE signature in one hypermutated sample that carried a functional mutation in POLE (DNA polymerase epsilon) that is predicted to affect its proofreading activity, and the ERCC2 signature.12 Unsupervised cluster analysis by mutation signature identified four clusters (MSig1 to MSig4). Patients with MSig1 cancers, characterized by high APOBEC signature mutation load, high mutation burden, and a high predicted neoantigen load, had the highest (75%) 5-year survival probability. In contrast, patients with MSig2 cancers, which had the lowest mutation rate, had the poorest (22%) 5-year survival probability. MSig4 samples had enrichment of ERCC2 signature mutations and ERCC2 mutations.

Таблица 67.1. Часто мутированные гены в NMIBC и MIBC

Генa Hurst et al., 2017b Nordentoft et al., 2004c Pietzak et al., 2017d Pietzak et al., 2017d Guo et al., 2013e TCGA 2017f
Ta (%) Ta (%) Ta (%) T1 (%) T1 (%) MIBC (%)
FGFR3 79 40 66 30 25 14
PIK3CA 54 25 36 22 6 22
KDM6A 52 65 50 43 50 26
STAG2 37 25 24 22 25 14
KMT2D 30 15 31 26 0 28
ARID1A 18 35 25 27 6 25
EP300 18 25 20 8 16 15
CREBBP 15 20 23 19 12 12
KMT2C 15 20 16 5 3 18
RHOB 13 0 ND ND 0 11
HRAS 12 10 2 8 16 9
KMT2A 11 0 9 11 9 11
TSC1 11 5 5 22 12 8
BRCA2 10 0 11 11 0 7
COL11A1 10 0 ND ND 0 5
RBM10 10 20 22 5 0 9
TP53 4 5 11 35 25 48
FAT1 (2) 10 13 17 0 12
KRAS 2 0 11 8 6 4
ATM (1) 5 13 19 3 14
CDKN1A (1) 0 11 13 0 9
ELF3 (1) 25 ND ND 12 12
ERCC2 (1) 0 21 13 6 9
ERBB2 (0) 0 11 19 3 12
ERBB3 (0) 0 9 19 3 10
RB1 (0) 5 0 5 9 17

a Genes mutated at ≥10% in the studies of Hurst et al.5 and The Cancer Genome Atlas (TCGA) 201711 are shown.

b Data from Hurst et al.5 Sample cohort consists of 79 TaG1/TaG2 and 3 TaG3 tumors. Exome sequencing of 24 tumors and targeted sequencing of 58 tumors was carried out. Mutation frequency is shown in parentheses where only exome sequencing data is available.

c Data from Nordentoft et al.6 Exome sequencing was carried out on 20 TaG1/TaG2 tumors.

d Data from Pietzak et al.21 Targeted sequencing of 55 Ta (23 grade 1/2; 32 grade 3) and 38 T1 tumors was carried out.

e Data from Guo et al.16 Exome sequencing was carried out on 32 T1 tumors.

f Data from TCGA 2017.11 Exome sequencing was carried out on 412 MIBCs.

NMIBC, non–muscle-invasive bladder cancer; MIBC, muscle-invasive bladder cancer; ND, genes were not covered by target capture design.

FGFR3, PIK3CA и RAS гены

Activating point mutations in fibroblast growth factor receptor 3 (FGFR3) are present in ≥70% of Ta cases.13 These are in hot-spot codons in exons 7, 10, and 15 and are all predicted to constitutively activate the receptor. The frequency of mutation is lower in stage T1 NMIBC (10% to 45%) and MIBC (approximately 15%) (see Table 67.1).13 In high-grade T1 (T1G3) tumors, FGFR3 and TP53 mutations show an independent distribution, unlike the situation in Ta tumors where these mutations are virtually mutually exclusive.14,15 Mutations are also found in urothelial papilloma, a likely precursor of NMIBC superficial UC.13 Increased expression of mutant FGFR3 protein is common in these tumors. Although only 15% of MIBC show FGFR3 mutation, protein expression is upregulated in 40% to 50% of nonmutant MIBC.13 An alternative mechanism of FGFR3 activation in a subset of cases (2% to 5%) is chromosomal translocation to generate fusion proteins.11,16,17 In cultured normal human urothelial cells, expression of mutant FGFR3 leads to activation of the RAS-MAPK pathway and PLCγ, leading to overgrowth of cells at confluence and suggesting a possible contribution of FGFR3 activation to urothelial hyperplasia in vivo.13

The phosphatidylinositol 3-kinase (PI3K) pathway plays a pivotal role in signaling from receptor tyrosine kinases. Activating mutations of the p110α catalytic subunit (PIK3CA) are common in low-grade, stage Ta (approximately 40% to 50%), compared to stage T1, NMIBC and MIBC (approximately 20%) (see Table 67.1).18 Missense mutations E542K and E545K in the helical domain are most common (22% and 60%, respectively), and the kinase domain mutation H1047R, which is the most common mutation in other cancers, is less frequent. A recent NGS-based study of primary stage Ta NMIBC reported that 17 of 48 PIK3CA mutations detected, some of which have confirmed gain of function, were not in these three major hotspot codons.5 This warrants caution when using assays that detect only hotspot mutations, especially in a clinical trial setting.

Activating hotspot mutations in the RAS gene family occur most commonly in HRAS or KRAS and, unlike FGFR3 and PIK3CA mutations, are not associated with either NMIBC or MIBC (mutations in approximately 10% overall) (see Table 67.1). PIK3CA mutation commonly co-occurs with either FGFR3 or RAS mutation in NMIBC.5,19 However, RAS and FGFR3 mutations are mutually exclusive,20 perhaps reflecting the fact that both activate the RAS-MAPK pathway. A recent study reported mutually exclusive alterations in FGFR3 and the receptor tyrosine kinase ERBB2 in 57% of high-grade (Ta, T1, CIS) UC.

Промотор обратной транскриптазы теломеразы

Mutations in the promoter of telomerase reverse transcriptase (TERT) represent the most common genomic alteration identified in UC to date, occurring at high frequency (60% to 80%) across all stages and grades.22,23 The high frequency of mutation suggests that this is an early event and a requirement in all pathways of urothelial tumorigenesis. Interestingly, the frequency of mutation in early-onset disease is reported to be much lower (46%), perhaps suggesting different mechanisms of tumorigenesis in young patients.24 Mutations are predominantly in two hotspot positions (−124 bp [G > A] and −146 bp [G > A] relative to the ATG translational start site), and this has facilitated the design of robust methods of detection. The ease with which these mutations can be detected in urine sediments22,23 is likely to make a major contribution to the development of noninvasive urine-based assays for detection of bladder tumors of all grades and stages in both diagnostic and disease monitoring settings. These mutations create binding sites for ETS/TCF transcription factors and are predicted to increase transcriptional activity.25 The effect of mutation on disease recurrence has been shown to be modified by the presence of a common polymorphism (rs2853669) within a preexisting ETS/TCF binding site in the promoter region, with mutations in the absence of the variant allele being associated with increased disease recurrence in NMIBC.26

