76 96
1.
Introduction
Although
the
natural
history
of
urothelial
carcinoma
of
the
bladder
(UCB)
from
radical
cystectomy
(RC)
to
disease
recurrence
(DR)
has
been
intensively
investigated
[1–7],
that
of
patients
who
have
experienced
DR
after
RC
for
UCB
still
remains
poorly
understood.
Indeed,
it
has
been
previously
reported
that
poor
patient
Karnofsky
perfor-
mance
status
(KPS) and
the presence of visceral metastases
(VMs) (ie,
the Bajorin criteria) are associated with decreased
survival
in
patients
with metastatic
urothelial
carcinoma
(UC)
treated with systematic cisplatin-based chemotherapy
and
could
be
used
to
stratify
patients
into
risk
groups
[8] .Recent
studies have
reported
that
a
shorter
time
to DR
after RC
is associated with unfavorable outcomes
[9,10].
Im-
proved understanding of the natural history of such patients
and accurate prognostication after DR
could help
in patient
counseling
and
in
the
design
of
clinical
trials.
The
aim
of
the
study was
therefore
threefold.
First, we
assessed
the
prognostic
value
of
the
Bajorin
criteria
in
a
large
multi-institutional
cohort
of
patients
who
experi-
enced DR after RC
for UCB. Second, we evaluated additional
clinical, pathologic, and/or biologic
factors at
the
time of DR
in
these
patients
for
an
association
with
cancer-specific
outcomes. Third, we
aimed
to
create
a multivariable model
based on
the
identified
variables
that were
associated with
cancer-specific
outcomes
and
to
compare
discrimination
versus
the
Bajorin
risk
grouping.
2.
Materials
and methods
2.1.
Patient
selection
This
study
was
approved
by
institutional
review
boards,
with
all
participating
sites
providing
the
necessary
data-sharing
agreements
before
initiation.
A
total
of
17
international
centers
provided
data.
The
database was
closed
in October
2012.
Our multi-institutional
cohort
consisted
of 967 patients who
experi-
enced DR after RC
for UCB between 1979 and 2012. Of these, we
identified
372 patients with complete data on all variables for analysis. Patientswere
excluded
even when missing
only
one
variable. We
consider
this
step
important
to
avoid
selection
bias. None
of
the
patients
had
evidence
of
distant metastases
at
the
time
of RC. All
patients underwent
an RC with
bilateral pelvic
lymph node dissection and urinary diversion as described
elsewhere
[4].
No
patient
received
neoadjuvant
chemotherapy
or
pre-
and/or postoperative
radiation
therapy
to
the bladder.
2.2.
Pathologic
evaluation
Tumors
were
staged
according
to
the
American
Joint
Committee
on
Cancer
Union
Internationale
Contre
le
Cancer
TNM
classification,
7th
edition
[11].
Tumor
grade was
assessed
according
to
the
1998 World
Health
Organization/International
Society
of
Urologic
Pathology
con-
sensus
classification
[12].
2.3.
Follow-up
regimen
Follow-up was
performed
according
to
institutional
protocols.
Postop-
eratively, patients were seen at
least every 3–4 mo
in
the first year, every
6 mo
in
the
second year,
and
annually
thereafter. Diagnostic
imaging of
the
upper
tract
(eg,
ultrasonography
and/or
intravenous
pyelography,
computed
tomography
of
the
abdomen/pelvis
with
intravenous
contrast)
and
chest
radiography
were
performed
annually
and
when
indicated
clinically. DR was
defined
as
a
tumor
relapse
in
the
operative
field,
regional
lymph nodes, and/or distant metastases. DR was managed
at
the patient’s
and
treating physician’s discretion
(ie,
administration of
salvage chemotherapy). Perioperative mortality
(ie, death within 30 d of
surgery)
was
censored
at
time
of
death
for
bladder
cancer–specific
survival
analyses.
2.4.
Statistical
analysis
For
statistical
analysis, we
only used patients with
complete data
on
all
variables
(
n
= 372). Our
first
aim was
to
describe
the
association
of
the
Bajorin
criteria with
cancer-specific mortality
(CSM)
[8].
Patients were
thus
categorized
into
three
risk
groups
determined
by
KPS
and/or
the
presence
of VMs
(no
risk
factors
[RFs]:
KPS
80%
and
no VMs;
one
RF:
KPS
<
80%
or
presence
of
VMs;
two
RFs
if
both).
Kaplan-Meier
curves
were
generated
(log
rank).
Our
next
aim was
to
identify
additional
characteristics
associated
with
cancer-specific
survival
(CSS)
in
these
patients.
There were
four
characteristics
of
interest
( Table 1 ):
clinical
characteristics,
character-
istics
at
RC,
biologic
characteristics
at
DR,
and
treatment-related
Design,
setting, and participants:
We
identified 967 patients with UCB who underwent
RC at 17 centers between 1979 and 2012 and experienced DR. Of
these, 372 patients had
complete
data we
used
for
analysis.
Outcomes measurements
and
statistical
analysis:
Univariable
Cox
regressions
analysis
was performed. We used a
forward stepwise selection process
for our final multivariable
model.
Results
and
limitations:
Within
a median
follow-up
of
18 mo,
266
patients
died
of
disease.
Cancer-specific
survival
at
1
yr was
79%,
76%,
and
47%
for
patients with
no
(
n
= 105),
one
(
n
= 180),
and
two
(
n
= 87)
risk
factors
(
p
<
0.001;
c-index:
0.604).
On
multivariable
analyses, we
found
that
KPS
<
80%,
higher
American
Society
of
Anesthe-
siologists
score, anemia,
leukocytosis, and
shorter
time
to DR
(all
p
values
<
0.034) were
independently
associated with
increased
CSM.
The
combination
of
time
to DR
and KPS
resulted
in
improved
discrimination
(c-index:
0.694).
Conclusions:
We
confirmed
the
prognostic
value
of
KPS
and
VMs
in
patients with
DR
following RC
for UCB. We also
found several other clinical variables
to be associated with
worse CSM. We developed a model
for predicting survival after DR
inclusive of
time
to DR
and
KPS
assessed
at DR.
If
validated,
this model
could
help
clinical
trial
design.
Patient
summary:
We
developed
a model
to
predict
survival
following
disease
recur-
rence
after
radical
cystectomy
for urothelial
carcinoma of
the bladder, based on
time
to
disease
recurrence
and
Karnofsky
performance
status.
#
2015 European Association
of Urology. Published by Elsevier B.V. All
rights
reserved.
Bladder
cancer
Transitional
cell
carcinoma
Metastasis
Disease
recurrence
Prognosis
Survival
Model
E U R O P E A N
U R O L O G Y
F O C U S
1
( 2 0 1 5
)
7 5 – 8 1
76