Abstract

Studies of circulating estrogen levels in relation to pre-menopausal breast cancer risk have yielded inconsistent results. Various estrogen metabolites might affect the risk differently. Estradiol metabolism occurs primarily via two mutually exclusive pathways, yielding 2-hydroxyestrone (2-OHE) and 16α-hydroxyestrone (16α-OHE). Most, but not all, studies have found that a relatively high 2-OHE/16α-OHE ratio is associated with a low breast cancer risk. Our objective was to determine if the 2-OHE/16α-OHE ratio in plasma correlates with suspected breast cancer risk factors and other lifestyle factors, such as ethnicity, body size, age at menarche, oral contraceptive use, smoking, vegetarian diet, coffee and alcohol consumption in 513 nulliparous women, aged 17–35. Oral contraceptive users had significantly lower 2-OHE/16α-OHE ratios than pill non-users ( P = 10 −21 ). Among women who were not using oral contraceptives, the median 2-OHE/16α-OHE ratio in plasma was similar for white, black, Indian/Pakistani and Asian women, after adjustment for age and menstrual cycle phase. Among oral contraceptive users, Asian women had significantly lower 2-OHE/16α-OHE ratios than white women, and this result remained after adjustment for age and day of menstrual cycle. Daily coffee consumption was significantly positively correlated with 2-OHE/16α-OHE ratios ( rs = 0.18, P = 0.002) only among pill non-users. Our findings suggest that the plasma 2-OHE/16α-OHE ratio is associated with constitutional factors and with modifiable lifestyle factors. The reported elevated risk of early onset breast cancer among young oral contraceptive users could be mediated in part through altered estrogen metabolism induced by synthetic estrogens and progestins.

Introduction

Breast cancer is the most common cancer in women. The role of endogenous estrogens in the development of breast cancer has been inferred from the observed protective effects of late age at menarche, early menopause and bilateral oophorectomy before age 40 ( 1 , 2 ). Each of these factors shortens the time that the breast tissue is exposed to endogenous estrogen. Several case–control and prospective studies have demonstrated an association between post-menopausal breast cancer and high estrogen levels ( 3 , 4 ). However, studies of circulating estrogen levels on pre-menopausal breast cancer risk have yielded inconsistent results ( 5 , 6 ) and it is possible that different products of estrogen metabolism may differentially influence the risk ( 7 , 8 ). Estradiol metabolism occurs primarily via two mutually exclusive pathways, yielding 2-hydroxyestrone (2-OHE) and 16α-hydroxyestrone (16α-OHE). Studies have shown that 2-OHE acts as a weak estrogen or an anti-estrogen ( 9 ) whereas 16α-OHE is pro-carcinogenic ( 10 ). The 2-OHE/16α-OHE ratio reflects the relative activities of these two mutually exclusive metabolic pathways and is independent of the absolute levels of the two metabolites. While the absolute levels of these two compounds vary over the menstrual cycle, Westerlind et al . have shown that the urinary 2-OHE/16α-OHE ratio is stable over the menstrual cycle ( 11 ). Most ( 7 , 1215 ), but not all ( 16 , 17 ), epidemiological studies have found that women with a high urinary ratio of the metabolites 2-OHE and 16α-OHE are at reduced risk for breast cancer. It is possible that the influence of the 2-OHE/16α-OHE ratio on breast cancer risk is different for pre- and post-menopausal women. However, the association has been demonstrated in both groups ( 13 , 14 ).

Enzyme immunoassays for measurement of 2-OHE and 16α-OHE in urine ( 18 ) have recently been modified to provide for measurement of these metabolites in serum and plasma. One pilot study has examined the relationship between levels of circulating estrogen metabolites and breast cancer. In a small prospective case–control study, post-menopausal women with a high serum 2-OHE/16α-OHE ratio were at reduced risk of breast cancer ( 19 ).

We measured levels of 2-OHE and 16α-OHE in nulliparous young women in plasma and calculated the ratio between the two. We have correlated the 2-OHE/16α-OHE ratio with known and suspected risk factors for breast cancer, including a vegetarian diet, fruit and vegetable consumption, body mass index (BMI), smoking, alcohol use, coffee consumption, oral contraceptive use, age at menarche, ethnicity and a family history of breast cancer.

Materials and methods

Study subjects

Healthy volunteer women, between the ages of 17 and 35 years, were recruited from the University of Toronto, from the Bay Center for Birth Control and from the Toronto community between July 1998 and January 2000. The University of Toronto ethics committee approved the protocol. The present study deals with the results on 513 women from the four most common ethnic groups. These women reported never to have been pregnant and had no history of cancer. Forty-four women of other or mixed ethnicity were excluded. In addition, two women on Depo-Provera, one woman on Norplant, one hysterectomized woman taking estrogen replacement therapy and one woman who did not answer if she was taking oral contraceptives were excluded. Asian women included those of Japanese, Chinese, Korean, Vietnamese or Thai descent, while the majority of these women were of Chinese (50%) or Korean (31%) descent. The black women originated from Canada, the USA, the West Caribbean, England and Africa. The study protocol included a morning visit between 8 a.m. and 12 noon to the study center on a random day of the menstrual cycle. Women were asked not to consume any grapefruit or grapefruit juice during the 2 days prior to the study visit because grapefruit may inhibit CYP3A4 ( 20 ) and thus alter the 2-OHE/16α-OHE ratio ( 21 ). All women completed a self-administered questionnaire, detailing information on age, ethnic group, menstrual history, diet (vegetarian or not), fruit and vegetable consumption (number of daily servings), alcohol and coffee consumption and current smoking (yes/no). A brief medical history was taken and included the current use of medications, including oral contraceptives. Samples of various packages of oral contraceptive brands were displayed at the center and the women were asked to identify their current brand. All contraceptive formulations contained 17α-ethinyl estradiol as estrogen, but the type of progestin varied. Each type of contraceptive was coded as to the amount of estrogen and type of progestin it contained. The average daily dose of ethinyl estradiol was calculated for women who used triphasic pills. Seven different types of progestins were recorded.

Women were asked to indicate the date of the first menstrual flow of the previous cycle and were required to report by telephone when the onset of menses began for the next cycle. The date of blood draw was then classified according to the calculated day of the menstrual cycle. Menstrual cycle day information was missing for 37 white, eight Asian, four black and eight Indian/Pakistani women, i.e. a total of 57 women, 14 women did not call back with the date of their next period and 43 women had irregular periods and could not be classified. Menstrual cycle day was categorized into seven categories in relation to onset of bleeding [days 1–4, days 5–8, days 9–12, mid-cycle (days 13–14 and days −14 to −13), days −12 to −9, days −8 to −5 and days −4 to −1] until onset of the next menstrual period.

Enzyme immunoassays

Blood samples were taken for analysis of levels of 2-OHE and 16α-OHE in EDTA–plasma. Samples were blind to the laboratory doing the analyses (samples were labeled with serial numbers only). Levels of 2-OHE and 16α-OHE in EDTA–plasma were measured by a monoclonal antibody-based enzyme assay (ESTRAMET™ 2 and 16; Immuna Care Inc., Bethlehem, PA). The enzyme immunoassays for estrogen metabolites 2-OHE and 16α-OHE in serum were developed from reagents and buffers previously designed for the measurement of these metabolites in urine ( 18 , 2224 ). The assays for urinary estrogen metabolites have been validated against gas chromatography–mass spectroscopy methods ( 18 , 22 ). Serum standards were calibrated against pure estrogen metabolites. The sample diluent buffer for the serum assays was modified to contain biotinylated monoclonal antibody to the respective metabolites, and biotinylated antibodies were captured directly on streptavidin-coated wells of polystyrene microtiter plates. The sample diluent for the serum 2-OHE assay contained deconjugating enzymes from Helix pomatia , whereas the sample diluent for the serum 16α-OHE assay did not, as the antibody bound the free form and 3-glucuronide equivalently. Standards for the serum assays were prepared by adding known amounts of 2-hydroxyestrone 3-glucuronide and 16α-hydroxyestrone 3-glucuronide to charcoal-stripped pooled human serum. Standard concentrations for serum 2-OHE and 16α-OHE were 0, 50, 100, 200, 400 and 800 pg/ml. Duplicate 25 µl aliquots of standards, positive controls and test sera were diluted sequentially with 150 µl of serum diluent buffer with antibodies and 150 µl of conjugate diluent buffer containing diluted estrogen metabolite–alkaline phosphatase conjugate. After 15 min, 150 µl of this mixture was added to each well of the respective microtiter plates. Serum positive controls were used in every assay on every assay plate. After incubation for 16 h at 4°C, the plates were washed and enzyme substrate added. The optical densities in the assay plate wells were determined after 2 h incubation at room temperature. Standard curves were fitted by a four parameter fit as previously described ( 18 ).

