Prolactin Determinants in Healthy Women: A Large Cross-Sectional Study within the EPIC Cohort

Prolactin is a polypeptide hormone mainly implicated in growth and differentiation of breast epithelial cells during different stages of the reproductive life history of women (1). It is synthesized primarily by the lactotrophs of the anterior pituitary and the secretion of prolactin is suggested to be controlled by various prolactin-releasing factors (including estrogen and progesterone) and prolactin-inhibiting factors (dopamine; refs. 1, 2).

Recent epidemiologic data suggest that higher circulating prolactin levels are associated with increased risk of breast cancer (3–5), and several potential breast cancer risk factors have been examined in relation to circulating prolactin levels (6–9), although these latter results are inconsistent. Regarding reproductive factors, it is well known that women who have experienced a full-term pregnancy (FTP) have lower levels of prolactin compared with women who have not been pregnant (6, 10, 11). The impact of other factors such as the number of FTPs, breastfeeding, previous use of oral contraceptives, age at and years since menopause, lifestyle factors, and adiposity indices on circulating prolactin levels is less clear. Furthermore, results regarding the effect of postmenopausal hormone therapy use on the blood prolactin levels are inconsistent.

We therefore investigated in a comprehensive manner the cross-sectional associations between several reproductive and lifestyle risk factors for breast cancer and circulating levels of prolactin in healthy pre- and postmenopausal women, taking into account the use of postmenopausal hormone therapy (HT) at the time of blood donation. The study population consisted of control subjects from a nested case–control study on breast cancer and prolactin levels within the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort.

The EPIC is a multicenter cohort study designed to investigate the relationships between diet, metabolic factors and cancer, consisting of approximately 360,000 women and 150,000 men aged mostly between 25 and 70 years. All participants were enrolled between 1992 and 2000 from 23 centers in 10 European countries: Denmark, France, Germany, Greece, Italy, Norway, Spain, Sweden, the Netherlands, and the United Kingdom. Extensive details about the cohort population and data collection procedures at baseline are given elsewhere (12). Briefly, participants provided questionnaire information, covering dietary habits and other lifestyle factors, history of previous illness, and invasive surgical operations. Women provided detailed information about their menstrual cycle, menopausal status, reproductive history, and use of oral contraceptives and postmenopausal hormone therapy. In addition to the questionnaire assessments, measurements of height, weight, and waist and hip circumferences were performed according to standardized protocols, and approximately two thirds of study participants also provided a blood sample for the laboratory analyses (12).

Women included in this analysis were selected as control subjects in a large case–control study on breast cancer and prediagnostic prolactin levels nested within the EPIC cohort (n = 2,578). Eligible were women who could be clearly classified as being either premenopausal or postmenopausal at the time of blood donation, did not have any previous diagnosis of cancer (except nonmelanoma skin cancer), and had given information on postmenopausal exogenous hormone use at blood donation. Sweden was not included in the analysis because of incomplete data on exogenous hormone use. One woman with a prolactin value above the detection limit of the assay and 17 women with missing prolactin values were excluded from the study, leaving a total of 2,560 subjects for analyses.

Women who were ≤42 years of age or reported having had at least 9 menstrual periods in the last 12 months were classified as premenopausal. Women were classified as postmenopausal when they reported not having any menses over the past 12 months, when older than 55 years of age (if the menstrual cycle history was missing), or when reporting bilateral ovariectomy. Women aged 42 years or older whose questionnaire data on menopausal status was incomplete were not included to the present study.

Information on postmenopausal exogenous hormone use was derived from country-specific baseline questionnaires covering never, previous, and current use of hormone therapy together with the brand name and duration of use. Current users of hormone therapy were subgrouped according to therapy regimens as estrogen alone users and combined estrogen–progestin users (estrogen with continuous or sequential daily use of progestin).

