Serum and Urinary Androgens and Risk of Breast Cancer in Postmenopausal Women1

Home , Androstanediol

(CANCER RESEARCH 51. 2572-2576. May 15. 1991] Serum and Urinary Androgens and Risk of Breast Cancer in Postmenopausal Women1

Giorgio Secreto,2 Paolo Toniolo, Franco Serrino, Camilla Recchione, Adalberto Cavalieri, Paola Pisani, Amadio Tods, Giuseppe Pariseli!, and Sergio Di Pietro Endocrinology Unit, Division of Experimental Oncology C [G. S., C. R., A. C.], Division of Epidemiology Â¡F.B., P. P., A. T.], and Division of Clinical Oncology [G. F., S. D. P.], Istituto Na:ionale Tumori, Milan. Italy; and Institute of Environmental Medicine and Kaplan Cancer Center, New York University Medical Center, New York, A'iw York [P. T.I

ABSTRACT (16, 17). Grattarola (16) also indicated that the resected ovaries show increased cellularity of the ovarian stroma. Ovarian inter Serum levels of testosterone, dihydrotestosterone, androstenedione, dehydroepiandrosterone sulfate, and sex hormone-binding globulin and stitial cells secrete mainly androstenedione and testosterone, urinary levels of testosterone and androstanediol were compared in 75 the immediate precursors for their aromatization products es women with breast carcinoma and 150 age-matched healthy controls. trone and estradiol at the ovarian and peripheral levels. The Odds ratios for quartiles of hormones, adjusted for known potential data of Grattarola are in agreement with previous observations confounders, were computed using conditional logistic regression. Risk by Sommers (18) that women dying of breast cancer had a of breast cancer was positively associated with levels of all androgens in considerably higher prevalence of ovarian stromal hyperplasia serum and urine but appeared stronger for testosterone (for trend, P = than women dying of causes other than cancer. 0.03) and dehydroepiandrosterone sulfate (for trend, P = 0.06) in serum We hypothesized that conditions causing increased cellularity and for testosterone (for trend, /' = 0.001) and androstanediol (for trend, of the ovarian stroma, such as follicular cyst, might alter the /' = 0.04) in urine. The adjusted odd ratios for high versus low quartiles balance of steroid hormones by permanently increasing the were 2.7 (95% confidence interval, 1.1-6.5) for serum testosterone, 2.8 (1.1-7.4) for dehydroepiandrosterone sulfate, 4.7 (1.8-12.1) for urinary amount of androgenic substrates that are available for ovarian testosterone, and 3.4 (1.4-8.7) for urinary androstanediol. These obser or peripheral aromatization or for direct action at the target vations suggest that endogenous androgenic hormones may play an level and thereby increase the risk of breast cancer. If this is important role in the epidemiology of postmenopausal breast cancer in true, and if one assumes that steroid hormones act as promoters women. rather than inducers, serum levels of androgens of ovarian origin, as well as levels of their metabolites in blood and urine, should be elevated in non-oophorectomized women with breast INTRODUCTION cancer, even at a late stage in the natural history of the disease Several lines of research suggest that reproductive hormones such as at, or well after, the time of clinical diagnosis. influence breast cancer risk. Specifically, breast cancer risk is Recently, we reported the results of a case-control study in known to be related to major hormone-dependent events in a which a strong positive association between increasing serum woman's life, and early oophorectomy has been shown to be and urine levels of testosterone and breast cancer was