The current immunoassays have a sensitivity of <20 pg/ml serum for 2-OHE and 16α-OHE. Thirty women had a 2-OHE value below the limit of detection of 20 pg/ml and their values were approximated to 20 pg/ml. None of the women had a 16α-OHE concentration below the limit of detection.

Results from recent studies of 200–500 women indicate that duplicate measures of serum 2-OHE correlate highly with a correlation coefficient (Spearman) of rs > 0.90, whereas duplicate measures of serum 16α-OHE correlate very highly ( rs = 0.95). In a recent study of 249 sera from post-menopausal women, assayed in five assays over a 1 week period, intra- and interassay variability for 2-OHE and 16α-OHE in six positive control sera were <4 and 7% and <5 and 12%, respectively (T.L.Klug, personal communication).

Data analysis

The values of 2-OHE and 16α-OHE and the 2-OHE/16α-OHE ratio were not normally distributed, as shown in the histograms ( Figure 1 ), and therefore a non-parametric test (Mann–Whitney U -test) was used for univariate analyses. We assessed univariate differences in the ratio using the Mann–Whitney U -test for categorical variables such as ethnic group (white women were used as the reference group), current oral contraceptive use (yes/no), current smoking (yes/no) and vegetarian diet (yes/no). Spearman's rank order correlation ( rs ) was used for correlations between the values of 2-OHE, 16α-OHE and the 2-OHE/16α-OHE ratio and age (continuous), age at menarche (continuous), day of the menstrual cycle (seven categories), BMI (continuous), coffee consumption (six categories, 0–≥5), alcohol consumption (categorical, no drinks, 0–3 drinks, 4–9 drinks, 10+ drinks) and fruit and vegetable consumption (seven categories, 0–≥6). The values of 2-OHE, 16α-OHE and the 2-OHE/16α-OHE ratio were transformed using the natural logarithm for analyses in multivariate models. For Table II , analysis of covariance was used to evaluate age-adjusted mean differences in log levels of 2-OHE and 16α-OHE and the log 2-OHE/16α-OHE ratio between two ethnic groups and the two menstrual cycle phases, using white women as the reference. These average differences correspond to ratios between geometric means on the original scale.

Fig. 1.

( A ) A histogram that demonstrates that the 2-OHE/16α-OHE ratio was not normally distributed. ( B ) A histogram that demonstrates that the levels of 2-OHE were not normally distributed. ( C ) A histogram that demonstrates that the levels of 16α-OHE were not normally distributed.

Fig. 1.

( A ) A histogram that demonstrates that the 2-OHE/16α-OHE ratio was not normally distributed. ( B ) A histogram that demonstrates that the levels of 2-OHE were not normally distributed. ( C ) A histogram that demonstrates that the levels of 16α-OHE were not normally distributed.

Table II.

Comparison of levels of 2-OHE and 16α-OHE and the 2-OHE/16α-OHE ratio, in different ethnic groups by time of menstrual cycle among women not using oral contraceptives

  White
 
   Black
 
   Asian
 
   Indian/Pakistani
 
  

 
GM25
 
GM ratio
 
P
 
GM25
 
GM ratio
 
P
 
GM25
 
GM ratio
 
P
 
GM25
 
GM ratio
 
P
 
2-OHE/16α-OHE ratio             
    Follicular phase & midcycle  0.56 ( n = 72)  1.00 ref.  0.68 ( n = 32)  1.22 0.16  0.43 ( n = 24)  0.77 0.17  0.63 ( n = 6)  1.12 0.73 
    Luteal phase  0.78 ( n = 70)  1.00 ref.  0.69 ( n = 25)  0.89 0.36  0.72 ( n = 21)  0.92 0.56  0.56 ( n = 4)  0.72 0.28 
2-OHE (pg/ml)             
    Follicular phase & midcycle  160 ( n = 72)  1.00 ref.  246 ( n = 32)  1.54 0.002  123 ( n = 24)  0.77 0.18  139 ( n = 6)  0.87 0.66 
    Luteal phase  242 ( n = 70)  1.00 ref.  271 ( n = 25)  1.12 0.34  211 ( n = 21)  0.87 0.30  172 ( n = 4)  0.71 0.19 
16α-OHE (pg/ml)             
    Follicular phase & midcycle  287 ( n = 72)  1.00 ref.  363 ( n = 32)  1.26 0.001  287 ( n = 24)  1.00 0.98  224 ( n = 6)  0.78 0.11 
    Luteal phase  312 ( n = 70)  1.00 ref.  391 ( n = 25)  1.25 0.003  297 ( n = 21)  0.95 0.52  308 ( n = 4)  0.99 0.93 
  White
 
   Black
 
   Asian
 
   Indian/Pakistani
 
  

 
GM25
 
GM ratio
 
P
 
GM25
 
GM ratio
 
P
 
GM25
 
GM ratio
 
P
 
GM25
 
GM ratio
 
P
 
2-OHE/16α-OHE ratio             
    Follicular phase & midcycle  0.56 ( n = 72)  1.00 ref.  0.68 ( n = 32)  1.22 0.16  0.43 ( n = 24)  0.77 0.17  0.63 ( n = 6)  1.12 0.73 
    Luteal phase  0.78 ( n = 70)  1.00 ref.  0.69 ( n = 25)  0.89 0.36  0.72 ( n = 21)  0.92 0.56  0.56 ( n = 4)  0.72 0.28 
2-OHE (pg/ml)             
    Follicular phase & midcycle  160 ( n = 72)  1.00 ref.  246 ( n = 32)  1.54 0.002  123 ( n = 24)  0.77 0.18  139 ( n = 6)  0.87 0.66 
    Luteal phase  242 ( n = 70)  1.00 ref.  271 ( n = 25)  1.12 0.34  211 ( n = 21)  0.87 0.30  172 ( n = 4)  0.71 0.19 
16α-OHE (pg/ml)             
    Follicular phase & midcycle  287 ( n = 72)  1.00 ref.  363 ( n = 32)  1.26 0.001  287 ( n = 24)  1.00 0.98  224 ( n = 6)  0.78 0.11 
    Luteal phase  312 ( n = 70)  1.00 ref.  391 ( n = 25)  1.25 0.003  297 ( n = 21)  0.95 0.52  308 ( n = 4)  0.99 0.93 

The levels were log transformed and analysis of covariance was used to adjust for age. Estimated geometric means at age 25 (GM25), age adjusted geometric mean (GM) ratios, corresponding to age adjusted mean differences on the logarithmic scale and P values are presented.

Results

The total sample of 513 women included 340 white women (66%), 77 black women (15%), 70 Asian women (14%) and 26 women of Indian or Pakistani origin (5%). All women were between the ages of 17 and 35 years. The characteristics of the women in the four ethnic groups are presented in Table I .

Table I.

Characteristics of women by ethnic background

  White ( n = 340)
 
  Black ( n = 77)
 
  Asian ( n = 70)
 
  Indian/Pakistani ( n = 26)
 
 

 
Median
 
(Range)
 
Median
 
(Range)
 
Median
 
(Range)
 
Median
 
(Range)
 
Age (years) 25 (17–35) 24 (17–35) 24 (19–31) 22 (19–30) 
Height (cm) 165 (147–188) 164 (150–182) 159 (150–173) 159 (154–166) 
Weight (kg) 61.3 (41.3–114.9) 62.6 (44.5–95.3) 54.3 (39.5–78.5) 52.7 (40.4–79.4) 
BMI (kg/m 2 )  22.0 (16.9–46.3) 23.0 (15.8–32.4) 21.2 (16.4–27.4) 20.7 (16.0–32.6) 
Age at menarche (years) 13 (9–18) 12 (8–17) 12 (8–16) 12 (10–15) 
First or second degree family history of breast cancer 23%  7%  12%  21%  
Current oral contraceptive user 50%  21%  24%  35%  
Current smoker 15%  9%  16%  15%  
Current vegetarian diet 15%  3%  4%  19%  
Fruit and vegetable servings/day (0–6+) (0–6+) (2–6+) (0–6+) 
Coffee consumption (cups/day) (0–5+) (0–4) (0–4) (0–5+) 
Alcoholic drinks per week         
    None 11%  34%  32%  20%  
    0–3 61%  64%  59%  68%  
    4–9 24%  3%  7%  12%  
    10–20 4%    1%    
  White ( n = 340)
 
  Black ( n = 77)
 
  Asian ( n = 70)
 
  Indian/Pakistani ( n = 26)
 
 

 
Median
 
(Range)
 
Median
 
(Range)
 
Median
 
(Range)
 
Median
 
(Range)
 
