|

 

Ai049900270p

Elevated Serum Estradiol and Testosterone Concentrations Are
Associated with a High Risk for Breast Cancer
Jane A. Cauley, DrPH; Frances L. Lucas, PhD; Lewis H. Kuller, MD, DrPH; Katie Stone, PhD;
Warren Browner, MD, MPH; and Steven R. Cummings, MD, for the Study of Osteoporotic Fractures Research Group Background: The relation between endogenous steroid
One in eight women in the United States will hormones and risk for breast cancer is uncertain. Measure- develop breast cancer, and deaths from breast ment of sex hormone levels may identify women at high cancer account for 17% of all cancer deaths in risk for breast cancer who should consider preventive ther- women in the United States (1, 2). In 1997, more than 180 000 new cases of breast cancer occurred in Objective: To test the hypothesis that serum concentra-
women in the United States (2); about half oc- tions of estradiol and testosterone predict risk for breast curred in women 65 years of age or older. About 1 in 14 women aged 60 to 79 years will develop breast Design: Prospective case– cohort study.
cancer compared with 1 in 26 women aged 40 to 59 Setting: Four clinical centers in the United States.
Endogenous estrogens may play an important Participants: 97 women with confirmed incident breast
role in the development of breast cancer (3). Some cancer and 244 randomly selected controls; all women (4–8) but not all (9–12) prospective studies have were white, 65 years of age or older, and were not receiv- found statistically significant positive associations ing estrogen.
between endogenous concentrations of estrogens Measurements: Sex-steroid hormone concentrations
and subsequent risk for breast cancer. Two recent were assayed by using serum that was collected at baseline reviews concluded that increasing evidence supports and stored at 2190 °C. Risk factors for breast cancer were a relation between estrogen concentrations and risk ascertained by questionnaire. Incident cases of breast can-cer were confirmed by review of medical records during an for breast cancer (3, 13). Women with higher bone average period of 3.2 years.
mineral density, which is a cumulative measure ofendogenous estrogen, have an increased risk for Results: The relative risk for breast cancer in women with
breast cancer (14–16). Elevated endogenous serum the highest concentration of bioavailable estradiol ($6.83pmol/L or 1.9 pg/mL) was 3.6 (95% CI, 1.3 to 10.0) com- concentrations of androgens may also be related to pared with women with the lowest concentration. The risk an increased risk for breast cancer (5, 7, 17), but for breast cancer in women with the highest concentration this relation may not be independent of serum es- of free testosterone compared with those with the lowest trogens (8, 18, 19). The best estrogen fraction with concentration was 3.3 (CI, 1.1 to 10.3). The estimated which to predict risk has not been identified (3).
incidence of breast cancer per 1000 person-years was 0.4 Most studies have included measurements of total (CI, 0.0 to 1.3) in women with the lowest levels of bioavail- hormone levels; the concentrations of free hor- able estradiol and free testosterone compared with 6.5 (CI, mones may have even stronger associations. Most of 2.7 to 10.3) in women with the highest concentrations of the women in these studies were postmenopausal these hormones. Traditional risk factors for breast cancer and younger than 65 years of age.
were similar in case-patients and controls. Adjustments forthese risk factors had little effect on the results.
Two randomized trials (20, 21) have shown a reduction in the occurrence of primary breast can- Conclusions: Estradiol and testosterone levels may play
cer in patients who have received tamoxifen and important roles in the development of breast cancer in raloxifene. In the Breast Cancer Prevention Trial older women. A single measurement of bioavailable estra-diol and free testosterone may be used to estimate a (20), 4 years of tamoxifen use led to a 45% reduc- woman's risk for breast cancer. Women identified as being tion in breast cancer incidence in 13 388 women.
at high risk for breast cancer as determined by these Women in this study were considered to be at high hormone levels may benefit from antiestrogen treatment risk for breast cancer on the basis of the presence for primary prevention.
of certain risk factors, including age of 60 years orolder; about 30% of women were 60 years of age or This paper is also available at http://www.acponline.org.
older. The Multiple Outcomes of Raloxifene Eval- Ann Intern Med. 1999;130:270-277.
uation (MORE) trial (21) found a 70% reduction in For author affiliations and current author addresses, see end of risk for breast cancer, especially cases of estrogen receptor–positive cancer, after 33 months of treat- 1999 American College of Physicians–American Society of Internal Medicine ment with raloxifene (21). About 80% of the 7704 to a central repository and stored in liquid nitrogen women in this trial were older than 60 years of age.
at 2190 °C until the assays were performed. We Because treatment with tamoxifen or raloxifene measured total estradiol, bioavailable estradiol or entails costs and risk (20–22), it is important to estradiol that was not bound by sex hormone–bind- identify women who are at greatest risk for breast ing globulin (SHBG), free estradiol, estrone, es- cancer and are therefore most likely to benefit from trone sulfate, androstenedione, dehydroepiandros- antiestrogen therapies. Our study tests the hypoth- terone sulfate, total and free testosterone, and esis that serum concentrations of estradiol and tes- SHBG. All assays were done by Corning Nichols tosterone, measured an average of 3 years before Institute (San Juan Capistrano, California); re- the clinical diagnosis of breast cancer, are related to searchers were blinded to participants' breast cancer risk for breast cancer in women 65 years of age or status. The sensitivity of the assays refers to the older. We hypothesized that measurements of se- lower limit of detection. Assays were performed rum hormones could be used to identify women at concurrently on serum specimens from case-patients high risk for developing breast cancer. We used a and controls.
case–cohort approach to compare serum hormone The intra-assay and total assay variability is ex- concentrations in 97 incident case-patients with pressed as a coefficient of variation. In this study, a breast cancer and 244 randomly selected controls.
range of coefficient of variation includes values for a low-concentration quality-control sample to those of a high-concentration quality-control sample. When no range is reported, the coefficient of variation was similar for both low- and high-concentration quality- control samples.
