Introduction
In situ breast tumours account for about 20% of all breast cancers diagnosed by mammography [
1] and approximately a third of invasive breast cancers are reported to originate from breast carcinoma
in situ [
1-
3], which may be a precursor of invasive breast cancer [
4]. The focus on
in situ breast cancer, therefore, offers the advantage of exploring the associations with risk factors that are important early in the carcinogenic process (that is, prior to development of invasive disease) and early in the natural history of breast cancer, increasing our understanding of the aetiology of breast cancer. Identifying common versus distinct risk factors for
in situ and invasive disease will, in part, provide insight into the common versus distinct mechanisms through which these cancers develop. Further,
in situ breast cancers are relatively aggressively treated with surgery, radiation and/or hormone therapies [
1], underscoring the importance of identifying risk factors for this breast cancer subtype.
The relationship between circulating prolactin and invasive breast cancer risk has been investigated previously [
5,
6]. In our previous study we found a modest positive association between circulating prolactin levels and invasive breast cancer risk among postmenopausal women (odds ratio (OR)
Q4-Q1 = 1.29 (95% CI 1.05,1.58),
P
trend = 0.09) [
5].
The association between prolactin and
in situ breast cancer risk, however, has received less attention. In the Nurses’ Health Study, the only large-scale prospective study reporting estimates separately for women with invasive and
in situ lesions, the increased risk appeared to be confined primarily to invasive cancers (OR
Q4-Q1 = 1.38 (95% CI 1.11, 1.73),
P
trend = 0.0005 for invasive versus OR
Q4-Q1 = 1.16 (95% CI 0.77, 1.74),
P
trend = 0.23 for
in situ breast cancer among postmenopausal women], although there was no heterogeneity comparing
in situ versus invasive cancer (
P-value for heterogeneity (
P
het) = 0.81) [
6]. The only other study to date that included subjects with
in situ breast cancer did not provide estimates separately by
in situ versus invasive disease [
7].
Therefore, in this study we examined the association between pre-diagnostic prolactin concentrations among pre- and postmenopausal women with subsequent risk of in situ breast cancer overall, by menopausal status and by use of postmenopausal hormone therapy (HT) at the time of blood donation within the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort.
Results
This nested case-control study consisted of 307 cases diagnosed with
in situ breast cancer, and an equal number of controls. For the premenopausal case-control pairs, the median age at blood donation was 46.4 years (range 34.1 to 56.4 years), and the median age at diagnosis for the cases was 50.6 years (range 37.3 to 61.3) (Table
1). Among postmenopausal women the ages at blood donation and diagnosis were 58.1 years (range 47.7 to 71.4) and 62.8 years (range 51.0 to 78.3), respectively. The median follow-up time between blood donation and date of diagnosis was 4.2 years (range 0.02 to 10.7 years). This study included 82 postmenopausal case-control pairs who used HT at the time they provided their baseline blood sample. Out of 307
in situ cases selected to this analysis, 15 patients developed invasive breast cancer after diagnosis of
in situ disease during further follow up. The median levels of prolactin were 8.3 ng/mL (range 3.8 to 28.1 ng/mL) for patients who developed invasive disease after diagnosis of
in situ disease and 6.8 ng/mL (range 2.6 to 128.3 ng/mL) in patients who did not develop invasive disease during further follow up (data not shown).