observed strongly protective (1). Furthermore, oral contraceptive use and in premenopausal women (19). In the present article, we report replacement estrogen therapy appear to increase risk (2), and the results of a parallel case-control study conducted in post antiestrogens appear to be effective chemotherapeutic agents menopausal women. Its scope was to evaluate whether women (3). However, consistent evidence indicating specific abnormal with localized breast carcinoma, as compared to women free of ities of the endocrine system has not yet been obtained. cancer, had higher serum levels of androstenedione and testos During the past several years, we have focused our attention terone, as well as of DHT\ the major metabolite of testosterone on the role of androgens in breast cancer. This interest was at the cellular level, and increased urinary testosterone and stimulated by a series of early studies by our group (4, 5) in androstanediol excretion. We were interested also in evaluating which we showed that premenopausal and postmenopausal the association between breast cancer and serum levels of the breast cancer patients, as well as women with epithelial hyper- carrier protein sex hormone-binding globulin, as well as the plasia of the breast (6), excrete excessive amounts of urinary major androgen of adrenal origin, DHEAS. testosterone and androstanediol. We also reported positive associations between serum testosterone and androstenedione MATERIALS AND METHODS levels and premenopausal (7) and postmenopausal (8) breast cancer. These findings are in agreement with those of other Study Subjects and Data Collection. Postmenopausal women residing investigators (9-11), including studies in which multiple sam in the Milan metropolitan area were studied. Eligible were postmeno pling (12, 13) was used, although some reports were inconclu pausal women 69 years or younger who had not been bilaterally oopho- sive (14, 15). rectomized, had a medical history negative for cancer and for major We suggested that the observed hyperandrogenism is of ovar endocrinological conditions, and did not report the use of hormonal ian origin by showing that ovariectomy (in advanced disease) is preparations of any type during the 3 months preceding enrollment. followed by rapid normalization of androgen excretion levels Cases were 75 eligible patients referred to the Breast Clinic of the Outpatient Department of the Italian National Tumor Institute, in Received 11/19/90; accepted 3/8/91. Milan, for breast examination, who subsequently received surgical The costs of publication of this article were defrayed in part by the payment treatment for primary breast cancer, between October 1983 and Decem of page charges. This article must therefore be hereby marked advertisement in ber 1985. We considered only women with histologically confirmed accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' Supported by research Grant 86/02609/44. Progetto Finalizzato Oncologia, invasive tumors 4 cm in diameter or smaller who had no clinical from the Italian National Research Council (Consiglio Nazionale delle Ricerche). P. T. received financial support from Grants CAI3343 and CAI6087 from the 'The abbreviations used are: DHT, dihydrotestosterone; DHEAS, dehydro National Cancer Institute. epiandrosterone sulfate; EGF, epidermal growth factor; OR, odds ratio; SHBGb, 3To whom requests for reprints should be addressed, at the Istituto Nazionale sex hormone-binding globulin binding capacity; SI IHd,,. sex hormone-binding Tumori, via Venezian 1, 20133 Milano, Italy. globulin, direct measurement. 2572

