Table 4.

Circulating γ-tocopherol before diagnosis and risk of lethal prostate cancer by two SNPs in GPX4 among men initially diagnosed with nonmetastatic prostate cancer in the HPFS and PHS

SNP allelesEvents/person-yearsBelow batch-specific median HR (95% CI)aBatch-specific median or higher HR (95% CI)aP valueInteraction P valueb
 AA30/2,7311.0 (ref.)0.86 (0.41–1.81)0.69
 AG63/4,8501.0 (ref.)1.07 (0.64–1.79)0.80
 GG21/2,6241.0 (ref.)3.52 (1.27–9.72)0.02
 AA28/2,7931.0 (ref.)0.66 (0.30–1.46)0.31
 AG45/4,6661.0 (ref.)1.12 (0.61–2.04)0.71
 GG20/2,0321.0 (ref.)6.35 (1.78–22.74)0.005
  • aAdjusted for age at diagnosis (continuous), cohort (HPFS vs. PHS), circulating cholesterol levels (batch-specific quartiles), and time between blood draw and diagnosis (continuous).

  • bInteraction P value calculated by adding a cross-product term between the dichotomized γ-tocopherol levels (at or above batch-specific median vs. below) and genotype (additive model) in a model that included the genotype, dichotomized γ-tocopherol levels, circulating cholesterol (batch-specific quartiles), cohort (HPFS vs. PHS), and time between blood draw and diagnosis (continuous).