J Clin Oncol 25: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 25 NUMBER 19 JULY JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T From the Department of Preventive and Predictive Medicine; Scientific Directorate; and Chemoprevention Unit, Fondazione IRCCS Istituto Nazionale dei Tumori ; and the European Institute of Oncology, Milan, Italy. Submitted September 1, 2006; accepted April 9, 2007; published online ahead of print at on June 4, Supported by the Italian Association for Cancer Research (ref. 47/03) and the US National Cancer Institute, US Department of Health and Human Services, National Institutes of Health, and the Italian National Research Council. Presented in part at the Reunião do Grupo para a Epidemiologia e o Registro do Cancro nos Países de Língua Latina, May 4-6, 2005, Lisbon, Portugal. Authors disclosures of potential conflicts of interest and author contributions are found at the end of this article. Address reprint requests to Andrea Micheli, PhD, Fondazione IRCCS Istituto Nazionale dei Tumori, Via Venezian 1, Milan, Italy; andrea.micheli@istitutotumori.mi.it by American Society of Clinical Oncology X/07/ /$20.00 DOI: /JCO Plasma Testosterone and Prognosis of Postmenopausal Breast Cancer Patients Andrea Micheli, Elisabetta Meneghini, Giorgio Secreto, Franco Berrino, Elisabetta Venturelli, Adalberto Cavalleri, Tiziana Camerini, Maria G. Di Mauro, Elena Cavadini, Giuseppe De Palo, Umberto Veronesi, and Franca Formelli A B S T R A C T Purpose High endogenous testosterone is associated with increased breast cancer (BC) risk. We designed this study specifically to assess the long-term prognostic role of testosterone in a cohort of postmenopausal BC patients. Patients and Methods We considered 194 postmenopausal women, operated on for early BC (T1-2N0M0), who never received chemotherapy or hormonal therapy, and who participated in a fenretinide BC prevention trial as untreated controls. Blood samples were collected 3 months (median) after surgery; plasma samples, stored at 80 C, were radioimmunoassayed for testosterone. Median follow-up was 14 years. The main end point was any cancer event. Event-free survival was estimated by the Kaplan-Meier method. Hazard ratios (HRs) of events by testosterone level were estimated by the Cox model, adjusting for age, tumor size, and histology. Results Patients with high testosterone (, median of distribution) had significantly lower event-free survival than those with low testosterone (log-rank P.004). The adjusted HR of patients with high versus low testosterone was 2.05 (95% CI, 1.28 to 3.27). High testosterone was also associated with a significantly higher risk of BC events (relapse and second primary) with an adjusted HR of 1.77 (95% CI, 1.06 to 2.96). Eleven second primaries (non-bc) occurred in the high-testosterone group, four in the low-testosterone group. Conclusion High plasma testosterone strongly predicts poorer prognosis in postmenopausal BC patients not administered adjuvant therapy. Testosterone levels should be determined as part of the prognostic work-up. J Clin Oncol 25: by American Society of Clinical Oncology INTRODUCTION The risk of developing breast cancer (BC) is related to events of reproductive life and lifestyle factors that modify plasma levels of endogenous sex hormones. 1-2 Soon after the discovery of sex hormones, in the 1930s, it was suggested that BC risk was related to endogenous estrogen levels. 3 Later in vitro and in vivo studies showed that estrogens increase the proliferation of normal and cancerous breast epithelium cells. 4-5 In the 1970s, other research had indicated that BC risk was increased in women with excess ovarian androgen production, and also with chronic anovulation and associated reduction of luteal-phase progesterone. 6-7 Large prospective cohort studies on postmenopausal women now make it clear that high serum levels of both estrogens and androgens are associated with increased BC risk. A reanalysis of nine prospective studies found that the strengths of these associations were similar, with relative risks estimated of 2 to 3 for women in the highest quintile of hormone level compared with those in the lowest quintile 8 ; furthermore, high testosterone was a marker of increased risk irrespective of estradiol levels. 