symptoms other than a lump and no evidence of mÃ©tastasesother than Table 1 Distribution afeases and controls and breast cancer odds ratios for to the ipsilateral regional lymph nodes. All of the 75 patients contacted potential confounders agreed to collaborate. Risk indicator confidence (n =75)32385641123272231717182315451585116232230Controls(n=150)C4087 interval0.3-1.00.1-0.80.5-3.80.5-2.70.4-2.30.04-0.70.3-1.60.5-2.80.5-2.60.5-2.30.5-3.20.4-2.50.4-3.60.5-2.10.7-2.6 Cases were individually matched to two controls by age (Â±6months). andlevelOccupationBlue The 150 controls were recruited from among women free of cancer collarWhite undergoing cervical screening at the Outpatient Department of the collarProfessionalMarital (23 same hospital and waiting in a reception area adjacent to the breast (13713384143 clinic. Within a few days of obtaining pathological confirmation of a statusEver case diagnosis, the first two women listed in the roster of daily appoint marriedNever marriedNo. ments for cervical screening satisfying the eligibility requirements and ofchildrenNulliparous123+Age the age-matching criterion were invited to collaborate. In the rare instance one refused the next eligible woman in the waiting room was (28 contacted. (3448 Cases and controls were interviewed at the time of the initial contact at firstfull-termpregnancy which, for cases, was the time of first clinical detection of a suspicious (yr)<2525-29>29NulliparousAge breast lesion. To obtain information concerning sociodemographic characteristics and known breast cancer risk indicators, we used a short 130383292251610727484953)R.01.51.3.0.7.0.1).9).2.0).7.2.2.0.1.3.0.0.2.0.0.395% questionnaire which was developed for this study. All interviews were administered in a waiting area inside the hospital by the same inter (vr)Â°<1212-14>14Ageat menarche viewer who was aware of the status of a subject case or control. At the end of the interview, all study members were given plastic containers and received oral and written instructions on how to collect (yr)<4545-52>52Bodyat menopause one complete 24-h specimen of urine. Collection was to be done as soon as possible and, for cases, always prior to hospital admission for surgery. When the study members had completed the urine collection, massindex<21.8321.83-24.88>24.88Cases they were asked to return to the hospital between 9 and 11:30 a.m. of the same morning to deliver the samples and to have 20 ml of peripheral venous blood drawn. The exact time of blood sampling was recorded. Whole blood was immediately transferred to glass tubes without anti " One control missing. coagulant. All biological specimens were labeled using codes and were delivered to the laboratory for serum separation and temporary storage mined by the two-tailed Mantel-Haenszel x2 test. Pearson product- at -20Â°C. Laboratory Tests. Hormones in urine were measured by gas chro- moment correlation coefficients were used to investigate colinearity among covariates. All statistical tests were considered significant at the matography according to the method of Mauvais-Jarvis et al. (20), with minor modifications. Circulating hormones were measured by radio- P < 0.05 level. Conditional logistic regression (23) was used to obtain estimates of immunoassay using commercial kits purchased from Biomerieux (Char- OR simultaneously adjusted for all the potential confounders listed bonnier les Bains, France) for testosterone, DHT, and androstenedione above. A multiple confounder model was developed for each exposure and from Sciavo (Milan, Italy) for DHEAS. Sex hormone-binding variable by entering all confounders at once and then proceeding to globulin was measured both directly