Melatonin and Breast Cancer: A Prospective Study

Transcription

1 Melatonin and Breast Cancer: A Prospective Study Ruth C. Travis, Diane S. Allen, Ian S. Fentiman, Timothy J. Key Background: Experimental data from animals suggest a protective role for the pineal hormone melatonin in the etiology of breast cancer, but results from the few retrospective case control studies that examined the association in humans have been inconsistent. To determine whether low levels of endogenous melatonin are associated with an increased risk for developing breast cancer, we conducted a prospective nested case control study among British women. Methods: Concentrations of 6-sulfatoxymelatonin, the main metabolite of melatonin in urine and a validated marker of circulating melatonin levels, were measured by radioimmunoassay in 24-hour urine samples collected from women shortly after enrollment in the prospective Guernsey III Study. Levels of 6-sulfatoxymelatonin were compared among 127 patients diagnosed with breast cancer during follow-up and among 353 control subjects, matched for age, recruitment date, menopausal status, and day of menstrual cycle for premenopausal women or number of years postmenopausal for postmenopausal women. Associations were examined by analyses of covariance and conditional logistic regression. All tests of statistical significance were two-sided. Results: No statistically significant differences in urinary 6-sulfatoxymelatonin concentrations were observed between women who developed breast cancer and control subjects among premenopausal or postmenopausal women (P.8 and P.9, respectively). When data from premenopausal and postmenopausal women were combined in a multivariable analysis adjusted for potential confounders and grouped into three categories defined by 6-sulfatoxymelatonin tertiles of control subjects, the level of 6-sulfatoxymelatonin excreted was not statistically significantly associated with the risk of breast cancer (odds ratio [OR] for breast cancer 0.95, 95% confidence interval [CI] 0.55 to 1.65, comparing the middle category with the lowest category of 6-sulfatoxymelatonin concentration, and OR 0.99, 95% CI 0.58 to 1.70, comparing the highest category with the lowest category). Conclusion: We found no evidence that the level of melatonin is strongly associated with the risk for breast cancer. [J Natl Cancer Inst 2004; 96:475 82] Recent research suggests that the risk of breast cancer is increased among women who work night shifts (1), may be elevated in flight attendants (2,3), and may be increased by exposure to electromagnetic fields (4,5) and to artificial light at night (6). There is also some evidence that the risk of breast cancer may be reduced among the visually impaired (7). The principal mechanism that has been proposed to account for these associations involves a protective effect of the pineal hormone melatonin (i.e., N-acetyl-5-methoxytryptamine) (8,9), with production of melatonin being suppressed by exposure to light at night or to electromagnetic fields. Consequently, individuals with certain forms of visual impairment may be less susceptible to suppression of melatonin production by light. Results from research in animal models support a protective association between melatonin and the development of breast cancer (10). Few clinical or epidemiologic studies, however, have investigated this association, and results have been inconsistent (11 20). Because melatonin levels in case patients in these studies were measured in samples collected after the diagnosis of cancer, it was not possible to assess whether melatonin levels influenced the development of the disease or whether the disease in some way modified levels of endogenous melatonin. Assessing the role of melatonin in prospective studies has become feasible in recent years with the development of methods for estimating circulating levels of melatonin by measuring 6-sulfatoxymelatonin, the main metabolite of melatonin, in urine samples. The measurement obtained has been found to be a reliable estimate of melatonin secretion for the period during which the urine sample was collected (21,22) and to be a moderately reliable estimate of melatonin production over several years (23). In this article, we present results from, to our knowledge, the first prospective study of the association between melatonin and the risk of breast cancer, which has been conducted in a cohort of British women. SUBJECTS AND METHODS Subjects Between April 25, 1977, and October 28, 1985, 5093 women living on the island of Guernsey in the British Isles were recruited into a prospective study of hormones and breast cancer, known as the Guernsey III Study. Approval was given by the local ethics committee, and each woman gave written informed consent to the study and follow-up. Height and weight were measured at interview, and a questionnaire was completed with details of reproductive history, menopausal status, and past use of oral contraceptives and other hormones. Body mass index, defined as weight in kilograms divided by height in meters squared, was calculated. Women were also screened for breast cancer by mammography. A 24-hour urine sample was collected shortly after recruitment (median number of days after recruitment 16 days; 25 th and 75 th percentiles 9 and 29 days, respectively). In premenopausal women, the sample was collected irrespective of the stage of their menstrual cycle, but the dates of onset of menses preceding and following urine collection were recorded (the Affiliations of authors: Cancer Research UK, Epidemiology Unit, University of Oxford, Oxford, U.K. (RCT, TJK); Academic Oncology Unit, Thomas Guy House, Guy s Hospital, London, U.K. (DSA, ISF). Correspondence to: Ruth C. Travis, MSc, Cancer Research UK, Epidemiology Unit, University of Oxford, Oxford OX2 6HE, U.K. ( ruth.travis@cancer.org.uk). See Notes following References. DOI: /jnci/djh077 Journal of the National Cancer Institute, Vol. 96, No. 6, Oxford University Press 2004, all rights reserved. Journal of the National Cancer Institute, Vol. 96, No. 6, March 17, 2004 ARTICLES 475