Age (years) 25 (17–35) 24 (17–35) 24 (19–31) 22 (19–30) 
Height (cm) 165 (147–188) 164 (150–182) 159 (150–173) 159 (154–166) 
Weight (kg) 61.3 (41.3–114.9) 62.6 (44.5–95.3) 54.3 (39.5–78.5) 52.7 (40.4–79.4) 
BMI (kg/m 2 )  22.0 (16.9–46.3) 23.0 (15.8–32.4) 21.2 (16.4–27.4) 20.7 (16.0–32.6) 
Age at menarche (years) 13 (9–18) 12 (8–17) 12 (8–16) 12 (10–15) 
First or second degree family history of breast cancer 23%  7%  12%  21%  
Current oral contraceptive user 50%  21%  24%  35%  
Current smoker 15%  9%  16%  15%  
Current vegetarian diet 15%  3%  4%  19%  
Fruit and vegetable servings/day (0–6+) (0–6+) (2–6+) (0–6+) 
Coffee consumption (cups/day) (0–5+) (0–4) (0–4) (0–5+) 
Alcoholic drinks per week         
    None 11%  34%  32%  20%  
    0–3 61%  64%  59%  68%  
    4–9 24%  3%  7%  12%  
    10–20 4%    1%    

Height is missing for two white and one Indian/Pakistani woman. Weight is missing for 15 white, one black and four Asian women. BMI is missing for 17 white, four Asian, one black and one Indian/Pakistani woman. Coffee consumption is missing for five white, one Asian and one black woman. Family history of breast cancer was missing for 31 white, 18 Asian, 19 black and seven Indian/Pakistani women. Fruit and vegetable consumption is missing for three white women. Alcohol consumption is missing for one Asian and one Indian/Pakistani woman.

Age

Among women who did not use oral contraceptives, there was a significant positive correlation between age and the 2-OHE/16α-OHE ratio ( rs = 0.16, P = 0.004) ( Figure 2A ). No significant correlations were present between age and the plasma level of 2-OHE ( rs = 0.10, P = 0.09) ( Figure 2B ) or the plasma level of 16α-OHE ( rs = −0.07, P = 0.22) ( Figure 2C ). Among women who used oral contraceptives there was no significant correlation between age and the 2-OHE/16α-OHE ratio in plasma ( rs = −0.07, P = 0.28) or the plasma levels of 2-OHE ( rs = −0.12, P = 0.07) or 16α-OHE ( rs = −0.11, P = 0.13) ( Figure 2A–C ). Similar correlations were observed for white women and for the whole study population, but the positive correlation between age and the plasma level of 2-OHE became significant among pill non-users ( rs = 0.16, P = 0.04). After adjustment for ethnicity and menstrual cycle day, the negative relationship between age and 16α-OHE among pill non-users became of borderline significance ( P = 0.048), whereas the rest of the results remained essentially the same after adjustments for these two factors.

Fig. 2.

( A ) A significant positive correlation existed between age and the 2-OHE/16α-OHE ratio ( rs = 0.16, P = 0.004) among women who did not use oral contraceptives, which remained significant after adjustment for ethnicity and menstrual cycle day ( P = 0.01). No significant correlation was observed in the subgroup of oral contraceptive users. ( B ) There was no significant correlation between the levels of 2-OHE and age among pill non-users or among oral contraceptive users. ( C ) There was no significant correlation between the levels of 2-OHE and age among pill non-users or among oral contraceptive users. After adjustment for ethnicity and menstrual cycle day, the negative relationship between age and 16α-OHE among pill non-users became of borderline significance ( P = 0.048).

Fig. 2.

( A ) A significant positive correlation existed between age and the 2-OHE/16α-OHE ratio ( rs = 0.16, P = 0.004) among women who did not use oral contraceptives, which remained significant after adjustment for ethnicity and menstrual cycle day ( P = 0.01). No significant correlation was observed in the subgroup of oral contraceptive users. ( B ) There was no significant correlation between the levels of 2-OHE and age among pill non-users or among oral contraceptive users. ( C ) There was no significant correlation between the levels of 2-OHE and age among pill non-users or among oral contraceptive users. After adjustment for ethnicity and menstrual cycle day, the negative relationship between age and 16α-OHE among pill non-users became of borderline significance ( P = 0.048).

Ethnicity

Among women who were not using oral contraceptives, the 2-OHE/16α-OHE ratio in plasma was lower among Asian women than white women ( P = 0.02) ( Figure 3A ), but this difference was not significant after adjustment for age and menstrual cycle phase. Table II shows the estimated geometric means of the 2-OHE/16α-OHE ratios at age 25 for the two menstrual cycle phases. The 2-OHE/16α-OHE ratios were similar for white, black and Indian/Pakistani women during the two menstrual cycle phases. Compared with white women, unadjusted 2-OHE levels were lower in Asian women ( P = 0.07) and in Indian/Pakistani women ( P = 0.04) and higher in black women ( P = 0.003) ( Figure 3B ). After adjustment for age and menstrual cycle phase the 2-OHE levels were significantly higher during the follicular phase and mid-cycle only among black women compared with white women ( P = 0.002). The unadjusted plasma levels of 16α-OHE were similar in white, Asian and Indian/Pakistani women, but black women had significantly higher plasma levels of 16α-OHE than white women ( P = 2 × 10 −6 ) ( Figure 3C ). This difference remained significant after adjustment for age and menstrual cycle phase.

Fig. 3.

( A ) The 2-OHE/16α-OHE ratio in plasma in relation to ethnic background among women not using oral contraceptives. The 2-OHE/16α-OHE ratios were similar for white, black and Indian/Pakistani women. The 2-OHE/16α-OHE ratio was lower among Asian women than in white women ( P = 0.02), but this was no longer significant after adjustment for age and menstrual cycle phase. ( B ) The 2-OHE levels in relation to ethnic background. Compared with white women, 2-OHE levels were lower in Asian women ( P = 0.07) and in Indian/Pakistani women ( P = 0.04) and higher in black women ( P = 0.003). After adjustment for age and menstrual cycle phase, only black women had significantly higher 2-OHE levels compared with white women ( P = 0.002) in the follicular phase and mid-cycle. ( C ) The plasma levels of 16α-OHE in relation to ethnic background. The levels of 16α-OHE were similar in white, Asian and Indian/Pakistani women, but black women had significantly higher plasma levels of 16α-OHE than white women ( P = 2 × 10 −6 ). After adjustment for age and menstrual cycle phase, black women had significantly higher 16α-OHE levels compared with white women in both menstrual cycle phases ( P = 0.001 and P = 0.003, respectively).

Fig. 3.

( A ) The 2-OHE/16α-OHE ratio in plasma in relation to ethnic background among women not using oral contraceptives. The 2-OHE/16α-OHE ratios were similar for white, black and Indian/Pakistani women. The 2-OHE/16α-OHE ratio was lower among Asian women than in white women ( P = 0.02), but this was no longer significant after adjustment for age and menstrual cycle phase. ( B ) The 2-OHE levels in relation to ethnic background. Compared with white women, 2-OHE levels were lower in Asian women ( P = 0.07) and in Indian/Pakistani women ( P = 0.04) and higher in black women ( P = 0.003). After adjustment for age and menstrual cycle phase, only black women had significantly higher 2-OHE levels compared with white women ( P = 0.002) in the follicular phase and mid-cycle. ( C ) The plasma levels of 16α-OHE in relation to ethnic background. The levels of 16α-OHE were similar in white, Asian and Indian/Pakistani women, but black women had significantly higher plasma levels of 16α-OHE than white women ( P = 2 × 10 −6 ). After adjustment for age and menstrual cycle phase, black women had significantly higher 16α-OHE levels compared with white women in both menstrual cycle phases ( P = 0.001 and P = 0.003, respectively).

Menstrual cycle

We divided the menstrual cycle into seven different categories depending on what day of the cycle the blood sample was drawn in relation to onset of bleeding of the previous and of the next menstrual period. Among pill non-users, the 2-OHE/16α-OHE ratio in plasma increased during the menstrual cycle ( rs = 0.26, P = 4 × 10 −5 ) ( Figure 4A ). The same was true for 2-OHE ( rs = 0.34, P = 10 −6 ) ( Figure 4B ) and 16α-OHE ( rs = 0.14, P = 0.03) ( Figure 4C ). It is common to take active contraceptive pills for 21 days followed by 7 days of placebo pills (or no pills). During the placebo period the woman will experience an artificial menstrual bleeding. The 2-OHE/16α-OHE ratio was low during cycle days 1–4 when no pills or the placebo pills were consumed. The 2-OHE/16α-OHE ratio increased during cycle days 5–8 when the active pills were introduced, declined throughout the 3 weeks on active pills and rose again just prior to onset of the next bleeding when the placebo pill was introduced. There was a significant negative correlation between the 2-OHE/16α-OHE ratio from cycle day 5 until 4 days prior to onset of the next period ( rs = –0.17, P = 0.04) (the days when active pills are typically consumed). The plasma level of 2-OHE showed a pattern similar to the 2-OHE/16α-OHE ratio ( rs = – 0.16, P = 0.05), whereas there was no correlation between the level of 16α-OHE and the menstrual cycle day ( Figure 4 ). The results remained essentially the same after adjustment for age and ethnicity.