by radioimmunoassay (SHBGd) eliminate single variables sequentially from the model, using a stepwise and by its binding capacity (SHBGb). SHBGd was quantitated using technique, if their deletion produced minimal changes (<20%) in the commercial kits purchased from Farmos Group Ltd. (Oulunsalo, Fin estimated exposure effect. Invariably, these were also the least signifi land), and SHBGb was quantitated by evaluating its ability to bind cant terms, as determined by their standardized regression coefficients DHT. The binding protein saturated with [1,2-3H]DHT (Biodata, (P > 0.10 by Wald's test) (23). Tests of log linearity for continuous Milan, Italy) was separated by the ammonium sulfate technique of variables were accomplished by means of likelihood ratio comparisons Rosner (21). of the multiple logistic model including an ordinal categorization of Serum testosterone, androstenedione, and DHT were measured after the continuous variable to that including the corresponding nonordinal ether extraction followed by partition chromatography on a celile categorization (23). None of the continuous variables in the study column. Elution was carried out stepwise using 5 ml of isooctane showed evidence of an appreciable departure from log linearity. (androstenedione fraction), followed by 6 ml of 6% ethyl acetate in To assess the possible influence of circadian fluctuations, models for isoctane (DHT fraction) and then 6 ml of 20% ethyl acetate in isooctane serum were evaluated with and without the inclusion of time at blood (testosterone fraction). collection (before 10 a.m., after 10 a.m.). To take into account differ The coefficients of intra- and interassay variation were 9.7 and 11.0% ences in the completeness and accuracy of urine collection, models for for testosterone, 7.9 and 12.3% for DHT, 6.1 and 13.5% for androstene urine were evaluated with and without the inclusion of urinary creati- dione, 6.3 and 6.8% for DHEAS, 3.5 and 6.2% for SHBGd, and 3.5 nine and volume of urine (as continuous variables). To account for the and 5.2% for SHBGb, respectively. possible influence of time since menopause (1-4, 5-8, 9-12, or 13+ Statistical Analyses. Potential confounders were known risk indica years), models were evaluated with and without the inclusion of this tors for breast cancer. These included a three-level indicator of social variable. None of these variables exerted any appreciable influence on status (low, medium, high income), based on the most representative risk estimates. occupation(s) in a woman's family (her own or her spouse's) prior to retirement, marital status (ever or never married), age at menarche, parity, age at first full-term pregnancy, age at menopause, and Quetelet RESULTS index (kg/m2), based on reported weight and height. The median age of cases was 59.0 years and of controls was Continuous exposure variables (testosterone, DHT, DHEAS, andro stenedione, SHBGb, and SHBGd in serum and testosterone and andro- 58.5 years. The distribution of cases and controls according to stanediol in urine) were categorized using the quartiles of the frequency potential confounders is summarized in Table 1. Fewer cases distributions of cases and controls combined. Initially, contingency (6.7%) than controls (15.3%) were members of families of tables were used statistical analyses to relate variables to the risk of higher social status, and more cases than controls were unmar malignancy. For contingency tables, the age-adjusted relative risk, as ried (14.7 versus 8.7%). More controls (18.7%) than cases estimated by the OR (22), was used. Statistical significance was deter- (4.0%) reported having given birth to 3 or more children. There 2573