2 latter by postcard). The 24-hour urine samples were frozen and stored at 20 C. Women were eligible for this study if they were not using any exogenous sex hormones at the time of recruitment, had not previously had cancer other than nonmelanoma skin cancer, had provided a 24-hour urine specimen on a known date, and could be classified as premenopausal or postmenopausal at recruitment. Women were classified as being premenopausal if they reported at interview that they had menstruated in their usual pattern in the previous 6 months and had a cycle length not longer than 42 days; they were classified as being postmenopausal if they were naturally postmenopausal (defined as not having had a menstrual period for at least 1 year) or had undergone a hysterectomy without bilateral oophorectomy before the menopause and were aged 60 years or older at recruitment. Follow-up for the diagnosis of breast cancer was accomplished by searching pathology reports, Guernsey death certificates, and records of the Wessex Cancer Registry for the names of participants in the study. Eligible case patients were women who had been diagnosed with carcinoma of the breast between their enrollment in the Guernsey III Study and October 31, Case patients diagnosed as a result of the mammographic screening at recruitment were excluded. Patients with carcinoma in situ of the breast were included as case patients in the main analyses; case patients with carcinoma in situ were then excluded, and the analysis was repeated. Selection of women for the nested case control study was initially made at the end of October 1994 for an analysis of the association of serum estrogen concentrations with the risk for breast cancer (24,25). During January 2002, additional women who had developed breast cancer between the end of October 1994 and October 31, 2001, were identified from those recorded as having provided 24-hour urine samples at recruitment. Matched sets of a case patient and control subjects from the first analysis were studied again, and each new case patient was matched to three control subjects randomly selected from those identified as suitable on the basis of the matching criteria used in the initial study, i.e., matching on age (within 2 years), date of recruitment (within 1 year), and menopausal status (premenopausal or postmenopausal). To be consistent with matching criteria used previously (24,25), premenopausal case patients were also matched on day of the menstrual cycle that blood was collected, within 1 day in the category of 1 29 days until the next menstrual period and within 2 days in the category of 30 days or more. It was necessary to relax the matching criteria to select 30 women as control subjects for 11 premenopausal case patients. Postmenopausal case patients with a natural menopause were also matched to control subjects on number of years postmenopausal in categories of 1 2 years or 3 or more years, but this matching was relaxed in five instances where a naturally postmenopausal case patient was matched to a control subject who had had a hysterectomy. For the three case patients who had undergone a hysterectomy, control subjects were matched on this criterion when possible, but when it was not possible, the control subjects were selected to be naturally postmenopausal for 3 or more years because these three case patients were aged 60 years or older at recruitment. After a control subject was matched to a case patient, she was unavailable for matching with further case patients. Women who had previously been studied as control subjects were included again as case patients if they were subsequently diagnosed with breast cancer (n 7 women). As far as was known, control subjects had not died or had not been diagnosed with breast cancer by the date at which the case patient to which they were matched was diagnosed with breast cancer. Urine samples were successfully located for 127 case patients and 353 of their corresponding control subjects of the 158 case patients and 474 control subjects identified from the data. Hormone Assays Hormone assays were performed by Stockgrand Ltd. (at the School of Biomedical and Life Sciences, University of Surrey, Guildford, U.K.). In 2002, the 480 urine samples were thawed, separated into aliquots, and assayed. The 1.5-mL urine specimens were analyzed in nine batches, with all samples from any one case control set analyzed in the same batch. 6-Sulfatoxymelatonin was assayed by radioimmunoassay with a 125 I-labeled tracer (26). Urine samples diluted in a tricine buffer (Sigma-Aldrich, Poole, Dorset, U.K.) were incubated with a specific antiserum against 6-sulfatoxymelatonin raised in sheep (26), and then trace amounts of 125 I-labeled 6-sulfatoxymelatonin were added. The free and antibody-bound fractions of 6-sulfatoxymelatonin were separated with a dextrancoated charcoal suspension (26). The free 6-sulfatoxymelatonin fraction was precipitated with charcoal by centrifugation at 2500g at 4 C for 15 minutes, and the radioactivity was measured in a gamma counter. A standard curve was fitted from standards constructed with charcoal-stripped urine. The lower limit of detection for 6-sulfatoxymelatonin was 0.2 ng/ml. Samples were assayed in duplicate, and the duplicate results were averaged for the statistical analyses. The assays also included samples from in-house quality-control subjects, and the interassay coefficients of variation for 6-sulfatoxymelatonin were 15.2%, 7.8%, and 9.6% for control samples with low (mean value of 2.8 ng/ml), medium (mean value