Fig. 4.

( A ) The plasma ratio of 2-OHE to 16α-OHE in relation to the day of the menstrual cycle. We divided the menstrual cycle into seven different categories depending on what day of the cycle the blood sample was drawn in relation to onset of bleeding of the previous and of the next menstrual period. Among pill non-users, the 2-OHE/16α-OHE ratio in plasma increased during the menstrual cycle ( rs = 0.26, P = 2 × 10 −5 ). Among oral contraceptive users the plasma ratio of 2-OHE to 16α-OHE was low during cycle days 1–4 when no pills or the placebo pills were consumed. There was a significant negative correlation between the 2-OHE/16α-OHE ratio from cycle day 5 until 4 days prior to onset of the next period, during which days active pills would have been consumed ( rs = −0.17, P = 0.04). The results remained essentially the same after adjustment for age and ethnicity. ( B ) The relationship between day of the menstrual cycle and the plasma level of 2-OHE. Among pill non-users the levels of 2-OHE in plasma increased during the menstrual cycle ( rs = 0.35, P = 10 −6 ). Among oral contraceptive users the plasma level of 2-OHE declined during the days when active pills were consumed ( rs = – 0.16, P = 0.05). The results remained essentially the same after adjustment for age and ethnicity. ( C ) The relationship between day of the menstrual cycle and 16α-OHE. Among pill non-users the levels of 16α-OHE in plasma increased during the menstrual cycle ( rs = 0.14, P = 0.03). There was no consistent pattern between day of menstrual cycle and the levels of 16α-OHE among oral contraceptive users. The results remained essentially the same after adjustment for age and ethnicity.

Fig. 4.

( A ) The plasma ratio of 2-OHE to 16α-OHE in relation to the day of the menstrual cycle. We divided the menstrual cycle into seven different categories depending on what day of the cycle the blood sample was drawn in relation to onset of bleeding of the previous and of the next menstrual period. Among pill non-users, the 2-OHE/16α-OHE ratio in plasma increased during the menstrual cycle ( rs = 0.26, P = 2 × 10 −5 ). Among oral contraceptive users the plasma ratio of 2-OHE to 16α-OHE was low during cycle days 1–4 when no pills or the placebo pills were consumed. There was a significant negative correlation between the 2-OHE/16α-OHE ratio from cycle day 5 until 4 days prior to onset of the next period, during which days active pills would have been consumed ( rs = −0.17, P = 0.04). The results remained essentially the same after adjustment for age and ethnicity. ( B ) The relationship between day of the menstrual cycle and the plasma level of 2-OHE. Among pill non-users the levels of 2-OHE in plasma increased during the menstrual cycle ( rs = 0.35, P = 10 −6 ). Among oral contraceptive users the plasma level of 2-OHE declined during the days when active pills were consumed ( rs = – 0.16, P = 0.05). The results remained essentially the same after adjustment for age and ethnicity. ( C ) The relationship between day of the menstrual cycle and 16α-OHE. Among pill non-users the levels of 16α-OHE in plasma increased during the menstrual cycle ( rs = 0.14, P = 0.03). There was no consistent pattern between day of menstrual cycle and the levels of 16α-OHE among oral contraceptive users. The results remained essentially the same after adjustment for age and ethnicity.

Oral contraceptives

Women who were using oral contraceptives had a much lower median 2-OHE/16α-OHE ratio in plasma than pill non-users (0.39 versus 0.69, P = 10 −21 ). This association remained significant after adjustment for age, ethnicity and menstrual cycle day. However, the difference was only significant among white ( P = 4 × 10 −19 ) and Indian/Pakistani women ( P = 0.003), and remained significant in these two groups after adjustment for age and menstrual cycle day ( Figure 5A ). Oral contraceptive users also had significantly lower levels of 2-OHE than pill non-users ( P = 4 × 10 −30 ); the difference remained significant after adjustment for age, ethnicity and menstrual cycle day. Lower 2-OHE levels among oral contraceptive users were observed in all four ethnic groups and the differences were significant among white ( P = 4 × 10 −26 ), black ( P = 0.02) and Indian/Pakistani women ( P = 0.01) ( Figure 5B ). After adjustment for age and menstrual cycle day, 2-OHE levels remained significantly lower among white women. The levels of 16α-OHE in plasma were suppressed by oral contraceptive use among white women ( P = 2 × 10 −6 ), but not among Asian, black or Indian/Pakistani women. After adjustment for age and menstrual cycle day, 16α-OHE levels remained significantly lower among white women ( Figure 5C ). The 2-OHE/16α-OHE ratio in plasma was not significantly associated with the dose of 17α-ethinyl estradiol in the oral contraceptives, or with the type of progestin.

Fig. 5.

( A ) The relationship between the 2-OHE/16α-OHE ratio in women in relation to oral contraceptive use among women of different ethnic background. The 2-OHE/16α-OHE ratio was significantly lower among white ( P = 4 × 10 −19 ) and Indian/Pakistani oral contraceptive users ( P = 0.003) compared with pill non-users of the same ethnic background, and remained significant in these two groups after adjustment for age and menstrual cycle day. ( B ) The relationship between the levels of 2-OHE and oral contraceptive use among women of different ethnic background. Lower 2-OHE levels among oral contraceptive users were observed in all four ethnic groups and was significant among white ( P = 4 × 10 −26 ), black ( P = 0.02) and Indian/Pakistani women ( P = 0.01). After adjustment for age and menstrual cycle day, 2-OHE levels remained significantly lower among white women. ( C ) The relationship between the levels of 16α-OHE and oral contraceptive use among women of different ethnic background. The levels of 16α-OHE in plasma were suppressed by oral contraceptive use among white women ( P = 2 × 10 −6 ), but not among Asian, black or Indian/Pakistani women, and remained significant in white women after adjustment for age and menstrual cycle day.

Fig. 5.

( A ) The relationship between the 2-OHE/16α-OHE ratio in women in relation to oral contraceptive use among women of different ethnic background. The 2-OHE/16α-OHE ratio was significantly lower among white ( P = 4 × 10 −19 ) and Indian/Pakistani oral contraceptive users ( P = 0.003) compared with pill non-users of the same ethnic background, and remained significant in these two groups after adjustment for age and menstrual cycle day. ( B ) The relationship between the levels of 2-OHE and oral contraceptive use among women of different ethnic background. Lower 2-OHE levels among oral contraceptive users were observed in all four ethnic groups and was significant among white ( P = 4 × 10 −26 ), black ( P = 0.02) and Indian/Pakistani women ( P = 0.01). After adjustment for age and menstrual cycle day, 2-OHE levels remained significantly lower among white women. ( C ) The relationship between the levels of 16α-OHE and oral contraceptive use among women of different ethnic background. The levels of 16α-OHE in plasma were suppressed by oral contraceptive use among white women ( P = 2 × 10 −6 ), but not among Asian, black or Indian/Pakistani women, and remained significant in white women after adjustment for age and menstrual cycle day.

Smoking

Among women who were not using oral contraceptives, current smokers were found to have higher plasma 2-OHE/16α-OHE ratios than non-smokers (0.81 versus 0.68, P = 0.06) ( Figure 6A ). After adjustment for age, ethnicity and menstrual cycle day, this relationship was not significant ( P = 0.76). The number of cigarette packs smoked per week had no effect on the plasma 2-OHE/16α-OHE ratio. Among oral contraceptive users, smoking had no effect on the 2-OHE/16α-OHE ratio ( P = 0.72). No significant effect of smoking was seen on the plasma levels of 2-OHE ( Figure 6B ) or 16α-OHE ( Figure 6C ).

Fig. 6.

( A ) The relationship between 2-OHE/16α-OHE ratio and smoking. Among women who were not using oral contraceptives, smoking was associated with a non-significantly higher plasma 2-OHE/16α-OHE ratio ( P = 0.06). Smoking had no effect on the plasma ratio of 2-OHE to 16α-OHE among oral contraceptive users ( P = 0.72). This was also true after adjustment for age and menstrual cycle day. ( B ) The relationship between 2-OHE levels and smoking. No significant effect of smoking was seen on the plasma levels of 2-OHE. This was also true after adjustment for age and menstrual cycle day. ( C ) The relationship between the 16α-OHE levels and smoking. No significant difference was observed between non-smokers and smokers. This was also true after adjustment for age and menstrual cycle day.