Table 1
Baseline characteristics of
in situ
breast cancer case and control subjects (data presented as median [min, max] or
n
[%])
Age at blood donation, years | 46.4 (34.1, 56.4) | 46.5 (34.4, 55.6) | 58.1 (47.7, 70.6) | 58.0 (47.7, 71.4) |
Age at menopause, years | | | 50.0 (33.0, 59.0) | 50.0(29.0,58.0) |
Age at diagnosis, years | 50.6 (37.3, 61.3) | | 62.8 (51.0, 78.3) | |
Lag time till diagnosis, years | 4.1 (0.02, 10.0) | | 4.4 (0.02, 10.7) | |
Body mass index, kg/m2
| 23.3 (17.1, 32.8) | 24.0 (17.4, 37.2) | 24.9 (17.8, 40.8) | 24.5 (18.2, 45.9) |
Ever had a full-term pregnancy | 68 (82.9) | 70 (81.4) | 190 (86.8) | 188 (85.5) |
Baseline smoking | 15 (17.4) | 20 (23.3) | 25 (11.3) | 33 (14.9) |
Baseline use of hormone therapy | | | 82 (37.1) | 82 (37.1) |
Prolactin levels, ng/mL | 8.9 (3.4, 51.7) | 7.7 (2.9, 133.0) | 6.2 (2.6, 128.3) | 5.9 (2.6, 55.3) |
In conditional logistic regression analyses adjusting for parity, smoking status, and BMI, there was a positive association between prolactin concentrations and
in situ breast cancer risk among all women (pre-and postmenopausal combined), with a statistically significant 35% increase in risk for each one unit increase in prolactin on the continuous log
2 scale (OR
log2 = 1.35 (95% CI 1.04, 1.76), P
trend = 0.03) (Table
2). The association was similar comparing the top to the bottom tertiles of prolactin, but did not reach statistical significance (OR
Q3-Q1 = 1.37 (95% CI 0.87, 2.16)). We observed no statistically significant heterogeneity in stratified analyses by menopausal status (
P
het = 0.98). However, after additional stratification by baseline HT use among postmenopausal women, we observed a suggestive association among the HT users (OR
log2 = 1.77 (95% CI 0.98, 3.21),
P
trend = 0.06), but not among the non-users (OR
log2 = 1.20 (95% CI 0.82, 1.76), P
trend = 0.35). However, case numbers in these subgroups were small, CIs overlapping and the test for heterogeneity between HT users and non-users was not statistically significant (
P
het = 0.20).
Table 2
Adjusted odd ratios
a
(OR) for
in situ
breast cancer by tertile levels and on a continuous log
2
scale of circulating prolactin
All women
| | | | | | |
Ca/Co | 89/101 | 94/104 | 124/102 | | | |
OR | Ref. | 0.98 (0.65, 1.49) | 1.37 (0.87, 2.16) | 1.35 (1.04, 1.76) | 0.03 | |
Premenopausal
| | | | | | |
Ca/Co | 27/29 | 19/28 | 40/29 | | | |
OR | Ref. | 0.63 (0.25, 1.58) | 1.49 (0.59, 3.74) | 1.30 (0.80, 2.10) | 0.28 | |
Postmenopausal (all)
| | | | | | |
Ca/Co | 65/72 | 69/75 | 87/74 | | | 0.98 |
OR | Ref. | 1.00 (0.61, 1.63) | 1.31 (0.77, 2.22) | 1.38 (1.00, 1.91) | 0.05 | |
Postmenopausal non-HT users
| | | | | | |
Ca/Co | 50/51 | 39/49 | 50/39 | | | |
OR | Ref. | 0.84 (0.44, 1.6) | 1.27 (0.65, 2.47) | 1.20 (0.82, 1.76) | 0.35 | |
Postmenopausal HT users
| | | | | | |
Ca/Co | 15/21 | 30/26 | 37/35 | | | 0.20 |
OR | Ref. | 1.63 (0.70, 3.81) | 1.62 (0.64, 4.09) | 1.77 (0.98, 3.21) | 0.06 | |
In subgroup analyses (Table
3), we observed significant heterogeneity in the strength of the association with
in situ breast cancer risk by time between blood donation and diagnosis (that is, lag time) among all women (
P
het = 0.04). Higher concentrations of prolactin were significantly associated with
in situ breast cancer diagnosed less than 4 years since blood donation (OR
log2 = 1.78 (95% CI 1.12, 2.84),
P
trend = 0.01), but not with breast cancer diagnosed 4 or more years since blood donation (OR
log2 = 1.09 (95% CI 0.77, 1.55),
P
trend = 0.63;
P
het = 0.04). Similarly, the estimated association of prolactin with
in situ breast cancer risk appeared stronger among nulliparous women compared to parous women, more pronounced in analysis restricted to postmenopausal women (OR
log2 = 5.10 (95% CI 1.23, 21.15),
P
trend = 0.02 in nulliparous women versus OR
log2 = 1.22 (95% CI 0.88, 1.68),
P
trend = 0.24 in parous women;
P
het = 0.02). It should be noted, however, that the case numbers for nulliparous women were small and the CIs for risk estimates wide. There was no significant heterogeneity by age at tumour diagnosis in either premenopausal or postmenopausal women (
P
het = 0.87) (data not shown).