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1991 American Association for Cancer Research. ANDROGENS AND BREAST CANCER Table 2 Missing values for hormone-related variables in cases and controls SerumCasesMeanSDMedianRangeMissing SHBGb SHBGd (pg/ml)250.2119.0243.0560.0219.4106.6203.0772.0DHT(pg/ml)75.151.263.0362.062.432.055.5170.0Androstenedione(pg/ml)715.3441.7546.02265.08576.2299.6518.02037.011DHEAS(ng/ml)(pg/ml)1134.8733.71001.33597.28903.0575.7776.13347.811206.051.4202.2232.03216.071.8205.0500.45409.9165.1400.0830.010450.3230.0407.51451.014UrineTestosterone (ngDHT/lOOml) (Pg/ml)54.041.845.0302.0338.936.531.0323.07Androstanediol(PB/ml)611.5628.3370.04137.02521.4723.0298.05306.09

(no.)ControlsMeanSDMedianRangeMissing

(no.)Testosterone

Table 3 Breast cancer OR within quarliles of hormone-related variables in serum were only minor differences between cases and controls in age at first full-term pregnancy, age at menarche, age at menopause, confidence test(/>)0.030.180.200.060.490.53and Quetelet index. (n =75)161618251520162415161422111718211818221416181813Controls(n=150)404038323740383536363730403434313636324133343039ORÂ°1.01.01.32.11.01.31.01.61.01.10.92.31.01.92.72.71.00.91.40.71.01.01.10.6OR*1.01.21.52.71.01.61.32.01.00.80.72.01.02.72.82.81.01.11.30.71.01.51.80.895%interval*0.5-3.00.6-3.71.1-6.70.7-3.70.5-3.10.8-5.00.3-2.10.3-1.90.8-4.91.0-7.31.0-7.71.1-7.40.5-2.50.6-3.10.3-1.60.6-3.60.6-5.00.3-2.1TrendThe distribution of levels of hormones in serum and urine Testosterone(pg/ml)<146146-212213-275>275DHT among cases and controls is presented in Table 2. Mean and median levels of serum testosterone, DHT, androstenedione, and DHEAS and urinary testosterone and androstanediol were consistently higher in cases than in controls. Mean and median levels of SHBGd were slightly lower among cases; there were (pg/ml)<3636-5738-82>82Androstenedione no differences in SHBGb. Serum testosterone, DHT, androstenedione, and DHEAS and urinary testosterone were positively intercorrelated among cases and controls. In particular, serum testosterone was highly (pg/ml)<389389-530531-765>765DHEAS correlated with DHT, DHEAS, and androstenedione. Pearson product-moment correlation coefficients were 0.66, 0.50, and 0.59 among cases and 0.59, 0.20, and 0.52 among controls, respectively. The coefficients between DHT and DHEAS and (ng/ml)<497.9497.9-826.0826.0-1268.8>1 between DHT and androstenedione were 0.60 and 0.53 among cases and 0.44 and 0.38 among controls, respectively. Those between DHEAS and androstenedione were 0.47 among cases 268.8SHBGb and 0.39 among controls. Urinary testosterone was highly correlated with DHEAS among cases (0.34) and controls (0.53) ml)<167.9168.0-202.2202.3-247.9>247.9SHBGd(^gDHT/100 and with androstenediol among controls (0.44) but less so among cases (0.17). SHBGb and SHBGd were highly intercor related (0.93 among cases, 0.91 among controls). Age was negatively, although only slightly, correlated with all exposure (pg/ml)<293294-400401-528>528Cases variables among both cases and controls. The age-adjusted and covariate-adjusted ORs for quartiles of hormones in serum are presented in Table 3. Occupation and number of children showed an appreciable negative confound Â°Age-adjusted. ing effect on the ORs for testosterone and DHEAS and were * Adjusted for occupation and number of children. forced in all adjusted models. The adjusted ORs for testosterone increased progressively and significantly (for linear trend, P = 0.03) to 2.7 (95% CI, 1.1-6.5). The adjusted ORs for DHT and Table 4 Breast cancer OR within quartiles of hormone-related variables in urine androstenedione also doubled (to 2.0) in the highest quartile, but there were no statistically significant trends. For DHEAS, confidence test adjusted ORs increased sharply to 2.7 (1.0-7.3) in the second interval*0.5-2.90.8-5.71.8-12.11.0-6.00.7-4.61.4-8.7Trend =75)1114173011191825Controls(n=150)4338372642353529OR"1.01.02.25.01.02.22.33.2OR*1.01.22.24.71.02.41.93.495%(n<0.0010.03lowest quartile and flattened out to 2.8 in the two highest Testosterone(pg/ml)<1818-3132-46>46Androstanediol quartiles. No associations were evident for SHBGb and SHBGd. A highly significant trend of increasing risk (P < 0.001) was observed for urinary testosterone (Table 4). The OR, adjusted for occupation and number of children, increased progressively to 4.7 (1.8-12.1) in the highest quartile. Adjusted ORs for (pg/ml)<170170-281282-583>583Cases(n androstanediol showed a similar pattern of increase, reaching 3.4 (1.4-8.7) in the highest quartile (for trend, P = 0.03). To evaluate confounding or effect modification between " Age-adjusted. serum testosterone and DHEAS, between these two hormones * Adjusted for occupation and number of children. and androstenedione or DHT, and between urinary testosterone 2574