Fig. 6.

( A ) The relationship between 2-OHE/16α-OHE ratio and smoking. Among women who were not using oral contraceptives, smoking was associated with a non-significantly higher plasma 2-OHE/16α-OHE ratio ( P = 0.06). Smoking had no effect on the plasma ratio of 2-OHE to 16α-OHE among oral contraceptive users ( P = 0.72). This was also true after adjustment for age and menstrual cycle day. ( B ) The relationship between 2-OHE levels and smoking. No significant effect of smoking was seen on the plasma levels of 2-OHE. This was also true after adjustment for age and menstrual cycle day. ( C ) The relationship between the 16α-OHE levels and smoking. No significant difference was observed between non-smokers and smokers. This was also true after adjustment for age and menstrual cycle day.

Diet

A vegetarian diet was not associated with the plasma 2-OHE/16α-OHE ratio or with the plasma levels of 2-OHE or 16α-OHE. However, a higher number of daily servings of fruits and vegetables correlated with low plasma levels of 16α-OHE ( rs = −0.14, P = 0.02) among pill non-users, but not with the 2-OHE/16α-OHE ratio ( P = 0.97) or the plasma levels of 2-OHE ( P = 0.29). After adjustment for age, ethnicity and menstrual cycle day, the relationship between the plasma levels of 16α-OHE and fruit and vegetable consumption was no longer significant. Among oral contraceptive users, no significant effect of fruit and vegetable consumption was seen.

Coffee

Daily coffee consumption was positively correlated with the plasma 2-OHE/16α-OHE ratio among women not taking oral contraceptives ( rs = 0.18, P = 0.002), but not among oral contraceptive users ( rs = −0.01, P = 0.83) ( Figure 7A ). Among pill non-users, coffee intake did not affect plasma levels of 2-OHE ( rs = 0.08, P = 0.18) ( Figure 7B ), but was correlated with plasma levels of 16α-OHE ( rs = −0.16, P = 0.007) ( Figure 7C ). After adjustment for age, ethnicity and menstrual cycle day, high coffee consumption was significantly associated with high plasma 2-OHE/16α-OHE ratios ( P = 0.003) and high levels of 2-OHE ( P = 0.03) and non-significantly with lower levels of 16α-OHE ( P = 0.11).

Fig. 7.

( A ) The relationship between daily coffee consumption and the plasma ratio of 2-OHE to 16α-OHE. Daily coffee consumption was positively correlated with the plasma ratio of 2-OHE to 16α-OHE among women not taking oral contraceptives ( rs = 0.18, P = 0.002), but not among oral contraceptive users ( rs = −0.01, P = 0.83). After adjustment for age, ethnicity and menstrual cycle day a high coffee consumption remained significantly associated with high plasma 2-OHE/16α-OHE ratios ( P = 0.003) among pill non-users. ( B ) The relationship between the plasma levels of 2-OHE and coffee consumption. Coffee consumption did not significantly affect the plasma levels of 2-OHE among pill non-users or among oral contraceptive users. After adjustment for age, ethnicity and menstrual cycle day a high coffee consumption remained significantly associated with high levels of 2-OHE ( P = 0.03) among pill non-users. ( C ) The relationship between coffee consumption and plasma levels of 16α-OHE. A high daily coffee consumption was correlated with suppressed plasma levels of 16α-OHE ( rs = −0.16, P = 0.007) among pill non-users, but not among oral contraceptive users. After adjustment for age, ethnicity and menstrual cycle day a high coffee consumption was only non-significantly associated with lower levels of 16α-OHE ( P = 0.11) among pill non-users.

Fig. 7.

( A ) The relationship between daily coffee consumption and the plasma ratio of 2-OHE to 16α-OHE. Daily coffee consumption was positively correlated with the plasma ratio of 2-OHE to 16α-OHE among women not taking oral contraceptives ( rs = 0.18, P = 0.002), but not among oral contraceptive users ( rs = −0.01, P = 0.83). After adjustment for age, ethnicity and menstrual cycle day a high coffee consumption remained significantly associated with high plasma 2-OHE/16α-OHE ratios ( P = 0.003) among pill non-users. ( B ) The relationship between the plasma levels of 2-OHE and coffee consumption. Coffee consumption did not significantly affect the plasma levels of 2-OHE among pill non-users or among oral contraceptive users. After adjustment for age, ethnicity and menstrual cycle day a high coffee consumption remained significantly associated with high levels of 2-OHE ( P = 0.03) among pill non-users. ( C ) The relationship between coffee consumption and plasma levels of 16α-OHE. A high daily coffee consumption was correlated with suppressed plasma levels of 16α-OHE ( rs = −0.16, P = 0.007) among pill non-users, but not among oral contraceptive users. After adjustment for age, ethnicity and menstrual cycle day a high coffee consumption was only non-significantly associated with lower levels of 16α-OHE ( P = 0.11) among pill non-users.

Alcohol consumption

Among pill non-users, women who drank alcohol had a higher plasma 2-OHE/16α-OHE ratio than those who did not ( P = 0.046), but this difference was not significant after adjustment for age, ethnicity and menstrual cycle day. There was no consistent correlation between the number of alcoholic drinks per week and plasma 2-OHE/16α-OHE ratio or with the plasma levels of 2-OHE or 16α-OHE among pill non-users. No effect was seen among oral contraceptive users.

Menarche

Among women not using oral contraceptives, age at menarche was not correlated with the 2-OHE/16α-OHE ratio or 16α-OHE. Age at menarche was positively correlated with the plasma level of 2-OHE ( rs = 0.11, P = 0.06), but not after adjustment for age, ethnicity and menstrual cycle day. Age at menarche was not associated with either the 2-OHE/16α-OHE ratio or the plasma levels of 2-OHE or 16α-OHE among oral contraceptive users.

BMI, height and weight

There was no correlation between BMI (kg/m 2 ) and the 2-OHE/16α-OHE ratio ( Figure 8A ) or between BMI and the plasma levels of 2-OHE ( Figure 8B ) and 16α-OHE either among users or non-users of oral contraceptives ( Figure 8C ). There was also no correlation between height or weight and the 2-OHE/16α-OHE ratio or between height or weight and the plasma levels of 2-OHE and 16α-OHE either among users or non-users of oral contraceptives.

Fig. 8.

( A ) There was no relationship between the 2-OHE/16α-OHE ratio and BMI, as divided in quartiles. This was also true after adjustment for age and menstrual cycle day. ( B ) There was no relationship between the levels of 2-OHE and quartile of BMI. This was also true after adjustment for age and menstrual cycle day. ( C ) There was no relationship between the levels of 16α-OHE and quartile of BMI. This was also true after adjustment for age and menstrual cycle day.

Fig. 8.

( A ) There was no relationship between the 2-OHE/16α-OHE ratio and BMI, as divided in quartiles. This was also true after adjustment for age and menstrual cycle day. ( B ) There was no relationship between the levels of 2-OHE and quartile of BMI. This was also true after adjustment for age and menstrual cycle day. ( C ) There was no relationship between the levels of 16α-OHE and quartile of BMI. This was also true after adjustment for age and menstrual cycle day.

Family history of breast cancer

Neither the plasma 2-OHE/16α-OHE ratio, the plasma level of 2-OHE nor the plasma level of 16α-OHE were associated with a history of breast cancer among either first or second degree relatives. This was true both among pill non-users and among oral contraceptive users.

Multivariate models

Overall, the strongest predictor of the 2-OHE/16α-OHE ratio in this study was oral contraceptive use. We created a stepwise model which initially included age, ethnicity, BMI, coffee consumption, current smoking and number of servings of fruits and vegetables. Among pill non-users, variables that predicted the 2-OHE/16α-OHE ratio included coffee consumption ( P = 0.004) and Asian ethnicity ( P = 0.015), and a higher BMI was associated with a lower 2-OHE/16α-OHE ratio ( P = 0.049). After inclusion of menstrual cycle day in the model, menstrual cycle day, coffee consumption and age were significantly associated with the ratio. Among oral contraceptive users, the only predictors of the 2-OHE/16α-OHE ratio were black ( P = 0.0002) and Asian ethnicity ( P = 0.04). These remained significant when menstrual cycle day was included in the model.

