an advanced stage and grade
[[30_TD$DIFF]
33] ,and in more recent
studies,
MYC
amplification was observed in about 13% of
MIBCs
[[5_TD$DIFF]
18] .Inactivating mutations in genes encoding DNA repair
proteins were also relatively common in MIBCs
[[31_TD$DIFF]
17,34]. The
most prevalent were inactivating mutations in
ERCC2
(12%
of tumors)
[[5_TD$DIFF]
18], which were linked to sensitivity to
neoadjuvant cisplatin-based combination chemotherapy
[[32_TD$DIFF]
35]. Inactivating mutations in several other DNA repair
proteins were also linked to cisplatin sensitivity
[[33_TD$DIFF]
36].
Inactivating mutations in
STAG2
, a component of the
cohesin complex that functions in chromosome segrega-
tion, were common in NMIBCs and MIBCs
[[34_TD$DIFF]
34,37–[35_TD$DIFF]
40], but
their biological significance remains unclear; the canonical
role of the cohesin complex would suggest that inactivation
of
STAG2
should produce genomic instability, but the
significant enrichment of
STAG2
mutations in low-grade
tumors that largely lack aneuploidy
[[35_TD$DIFF]
40]argues against this
being their most relevant effect in bladder cancers
[[34_TD$DIFF]
34,37–[35_TD$DIFF]
40]. Alternative mechanisms include alterations in high-
order chromatin organization and gene expression.
Activating mutations in the telomerase (
TERT1
) promot-
er were common in both NMIBCs and MIBCs
[[36_TD$DIFF]
41–[37_TD$DIFF]
48],
making them attractive biomarkers for early detection of
recurrence
[[38_TD$DIFF]
44,45]and potentially as therapeutic targets
across the course of disease progression.
Bladder cancers often contained DNA alterations involv-
ing oncogenes or tumor suppressor genes that regulate
activation of the Ras–MEK–ERK and PI3 kinase–AKT–mTOR
pathways. These pathways control progression through the
RB1-dependent G1-S cell cycle restriction point, anabolic
metabolism, and cell survival. Activating mutations in
FGFR3
were detected in up to 80% of NMIBCs and
approximately 15–20% of MIBCs, consistent with earlier
studies
[[39_TD$DIFF]
1,49,50]. Preclinical studies demonstrated that
these
FGFR3
mutations, which cause constitutive receptor
activation, functioned to promote proliferation via down-
stream activation of the ERKs
[[40_TD$DIFF]
51,52]. Papillary and
nonpapillary cancers contained similar frequencies of
activating
RAS
mutations (5–10%)
[[41_TD$DIFF]
51] ,which also function
to promote downstream ERK activation.
RAS
and
FGFR3
mutations occurred in a mutually exclusive fashion
[[5_TD$DIFF]
18] ,so
together they probably accounted for enhanced ERK
activation in almost 90% of NMIBCs. Although activating
FGFR3
mutations were less common in MIBCs, some MIBCs
contained activating
FGFR3
fusions
[[42_TD$DIFF]
53]. They also con-
tained activating mutations, fusions, or amplification of
genes encoding members of the epidermal growth factor
receptor (EGFR) family
[[15_TD$DIFF]
17], including the
EGFR
itself (about
10% of tumors),
ERBB2
(about 10% of tumors),
ERBB3
(about
10% tumors), and
ERBB4
(about 6% of tumors). Other tumors
contained inactivating mutations in the RAS inhibitor,
NF1
(over 10%)
[[28_TD$DIFF]
1,18]; so, together, these alterations probably
promoted RAS pathway activation in over 50% of MIBCs.
With respect to the PI3 kinase/AKT pathway, activating
PIK3CA
mutations—predominantly in the region coding for
the helical domain, probably caused by APOBEC3B-mediat-
ed mutagenesis
[[20_TD$DIFF]
24]—were common in both NMIBCs and
MIBCs. Amplification of
AKT3
or inactivating mutations in
various negative regulators of the PI3 kinase/AKT/mTOR
pathway (including the lipid phosphatase
PTEN
) were also
observed, leading to predicted pathway activation in almost
75% of MIBCs
[[28_TD$DIFF]
1,18].
Mutations and/or amplification of transcription factors
implicated in urothelial terminal differentiation were found
in MIBCs
[[5_TD$DIFF]
18] .Amplifications of peroxisome proliferator
activator receptor-gamma (
PPARG
),
GATA3
, and
SOX4
were
frequently detected, occurring in about 10–15% of tumors
[[5_TD$DIFF]
18]. Mutations in
ELF3
,
RXRA
, and
KLF5
were also relatively
common, occurring in 5–10% of tumors
[[5_TD$DIFF]
18]. Mutations in
FOXA1
were observed in about 5% of tumors, and deletion of
FOXQ1
occurred in about 10% of tumors
[[5_TD$DIFF]
18]. Inactivating
mutations in
NOTCH1
and
NOTCH2
have also been reported
in MIBCs
[[43_TD$DIFF]
54], and preclinical studies in mouse models
suggested that they promoted tumor progression by
facilitating epithelial-to-mesenchymal transition (EMT)
[[44_TD$DIFF]
55]. Finally, mutations in the ubiquitin ligase and NOTCH
pathway regulator,
FBXW7
, occurred in about 7% of MIBCs
[[5_TD$DIFF]
18]. Although inactivating mutations in other develop-
mental pathways were less common, preclinical studies
have suggested that activation of the Wnt/
b
-catenin
pathway and downregulation of the sonic hedgehog
pathway also contribute to bladder cancer progression
[[45_TD$DIFF]
56,57].
3.4.
Molecular subtypes of bladder cancer
The identification and validation of molecular subtypes in
other malignancies provided the impetus to use transcrip-
tome profiling to search for molecular subtypes in bladder
cancers. The initial results established that unsupervised
analyses of gene expression could distinguish most NMIBCs
from most MIBCs
[[46_TD$DIFF]
58–[47_TD$DIFF]
60] .Furthermore, early studies of
NMIBCs identified gene expression signatures associated
with disease aggressiveness using unsupervised analyses
[[48_TD$DIFF]
61–[49_TD$DIFF]
63]. These studies showed the first indications of the
presence of major molecular subtypes in bladder cancer
[[50_TD$DIFF]
19,64]. One of our groups (M.H., Lund University, Lund,
Sweden) extended these findings by implicating differences
in
[51_TD$DIFF]
TP53
mutation frequencies and/or genomic instability to
the formation of these two major gene expression subtypes
[[52_TD$DIFF]
28]. Subsequently, they used a large cohort of NMIBCs and
MIBCs (
n
= 308) to identify additional subtypes within the
two major clusters
[[53_TD$DIFF]
65]. The Lund classification revealed
that bladder cancers could be segregated into at least five
molecular subtypes, termed urobasal A (uroA), urobasal B
(uroB), genomically unstable (GU), infiltrated, and SCC like
(SCCL)
[[53_TD$DIFF]
65]. The uroA and uroB tumors were characterized
by stratified expression of differentiation-associated bio-
markers reminiscent of what is observed in the normal
urothelium, whereas differentiation-associated biomarkers
displayed abnormal expression in the GU and SCCL tumors.
The SCCL subtype was characterized by expression of
squamous keratins (KRT5, KRT6, and KRT14) and keratini-
zation-associated genes
[[54_TD$DIFF]
66], and the SCCL and uroB tumors
were both enriched with various degrees of squamous
differentiation markers
[[55_TD$DIFF]
65,66]. As its name implies, the
infiltrated subtype was characterized by the expression of
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
357