8 These results were supported by the large European Prospective Investigation into Cancer and Nutrition (EPIC) study and by the Nurses Health study Before menopause, the key hormonal imbalances that precede BC development are high serum androgens and low luteal-phase progesterone As regard prognosis, high urinary testosterone was found to be associated with poor prognosis in BC patients More recently, the finding that antiestrogen treatments are effective in preventing BC 2685

2 Micheli et al recurrence 16 has further established the role of estrogens in BC prognosis. Our own study 17 on 110 postmenopausal BC patients with 5.5 years of follow-up found that high serum testosterone was an important predictor of BC progression; serum estradiol was also found to be significantly associated with poor prognosis, but its effect disappeared after adjusting for serum testosterone. 17 We also found that high serum testosterone was related to the presence of metabolic syndrome, and that presence of metabolic syndrome was strongly associated with BC progression. 18 These data suggest that high circulating testosterone is a marker of a more general dysmetabolic condition that favors cancer development and progression. Metabolic syndrome and its components (central adiposity, impaired fasting glucose, elevated blood pressure, high triglyceride, and low high-density lipoprotein cholesterol) have been found to be associated with the occurrence and progression of several neoplasms. In 2003, after the results of our previous studies, we designed a new study to investigate the role of high testosterone in predicting new cancer events in BC patients. This new study took advantage of a larger independent set of postmenopausal BC patients with longer follow-up. The set of patients considered was the untreated control group of a clinical trial 28 performed to test the efficacy of the retinoid fenretinide in preventing BC recurrence. This article presents the findings of this new study. PATIENTS AND METHODS Study Patients We used the data on the untreated postmenopausal patients recruited at the Istituto Nazionale dei Tumori (INT; Milan, Italy) to the fenretinide BC prevention trial. 29 This trial began on March 1, 1987, and accrual closed on July 31, Trial design, protocol, and results have been published elsewhere Patients eligible for inclusion were operated on for T1N0M0 or T2N0M0 BC; did not receive adjuvant chemotherapy or hormone therapy; and had no new cancer event after surgery. At baseline, patients were assessed for metastases by chest x-ray, bone scan, blood tests for liver function, and liver ultrasonography, and, for contralateral BC, by mammography. Those recruited also had normal metabolic function tests and normal erythrocyte, leukocyte, and platelet counts. Postmenopausal status was defined as amenorrhea for more than 1 year before recruitment. Patients who had undergone hysterectomy without ovariosalpingectomy were considered postmenopausal if they were older than 50 years at enrollment. At baseline, patients donated blood for retinol measurements and other ancillary studies The samples were conserved at 80 C; no thawing accident occurred during storage. In view of the small quantities of cryopreserved plasma available and the fact that testosterone measurements are more reliable than estradiol measurements in postmenopausal women, only testosterone was assayed for this study. Material was available to determine testosterone in 194 of the 235 control patients prospectively enrolled in the fenretinide trial. 28 Samples from these 194 women (age: mean, years; range, 48 to 70 years) at recruitment had been taken at median of 2.8 months after surgery (89% within 12 months; 37% within a month); for the 10% who were recruited later, samples were taken 12 to 55 months after surgery. Life status was checked annually. At July 31, 2003 (end of follow-up), 15 years after the initiation of the trial, no cohort member had been lost to life-status follow-up. However 36 women (19%) had been lost to clinical follow-up (8% lost at 10 years). These women were not covered by a cancer registry or included in other clinical files. Clinical follow-up consisted of an annual check with mammography and chest x-ray, and a biannual visit including clinical examination and blood tests; bone scan was programmed every 18 months. 