Discussion

Our study is the first to apply the technique of measuring the 2-OHE/16α-OHE ratio in plasma in a large sample of young women. Current oral contraceptive use was the single most important predictor of the plasma 2-OHE/16α-OHE ratio. Previously, a difference in the concentration of urinary estrogen metabolites was observed between users of oral contraceptives and pill non-users in a study of 36 women ( 25 ). A second small study reported that oral administration of the synthetic progestin norethisterone acetate, in combination with estradiol, was associated with a decrease in the urinary ratio of 2-OHE/16α-OHE in post-menopasual women ( 26 ).

We chose to study the 2-OHE/16α-OHE ratio in young nulliparous women because breast tissue is vulnerable to cancer induction before the first full-term pregnancy, when the cells are proliferating at a high rate but are poorly differentiated ( 27 , 28 ). Although breast cancer may not be clinically detectable until decades later, tumor initiation may take place in the terminal duct lobular units before the first full-term pregnancy. Therefore, factors that affect estrogen metabolism prior to reproduction may have an impact on the development of breast cancer later in life. Because all the women in our study were nulliparous we could not analyze the effect of parity or lactation on the 2-OHE/16α-OHE ratio.

Most ( 7 , 1214 , 19 ), but not all ( 16 , 17 ), epidemiological studies on estrogen metabolites and breast cancer risk have found that a high level of 2-OHE is protective. In contrast, higher levels of 16α-OHE have been found in the blood and breast tissue of breast cancer patients than of controls ( 29 , 30 ). In cell cultures, 2-OHE appears to act either as a weak estrogen or as an anti-estrogen ( 9 , 31 ). Several studies have found an association between the 2-OHE/16α-OHE ratio and breast cancer risk ( 7 , 1215 , 19 ). These included both prospective and case–control studies. Ursin et al . failed to establish an association, but included only cases that were long-term survivors and excluded cases with severe disease ( 16 , 17 ). The technique of measuring circulating 2-OHE and 16α-OHE levels is new and only one pilot study has been completed using this technique, which reported significantly increased risk of post-menopausal breast cancer in women with low serum 2-OHE/16α-OHE ratios ( 19 ).

Our finding of a lowered 2-OHE/16α-OHE ratio among oral contraceptive users is consistent with reports of an increased risk of breast cancer among young oral contraceptive users in a re-analysis of data from >50 000 women with breast cancer and over 100 000 control women ( 32 ). However, in women aged 35–64 years a recent study found no increased risk of breast cancer among oral contraceptive users ( 33 ). One study reported that the risk of breast cancer associated with oral contraceptive use was highest in women with a first degree relative with breast cancer ( 34 ). In the present study we found no significant effect of a family history of breast cancer on the 2-OHE/16α-OHE ratio in either oral contraceptive users or pill non-users. Kabat et al . and Ursin et al . also reported similar 2-OHE/16α-OHE ratios in women with and without a family history of breast cancer ( 7 , 35 ). In contrast to our findings of similar 2-OHE and 16α-OHE levels in women with and without a family history of breast cancer, Ursin et al . reported lower absolute levels of both metabolites in women with a family history of breast cancer ( 35 ).

We found that endogenous and exogenous hormones exerted opposite effects on the 2-OHE/16α-OHE ratio. Among pill non-users, the 2-OHE/16α-OHE ratio was highest in women whose sample was taken in the luteal phase of the menstrual cycle, when endogenous estrogen and progesterone levels are high. Conversely, among oral contraceptive users the 2-OHE/16α-OHE ratio was lowest among women who had their sample taken towards the end of the 3 week period when active oral contraceptive pills had been consumed. The opposite effects of ethinyl estradiol/progestin and endogenous estrogens may be due to structural differences between 17β-estradiol and 17α-ethinyl estradiol. In post-menopausal women oral conjugated estrogens ( 36 ) increase the 2-OHE/16α-OHE by 2- to 3-fold, whereas oral estradiol in combination with the synthetic progestin norethisterone acetate decreases the urinary 2-OHE/16α-OHE ratio ( 26 ).

2-Hydroxylation of endogenous estradiol is catalyzed primarily by the action of CYP1A1 and CYP1A2 ( 8 , 37 ). 2-Hydroxylation of the synthetic 17α-ethinyl estradiol is carried out by CYP3A4 ( 38 ). CYP3A4 is also responsible for 16α-hydroxylation of endogenous estrogen ( 37 ).

Among oral contraceptive users, Asian and black women had the highest 2-OHE/16α-OHE ratios. Functional CYP1A2 polymorphisms have been described in Asian populations ( 39 ) and a functional CYP3A4 polymorphism is significantly more common among black women than among white women ( 40 , 41 ).

Asian women had the lowest 2-OHE/16α-OHE ratio, although this relationship was no longer significant among women who did not use oral contraceptives after adjustments for age and menstrual cycle day. Among oral contraceptive users, Asian women had significantly lower 2-OHE/16α-OHE ratios than white women, and this result remained after adjustment for menstrual cycle day. Based on the lower risk of breast cancer reported among Asian women compared with white women ( 42 ), this finding was unexpected. However, it still appears that differences in estrogen metabolism have relevance in Asian women. One recent study found significantly reduced 2-OHE/16α-OHE ratios in Chinese breast cancer cases from Shanghai compared with controls from the same area ( 15 ). The risk of breast cancer is higher for Asian women living in North America than for those living in Asia ( 43 ). Thirty-five of the 70 Asian women were born in Asia. Among Asian women who did not use oral contraceptives, there was no significant difference in the median 2-OHE/16β-OHE ratios between Asian women born in Asia (median 0.50) and those who were born in North America, Europe or the West Indies (median 0.69, P = 0.08). The BMI was also lower among Asian women than among white women, another reason why we expected higher 2-OHE/16α-OHE ratios among Asian women. As all the women live in the Toronto area, their current exposure to environmental effects should be similar across the population. The low 2-OHE/16α-OHE ratios observed among Asian women cannot be explained by the lifestyle factors that were evaluated in this study, but might be due to either other lifestyle factors, such as certain dietary constituents, or genetic factors (e.g. a functional polymorphism in the CYP1A2 gene).

Coffee consumption was the second most important lifestyle factor associated with an increased plasma 2-OHE/16α-OHE ratio, mainly due to the effect on lowering the 16α-OHE levels. A prospective study of >14 500 Norwegian women showed a significant negative association between daily intake of coffee and risk of subsequent breast cancer in women with a BMI under 24 kg/m 2 and a weak positive association in women with a higher BMI ( 44 ). In in vivo animal models caffeine has been shown to give rise to a dose-dependent increase in hepatic CYP1A2 expression ( 45 ) and in vitro studies of rat hepatocytes have shown a dose-dependent increase due to caffeine of both CYP1A1 and CYP1A2 ( 46 ). Both CYP1A1 and CYP1A2 are responsible for 2-hydroxylation, therefore, we expected that the high 2-OHE/16α-OHE ratios seen in women with high coffee consumption would have been due to an increase in the 2-OHE levels rather than to the observed decrease in 16α-OHE. No effect of coffee consumption was seen among current oral contraceptive users. Caffeine is mainly metabolized by CYP1A2 and oral contraceptives are known to reduce CYP1A2 activity ( 47 ). Oral contraceptives may therefore counteract the caffeine-induced up-regulation of CYP1A2.

In the present study, smokers who were not using oral contraceptives had slightly higher 2-OHE/16α-OHE ratios than non-smokers, but this effect was not significant after adjustment for other factors. We also did not find any effect on the 2-OHE/16α-OHE ratios depending on the number of cigarette packs per week the women smoked. Michnovicz et al . found increased 2-hydroxylation in smokers and proposed this to be a possible mechanism for the anti-estrogenic effect of smoking ( 48 ). Smoking is associated with reduced risk of estrogen-dependent cancers such as endometrial cancer ( 49 ). Key et al . found that the effect of smoking on urinary excretion of estrone, estradiol and estriol was limited to post-menopausal women ( 50 ). Our finding suggests that if there is any effect of smoking on the 2-OHE/16α-OHE ratio among pre-menopausal women, it is limited to those not using oral contraceptives.

Women reported their use of medications during the 2 months prior to the study visit. We cross-referenced these medications with reported interactions between medication use and estrogen metabolism. Only two women used medications with a suspected effect on estrogen metabolism: St John's Wort ( n = 1) and ranitidine ( n = 1). Exclusion of these two women would not change the results.

In the present study, daily intake of fruits and vegetables was not associated with the 2-OHE/16α-OHE ratio or with plasma levels of 2-OHE. Women who had a higher fruit and vegetable consumption had lower levels of 16α-OHE. Michnovicz et al . found that cruciferous vegetables containing indole-3-carbinol increase 2-hydroxylation ( 51 ), but we could not confirm this finding. We did not specifically ask for the type of vegetables consumed and could not estimate the indole-3-carbinol content.