Study Sample
Total estradiol was measured by using liquid– All women were participants in the Study of Os- liquid organic extraction, column chromatography, teoporotic Fractures, a prospective study of 9704 and radioimmunoassay (coefficient of variation for white, community-dwelling women who were at intra-assay and total assay, 4% to 12% and 9% to least 65 years of age and were recruited at four 11%, respectively; sensitivity, 7.3 pmol/L). Free es- clinical centers across the United States (23).
tradiol was measured by using equilibrium dialysis Women were excluded from the study if they re- and calculated by using the percentage of dialyzable ported a bilateral hip replacement or the inability to estradiol and total estradiol (coefficient of variation walk without the assistance of another person. Dur- for intra-assay and total assay, 3% to 4% and 5%, ing 3.2 years of follow-up, we confirmed 121 cases respectively; sensitivity, 0.37 pmol/L). Percentage of of breast cancer, including 4 cases of carcinoma in non–SHBG-bound estradiol or bioavailable estra- situ, through review of medical records by a physi- diol was measured by ammonium sulfate precipita- cian-epidemiologist (14). We excluded women who tion of SHBG-bound steroids (coefficient of varia- reported current estrogen replacement therapy at tion for intra-assay and total assay, 3% and 6%, baseline; remaining were 97 confirmed cases of in- respectively). The amount of estradiol that was cident breast cancer. Using a case–cohort approach, non–SHBG-bound was then calculated as the prod- we chose as controls a random sample of 247 uct of the total amount of estradiol and the per- women who survived to the first annual visit, denied centage of nonbound estradiol.
a history of breast cancer, and did not report use of Estrone was measured by using extraction, chro- estrogen at baseline. Three of these women subse- matography, and radioimmunoassay (coefficient of quently developed incident breast cancer and were variation for intra-assay and total assay, 6% to 12% included in the case-patient group. This study was and 8% to 17%, respectively; sensitivity, 37 pmol/L).
approved by the biomedical institutional review Estrone sulfate was measured by using organic ex- board at each of the participating institutions. All traction, enzymatic hydrolysis, celite chromatogra- participants provided informed consent.
phy, and radioimmunoassay (coefficient of variation for intra-assay and total assay, 6% to 7% and 7% to 8%, respectively; sensitivity, 143 pmol/L).
Serum specimens were obtained from all partic- Androstenedione was measured by using a radio- ipants at a baseline examination in 1986 to 1988. All immunoassay after preparation for analysis by or- participants were instructed to adhere to a fat-free ganic extraction and chromatography (coefficient of diet during the night and morning before the exam- variation for intra-assay and total assay, 6% to 10% ination to minimize lipemia that might interfere and 7% to 16%, respectively; sensitivity, 0.10 nmol/ with assays. Blood was drawn between 8:00 a.m. and L). Dehydroepiandrosterone sulfate was measured 2:00 p.m., and serum was immediately frozen to by using radioimmunoassay after preparation for 220 °C. Within 2 weeks, all samples were shipped the analysis by serial dilution (coefficient of varia- 16 February 1999 • Annals of Internal Medicine • Volume 130 • Number 4 (Part 1) tion for intra-assay and total assay, 6% to 11% and to compare the distribution of hormones in case- 9% to 12%, respectively; sensitivity, 0.17 mmol/L).
patients and controls.
Total testosterone was measured by using radioim- For all hormones except free estradiol, the rela- munoassay with chromatographic purification (coef- tive hazard (RH) for breast cancer was calculated ficient of variation for intra-assay and total assay, (using the lowest quartile as the reference group) 6% to 14% and 5% to 13%, respectively; sensitivity, across quartiles of sex-steroid hormone levels by 0.03 nmol/L). Free testosterone was measured by using a modification of the Cox proportional haz- using equilibrium dialysis. Calculation of free testos- ards model that accounted for the case–cohort sam- terone was adjusted for albumin concentration (co- pling design; this modified model has been success- efficient of variation for intra-assay and total assay, fully applied in previous studies (25). Cut-points for 5% and 5.4%, respectively; sensitivity, 34.7 pmol/L).
quartiles were based on distribution within the ran- Sex hormone–binding globulin was measured by us- dom subset of the cohort. The distribution of free ing radioimmunoassay (coefficient of variation for estradiol did not allow division by quartiles; four intra-assay and total assay, 7% and 7.8%, respec- levels of free estradiol were assigned to approximate tively; sensitivity, 5.0 nmol/L).
quartiles as closely as possible. A test was done for We formed the ratio of estrone sulfate to estrone linear trend of increasing risk for breast cancer to test the hypothesis suggested by Dorgan and across quartiles of hormones.
coworkers (5) that women who develop breast can- We initially adjusted for age and modified body cer may be less able than other women to metabo- mass index. Multivariate models included adjust- lize estrone to a less active form.