TP53, RB1 и CDKN2A

As in other aggressive cancers, the tumor suppressor genes TP53, RB1, and CDKN2A are implicated in MIBC. The pathways controlled by p53 and RB1 regulate cell-cycle progression and responses to stress. The TCGA study has reported alterations in the p53/cell-cycle pathway, including TP53 mutation, MDM2 amplification or overexpression, RB1 mutation or deletion, and CDKN2A mutation or deletion, in 89% of MIBC.11

TP53 is the most commonly mutated gene in MIBC (approximately 50%).11 Mutations are very infrequent in low-grade Ta tumors (approximately 1%) but occur at higher frequency in T1 tumors (see Table 67.1).27 Detection of mutation or p53 protein accumulation is associated with poor prognosis. Although immunohistochemical detection of p53 with increased half-life identifies many mutant p53 proteins and has commonly been used as a surrogate marker for mutation, some TP53 mutations (approximately 20%) yield unstable or truncated proteins that cannot be detected in this way. Thus, p53 protein accumulation is not a useful prognostic marker. Two meta-analyses indicate only a small association between p53 positivity and poor prognosis.28,29 However, examination of both protein expression and mutation of TP53 provides more useful prognostic information.30

The RB pathway regulates cell-cycle progression from G1 to S phase. Deletion of 13q14 and loss of RB1 protein expression are common in MIBC.1 Loss of p16 expression is inversely related to positive RB1 expression, and high-level p16 expression results from negative feedback in tumors with loss of RB1. Thus, both loss of expression and high level expression of p16 are associated with RB pathway deregulation, and these are adverse prognostic biomarkers found in >50% of MIBC.1 Interestingly, in MIBC with FGFR3 mutation, a high frequency of CDKN2A homozygous deletion (HD) has been reported, which may identify a progression pathway for noninvasive FGFR3-mutant tumors to muscle invasion via loss of CDKN2A.11,31,32 Amplification and overexpression of E2F3, which is normally repressed by RB1, is associated with RB1 or p16 loss in MIBC.33 p16 and p14ARF proteins link the RB and p53 pathways, and due to multiple regulatory feedback mechanisms, inactivation of both pathways together is predicted to have greater impact than inactivation of either pathway alone. This is borne out by the achievement of greater predictive power in studies using concurrent analyses of multiple changes that deregulate the G1 checkpoint.34

Гены, вовлеченные в модификацию и архитектуру хроматина

One of the major findings of genome sequencing studies in UC is the high frequency of mutations in chromatin modifier (CM) genes including the histone demethylase (KDM6A), the histone methyltransferases (KMT2A, KMT2C, KMT2D), the histone acetyltransferases (CREBBP, EP300), the SWI/SNF complex genes (ARID1A, ARID4A), and the polycomb group genes (ASXL1, ASXL2) (see Table 67.1). Mutations were reported in the first exome sequencing study of UC,35 and subsequent studies have demonstrated that such mutations are common in tumors of all stages and grades and are most frequent in NMIBC.5,6,21 Many mutations in these genes are inactivating (small deletions/insertions, nonsense, essential splice site). ARID1A, CREBBP, and KDM6A are also targets in large deletions, suggesting that all of these genes have a tumor suppressor function.11

KDM6A, a histone demethylase that catalyzes the demethylation of tri-/dimethylated histone H3 lysine 27 (H3K27me2/3), is the most frequently mutated CM gene in NMIBC,5,6,21 with mutations occurring at higher frequency (38% to 65%) than in MIBC (26%) (see Table 67.1).11 KDM6A associates with KMT2C/D in a COMPASS-like complex, which acts to maintain gene expression. KDM6A, through the removal of methyl groups from H3K27me2 and H3k27me3 that are associated with promoter silencing, antagonizes the methyltransferase activity of EZH2 in polycomb-repressive complex 2 (PRC2). It also erases H3K27me3 at poised/silenced enhancers. KMT2C/D are methyl transferases that write the H3K4Me3 mark associated with active promoters and the H3K4me mark at enhancers. The predicted effect of inactivation of these genes is therefore to silence transcription via effects on both promoters and enhancers. Loss of KDM6A results in enrichment of PRC2-regulated pathway signaling, and thus, KDM6A-null or PRC2-enriched cells are sensitive to EZH2 inhibition.36 Studies have implicated loss of KDM6A function as an early event in tumorigenesis.6,37 Interestingly, KDM6A (Xp11.3) has been reported to show more mutations in noninvasive tumors from females than males, perhaps indicating gender differences in the epigenetic landscape of the normal bladder.5 Interrogation of publicly available exome-sequencing data has not revealed a similar association in MIBC.

Инактивирующие мутации в ARID1A ассоциированы с высоко дифференцированным раком мочевого пузыря (таблица 67.1). Пациенты с опухолями, несущими мутации ARID1A, имеют значительно худшую безрецидивную выживаемость после индукционного курса бациллами Calmette-Guérin (BCG). Будущие исследования необходимы, чтобы выяснить, могут ли мутации ARID1A предсказать ответ на BCG или они просто идентифицируют подгруппу пациентов с плохим прогнозом. Недавнее исследование светлоклеточной карциномы яичника (OCCC) показало, что тераптя ингибитором EZH2 метилтрансферазы (GSK126) является синтетически летальным в ARID1A мутантной OCCC. Такое лечение может представлять валидный подход для UC с инактивированным ARID1A.

STAG2

Inactivating mutations in the cohesin complex component STAG2 (Xq25) have been identified in NMIBC and MIBC by single gene analyses40,41 and whole-exome sequencing (see Table 67.1).5,16,38 Cohesin plays an important role in ensuring accurate chromosome segregation at mitosis, and in some cancer types, STAG2 mutation has been associated with aneuploidy.42 However, studies in bladder cancer have reported conflicting results. Some have reported that mutations are more prevalent in genomically stable NMIBC,5,38,41 whereas others have associated loss of STAG2 with aneuploidy.16,40 The relationship of STAG2 mutation status to prognosis is also ambiguous, with two studies reporting worse prognosis for patients with STAG2-mutant tumors16,40 and another indicating that loss of STAG2 was associated with better prognosis for patients with either NMIBC or MIBC.38 This may reflect differences in the stages and grades of the tumor cohorts used in individual studies and/or alternative roles for cohesin in bladder cancer.43 For example, in addition to its role in chromosome segregation, cohesin plays a role in anchoring the base of chromatin loops to facilitate long range chromatin interactions that regulate transcription. As STAG2 mutations are commonly found in association with mutations in other CM genes, loss of function may contribute to an overall pattern of gene silencing.

