EURURO Vol. 72 No. 2

E29 320

Letter to the Editor
Re: Giorgio Gandaglia, Nicola Fossati, Armando Stabile,
et al. Radical Prostatectomy in Men with Oligometa-
static Prostate Cancer: Results of a Single-institution
Series with Long-term Follow-up. Eur Urol 2017;72:
289–92
Do the Data Violate Kaplan-Meier Assumptions?
I read with great interest the article of Gandaglia et al
. While there is increasing interest in the potential
oncologic benefit of treating the primary tumor in the
setting of metastatic prostate cancer (PCa), the authors
analyze a series of 11 patients with oligometastatic PCa
managed with radical prostatectomy (RP) in a multimodal
setting between 2006 and 2011.
When reading this article, two questions came to mind:
(1) should these data be used as evidence to support the
treatment of the primary tumor in patients with metastatic
PCa; and (2) should we neglect the methodology in favor of
the originality of approach.
In their conclusion, the authors mention that 7-yr clinical
progression–free survival and cancer-specific mortality
(CSM)–free survival rates were 45% and 82%, respectively,
and that their findings support the safety and effectiveness
of RP in a highly selected cohort of PCa patients with bone
metastases and long-term follow-up. Unfortunately, the
interpretation of the Kaplan-Meier (KM) curves is mislead-
ing, so the authors’ conclusion is not supported by the data
and the methodology.
The KM curve clearly showed that the 7-yr CSM-free
survival was 82%. How could this survival estimate for
patients with metastatic PCa be better than the 7-yr CSM-
free survival probability for high-risk localized PCa treated
with RP or external-beam radiation therapy .
It should be noted that the KM curve showed that the
2.5-yr CSM-free survival was also 82%. Is it logical to have
such a constant survival estimate throughout a long time
interval (5 yr) in the oncology setting? Does 2.5-yr survival
predict long-term survival in this population? Is the
information given by the survival curve after 2.5 yr of
follow-up reliable?
The confidence intervals (CIs) for the survival curves
are wide. The 95% CI varies from 62% to 99% for 7-yr and for
2.5-yr CSM-free survival. In other words, the 7-yr and 2.5-yr
CSM-free survival could range from 62% to 99%. These
results are uninterpretable in a clinical setting.
Contrary to the affirmation by the authors that ‘‘the
long-term disease-free survival observed was higher than
that observed in men managed with ADT alone, for whom
the 5-yr CSM-free survival rates ranged between 48% and
55%’’, the KM curve showed that the 62% CSM-free survival
at 2.5 yr is worse than the 3-yr CSM-free survival rate,
which ranged between 66% and 73% for patients managed
with ADT alone using references as cited by the
authors.
Finally, >65% of patients were censored by 5-yr follow-
up. This number limits the interpretation of the survival
estimates and makes the estimates unreliable beyond this
time point.
Some points need to be clarified by recalling the KM
assumptions. Violation of these assumptions means that the
KM analysis not useful. In KM analyses, the cumulative
probability defines the probability at the beginning and
throughout a time interval. The time intervals are deter-
mined by the distinct noncensored survival times. This
means that the smaller the sample size, the longer the
intervals will be, raising the question of whether the
assumption of a constant survival probability within each
interval is appropriate.
It should also be remembered that after the first patient
is censored, the survival curve becomes an estimate, since
we do not know if censored patients would have experi-
enced an event at some point later in life. Thus, the more
patients are censored in a study, the less reliable is the
survival curve.
Conflicts of interest: The author has nothing to disclose.
References
[1]
.
[2]
.
ava ilable at
journal homepage:
DOI of original article: .
0302-2838/# 2017 European Association of Urology. Published by Elsevier B.V. All rights reserved.