ment for conventional risk factors for breast cancer, We determined the reproducibility of selected including age; modified body mass index; age at hormone measurements in 20 postmenopausal menarche, first birth, and menopause; surgical women by assaying hormone levels in duplicate in menopause (yes or no); nulliparity (yes or no); fam- different batches. Pearson correlations (all signifi- ily history of breast cancer in a mother or sister (yes cant at P , 0.001) between the two measures were or no); past estrogen use (yes or no); walking for as follows: total testosterone, r 5 0.98; free testos- exercise (yes or no); and alcohol consumption. Un- terone, r 5 0.97; total estradiol, r 5 0.56; non–SHBG- less otherwise noted, variables were entered as con- bound estradiol, r 5 0.83; estrone, r 5 0.67; estrone tinuous variables. Alcoholic drinks were converted sulfate, r 5 0.70; androstenedione, r 5 0.77; dehydro- to grams per day, assuming an average of 11.5 epiandrosterone sulfate, r 5 0.97; and SHBG, r 5 0.97.
grams per drink. Average number of grams con- Initial and repeated mean values were similar.
sumed per day was categorized to conform with five categories (0 g/d, 1.5 g/d, 1.5 g/d to ,5.0 g/d, 5.0 g/d to ,15 g/d, and 15 or more g/d) that were typically used in other studies (25, 26). These cate- Weight (in lightweight clothing with shoes re- gories were entered as dummy variables in the mul- moved) was recorded with a balance-beam scale.
tivariate model.
Self-reported height at 25 years of age was used to In the Nurses' Health Study (8), the association calculate the modified body mass index because between serum hormones and breast cancer was women with low bone mass experience height loss particularly strong in women who had never used secondary to vertebral fractures. A reproductive his- estrogen. Therefore, we excluded past estrogen us- tory, obtained by questionnaire and interview, in- ers in our study and redid our analyses.
cluded information on ages at menarche, meno- We estimated the incidence of breast cancer per pause, and first birth; parity; and family history of 1000 person-years and 95% CIs by levels of both breast cancer. Participants were asked about past bioavailable estradiol and free testosterone. For use of estrogen replacement therapy, current and these analyses, we combined the two middle quar- lifetime use of cigarettes and alcohol, and whether tiles of hormones and calculated the incidence of they walked for exercise. We calculated the number breast cancer in relation to levels of bioavailable of alcoholic drinks per week; nondrinkers were estradiol and free testosterone. To calculate inci- coded as having zero intake.
dence, we estimated total person-years within each category of bioavailable estradiol and free testoster- one by applying the person-year distribution of the random sample of the cohort (controls) to the total Characteristics of case-patients and controls were number of person-years in the cohort. Rates were compared by t-test (continuous variables) or by chi- then obtained in the usual fashion by multiplying square test (categorical variables). Sex-steroid hor- the ratio of the number of case-patients to the mone levels were not normally distributed. The number of person-years by 1000 (to express in units nonparametric (Wilcoxon two-sample) test was used per 1000 person-years). We estimated standard er- 16 February 1999 • Annals of Internal Medicine • Volume 130 • Number 4 (Part 1) Descriptive Characteristics of Case-Patients
smoking, exercise, and other conventional risk fac- tors for breast cancer (Table 1). The mean body
weight and body mass index tended to be higher in case-patients. Case-patients reported more con- sumption of alcohol in the past year. About one Age 6 SD, y third of case-patients and one third of controls re- Weight 6 SD, kg Body mass index 6 SD, ported past use of estrogen replacement therapy.
Case-patients and controls did not differ signifi- Height at 25 years of age 6 SD, cm cantly in the number of years since discontinuing Age at menarche 6 SD, y use of estrogen or in duration of estrogen use.
Age at first birth 6 SD, y† Age at menopause 6 SD, y Live births 6 SD, n Sex-Steroid Hormones and Breast Cancer
Surgical menopause, % Ever pregnant, % Median concentrations of sex-steroid hormone Nulliparous, % were higher in case-patients than in controls (Table
Family history of breast 2). In particular, total estradiol and bioavailable
Walks for exercise, % estradiol concentrations were about 30% higher and Current smoker, % Drank alcohol within the free testosterone concentrations were 28% higher.
past 12 months, % Case-patients and controls differed significantly in Median drinks per week (range), n distribution of all hormones except SHBG.
Past estrogen use, % The association between serum hormone level Time elapsed since stopping estrogen 6 SD, and breast cancer was strongest for bioavailable es- tradiol: Women in the highest quartile of estradiol Duration of estrogen use 6 SD, y‡ concentration had a 3.6-fold greater risk for breast cancer than women in the lowest quartile of estra- * Values expressed with SDs are means.
† Among parous women.
diol concentration (95% CI, 1.3-fold to 10.0-fold) ‡ Estrogen users only.