Reproductive factors of interest included in this study were age at menarche (<12, 12–14, ≥15 years), parity (never vs. ever), number of full-term pregnancies (FTP, sum of live and stillbirths, 1, 2 ≥3), age at first FTP (<25 vs. ≥25), years since first FTP (<20, 20–24, 25–29, 30–34, ≥35 years), years since last FTP (<15, 15–19, 20–24, 25–29, ≥30 years), breastfeeding (yes vs. no), duration of first breastfeeding [never (parous), 0.1–3, 4–8, ≥9 months], cumulative duration of breastfeeding (number of pregnancies multiplied by the mean of each breastfeeding duration, <1, 1–3, 4–8, 9–12, ≥13 months), previous use of oral contraceptives (never vs. previous), previous duration of oral contraceptives (<5, 5–9, ≥10 years), age at menopause (<45, 45–49, 50–54, ≥55 years), years since menopause (<5, 5–9, ≥10 years), previous use of hormone therapy (never vs. previous), duration of hormone therapy (previous and current, <1, 1–3, ≥4 years), and type of hormone therapy (estrogen alone, combined estrogen–progestin). Lifestyle and anthropometric factors of interest included smoking status (never, former, current), lifetime mean alcohol use (<6, 6–11, 12–23, ≥24 g/day), alcohol use at recruitment (0, 0.1–5, 6–10, 11–15, ≥16 g/day), a physical activity score (inactive, moderately inactive, moderately active, active), practice of vigorous physical activity (yes vs. no), body mass index, (BMI, <25, 25–29, ≥30 kg/m²), waist circumference (<72, 72–77, 78–85, ≥86 cm), hip circumference (<95, 95–99, 100–104, ≥105 cm), body weight (<57, 57–62, 63–70, ≥71 kg), and height (<157, 157–160, 161–165, ≥166.0 cm).

The prolactin analyses were performed in the laboratory of the Division of Cancer Epidemiology at the German Cancer Research Centre (DKFZ) and determined by immunoradiometric assay [IRMA (CT)]. All samples were analyzed at the same timeframe. The same quality controls were included in each analytic batch and laboratory personnel were blinded to the status of samples and quality controls. The detection range of the assay was 0.35 to 133 ng/mL. The mean inter- and intra-assay coefficients of variation were 4.62% and 2.17%, respectively.

All statistical analyses were conducted separately for 3 groups: premenopausal women, postmenopausal women using hormone therapy, and not using hormone therapy at the time of blood donation. Levels of prolactin were logarithmically transformed to normalize the distribution. Partial correlations between prolactin and age at blood donation, years since menopause, and endogenous hormones (e.g., logarithmically transformed estrogen, testosterone, progesterone, DHEAS, SHBG, IGFI) were calculated using the Spearman correlation coefficient. To examine the effect of age at prolactin measurement, a linear regression model with separate slopes and intercepts for pre- and postmenopausal women was used.

Adjusted geometric mean prolactin levels by categories of various factors, together with their 95% confidence intervals (95% CI), were calculated using linear regression. All analyses were adjusted for recruitment center, age (continuous), and time at blood donation (before 10 am, 10 am–1 pm, after 1 pm), fasting status (<3, 3–6, >6 hours), smoking status (never, former, current, unknown), and parity (never, ever). Years since first and last FTP were additionally adjusted for number of FTPs, and in postmenopausal women, adjustments for years since menopause were performed for age-related factors (age at blood donation, age at menopause). Adjustments for other factors such as phase of menstrual cycle in premenopausal women and BMI did not alter the results and were not included to the final model. The percent difference in the adjusted geometric means for the highest versus lowest category of established significant factors was calculated as (e^b − 1) × 100. Difference of adjusted geometric mean among exposure categories (P_cat) was assessed by the type III test using F statistics obtained from linear regression models. Linear trends (P_trend) between prolactin levels and quantitative variables were tested by fitting linear regression models with continuous exposure variables. Most of the studied factors had unknown values for some women (usually less than 2%, except “age at menopause” and “years since menopause” among women who used currently hormone therapy where nearly 50% of the data were missing) and those individuals were excluded from the analyses in relation to these factors. The precise number of missing data for each studied factor can be calculated on the basis of the information provided in columns labeled “n (%)” in each table. All statistical tests were two sided, and P < 0.05 was considered statistically significant. Statistical analyses were conducted using SAS software, version 9.2 (SAS Institute).

