• © 2006 by American Society of Clinical Oncology

Have We Found the Ultimate Risk Factor for Breast Cancer?

The accepted risk factors for breast cancer (eg, ages at menarche, first live birth, menopause, and alcohol consumption) can be thought of as measures of the cumulative exposure of the breast epithelium to estrogenic substances.¹ In pregnancy, prolactin and estradiol levels fall and sex hormone–binding globulin levels (SHBG) rise, and some premalignant cells terminally differentiate and lose their malignancy potential. How these exposures mediate the risk of breast cancer remains somewhat obscure, however.

The worldwide data from prospective studies of the relationship between the levels of endogenous sex hormones and breast cancer risk in postmenopausal women also show multiple and complex relationships. Nine prospective studies of women not taking exogenous sex hormones when their blood was collected to determine hormone levels showed that the risk for breast cancer increased significantly with increasing concentrations of all sex hormones examined: total estradiol, free estradiol, non-SHBG–bound estradiol, estrone, estrone sulfate, androstenedione, dehydroepiandrosterone, dehydroepiandrosterone sulfate, and testosterone.² The relative risks (RRs) for women with increasing quintiles of all hormone concentrations, relative to the lowest quintile, ranged from 1.04 to 2.58, and high SHBG was associated with a decrease in breast cancer risk. Interestingly, estradiol levels are generally higher in North American women than Asian women, and urinary estrogens are higher in North American teenagers who face higher lifetime risks for breast cancer.

In an earlier report from the Nurses' Health Study of 11,169 postmenopausal women not using hormone-replacement therapy (HRT) at blood collection,³ there was a statistically significant positive association with risk of breast cancer for circulating levels of estradiol, estrone, estrone sulfate, and dehydroepiandrosterone sulfate. There were no substantial associations found, however, with percent free or percent bioavailable estradiol, androstenedione, testosterone, or dehydroepiandrosterone.

In this issue of the Journal of Clinical Oncology, Eliassen et al⁴ test the hypothesis that the risk of breast cancer, quantitatively assessed using either of two multivariate risk models, is associated with measured levels of circulating endogenous estrogen and related compounds in postmenopausal women. Their findings strongly support the conclusion that the traditional risk factors for breast cancer mediate their effect through synthesis and/or metabolism pathways of endogenous human estrogen, and give credence to a possible association between hormone levels and tumor receptor status or invasive versus in situ tumor status. Their data are supported, in part, by another study showing a statistically significant association between breast cancer risk and the levels of both estrogens and androgens, but no statistically significant associations between this risk and the level of progesterone or SHBG.⁵ When that analysis was limited to case subjects with estrogen receptor (ER)–positive/progesterone receptor (PR)–positive breast tumors, and compared the highest with the lowest fourths of plasma hormone concentration, there were large increased risks of breast cancer associated with estradiol (RR = 3.3), testosterone (RR = 2.0), androstenedione (RR = 2.5), and dehydroepiandrosterone sulfate (RR = 2.3), and all hormones tended to be associated most strongly with in situ disease.

In agreement with these observations, earlier studies among women who were at increased risk for osteoporotic fractures who had traditional risk factors for breast cancer showed that the risk for breast cancer was more than three times greater among women with the highest concentrations of bioavailable estradiol compared with women with the lowest concentration.⁶

Women with the highest estradiol levels (≥ 12 pmol/L) in the Multiple Outcomes of Raloxifene Evaluation (MORE) trial had a twofold increased risk for invasive breast cancer compared with women with lower levels of estradiol.⁷ In the placebo group, women with estradiol levels greater than 10 pmol/L had a 6.8 times higher rate of breast cancer than breast cancer risk of women with undetectable estradiol levels.⁸ Women with estradiol levels greater than 10 pmol/L in the raloxifene group had a rate of breast cancer that was 76% lower than that of women with estradiol levels greater than 10 pmol/L in the placebo group. Importantly, raloxifene reduced breast cancer risk in both the low- and high-estrogen subgroups for all risk factors examined, but the reduction was greatest in those with the highest estradiol levels. In contrast, women with undetectable estradiol levels had similar breast cancer risk whether or not they were treated with raloxifene. Older postmenopausal women whose testosterone levels were in the highest two quintiles had a four-fold increased risk of ER-positive breast cancer.⁹

