EURURO Vol. 72 No. 6

930 1030

higher versus lower use of orchiectomy. Primary analyses
demonstrated that the risk of cardiovascular disease was
similar for units with the highest and lowest proportions of
GnRH agonist use (relative risk 1.01, 95% confidence interval
0.93–1.11). The semiecologic study design seeks to limit bias
by assuming that menwere assigned to the treatment based
on the providers that they happened to see at the time they
were diagnosed, rather than by other factors such as disease
status or health status that might confound any association
of treatment with outcome. Thus, the use of orchiectomy is
less driven by selection of older or sicker patients, and
comparisons of treatment are more likely to mimic random
treatment allocation. Nevertheless, such designs are at risk
for bias if there remains within-unit variability in the use of
orchiectomy or if there is within-unit confounding. As
noted, even after assignment to groups based on treatment
unit, men in the orchiectomy group were older and had
higher-risk tumors. Of more concern is the possibility of
residual confounding based on time period, since nearly all
the units with very high rates of orchiectomy occurred in
the first one or two time periods (Fig. 1 in the study). With a
decreasing risk of cardiovascular disease incidence in
Sweden over time , a potentially higher risk of
cardiovascular disease associated with more recent treat-
ment, most of which was with GnRH agonist therapy, could
be masked.
Another limitation of the study involves the ascertain-
ment[6_TD$DIFF] of cardiovascular disease. It is noteworthy that the
definition of the cardiovascular disease end point includes
hypertension (not included in the definition of cardiovas-
cular disease in other studies). The very low rate of
hypertension at baseline in this cohort of older men (6–
7%) raises questions about how well medical conditions are
documented. In addition, the data source lacked informa-
tion about the start date and duration of GnRH agonist
therapy. Although, as the authors note, men with advanced
prostate cancer are likely to start treatment immediately
and may continue GnRH agonist therapy indefinitely, only
about one-third of patients in the study had metastatic
disease, and men receiving adjuvant GnRH therapy may
start later and continue therapy for only 6–36 mo Use of
GnRH agonist therapy also requires regular visits, which
could increase the likelihood of identifying incident
cardiovascular disease. It is difficult to predict the effect
of these potential biases on outcomes, since the shorter
duration of androgen suppression with GnRH agonist
therapy would suggest less time at risk and thus less
cardiovascular disease, but their additional visits could lead
to more cardiovascular diagnoses.
The mechanism for the association of androgen depriva-
tion therapy and risk of cardiovascular disease are uncertain
but may be linked to metabolic effects, including central
adiposity, decreased insulin sensitivity, and changes in
lipoproteins and arterial stiffness . Such effects seem to be
related to androgen suppression, and thus there is a reason
to believe that the associations should be similar for
orchiectomy and GnRH agonists, as this study suggests.
The likely harm of androgen deprivation therapy of any type
underscores the need for clinicians toweigh the benefits and
harm of treatment, and engage in shared decision making
with patients. This is of particular importance when
androgen deprivation therapy is considered for clinical
scenarios for which benefits are uncertain, such as primary
treatment of local/regional prostate cancers and biochemical
recurrence[7_TD$DIFF], regardless of the type of androgen deprivation
therapy.
Conflicts of interest: The author has no potential confl icts of interest to
disclose.
Funding support: Dr. Keating's effort is supported by K24CA181510 from
the United States National Cancer Institute.
References
[1]
.
[2]
.
[3]
.
[4]
.
[5]
.
[6]
.
[7]
.
[8] National Comprehensive Cancer Network. Clinical practice guide-
lines in oncology. Prostate cancer Version 2 February 212017
E U R O P E A N U R O L O GY 7 2 ( 2 0 17 ) 9 2 9 – 9 3 0 930