(Table 3). Among the androgens, total and free
testosterone concentrations were strongly linked to rors by assuming a Poisson distribution for occur- subsequent risk for breast cancer; risk was three rence of events and by using a Taylor expansion to times greater in women with the highest concentra- account for the additional variability introduced by tions of testosterone. These associations were inde- the estimation of person-years.
pendent of age, body mass index, and other conven- To test the hypothesis that the association be- tional risk factors for breast cancer.
tween breast cancer and the precursor hormone Women in the highest quartile of estrone, es- (androstenedione or dehydroepiandrosterone sul- trone sulfate, androstenedione, and dehydroepi- fate) could be explained by levels of bioavailable androsterone sulfate concentrations also had an in- estradiol and free testosterone, we calculated the RH for breast cancer in multivariate models that included all three hormones: androstenedione (or Median and Range of Concentrations of
dehydroepiandrosterone sulfate), bioavailable estra- Sex-Steroid Hormones in Case-Patients
diol, and free testosterone. For these analyses, we dichotomized the hormone variables and compared Sex-Steroid Hormones women in the top three quartiles with those in the lowest quartile. Adjustment for age and body mass index was included in these models.
Estradiol, pmol/L Role of the Funding Source
estradiol, pmol/L 4.8 (1.10 –23.9) 3.7 (0.70 –38.9) The funding sources did not participate in the Free estradiol, pmol/L design and conduct of the study or reporting of Estrone, pmol/L 88.8 (0 –255.2) 74.0 (0 –248.0) Estrone sulfate, pmol/L 630.7 (120 –2957) 459.5 (0 –3108) results and had no role in the decision to submit this paper for publication.
1.54 (0.17–5.31) sulfate, mmol/L 2.04 (0.24 –9.69) Total testosterone, Free testosterone, Case-patients and controls (random sample of SHBG, nmol/L the cohort) were similar with respect to age, repro- ductive history, family history of breast cancer, * SHBG 5 sex hormone– binding globulin.
† Wilcoxon two-sample test.
16 February 1999 • Annals of Internal Medicine • Volume 130 • Number 4 (Part 1) Relative Hazard for Breast Cancer by Concentration of Sex-Steroid Hormones*
Sex-Steroid Hormones Relative Hazard Adjusted for Conventional Breast Cancer Risk Factors Level 1 (,18.4 pmol/L) Level 2 (18.4 –,22.0 pmol/L) Level 3 (22.0 –,29.4 pmol/L) Level 4 ($29.4 pmol/L) Bioavailable estradiol Quartile 1 (,2.20 pmol/L) Quartile 2 (2.2–,3.78 pmol/L) Quartile 3 (3.78 –,6.83 pmol/L) 1.8 (0.8 – 4.4) 2.2 (0.8 – 6.6) Quartile 4 ($6.83 pmol/L) 3.4 (1.4 – 8.3) Level 1 (,0.37 pmol/L) Level 2 (0.37–,0.73 pmol/L) 1.6 (0.4 – 6.7) Level 3 (0.73–,1.10 pmol/L) 4.8 (0.9 –25.4) Level 4 ($1.10 pmol/L) Quartile 1 (,51.8 pmol/L) Quartile 2 (51.8 –,74.0 pmol/L) Quartile 3 (74.0 –,103.6 pmol/L) Quartile 4 ($103.6 pmol/L) Quartile 1 (,305.4 pmol/L) Quartile 2 (305.4 –,476.6 pmol/L) Quartile 3 (476.6 –,756.3 pmol/L) Quartile 4 ($756.3 pmol/L) 2.0 (0.9 – 4.2) Quartile 1 (,0.84 nmol/L) Quartile 2 (0.84 –,1.26 nmol/L) Quartile 3 (1.26 –,1.78 nmol/L) Quartile 4 ($1.78 nmol/L) Quartile 1 (,1.00 mmol/L) Quartile 2 (1.00 –,1.74 mmol/L) Quartile 3 (1.74 –,2.71 mmol/L) Quartile 4 ($2.71 mmol/L) 2.1 (1.0 – 4.4) 2.4 (0.9 – 6.3) Total testosterone Quartile 1 (,0.42 nmol/L) Quartile 2 (0.42–,0.62 nmol/L) Quartile 3 (0.62–,0.97 nmol/L) 5.5 (1.8 –17.0) Quartile 4 ($0.97 nmol/L) Free testosterone Quartile 1 (,5.54 pmol/L) Quartile 2 (5.54 –,8.32 pmol/L) Quartile 3 (8.32–,13.17 pmol/L) Quartile 4 ($13.17 pmol/L) Sex hormone– binding globulin Quartile 1 (,29 nmol/L) Quartile 2 (29 –,43 nmol/L) Quartile 3 (43–,59 nmol/L) Quartile 4 ($59 nmol/L) Ratio of estrone sulfate to estrone Quartile 1 (,6.13) Quartile 2 (6.13–,9.00) Quartile 3 (9.00 –13.10) Quartile 4 ($13.10) * BMI 5 body mass index.
† Adjusted for age; body mass index; ages at menarche, first birth, and menopause; nulliparity; family history of breast cancer; physical activity; surgical menopause; and alcohol creased risk for breast cancer (Table 3). Sex
diol and free testosterone (Figure). In contrast, the
hormone–binding globulin and the ratio of estrone incidence of breast cancer was 6.5 per 1000 person- sulfate to estrone were not associated with breast years (CI, 2.7 to 10.3) in women with the highest cancer. Results were similar when we excluded concentration of both hormones.
women who had used estrogen in the past.