Изменения в путях репарации ДНК

DNA-damaging agents play an important role in the treatment of both NMIBC and MIBC. Mitomycin C is an alkylating agent that is administered intravesically following transurethral resection of NMIBC, and first-line systemic chemotherapy for MIBC involves the use of platinum-based DNA-damaging agents such as cisplatin administered in adjuvant or neoadjuvant settings. The therapeutic efficacy of these drugs exploits deficiencies in DNA repair pathways, and there is much interest in exploring alterations in these pathways and identifying markers that can help guide therapy. Somatic mutations in DNA damage repair (DDR) genes including ATM, ATR, ERCC2, ERCC4, BRCA1, BRCA2, CHEK2, PALB2, POLE, FANCA, FANCC, FANCD2, FANCM, and MSH6 have been reported in NMIBC and MIBC (see Table 67.1).5,11,21,38 In the 2017 TCGA study, the most frequently mutated DDR genes identified were ATM (14%) and ERCC2 (9%) (see Table 67.1). Recurrent somatic mutations in ERCC2 have been reported in several studies12,16,21,44 and are associated with improved outcomes in cisplatin-treated patients.44 A recent study employing a targeted capture-based NGS test, MSK-IMPACT, identified a high frequency (30%) of alterations in DDR genes, including ERCC2, in high-grade NMIBC.21 These alterations were associated with a larger mutational burden and a predicted higher neoantigen burden, suggesting that treatment with BCG or immune checkpoint inhibitors may be viable therapeutic approaches in these patients. The expression levels of DDR pathway components have been much studied in relation to outcome, prognosis, and treatment response. For example, low expression of the NER pathway component ERCC1 has been associated with better overall survival in patients with metastatic UC treated with cisplatin-based chemotherapy,45 and high levels of MRE11A, a homologous recombination pathway protein involved in the repair of double-strand breaks, was associated with better outcome in radiation-treated patients.46

Структурные изменения в геноме

Structural alterations in the genomes of bladder tumors include allelic loss, DNA copy number gains and losses, and rearrangements. The genomes of NMIBC and MIBC are very different. NMIBC, especially stage Ta tumors, are typically diploid or near-diploid and exhibit very few copy number alterations.5 In contrast, MIBC can be highly aneuploid and exhibit many genomic alterations.11,15 It is notable that some stage T1 tumors show similar profiles to MIBC, suggesting that these tumors with the ability to break through the basement membrane may be aggressive lesions. However, other T1 tumors show remarkable similarity in their copy number profiles to Ta tumors, suggesting that distinct biologic subgroups exist.15

The most common genomic alteration in both NMIBC and MIBC is loss of heterozygosity (LOH) or copy number loss of chromosome 9. More than 50% of UC of all grades and stages show chromosome 9 loss. A critical region on 9p contains CDKN2A (9p21), which encodes the two cell-cycle regulators, p16 and p14ARF.1 p16 is a negative regulator of the RB pathway, and p14ARF is a negative regulator of the p53 pathway. TSC1 is the best validated 9q tumor suppressor gene, with biallelic inactivation in approximately 12% to 16% of cases.47,48 TSC1 in complex with TSC2 negatively regulates the mammalian target of rapamycin (mTOR) branch of the PI3K pathway.

High-level DNA amplification occurs at a low frequency in NMIBC and is mainly associated with high-grade and T1 tumors.15,49 Other alterations reported in stage Ta NMIBC include losses of 10q, 11p, 11q, 17p, 19p, and 19q and gains of 20q.15,50 In contrast, the genomes of MIBC are very complex, exhibiting many copy number alterations and rearrangements.11,15,50 High-level amplifications are common, with candidate regions containing key genes involved in bladder cancer including CCND1, CCNE1, E2F3, EGFR, ERBB2, FGFR3, MDM2, MYCL1, PPARG, and YWHAZ.11,15,33,50 The most common region of homozygous deletion (HD) reported is 9p21 containing CDKN2A. Other key regions of HD include 10q23 containing PTEN, a key regulator of the PI3K pathway, and 13q14 (RB1). Recurrent focal deletions at 14q24 containing RAD51B are also reported.11

Genome doubling in combination with increased tolerance to chromosome aberrations is proposed to accelerate cancer genome evolution,51 and this event has been reported in bladder cancer.11,52 The 2017 TCGA study reported that TP53 mutations were enriched in tumors with genome-doubling events, supporting the suggestion that loss of p53 facilitates this.11,52

Several gene fusions have been reported in bladder cancer. The most common is an intrachromosomal FGFR3- TACC3 fusion.11,16,17 All fusions identified to date show loss of the final exon of FGFR3 with frequent fusion in- frame to transforming acid coiled-coil containing protein 3 (TACC3). These fusion proteins are highly activated and transforming oncogenes.17 Other FGFR3 fusions include FGFR3-BAIAP2L117,53 and FGFR3-TNIP2.21 FGFR3 fusions have mainly been reported in MIBC but have also been found in two NMIBC-derived cell lines and a low-grade Ta tumor.21 The 2017 TCGA study reported fusions involving PPARG (TSEN2-PPARG, MKRN2-PPARG), with PPARG expression being higher in samples with fusions than in those without.11 The majority retained the DNA- and ligand-binding domains of PPARG, suggesting that they are functional. Two other studies have described PPARG activation in bladder cancer cells through PPARG amplification or mutation or RXRA S427F mutation, suggesting that PPARG may represent a candidate therapeutic target in these tumors.54,55

Гетерогенность и клональная эволюция

Multifocality and/or development of multiple recurrent tumors in the same patient is a common feature of UC. Although some patients develop more than one molecularly distinct tumor (oligoclonal disease),56 in most cases, tumors from the same patient are molecularly related (monoclonal).57

Several studies in NMIBC have sequenced individual tumors, synchronous multifocal tumors, primary and recurrent tumors from the same patient, and samples collected before and after disease progression.6,8,10,58 A higher intrapatient variation of the tumor mutation spectrum and a higher frequency of APOBEC- related mutations was reported in patients with progressive disease, implying that APOBEC activity in these tumors was a later tumor-specific event.10 Monoclonality was also confirmed in this study. Nordentoft et al.6 analyzed paired samples from patients with progressive disease and showed that although noninvasive and invasive tumors shared multiple identical mutations indicating a common origin, progressed tumors also exhibited much divergence.6

Recent genome-sequencing studies have revealed much information regarding intratumor heterogeneity (ITH), with clonal diversity being strongly associated with higher tumor stage and grade.59 Sequencing of primary bladder tumors and metastatic lesions from the same patient has revealed low spatial heterogeneity in primary tumors and a much higher level of ITH in metastases.60 Multiregional analysis of cystectomy specimens from patients with multifocal or unifocal disease has also revealed higher spatial heterogeneity in multifocal lesions.61 Analysis of adjacent “normal cells” in this study detected more mutations in the samples from patients with multifocal disease than in those with unifocal disease, suggesting intraepithelial migration or seeding from tumors. The presence of genomic alterations in morphologically normal urothelium in tumor-bearing bladders has been widely reported.62,63 Where the normal cell population uniformly contains alterations, this has been interpreted as clonal expansion of altered cells within the urothelium to generate “fields” of altered cells within which tumor development occurs following acquisition of additional changes.