A recent evaluation among women who were using postmenopausal HRT found that HRT users had statistically significantly higher estradiol, free estradiol, sex hormone–binding globulin, and testosterone, and lower free testosterone concentrations than non-HRT users.¹⁰ Among the HRT users, there were modest associations with breast cancer risk when comparing the highest versus lowest quartiles of free estradiol, free testosterone, and SHBG, but not of estradiol or of testosterone. However, estradiol and free estradiol were significantly and positively associated with breast cancer risk among women older than 60 years (with a relative risk of nearly 3) and among women with a body mass index of less than 25 kg/m². The etiological issues are complex, however, given the findings of the clinical trials of the Women's Health Initiative that showed an increased risk of breast cancer among women taking conjugated equine estrogens plus progesterone,¹¹ but not among those women with hysterectomy who were taking only conjugated equine estrogens.¹²

Consistent with this latter observation, not all studies have shown an association between increased serum levels of estradiol and/or testosterone and breast cancer risk in postmenopausal women. Among women from the placebo group in the National Surgical Adjuvant Breast and Bowel Project Breast Cancer Prevention Trial (BCPT), the median levels of estradiol, testosterone, and SHBG did not differ statistically between patient cases and the entire cohort.¹³ There was a statistically significant direct relationship between estradiol and testosterone and a statistically significant inverse relationship between estradiol and SHBG. There was no significant difference, however, in the relative risk of breast cancer across quartiles of estradiol, testosterone, or SHBG. Even for ER-positive breast cancer, the relative risks across quartiles were not different from 1.0. Therefore, in contrast to prior studies (all in lower-risk populations), sex hormone levels were not associated with breast cancer among the high-risk population in BCPT.

Although the data from BCPT did not suggest that hormone levels can further stratify risk, the data from Eliassen et al⁴ indicate a continuum of serum hormone levels that are directly correlated with the quantitative risk of developing invasive breast cancer. It is likely that there was a plateau in circulating hormone concentrations in BCPT, given that the mean quantitative risk scores among the participants was high (average 5-year Gail risk = 4); the Eliassen et al study suggests a continuum of hormone levels that peaks among those women who are at highest risk. Do these observations suggest that endogenous hormones will better serve to select women at moderate risk for selective estrogen receptor modulators (SERM) or aromatase risk reduction interventions than quantitative risk assessment alone? The data of Eliassen and her colleagues imply that they will.

Polymorphisms in the genes coding for enzymes that metabolize sex steroids may also alter the risk of breast cancer. The CAG repeat polymorphism in exon 1 of the androgen receptor gene, for example, is associated with breast cancer risk among some groups of white and Asian women. Among women with a first-degree family history of breast cancer, longer CAG repeats are associated with a significantly increased risk.¹⁴ Women carrying at least one longer allele have a three-fold increased risk compared with those women with two shorter alleles. In the Multiethnic Cohort Study,¹⁵ a large prospective study of men and predominantly postmenopausal women of Japanese, white, African American, Latino, and Native Hawaiian ancestry, there was no association between breast cancer and polymorphisms in several genes in the estradiol/estrone metabolism pathway, except for CYP1A2*1F, which was inversely associated with risk. This association was stronger for estrogen receptor–and progesterone receptor–negative tumors than for ER/PR-positive tumors, and there was no statistically significant interaction with estrogen-related risk factors. These findings provide no evidence for unique roles for these specific polymorphisms in the etiology of breast cancer but do provide support for an inverse association between CYP1A2*1F and breast cancer that is consistent with the observation of lower circulating estrogen levels in premenopausal women with the CC genotype.¹⁶

The HSD17B1 (17β-hydroxysteroid dehydrogenase type 1) gene produces the enzyme that catalyzes the final step of estradiol biosynthesis, ie, the conversion of estrone to estradiol. Polymorphisms in estrogen-metabolizing genes may potentially cause alterations in their biologic function and thus contribute to the susceptibility of an individual to hormone-related cancers. These polymorphisms may increase the activity of HSD17B1 enzyme, increase estradiol level, and, thus, increase breast cancer risk. The +1954A/A genotype is associated with higher estradiol levels in lean women, and there is a possible interaction between the +1954 genotype with body mass index in postmenopausal breast cancer.¹⁷ Thus, the HSD17B1 may be associated with circulating estradiol levels and interact with body mass index in postmenopausal breast cancer.