29 During the 15-year follow-up, no participant received adjuvant chemotherapy or hormone therapy. All causes of death were recorded; 46 deaths occurred, 41 resulting from cancer (39 from BC). All participants who died as a result of cancer had a new cancer event before death and were retained in the survival analysis to the date of the new event. At the end of follow-up, median potential follow-up 35 was 13.9 years (interquartile range, 12.8 to 14.8 years). The main end point was any cancer event (BC event or second primary at nonbreast site). BC events comprised local relapse, regional relapse, distant metastasis, contralateral primary BC, and ipsilateral primary BC unrelated to the first. Secondary end points were any BC event and second primary at nonbreast site. Lobular carcinoma in situ was not considered an event. Death from causes other than cancer without previous cancer event was not considered an event. Blood Sample Storage and Testosterone Assay Blood samples were collected between 9 AM and 3 PM using heparin. Plasma was separated and divided into samples, one for immediate measurements of retinol and drug levels 28,31-32 and others for long-term storage at 80 C. In early 2004, plasma samples were extracted from the freezers and testosterone determined. The samples were analyzed in batches by technicians blinded to patient life/disease status. Each batch contained samples from 40 women plus in-house quality-control plasma samples inserted at the beginning and end of each batch. Plasma testosterone was assayed in duplicate using a commercial RIA kit (Orion Diagnostica, Espoo, Finland), according to manufacturer s instructions. Kit detection limit was 0.03 ng/ml; interassay coefficient of variation was 9.6% for a mean testosterone value of ng/ml. Statistical Analysis For the main analyses, the 194-woman cohort was divided into two groups according to testosterone levels, with the median as cutoff. Differences in clinical and pathologic characteristics between these groups were tested by 2 test. For some ancillary analyses, testosterone levels were divided into tertiles. Survival analyses were performed across 15 years and included the 36 women lost to follow-up and excluded after last clinical check-up. Deaths resulting from noncancer causes were excluded at date of death. Potential follow-up was calculated by the reverse Kaplan-Meier method, excluding observations in women who died or were lost to follow-up. 35 Event-free survival (no cancer event of any kind) by testosterone level was represented by the Kaplan-Meier method. Differences between the survival curves were investigated by the log-rank test. Hazard ratios (HRs) for events were estimated, adjusting for various covariates, by Cox multivariable modeling. The proportional hazards assumption was checked by analyses of scaled Schoenfeld residuals and inspection of log-log plots. The following variables were evaluated as possible covariates: age at recruitment (continuous variable), size of first primary BC (dichotomized as pt1 v pt2 according to the 1987 TNM classification 36 ; one woman originally classified as pt1-pt2 was considered pt2), histology (infiltrating ductal carcinoma v other), time between surgery and blood sampling, and type of treatment (breast-conserving surgery v mastectomy). As expected, 37 there was a weak inverse relationship between testosterone levels and age; therefore, age at recruitment was considered a potential confounder and included in all models. Cancer size and histology had modest influences on HR estimates and were retained in the final models. By contrast, type of primary treatment and time between surgery and blood sampling did not substantially influence HRs and were excluded from final models. At the time of recruitment, hormonereceptor status was determined in a selected group of women; information was missing in 20% of the cohort and was not considered in the main analyses. To calculate the sample size, we scrutinized the results of our previous study (to our knowledge, the only one available on testosterone and BC progression) in which high testosterone had an HR of 7.2 (95% CI, 2.4 to 21.4) for BC progression versus low testosterone and 5-year event-free survival was 87% in the low testosterone group. 