All EPIC study participants provided written consent at the time of recruitment to participate in the study. The study was approved by local ethics committees in participating countries/study centers and by the institutional review board of the International Agency for Research on Cancer (IARC, Lyon, France).

The study population consisted in total of 2,560 healthy control subjects, of which 596 were classified as premenopausal and 1,964 as postmenopausal, including 900 postmenopausal women who were using hormone therapy at the time of blood donation (Table 1). Age at menarche, ever having had a FTP, and age at first FTP were similar between these groups. Postmenopausal women who were using hormone therapy were slightly younger, had lower BMI, lower waist circumference, and higher lifetime mean alcohol consumption compared with the postmenopausal women who did not use hormone therapy. Moreover, a higher percentage of hormone therapy users were previous users of oral contraceptives with a longer median duration of use compared with nonusers of hormone therapy. Geometric mean prolactin levels were significantly higher in premenopausal women and in women who were using postmenopausal hormone therapy compared with women who did not currently use hormone therapy (adjusted geometric means 8.2, 7.0, and 6.3 ng/mL, respectively; P_cat ≤ 0.0001 for premenopausal vs. postmenopausal women, and P_cat ≤ 0.0001 for postmenopausal hormone therapy users vs. nonusers, Table 1; Fig. 1).

In analysis evaluating associations between prolactin levels and characteristics at the time of blood donation (Supplementary Table S1), circulating levels of prolactin were generally highest in blood taken before 10 am and lowest around mid-day (10 am–1 pm), regardless of menopausal status and postmenopausal hormone therapy use. Among postmenopausal women, prolactin levels decreased with increasing age; however, this effect was eliminated when controlling for years since menopause in our models. Moreover, postmenopausal women who were current smokers had up to 20% lower levels of prolactin compared with never smokers. In premenopausal women, prolactin levels were not significantly associated with age at blood donation, fasting, and smoking status but varied modestly, depending on the phase of the menstrual cycle, with the highest levels in the ovulatory and luteal phase of the cycle. This variation, however, was not statistically significant (P_cat = 0.16).

With regard to reproductive factors in all three groups, parous women had significantly lower prolactin levels compared with nulliparous women (Table 2). This effect was much greater in premenopausal women (geometric mean prolactin levels comparing parous vs. nulliparous women: 8.61 vs. 10.95 ng/mL; P_cat = 0.0002) than in postmenopausal women with or without hormone therapy (geometric mean prolactin levels comparing parous vs. nulliparous women: 6.24 vs. 7.07 ng/mL, P_cat = 0.01 among hormone therapy users, and 5.42 vs. 5.89 ng/mL, P_cat = 0.04 among hormone therapy nonusers, respectively). Moreover, a higher number of FTPs was significantly associated with lower levels of prolactin (prolactin levels comparing ≥3 vs. 1 FTP; 7.77 vs. 9.49 ng/mL, P_trend = 0.01 among premenopausal women; and 5.91 vs. 6.58 ng/mL, P_trend = 0.01 among postmenopausal women using hormone therapy), although the trend test was not significant for postmenopausal women without hormone therapy (prolactin levels comparing ≥3 vs. 1 FTP, 5.55 vs. 6.02 ng/mL, P_trend = 0.38). Prolactin levels did not vary significantly by age at menarche, ages at and years since first or last FTP, ever breastfeeding, duration of breastfeeding or previous use and duration of oral contraceptives in either premenopausal or postmenopausal women.

Regarding postmenopausal characteristics (Table 3), higher age at menopause and fewer years since menopause were associated with slightly higher levels of prolactin; however, after mutual adjustment these associations were both eliminated. A tendency for higher levels of prolactin with increasing duration of hormone therapy use was also found. In subgroup analyses stratified by the type of hormone therapy regimen, circulating prolactin levels were significantly higher among women who currently used combined estrogen–progestin therapy compared with the women who used estrogens alone (geometric mean prolactin levels 6.66 and 5.90 ng/mL, P_cat = 0.001 Table 3; Fig. 2).