ITH in 16 matched sets of primary and advanced tumors prospectively collected before and after chemotherapy has also been assessed.64 Intrapatient mutational heterogeneity in the chemotherapy-treated samples was evident, with most mutations not shared with pretreatment samples. This suggests that care is warranted when using primary samples to guide treatment of metastatic disease.

Молекулярные подтипы

Bladder tumors of similar grade and stage commonly show divergent clinical behavior. In particular, stage T1 tumors show considerable molecular and clinical diversity. Until recently, molecular features have failed to explain or predict this heterogeneity. The two tumor groupings (NMIBC and MIBC) that have for so long dominated the bladder cancer literature are insufficient for this. Recent whole-genome DNA-based and RNA- based (transcriptome) studies have now begun to unravel this complexity, revealing multiple subgroups that are independent of conventional grade and stage groupings. The following sections describe the main findings of these studies to date.

ДНК-базисные подтипы

DNA copy number and mutation status has identified multiple subgroups of tumors within the conventional grade and stage groupings,15 although these have not been as extensively studied as expression-based subtypes and classification signatures are not yet fully defined. Hierarchical cluster analysis of copy number data for 49 high- grade stage T1 tumors defined three clusters, one of which was associated with disease progression.15 This study also separated 58 stage Ta tumors into two copy number groups. In a larger panel of tumors (n = 140), the existence of two major genomic subtypes of primary stage Ta tumors with distinct copy number profiles was confirmed.5 Genomic subtype 1 (GS1) contained no or few copy number alterations, whereas genomic subtype 2 (GS2) was characterized by a higher level of genomic instability, in particular loss of 9q (including the TSC1 genomic region). Whole-exome and targeted sequencing analysis revealed that GS2 tumors have a higher mutation rate, enrichment of APOBEC-related signatures, and more mutations in TSC1. Consistent with loss of either one or both copies of TSC1 (a regulator of mTORC1 activity), GS2 tumors had upregulated mTORC1 signaling.

Транскриптом-базисные подтипы

Initial assessment of messenger RNA (mRNA) expression profiles of UC of all grades and stages by the Lund group identified two major molecular subtypes (MS1 and MS2), separated mainly, though not entirely, according to grade and stage with stage T1 tumors distributed relatively equally between the two subtypes.65,66 The same group subsequently reported a novel molecular taxonomy of UC based on transcriptional profiling of 308 bladder tumors of all stages and grades.67 Five major subtypes were identified: urobasal A (UroA), urobasal B (UroB), genomically unstable (GU), squamous cell carcinoma–like (SCCL), and “infiltrated” (Fig. 67.1). Tumors in the latter group were highly infiltrated with nontumor cells, whereas the definition of the other groups reflected tumor cell–specific criteria. Clear differences in expression of cell-cycle regulators, keratins, receptor tyrosine kinases, and cell adhesion molecules were evident. UroA and UroB subtypes expressed high levels of FGFR3, CCND1, and TP63; GU tumors showed low levels of these proteins but expressed high levels of ERBB2 and E-cadherin, and SCCL tumors expressed P-cadherin and high levels of KRT5, KRT14, and proteins involved in keratinization. These subtypes showed distinct clinical outcome. UroA had good prognosis, GU had intermediate prognosis, and SCCL and UroB, the worst prognosis. UroB tumors shared epithelial characteristics with UroA tumors including FGFR3 mutation, but they also had TP53 mutation and were often invasive. Stage T1 tumors appeared evenly distributed between molecular subtypes. The Lund group subsequently reported immunohistochemistry markers that could distinguish these subtypes.

Subsequent RNA-profiling studies of MIBC by three other groups (MD Anderson Cancer Center [MDA], University of North Carolina [UNC], TCGA) have all identified two major subtypes that show considerable overlap with the Lund subtypes (see Fig. 67.1).2,69,70 These subtypes were termed “luminal” and “basal” because they express markers of urothelial differentiation and normal basal cells of the urothelium, respectively, and show similarity to basal and luminal subtypes of breast cancer.71 Basal tumors typically express high levels of KRT5, KRT6, KRT14, and CD44, and luminal tumors are characterized by high expression of FGFR3, uroplakins, and the transcription factors PPARG, GATA3, and FOXA1. The MDA group also described a “p53-like” subset of luminal MIBC characterized by the expression of luminal markers and genes expressed by cancer-associated fibroblasts, which corresponds to the infiltrated subtype described in the Lund study. The first TCGA study of 131 MIBC identified four clusters (I to IV), enriched for luminal (I and II) and basal (III and IV) markers. Clusters I and II corresponded to the luminal and “p53-like” subtypes, respectively, described in the MDA study. Cluster III overlapped with the basal subtype, and cluster IV was similar to the claudin-low breast cancer subtype.72 These expression subtypes have shown relationships to outcome and therapeutic response.

The TCGA have reported on the most comprehensive omic- based study of MIBC to date.11 This confirms overlap with the basal and luminal subtypes, and refines and adds to the current consensus. Using NMF consensus clustering of RNA-seq data, five mRNA expression-based subtypes (luminal, luminal-papillary, luminal- infiltrated, basal-squamous, and a new “neuronal” subtype) were identified (see Fig. 67.1). The majority of tumors within the luminal subtypes express high levels of uroplakins (UPK1a and UPK2) and urothelial differentiation markers (FOXA1, GATA3, PPARG). The luminal-papillary subtype consists mainly of tumors with papillary architecture, lower stage (T2), and high purity. These are characterized by FGFR3 overexpression and enriched for FGFR3 mutations and amplifications, and FGFR3-TACC3 fusions. A low CIS expression signature score73 and high expression of genes involved in sonic hedgehog signaling (SHH and BMP5) are also characteristic. Based on these observations, it was suggested that tumors in this group may have developed from precursor NMIBC. Uroplakins are highest in the luminal subtype as are genes expressed in terminally differentiated urothelial umbrella cells (KRT20, SNX31). The luminal-infiltrated subtype is less pure than the other luminal subtypes, containing lymphocytic infiltrates and expressing high levels of smooth muscle and myofibroblast signature genes. This subtype shares features with the MDA “p53-like” subtype that has been associated with chemoresistance.69 The majority of luminal-infiltrated tumors were classified as cluster II in the previous TCGA study and have increased expression of the immune checkpoint markers, programmed cell death protein 1 (PD-1) and programmed cell death protein ligand 1 (PD-L1). It was reported that patients with cluster II subtype tumors respond best to the anti–PD-L1 treatment atezolizumab.74 The basal-squamous subtype expresses high levels of basal and stem cell markers (CD44, KRT5, KRT6A, KRT14) and markers of squamous differentiation (TGM1, DSC3, PI3). This subtype was more common in females, and a high proportion of tumors had squamous histology. Enrichment of TP53 mutations, a strong CIS signature, and low sonic hedgehog signature gene expression led to the suggestion that these tumors may have developed from basal cells and CIS lesions. Expression of immune markers is highest in this subtype, reflecting the relatively low purity of the samples.

