Do Urinary Estrogen Metabolites Reflect the Differences in Breast Cancer Risk between Singapore Chinese and United States African-American and White Women?

INTRODUCTION

Ovarian hormones play an important role in breast cancer etiology. Elevated serum estrogen levels and increased urinary excretion rates of E₁, 3 E_2, and E₃ have been found in breast cancer cases as compared with controls (1) .

The extent to which E₂ is metabolized via the 16α-hydroxylation pathway may also be associated with breast cancer risk. The two main pathways for metabolizing E₂ are via 16α-hydroxylation and via 2-hydroxylation of E₁. There are some data suggesting that 2-hydroxylated compounds are less biologically active than 16α-hydroxylated compounds (2 , 3) , and that women metabolizing more E₁ through the 16α pathway may have a prolonged estrogenic effect of E₁ (4 , 5) . It has therefore been suggested that the ratio of 2-hydroxylation:16α-hydroxylation is important for breast cancer development. It has also been suggested that the estrogen metabolite ratio of urinary 2-OHE₁:16α-OHE₁ obtained from a single spot urine represents an important biomarker for breast cancer risk; the lower the ratio the higher the risk (6) . The epidemiological studies that have addressed this hypothesis are not entirely supportive (6, 7, 8) , and it has been argued that perhaps some of the discrepancy is attributable to some case-control studies measuring alterations in metabolism that occur after the onset of cancer. We therefore set out to determine whether this estrogen metabolite ratio differs in women known to be at widely different breast cancer risk.

Breast cancer incidence rates are still more than twice as high in United States women as in women living in Asia (9) . Much of this difference can be explained by differences in estrogen levels (10 , 11) between postmenopausal Asian women living in Asia and United States women living in the United States. However, it is unknown whether the average level of urinary 2-OHE₁:16α-OHE₁ is different between these populations. We conducted a study of postmenopausal women participating in ongoing cohort studies in Los Angeles and in Singapore to determine whether there is a difference in this ratio that parallels the different breast cancer risks in these two populations.

MATERIALS AND METHODS

We obtained urine samples from women who had participated in either the Los Angeles part of the Multiethnic Cohort study (12) or in the Singapore Chinese Health Study (13) . This study was approved by the Institutional Review Board at the University of Southern California. The urine samples were collected as part of separate protocols approved by the Institutional Review Board at the University of Southern California and at the National University of Singapore.

Multiethnic Cohort Study.

Participants for the multiethnic cohort were recruited in Hawaii and California from 1993 through 1996. In California, areas with a high population of African Americans and Latinos were targeted. Subjects were selected primarily from drivers’ license files in the two states. Additional African Americans in California were selected from the Health Care Financing Administration files. Self-administered questionnaires were mailed to randomly selected persons to obtain both demographic data and information on several lifestyle factors, including diet, during the previous 12 months. A total of 215,251 male and female subjects between the ages of 45 and 75 years were included in the cohort. As part of a dietary calibration study, biological specimens were collected from 260 subjects in each ethnic/sex group. Urine samples were predominantly first-void samples. Samples were obtained in styrofoam cups, and the interviewer subsequently transferred ∼45 ml to urine containers. Specimens were transported to the laboratory in a cooler containing ice or a frozen coolant, aliquoted, and stored at −20°C. Processing was conducted within 6 h of obtaining the sample.

Singapore Chinese Health Study.

Chinese men and women (n = 63,257), 45 to 74 years of age, who were residents of government housing estates (86% of the Singapore population reside in such facilities) enrolled in the study from April 1993 through December 1998. An in-person interview was conducted in the subject’s home by a trained interviewer using a structured questionnaire that included a validated dietary component soliciting intake over the previous 12 months. The questionnaire also requested demographic information, reproductive history (women only), occupational exposure, medical history, and a brief family history of cancer. Collection of blood and spot urine specimens from a random 3% sample of study enrollees began 1 year after the start of the cohort study.

The current study was based on a random sample of subjects from the Los Angeles part of the Hawaii/Los Angeles Multiethnic Cohort Study (n = 71) and on the first batch of subjects from the Singapore Chinese Health Study (n = 79) who donated urine and who were postmenopausal and currently not using hormone replacement therapy.