of 16.8 ng/ml), and high (mean value of 34.8 ng/ml) levels of 6-sulfatoxymelatonin, respectively. We included additional quality-control samples from the 24-hour urine specimens of two women (29 samples from one woman and 27 from the other), with at least three samples from each woman inserted evenly across each batch. Intra-assay and inter-assay coefficients of variation were 13.1% and 7.6%, respectively, for the first set of quality controls (mean value 8.0 ng/ml) and 29.7% and 14.0%, respectively, for the second set (mean value 10.0 ng/ml). Urinary creatinine concentrations were assayed for each sample with a kinetic Jaffé method, using a Sigma kit (Sigma- Aldrich); measurements were made in a spectrophotometer. Urinary 6-sulfatoxymelatonin levels are presented as creatininestandardized values (6-sulfatoxymelatonin concentration divided by the creatinine concentration in each sample) (21,22). Statistical Analyses Statistical analyses were performed with the Stata 7 statistical software package (27). The hormonal values were logarithmically transformed for statistical analyses to approximately normalize their frequency distributions, and geometric means and 95% confidence intervals (CIs) were calculated. All tests of statistical significance were two-sided. We compared characteristics of case patients and control subjects with the chi-square test (for categorical variables) and the two-sample t test (for 476 ARTICLES Journal of the National Cancer Institute, Vol. 96, No. 6, March 17, 2004

3 continuous variables), except in the case of length of menstrual cycle, for which variances were not equal and a t test for unequal variances was used. Analyses of covariance were used to examine whether breast cancer risk factors and other subject characteristics were associated with the level of 6-sulfatoxymelatonin excreted. Certain psychotropic medications, such as antidepressants, antipsychotics, -blockers, calcium antagonists, and nonsteroidal anti-inflammatory drugs influence melatonin levels (28). Therefore, self-reported medication at the time of recruitment was coded and assessed in relation to the level of 6-sulfatoxymelatonin in the 24-hour urine sample. Conditional logistic regression analyses were used to calculate the relative risks of breast cancer in relation to three categories of 6-sulfatoxymelatonin excretion using cut points defined by the tertiles of 6-sulfatoxymelatonin excretion among the control subjects; these analyses were adjusted for potential confounders, including age (continuous variable) as an a priori confounder. In a second analysis, we also investigated body mass index (continuous), use of any medication thought to suppress melatonin production (yes, no), duration of urine storage (continuous), first-degree family history of breast cancer at recruitment (yes, no), age at menarche (continuous), parity and age at first birth (nulliparous, parous with age at first birth 25 years, and parous with age at first birth 25 years), previous use of oral contraceptives (yes, no), previous use of other hormones (yes, no), season of urine collection (March through May, June through August, September through November, December through February), stage of menstrual cycle (early follicular, late follicular, mid-cycle, early luteal, and late luteal, defined as 22, 16 21, 12 15, 3 11, and 0 2 days before next menstrual period, respectively, among premenopausal women) and years postmenopausal (continuous; among postmenopausal women only). No information was available on alcohol consumption; therefore, no adjustment could be made for this potential confounder. Logistic regression analyses were stratified by menopausal status because mechanisms by which melatonin may influence the risk of breast cancer, such as through interaction with ovarian hormones, may be different for premenopausal and postmenopausal women. RESULTS Characteristics of Case Patients and Control Subjects Breast cancer diagnosis followed urine collection by a mean of 12.6 years among premenopausal case patients (range 0.9 to 21.8 years) and 12.4 years among postmenopausal case patients (range 3.5 to 21.3 years). The characteristics of case patients with breast cancer and control subjects are shown in Table 1. Premenopausal and postmenopausal case patients and control subjects were similar with respect to age, age at menarche, weight, height, body mass index, age at first birth, and the time for which urine samples were stored between collection and analysis. Premenopausal case patients and control subjects were also similar with respect to parity, past use of hormones other than oral contraceptives, and menstrual cycle length. Among premenopausal women, a statistically significantly lower proportion of case patients than control subjects was using medication thought to suppress melatonin production (P.02), and a statistically significantly higher proportion of case patients than control subjects had a first-degree family history of breast cancer (P.02). A greater proportion of premenopausal case patients than of control subjects had previously used oral contraceptives (72.7% versus 64.5%), although this difference was not statistically significantly (P.2). Postmenopausal case patients and control subjects were similar with respect to family history of breast cancer, past use of oral contraceptives, and age at menopause. A lower proportion of postmenopausal case patients than postmenopausal control subjects was parous, had previously taken hormones other than oral contraceptives, and was currently using medications thought to suppress melatonin production; these differences were not statistically significant. Association Between 6-Sulfatoxymelatonin and Other Variables in Control Subjects Of the 15 reproductive, lifestyle, and sample or assay characteristics examined among control subjects, we found four to be statistically significantly associated with the concentration of 6-sulfatoxymelatonin excreted in the urine sample. Tables 2 and 3 show 6-sulfatoxymelatonin excretion among control subjects by selected characteristics. 