Table 5.

EPA measurements and risk of prostate cancer progression

Univariable modelsMultivariable modelsa
MethodEPA tertilenHigh-grade prostate cancer (%)OR (95% CI)PbPcOR (95% CI)PbPc
Web-FFQ11631%1.000.231.000.06
21625%0.80 (0.17–3.80)0.780.78 (0.11–5.61)0.80
31520%0.55 (0.11–2.86)0.480.15 (0.01–1.52)0.11
RBC11533%1.000.331.000.64
21729%0.83 (0.19–3.72)0.810.45 (0.03–6.62)0.56
31625%0.67 (0.14–3.17)0.610.62 (0.07–5.80)0.68
Prostate tissue11850%1.000.031.000.01
21625%0.33 (0.08–1.44)0.140.12 (0.01–1.22)0.07
3147%0.08 (0.01–0.72)0.020.008 (<0.001–0.56)0.03

NOTE: Logistic regression models in which outcome is the presence of high-risk prostate cancer defined as Gleason score 7 at the first repeat biopsy session versus low-risk prostate cancer or absence of prostate cancer at the first repeat biopsy session.

Bold font indicates significance at P < 0.05.

  • aMultivariable models were adjusted for age, PSA level, total energy intake, smoking status, time between the initial and the first repeat biopsies, and number of positive biopsy zones at the first repeat biopsy session.

  • bP value of category relative to referent (lowest tertile).

  • cP value of the nutrient variable modeled continuously.