Laboratory Analyses

Enzyme Immunoassay of 16α-OHE₁ and 2-OHE₁.

The urinary 16α-OHE₁ and 2-OHE₁ metabolites were measured at the Strang Cornell Cancer Research Laboratory in New York, NY, by Dr. Daniel Sepkovic. Commercially available monoclonal antibody-based enzyme assay kits (Estramet; Immuna Care Corporation, Bethlehem, PA) were used to measure 2-OHE₁ and 16α-OHE₁ directly in urine (14) The monoclonal antibodies have been tested against other estrogen and androgen metabolites with similar chemical structures and found to show high affinity and specificity for 2-OHE₁ and 16α-OHE₁ (14) . Results from the 2-OHE₁ and 16α-OHE₁ assays have been found to correlate highly (correlation coefficients for the comparisons were 0.947 and 0.936, respectively) with corresponding results from gas chromatography-mass spectroscopy (15 , 16) . We used the revised version of the assay (16) . The intra-assay coefficients of variation estimated from duplicate measures of all samples were 4.8% for 2-OHE₁ and 4.5% for 16α-OHE₁. Between-assay CV estimated from 4 samples measured once or twice in two to six batches of samples were 20.2% for 2-OHE_1, 31.3% for 16α-OHE_1, 15.6% for the 2-OHE₁:16α-OHE₁ ratio, and 13.3% for creatinine. The inter-assay CVs for 2-OHE₁ and 16α-OHE₁ were somewhat higher than what has been reported previously (10% and 17%, respectively; Ref. 16 ); however, this was driven by one measurement, without which the CVs were 12% and 22%, respectively. The CV for the ratio was identical to what has been reported previously in postmenopausal women.

We submitted all of the samples in a total of 7 batches, each batch containing the same proportion of Asian and white/African American women. Because of the known problems with errors over time, the samples were run closely together in time, with 90% of the samples run during a one month period.

RIA of Urinary E₁, E₂, and E₃.

The urinary E₁, E₂, and E₃ metabolites were measured in the laboratory of Dr. Frank Z. Stanczyk at the Los Angeles County/University of Southern California Medical Center using radioimmunoassays with preceding high-performance liquid chromatography and radioimmunoassays (8 , 17) . Each sample of urine was acidified and subjected to β-glucuronidase/aryl sulfatase hydrolysis before being subjected to chromatography and separation of E₁, E₂, and E₃. Quality control samples were included with each set of samples assayed. The coefficients of variation (estimated from one sample estimated six times in each batch) were 17.8% for E₁, 17.1% for E₂, and 15.7%for E₃.

Statistical Analyses.

Within the United States population, African Americans and whites have a similar 2-OHE₁:16α-OHE₁ ratio after weight has been taken into consideration (18) . Although African Americans have slightly lower breast cancer incidence rates than whites (19) , their rates are still twice as high as those of Chinese women in Singapore (9) . We therefore combined the data from African Americans and whites in the analyses.

We adjusted all of the hormone metabolites for creatinine excretion, logarithmically transformed the values, and evaluated the statistical significance of the difference in these variables between women from the United States (African Americans/whites) and Chinese women from Singapore using analysis of covariance (20) . We adjusted for age and age at menopause (<40, 40–44, 45–49, 50–54, or 55+) in the main analyses. In additional analyses, we also adjusted for a number of variables that are known to affect breast cancer risk and which may be associated with hormone levels: age, BMI, age at menarche (≤12, 13–14, 15+), and parity (0–1, 2–3, 4+ children). Post-menopausal status was defined as an affirmative response to the question: “Have your menstrual cycles stopped permanently?” There were 4 women from the United States and 3 women from Singapore who had undergone simple hysterectomy and were under 55 years of age. Exclusion of these subjects did not materially change any of the study results. They were therefore included in all results presented in this report. Five African American women had unknown age at menopause, and 2 African American, 1 white, 2 with mixed ethnicity, and 14 Singapore Chinese women had missing BMIs. To keep these women in the adjusted analyses, we replaced the missing values with the median age at menopause or the BMI level for each ethnic group (Singapore Chinese, African-American, or white/other). All Ps are from two-sided tests. The levels of 2-OHE₁ and 16α-OHE₁ were below detection levels in 12 women from the United States and in 12 women from Singapore. These women were excluded from all analyses. We also excluded one United States woman with missing information on parity. The final study sample consisted of 58 United States women and 67 Singapore Chinese women. E₁, E₂, or E₃ levels were below detection levels for some women; these women were included in all analyses in which they had known values.