6-Sulfatoxymelatonin excretion among control subjects decreased statistically significantly with increasing age at interview (regression coefficient 0.023; P.001) and varied statistically significantly by analysis batch (P.02, after adjustment for age). After adjustment for age and analysis batch, we found no statistically significant difference in 6-sulfatoxymelatonin excretion between premenopausal and postmenopausal control subjects (P 1.0). Excretion of 6-sulfatoxymelatonin decreased statistically significantly with increasing body mass index (regression coefficient 0.024; P.04, adjusted for age and analysis batch). However, after stratification by age group (10-year intervals), this association was apparent only in women younger than 50 years. The mean concentration of 6-sulfatoxymelatonin excreted was 29.8% higher in parous women than in nulliparous women (P.03), but no statistically significant association was observed among parous women between number of children and the concentration of 6-sulfatoxymelatonin excreted (P.09; results not shown). Mean concentration of 6-sulfatoxymelatonin excreted was 15.2% lower in women using medication thought to suppress melatonin production than in those who were not using such medication, although this difference was not statistically significant (P.1). After adjustment for age at recruitment and batch of analysis, there were no statistically significant associations between 6-sulfatoxymelatonin levels and age at menarche, height, age at first birth among the parous, time between urine sample collection and assaying, previous use of oral contraceptives or other hormones, length or stage of menstrual cycle (premenopasual women only), or age at menopause or number of years postmenopausal (postmenopausal women only). Urinary 6-Sulfatoxymelatonin Excretion in Case Patients and Control Subjects Table 4 shows the geometric mean concentration of 6-sulfatoxymelatonin excreted by case patients and control subjects, adjusted for age and analysis batch and, secondly, also adjusted for other factors that may confound the relationship between 6-sulfatoxymelatonin and the risk of breast cancer. Age-adjusted and multivariable-adjusted results were Journal of the National Cancer Institute, Vol. 96, No. 6, March 17, 2004 ARTICLES 477

4 Table 1. Characteristics of case patients with breast cancer and control subjects by menopausal status Case patients (SD) Control subjects (SD) P value* Premenopausal at baseline Total No Age at interview, y 41.0 (4.9) 41.0 (4.6).9 Age at menarche, y 13.1 (1.5) 13.0 (1.4).8 Length of menstrual cycle, days 27.9 (3.8) 27.9 (5.0) 1.0 Weight, kg 63.9 (11.1) 63.6 (9.6).8 Height, cm (6.2) (6.6).5 Body mass index, kg/m (3.7) 24.5 (3.5).8 % parous Age at first birth among parous, y 24.5 (3.9) 24.4 (4.2).9 % first-degree family history of breast cancer % past hormone use, % past use of oral contraceptives % current medication use# Duration of storage for urine, y 22.3 (2.3) 22.3 (2.2).9 Postmenopausal at baseline Total No Age at interview, y 58.9 (6.6) 58.8 (6.5).9 Age at menarche, y** 13.6 (1.5) 13.3 (1.4).3 Age at menopause, y 51.9 (9.1) 51.5 (8.6).8 Time postmenopausal, y 7.8 (6.1) 8.1 (6.2).7 Weight, kg 64.0 (9.9) 63.8 (8.6).9 Height, cm (6.0) (5.8) 1.0 Body mass index, kg/m (3.5) 25.3 (3.4).9 % parous Age at first birth among parous, y 26.9 (5.1) 26.4 (5.5).6 % first-degree family history of breast cancer % past hormone use, % past use of oral contraceptives % current medication use# Duration of storage for urine, y 22.4 (1.7) 22.5 (1.6).8 *P values were from chi-square tests (for categorical variables) and two-sample t tests (for continuous variables). All t tests were two-sided. SD standard deviation. Data were available for 76 case patients and 212 control subjects. Data were available for 75 case patients and 213 control subjects. Family history of breast cancer in mother, sister, or daughter. Previous use of hormones other than oral contraceptives. Data were available for 212 control subjects. #Current use of medications thought to suppress melatonin production, including -blockers, psychotropic medications, and nonsteroidal anti-inflammatory drugs. **Data were available for 138 control subjects. Data were available for 48 case patients and 134 control subjects. Data were available for 36 case patients and 113 control subjects. Data were available for 49 case patients and 137 control subjects. very similar, partly because several of the factors adjusted for were matching criteria. The mean concentration of 6-sulfatoxymelatonin excreted was similar among premenopausal and postmenopausal case patients and control subjects. The mean concentration of 6-sulfatoxymelatonin excreted was 2% higher in premenopausal case patients than in premenopausal control subjects (P.8) and was 2% lower in postmenopausal case patients than in postmenopausal control subjects (P.9). The exclusion of one premenopausal case patient diagnosed with breast cancer less than 2 years after urine collection (and her corresponding control subjects) did not materially alter these findings. Table 2. Relationships between 6-sulfatoxymelatonin (natural logarithmic values) and selected characteristics of control subjects and their samples (continuous variables), after adjustment for the effects of age and analysis batch, where appropriate* No. of control subjects % change in 6-sulfatoxymelatonin, ng/mg creatinine (95% CI) P value Age at interview, y ( 3.2 to 1.4).001 Age at menarche, y ( 3.0 to 8.2).4 Age at first birth, y ( 0.6 to 3.0).2 Time postmenopausal, y ( 3.4 to 6.5).5 Body mass index, kg/m ( 4.7 to 0.1).04 Duration of storage for urine, y ( 8.7 to 1.5).2 *All values were adjusted for age at interview and analysis batch unless specified. CI confidence interval. P values for tests of linear trend were obtained from analyses of covariance. Values were adjusted for analysis batch. Age at first birth among parous subjects. Values were for postmenopausal women only. 478 ARTICLES Journal of the National Cancer Institute, Vol. 96, No. 6, March 17, 2004