In addition, several other genetic polymorphisms may influence sex hormone concentrations, including CYP17, CYP19, and CYP1B1. Associations between these polymorphisms and serum concentrations of estrogens, androgens, and SHBG and urinary concentrations of 2- and 16 α-hydroxyestrone have been studied.¹⁸ Compared with noncarriers, women carrying two CYP19 7r(−3) alleles had 26% lower estrone, 19% lower estradiol, 23% lower free estradiol, and 22% higher SHBG concentrations. Compared with noncarriers, women carrying at least one CYP19 8r allele had 20% higher estrone, 18% higher estradiol, and 21% higher free estradiol concentrations. Women with the COMT Met/Met genotype had 28% higher 2-hydroxyestrone and 31% higher 16 α-hydroxyestrone concentrations, compared with Val/Val women. It is evident, therefore, that genetic variation may appreciably alter sex hormone concentrations in postmenopausal women, but the mechanisms whereby these polymorphisms affect the risk of breast cancer and to what degree is the subject of several ongoing studies.

The role of urinary estrogen metabolites on breast cancer risk has been controversial mainly because of the small sample size of the case-control studies conducted in the past. A recent report from the Long Island Breast Cancer Study¹⁹ showed that the odds ratio for invasive breast cancer was significantly inversely associated with the ratio of 2- and 16-hydroxy estrone metabolites among premenopausal women, whereas the result was less evident in postmenopausal women. In this study, measurement was limited to the two main products of estrogen metabolism, but did not include the 4-hydroxy metabolite that is known to be a carcinogen in animal models. A complete battery of all the estrogen congeners deriving from estrogen metabolites should be carefully analyzed as predictors of future breast cancer. In addition, ethnic differences in urinary estrogen metabolites could explain some of the uncertainty of the studies of estrogen metabolites and breast cancer risk. Women of Asian origin, for example, have consistently higher 2:16 urinary hydroxyestrone ratios than Western women, whereas African American women have lower levels.

Some studies have shown that adrenal androgen synthesis (DHEA sulfate, Δ4-androstenedione, testosterone) is a risk factor for breast cancer,²⁰ but other studies have given negative results.²¹ Prolactin appears to be associated with a modestly increased risk of ER-positive/PR-positive and for ER-positive/PR-negative postmenopausal breast cancer, but less for estrogen receptor–negative/progesterone receptor–negative breast cancer.²²

Additional indirect evidence of the role of endogenous hormones in the etiology of breast cancer comes from the adjuvant therapy trials with aromatase inhibitors. These agents interrupt hormonal synthesis, cause greater than 90% decrease in the levels of circulating estrogens in postmenopausal women, and have shown a 58% reduction in primary contralateral breast cancers as a first event among women treated with anastrozole when compared with women receiving tamoxifen.²³ Similarly, the rate of contralateral breast cancer was half as great among postmenopausal women receiving letrozole as it was among women taking tamoxifen in two adjuvant treatment trials.^24,25

There are 50 million white women aged 35 to 79 years in the United States, and more than 2 million would have a positive benefit/risk index for tamoxifen chemoprevention based on quantitative risk modeling.²⁶ Would this number be even larger if serum hormone levels were incorporated into risk assessment profiles? Could we reduce this number and be more specific in our predictions if we used serum hormonal profiling in our routine breast cancer risk assessments? These questions should be addressed immediately in prospective clinical trials.

The known risk factors for breast cancer account for a small fraction of the population attributable risk, and new risk factors are required. Mammographic breast density may be one such new factor, and estradiol may serve as another for postmenopausal women. The data from Eliassen et al⁴ along with the published literature lead us to ask whether measurement of serum estradiol could stratify subjects for risk reductions with SERMs, aromatase inhibitors, or other hormonally active agents. Can serum estradiol levels allow selection of case subjects for upcoming aromatase inhibitor risk reduction intervention trials?

The data also suggest that family history and its related metabolism gene polymorphisms may drive familial patterns of endogenous estrogen levels. It is also remarkable that androgens predict risk in a completely different way from the traditional risk scores, and these, too, should also be refined as a possible predictive tool for chemoprevention trials. Integration of serum hormone levels into risk stratification models should proceed directly, and validation studies should be incorporated into both ongoing and upcoming breast cancer risk reduction trials.