RESULTS

The United States women were predominantly African American (n = 44); 7 were white, and the rest (n = 7) were mixed African American/white or African American/other. The characteristics of the women in this study are displayed in Table 1 ⇓ . Singapore Chinese women were substantially lighter, somewhat younger, less educated, had a later menarche and menopause, and were less likely to have a family history of breast cancer than United States women.

Table 2 ⇓ shows the geometric mean values of urinary estrogens in United States and Singapore Chinese women. Unadjusted mean levels of all estrogen metabolites were higher in United States than in Singapore women. Adjusted mean levels of E₁, E₂, and E₃ were 162% (P < 0.0001), 152% (P < 0.0001), and 92% (P = 0.0009) higher, respectively, in United States than in Singapore Chinese women. However, mean 2-OHE₁ was only 23% lower in United States women (P = 0.03), and there were no statistically significant differences in 16α-OHE₁ or in the ratio of 2-OHE₁:16α-OHE₁ between the two groups.

Additional adjustment for BMI, parity, and age at menarche completely obliterated any difference between the two groups with respect to 2-OHE₁, whereas E₁, E₂, and E₃ were still 167% (P = 0.0001), 136% (P = 0.0005), and 44% (P = 0.12) greater in United States women than in Singapore Chinese. Similar results were obtained when we adjusted for body weight. The results were also similar after adjustments for intake of tofu or cruciferous vegetables over the past year (data not shown). The results were essentially unchanged after excluding women who were current smokers, who were <55 years of age and had undergone a simple hysterectomy, or who reported drinking >3 alcoholic drinks/week over the past year prior to urine collection (data not shown). There were too few white women to determine differences in estrogen metabolism between the whites and the African Americans. However, excluding the whites from the analyses yielded very similar results.

DISCUSSION

Urinary levels of E₁, E₂, and E₃ were all clearly higher in United States women than in Singapore Chinese. Although there was a statistically significant difference in 2-OHE₁ between the two ethnic groups, this difference was much smaller than the corresponding differences for E₁, E₂, and E₃. Also, the ratio of urinary 2-OHE₁:16α-OHE₁ was not significantly different between the two groups.

It is theoretically possible that the low levels of estrogens in this sample of postmenopausal women resulted in substantial misclassification of 2-OHE₁ and/or 16α-OHE₁ so that we were unable to find a difference in the 2-OHE₁:16α-OHE₁ ratio between the United States and the Singapore Chinese women. However, we think this is very unlikely. We used a revised version of the 2-OHE₁:16α-OHE₁ immunoassay that has been found to yield results that correlate highly with results obtained with gas chromatography-mass spectroscopy in postmenopausal women (15 , 16) , and the CV were reasonable and similar to those that have been reported previously.

The samples were stored at −20°C, and not at −80°C; however, it is unlikely that this affected our results. First, we observed the expected differences in E₁, E₂, and E₃ between the two ethnic groups. Second, freeze-thaw cycles (including storage at −30°C) has not been found to affect the 2-OHE₁ or 16α-OHE₁ measurements (16 , 21) . Furthermore, our absolute values of estrogen metabolites were very similar to what we have observed in data from a previous study in Los Angeles where samples were stored at −80°C (8) .

Our results are consistent with previous literature on the differences in estrogen levels between postmenopausal women of different ethnicities (10 , 11 , 22) . Goldin et al. (22) found that urinary E₁, E₂, E₃, and plasma E₂ were significantly higher (2–4 times higher) among United States whites than among recent Asian immigrants to Hawaii. Shimizu et al. (10) found that United States whites living in Southern California had significantly higher serum E₁ and E₂ levels than women living in rural areas in Japan, even after adjusting for body weight. Key et al. (11) found that, among women 55–64 years of age, plasma levels of E₂ were almost three times as high in British women as in Chinese women living in rural China. Only the study of British, Japanese, and Hawaiian-Japanese women by Hayward et al. (23) failed to find a difference in urinary levels of E₁, E₂, or E₃ between the three ethnic groups; the reason for this is unknown.