Фигура 67.1. Молекулярные подтипы рака мочевого пузыря. A: Three messenger RNA (mRNA) expression subtypes of non–muscle-invasive bladder cancer (NMIBC) (class 1, class 2, and class 3) were defined by the UROMOL study.9 The main molecular features of each subtype are shown. B: Overview and proposed overlap of mRNA expression subtypes defined in six studies carried out by four groups: MD Anderson Cancer Center (MDA),69 University of North Carolina (UNC),70 Lund University (LUND),67,75 and The Cancer Genome Atlas (TCGA).2,11 The LUND studies included both NMIBC and muscle-invasive bladder cancer (MIBC), whereas the MDA, UNC, and TCGA studies included only MIBC. The newly described neuronal (TCGA) and small- cell/neuroendocrine-like (Sc/NE) (LUND) subtypes are shown on the right-hand side of the figure.

The overlap of this subtype with the other described subtypes is yet to be fully defined. Key molecular features of the five subtypes recently described in the TCGA 2017 study are shown in the bottom panel. UroA, urobasal A; SCC, squamous cell carcinoma; UroB, urobasal B; Epi-Inf, epithelial-infiltrated; Mes-Inf, mesenchymal-infiltrated; SCCL, squamous cell carcinoma–like.

A newly described “neuronal” subtype is characterized by high expression of neuronal differentiation and development genes, and typical neuroendocrine markers, although the majority of tumors lacked small-cell or neuronal histology. Alterations in genes affecting the p53/cell-cycle pathway, including TP53 and RB1 mutations and E2F3 amplifications, are common in this subtype, which is also characterized by the poorest survival. A similar subtype (small-cell/neuroendocrine-like) and refinement of their existing molecular classification system was recently described by the Lund group.75

Fewer studies have described mRNA-based subtypes of NMIBC.5,9,67 Low-grade Ta tumors included in the Lund study were classified mainly as UroA, expressing high levels of urothelial differentiation markers, FGFR3 signature genes, early cell-cycle genes, and cell adhesion genes. Stage T1 and high-grade tumors, on the other hand, were shown to be very heterogenous being classified into UroA, genomically unstable, or infiltrated subtypes.67 The UROMOL study reported transcriptome profiling of 460 NMIBC of all stages and grades, including CIS, and 16 MIBC.9 Three molecular subtypes (class 1, class 2, and class 3) were defined. Class 1 contained mainly low-grade Ta tumors with the best prognosis and overlapped with the Lund UroA subtype. The other two classes were associated with tumors that had higher European Organisation for Research and Treatment of Cancer (EORTC) risk scores and contained more T1 and high-grade tumors, and CIS. Class 2 also contained the majority of MIBC samples and tumors from patients with progression events. These tumors expressed high levels of uroplakins, characteristic of luminal cells, but also expressed high levels of late cell-cycle genes, cancer stem cell markers, epithelial-mesenchymal transition markers, and progression and CIS signatures, leading to the suggestion that Class 2 may represent tumors of origin for luminal MIBC. Class 3 tumors not only had some luminal features (GATA3+, FGFR3 mutation) but also exhibited features characteristic of basal MIBC (CD44+, KRT5+, KRT14+, KRT15+, KRT20−, PPARG−). This class may represent a dormant tumor state as it is also characterized by the expression of many long noncoding RNAs, some of which have been found to be upregulated in oncogene-induced senescence.9 In the study of Hurst et al.,5 expression profiling of primary stage Ta tumors confirmed the overall luminal status of GS1 and GS2 samples and showed alignment mainly to the UroA subgroup.67

Терапевтические возможности и перспективы на будущее

The overall therapeutic efficacy of standard-of-care treatments for NMIBC and MIBC is relatively poor, and until recently, little progress had been made in identifying new therapeutic approaches. Patients with NMIBC have a very high recurrence rate (approximately 70%), and 10% to 15% progress to MIBC despite intravesical chemotherapy or BCG. Novel approaches to localized therapy are urgently needed to reduce their requirement for long term cystoscopic surveillance and its associated costs. Platinum-based chemotherapy has long been recognized as the standard-of-care treatment for metastatic bladder cancer, but only approximately 40% of patients respond, and relapse is common. Recently, checkpoint immunotherapy was approved for second-line therapy in such patients whose first-line chemotherapy has failed. Although impressive and durable responses have been reported, the overall response rates are modest, and robust predictive biomarkers are currently lacking. Over the past 5 years, molecular profiling studies employing whole-genome technologies have greatly enhanced our knowledge of the molecular landscape of bladder cancer. These have revealed clinically actionable alterations (activating mutations, amplifications, fusions) and discovered molecular signatures with predictive relevance.

High frequencies of alterations in CM genes point to chromatin modification as a key driver of bladder cancer. The reversible nature of epigenetic changes highlights a potential therapeutic opportunity in patients carrying such alterations. Molecules targeting epigenetic alterations are being developed and several are in clinical trials. These include DNA methyltransferase inhibitors (e.g., 5-aza-2′-deoxycytidine), histone deacetylase inhibitors (e.g., vorinostat, romidepsin, mocetinostat), and histone methyl transferase inhibitors (e.g., tazemetostat). A phase II trial (NCT02236195) aimed at evaluating the efficacy of mocetinosat (a histone deacetylase inhibitor) in patients with advanced UC with gene deletions or inactivating mutations in the histone acetyltransferase (HAT) family genes CREBBP and/or EP300 has recently been completed and results are awaited. However, a recent study showed that some mutations in CREBBP or EP300 did not abrogate HAT activity, highlighting the need for a full understanding of the functional impact of variants detected in such genes. This study also developed a gene expression signature associated with loss of HAT activity that could be used to stratify patients.76

Many canonical pathways that can be targeted with available drugs are altered in bladder cancer. As p53/cell- cycle pathway alterations occur in 89% of MIBC,11 targeting of components of the cell cycle may represent a therapeutic option. For example, palbociclib, a selective inhibitor of the cyclin-dependent kinases CDK4 and CDK6, has been licensed for use in some breast cancer patients and may be suitable as a second-line treatment in advanced bladder cancer with alterations in cell-cycle regulators such as RB1 and CDKN2A. Frequent mutations and copy number alterations in the ERBB genes family of receptor tyrosine kinases (ERBB2, ERBB3, and EGFR) are potential targets for treatment with tyrosine kinase inhibitors. For example, in platinum-refractory metastatic UC, alterations in ERBB2 and ERBB3 have been associated with response to treatment with the tyrosine kinase inhibitor afatinib.77 Similarly, FGFR3 is considered a good therapeutic target, and several inhibitors are currently in clinical trials. Good responses of FGFR3 mutant bladder cancers to the selective FGFR1-3 tyrosine kinase inhibitor BGJ398 have been reported.78 Alterations to the PI3K/AKT/mTOR pathway, such as PIK3CA mutation and activation of mTOR, may also represent actionable targets. Indeed, durable responses to everolimus have been reported in patients with MTOR or TSC1 mutations.79,80

The use of molecular markers holds promise for predicting the response of patients to chemotherapy prior to surgery. Alterations in DNA repair genes show some association with response to chemotherapy. For instance, mutations in ERCC2 are associated with improved outcomes in cisplatin-treated patients.44 Prospective screening of such genes in clinical trial samples should help identify robust markers of chemosensitivity. A clinical trial (Southwest Oncology Group [SWOG] 1314) has been testing the efficacy of a gene expression profiling–based algorithm (CoXEN) to predict a patient’s response to neoadjuvant chemotherapy (NAC).