17 We calculated that 188 women would be sufficient to detect an HR of 2.5 for any cancer event in the high- versus low-testosterone group, with 85% power and an level of We, in fact, analyzed cancer progression over a longer period (15 years) than considered in the power estimate, resulting in a greater study power than originally assumed JOURNAL OF CLINICAL ONCOLOGY

3 Testosterone and Breast Cancer Prognosis Thus, the 194 available plasma aliquots were more than sufficient to reveal a difference according to the sample size calculation. Ninety-five percent CIs were estimated for all HRs. All P values refer to two-sided statistical tests; differences with P.05 were considered significant. The analyses were performed with the STATA statistical software package, release 7.0 (2001; Stata Corp, College Station, TX). RESULTS Table 1 shows the baseline characteristics of the 194 women dichotomized by testosterone levels. Except for cancer size, the women in the two testosterone groups were well balanced, although no differences were significant by 2 test. The 55- to 59-year age group contained the highest proportions of women (33% in the group and 35% in the group); there were few women in the youngest (45 to 49 years; 4%) and oldest (65 years; 12% and 9%) age groups. A total of 155 (75% and 85%) patients had stage I BC (pt1n0m0), 120 (62%) had infiltrating ductal carcinoma, with or without intraductal component. Most (72% and 76%) underwent breast-conserving surgery. Patients with high plasma testosterone ( the median of 0.40 ng/ml) had significantly worse (P.004) event-free survival than did patients with plasma testosterone below the median (Fig 1). The advantage for the low-testosterone group became evident from the third year of follow-up, as illustrated by the event-free survival curves (Fig 1). Five-year event-free survival was 83% in the low-testosterone group and 76% in the high-testosterone group; 10-year event-free survival was 70% versus 52%; 15-year event-free survival was 67% versus 36% (Fig 1). Survival (%) (97) (97) (86) (81) T < 0.40 ng/ml T 0.40 ng/ml Log-rank P =.004 (76) (61) Follow-Up (years) Table 2 shows all cancer events and vital status at the end of follow-up according to testosterone level. Eighty women (41%) had no event, 50 (63%) in the low- and 30 (37%) in the high-testosterone group. Sixty-three patients (32%) had a BC event: 26 (41%) in the low- and 37 (59%) in the high-testosterone group. Fifteen patients had a second nonbreast primary, four (27%) in the low- (one each stomach, lung, and kidney cancers, and one leukemia) and 11 (73%) in the (65) (43) (38) (25) (7) (5) Fig 1. Fifteen-year event-free survival by median testosterone levels in postmenopausal patients operated on for early breast cancer. Event-free survival was estimated by the Kaplan-Meier method; events considered were local relapse, regional relapse, distant metastases, ipsilateral breast cancer, contralateral breast cancer, and second primary cancer at nonbreast site. The number of women at risk appears in parentheses. T, plasma testosterone. Table 1. Clinical and Pathologic Characteristics at Recruitment of Postmenopausal Women Operated on for Early Breast Cancer in Relation to Testosterone Levels (N 194) Testosterone Concentration Characteristic No. % No. % 2 Test P Age, years Tumor size pt pt Histology IDC IDC with intraductal component ILC IDC plus ILC Other infiltrating histotype Treatment Breast-conserving surgery Mastectomy Total NOTE. Median testosterone concentration is 0.40 ng/ml. Abbreviations: IDC, infiltrating ductal carcinoma; ILC, infiltrating lobular carcinoma; pt1, 2 cm in diameter; pt2, 2-5 cm in diameter. International Union Against Cancer TNM classification