Prolactin levels were generally not associated with anthropometric and lifestyle factors, except among women using hormone therapy, where inverse trends of circulating prolactin levels with BMI (P_trend = 0.06), waist circumference (P_trend = 0.08), body weight (P_trend = 0.01), body height (P_trend = 0.09), and hip circumference (P_trend = 0.002) were found (Supplementary Table S2). No significant variation with lifetime mean and baseline alcohol consumption and various physical activity categories were observed in any of the groups (Supplementary Table S3).

All correlations between prolactin and other endogenous hormones such as estradiol, calculated free estradiol, estrone, progesterone, testosterone, free testosterone, DHEAS, SHBG, C-peptide, and IGFI were lower than 0.14 among pre- and postmenopausal women (data not shown).

In this large cross-sectional study, we observed that postmenopausal women who used hormone therapy at the time of blood donation had significantly higher levels of circulating prolactin compared with women who did not currently use hormone therapy. However, these higher levels seemed to be confined to those postmenopausal women who used combined estrogen–progestin therapy, and not to women who used estrogens alone. Regardless of the hormone therapy use (yes or no, combined or estrogen-only), postmenopausal women overall had lower levels of prolactin than premenopausal women. Our results also showed lower circulating prolactin levels among parous women compared with nulliparous women in all groups, although this effect was much greater in premenopausal women (27.2% lower levels) than in postmenopausal women with or without hormone therapy (13.2% and 8.6% lower levels, respectively). Furthermore, the magnitude of this reduction was associated with the number of FTPs. In contrast, prolactin levels did not vary by age at menarche, ages at or years since first and last FTP, breastfeeding, previous use, and duration of oral contraceptives and lifestyle or anthropometric factors, and were not correlated with other endogenous hormones.

Our finding of a higher level of circulating prolactin among postmenopausal women who were currently using hormone therapy compared with the women who did not use hormone therapy is generally in line with findings from a few other large studies (3, 13). However, whether these higher levels of prolactin depend on the composition/type of the used hormone therapy regimen is less clear. Our results are consistent with studies showing elevated levels of prolactin in postmenopausal women who used combined estrogen–progestin therapy (13–15), although some studies showed no significant effect of combined hormone therapy on blood prolactin levels (16–18). Likewise, in other studies the use of estrogens alone has shown to either increase (19, 20) or not to have any significant influence on circulating prolactin levels (16, 21, 22). In contrast with the postmenopausal hormone therapy use, prolactin levels are shown not to be largely influenced by oral contraceptive use in healthy premenopausal women (23).

Despite inconsistent results with regard to the various hormone therapy regimens on the circulating prolactin levels in women, it has been long proposed that estrogens stimulate prolactin secretion. However, the pattern of estrogen, progesterone, and prolactin changes during puberty (24), the menstrual cycle (25), pregnancy (26), breastfeeding (8), and the transition into menopause (25) provides evidence for a joint role of estrogens and progesterone in regulating the pituitary synthesis and release of prolactin (27). Despite hypothesized associations between prolactin and other hormones (28–30), we and others (31, 32) did not observe correlations between prolactin and endogenous hormones (i.e., estradiol and progesterone). The mechanisms behind prolactin surges are poorly understood but it has been proposed that in addition to the direct action of steroids on the lactotrophs, steroid-induced reduced dopaminergic activity likely also increases prolactin release (28, 33).

Recent prospective studies show that higher circulating prolactin levels are associated with an increased risk of breast cancer among postmenopausal women (3–5). Given the relatively large number of women with history of hormone therapy use and the association of prediagnostic prolactin levels with subsequent breast cancer risk more pronounced among women using hormone therapy (5), the current study has an importance also from a clinical perspective. However, further studies are needed to evaluate the effects of dose, exact formulation, specific mode of administration, and duration of hormone therapy on circulating prolactin levels (34–36).