The recently described mRNA subtypes also hold promise for prediction of response.11 Patients with basal- squamous MIBC have the worst prognosis and, when treated with cystectomy alone, have significantly shorter disease-specific and overall survival. However, when treated with cisplatin-based NAC, patients with basal-like tumors show better outcome than those with luminal or “p53-like” tumors.69 This superior response of basal tumors has been confirmed recently.81 Cisplatin-based combination therapies (e.g., etoposide-cisplatin) are currently in use for the treatment of neuroendocrine tumors in other tissues and may be appropriate for the newly identified bladder neuronal subtype.11,75

Recent data suggests that tumors with the luminal-papillary subtype identified in the 2017 TCGA study show poor response to cisplatin-based NAC.81 This subtype has better overall survival and is characterized by alterations in FGFR3 (activating mutations, FGFR3-TACC3 fusions, DNA amplifications), suggesting that treatment with FGFR inhibitors may be a valid approach.

The 2017 TCGA luminal-infiltrated subtype is predicted to be resistant to cisplatin-based chemotherapy as it shares features with the 2014 TCGA cluster II and the p53-like subtype identified by Choi et al.2,69 However, treatment with checkpoint inhibitors may be a valid approach in such patients as the cluster II subtype has previously been shown to respond well to treatment with atezolizumab.74 Basal-squamous tumors express high levels of immune markers, but basal subtypes III and IV are reported to show a reduced response to immune checkpoint inhibitors compared to cluster II, perhaps suggesting that other immunosuppressive factors exist in the basal subtype.74

Tumors with the mutation signature MSig1 may benefit from immunotherapy. These tumors have a high mutational burden, high APOBEC signature mutational load and high predicted neoantigen load. The 5-year survival probability in these patients was very high (75%) compared to the cluster with the lowest mutational burden (22%). It was hypothesized that this improved survival may be related to a greater host immune response.11 It will be important to examine such features in future and ongoing clinical trials, especially those using immune checkpoint therapy.

In NMIBC, particularly patients with stage Ta disease, the use of systemic therapies is unlikely to be suitable. However, for patients with high-risk NMIBC, particularly BCG refractory disease, some systemic therapies may be considered. These include immune checkpoint inhibitors and some targeted therapies (e.g., FGFR or ERBB family inhibitors). Ultimately, improved understanding of the molecular features of NMIBC may lead to reformulation of drugs for localized application and the development of novel approaches for delivery of therapeutics in this setting.

Genome-wide profiling has generated a wealth of data and is suggesting exciting potential therapeutic advances in the treatment of bladder cancer. However, application in the clinic will require careful validation by retrospective analysis of samples from previous studies and by carefully designed prospective studies. To increase predictive power, these should take into consideration not only alterations in the therapeutic target but also the overall molecular landscape in which the target functions. The ultimate goal will be the development of robust markers, suitable for routine application in the clinical setting.