5 Table 3. Relationships between 6-sulfatoxymelatonin (natural logarithmic values) and selected characteristics of control subjects and their samples (categorical variables), after adjustment for the effects of age and analysis batch, where appropriate* No. Mean level of 6-sulfatoxymelatonin, ng/mg creatinine (95% CI) P value Menopausal status Premenopausal (5.99 to 8.40) 1.0 Postmenopausal (5.52 to 8.97) Parity Nulliparous (4.54 to 7.02).03 Parous (6.73 to 7.97) Menstrual cycle phase Early follicular (6.25 to 12.13) 1.0 Late follicular (7.10 to 10.60) Mid-cycle (7.26 to 11.33) Early luteal (7.02 to 9.84) Late luteal (6.74 to 12.49) First-degree family history No (6.51 to 7.68) 1.0 Yes (5.28 to 9.48) Current medication use No (6.69 to 7.99).1 Yes (5.18 to 7.43) Season of urine collection Spring (6.37 to 8.86).3 Summer (6.50 to 8.68) Fall (5.92 to 8.32) Winter (5.28 to 7.32) Analysis batch (5.33 to 8.52) (5.25 to 8.76) (5.99 to 9.51) (6.68 to 10.90) (7.03 to 11.20) (6.50 to 10.29) (4.62 to 7.32) (4.09 to 6.47) (4.85 to 9.61) *All values were adjusted for age at interview and analysis batch unless specified. CI confidence interval. P values for tests of heterogeneity were obtained from analyses of covariance. Values were for premenopausal women only. Current use of medications thought to suppress melatonin production, including -blockers, psychotropic medications, and nonsteroidal anti-inflammatory drugs. Values were adjusted for age at interview. Table 5 shows the relative risk of breast cancer by three categories of 6-sulfatoxymelatonin excretion defined by levels in control subjects. When women who had melatonin levels in the highest melatonin category were compared with women who had levels in the lowest category, the ageadjusted relative risk for breast cancer associated with melatonin was not substantially different from unity for either premenopausal women (odds ratio [OR] 1.20, 95% CI 0.61 to 2.37) or postmenopausal women (OR 1.26, 95% CI 0.60 to 2.68). After adjusting for other potential confounders, the association was also not statistically significant for premenopausal women (OR 0.99, 95% CI 0.45 to 2.17) and for postmenopausal women (OR 1.09, 95% CI 0.46 to 2.60). Table 4. 6-Sulfatoxymelatonin excretion (geometric mean) among case patients and control subjects by menopausal status No. Age- and analysis batch-adjusted* 6-Sulfatoxymelatonin, ng/mg of creatinine (95% CI) P value Multivariable-adjusted 6-Sulfatoxymelatonin, ng/mg of creatinine (95% CI) P value Premenopausal Case patients (6.32 to 9.50) (6.10 to 9.16).8 Control subjects (6.36 to 8.65) 7.34 (6.30 to 8.56) Postmenopausal Case patients (4.93 to 8.51) (5.03 to 8.71).9 Control subjects (5.29 to 8.15) 6.75 (5.44 to 8.38) *Values, adjusted for age and analysis batch, were for premenopausal and postmenopausal women. CI confidence interval. Values, adjusted for age, body mass index, storage time, season of urine collection, analysis batch, parity, and use of medication thought to affect melatonin, were for premenopausal women and postmenopausal women. P values for tests of heterogeneity were from analyses of covariance. Journal of the National Cancer Institute, Vol. 96, No. 6, March 17, 2004 ARTICLES 479

6 Table 5. Association between the risk of breast cancer and categories of urinary 6-sulfatoxymelatonin excretion, by menopausal status 6-Sulfatoxymelatonin, ng/mg of creatinine* No. of case patients No. of control subjects Age-adjusted OR (95% CI) Multivariable-adjusted Premenopausal (referent) 1.00 (referent) 9.27 to (0.65 to 2.45) 1.06 (0.48 to 2.34) (0.61 to 2.37) 0.99 (0.45 to 2.17) Postmenopausal (referent) 1.00 (referent) 5.06 to (0.37 to 1.97) 0.76 (0.31 to 1.84) (0.60 to 2.68) 1.09 (0.46 to 2.60) All Low (referent) 1.00 (referent) Medium (0.65 to 1.83) 0.95 (0.55 to 1.65) High (0.75 to 2.04) 0.99 (0.58 to 1.70) *Cut points for premenopausal and postmenopausal women are defined by tertiles in control subjects. Cut points for all women for the low, medium, and high categories in the pooled analyses are defined by the tertile cut points provided above for premenopausal and postmenopausal women separately. OR odds ratio; CI confidence interval. Values were adjusted for age, body mass index, season of urine collection, duration of storage for urine, first-degree family history of breast cancer, age at menarche, ever use of oral contraceptives, ever use of other hormones, parity and age at first birth, and use of medication thought to influence melatonin production. Values for premenopausal women were additionally adjusted for menstrual cycle phase and length, and values for postmenopausal women were additionally adjusted for the number of years postmenopausal. Similar results were observed when these analyses were repeated on combined data from all women, stratified by menopausal status. When women who had melatonin levels in the middle and highest categories were compared with women who had levels in lowest melatonin category, neither the age-adjusted relative risk for breast cancer associated with melatonin (OR 1.09 [95% CI 0.65 to 1.83] and OR 1.23 [95% CI 0.75 to 2.04], respectively) nor the relative risk adjusted for other potential confounders (OR 0.95 [95% CI 0.55 to 1.65] and OR 0.99 [95% CI 0.58 to 1.70], respectively) was statistically significantly different from unity. Findings were similar