Regarding reproductive factors, in line with numerous previous reports, we found that prolactin levels among parous women were lower than those among nulliparous women (6, 9–11). Furthermore, the observed gradual decrease in prolactin levels with increasing number of FTP is in agreement with some studies (11, 37), but not with other studies, where the parity related decrease was only related to the first FTP (6, 9, 10). Mechanisms of the reduced prolactin secretion after pregnancy are currently still unclear. Possibly, long-term changes in secretion and metabolism of estrogens together with suggested increased endogenous dopaminergic activity following pregnancy may help to account for the reduction in prolactin secretion in parous women (38–40). Furthermore, consistent with previous study, we also found that parity mediated variations in prolactin levels were more pronounced in premenopausal women compared with postmenopausal women (6). As prolactin declines after menopause (6, 16, 25), this most likely also reduces the difference observed in prolactin levels between nulliparous and parous postmenopausal women (6).

As observed in numerous previous studies, ages at which women had their first or last FTP and time passed since deliveries were not associated with the variation in circulating prolactin levels (6, 8, 10, 11, 39), indicating permanent alterations in the postpregnancy prolactin response which do not depend on ages at births. With respect to breast cancer, in contrast, decreased risk of breast cancer in parous women correlates with age at first FTP (41). Although the pituitary gland is the major source of prolactin synthesis, extrapituitary prolactin produced locally in other tissues, including breast tissue, potentially acts as an autocrine–paracrine factor within the breast (42). However, less is understood about the role of extra pituitary prolactin during normal physiology and about mechanisms that control expression of prolactin at extra pituitary sites.

Because of a strong influential role of prolactin on the proliferation process of the lobuloalveolar epithelium of the breast during pregnancy and lactation (1), it would be plausible that breastfeeding influences prolactin levels beyond the effect of pregnancy. Indeed, a few small-scale studies with relatively young cohorts of women (median age 29 and 39 years) showed that prolactin was lowered by breastfeeding duration for the first child, with no substantial effect on feeding subsequent children (8, 9). However, the clear null result with respect to breastfeeding duration and prolactin levels in our study and another large prospective study (6) suggests that the association is diminished with time, if present at all. Nevertheless, it should be noted that the median total breastfeeding duration in our study population was 6.5 months and this relatively short duration and little variability across our population might result underestimation of the effect of breastfeeding on prolactin levels. Moreover, although prolactin is important in breast development and levels modestly increase during puberty (43), age at menarche has not been found to influence prolactin levels in current or in other studies (6, 11).

The lack of variation in prolactin levels in relation to physical activity or alcohol consumption suggests limited associations between prolactin and breast cancer–related lifestyle risk factors (7, 9, 44). However, the lower prolactin response related to smoking exposure confirmed in our study is believed to be the consequence of nicotine induced activation of dopamine secretion and its antiestrogenic action, which inhibit prolactin release among smokers (45, 46).

Obesity-related anthropometric parameters were inversely linearly associated with circulating prolactin in one large study (47) and among postmenopausal hormone therapy users in our study, but not in several other studies (9, 11, 44, 48). High serum prolactin has been also associated with favorable glucose metabolic profile and lower prevalence of diabetes (49, 50), which taken together could provide some evidence of obesity-related prolactin actions; however, substantially more data are needed to explore this relationship before conclusions can be drawn.

The strength of this study includes its extensive collection of information on reproductive and lifestyle factors and a large sample size, which allowed us to stratify the analyses by menopausal status and by postmenopausal hormone therapy use at the time of blood donation. In addition, our inter- and intra-assay correlations between prolactin measurements were low, showing a good validity of the data. The analyses presented here have some limitations. Our study has a cross-sectional design that cannot establish causality of the relationship seen between prolactin and studied factors. Prolactin has a strong circadian pattern, which could make the interpretation of single blood prolactin value rather complex. However, the observed circadian pattern together with other factors was controlled in our analyses. Because of no or limited information on family history of breast cancer, acute psychological stress and use of antidepressants/thyroid hormones in the EPIC study, we were not able to address the influence of these factors on prolactin levels.

In conclusion, our large-scale study shows that current hormone therapy use at the time of blood donation is associated with the higher circulating prolactin levels in postmenopausal women, with the effect confined to the women who were using combined estrogen–progestin therapy. Our study also confirmed previous findings indicating the permanent reduction of prolactin secretion after FTP. Moreover, the magnitude of this reduction depended on the number of FTPs. No major variation in prolactin levels was seen in relation to the other reproductive and lifestyle risk factors for breast cancer, with the exception of lower levels among postmenopausal women who were current smokers compared with never smokers.

