HRs and 95% CIs of breast cancer by quartiles of steroid hormone levels
| Quartiles* | Doubling hormone concentration† | Ptrend‡ | ||||
|---|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | |||
| Total estradiol | ||||||
| Cases/PY§ | 39/1,950 | 47/1,946 | 49/1,917 | 60/1,904 | ||
| HR (95% CI) | Reference | 1.09 (0.66-1.79) | 1.21 (0.74-1.99) | 1.44 (0.89-2.35) | 1.45 (0.91-2.30) | 0.12 |
| Free estradiol | ||||||
| Cases/PY§ | 37/1,934 | 39/1,948 | 53/1,904 | 66/1,923 | ||
| HR (95% CI) | Reference | 1.05 (0.62-1.77) | 1.49 (0.90-2.44) | 1.75 (1.06-2.89) | 1.61 (1.12-2.33) | 0.01 |
| Estrone sulfate | ||||||
| Cases/PY§ | 35/1,954 | 48/1,914 | 43/1,939 | 66/1,898 | ||
| HR (95% CI) | Reference | 1.53 (0.93-2.52) | 1.20 (0.71-2.03) | 2.05 (1.24-3.37) | 1.52 (1.11-2.07) | <0.01 |
| Testosterone | ||||||
| Cases/PY§ | 45/2,007 | 47/1,934 | 48/1,930 | 56/1,946 | ||
| HR (95% CI) | Reference | 1.09 (0.68-1.74) | 1.10 (0.68-1.78) | 1.25 (0.78-2.01) | 1.11 (0.88-1.40) | 0.37 |
| DHEAS | ||||||
| Cases/PY§ | 45/1,984 | 47/1,940 | 44/1,957 | 61/1,948 | ||
| HR (95% CI) | Reference | 1.06 (0.65-1.74) | 1.06 (0.64-1.73) | 1.41 (0.88-2.27) | 1.15 (0.94-1.42) | 0.17 |
| Androstenedione | ||||||
| Cases/PY§ | 36/1,958 | 49/1,944 | 56/1,950 | 55/1,970 | ||
| HR (95% CI) | Reference | 1.39 (0.85-2.28) | 1.66 (1.03-2.69) | 1.49 (0.91-2.44) | 1.24 (0.97-1.57) | 0.08 |
| SHBG | ||||||
| Cases/PY§ | 79/1,985 | 51/1,960 | 36/1,940 | 31/1,955 | ||
| HR (95% CI) | Reference | 0.57 (0.37-0.88) | 0.38 (0.24-0.61) | 0.33 (0.19-0.55) | 0.41 (0.29-0.60) | <0.01 |
NOTE: HRs from the Cox model were adjusted for country of birth, age at menarche, parity and age at first pregnancy, duration of lactation, oral contraceptive use, HRT use, alcohol consumption, energy from diet, smoking, BMI, level of education, and level of physical activity. Estimates were based on the following number of women (cases): total estradiol, 1,019 (195); free estradiol, 1,018 (195); estrone sulfate, 1,014 (192); testosterone, 1,031 (196); DHEAS, 1,032 (197); androstenedione, 1,031 (196); SHBG, 1,034 (197).
↵*Quartiles were adjusted for variations between batches and by age at time of blood collection, according to the procedure described in the method section.
↵†Estimates from the model including the pseudocontinuous variable log2 transformed.
↵‡Test for linear trend using the pseudocontinuous variable log2 transformed.
↵§Breast cancer cases and person-years (PY).