Table 3.

Association of plasma vitamin D concentration with prostate cancer risk: SELECT

ModelVitamin D (nmol/L)N (case)N (cohort)HR95% CIP
Overall prostate cancer
 Model 1a<37.51994641.00
37.5–<502594701.080.83–1.410.572
50–<755501,0700.890.70–1.120.328
≥757231,1990.980.78–1.240.897
 Model 2a,b<37.51834261.00
37.5–<502394750.900.68–1.180.444
50–<755881,1230.850.67–1.090.200
≥757211,1790.960.75–1.230.763
 Model 3a,b,c<44.13086391.00
44.1–<58.23186450.830.66–1.030.092
58.2–<72.93206380.740.59–0.920.008
72.9–<90.73636410.860.69–1.070.181
≥90.74226400.980.78–1.210.823
Gleason 2–6 prostate cancer
 Model 1a<37.51074641.00
37.5–<501574701.200.87–1.650.269
50–<753091,0700.900.67–1.200.454
≥754411,1991.050.79–1.400.738
 Model 2a,b<37.5974261.00
37.5–<501374750.940.67–1.320.732
50–<753451,1230.890.66–1.210.467
≥754351,1791.010.75–1.370.943
 Model 3a,b,c<44.11676391.00
44.1–<58.21906450.870.66–1.140.302
58.2–<72.91906380.760.58–1.000.048
72.9–<90.72106410.860.66–1.130.276
≥90.72576401.010.77–1.310.957
Gleason 7–10 prostate cancer
 Model 1a<37.5674641.00
37.5–<50724700.850.56–1.280.435
50–<751571,0700.750.52–1.070.107
≥752061,1990.860.60–1.220.402
 Model 2a,b<37.5604261.00
37.5–<50734750.860.56–1.310.477
50–<751631,1230.750.51–1.090.132
≥752061,1790.910.63–1.330.642
 Model 3a,b,c<44.11046391.00
44.1–<58.2816450.630.45–0.900.010
58.2–<72.9916380.660.47–0.920.016
72.9–<90.71076410.790.56–1.100.165
≥90.71196400.880.63–1.220.436
Gleason 8–10 prostate cancer
 Model 1a<37.5164641.00
37.5–<50144700.710.32–1.580.406
50–<75271,0700.410.20–0.830.013
≥75471,1990.700.36–1.360.288
 Model 2a,b<37.5124261.00
37.5–<50154750.960.41–2.250.926
50–<75291,1230.550.25–1.200.131
≥75481,1790.950.45–2.020.894
 Model 3a,b,c<44.1206391.00
44.1–<58.2206450.680.34–1.340.267
58.2–<72.9136380.360.16–0.780.010
72.9–<90.7276410.850.44–1.650.630
≥90.7246400.780.40–1.540.477
  • aHRs adjusted for age and race (though matching) and family history of prostate cancer, BMI, baseline diabetes, and SELECT treatment arm (as covariates).

  • bVitamin D values adjusted for month of serum sample.

  • cQuintiles are calculated based on the distribution among the cohort.