EURURO Vol. 72 No. 3

355 476

1. Introduction
Recent whole genome mRNA expression profiling studies
revealed that bladder cancers can be grouped into molecu-
lar subtypes, some of which share clinical properties and
gene expression patterns with the intrinsic subtypes of
breast cancer and the molecular subtypes found in other
solid tumors. The molecular subtypes in other solid tumors
are enriched with specific mutations and copy number
aberrations (CNAs) that are thought to underlie their
distinct progression patterns, and biological and clinical
properties.
2. Evidence acquisition
We used the complete The Cancer Genome Atlas (TCGA)
RNA-seq dataset and three different published classifiers
developed by our groups to assign TCGA’s bladder cancers
to molecular subtypes, and examined the prevalence of the
most common DNA alterations within them (Supplemen-
tary material). We interpreted the results against the
background of what was known from the published
literature about the prevalence of these alterations in
nonmuscle-invasive and muscle-invasive bladder cancers.
3. Evidence synthesis
3.1. Clinical issues in bladder cancer
Clinical experience and emerging genomic data support the
idea that bladder cancers progress along two largely
nonoverlapping tracks (‘‘papillary’’ and ‘‘nonpapillary’’)
that pose distinct challenges for clinical management
. Most nonmuscle-invasive bladder cancers (NMIBCs)
belong to the papillary pathway and are characterized by
the presence of activating type-3 receptor for fibroblast
growth factor (FGFR3) mutations, downstream Ras pathway
activation, wild-type TP53, and stable genomes . Clini-
cally, papillary NMIBCs are rarely lethal but recur almost
always, necessitating that patients receive lifelong surveil-
lance; the repeated surgical procedures required to deal
with recurrences cause significant anxiety, discomfort, and
morbidity, making bladder cancer the most expensive
tumor on a per patient basis. A significant proportion of
cases (15–20%) of NMIBCs progress to become muscle
invasive . However, currently no reliable tools are
available to identify them before they become life
threatening. The nonpapillary pathway is characterized
by loss-of-function mutations and CNAs involvingTP53 and
RB1 and genomic instability . It gives rise to aggressive,
muscle-invasive bladder cancers (MIBCs), representing
approximately 20–25% of all bladder cancers and causing
death in approximately half of affected patients. Carcinoma
in situ (CIS) is generally considered to be the precursor
lesion for nonpapillary MIBCs , but comprehensive
genomic data for CIS are not yet available, so this
assumption awaits direct experimental validation. Patients
with either high-grade papillary nonmuscle-invasive dis-
ease or CIS are currently treated with the same adjuvant
therapy (intravesical Bacillus Calmette–Guerin [BCG] im-
munotherapy), but it is by no means clear that BCG
produces comparable benefit in CIS and high-grade papil-
lary tumors . Many high-grade papillary tumors
ultimately become BCG unresponsive, so clinicians are
then faced with the dilemma of whether to continue using a
bladder-sparing regimen or to employ definitive surgery.
The latter is certainly too aggressive for those patients
whose tumors could be controlled by local therapy, but
again there are no reliable tools to distinguish the tumors
that have the potential to metastasize from those that do
not. Muscle-invasive disease is managed with definitive
local therapy (chemoradiation) or surgery (cystectomy)
with or without perioperative systemic cisplatin-based
chemotherapy to treat subclinical metastatic disease, but it
is still not possible to distinguish the patients who warrant
chemotherapy from those who will not benefit from it. It
would also be tremendously useful to have biomarkers that
would enable patients and their physicians to choose
between bladder-sparing regimens such as chemoradiation
and cystectomy. Overall, it is hoped that by understanding
the molecular mechanisms that give rise to papillary and
nonpapillary bladder cancers, it will be possible to develop
methods to inform clinical decision making at every step of
disease progression and management.
3.2. Intrinsic subtypes of cancer
The widespread use of genomics to investigate cancer
heterogeneity is transforming our understanding of cancer
biology. A pioneering study in leukemia demonstrated that
mRNA expression profiling could be used to distinguish ALL
from AML with a high degree of accuracy , and a
subsequent study used gene expression profiling to identify
two previously unrecognized molecular subtypes of diffuse
large B-cell lymphoma Importantly, patients whose
tumors belonged to one of the subtypes (‘‘germinal center-
like DLBCL’’) had better clinical outcomes than patients with
the other (‘‘activated B-like DLBCL’’) . Parallel studies in
breast cancer revealed that they could also be grouped into
‘‘intrinsic subtypes’’ that had very different biological
properties and behaved clinically as distinct disease entities
. Patients with basal-like or HER2-enriched breast
tumors had poor clinical outcomes in the absence of
systemic therapy, but many of them benefited greatly from
neoadjuvant chemotherapy (NAC) [7_TD$DIFF] . Patients with
HER2-enriched tumors also obtained significant clinical
benefit from ERBB2 antagonists [8_TD$DIFF] . In the absence of
perioperative chemotherapy, women with luminal tumors
had better prognoses [9_TD$DIFF] and, when given perioperative
chemotherapy, most patients also obtained little to no
benefit [10_TD$DIFF] . Rather, they obtained major chemopreven-
tive clinical benefit from adjuvant therapy with selective
estrogen receptor modulators (SERMs), which reduced
disease recurrence by about 50% [9_TD$DIFF] . In contrast, SERMs
produced no benefit in patients with basal-like or HER2-
enriched tumors [9_TD$DIFF] . Subsequent studies identified molec-
ular subtypes in head and neck squamous cell carcinomas
(SCCs) [11_TD$DIFF] , glioblastomas [12_TD$DIFF] , and pancreatic cancers [13_TD$DIFF] ,
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 3 5 4 – 3 6 5 355