4 Micheli et al Table 2. Disease and Vital Status at the End of Follow-Up According to Plasma Testosterone at Baseline in Postmenopausal Patients Who Received Surgery for Early Breast Cancer Testosterone Concentration Patients Status No. % No. % No. % Disease status Cancer free Breast events Second primary cancer Lost to clinical follow-up (known to be alive at end of follow-up) Vital status Deceased From breast cancer From other causes Alive Total Six local relapses, 37 distant metastases, 15 contralateral breast cancers, three ipsilateral breast cancers, and two ipsilateral and contralateral breast cancers. One oral cavity cancer, one esophageal cancer, one stomach cancer (low-testosterone group), two colon cancers, one liver cancer, two lung cancers (one in lowand one in high-testosterone group), two melanomas, one endometrial cancer, one ovarian cancer, one kidney cancer (low-testosterone group), one parotid cancer, one leukemia (low-testosterone group). One lung cancer, one oral cavity cancer, one stroke, one cardiac arrest, one amyotrophic lateral sclerosis, and two unknown causes. high-testosterone group (one each oral cavity, esophageal, liver, lung, endometrial, ovarian, and parotid cancers; two each colon cancer and melanoma). Of 46 deaths 39 from BC 20 (44%) were in the lowand 26 (56%) in the high-testosterone groups. Table 3 shows age-adjusted HRs and HRs adjusted for age, tumor size, and histology, where appropriate, for different events. For all end Table 3. HRs for Events at End of Follow-Up in High- Versus Low-Testosterone Group in Postmenopausal Patients Operated on for Early Breast Cancer End Point Testosterone Level (n 97) (n 97) Any cancer event No. of events Age-adjusted HR % CI 1.23 to 3.07 Multivariable adjusted HR % CI 1.28 to 3.27 Any breast cancer event No. of events Age-adjusted HR % CI 1.01 to 2.76 Multivariable adjusted HR % CI 1.06 to 2.96 Second primary cancer at nonbreast site No. of events 4 11 Age adjusted HR % CI 1.19 to 11.8 NOTE. HRs estimated by Cox regression model. Abbreviation: HR, hazard ratio. Local relapse, regional relapse, distant metastasis, ipsilateral breast cancer, contralateral breast cancer, and second primary cancer at nonbreast site. Adjusted for age, size of first primary, and histology of first primary variables. Local relapse, regional relapse, distant metastasis, ipsilateral breast cancer, and contralateral breast cancer. points, women with high testosterone had significantly greater HRs than did low-testosterone women. In the high-testosterone group, HRs were 2.05 (95% CI, 1.28 to 3.27) for any cancer event, 1.77 (95% CI, 1.06 to 2.96) for any BC event, and 3.74 (95% CI, 1.19 to 11.80) for second primary at nonbreast site. When testosterone levels were categorized into tertiles, HRs showed a significant linear trend: The multivariable adjusted HRs for increasing tertiles were 1.0, 2.0 (95% CI, 1.09 to 3.65), and 2.41 (95% CI, 1.33 to 4.35) for any cancer event (P for the trend.004); 1.0, 1.83 (95% CI, 0.96 to 3.51), and 1.98 (95% CI, 1.03 to 3.81) for any BC event (P for the trend.041); and 1.0, 3.36 (95% CI, 0.64 to 17.6), and 5.46 (95% CI, 1.16 to 25.8) for second primary cancer at nonbreast site (P for the trend.025). DISCUSSION The results of this study are in full accord with our previous finding 17 of an association between high endogenous testosterone levels and worse prognosis for postmenopausal BC patients. In this larger cohort with longer follow-up, women with high plasma testosterone level had significantly lower 15-year event-free survival after breast surgery than did those with low testosterone. High endogenous sex hormones, particularly testosterone, are established risk factors for developing BC However, to our knowledge, no other prospective studies indicating an association between sex hormones and BC prognosis have been published (except our previous one 17 ). We investigated testosterone but not estrogens. Interpretations of the results of prospective studies that analyzed the effects of estrogens and testosterone on BC risk in the same model have been contrasting. Some studies suggested that testosterone might be more strongly associated with BC risk than estradiol 8,39 ; others indicated that the contribution of androgens to BC risk was largely due to their role as estrogen precursors JOURNAL OF CLINICAL ONCOLOGY