The estimated incidence of breast cancer was lowest (0.4 per 1000 person-years [CI, 0 to 1.29]) in We tested the hypothesis that the precursor hor- women with the lowest levels of bioavailable estra- mones, androstenedione or dehydroepiandrosterone 16 February 1999 • Annals of Internal Medicine • Volume 130 • Number 4 (Part 1)

sulfate, were not independently related to breast cancer. In a model that included levels of bioavail- able estradiol, free testosterone, and androstenedi- one, bioavailable estradiol (RH, 2.5 [CI, 1.2 to 5.3]) was independently related to breast cancer. A two- fold increased risk for breast cancer was related to the level of free testosterone, but the CI included 1.0 (RH, 2.1 [CI, 0.9 to 4.7]). Androstenedione was not related to the risk for breast cancer (RH, 1.2 [CI, 0.6 to 2.4]). Similar results were obtained in models that used bioavailable estradiol (RH, 2.5 [CI, 1.2 to 5.3]), free testosterone (RH, 2.1 [CI, 0.9 Incidence of breast cancer (95% CI) in relation to concen-
to 4.6]), and dehydroepiandrosterone sulfate (RH, trations of bioavailable estradiol and free testosterone. Incidence
1.2 [CI, 0.6 to 2.3]). Inclusion of androstenedione expressed per 1000 person-years. Quartile 1: free testosterone, up-slantingdiagonally striped bar; quartiles 2 and 3: free testosterone, white bar; quar- and dehydroepiandrosterone sulfate in the same tile 4: free testosterone, down-slanting diagonally striped bar.
model yielded similar results.
treatment. Clinical trials to test the effects of anti- estrogen therapies on breast cancer risk related to estrogen concentrations are ongoing.
The results of this study support the hypothesis Our results also suggest that interventions to re- that sex hormones are an important factor in the duce serum hormone concentrations may reduce development of breast cancer in older women. In risk for breast cancer. A low-fat diet (29–31), weight particular, women with a bioavailable estradiol con- reduction (32), and a vegetarian diet (33) have been centration greater than 7 pmol/L (1.9 pg/mL) had a shown to reduce levels of sex-steroid hormones. We risk for breast cancer that was 3.6-fold greater than previously reported an inverse association between that in women with the lowest concentration of concentrations of serum estrone and physical activ- bioavailable estradiol. We also found a strong rela- ity (34). In the Women's Health Trial (31), a 10- to tion between the unbound portion of testosterone 22-week low-fat diet intervention was associated and the risk for breast cancer. Our results are con- with a 17% reduction in estradiol concentrations in sistent with those of other prospective studies of the healthy postmenopausal women. The Women's relation between sex-steroid hormone levels and the Health Initiative (35) is directly testing whether low- risk for breast cancer in somewhat younger women fat dietary patterns will reduce the incidence of breast cancer.
The average incidence of breast cancer in white Sources of testosterone in postmenopausal women women 65 years of age and older in the United include direct secretion from the ovary and from States is 4.6 per 1000 person-years (27). On the the precursor hormones, androstenedione or dehy- basis of our results, we estimate that the incidence droepiandrosterone sulfate. Testosterone could in- of breast cancer in women with the highest concen- fluence the risk for breast cancer directly or indi- trations of bioavailable estradiol and free testoster- rectly (as a source of estradiol). Androgen receptors one is about 40% higher than this expected rate.
have been identified in human breast cancer cells, The magnitude of the relative risk is similar to that although in vitro activation of the androgen recep- of the strongest risk factors for breast cancer (per- tor tends to suppress the proliferation of breast sonal history of ductal carcinoma in situ and atypi- cancer cells (36). In three studies, the association cal hyperplasia) (28).
between levels of total testosterone and breast can- The absolute concentrations of hormones, espe- cer was not independent of levels of bioavailable cially estradiol, were very low but are consistent estradiol (8, 18, 19). However, these studies did not with those previously reported in postmenopausal measure concentrations of free testosterone. In our women (8). Nonetheless, a gradient of risk was ob- study, we found an association of free testosterone served across increasing concentrations. This gradi- levels to breast cancer that was independent of bio- ent of risk is greater than that observed between available estradiol levels, thereby suggesting a direct serum cholesterol concentrations and coronary heart disease, especially in older women (3). Our The primary source of estrogens in postmeno- results suggest that measurement of bioavailable es- pausal women is the aromatization of androstenedi- tradiol and free testosterone may be used as a clin- one, an adrenal hormone (37). We and others (5) ical measure to identify women at high-risk for found an association between breast cancer and breast cancer who may benefit from antiestrogen higher concentrations of androstenedione and dehy- 16 February 1999 • Annals of Internal Medicine • Volume 130 • Number 4 (Part 1) droepiandrosterone sulfate. However, in our study, testosterone (46, 47). Future studies should include the association between breast cancer and concen- these measures.
trations of androstenedione and dehydroepiandros- Measures of traditional risk factors for breast terone sulfate was no longer significant in models cancer, such as age at first birth, nulliparity, early that included concentrations of bioavailable estra- menarche, and family history of breast cancer, were diol and free testosterone; this finding is consistent remarkably similar between case-patients and con- with the hypothesis that increased concentrations of trols; our findings suggest that these conventional androstenedione or dehydroepiandrosterone sulfate, risk factors cannot accurately identify older women as a precursor to estradiol and testosterone, may at high risk for breast cancer. Our results are con- contribute to the increased risk for breast cancer.
