Table 2.

Relative risk of prostate cancer by quartile of hormone levels and by tumor aggressiveness

Q1*Q2, HR (95% CI)Q3, HR (95% CI)Q4, HR (95% CI)PtrendP§
All prostate cancer
    IGF-IReference0.88 (0.66-1.18)1.08 (0.80-1.44)1.07 (0.79-1.46)0.5
    IGFBP-3Reference1.03 (0.76-1.39)1.19 (0.88-1.60)1.49 (1.11-2.00)0.008
    IGF-I/IGFBP-3Reference0.92 (0.69-1.22)0.95 (0.71-1.28)0.82 (0.61-1.11)0.2
Nonaggressive prostate cancer
    IGF-IReference0.87 (0.64-1.18)0.99 (0.72-1.36)1.09 (0.79-1.51)0.6
    IGFBP-3Reference1.02 (0.74-1.40)1.16 (0.84-1.59)1.40 (1.02-1.92)0.04
    IGF-I/IGFBP-3Reference0.94 (0.69-1.27)0.95 (0.70-1.31)0.84 (0.61-1.17)0.3
Aggressive prostate cancer
    IGF-IReference1.03 (0.56-1.91)1.58 (0.87-2.87)1.07 (0.54-2.11)0.50.7
    IGFBP-3Reference1.06 (0.57-2.00)1.26 (0.68-2.36)1.88 (1.03-3.41)0.050.4
    IGF-I/IGFBP-3Reference0.86 (0.47-1.57)1.02 (0.56-1.85)0.78 (0.42-1.47)0.6>0.9
Late-onset prostate cancer (>64 y)
    IGF-IReference0.88 (0.63-1.22)1.12 (0.80-1.57)1.08 (0.75-1.54)0.5
    IGFBP-3Reference1.04 (0.74-1.45)1.19 (0.85-1.66)1.44 (1.03-2.03)0.04
    IGF-I/IGFBP-3Reference0.90 (0.65-1.26)1.03 (0.73-1.44)0.84 (0.60-1.19)0.5
Early-onset prostate cancer (≤64 y)
    IGF-IReference0.89 (0.52-1.53)0.93 (0.53-1.62)1.05 (0.62-1.78)0.80.8
    IGFBP-3Reference1.03 (0.56-1.87)1.21 (0.67-2.20)1.63 (0.93-2.85)0.060.6
    IGF-I/IGFBP-3Reference0.97 (0.58-1.60)0.74 (0.43-1.26)0.74 (0.43-1.29)0.20.5
Follow-up from 2 y onwards
    IGF-IReference0.95 (0.70-1.30)1.15 (0.84-1.58)1.02 (0.73-1.42)0.60.6
    IGFBP-3Reference0.95 (0.69-1.31)1.08 (0.78-1.48)1.45 (1.06-1.98)0.030.5
    IGF-I/IGFBP-3Reference0.93 (0.69-1.26)1.00 (0.73-1.37)0.80 (0.58-1.11)0.30.9
  • NOTE: A tumor was classified as aggressive if Gleason score was >7 or stage was advanced (T4 or N+ or M+). We were not able to define aggressiveness for six cases because Gleason score and tumor stage were not available (clinical diagnoses only).

  • * Quartiles were assigned within each laboratory batch to adjust for any variation between batches.

  • HRs from Cox regression models adjusted for country of birth (Australia/New Zealand, United Kingdom, Italy, or Greece) and alcohol consumption. The Prentice method has been used to take into account the case-cohort sampling (see Materials and Methods). HRs by tumor aggressiveness and age at diagnosis were obtained from Cox regression models fitted using competing risks methods.

  • The hypothesis of a linear trend in the HR was tested, including in the model a pseudo-continuous variable computed assigning the median level of the specific hormone for each quartile.

  • § Test for difference in the estimates for the pseudo-continuous variables (i.e., linear trend) between aggressive and nonaggressive cases and between early-onset and late-onset cases.

  • Test for difference in the estimates for the pseudo-continuous variables (i.e., linear trend) between the first 2 years and the rest of the follow-up.