5 Testosterone and Breast Cancer Prognosis In this study, 11 (73%) of the 15 primary cancers at nonbreast sites developed in the high-testosterone group. In addition to a case of endometrial carcinoma, which is known to be related to both estrogens and androgens, these second primaries included seven other tumors for which a relationship with reproductive factors or sex hormones has been hypothesized, including melanoma, 43 colon cancer, 44 ovarian cancer, 45 liver cancer, 46 and lung cancer. 47 For none of these cancer sites, however, has a firm etiologic relationship with sex hormones been established. It is unclear whether testosterone per se is directly responsible for the observed poor prognosis or is just a marker of a general metabolic imbalance that provides a milieu in which cancer cell proliferation in the breast or another site is facilitated. In a recent study, 18 we found that presence of metabolic syndrome was strongly associated with BC progression, and that high serum testosterone was related to the presence of metabolic syndrome. We also found, however, that, irrespective of metabolic syndrome status, women with high testosterone had poorer prognosis, 18 indicating an independent role of testosterone in BC progression. A number of limitations of the study must be considered. During the 15-year follow-up, 36 women were lost (19%; Table 2). We expected approximately 10 deaths in this group, but know that all were alive at the end of follow-up; they therefore constitute a good prognosis subgroup compared with compliant women. Nevertheless, this subgroup showed no testosterone selection (17 had low and 19 high testosterone). The noncompliance of these patients should have no systematic effect on the event-free survival curves because dropouts were spread over entire follow-up period (8% lost at 10 years) and occurred at the same rate in the low- and high-testosterone groups. Furthermore, we found no significant differences in age, tumor size, histology, and treatment between dropouts and compliers (data not presented). We made retrospective use of data from a clinical trial designed two decades ago; as a result, another limitation is lack of systematic information on more recently established prognostic factors such as hormone receptor, HER-2, and p53 expression in cancer specimens. At the time of recruitment, hormone-receptor status was not determined in all women, and those in whom it was determined (158 women, 127 compliant) are not a random subgroup of the entire cohort. When estrogen-receptor status was included in a multivariable analysis, adjusted HRs in relation to testosterone did not change substantially (data not shown). It is also important to note that study participants did not receive adjuvant therapy. Today s hormone receptor positive patients would receive hormone therapy, and selected patients would be offered chemotherapy; the generalizability of our findings may be limited for this reason. Finally, participants were recruited with varying intervals between surgery and blood sampling. However, this did not affect our results because mean testosterone levels did not significantly change with time between surgery and blood sampling, and the time to sampling did not substantially influence HRs (data not shown). To conclude, these results corroborate our previous finding 17 that endogenous testosterone levels strongly influence the prognosis in women operated on for BC. The implication is that testosterone assays should be determined as part of the prognostic work-up in postmenopausal women with this disease. AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest. AUTHOR CONTRIBUTIONS Conception and design: Andrea Micheli, Giorgio Secreto, Franco Berrino, Elisabetta Venturelli, Giuseppe De Palo, Franca Formelli Financial support: Andrea