EURURO Vol. 72 No. 4

556 656

apropos . Simon and colleagues acknowledge the
value of retrospective biomarker studies given the multi-
tude of challenges with prospective studies. However,
cohorts ‘‘of convenience’’ are assigned a category D ranking
given the multiple biases introduced by:
(1) Retrospective selection of patients and specimens;
(2) Lack of stipulation of treatment or follow-up;
(3) Data collection via retrospective chart review;
(4) Specimens collected without standard operating pro-
cedures; and
(5) A lack of power to address the biomarker question.
Given these considerations and the major concerns
associated with cross-cohort comparisons, the finding that
MIBC subtypes are predictive of NAC benefit must still be
considered hypothesis-generating. The relationship be-
tween the MIBC subtypes, pathologic complete response,
and survival reported by Seiler et al is also difficult to
reconcile and further complicates interpretation of the data.
That said, the fact that independent groups are reaching
similar conclusions is intriguing and suggests that clinical
validity may indeed be established with further study
We are faced with a self-perpetuating problem
in advancing perioperative systemic therapy approaches
in MIBC. Randomized trials comparing perioperative
chemotherapy with local therapy alone have been small
and/or closed early because of poor accrual, and biospeci-
mens are not available from the completed trials. The lack of
such a resource suitable for ‘‘prospective-retrospective’’
biomarker studies markedly hampers efforts to clinically
validate putative predictive biomarkers and may necessi-
tate proceeding directly with large and costly prospective
studies to definitively establish both clinical validity and
utility.
Establishing the clinical utility of a predictive biomarker
for NAC benefit inMIBC is no easy task. Because current data
support the use of NAC in all patients with MIBC, trials to
establish the clinical utility of biomarker-based decision-
making would not be expected to improve survival, but
rather would need to establish the safety of withholding
treatment for specific subsets of patients predicted as being
unlikely to benefit. Two general trial designs can be
considered in this setting. Patients could be randomized
to biomarker-based decision-making versus standard-of-
care decision-making using the ‘‘marker strategy design’’
proposed by Simon and Wang However, if ‘‘the analysis
is to demonstrate that withholding a standard therapy for
test-negative patients is not inferior, then sample size
problems are compounded, and even with a huge sample
size, the results are unlikely to be convincing’’ . Alterna-
tively, all patients could be tested for the biomarker with
treatment decisions based on the test results. This approach
was utilized to establish the clinical utility of the Oncotype
Dx assay for decision-making on perioperative systemic
therapy in breast cancer in the TAILORx study . In the
TAILORx study, patients with a gene expression score
predicting a low risk of recurrence were observed without
chemotherapy, those with a high risk score received
perioperative chemotherapy, and those with intermediate
risk scores were randomized to perioperative chemothera-
py versus observation. An analogous design testing the
molecular classifier developed by Seiler and colleagues
could be envisioned in which patients with luminal tumors
are treated with cystectomy alone, those with basal tumors
receive NAC followed by cystectomy, and those with
claudin-low and luminal-infiltrated tumors are randomized
to receive NAC followed by cystectomy versus cystectomy
alone. Importantly, TAILORx enrolled 10 253 patients;
whether there are sufficient data to make the leap to
prospective testing and whether such a trial is feasible in
MIBC should be a high priority topic of discussion for the
bladder cancer community that has longed for tools to
‘‘personalize’’ NAC treatment decisions. In the meantime,
the challenges of establishing clinical validity and utility in
this setting should not be used to justify acceptance of a
lower level of evidence to guide clinical practice.
Conflicts of interest: Matthew D. Galsky has received research funding
from Bristol Myers Squibb, Novartis, Dendreon, and Merck, and has
participated in advisory boards for Agensys, Genentech, Merck-Serono,
and AstraZeneca. John P. Sfakianos and Bart S. Ferket have nothing to
disclose.
References
[1] Plimack ER, Dunbrack RL, Brennan TA, et al. Defects in DNA repair
genes predict response to neoadjuvant cisplatin-based chemothera-
py in muscle-invasive bladder cancer. Eur Urol 2015;68:959–67.
.
[2] Van Allen EM, Mouw KW, Kim P, et al. Somatic ERCC2 mutations
correlate with cisplatin sensitivity in muscle-invasive urothelial
carcinoma. Cancer Discov 2014;4:1140–53.
.
[3] 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;25:152–
65. .
[4]
[5]
.
[6] Simon RM, Paik S, Hayes DF. Use of archived specimens in evaluation
of prognostic and predictive biomarkers. J Natl Cancer Inst
2009;101:1446–52.
[7] Choi W, Sundi D, Metcalfe M, et al. Impact of molecular subtypes on
predictive chemotherapy response and survival in muscle invasive
bladder cancer. Presented at the 2017 American Association for
Cancer Research Annual Meeting.
[8] Simon R, Wang S-J. Use of genomic signatures in therapeutics
development in oncology and other diseases. Pharmacogenomics J
2006;6:166–73. .
[9] Sparano JA, Gray RJ, Makower DF, et al. Prospective Validation of a
21-gene expression assay in breast cancer. N Engl J Med
2015;373:2005–14. .
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 5 5 5 – 5 5 6 556