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 1991 American Association for Cancer Research. ANDROGENS AND BREAST CANCER and androstanediol, models were developed which always in had been present for months or, likely, for years. Therefore, cluded occupation and number of children and two hormone one cannot dismiss the possibility that high androgen levels are variables. The effects of serum testosterone and DHEAS were contributed entirely or in part by the disease, either through marginally reduced after reciprocal adjustments. Serum testos direct secretion of hormones by the tumor or as a consequence terone and DHEAS were not evidently affected by adjustments of the emotional stress on the endocrine system induced by the for androstenedione and DHT. Urinary testosterone was mar diagnosis. Contrary to this view, we previously showed that ginally influenced by adjustment for androstanediol, but ORs breast cancer patients with elevated urinary testosterone levels for androstanediol were reduced drastically toward unity after at the time of diagnosis experienced a dramatic decrease in the adjustment for urinary testosterone (ORs 1.5, 1.1, and 1.6 for excretion levels of this hormone after surgical removal of the increasing quartiles, respectively). ovaries, while no change was observed in patients with normal levels at the time of diagnosis prior to oophorectomy (5, 17). Histological examination of the excised ovaries revealed hyper- DISCUSSION plasia of interstitial cells in all hyperandrogenic patients (5), The results of this study confirm previous indications of thus suggesting that the ovaries, rather than the tumor, may positive associations between serum and urinary levels of an- have been responsible for the observed androgenic excess. Sec drogens and breast cancer risk in postmenopausal women and ond, in all epidemiological studies conducted in hospital set substantiate our recently reported findings in premenopausal tings, interpretation of the results is limited by the inherent women (19). Positive associations were apparent for all andro- lack of comparability between cases and controls who may be selected or self-selected on the basis of the factor(s) under gens in serum (testosterone, androstenedione, DHT, and DHEAS) and in urine (testosterone and androstanediol). investigation. Indeed, in our study, cases and controls were Results related to serum and urinary testosterone are in drawn from different populations, as indicated by the greater agreement with previous observations in Caucasian postmeno percentage of controls than cases (15.3 versus 6.7) in the highest pausal women made by us (5, 7, 24) and by other investigators social stratum (Table 1). This may reflect social differences (9, 12, 13). Elevated plasma testosterone levels (10), as well as between women attending the breast and cervical cancer clinics. non-protein-bound (free) testosterone (25), were found also in While one is the most popular among a few breast clinics in Japanese postmenopausal breast cancer patients as compared the city, thus attracting many women of middle or low income to non-cancer controls. Similar findings were reported by Chen who could hardly afford the cost of a private visit, the second serves upper middle-class women more than many other screen et al. (11) among Chinese postmenopausal women. No associ ations between testosterone or androstenedione and breast can ing centers in the metropolitan area. In our data, social class, as defined by occupation, was associated both with the disease cer were reported in two small studies by Jones et al. (26), in and with the exposure (women in the "professional" group had the United Kingdom, by Malarkey et al. (14), in the United States, and in the only prospective study reported thus far (15). lower levels of most androgens than other women). Thus, social Overall, the majority of epidemiological studies known to us class was a confounder. Indeed, it was an important, negative show a tendency toward more elevated levels of testosterone confounder, one which tended to obscure the associations be and/or androstenedione in breast cancer patients than in con tween androgens and breast cancer. After adjusting for occu trols. These conclusions are further, but indirectly, supported pation, residual confounding may still have been present. It by reports of (a) elevated plasma testosterone levels in healthy seems likely, however, that residual confounding would have Caucasians as compared to Japanese and South African black contributed to reducing, rather than inflating, the association. women (27), two groups experiencing a considerably lower risk Finally, adjustment for body weight and body mass index in for breast cancer than white women, and (b) about 53% higher statistical analyses should have reduced the possibility that plasma concentrations of testosterone in British women com differences in body fat may explain differences in risk, since pared with Chinese women of all age groups.4 This latter study androgen levels are positively related to weight and body fat also showed that, within China, weight-adjusted plasma testos distribution (30, 31). terone was significantly correlated with between-county varia It is not known why increased levels of androgenic hormones tion in breast cancer mortality. might be associated with increased risk of postmenopausal As reviewed by Zumoff (28), several investigators have re breast carcinoma. The source of androgens may be the intersti ported plasma levels of DHEAS or of its urinary metabolites tial tissue of the ovary which will continue secreting relevant amounts of androstenedione and testosterone throughout a inversely associated with breast cancer, findings which are in woman's lifetime, as we have hypothesized. Alternatively, the conflict with the results reported in this study. However, the evidence is not consistently in favor of an inverse association adrenal may contribute to, or even may be solely responsible in postmenopausal women. Several reports indicate normal for, the observed hyperandrogenism. A role for adrenal secre levels of DHEAS in blood and urine (13, 26) or even increased tion is suggested by the recent findings reported by Gordon et plasma levels (28). Recently, high dehydroepiandrosterone and al. (29), as well as by our own data. The association does not DHEAS levels were reported to be associated with a substan seem to operate through known reproductive risk indicators for tially increased risk of postmenopausal breast cancer in a case- breast cancer or through increased body weight or body mass. control study in which blood donations had been collected 9 or Androgens may directly stimulate breast epithelium through more years prior to diagnosis (29). binding to androgen receptors (32) or they may act indirectly Before causal implications for the findings from this study after aromatization to estrogen in peripheral tissues, including can be considered, the methods used warrant some discussion. breast fat and breast cells (33, 34). Alternatively, it could be First, blood and urine specimens were obtained from breast hypothesized that androgens stimulate the synthesis of EGF cancer patients at the time of clinical diagnosis, when the tumor and, possibly, of other growth factors inside breast epithelium and thus play a role in the autocrine and paracrine growth 4 T. J. Key el al., manuscript in press. regulation of human breast cancer. The ability of mammary 2575