Micheli Provision of study materials or patients: Maria G. Di Mauro, Elena Cavadini, Franca Formelli Collection and assembly of data: Tiziana Camerini, Maria G. Di Mauro Data analysis and interpretation: Andrea Micheli, Elisabetta Meneghini, Giorgio Secreto, Elisabetta Venturelli, Adalberto Cavalleri, Franca Formelli Manuscript writing: Andrea Micheli, Elisabetta Meneghini, Franco Berrino, Giuseppe De Palo, Franca Formelli Final approval of manuscript: Andrea Micheli, Elisabetta Meneghini, Giorgio Secreto, Franco Berrino, Elisabetta Venturelli, Adalberto Cavalleri, Tiziana Camerini, Maria G. Di Mauro, Elena Cavadini, Giuseppe De Palo, Umberto Veronesi, Franca Formelli REFERENCES 1. Key TJ, Verkasalo PK, Banks E: Epidemiology of breast cancer. Lancet Oncol 2: , Kelsey JL: A review of the epidemiology of human breast cancer. Epidemiol Rev 1:74-109, MacKenzie I: The production of mammary cancer in rats using oestrogens. Br J Cancer 9: , Henderson BE, Ross RK, Judd HL, et al: Do regular ovulatory cycles increase breast cancer risk? Cancer 56: , Russo J, Russo IH: Toward a physiological approach to breast cancer prevention. Cancer Epidemiol Biomarkers Prev 3: , Grattarola R: The premenstrual endometrial pattern of women with breast cancer: A study of progestational activity. Cancer 17: , Grattarola R: Androgens in breast cancer. I. Atypical endometrial hyperplasia and breast cancer in married premenopausal women. Am J Obstet Gynecol 116: , Key T, Appleby P, Barnes I, et al: Endogenous Hormones and Breast Cancer Collaborative Group: Endogenous sex hormones and breast cancer in postmenopausal women Reanalysis of nine prospective studies. J Natl Cancer Inst 94: , Kaaks R, Rinaldi S, Key TJ, et al: Postmenopausal serum androgens, oestrogens and breast cancer risk: The European prospective investigation into cancer and nutrition. Endocr Relat Cancer 12: , Eliassen AH, Missmer SA, Tworoger SS, et al: Endogenous steroid hormone concentrations and risk of breast cancer: Does the association vary by a woman s predicted breast cancer risk? J Clin Oncol 24: , Micheli A, Muti P, Secreto G, et al: Endogenous sex hormones and subsequent breast cancer in pre-menopausal women. Int J Cancer 112: , Kaaks R, Berrino F, Key T, et al: Serum sex steroids in premenopausal women and breast cancer risk within the European Prospective Investigation into Cancer and Nutrition (EPIC). J Natl Cancer Inst 97: , Eliassen AH, Missmer SA, Tworoger SS, et al: Endogenous steroid hormone concentrations and risk of breast cancer among premenopausal women. J Natl Cancer Inst 98: , Secreto G, Zumoff B: Paradoxical effects associated with supranormal urinary testosterone excretion in premenopausal women with breast cancer: Increased risk of postmastectomy recurrence and higher remission rate after ovariectomy. Cancer Res 43: , Secreto G, Zumoff B: Abnormal production of androgens in women with breast cancer. Anticancer Res 14: , Tamoxifen for early breast cancer: An overview of the randomised trials: Early Breast Cancer Trialists Collaborative Group. Lancet 351: ,

6 Micheli et al 17. Berrino F, Pasanisi P, Bellati C, et al: Serum testosterone levels and breast cancer recurrence. Int J Cancer 113: , Pasanisi P, Berrino F, De Petris M, et al: Metabolic syndrome as a prognostic factor for breast cancer recurrences. Int J Cancer 119: , Ahmed RL, Schmitz KH, Anderson KE, et al: The metabolic syndrome and risk of incident colorectal cancer. Cancer 107:28-36, Morita T, Tabata S, Mineshita M, et al: The metabolic syndrome is associated with increased risk of colorectal adenoma development: The Self- Defense Forces health study. Asian Pac J Cancer Prev 6: , Tulinius H, Sigfusson N, Sigvaldason H, et al: Risk factors for malignant diseases: A cohort study on a population of 22,946 Icelanders. Cancer Epidemiol Biomarkers Prev 6: , Park SK, Joo JS, Kim DH, et al: Association of serum lipids and glucose with the risk of colorectal adenomatous polyp in men: A case-control study in Korea. J Korean Med Sci 15: , Yamada K, Araki S, Tamura M, et al: Relation of serum total cholesterol, serum triglycerides and fasting plasma glucose to colorectal