when analyses were confined to invasive breast cancer only (112 case patients of a total of 127 case patients; results not shown). There were too few participants to allow meaningful analysis of whether the risk for carcinoma in situ was associated with melatonin excretion. When the hormone data were divided into four categories with cut points defined by quartiles of 6-sulfatoxymelatonin distribution among control subjects (instead of tertiles) to assess the robustness of the data, similar results were again noted. DISCUSSION To our knowledge, this is the first prospective study of the association between melatonin and breast cancer risk in humans. The results of this study do not support the hypothesis that endogenous melatonin concentration is a major factor in breast cancer etiology. Detailed data on reproductive and lifestyle characteristics of women in this study have enabled us to adjust for the effects of factors that may confound any relationship between melatonin and breast cancer risk. Of particular interest in this study were the findings of lower concentrations of melatonin excreted among nulliparous women than parous women and among women taking medications thought to suppress melatonin production than among women not taking such medications. The relationship between parity and the risk of breast cancer is well established, but there has been little research into whether melatonin and parity are associated. To our knowledge, the only report of such a study (13) showed no statistically significant difference in melatonin by the number of children, although levels of melatonin were higher among all parous women than among nulliparous women, as we also have found. Our findings concur with results of earlier research (28 32) that lower concentrations of melatonin are excreted by women taking certain medications, including nonsteroidal anti-inflammatory drugs, -blockers, calcium channel blockers, and psychotropic medications, such as antidepressants, antipsychotics, and anticonvulsants. Because we also found that use of these drugs was higher among control subjects than among case patients and because previous research has suggested that the use of some of these drugs, particularly certain nonsteroidal anti-inflammatory drugs, may modify the risk of breast cancer (33 35), the use of these medications might mask any protective effect that melatonin may have on the risk of breast cancer. Thus, it was important to make appropriate adjustments for the effects of such medication. Data on alcohol intake were not available, so we were unable to adjust for the potentially confounding effects of this factor (36). Several of the retrospective case control studies have observed lower concentrations of melatonin or metabolites of melatonin in the blood and/or urine of patients with breast cancer than in the blood and/or urine of breast cancer free women (11,15,19,20); one study (19) of the association between excretion of the main melatonin metabolite and plasma melatonin suggested that lower levels in case patients with breast cancer resulted from decreased secretion of the hormone by the pineal gland rather than increased peripheral metabolism. One early study (12) also found nighttime plasma melatonin levels to be lower in patients with estrogen receptor positive breast cancer than in those with estrogen receptor negative breast cancer and healthy control women. However, other case control studies have found that the daytime serum concentration of melatonin was higher in patients with breast cancer than in healthy control subjects (14,16), that there was no association between the risk of breast cancer and the mean daytime nadir and nighttime peak plasma concentrations (13), and that the amount of 6-sulfatoxymelatonin excreted in 24-hour urine samples from 480 ARTICLES Journal of the National Cancer Institute, Vol. 96, No. 6, March 17, 2004

7 women with malignant tumors was similar to that from women with benign breast disease (17). It is difficult to interpret the meaning of these previous findings in patients with breast cancer and breast cancer free control subjects because the disease itself, treatment, and/or changes in behavior that might occur after diagnosis or before surgery or treatment may influence melatonin concentrations in the blood. Indeed, several studies have found melatonin secretion to be positively or negatively associated with the severity of the disease in terms of tumor size (20), whether the cancer has metastasized (14,18), and whether or not the patients have primary or secondary tumors (19). Furthermore, these studies were small, and some studies did not include adequate steps to exclude the potential confounding effects of age, parity, medication use, and body mass index. Studies in animal models generally support the melatonin hypothesis, indicating that melatonin can influence the frequency and growth of spontaneous and induced tumors: a pinealectomy increases tumorigenesis and shortens survival time, whereas administration of melatonin reverses these trends and inhibits tumor growth (37). Evidence from experimental studies with breast cancer and malignant melanoma cell lines and with animals suggests that antiproliferative pathways (38), antioxidant pathways (39), and immunostimulatory mechanisms (40) may link the level of melatonin with the risk of breast cancer. Melatonin may influence the action and/or release of growth factors stimulating neoplastic growth (41,42), may interact through the endogenous opioid system (43,44), or may act indirectly through the endocrine system [e.g., altering concentrations of steroid hormones such as estradiol or modulating the activity of estrogen receptors in tumor cells (45 47)]. There are several limitations to the design of this study. Although the prospective design allows