Obesity has been linked to decreased 2-hydroxylation ( 8 ) and we found a lower 2-OHE/16α-OHE ratio among women with a higher BMI in one multivariate model. However, this effect was not significant after adjustment for other factors, including menstrual cycle day, and is unlikely to play a major role in determining estrogen hydroxylation pathways among pre-menopausal women.

This is the first study where the new technique of measuring the 2-OHE/16α-OHE ratio in plasma has been applied to a large sample of healthy women. The urinary 2-OHE/16α-OHE ratio has been reported to be relatively stable over the menstrual cycle ( 11 ) or to increase slightly just before and after ovulation due to the surge in 2-OHE production ( 52 ). However, this is not true for the plasma 2-OHE/16α-OHE ratio. We observed significant variation in the plasma 2-OHE/16α-OHE ratio depending on the day of the menstrual cycle. It will be important to control for the day of the menstrual cycle if the 2-OHE/16α-OHE ratio is to be used in cohort studies of pre-menopausal women. Among women who did not use oral contraceptives, the 2-OHE/16α-OHE ratio was increased by coffee consumption and by smoking and was lower in women of Asian descent. Among current oral contraceptive users, smoking, coffee consumption or a high fruit and vegetable intake did not influence the 2-OHE/16α-OHE ratio. Thus it appears that the synthetic hormones given in oral contraceptives ‘standardize’ the 2-OHE/16α-OHE ratio and mask the effects of other lifestyle factors. Therefore, studies designed to investigate whether or not certain breast cancer risk factors are mediated through an altered estrogen metabolism should exclude women who use oral contraceptives. Furthermore, our results suggest that epidemiological studies of estrogen metabolites will need to consider ethnic group and that cross-ethnic studies with patients in one group and controls in another may generate misleading results.

In conclusion, we found that the plasma 2-OHE/16α-OHE ratio is associated with both constitutional factors and modifiable lifestyle factors. The reported elevated risk of early onset breast cancer among young oral contraceptive users could be mediated in part through altered estrogen metabolism.

We would like to thank Dr Phillipa Holowaty, Elizabeth Hoodfar, Kelly Metcalf, Jalil Hakimi, Caitlin Springate, Danielle Hanna, Minnie Ho, Patricia de los Rios and Amy Finch for their contributions to this research and Dr Pär-Ola Bendahl for statistical assistance with calculations of geometric means for Table II . This work was supported by MFR (the Swedish Medical Research Council) and the Murray and Isabella Rayburn Fund.