premenopausal women. Cancer Res.. 49: 471-476. 1989. epithelium to synthesize EGF under androgen modulation 20. Mauvais-Jarvis. P.. Charransol. G.. and Bobas-Masson, F. Simultaneous could be inferred on the basis of the large amounts of EGF determination of urinary androstanediol and testosterone as an evaluation of found in human breast cysts (35-37), especially in those lined human androgenicity. J. Clin. Endocrinol. Metab.. .J6:452-459, 1973. 21. Rosner. W. A simplified method for the quantitative determination of testos- with apocrine epithelium (38). It was suggested that women terone-estradiol-binding globulin activity in human plasma. J. Clin. Endocri with apocrine cysts may be at increased risk of breast cancer nol. Metab.. 34: 983-989. 1972. (39). Foci of apocrine metaplasia are frequently present in 22. Mantel, N., and Haens/el. W. Statistical aspects of the analysis of data from retrospective studies of disease. J. Nati. Cancer Inst., 22: 719-748. 1959. normal breast tissue of women with breast cancer (40). 23. Breslow, N. E.. and Day. N. E. Statistical Methods in Cancer Research. Vol. I. The Analysis of Case-Control Studies. Lyon, France: International Agency for Research on Cancer, 1980. REFERENCES 24. Grattarola. R., Secreto, G., and Recchione, C. Androgens in breast cancer. III. Breast cancer recurrences years after mastectomy and increased andro- 1. Kelscy, J. L.. and Bcrkuwit/. G. S. Breast cancer epidemiology. Cancer Res.. geniv activity. Am. J. Obstet. Gynecol., 121: 169-172, 1975. 48: 5615-5623. 1988. 25. Takatani. O.. Kosano. H.. Okumoto. T., Akamatsu. K.. Tamakuma, S., and 2. Key. T. J. A., and Pike, M. C. The role of oestrogens and progcstagens in Hiraide. H. Distribution of estradiol and percentage of free testosterone in the epidemiology and prevention of breast cancer. Eur. J. Cancer Clin. sera of Japanese women: preoperative breast cancer patients and normal Oncol., 24: 29-43. 1988. controls. J. Nail. Cancer Inst.. 79: 1199-1204. 1987. 3. Love. R. R. Prospects for antiestrogen chemoprevention of breast cancer. J. 26. Jones. D. L.. and James. V. H. T. Determination of dehydroepiandrosterone Nati. Cancer Inst.. 82: 18-21. 1990. and dehydroepiandrosterone sulfate in blood and tissue studies of normal 4. Grattarola. R. Androgens in breast cancer. I. Atypical endometrial hyperpla- women and women with breast or endometrial cancer. J. Steroid Biochem.. sia and breast cancer in married premenopausal women. Am. J. Obstet. 26: 151-159. 1987. Gynccol., 116:423-428. 1973. 27. Hill. P.. Garbaczewski. L.. Kasumi, F.. Kuno, K., Helman, P., and Wynder, 5. Grattarola. R.. Secreto. G.. Recchione. C., et al. Androgens in breast cancer. E. L. Breast cancer: diet and hormone metabolism. In: M. C. Pike, P. K. II. Endometrial adenocarcinoma and breast cancer in married postmenopau- Siiteri. and C. W. Welsh (eds.), Banbury Report. Vol. 8, Hormones and sal women. Am. J. Obstet. Gynecol., 118: 173-181, 1974. Breast Cancer, pp. 257-266. Cold Spring Harbor. NY: Cold Spring Harbor 6. Secreto. G.. Fariselli, G.. Bandieramonte. G.. et al. Androgen excretion in Laboratory. 1981. women with a family history of breast cancer or with epithelial hyperplasia 28. Zumoff. B. Hormonal profiles in women with breast cancer (review). Anti- or cancer of the breast. Eur. J. Cancer Clin. Oncol., 19: 5-9, 1983. cancer Res.. * 627-636. 1989. 7. Secreto. G.. Toniolo, P.. Bcrrino. F.. Recchione. C., Di Pietro, S.. Fariselli. 29. Gordon. G. B.. Bush. T. L.. Helzlsouer. K. J., Miller, S. R., and Comstock, G., and Decarli, A. Increased androgenic activity and breast cancer risk in G. W. Relationship of serum levels of dehydroepiandrosterone and dehydro premenopausal women. Cancer Res.. 44: 5902-5905. 1984. epiandrosterone sulfate to the risk of developing postmenopausal breast 8. Secreto. G.. Recchione. C.. Cavalieri. A., et al. Circulating levels of testoster cancer. Cancer Res.. 50: 3859-3862. 1990. one, 17Â¡Ã-oestradiol.luteinisinghormone and prolactin in postmenopausal 30. Evans. D. J.. Hoffman. R. G.. Kalkhoff, R. K.. et al. Relationship of breast cancer patients. Br. J. Cancer. 47: 269-273. 1983. androgenic activity to body fat topography, fat cell morphology, and meta 9. Adami. H. O., Johansson, E. D. B.. Vegelius. J., and Victor, A. Serum bolic aberrations in premenopausal women. J. Clin. Endocrinol. Metab., 57: concentrations of estrone, androstenedione. testosterone and sex-hormonc- 304-309. 