EURURO Vol. 72 No. 6

The main limitation of the present study regarding

exposure was the nonrandom allocation to type of ADT,

with ensuing selection bias for younger and healthier men

with less advanced cancer to receive GnRH agonists. Thus,

despite the semi-ecologic design, residual confounding

cannot be excluded. Since the NPCR does not register date of

treatment, we started follow-up 3 mo after the date of

diagnosis, at which time point 90% of men diagnosed in a

later calendar period had received their primary

[16_TD$DIFF]

treatment.

There was no information on duration and adherence to

GnRH agonists, but it is rare for men with advanced prostate

cancer to stop ADT. We also lacked information on smoking,

body mass index, and use of cardiovascular drugs. Limita-

tions regarding the endpoints are that we used administra-

tive data from the Patient Registry and the Cause of Death

Registry to define CVD events. However, several investiga-

tions have shown high validity for diagnosis of CVD (eg,

heart failure, acute myocardial infarction, and stroke)

[34– 37]

and there are no reasons to assume a systematic bias

according to type of ADT. Strengths of the study included

the nationwide, population-based cohort of men with

comprehensive data from several high-quality health care

registers

[23–26]

as well as the use of three different

statistical methods to assess the association between type

of ADT and risk of CVD.

In accordance with previous studies, men treated with

orchiectomy for prostate cancer in the current study were

older, had more comorbidities, and presented with more

advanced stage of prostate cancer compared to men treated

with GnRH agonists

[12,17–19]

. A meta-analysis including

12 randomised clinical trials found no difference in overall

or prostate cancer survival between men treated with GnRH

agonists and orchiectomy

[38]

. Accordingly, the higher rate

of prostate cancer death among men treated with orchiec-

tomy in our study was the result of more advanced cancer in

comparison to men on GnRH agonists, which in turn

influenced the risk of CVD. The semi-ecologic study design

decreased the influence of an individual’s cancer character-

istics and general health on selection of ADT, but did not

fully eliminate it. However, the risk of CVD among men

treated in units with high use of GnRH agonists was similar

to that in units with low use.

In separate analyses, we determined the net and crude

probability of death from CVD, prostate cancer, and other

causes. Crude probability is estimated using a competing-

risks analysis in which death from causes other than the

Table 3 – Multivariable Cox proportional hazards model

s *

[12_TD$DIFF]

f cardiovascular disease, fractures, and diabetes

CVD

a (

= 5145)

Fracture (

= 1262)

Diabetes

b (

= 705)

HR (95% CI)

value

HR (95% CI)

value

HR (95% CI)

value

Primary treatment

Orchiectomy

1.00 (reference)

GnRH agonists

1.02 (0.96–1.09)

0.5

0.91 (0.80–1.04)

0.16

1.02 (0.85–1.22)

0.8

Year of diagnosis (continuous)

1.02 (1.00–1.04)

0.017

1.11 (1.07–1.15)

0.001

1.02 (0.98–1.07)

0.3

T stage

1.00 (reference)

–

1.00 (reference)

–

1.00 (reference)

–

0.97 (0.86–1.09)

0.6

0.88 (0.70–1.09)

0.2

1.20 (0.86–1.67)

0.3

0.97 (0.87–1.09)

0.6

0.89 (0.72–1.11)

0.3

1.22 (0.89–1.68)

0.2

1.02 (0.89–1.17)

0.8

1.07 (0.82–1.40)

0.6

1.19 (0.80–1.76)

0.4

M stage

M0/MX

1.00 (reference)

–

1.00 (reference)

–

1.00 (reference)

–

1.00 (0.93–1.08)

0.9

1.32 (1.14–1.53)

0.001

1.06 (0.87–1.29)

0.6

Prostate-specific antigen

25 ng/ml

1.00 (reference)

–

1.00 (reference)

–

1.00 (reference)

–

25–

50 ng/ml

0.97 (0.89–1.05)

0.5

1.04 (0.88–1.23)

0.7

0.83 (0.67–1.04)

0.11

50–

100 ng/ml

1.03 (0.95–1.12)

0.5

1.00 (0.84–1.19)

0.86 (0.68–1.08)

0.19

100–

200 ng/ml

1.07 (0.97–1.18)

0.19

1.22 (1.01–1.49)

0.043

0.83 (0.63–1.09)

0.18

200–

500 ng/ml

1.09 (0.98–1.22)

0.11

1.11 (0.89–1.38)

0.4

1.25 (0.96–1.63)

0.10

500 ng/ml

1.05 (0.93–1.18)

0.4

1.01 (0.79–1.29)

0.9

0.93 (0.68–1.28)

0.7

Gleason score

2–6

1.00 (reference)

–

1.00 (reference)

–

1.00 (reference)

–

1.01 (0.93–1.08)

0.9

1.09 (0.94–1.27)

0.2

1.18 (0.96–1.44)

0.12

8–10

1.10 (1.01–1.18)

0.022

1.09 (0.93–1.27)

0.3

1.20 (0.97–1.49)

0.09

Previous CVD

1.00 (reference)

–

1.00 (reference)

–

1.00 (reference)

–

Yes

2.03 (1.90–2.17)

0.001

1.04 (0.90–1.21)

0.6

1.11 (0.91–1.35)

0.3

Previous hypertension

1.00 (reference)

–

1.00 (reference)

–

1.00 (reference)

–

Yes

1.13 (1.00–1.28)

0.051

1.14 (0.88–1.49)

0.3

1.33 (0.93–1.91)

0.12

Previous diabetes

1.00 (reference)

–

1.00 (reference)

–

Yes

1.50 (1.32–1.71)

0.001

1.43 (1.08–1.88)

0.012

CVD = cardiovascular disease; GnRH = gonadotropin-releasing hormone.

All models performed with age as time-sale.

A total of 564 (11%) men had hypertension as their first CVD event.

Only men without previous diabetes before their prostate cancer diagnosis.

Within 5 yr before prostate cancer diagnosis.

E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 9 2 0 – 9 2 8

925