Table 3
Adjusted odds ratios
a
(OR) for
in situ
breast cancer on a continuous log
2
scale of circulating prolactin by subgroup analysis
All women
| | | | |
Lag time till diagnosis | | | | |
<4 years | 127 | 1.78 (1.12, 2.84) | 0.01 | 0.04 |
≥4 years | 180 | 1.09 (0.77, 1.55) | 0.63 | |
Parityd
| | | | |
Nulliparous | 42/45 | 2.64 (1.07, 6.51) | 0.03 | 0.07 |
Parous | 251/251 | 1.21 (0.94, 1.57) | 0.15 | |
Premenopausal women
| | | | |
Lag time till diagnosis | | | | |
<4 years | 39 | 2.30 (0.85, 6.14) | 0.10 | 0.15 |
≥4 years | 47 | 0.97 (0.51, 1.83) | 0.92 | |
Postmenopausal women (all)
| | | | |
Lag time till diagnosis | | | | |
<4 years | 88 | 1.66 (0.97, 2.85) | 0.07 | 0.19 |
≥4 years | 133 | 1.18 (0.76, 1.82) | 0.46 | |
Parityd
| | | | |
Nulliparous | 29/32 | 5.10 (1.23, 21.15) | 0.02 | 0.02 |
Parous | 187/185 | 1.22 (0.88, 1.68) | 0.24 | |
Postmenopausal non-HT users
| | | | |
Lag time till diagnosis | | | | |
<4 years | 52 | 1.20 (0.70, 2.05) | 0.51 | 0.52 |
≥4 years | 87 | 1.08 (0.61, 1.93) | 0.79 | |
Postmenopausal HT users
| | | | |
Lag time till diagnosis | | | | |
<4 years | 36 | 6.02 (1.31, 27.72) | 0.02 | 0.04 |
≥4 years | 46 | 1.29 (0.64, 2.61) | 0.47 | |
Additional statistical analyses showed no evidence for major confounding effects by other lifestyle and reproductive factors nor was there an interaction between prolactin and BMI (P
interaction = 0.84) (data not shown).
Finally, we compared the results from our present analysis on
in situ breast cancer and from our prior analysis on invasive breast cancer [
5]. The median levels of prolactin comparing
in situ (n = 307) versus invasive breast cancer tumors (n = 2250) were 8.9 (
in situ) versus 8.6 ng/mL (invasive) for premenopausal women and 6.2 (
in situ) versus 6.1 (invasive) ng/mL for the postmenopausal women (data not shown). A heterogeneity test comparing
in situ versus invasive disease was not statistically significant (
P
het = 0.25 for premenopausal women and
P
het = 0.33 for postmenopausal women).
Discussion
In this prospective study, we observed a significant positive association between pre-diagnostic circulating prolactin levels and risk of in situ breast cancer. Our data showed no evidence for heterogeneity in the relationship of prolactin levels and in situ cancer risk by menopausal status or postmenopausal HT use at blood donation, although the observed positive association was more pronounced among HT users versus non-users. In addition, the associations were strongest among women diagnosed with in situ breast tumors less than 4 years after blood donation and also among nulliparous women.
Our findings of modest, positive associations between prolactin and
in situ breast cancer risk are similar to those observed for invasive breast cancer. Prior analyses limited to invasive cases found modest significant associations between circulating prolactin and risk of invasive breast cancer, with up to a 50% increase in risk contrasting top
versus bottom quartiles [
5,
6]. Moreover, comparable to our previous findings for invasive breast cancer [
5], the observed association in the present study of
in situ breast cancer was more pronounced among postmenopausal HT users as compared to non-users. Although
in situ carcinomas in the breast may not have invasive characteristics [
12], up to 50% of
in situ lesions, if left untreated, progress to invasive disease [
3]. Little is known about the exact mechanisms influencing the possible progression of
in situ lesions to invasive disease. However, the very similar associations between prolactin and breast cancer risk observed in our previous study of invasive disease [
5] and the current study of
in situ disease suggest that circulating prolactin may influence risk of invasive and
in situ breast cancer via the same aetiological pathway. Prolactin may also play a role in progression from
in situ to invasive breast cancer. However, out of 307
in situ cases selected for this analysis, only 15 developed invasive breast cancer during further follow up. Therefore, we were unable to assess the association between prolactin and progression from
in situ to invasive disease in this study.