1983. 31. Kirschner. M. A.. Samojilik. E., Drejka, M.,etal. Androgen-estrogen metab binding globulin in postmenopausal women with breast cancer and in age matched controls, l'psala J. Med. Sci.. 84: 259-275, 1979. olism in women with upper body versus lower body obesity. J. Clin. Endo crinol. Metab.. 70:473-479. 1990. 10. Hill, P.. Garbac/ewski. L.. and Kasumi. F. Plasma testosterone and breast cancer. Eur. J. Cancer Clin. Oncol.. 21: 1265-1266. 1985. 32. Bryan. R. M.. Mercer. R. J.. Bennett. R. C., Rennie, G. C, Lie, T. H., and Morgan, F. J. Androgen receptors in breast cancer. Cancer (Phila.), 54: 11. Chen.patients F., with Wang, breast S., cancer. Dai, Q.. Chin. et al. Med. A study J."(Engl. on plasma Ed.). sex 98: hormone 507-510. levels 1985. in 2436-2440. 1984. 33. Siiteri, P. K. Review of studies on estrogen biosynthesis in the human. Cancer 12. McFadyen, I. J.. Forrest. A. P. M., Prescott. R. J.. Groom. G. V.. Golder. Res., 42 (Suppl.): 3269s-3273s. 1982. M. P.. and Fahmy. D. R. Circulating hormone concentrations in women with breast cancer. Lancet. /: 1100-1102. 1976. 34. Bradlow, H. L. A reassessment of the role of breast tumor aromatization. Cancer Res.. 42 (Suppl.): 3382s-3386s, 1982. 13. Adlercreutz. H.. Hamalainen. E.. Gorbach. S. L.. Goldin. B. R.. Woods. M. 35. Jaspar. J. M.. and Franchimont, P. Radioimmunoassay of human epidermal N., and Dwyer, J. T. Diet and plasma androgens in postmenopausal vegetar growth factor in human breast cyst fluid. Eur. J. Cancer Clin. Oncol., 27: ian and onmivorous women and postmenopausal women with breast cancer. 1343-1348. 1985. Am. J. Clin. Nutr.. 49: 433-442. 1989. 36. Wang, D. Y., Hamed, H., Mockridge, C. I., and Fentiman. I. S. Radio- 14. Malarkey, W. B.. Schroeder. L. L., Stevens. V. C.. James. A. G., and LÃ¡Ã±ese. immunoassayable insulin-like growth factor-I in human breast cyst fluid. R. R. Twenty-four hour preoperative endocrine profiles in women with benign Eur. J. Cancer Clin. Oncol.. 25: 867-872. 1989. and malignant breast disease. Cancer Res.. 37: 4655-4659, 1977. 37. Collette. J.. Hendrick. J. C.. Jaspar. J. M., and Franchimont, P. Presence of 15. \\ysowski, D. K.. Comstock. G. W.. Helsing. K. J.. and Lau, H. L. Sex alpha-lactalbumin, epidermal growth factor, epithelial membrane antigen, hormone levels in serum in relation to the development of breast cancer. and gross cystic disease fluid protein (15,000 Dallons) in breast cyst fluid. Am. J. Epidemiol.. 125: 791-799, 1987. Cancer Res.. 46: 3728-3733. 1986. 16. Grattarola. R. Ovaricctomy alone or in combination with dexamethasone in 38. Boccardo. F.. Valenti. G.. Zanardi. S., Cerruti, G., Passio, T., Bruzzi. P., De patients with advanced breast cancer and high levels of testosterone excretion. Franchis, V., Barreca, A., Del Monte, P., and Minuto, F. Epidermal growth J. Nati. Cancer. Inst., 56: 11-19, 1976. factor in breast cyst fluid: relationship with intracystic cation and androgen 17. Secreto. G., and Zumoff. B. Paradoxical effects associated with supranormal conjugate content. Cancer Res.. 48: 5860-5863, 1988. urinary testosterone excretion in premenopausal women with breast cancer: 39. Dixon, J. M.. Lumsden, A. B., and Miller, W. R. The relationship of cyst increased risk of postmastectomy recurrence and higher remission rate after type to risk factors for breast cancer and the subsequent development of ovaricctomy. Cancer Res., 43: 3408-3412. 1983. breast cancer in patients with breast cystic disease. Eur. J. Cancer Clin. 18. Sommers. S. C. Endocrine abnormalities in women with breast cancer. Lab. Oncol.. 21: 1047-1050, 1985. Invest., 4: 160-174, 1955. 40. Mazoujian. G.. Pinkus. G. S., Davis, S., et al. Immunochemistry of a gross 19. Secreto, G., Toniolo, P.. Pisani. P., Recchione, C.. Cavalieri. A., Fariselli. cystic disease fluid protein (GCDFP-I5) of the breast. Am. J. Pathol., 770: G., Totis. A.. Di Pietro. S., and Berrino. F. Androgens and breast cancer in 105-112. 1983

2576

Giorgio Secreto, Paolo Toniolo, Franco Berrino, et al.

Cancer Res 1991;51:2572-2576.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/51/10/2572

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/51/10/2572. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.