us to assess the association of melatonin with the development of breast cancer, we were unable to specifically assess associations of melatonin exposure with the risk of the initiation or progression of the disease, and we were unable to evaluate the importance of the estrogen receptor status of case patients because those data were not available. Furthermore, the moderate sample size in this study did not allow us to exclude a modest association between the concentration of 6-sulfatoxymelatonin excreted in urine and the risk of breast cancer. This study had approximately 80% power at a statistical significance level of.05 to detect a relative risk for breast cancer of 0.5 among women in the highest third of the melatonin distribution compared with those in the lowest third, but it had limited ability to detect a more modest association. In this study, we focused specifically on the total daily production of melatonin. The possibility remains that, in addition to the quantity of melatonin produced, characteristics of the rhythm of melatonin secretion, such as the differences between peak secretion during the night and the low level of production during the day or the timing of the nocturnal peak, may be important for the development of breast cancer. This study uses measurements of 6-sulfatoxymelatonin taken from one 24-hour urine sample provided by each participant. Relatively little is known, however, about the reliability of a single measurement as a marker of long-term hormone levels in an individual. In a recently published reproducibility study (23), we found an intra-class correlation coefficient of.56 for urinary 6-sulfatoxymelatonin in three repeat first morning urine samples taken over 5 years in postmenopausal women. Thus, single measurements of urinary 6-sulfatoxymelatonin appear to be sufficiently reproducible to justify their use as markers for longterm exposure to endogenous melatonin in epidemiologic studies (23). However, further reproducibility data for urinary 6-sulfatoxymelatonin are needed to validate this method. It is possible that measurement error from the use of a single sample to estimate long-term hormone levels could obscure an association between melatonin and risk of breast cancer. The strong association between the risk of developing breast cancer and levels of another endogenous hormone, estradiol, among postmenopausal women has been established by a collaborative analysis of data from several prospective studies (48). The results of individual studies were inconsistent. Such research serves as a reminder of the limited conclusions that can be drawn from results in single, albeit prospective, studies such as this one. Results of this study cannot rule out a moderate association between melatonin and the risk of breast cancer, and other prospective studies are needed to further explore the relationship. REFERENCES (1) Hansen J. Light at night, shiftwork, and breast cancer risk. J Natl Cancer Inst 2001;93: (2) Pukkala E, Auvinen A, Wahlberg G. Incidence of cancer among Finnish airline cabin attendants, BMJ 1995;311: (3) Rafnsson V, Tulinius H, Jonasson JG, Hrafnkelsson J. Risk of breast cancer in female flight attendants: a population-based study (Iceland). Cancer Causes Control 2001;12: (4) Erren TC. A meta-analysis of epidemiologic studies of electric and magnetic fields and breast cancer in women and men. Bioelectromagnetics 2001;Suppl 5:S (5) Kliukiene J, Tynes T, Andersen A. Follow-up of radio and telegraph operators with exposure to electromagnetic fields and risk of breast cancer. Eur J Cancer Prev 2003;12: (6) Davis S, Mirick DK, Stevens RG. Night shift work, light at night, and risk of breast cancer. J Natl Cancer Inst 2001;93: (7) Verkasalo PK, Pukkala E, Stevens RG, Ojamo M, Rudanko SL. Inverse association between breast cancer incidence and degree of visual impairment in Finland. Br J Cancer 1999;80: (8) Stevens RG. Electric power use and breast cancer: a hypothesis. Am J Epidemiol 1987;125: (9) Brainard GC, Kavet R, Kheifets LI. The relationship between electromagnetic field and light exposures to melatonin and breast cancer risk: a review of the relevant literature. J Pineal Res 1999;26: (10) Cohen M, Lippman M, Chabner B. Role of pineal gland in aetiology and treatment of breast cancer. Lancet 1978;2: (11) Bartsch C, Bartsch H, Jain AK, Laumas KR, Wetterberg L. Urinary melatonin levels in human breast cancer patients. J Neural Transm 1981; 52: (12) Tamarkin L, Danforth D, Lichter A, DeMoss E, Cohen M, Chabner B, et al. Decreased nocturnal plasma melatonin peak in patients with estrogen receptor positive breast cancer. Science 1982;216: (13) Danforth DN Jr, Tamarkin L, Mulvihill JJ, Bagley CS, Lippman ME. Plasma melatonin and the hormone-dependency of human breast cancer. J Clin Oncol 1985;3: (14) Lissoni P, Bastone A, Sala R, Mauri R, Rovelli F, Viviani S, et al. The clinical significance of melatonin serum determination in oncological patients and its correlations with GH and PRL blood levels. Eur J Cancer Clin Oncol 1987;23: (15) Bartsch C, Bartsch H, Fuchs U, Lippert TH, Bellmann O, Gupta D. Stage-dependent depression of melatonin in patients with primary breast cancer. Correlation with prolactin, thyroid stimulating hormone, and steroid receptors. Cancer 1989;64: Journal of the National Cancer Institute, Vol. 96, No. 6, March 17, 2004 ARTICLES 481