sistent with those of other studies of older women Local formation of androgens and estrogens in (8). In addition, these risk factors are highly prev- the breast may also contribute to the development alent. In one study, more than 98% of the popula- of breast cancer. Breast fat has aromatase activity, tion had at least one of these risk factors (48), but and levels of aromatase activity in adipose tissue most women with one or more of these risk factors adjacent to malignant tumors were significantly do not develop breast cancer. Hence, it is unlikely higher than those in tissues adjacent to benign le- that they can be used to identify older women at sions in one study (38). Breast tissue also contains a risk for breast cancer.
sulfatase enzyme that can convert estrone sulfate to Our study has several limitations. Our cohort estrone, which can then be converted to estradiol, consists primarily of healthy, community-dwelling el- thereby increasing the level of estradiol in the derly white women; however, the overall rate of breast (39). In our study, both estrone and estrone breast cancer in our cohort (4.3 per 1000 person- sulfate were directly related to risk for breast can- years) was similar to that observed for white women cer. Additional enzymatic processes, including the aged 65 years and older in the United States (27).
17b-estradiol dehydrogenase, could allow high levels of sex-steroid hormones to accumulate in breast Our results may not apply to women of other ethnic tissue (40, 41). It is unlikely, however, that an in- groups. The concentrations of hormones in these crease in estrogen synthesis in the breast could ac- elderly women are relatively low and may be subject count for the increased blood levels of estradiol that to increased laboratory variability. However, when we observed in women with breast cancer.
the same laboratory was used, the reproducibility of Breast cancer generally requires several years to sex-steroid hormone concentrations in postmeno- become clinically or radiographically detectable pausal women was excellent (49). Hormone concen- (42). Breast tumor aromatase activity may be more trations were measured only once, and a single important than aromatase in breast fatty tissue for measure is always imprecise to some degree. We the maintenance of tumor estradiol levels (43).
used a specialized endocrine laboratory; however, Thus, we cannot exclude the possibility that higher laboratory methods must be standardized before levels of serum estrogens in case-patients reflect routine clinical laboratories are used to screen enhanced production of estrogen within the tumor women for risk for breast cancer in relation to itself. However, in the Nurses' Health Study (8), serum hormone concentrations. Although this is one exclusion of case-patients who had received a diag- of the largest cohort studies of breast cancer, we nosis of breast cancer within 1 year of initial blood had limited power to test for interactions among collection had no effect on the results. Longer follow- hormones and breast cancer. Current hormone lev- up will be needed to evaluate the long-term relation els may not reflect earlier levels. However, several between serum hormones and breast cancer.
studies have documented correlations of serum es- Weight gain, obesity, and increased intra-abdom- trogens over several years, especially in women inal fat have all been identified as possible risk whose weight remains stable (49–51). Thus, it is factors for breast cancer (44, 45), possibly because possible that the levels of hormones measured in of aromatization of androstenedione to estrone in these women may reflect exposures over a longer fatty tissue (46). In our study, adjustment for obe- sity as measured by body weight or body mass index Estradiol and testosterone play important roles in did not substantially influence the association be- the risk for breast cancer in older women. Concen- tween sex-hormone concentration and breast can- trations of these hormones predict the risk for cer. However, we did not measure thigh fat mass, breast cancer and may help clinicians decide about which has been associated with greater aromatase treatments to decrease breast cancer risk.
activity and therefore higher blood levels of estrone and estradiol (47), or intra-abdominal fat mass, From University of Pittsburgh, Pittsburgh, Pennsylvania; Maine which has been associated with greater concentra- Medical Center, Portland, Maine; University of California, San Francisco; and Veterans Affairs Medical Center, San Francisco, tions of insulin, free and bioavailable estradiol, and 16 February 1999 • Annals of Internal Medicine • Volume 130 • Number 4 (Part 1) Grant Support: In part by National Institutes of Health Public al. Raloxifene reduces the risk of breast cancer in postmenopausal women
Health Service research grants AM35584, AR35582, AG05407, with osteoporosis. [Abstract]. Proceedings of the American Society of Clinical AG05395, AR35583, and TG32AG00181 and by grant DAMD17- Oncology Program. May 1998.
96-6114 from the U.S. Army Medical Research and Material 22. Evista (package insert). Indianapolis, IN: Eli Lilly; 1998.
23. Cummings SR, Black DM, Nevitt MC, Browner WS, Cauley JA, Genant
HK, et al. Appendicular bone density and age predict hip fracture in women.
The Study of Osteoporotic Fractures Research Group. JAMA. 1990;263:665-8.
Requests for Reprints: Jane A. Cauley, DrPH, Department of 24. Prentice RL. A case-control design for epidemiologic cohort studies and
Epidemiology, Graduate School of Public Health, University of disease prevention trials. Biometrika. 1986;73:1-11.
Pittsburgh, 130 DeSoto Street, Crabtree Hall A524, Pittsburgh, 25. Schatzkin A, Jones DY, Hoover RN, Taylor PR, Brinton LA, Ziegler RG,
et al. Alcohol consumption and breast cancer in the epidemiologic follow-up
study of the first National Health and Nutrition Examination Survey. N Engl JMed. 1987;316:1169-73.
Current Author Addresses: Drs. Cauley and Kuller: Department of 26. Willett WC, Stampfer MJ, Colditz GA, Rosner BA, Hennekens CH,
Epidemiology, Graduate School of Public Health, University of Speizer FE. Moderate alcohol consumption and the risk of breast cancer. N
Pittsburgh, 130 DeSoto Street, Crabtree Hall A524, Pittsburgh, Engl J Med. 1987;316:1174-80.