Литература

  1. Knowles MA, Hurst CD. Molecular biology of bladder cancer: new insights into pathogenesis and clinical diversity. Nat Rev Cancer 2015;15(1):25–41.
  2. Cancer Genome  Atlas  Research    Comprehensive  molecular  characterization  of  urothelial  bladder carcinoma. Nature 2014;507(7492):315–322.
  3. Kandoth C, McLellan MD, Vandin F, et al. Mutational landscape and significance across 12 major cancer types. Nature 2013;502(7471):333–339.
  4. Lawrence MS, Stojanov P, Polak P, et al. Mutational heterogeneity in cancer and the search for new cancer- associated genes. Nature 2013;499(7457):214–218.
  5. Hurst CD, Alder O, Platt FM, et al. Genomic subtypes of non-invasive bladder cancer with distinct metabolic profile and female gender bias in KDM6A mutation frequency. Cancer Cell 2017;32(5):701–715.e7.
  6. Nordentoft I, Lamy P, Birkenkamp-Demtröder K, et al. Mutational context and diverse clonal development in early and late bladder cancer. Cell Rep 2014;7(5):1649–1663.
  7. Roberts SA, Lawrence MS, Klimczak LJ, et al. An APOBEC cytidine deaminase mutagenesis pattern is widespread in human cancers. Nat Genet 2013;45(9):970–976.
  8. Acar Ö, Özkurt E, Demir G, et al. Determining the origin of synchronous multifocal bladder cancer by exome sequencing. BMC Cancer 2015;15:871.
  9. Hedegaard J, Lamy P, Nordentoft I, et al. Comprehensive transcriptional analysis of early-stage urothelial carcinoma. Cancer Cell 2016;30(1):27–42.
  10. Lamy P, Nordentoft I, Birkenkamp-Demtroder K, et al. Paired exome analysis reveals clonal evolution and potential therapeutic targets in urothelial carcinoma. Cancer Res 2016;76(19):5894–5906.
  11. Robertson AG, Kim J, Al-Ahmadie H, et al. Comprehensive molecular characterization of muscle-invasive bladder cancer. Cell 2017;171(3):540–556.e25.
  12. Kim J, Mouw KW, Polak P, et al. Somatic ERCC2 mutations are associated with a distinct genomic signature in urothelial tumors. Nat Genet 2016;48(6):600–606.
  13. di Martino E, Tomlinson DC, Knowles MA. A decade of FGF receptor research in bladder cancer: past, present, and future challenges. Adv Urol 2012;2012:429213.
  14. Hernández S, López-Knowles E, Lloreta J, et al. FGFR3 and Tp53 mutations in T1G3 transitional bladder carcinomas: independent distribution and lack of association with prognosis. Clin Cancer Res 2005;11(15):5444– 5450.
  15. Hurst CD, Platt FM, Taylor CF, et al. Novel tumor subgroups of urothelial carcinoma of the bladder defined by integrated genomic analysis. Clin Cancer Res 2012;18(21):5865–5877.
  16. Guo G, Sun X, Chen C, et al. Whole-genome and whole-exome sequencing of bladder cancer identifies frequent alterations in genes involved in sister chromatid cohesion and segregation. Nat Genet 2013;45(12):1459–1463.
  17. Williams SV, Hurst CD, Knowles MA. Oncogenic FGFR3 gene fusions in bladder cancer. Hum Mol Genet 2013;22(4):795–803.
  18. Knowles MA, Platt FM, Ross RL, et al. Phosphatidylinositol 3-kinase (PI3K) pathway activation in bladder cancer. Cancer Metastasis Rev 2009;28(3–4):305–316.
  19. López-Knowles E, Hernández S, Malats N, et al. PIK3CA mutations are an early genetic alteration associated with FGFR3 mutations in superficial papillary bladder tumors. Cancer Res 2006;66(15):7401–7404.
  20. Jebar AH, Hurst CD, Tomlinson DC, et al. FGFR3 and Ras gene mutations are mutually exclusive genetic events in urothelial cell carcinoma. Oncogene 2005;24(33):5218–5225.
  21. Pietzak EJ, Bagrodia A, Cha EK, et al. Next-generation sequencing of nonmuscle invasive bladder cancer reveals potential biomarkers and rational therapeutic targets. Eur Urol 2017;72(6):952–959.
  22. Allory Y, Beukers W, Sagrera A, et al. Telomerase reverse transcriptase promoter mutations in bladder cancer: high frequency across stages, detection in urine, and lack of association with outcome. Eur Urol 2014;65(2):360– 366.
  23. Hurst CD, Platt FM, Knowles MA. Comprehensive mutation analysis of the TERT promoter in bladder cancer and detection of mutations in voided urine. Eur Urol 2014;65(2):367–369.
  24. Giedl J, Rogler A, Wild A, et al. TERT core promotor mutations in early-onset bladder cancer. J Cancer2016;7(8):915–920.
  25. Huang FW, Hodis E, Xu MJ, et al. Highly recurrent TERT promoter mutations in human melanoma. Science 2013;339(6122):957–959.
  26. Hosen I, Rachakonda PS, Heidenreich B, et al. Mutations in TERT promoter and FGFR3 and telomere length in bladder cancer. Int J Cancer 2015;137(7):1621–1629.
  27. López-Knowles E, Hernández S, Kogevinas M, et al. The p53 pathway and outcome among patients with T1G3 bladder tumors. Clin Cancer Res 2006;12(20 Pt 1):6029–6036.
  28. Malats N, Bustos A, Nascimento CM, et al. P53 as a prognostic marker for bladder cancer: a meta-analysis and review. Lancet Oncol 2005;6(9):678–686.
  29. Schmitz-Dräger BJ, Goebell PJ, Ebert T, et al. p53 immunohistochemistry as a prognostic marker in bladder cancer. Playground for urology scientists? Eur Urol 2000;38(6):691–700.
  30. George B, Datar RH, Wu L, et al. p53 gene and protein status: the role of p53 alterations in predicting outcome in patients with bladder cancer. J Clin Oncol 2007;25(34):5352–5358.
  31. Aine M, Eriksson P, Liedberg F, et al. Biological determinants of bladder cancer gene expression subtypes. Sci Rep 2015;5:10957.
  32. Rebouissou S, Hérault A, Letouzé E, et al. CDKN2A homozygous deletion is associated with muscle invasion in FGFR3-mutated urothelial bladder carcinoma. J Pathol 2012;227(3):315–324.
  33. Hurst CD, Tomlinson DC, Williams SV, et al. Inactivation of the Rb pathway and overexpression of both isoforms of E2F3 are obligate events in bladder tumours with 6p22 amplification. Oncogene 2008;27(19):2716–2727.
  34. Shariat SF, Ashfaq R, Sagalowsky AI, et al. Predictive value of cell cycle biomarkers in nonmuscle invasive bladder transitional cell carcinoma. J Urol 2007;177(2):481–487.
  35. Gui Y, Guo G, Huang Y, et al. Frequent mutations of chromatin remodeling genes in transitional cell carcinoma of the bladder. Nat Genet 2011;43(9):875–878.
  36. Ler LD, Ghosh S, Chai X, et al. Loss of tumor suppressor KDM6A amplifies PRC2-regulated transcriptional repression in  bladder  cancer  and  can  be  targeted  through  inhibition  of    Sci  Transl  Med 2017;9(378):eaai8312.
  37. Dancik GM, Owens CR, Iczkowski KA, et al. A cell of origin gene signature indicates human bladder cancer has distinct cellular progenitors. Stem Cells 2014;32(4):974–982.
  38. Balbás-Martínez C, Sagrera A, Carrillo-de-Santa-Pau E, et al. Recurrent inactivation of STAG2 in bladder cancer is not associated with aneuploidy. Nat Genet 2013;45(12):1464–1469.
  39. Bitler BG, Aird KM, Garipov A, et al. Synthetic lethality by targeting EZH2 methyltransferase activity in ARID1A-mutated cancers. Nat Med 2015;21(3):231–238.
  40. Solomon DA, Kim JS, Bondaruk J, et al. Frequent truncating mutations of STAG2 in bladder cancer. Nat Genet 2013;45(12):1428–1430.
  41. Taylor CF, Platt FM, Hurst CD, et al. Frequent inactivating mutations of STAG2 in bladder cancer are associated with low tumour grade and stage and inversely related to chromosomal copy number changes. Hum Mol Genet 2014;23(8):1964–1974.
  42. Solomon DA, Kim T, Diaz-Martinez LA, et al. Mutational inactivation of STAG2 causes aneuploidy in human cancer. Science 2011;333(6045):1039–1043.