8 (16) Lissoni P, Crispino S, Barni S, Sormani A, Brivio F, Pelizzoni F, et al. Pineal gland and tumor cell kinetics: serum levels of melatonin in relation to Ki-67 labeling rate in breast cancer. Oncology 1990;47: (17) Skene DJ, Bojkowski CJ, Currie JE, Wright J, Boulter PS, Arendt J. 6-Sulphatoxymelatonin production in breast cancer patients. J Pineal Res 1990;8: (18) Falkson G, Falkson HC, Steyn ME, Rapoport BL, Meyer BJ. Plasma melatonin in patients with breast cancer. Oncology 1990;47: (19) Bartsch C, Bartsch H, Bellmann O, Lippert TH. Depression of serum melatonin in patients with primary breast cancer is not due to an increased peripheral metabolism. Cancer 1991;67: (20) Bartsch C, Bartsch H, Karenovics A, Franz H, Peiker G, Mecke D. Nocturnal urinary 6-sulphatoxymelatonin excretion is decreased in primary breast cancer patients compared to age-matched controls and shows negative correlation with tumor-size. J Pineal Res 1997;23:53 8. (21) Cook MR, Graham C, Kavet R, Stevens RG, Davis S, Kheifets L. Morning urinary assessment of nocturnal melatonin secretion in older women. J Pineal Res 2000;28:41 7. (22) Graham C, Cook MR, Kavet R, Sastre A, Smith DK. Prediction of nocturnal plasma melatonin from morning urinary measures. J Pineal Res 1998;24: (23) Travis RC, Allen NE, Peeters PH, van Noord PA, Key TJ. Reproducibility over 5 years of measurements of 6-sulphatoxymelatonin in urine samples from postmenopausal women. Cancer Epidemiol Biomarkers Prev 2003; 12: (24) Thomas HV, Key TJ, Allen DS, Moore JW, Dowsett M, Fentiman IS, Wang DY. A prospective study of endogenous serum hormone concentrations and breast cancer risk in post-menopausal women on the island of Guernsey. Br J Cancer 1997;76: (25) Thomas HV, Key TJ, Allen DS, Moore JW, Dowsett M, Fentiman IS, et al. A prospective study of endogenous serum hormone concentrations and breast cancer risk in premenopausal women on the island of Guernsey. Br J Cancer 1997;75: (26) Aldhous ME, Arendt J. Radioimmunoassay for 6-sulphatoxymelatonin in urine using an iodinated tracer. Ann Clin Biochem 1988;25(Pt 3): (27) Stata Statistical Software, Release 7.0. College Station (TX): Stata Corp., (28) Levallois P, Dumont M, Touitou Y, Gingras S, Masse B, Gauvin D, et al. Effects of electric and magnetic fields from high-power lines on female urinary excretion of 6-sulfatoxymelatonin. Am J Epidemiol 2001;154: (29) Murphy PJ, Myers BL, Badia P. Nonsteroidal anti-inflammatory drugs alter body temperature and suppress melatonin in humans. Physiol Behav 1996; 59: (30) Murphy DL, Garrick NA, Tamarkin L, Taylor PL, Markey SP. Effects of antidepressants and other psychotropic drugs on melatonin release and pineal gland function. J Neural Transm Suppl 1986;21: (31) Stoschitzky K, Sakotnik A, Lercher P, Zweiker R, Maier R, Liebmann P, et al. Influence of beta-blockers on melatonin release. Eur J Clin Pharmacol 1999;55: (32) Davis S, Kaune WT, Mirick DK, Chen C, Stevens RG. Residential magnetic fields, light-at-night, and nocturnal urinary 6-sulfatoxymelatonin concentration in women. Am J Epidemiol 2001;154: (33) Johnson TW, Anderson KE, Lazovich D, Folsom AR. Association of aspirin and nonsteroidal anti-inflammatory drug use with breast cancer. Cancer Epidemiol Biomarkers Prev 2002;11: (34) Cakir Y, Plummer HK 3rd, Tithof PK, Schuller HM. Beta-adrenergic and arachidonic acid-mediated growth regulation of human breast cancer cell lines. Int J Oncol 2002;21: (35) Moorman PG, Grubber JM, Millikan RC, Newman B. Antidepressant medications and their association with invasive breast cancer and carcinoma in situ of the breast. Epidemiology 2003;14: (36) Stevens RG, Davis S, Mirick DK, Kheifets L, Kaune W. Alcohol consumption and urinary concentration of 6-sulfatoxymelatonin in healthy women. Epidemiology 2000;11: (37) Tamarkin L, Cohen M, Roselle D, Reichert C, Lippman M, Chabner B. Melatonin inhibition and pinealectomy enhancement of 7,12- dimethylbenz(a)anthracene-induced mammary tumors in the rat. Cancer Res 1981;41: (38) Karasek M, Pawlikowski M. Antiproliferative effects of melatonin and CGP Biol Signals Recept 1999;8:75 8. (39) Reiter RJ. The role of the neurohormone melatonin as a buffer against macromolecular oxidative damage. Neurochem Int 1995;27: (40) Maestroni GJ. The immunoneuroendocrine role of melatonin. J Pineal Res 1993;14:1 10. (41) Cos S, Blask DE. Melatonin modulates growth factor activity in MCF-7 human breast cancer cells. J Pineal Res 1994;17: (42) Lissoni P, Barni S, Meregalli S, Fossati V, Cazzaniga M, Esposti D, et al. Modulation of cancer endocrine therapy by melatonin: a phase II study of tamoxifen plus melatonin in metastatic breast cancer patients progressing under tamoxifen alone. Br J Cancer 1995;71: (43) Esposti D, Lissoni P, Tancini G, Barni S, Crispino S, Paolorossi F, et al. A study on the relationship between the pineal gland and the opioid system in patients with cancer. Preliminary considerations. Cancer 1988;62: (44) Lissoni P, Barni S, Tancini G, Fossati V, Figerio F. Pineal-opioid system interactions in the control of immunoinflammatory responses. Ann N Y Acad Sci 1994;741: (45) Danforth DN Jr, Tamarkin L, Lippman ME. Melatonin increases oestrogen receptor binding activity of human breast cancer cells. Nature 1983;305: (46) Molis TM, Spriggs LL, Hill SM. Modulation of estrogen receptor mrna expression by melatonin in MCF-7 human breast cancer cells. Mol Endocrinol 1994;8: (47) Baldwin WS, Barrett JC. Melatonin: receptor-mediated events that may affect breast and other steroid hormone-dependent cancers. Mol Carcinog 1998;21: (48) The 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 2002;94: NOTES Supported by Cancer Research UK and the Lloyds TSB Charitable Foundation for the Channel Islands. We thank John Hayward and the late Dr. Richard Bulbrook for the initiation, design, and management of the series of studies in Guernsey made over many years; the women of the island who volunteered to take part; Drs. Bryan Gunton-Bunn and David Jeffs, and the staff at The Greffe for assistance in follow-up; and Clare Hobson for help with separating urine samples into aliquots. Manuscript received September 5, 2003; revised January 20, 2004; accepted January 28, ARTICLES Journal of the National Cancer Institute, Vol. 96, No. 6, March 17, 2004