EURURO Vol. 72 No. 6

936 1030

but not those with high-risk tumors (p [9_TD$DIFF]interaction = [10_TD$DIFF]0.02).
A similar result was observed for walking; among men
with lower-risk tumors, those who walked for 7 h/wk
had a 47% lower risk of PCSM compared with those who
walked for 1–3 h/wk (HR: 0.53, 95% CI: 0.33–0.86, p
trend = 0.04).
For postdiagnosis recreational PA ), there was a
significantly lower risk of PCSM for men engaging in 17.5
MET-h/wk of recreational PA compared with those engag-
ing in 3.5–<8.75 MET-h/wk (HR: 0.69, 95% CI: 0.49–0.95, p
trend = 0.006). Further adjustment for prediagnosis METs
did not change the results (not shown). Postdiagnosis
walking, but not other recreational PA, was inversely
associated with PCSM, although the HRs were not statisti-
cally significant. The inverse association for recreational PA
did not differ by tumor risk category (p [9_TD$DIFF]interaction = [13_TD$DIFF]0.1).
We also examined recreational PA in relation to CVD
mortality and all-cause mortality. Engaging in 17.5 MET-
h/wk of recreational PA before diagnosis was associated
with a 20% lower risk of CVD mortality (HR: 0.80, 95% CI:
0.67–0.96, p trend = 0.008; Supplementary Table 2). In
postdiagnosis analyses, the inverse trend (p trend = 0.01) for
recreational PA was primarily driven by men in the bottom
category who had a significantly higher risk of CVD
mortality (HR: 1.49, 95% CI: 1.11–2.00) rather than by
those in the top category (HR: 0.89, 95% CI: 0.69–1.14). A
similar result was seen for walking but not for other
recreational PA. For all-cause mortality, same comparison of
prediagnosis recreational PA ( 17.5 vs 3.5–<8.75 MET-h/
wk) was associated with 12% lower risk (HR: 0.88, 95% CI:
0.80–0.97, p trend < 0.0001; Supplementary Table 3). In
postdiagnosis analyses, engaging in 17.5 MET-h/wk of
recreational PA was associated with a 14% lower risk of all-
cause mortality (HR: 0.86, 95% CI: 0.75–0.98, p trend <
0.0001). Similar results were seen for both walking and
other recreational PA.
The associations of pre- or postdiagnosis recreational PA
with PCSM did not differ by age at diagnosis (<70 and 70
yr), BMI (<25, 25–<30, and 30 kg/m2), smoking status
(never and ever), diagnosis year (1992–1998, 1999–2005,
and 2006–2011), or follow-up time (<10 and 10 yr for
prediagnosis analyses, <6 and 6 yr for postdiagnosis
analyses). Change in PA from pre- to postdiagnosis showed
no clear patterns of associations with PCSM (data not
shown).
4. Discussion
In this large prospective cohort study of men diagnosed
with nonmetastatic prostate cancer, higher prediagnosis
Table 4 – Prostate cancer–specific mortality by postdiagnosis recreational physical activity among men diagnosed with nonmetastatic
prostate cancer in the CPS-II Nutrition Cohort (1992–2012)
Median
MET-h/wk
All prostate
cancer [12_TD$DIFF]
Lower-risk prostate
cancer
High-risk prostate
cancer
p interaction
Deaths/
person-yr
HR
(95% CI)
Deaths/
person-yr
HR
(95% CI)
Deaths/
person-yr
HR
(95% CI)
Total recreational
physical activity [9_TD$DIFF]
(MET-h/wk)
[13_TD$DIFF]0.1
<3.5 0.9 45/3079 1.13 (0.77–1.66) 24/2116 1.16 (0.69–1.95) 15/489 1.25 (0.60–2.62)
3.5–<8.75 6.8 76/7526 1.00 (ref.) 49/5407 1.00 (ref.) 21/1204 1.00 (ref.)
8.75–<17.5 14 63/7462 0.81 (0.58–1.15) 29/5272 0.72 (0.44–1.15) 24/1305 0.77 (0.40–1.49)
17.5+ 29.5 77/12 427 0.69 (0.49–0.95) 41/8966 0.67 (0.43–1.04) 28/2084 0.69 (0.37–1.29)
p trend 0.006 0.04 0.1
Walking h/wk) 0.36
<1 0 60/4915 1.07 (0.77–1.49) 29/3346 0.92 (0.58–1.46) 23/775 1.77 (0.95–3.29)
1–3 7 106/11 679 1.00 (ref.) 68/8504 1.00 (ref.) 28/1832 1.00 (ref.)
4–6 14 52/6866 0.85 (0.60–1.20) 23/4834 0.68 (0.41–1.10) 21/1306 1.10 (0.58–2.06)
7 24.5 43/7034 0.76 (0.53–1.10) 23/5078 0.70 (0.43–1.15) 16/1169 1.01 (0.52–1.98)
p trend 0.07 0.2 0.2
Other recreational
physical activity [14_TD$DIFF]
(MET-h/wk)
0.49
<3.5 0 177/17 758 1.38 (0.96–1.99) 101/12 562 1.62 (0.97–2.69) 57/3 088 1.03 (0.56–1.92)
3.5–<8.75 5.5 38/5829 1.00 (ref.) 19/4137 1.00 (ref.) 16/917 1.00 (ref.)
8.75–<17.5 12.5 27/3901 1.05 (0.63–1.75) 13/2925 1.06 (0.51–2.20) 11/603 1.01 (0.42–2.41)
17.5+ 28 19/3006 1.15 (0.65–2.02) 10/2137 1.23 (0.56–2.70) 4/474 0.51 (0.15–1.71)
p trend 0.2 0.2 0.3
CPS = Cancer Prevention Study; HR = hazard ratio; CI = confidence interval; MET = metabolic equivalent task; ref. = reference.
a Includes T1–T2 cancers with unknown Gleason score not included in lower- or high-risk categories.
b Defined as Gleason score 2–7 and T1–T2.
c Defined as Gleason score 8–10 or T3–T4 or nodal involvement.
d Adjusted for age at diagnosis, race, calendar year of diagnosis, tumor extent, nodal involvement, Gleason score, education, family history of prostate cancer,
initial treatment, history of prediagnosis prostate-specific antigen testing, postdiagnosis cardiovascular disease history, postdiagnosis other cancer history,
postdiagnosis body mass index, postdiagnosis smoking status, prediagnosis red meat and processed meat intake, and postdiagnosis sitting time.
e Further adjusted for other recreational physical activity.
f MET hours of other physical activity is calculated as total recreational MET hours minus walking MET hours; model further adjusted for walking.
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 9 3 1 – 9 3 9 936