EURURO Vol. 71 No. 5

832 844

Re: Prediction of Lymph Node Metastasis in Patients
with Bladder Cancer Using Whole Transcriptome Gene
Expression Signatures
Seiler R, Lam LL, Erho N, et al
J Urol 2016;196:1036–41
Expert’s summary:
Seiler and colleagues recently published a 51-gene signature
(KNN51) comprising coding and noncoding transcripts that
allows prediction of lymph node (LN) metastases in patients
with muscle-invasive bladder cancer (MIBC). To generate
their new classifier, they performed whole-transcriptome
analysis on RNA isolated from cystectomy specimens from
a 133-patient discovery cohort and validated their findings
in a validation cohort of 66 patients. Interestingly, none of the
51 genes included in KNN51 overlapped with the 20 genes
from a previously published expression signature (20-gene
lymph node signature [LN20]) [1]. Moreover, KNN51 greatly
outperformed LN20, with an area under the receiver operating
characteristic curve of 0.82 versus 0.46. Every 10% increase in
the KNN51 score was associated with a significant increase in
the presence of LN metastases (odds ratio 2.65, p < 0.001).
Thus, KNN51 helps to improve clinical risk stratification for
guiding optimal therapy (eg, neoadjuvant chemotherapy
[NAC]) for patients with MIBC.
Expert’s comments:
Although level 1 evidence demonstrates that NAC improves
survival for patients with MIBC (6% difference in 10-yr overall
survival) [2], fewer than 20% of eligible patients receive the
recommended care [3]. The rate of NAC treatment is probably
so low because patients with organ-confined bladder cancer
( pT2), who comprise up to 50% of cases in contemporary
series, have excellent survival following cystectomy alone
(80% cure rate). As a consequence, only 20–25% of unselected
patients benefit fromNAC. This limited survival benefit of NAC
must be weighed against NAC-related toxicity and delay of
cystectomy. Because current staging is inadequate (25% of cN0
patients harbor LN metastases [4]; up to 50% of patients are
understaged and potentially undertreated [5]), there is a
clinical need for tools to guide treatment decisions regarding
NAC in patients with MIBC.
In KNN51, Seiler et al present such a tool. It has an
advantage over, for example, LN20 in that both coding and
noncoding transcripts are analyzed. Although the biological
significance of noncoding genes is limited, they are known
to play an important role in gene regulation, which is
undoubtedly essential in cancer formation and metastasis.
Thus, it is not surprising that most genes included in KNN51
are involved in cellular movement, tumor growth, and
metastasis formation.
The big advantage of this study is the cohort of cystectomy
patients used to develop and validate the classifier: all
patients underwent uniform surgery including extended
pelvic lymphadenectomy (ePLND). This stands in contrast to
the practice of other groups that have tried to develop their
own classifier. Because they did not performuniformPLND in
most of their cohorts, the LN counts for these groups have
rangedwidely from2 to >40 LNs. Owing to inadequate PLND,
there is a risk that the LN-negative cohort includes some LN-
positive patients, negating any possible difference in gene
expression between the two cohorts.
Some limitations have to be mentioned. First, the
platform used by Seiler and colleagues was whole-
transcriptome expression analysis. This is a laborious and
costly technique that is not widely available for standard
diagnostic procedures. A more widely used and easily
available technique needs to be validated, such as qualita-
tive real-time polymerase chain reaction. Second, the
samples used originated from radical cystectomy speci-
mens. However, clinical decision-making (eg, NAC or
upfront cystectomy) has to rely on transurethral resection
of bladder (TURB) material. Therefore, KNN51 must first be
validated on TURB samples before it is clinically applicable.
Conflicts of interest: The author has nothing to disclose.
References
[1]
[2]
[3]
.
[4]
[5]
Beat Roth *
Department of Urology, University of Bern, Bern, Switzerland
*Department of Urology, University Hospital Bern,
CH-3010 Bern, Switzerland.
E-mail address:
# 2017 European Association of Urology.
Published by Elsevier B.V. All rights reserved.
E U R O P E A N U R O L O G Y 7 1 ( 2 0 1 7 ) 8 3 1 – 8 3 6 832