The only other study that has compared 2- and 16α-hydroxylated estrogen metabolites in Asians and non-Asians was conducted by Adlercreutz et al. (24) , who compared 13 Asian premenopausal women (mostly Vietnamese) living in Hawaii with 12 omnivorous Finnish premenopausal women from a previous study. They found that, compared with the Asian women, the Finnish women had a higher extent of 2-hydroxylation, but a similar extent of 16α-hydroxylation. The ratio of 2-:16α-hydroxylation was four to five times higher among the Finnish women than among the Asian women.

The hypothesis that the ratio of 2-hydroxylation:16α-hydroxylation is important for breast cancer development (4 , 5) is supported by some, but not all, experimental evidence. The 2-hydroxylated compounds bind more weakly to the estrogen receptor and result in lower estrogenic effects than do the 16α-hydroxylated compounds (2 , 3) . There are experimental data showing an antiestrogen role of 2-hydroxylated compounds (25) , DNA adduct formation with 16α-hydroxylated compounds (26) , and higher levels of 16α-hydroxylated compounds in breast cancer tissue than in noncancerous tissue (27 , 28) . However, others have suggested that 2-hydroxylated estrogens, like the other catechol estrogens, the 4-hydroxylated compounds, can cause oxidative DNA damage (29) .

Results from epidemiological studies on the association between 2- and 16α- hydroxylation and breast cancer are mixed. Several (6 , 18 , 30, 31, 32) but not all (33) non-population-based studies found an increased risk of breast cancer associated with lower 2-OHE₁:16α-OHE₁ ratio. Thus far, there are two population-based samples addressing this question. In a study of 66 postmenopausal breast cancer cases and 76 control patients, we found no increased risk with a lower ratio of 2-OHE₁:16α-OHE₁ (8) . Meilahn et al. (7) reported a median of 1.6 for the ratio 2-OHE₁:16α-OHE₁ in 42 postmenopausal patients and 1.7 in 139 matched control subjects from the Guernsey III cohort follow-up. Compared with postmenopausal women in the lowest tertile category of 2-OHE₁:16α-OHE₁, women in the highest tertile had an odds ratio for breast cancer of 0.71, but the CI was wide and not statistically significant (95% CI, 0.29–1.75).

Apart from the last study, the other studies measured metabolites after breast cancer diagnosis. Thus, there is the possibility that the results obtained in case-control studies may have been affected by the cancer or the cancer treatment, especially in studies that have included recently diagnosed or treated breast cancer cases.

We only examined the ratio between two metabolites in the 2-hydroxylated and 16α-hydroxylated estrogen pathways. There is increasing experimental evidence that the 4-hydroxylated estrogen pathway may play a major role (29) . However, there is no readily available assay to determine 4-hydroxylated products.

A number of genetic and environmental variables seem to modify the role of these hydroxylation pathways. Smoking (34) and the intake of soy (35) up-regulate 2-hydroxylation. Also, indole-3 carbinol, which is found in cruciferous vegetables, up-regulates the 2-hydroxylation pathway and down-regulates the 4-hydroxylation pathway (36 , 37) . However, in our study, adjustment for soy consumption and cruciferous vegetable intake did not alter the results.

Adjustment for weight or BMI completely obliterated any difference in 2-OHE₁ between the two groups of women, whereas the differences in E₁ and E₂ were still statistically significant. Previous studies have reported that weight correlated inversely with 2-hydroxylation (18 , 38) , but had no effect (38) or was positively correlated (39) with 16α-hydroxylation. To what extent lack of adequate body weight or body mass adjustment may have caused the apparent protective association between the 2-OHE₁:16α-OHE₁ ratio and breast cancer in previous studies is unclear. In our previous case-control study of breast cancer (8) , we restricted participants to women weighing under 200 pounds and found no association between estrogen metabolites and breast cancer risk. The confounding effects of body weight or mass need to be evaluated carefully in future studies of estrogen metabolites and breast cancer risk.