No potential conflicts of interest were disclosed.

This research did not receive any specific grant from any funding agency in the public, commercial, or not-for-profit sector.

Conception and design: K. Tikk, L. Dossus, A. Tjønneland, K. Overvad, R. Tumino, A. Barricarte, H.B. Bueno-de-Mesquita, E. Weiderpass, K.-T. Khaw, R. Kaaks

Development of methodology: K. Tikk, T. Johnson, R. Tumino, E. Weiderpass

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Johnson, F. Clavel-Chapelon, A. Tjønneland, K. Overvad, H. Boeing, A. Trichopoulou, P. Lagiou, D. Trichopoulos, G. Masala, R. Tumino, C. Sacerdote, A. Mattiello, S. Sánchez, E. Molina-Montes, P. Amiano, A. Barricarte, H.B. Bueno-de-Mesquita, E.M. Monninkhof, A. Idahl, E. Lundin, E. Weiderpass, E. Lund, K.-T. Khaw, T.J. Key, R.C. Travis, R. Kaaks

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Tikk, D. Sookthai, T. Johnson, I. Romieu, A. Barricarte, N.C. Onland-Moret, E. Weiderpass, M.J. Gunter, R. Kaaks

Writing, review, and/or revision of the manuscript: K. Tikk, D. Sookthai, T. Johnson, L. Dossus, F. Clavel-Chapelon, A. Tjønneland, A. Olsen, K. Overvad, L. Baglietto, S. Rinaldi, I. Romieu, H. Boeing, A. Trichopoulou, P. Lagiou, D. Trichopoulos, G. Masala, C. Agnoli, C. Sacerdote, A. Mattiello, G. Buckland, S. Sánchez, E. Molina-Montes, P. Amiano, J.M.H. Castaño, A. Barricarte, H.B. Bueno-de-Mesquita, E.M. Monninkhof, N.C. Onland-Moret, A. Idahl, E. Lundin, E. Weiderpass, E. Lund, M. Waaseth, K.-T. Khaw, T.J. Key, R.C. Travis, M.J. Gunter, E. Riboli, R. Kaaks

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Tikk, D. Sookthai, T. Johnson, H. Boeing, G. Masala, R. Tumino, C. Sacerdote, E. Molina-Montes, A. Idahl, E. Weiderpass, K.-T. Khaw

Study supervision: R. Tumino, E. Weiderpass, R. Kaaks

The authors thank Britta Lederer and Sigrid Henke for conducting the laboratory analyses, and all the EPIC cohort participants.

The coordination of EPIC is financially supported by the European Commission (DG-SANCO) and the International Agency for Research on Cancer. The national cohorts are supported by Danish Cancer Society (Denmark); Ligue contre le Cancer, Mutuelle Générale de l'Éducation Nationale, Institut National de la Santé et de la Recherche Médicale (France); Deutsche Krebshilfe, Deutsches Krebsforschungszentrum and Federal Ministry of Education and Research (Germany); the Hellenic Health Foundation (Greece); Italian Association for Research on Cancer (AIRC) and National Research Council (Italy); Dutch Ministry of Public Health, Welfare, and Sports (VWS), Netherlands Cancer Registry (NKR), LK Research Funds, Dutch Prevention Funds, Dutch ZON (Zorg Onderzoek Nederland), World Cancer Research Fund (WCRF), Statistics Netherlands (The Netherlands); Norwegian Research Council, Norwegian Cancer Society, ERC-2009-AdG 232997 and Nordforsk, Nordic Centre of Excellence Programme on Food, Nutrition, and Health (Norway); Health Research Fund (FIS), The Spanish Ministry of Health (ISCIII RETICC RD06/0020/0091) and the Catalan Institute of Oncology, Regional Governments of Andalucía, Asturias, Basque Country, Murcia (no 6236) and Navarra, ISCIII RETIC (RD06/0020; Spain); Swedish Cancer Society, Swedish Scientific Council and Regional Government of Skåne and Västerbotten (Sweden); and Cancer Research UK, Medical Research Council (United Kingdom).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.