27. SEER Cancer Statistics Review, 1973-1990. Bethesda, MD: U.S. Department of
Health and Human Services; 1993. National Institutes of Health publication Dr. Lucas: Health Services Research, Maine Medical Center, 31 no. 93-2789.
Bramhall, Portland, ME 04102.
28. Kelsey JL, Bernstein L. Epidemiology and prevention of breast cancer. Annu
Drs. Stone, Browner, and Cummings: Department of Biostatistics Rev Public Health. 1996;17:47-67.
and Epidemiology, University of California, 74 New Montgomery 29. Rose DP, Connolly JM, Chlebowski RT, Buzzard IM, Wynder EL. The
effects of a low-fat dietary intervention and tamoxifen adjuvant therapy on Street, Suite 600, San Francisco, CA 94105.
the serum estrogen and sex hormone-binding globulin concentrations ofpostmenopausal breast cancer patients. Breast Cancer Res Treat. 1993;27:253-62.
30. Dorgan JF, Reichman ME, Judd JT, Brown C, Longcope C, Schatzkin A, et
al. Relation of energy, fat, and fiber intakes to plasma concentrations of estro-
gens and androgens in premenopausal women. Am J Clin Nutr. 1996;64:25-31.
31. Prentice R, Thompson D, Clifford C, Gorbach S, Goldin B, Byar D.
1. Miller BA, Feuer EJ, Hankey BF. The significance of the rising incidence of
Dietary fat reduction and plasma estradiol concentration in healthy post- breast cancer in the United States. Important Adv Oncol. 1994;193-207.
menopausal women. The Women's Health Trial Study Group. J Natl Cancer 2. Parker SL, Tong T, Bolden S, Wingo PA. Cancer statistics, 1997. CA
Cancer J Clin. 1997;47:5-27.
32. de Waard F, Poortman J, de Pedro-Alvarez Ferrero M, Baanders-van
3. Colditz GA. Relationship between estrogen levels, use of hormone replace-
Halewijn EA. Weight reduction and oestrogen excretion in obese post-
ment therapy, and breast cancer. J Natl Cancer Inst. 1998;90:814-23.
menopausal women. Maturitas. 1982;4:155-62.
4. Toniolo PG, Levitz M, Zeleniuch-Jacquotte A, Banerjee S, Koenig KL,
33. Armstrong BK, Brown JB, Clarke HT, Crooke DK, Hahnel R, Masarei
Shore RE, et al. A prospective study of endogenous estrogens and breast
JR, et al. Diet and reproductive hormones: a study of vegetarian and non-
cancer in postmenopausal women. J Natl Cancer Inst. 1995;87:190-7.
vegetarian postmenopausal women. J Natl Cancer Inst. 1981;67:761-7.
5. Dorgan JF, Longcope C, Stephenson HE Jr, Falk RT, Miller R, Franz C,
34. Cauley JA, Gutai JP, Kuller LH, LeDonne D, Powell JG. The epidemiology
et al. Relation of prediagnostic serum estrogen and androgen levels to breast
of serum sex hormones in postmenopausal women. Am J Epidemiol. 1989; cancer risk. Cancer Epidemiol Biomarkers Prev. 1996;5:533-9.
6. Moore JW, Clark GM, Hoare SA, Millis RR, Hayward JL, Quinlan MK,
35. Design of the Women's Health Initiative: clinical trial and observational study. The
et al. Binding of oestradiol to blood proteins and aetiology of breast cancer.
Women's Health Initiative Study Group. Control Clin Trials. 1998;19:61-109.
Int J Cancer. 1986;38:625-30.
36. MacIndoe JH, Etre LA. An antiestrogenic action of androgens in human
7. Berrino F, Muti P, Micheli A, Bolelli G, Krogh V, Sciajno R, et al. Serum
breast cancer cells. J Clin Endocrinol Metab. 1981;53:836-42.
sex hormone levels after menopause and subsequent breast cancer. J Natl 37. Jaffe RB. Pathophysiology: the menopause and perimenopausal period. In:
Cancer Inst. 1996;88:291-6.
Yen SS, Jaffe RB, eds. Reproductive Endocrinology: Physiology, Pathophysiology, 8. Hankinson SE, Willett WC, Manson JE, Colditz GA, Hunter DJ,
and Clinical Management. 2d ed. Philadelphia: WB Saunders; 1986:406-23.
Spiegelman D, et al. Plasma sex steroid hormone levels and risk of breast
38. O'Neill JS, Miller WR. Aromatase activity in breast adipose tissue from women
cancer in postmenopausal women. J Natl Cancer Inst. 1998;90:1292-9.
with benign and malignant breast diseases. Br J Cancer. 1987;56:601-4.
9. Helzlsouer KJ, Alberg AJ, Bush TL, Longcope C, Gordon GB, Comstock
39. Santen RJ, Leszczynski D, Tilson-Mallet N, Feil PD, Wright C, Manni A,
GW. A prospective study of endogenous hormones and breast cancer. Cancer
et al. Enzymatic control of estrogen production in human breast cancer:
Detect Prev. 1994;18:79-85.
relative significance of aromatase versus sulfatase pathways. Ann N Y Acad 10. Garland CF, Friedlander NJ, Barrett-Connor E, Khaw KT. Sex hormones
Sci. 1986;464: 126-37.
and postmenopausal breast cancer: a prospective study in an adult commu- 40. Thijssen JH, Blankenstein MA, Donker GH, Daroszewski J. Endogenous
nity. Am J Epidemiol. 1992;135:1220-30.
steroid hormones and local aromatase activity in the breast. J Steroid Biochem 11. Wysowski DK, Comstock GW, Helsing KJ, Lau HL. Sex hormone levels in
Mol Biol. 1991;39:799-804.
serum in relation to the development of breast cancer. Am J Epidemiol.