  43. De Koninck M, Losada A. Cohesin mutations in cancer. Cold Spring Harb Perspect Med 2016;6(12):a026476.
  44. Van Allen EM, Mouw KW, Kim P, et al. Somatic ERCC2 mutations correlate with cisplatin sensitivity in muscle- invasive urothelial carcinoma. Cancer Discov 2014;4(10):1140–1153.
  45. Mullane SA, Werner L, Guancial EA, et al. Expression levels of DNA damage repair proteins are associated with overall survival in platinum-treated advanced urothelial carcinoma. Clin Genitourin Cancer 2016;14(4):352–359.
  46. Choudhury A, Nelson LD, Teo MT, et al. MRE11 expression is predictive of cause-specific survival following radical radiotherapy for muscle-invasive bladder cancer. Cancer Res 2010;70(18):7017–7026.
  47. Knowles MA, Habuchi T, Kennedy W, et al. Mutation spectrum of the 9q34 tuberous sclerosis gene TSC1 in transitional cell carcinoma of the bladder. Cancer Res 2003;63(22):7652–7656.
  48. Platt FM, Hurst CD, Taylor CF, et al. Spectrum of phosphatidylinositol 3-kinase pathway gene alterations in bladder cancer. Clin Cancer Res 2009;15(19):6008–6017.
  49. Nord H, Segersten U, Sandgren J, et al. Focal amplifications are associated with high grade and recurrences in stage Ta bladder carcinoma. Int J Cancer 2010;126(6):1390–1402.
  50. Blaveri E, Brewer JL, Roydasgupta R, et al. Bladder cancer stage and outcome by array-based comparative genomic hybridization. Clin Cancer Res 2005;11(19 Pt 1):7012–7022.
  51. Dewhurst SM, McGranahan N, Burrell RA, et al. Tolerance of whole-genome doubling propagates chromosomal instability and accelerates cancer genome evolution. Cancer Discov 2014;4(2):175–185.
  52. Zack TI, Schumacher SE, Carter SL, et al. Pan-cancer patterns of somatic copy number alteration. Nat Genet 2013;45(10):1134–1140.
  53. Nakanishi Y, Akiyama N, Tsukaguchi T, et al. Mechanism of oncogenic signal activation by the novel fusion kinase FGFR3-BAIAP2L1. Mol Cancer Ther 2015;14(3):704–712.
  54. Goldstein JT, Berger AC, Shih J, et al. Genomic activation of PPARG reveals a candidate therapeutic axis in bladder cancer. Cancer Res 2017;77(24):6987–6998.
  55. Korpal M, Puyang X, Jeremy Wu Z, et al. Evasion of immunosurveillance by genomic alterations of PPARγ/RXRα in bladder cancer. Nat Commun 2017;8(1):103.
  56. Hafner C, Knuechel R, Zanardo L, et al. Evidence for oligoclonality and tumor spread by intraluminal seeding in multifocal urothelial carcinomas of the upper and lower urinary tract. Oncogene 2001;20(35):4910–4915.
  57. Sidransky D, Frost P, Von Eschenbach A, et al. Clonal origin of bladder cancer. N Engl J Med 1992;326(11):737– 740.
  58. Warrick JI, Hovelson DH, Amin A, et al. Tumor evolution and progression in multifocal and paired non- invasive/invasive urothelial carcinoma. Virchows Arch 2015;466(3):297–311.
  59. Cazier JB, Rao SR, McLean CM, et al. Whole-genome sequencing of bladder cancers reveals somatic CDKN1A mutations and clinicopathological associations with mutation burden. Nat Commun 2014;5:3756.
  60. Thomsen MB, Nordentoft I, Lamy P, et al. Spatial and temporal clonal evolution during development of metastatic urothelial carcinoma. Mol Oncol 2016;10(9):1450–1460.
  61. Thomsen MBH, Nordentoft I, Lamy P, et al. Comprehensive multiregional analysis of molecular heterogeneity in bladder cancer. Sci Rep 2017;7(1):11702.
  62. Majewski T, Lee S, Jeong J, et al. Understanding the development of human bladder cancer by using a whole- organ genomic mapping strategy. Lab Invest 2008;88(7):694–721.
  63. Stoehr R, Zietz S, Burger M, et al. Deletions of chromosomes 9 and 8p in histologically normal urothelium of patients with bladder cancer. Eur Urol 2005;47(1):58–63.
  64. Faltas BM, Prandi D, Tagawa ST, et al. Clonal evolution of chemotherapyresistant urothelial carcinoma. Nat Genet 2016;48(12):1490–1499.
  65. Lindgren D, Frigyesi A, Gudjonsson S, et al. Combined gene expression and genomic profiling define two intrinsic molecular subtypes of urothelial carcinoma and gene signatures for molecular grading and outcome. Cancer Res 2010;70(9):3463–3472.
  66. Lindgren D, Sjödahl G, Lauss M, et al. Integrated genomic and gene expression profiling identifies two major genomic circuits in urothelial carcinoma. PLoS One 2012;7(6):e38863.
  67. Sjödahl G, Lauss M, Lövgren K, et al. A molecular taxonomy for urothelial carcinoma. Clin Cancer Res 2012;18(12):3377–3386.
  68. Sjödahl G, Lövgren K, Lauss M, et al. Toward a molecular pathologic classification of urothelial carcinoma. Am J Pathol 2013;183(3):681–691.
  69. Choi W, Porten S, Kim S, et al. Identification of distinct basal and luminal subtypes of muscle-invasive bladder cancer with different sensitivities to frontline chemotherapy. Cancer Cell 2014;25(2):152–165.
  70. Damrauer JS, Hoadley KA, Chism DD, et al. Intrinsic subtypes of high-grade bladder cancer reflect the hallmarks of breast cancer biology. Proc Natl Acad Sci U S A 2014;111(8):3110–3115.
  71. Perou CM, Sørlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature 2000;406(6797):747– 752.
  72. Prat A, Karginova O, Parker JS, et al. Characterization of cell lines derived from breast cancers and normal mammary tissues for the study of the intrinsic molecular subtypes. Breast Cancer Res Treat 2013;142(2):237–255.
  73. Dyrskjøt L, Kruhøffer M, Thykjaer T, et al. Gene expression in the urinary bladder: a common carcinoma in situ gene expression signature exists disregarding histopathological classification. Cancer Res 2004;64(11):4040–4048.
  74. Rosenberg JE, Hoffman-Censits J, Powles T, et al. Atezolizumab in patients with locally advanced and metastatic urothelial carcinoma who have progressed following treatment with platinum-based chemotherapy: a single-arm, multicentre, phase 2 trial. Lancet 2016;387(10031):1909–1920.
  75. Sjödahl G, Eriksson P, Liedberg F, et al. Molecular classification of urothelial carcinoma: global mRNA classification versus tumour-cell phenotype classification. J Pathol 2017;242(1):113–125.
  76. Duex JE, Swain KE, Dancik GM, et al. Functional impact of chromatin remodeling gene mutations and predictive signature for therapeutic response in bladder cancer. Mol Cancer Res 2018;16(1):69–77.
  77. Choudhury NJ, Campanile A, Antic T, et al. Afatinib activity in platinumrefractory metastatic urothelial carcinoma in patients with ERBB alterations. J Clin Oncol 2016;34(18):2165–2171.
  78. Nogova L, Sequist LV, Perez Garcia JM, et al. Evaluation of BGJ398, a fibroblast growth factor receptor 1-3 kinase inhibitor, in patients with advanced solid tumors harboring genetic alterations in fibroblast growth factor receptors: results of a global phase I, dose-escalation and dose-expansion study. J Clin Oncol 2017;35(2):157–165.
  79. Wagle N, Grabiner BC, Van Allen EM, et al. Activating mTOR mutations in a patient with an extraordinary response on a phase I trial of everolimus and pazopanib. Cancer Discov 2014;4(5):546–553.
  80. Iyer G, Hanrahan AJ, Milowsky MI, et al. Genome sequencing identifies a basis for everolimus sensitivity. Science 2012;338(6104):221.
  81. Seiler R, Ashab HAD, Erho N, et al. Impact of molecular subtypes in muscleinvasive bladder cancer on predicting response and survival after neoadjuvant chemotherapy. Eur Urol 2017;72(4):544–554.