carcinoma in situ. Int J Epidemiol 27: , Muti P, Quattrin T, Grant BJ, et al: Fasting glucose is a risk factor for breast cancer: A prospective study. Cancer Epidemiol Biomarkers Prev 11: , Notarnicola M, Altomare DF, Correale M, et al: Serum lipid profile in colorectal cancer patients with and without synchronous distant metastases. Oncology 68: , Kitayama J, Hatano K, Kaisaki S, et al: Hyperlipidaemia is positively correlated with lymph node metastasis in men with early gastric cancer. Br J Surg 91: , Sako A, Kitayama J, Kaisaki S, et al: Hyperlipidemia is a risk factor for lymphatic metastasis in superficial esophageal carcinoma. Cancer Lett 208: 43-49, Veronesi U, De Palo G, Marubini E, et al: Randomized trial of fenretinide to prevent second breast malignancy in women with early breast cancer. J Natl Cancer Inst 91: , De Palo G, Camerini T, Marubini E, et al: Chemoprevention trial of contralateral breast cancer with fenretinide: Rationale, design, methodology, organization, data management, statistics and accrual. Tumori 83: , Veronesi U, Mariani L, Decensi A, et al: Fifteen-year results of a randomized phase III trial of fenretinide to prevent second breast cancer. Ann Oncol 17: , Formelli F, Camerini T, Cavadini E, et al: Fenretinide breast cancer prevention trial: Drug and retinol plasma levels in relation to age and disease outcome. Cancer Epidemiol Biomarkers Prev 12:34-41, Decensi A, Veronesi U, Miceli R, et al: Relationships between plasma insulin-like growth factor-i and insulin-like growth factor binding protein-3 and second breast cancer risk in a prevention trial of fenretinide. Clin Cancer Res 9: , Bolelli G, Muti P, Micheli A, et al: Validity for epidemiological studies of long-term cryoconservation of steroid and protein hormones in serum and plasma. Cancer Epidemiol Biomarkers Prev 4: , Rinaldi S, Déchaud H, Biessy C, et al: Reliability and validity of commercially available, direct radioimmunoassays for measurement of blood androgens and estrogens in postmenopausal women cancer. Cancer Epidemiol Biomarkers Prev 10: , Schemper M, Smith TL: A note on quantifying follow-up in studies of failure time. Control Clin Trials 17: , International Union Against Cancer: TNM Classification of Malignant Tumours (ed 4). Berlin, Germany, Springer-Verlag, Micheli A, Muti P, Pisani P, et al: Repeated serum and urinary androgen measurements in premenopausal and postmenopausal women. J Clin Epidemiol 44: , Marubini E, Valsecchi MG: Analysing Survival Data from Clinical Trials and Observational Studies (ed 2). Chichester, United Kingdom, John Wiley & Sons, Cummings SR, Lee JS, Lui LY, et al: Sex hormones, risk factors, and risk of estrogen receptor-positive breast cancer in older women: A long-term prospective study. Cancer Epidemiol Biomarkers Prev 14: , Zeleniuch-Jacquotte A, Shore RE, Koenig KL, et al: Postmenopausal levels of oestrogen, androgen, and SHBG and breast cancer: Long-term results of a prospective study. Br J Cancer 90: , Kaaks R, Lukanova A, Kurzer MS: Obesity, endogenous hormones, and endometrial cancer risk: A synthetic review. Cancer Epidemiol Biomarkers Prev 11: , Lukanova A, Lundin E, Micheli A, et al: Circulating levels of sex steroid hormones and risk of endometrial cancer in postmenopausal women. Int J Cancer 108: , Leslie KK, Espey E: Oral contraceptives and skin cancer: Is there a link? Am J Clin Dermatol 6: , Potter JD: Colorectal cancer: Molecules and populations. J Natl Cancer Inst 91: , Lukanova A, Kaaks R: Endogenous hormones and ovarian cancer: Epidemiology and current hypotheses. Cancer Epidemiol Biomarkers Prev 14:98-107, Mucci LA, Kuper HE, Tamimi R, et al: Age at menarche and age at menopause in relation to hepatocellular carcinoma in women. BJOG 108: , Liu Y, Inoue M, Sobue T, et al: Reproductive factors, hormone use and the risk of lung cancer among middle-aged never-smoking Japanese women: A large-scale population-based cohort study. Int J Cancer 117: , 2005 Acknowledgment We thank Don Ward for language assistance and the Fenretinide Breast Cancer Prevention Trial participants for their continuing cooperation JOURNAL OF CLINICAL ONCOLOGY