41. Perel E, Wilkins D, Killinger DW. The conversion of androstenedione to
estrone, estradiol, and testosterone in breast tissue. J Steroid Biochem. 1980; 12. Bulbrook RD, Moore JW, Clark GM, Wang DY, Millis RR, Hayward JL.
Relation between risk of breast cancer and biological availability of estradiol in the 42. Spratt JS, Meyer JS, Spratt JA. Rates of growth of human neoplasms: Part
blood: prospective study in Guernsey. Ann N Y Acad Sci. 1986;464:378-88.
II. J Surg Oncol. 1996;61:68-83.
13. Thomas HV, Reeves GK, Key TJ. Endogenous estrogen and postmenopausal
43. Blankenstein MA, Maitimu-Smeele I, Donker GH, Daroszewski J,
breast cancer: a quantitative review. Cancer Causes Control. 1997;8:922-8.
Milewicz A, Thijssen JH. On the significance of in situ production of oestrogens
14. Cauley JA, Lucas FL, Kuller LH, Vogt MT, Browner WS, Cummings SR.
in human breast cancer tissue. J Steroid Biochem Mol Biol. 1992;41:891-6.
Bone mineral density and risk of breast cancer in older women: the study of 44. Sellers TA, Gapstur SM, Potter JD, Kushi LH, Bostick RM, Folsom AR.
osteoporotic fractures. Study of Osteoporotic Fractures Research Group.
Association of body fat distribution and family histories of breast and ovarian cancer with risk of postmenopausal breast cancer. Am J Epidemiol. 1993;138: 15. Zhang Y, Kiel DP, Kreger BE, Cupples LA, Ellison RC, Dorgan JF, et al.
Bone mass and the risk of breast cancer among postmenopausal women.
45. Mannisto S, Pietinen P, Pyy M, Palmgren J, Eskelinen M, Uusitupa M.
N Engl J Med. 1997;336:611-7.
Body-size indicators and risk of breast cancer according to menopause and 16. Meema S, Meema HE. Possible estrogenic effect on bone in postmeno-
estrogen-receptor status. Int J Cancer. 1996;68:8-13.
pausal patients with mammary carcinoma. Cancer. 1966;19:433-6.
46. Killinger DW, Perel E, Daniilescu D, Kharlip L, Lindsay WR. The rela-
17. Dorgan JF, Stanczyk FZ, Longcope C, Stephenson HE Jr, Chang L,
tionship between aromatase activity and body fat distribution. Steroids. 1987; Miller R, et al. Relationship of serum dehydroepiandrosterone (DHEA), DHEA
sulfate, and 5-androstene-3 beta, 17 beta-diol to risk of breast cancer in 47. Rink JD, Simpson ER, Barnard JJ, Bulun SE. Cellular characterization of
postmenopausal women. Cancer Epidemiol Biomarkers Prev. 1997;6:177-81.
adipose tissue from various body sites of women. J Clin Endocrinol Metab.
18. Zeleniuch-Jacquotte A, Bruning PF, Bonfrer JM, Koenig KL, Shore RE,
Kim MY, et al. Relation of serum levels of testosterone and dehydroepi-
48. Rockhill B, Weinberg CR, Newman B. Population attributable fraction
androsterone sulfate to risk of breast cancer in postmenopausal women. Am J estimation for established breast cancer risk factors: considering the issues of high prevalence and unmodifiability. Am J Epidemiol. 1998;147:826-33.
19. Thomas HV, Key TJ, Allen DS, Moore JW, Dowsett M, Fentiman IS, et
49. Hankinson SE, Manson JE, Spiegelman D, Willett WC, Longcope C,
al. A prospective study of endogenous serum hormone concentrations and
Speizer FE. Reproducibility of plasma hormone levels in postmenopausal women
breast cancer risk in post-menopausal women on the island of Guernsey. Br J over a 2-3-year period. Cancer Epidemiol Biomarkers Prev. 1996;4:649-54.
50. Kuller LH, Gutai JP, Meilahn E, Matthews KA, Plantinga P. Relationship
20. Fisher B, Costantino JP, Wickerham DL, Redmond CK, Kavanah M,
of endogenous sex steroid hormones to lipids and apoproteins in postmeno- Cronin WM, et al. Tamoxifen for prevention of breast cancer: report of the
pausal women. Arteriosclerosis. 1990;10:1058-66.
National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer 51. Muti P, Trevisan M, Micheli A, Krogh V, Bolelli G, Sciajno R, et al.
Reliability of serum hormones in premenopausal and postmenopausal women 21. Cummings SR, Eckert S, Grady D, Glusman J, Krueger K, Norton L, et
over a one-year period. Cancer Epidemiol Biomarkers Prev. 1996;5:917-22.
16 February 1999 • Annals of Internal Medicine • Volume 130 • Number 4 (Part 1)

Source: http://sites.utoronto.ca/acdclab/pubs/PM/10068384.pdf