References

1.
Kelsey,J.L., Gammon,M.D. and John,E.M. (
1993
) Reproductive factors and breast cancer.
Epidemiol. Rev.
  ,
15
,
36
–47.
2.
Harris,J.R., Lippman,M.E., Veronesi,U. and Willett,W. (
1993
) Breast cancer.
N. Engl. J. Med.
  ,
327
,
319
–328.
3.
Thomas,H.V., Reeves,G.K. and Key,T.J. (
1997
) Endogenous estrogen and postmenopausal breast cancer: a quantitative review.
Cancer Causes Control
  ,
8
,
922
–928.
4.
The Endogenous Hormones and Breast Cancer Collaborative Group (
2002
) Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies.
J. Natl Cancer Inst.
  ,
94
,
606
–616.
5.
Bernstein,L. and Ross,R.K. (
1993
) Endogenous hormones and breast cancer risk.
Epidemiol. Rev.
  ,
15
,
48
–65.
6.
Key,T.J. and Pike,M.C. (
1988
) The role of oestrogens and progestagens in the epidemiology and prevention of breast cancer.
Eur. J. Cancer Clin. Oncol.
  ,
24
,
29
–43.
7.
Kabat,G.C., Chang,C.J., Sparano,J.A., Sepkovie,D.W., Hu,X.P., Khalil,A., Rosenblatt,R. and Bradlow,H.L. (
1997
) Urinary estrogen metabolites and breast cancer: a case-control study.
Cancer Epidemiol. Biomarkers Prev.
  ,
6
,
505
–509.
8.
Bradlow,L., Telang,N.T. and Osborn,P. (
1996
) Estrogen metabolites as bioreactive modulators of tumor initiators and promotors. In Snyder,R. et al . (eds), Biological Reactive Intermediates . Plenum Press, New York, NY, Vol. V, pp. 285–296.
9.
Schneider,J., Huh,M.M., Bradlow,H.L. and Fishman,J. (
1984
) Antiestrogen action of 2-hydroxyestrone on MCF-7 human breast cancer cells.
J. Biol. Chem.
  ,
259
,
4840
–4845.
10.
Telang,N.T., Suto,A., Wong,G.Y., Osborne,M.P. and Bradlow,H.L. (
1992
) Induction by estrogen metabolite 16 alpha-hydroxyestrone of genotoxic damage and aberrant proliferation in mouse mammary epithelial cells.
J. Natl Cancer Inst.
  ,
84
,
634
–638.
11.
Westerlind,K.C., Gibson,K.J. and Wolfe,P. (
1999
) The effect of diurnal and menstrual cyclicity and menopausal status on estrogen metabolites: implications for disease-risk assessment.
Steroids
  ,
64
,
233
–243.
12.
Ho,G.H., Luo,X.W., Ji,C.Y., Foo,S.C. and Ng,E.H. (
1998
) Urinary 2/16 alpha-hydroxyestrone ratio: correlation with serum insulin-like growth factor binding protein-3 and a potential biomarker of breast cancer risk.
Ann. Acad. Med. Singapore
  ,
27
,
294
–299.
13.
Meilahn,E.N., De Stavola,B., Allen,D.S., Fentiman,I., Bradlow,H.L., Sepkovic,D.W. and Kuller,L.H. (
1998
) Do urinary oestrogen metabolites predict breast cancer? Guernsey III cohort follow-up.
Br. J. Cancer
  ,
78
,
1250
–1255.
14.
Muti,P., Bradlow,H.L., Micheli,A. et al . (
2000
) Estrogen metabolism and risk of breast cancer: a prospective study of the 2:16alpha-hydroxyestrone ratio in premenopausal and postmenopausal women.
Epidemiology
  ,
11
,
635
–640.
15.
Fowke,J.H., Qi,D., Bradlow,H.L., Shu,X.O., Gao,Y.T., Cheng,J.R., Jin,F. and Zheng,W. (
2003
) Urinary estrogen metabolites and breast cancer: differential pattern of risk found with pre- versus post-treatment collection.
Steroids
  ,
68
,
65
–72.
16.
Ursin,G., London,S., Stanczyk,F.Z., Gentzschein,E., Paganini-Hill,A., Ross,R.K. and Pike,M.C. (
1997
) A pilot study of urinary estrogen metabolites (16alpha-OHE1 and 2-OHE1) in postmenopausal women with and without breast cancer.
Environ. Health Perspect.
  ,
105
,
601
–605.
17.
Ursin,G., London,S., Stanczyk,F.Z., Gentzschein,E., Paganini-Hill,A., Ross,R.K. and Pike,M.C. (
1999
) Urinary 2-hydroxyestrone/16alpha-hydroxyestrone ratio and risk of breast cancer in postmenopausal women.
J. Natl Cancer Inst.
  ,
91
,
1067
–1072.
18.
Klug,T.L., Bradlow,H.L. and Sepkovic,D.W. (
1994
) Monoclonal antibody-based enzyme immunoassay for simultaneous quantitation of 2- and 16 alpha-hydroxyestrone in urine.
Steroids
  ,
59
,
648
–655.
19.
Cauley,J.A., Zmuda,J.M., Klug,T., Ljung,B.M., Cummings,S.R. and Kuller,L.H. (
2000
) Estrogen metabolites and breast cancer: a prospective study. In Era of Hope Department of Defense Breast Cancer Research Program Meeting , Atlanta, GA, Vol. II, p. 471.
20.
Fukuda,K., Ohta,T., Oshima,Y., Ohashi,N., Yoshikawa,M. and Yamazoe,Y. (
1997
) Specific CYP3A4 inhibitors in grapefruit juice: furocoumarin dimers as components of drug interaction.
Pharmacogenetics
  ,
7
,
391
–396.
21.
Zheng,W., Jin,F., Dunning,L.A., Shu,X.O., Dai,Q., Wen,W.Q., Gao,Y.T. and Holtzman,J.L. (
2001
) Epidemiological study of urinary 6beta-hydroxycortisol to cortisol ratios and breast cancer risk.
Cancer Epidemiol. Biomarkers Prev.
  ,
10
,
237
–242.
22.
Ziegler,R.G., Rossi,S.C., Fears,T.R. et al . (
1997
) Quantifying estrogen metabolism: an evaluation of the reproducibility and validity of enzyme immunoassays for 2-hydroxyestrone and 16alpha-hydroxyestrone in urine.
Environ. Health Perspect.
  ,
105
,
607
–614.
23.
Bradlow,H.L., Sepkovic,D.W., Klug,T. and Osborne,M.P. (
1998
) Application of an improved ELISA assay to the analysis of urinary estrogen metabolites.
Steroids
  ,
63
,
406
–413.
24.
Falk,R.T., Rossi,S.C., Fears,T.R., Sepkovic,D.W., Migella,A., Adlercreutz,H., Donaldson,J., Bradlow,H.L. and Ziegler,R.G. (
2000
) A new ELISA kit for measuring urinary 2-hydroxyestrone, 16alpha-hydroxyestrone and their ratio: reproducibility, validity and assay performance after freeze–thaw cycling and preservation by boric acid.
Cancer Epidemiol. Biomarkers Prev.
  ,
9
,
81
–87.
25.
Martini,M.C., Dancisak,B.B., Haggans,C.J., Thomas,W. and Slavin,J.L. (
1999
) Effects of soy intake on sex hormone metabolism in premenopausal women.
Nutr. Cancer
  ,
34
,
133
–139.
26.
Seeger,H., Mueck,A.O. and Lippert,T.H. (
2000
) Effect of norethisterone acetate on estrogen metabolism in postmenopausal women [In Process Citation].
Horm. Metab. Res.
  ,
32
,
436
–439.
27.
Preston-Martin,S., Pike,M.C., Ross,R.K., Jones,P.A. and Henderson,B.E. (
1990
) Increased cell division as a cause of human cancer.
Cancer Res.
  ,
50
,
7415
–7421.
28.
Russo,J., Tay,L.K. and Russo,I.H. (
1982
) Differentiation of the mammary gland and susceptibility to carcinogenesis.
Breast Cancer Res. Treat.
  ,
2
,
5
–73.
29.
Schneider,J., Kinne,D., Fracchia,A., Pierce,V., Anderson,K.E., Bradlow,H.L. and Fishman,J. (
1982
) Abnormal oxidative metabolism of estradiol in women with breast cancer.
Proc. Natl Acad. Sci. USA
  ,
79
,
3047
–3051.
30.
Osborne,M.P., Bradlow,H.L., Wong,G.Y. and Telang,N.T. (
1993
) Upregulation of estradiol C16 alpha-hydroxylation in human breast tissue: a potential biomarker of breast cancer risk.
J. Natl Cancer Inst.
  ,
85
,
1917
–1920.
31.
Lottering,M.L., Haag,M. and Seegers,J.C. (
1992
) Effects of 17 beta-estradiol metabolites on cell cycle events in MCF-7 cells.
Cancer Res.
  ,
52
,
5926
–5932.
32.
Collaborative Group on Hormonal Factors in Breast Cancer (
1996
) Breast cancer and hormonal contraceptives: collaborative reanalysis of individual data on 53,297 women with breast cancer and 100,239 women without breast cancer from 54 epidemiological studies.
Lancet
  ,
347
,
1713
–1727.
33.
Marchbanks,P.A., McDonald,J.A., Wilson,H.G. et al . (
2002
) Oral contraceptives and the risk of breast cancer.
N. Engl. J. Med.
  ,
346
,
2025
–2032.
34.
Grabrick,D.M., Hartmann,L.C., Cerhan,J.R. et al . (
2000
) Risk of breast cancer with oral contraceptive use in women with a family history of breast cancer.
J. Am. Med. Assoc.
  ,
284
,
1791
–1798.
35.
Ursin,G., London,S., Yang,D., Tseng,C.C., Pike,M.C., Bernstein,L., Stanczyk,F.Z. and Gentzschein,E. (
2002
) Urinary 2-hydroxyestrone/16alpha-hydroxyestrone ratio and family history of breast cancer in premenopausal women.
Breast Cancer Res. Treat.
  ,
72
,
139
–143.
36.
Kuller,L.H., Matthews,K.A. and Meilhan,E.N. (
2000
) Estrogens and women's health: interrelation of coronary heart disease, breast cancer and osteoporosis.
J. Steroid Biochem. Mol. Biol.
  ,
74
,
297
–309.
37.
Yamazaki,H., Shaw,P.M., Guengerich,F.P. and Shimada,T. (
1998
) Roles of cytochromes P450 1A2 and 3A4 in the oxidation of estradiol and estrone in human liver microsomes.
Chem. Res. Toxicol.
  ,
11
,
659
–665.
38.
Guengerich,F.P. (
1990
) Metabolism of 17 alpha-ethynylestradiol in humans.
Life Sci.
  ,
47
,
1981
–1988.
39.
Nakajima,M., Yokoi,T., Mizutani,M., Kinoshita,M., Funayama,M. and Kamataki,T. (
1999
) Genetic polymorphism in the 5′-flanking region of human CYP1A2 gene: effect on the CYP1A2 inducibility in humans.
J. Biochem. (Tokyo)
  ,
125
,
803
–808.
40.
Jernström,H., Chu,W., Vesprini,D., Tao,Y., Majeed,N., Deal,C., Pollak,M. and Narod,S.A. (
2001
) Genetic factors related to racial variation in plasma levels of insulin-like growth factor-1: implications for pre-menopausal breast cancer risk.
Mol. Genet. Metab.
  ,
72
,
144
–154.
41.
Paris,P.L., Kupelian,P.A., Hall,J.M., Williams,T.L., Levin,H., Klein,E.A., Casey,G. and Witte,J.S. (
1999
) Association between a CYP3A4 genetic variant and clinical presentation in African-American prostate cancer patients.
Cancer Epidemiol. Biomarkers Prev.
  ,
8
,
901
–905.
42.
International Agency for Research on Cancer (
1997
) Cancer Incidence in Five Continents . IARC Scientific Publications, vol. VII no. 143, i–xxxiv, 1–1240. IARC, Lyon.
43.
Ziegler,R.G., Hoover,R.N., Pike,M.C. et al . (
1993
) Migration patterns and breast cancer risk in Asian-American women.
J. Natl Cancer Inst.
  ,
85
,
1819
–1827.
44.
Vatten,L.J., Solvoll,K. and Loken,E.B. (
1990
) Coffee consumption and the risk of breast cancer. A prospective study of 14,593 Norwegian women.
Br. J. Cancer
  ,
62
,
267
–270.
45.
Ayalogu,E.O., Snelling,J., Lewis,D.F., Talwar,S., Clifford,M.N. and Ioannides,C. (
1995
) Induction of hepatic CYP1A2 by the oral administration of caffeine to rats: lack of association with the Ah locus.
Biochim. Biophys. Acta
  ,
1272
,
89
–94.
46.
Goasduff,T., Dreano,Y., Guillois,B., Menez,J.F. and Berthou,F. (
1996
) Induction of liver and kidney CYP1A1/1A2 by caffeine in rat.
Biochem. Pharmacol.
  ,
52
,
1915
–1919.
47.
Rietveld,E.C., Broekman,M.M., Houben,J.J., Eskes,T.K. and van Rossum,J.M. (
1984
) Rapid onset of an increase in caffeine residence time in young women due to oral contraceptive steroids.
Eur. J. Clin. Pharmacol.
  ,
26
,
371
–373.
48.
Michnovicz,J.J., Hershcopf,R.J., Naganuma,H., Bradlow,H.L. and Fishman,J. (
1986
) Increased 2-hydroxylation of estradiol as a possible mechanism for the anti-estrogenic effect of cigarette smoking.
N. Engl. J. Med.
  ,
315
,
1305
–1309.
49.
Baron,J.A., La Vecchia,C. and Levi,F. (
1990
) The antiestrogenic effect of cigarette smoking in women.
Am. J. Obstet. Gynecol.
  ,
162
,
502
–514.
50.
Key,T.J., Pike,M.C., Brown,J.B., Hermon,C., Allen,D.S. and Wang,D.Y. (
1996
) Cigarette smoking and urinary oestrogen excretion in premenopausal and post-menopausal women.
Br. J. Cancer
  ,
74
,
1313
–1316.
51.
Michnovicz,J.J., Adlercreutz,H. and Bradlow,H.L. (
1997
) Changes in levels of urinary estrogen metabolites after oral indole-3-carbinol treatment in humans.
J. Natl Cancer Inst.
  ,
89
,
718
–723.
52.
Lu,L.J., Cree,M., Josyula,S., Nagamani,M., Grady,J.J. and Anderson,K.E. (
2000
) Increased urinary excretion of 2-hydroxyestrone but not 16alpha-hydroxyestrone in premenopausal women during a soya diet containing isoflavones.
Cancer Res.
  ,
60
,
1299
–1305.