Although a vast body of evidence from animal,
in vitro, and epidemiological studies strongly supports the involvement of prolactin in breast cancer development [
5,
6,
13,
14], the complex and diverse biological and molecular mechanisms through which prolactin may increase risk of breast cancer are not clear. Some proposed mechanisms include its proliferative effects on malignant breast cells, mitogenic action, and inhibition of apoptosis via signalling through the prolactin receptor [
14]. In addition to the endocrine (circulating) concentrations, locally produced prolactin may promote cancer development via autocrine and paracrine effects [
15]. Evidence is also emerging that drugs resulting in elevated prolactin (for example, neuroleptic and hormonal medications) may increase breast cancer risk [
16,
17].
The relationship between prolactin and
in situ breast cancer risk was confined to tumors diagnosed within the first 4 years from blood donation. Interestingly, this differential association by time between blood collection and diagnosis was not evident for invasive breast cancer risk in our previous analysis [
5]. Whether prolactin is involved in the early development or promotes late-stage growth of an established tumor is unclear. Consistent with the current analysis, recent results from the Nurses’ Health Study suggest that prolactin primarily plays a role in the late stage of tumor development, with stronger positive associations observed among participants providing blood samples closer to diagnosis [
6,
18]. This is also supported by studies observing elevated levels of circulating prolactin in breast cancer patients [
19], and demonstrating the ability of breast tumors to secrete prolactin [
13]. However, our present analysis suggests that prolactin may also operate early in the natural history of breast cancer by increasing risk of
in situ tumors, the earliest detectable breast carcinomas.
In the current study, prolactin was associated with
in situ breast cancer among postmenopausal nulliparous women but not in parous women. Parity, a well-established protective factor for breast cancer [
20], has been associated with a long-term post-pregnancy reduction in levels of circulating prolactin [
10,
21]. Our results support the hypothesis that lowered prolactin levels following pregnancy might serve as one of several possible mechanisms that mediate the long-term reduction in risk afforded by parity. However, our findings should be interpreted with caution as only about 15% of women in this case-control study were nulliparous and OR estimates for nulliparous women had wide confidence limits.
Our study is the second largest prospective study on the association between prolactin and breast cancer risk, to provide risk estimates for
in situ breast cancer. A general limitation of our study is that prolactin levels were measured only at a single point in time. However, reliability studies in which repeat blood samples were taken over intervals of up to three years have shown relatively high intra-class correlation (ICC coefficients of up to 0.76) between individuals’ serum prolactin levels over time [
22-
24]. In
situ breast cancer is largely diagnosed as a result of mammography; however, we were not able to adjust our models for the participation in mammographic screening, nor were we able to compare rates of screening in cases and controls. Our study had relatively limited numbers of case subjects for subgroup analysis, and thus the statistical power for detecting associations of prolactin with risk in these subgroups was limited. Due to very limited information available for the molecular characteristics of
in situ tumors, we were not able to perform the analysis by molecular subtypes of the lesions.
Acknowledgements
We thank all the EPIC cohort participants. Furthermore, we thank Britta Lederer and Sigrid Henke for their work in conducting the immunoassays and Bertrand Hémon for his help with the EPIC database. Ethical approval for the EPIC study was obtained from the ethical review boards of IARC (Lyon, France) and from The Committee of Bioethics and Deontology of the Hellenic Health Foundation (Athens; Greece), Norwich District Ethics Committee (Cambridge; UK), The National Committee on Health Research (Aarhus, Copenhagen; Denmark), Ethics Comité de Protection des Personnes (Paris; France), Ethics Committee of the Heidelberg University Medical School (Heidelberg; Germany), IARC Ethics Committee (France), Imperial College Research Ethics Committee (ICREC) (London; UK), Comitato Etico Indipendente, Fondazione IRCCS Istituto Nazionale dei Tumori, Milano (Milan, Naples, Ragusa, Turin; Italy), Comitato Etico Locale Azienda Sanitaria di Firenze (Florence; Italy), Ethics Committee of Lundst University (Malmo; Sweden), The Medical Ethical Committee (METC, Medisch Ethische Toetsingscommissie) of the University Medical Center Utrecht (UMCU)( Utrecht, Bilthoven; the Netherlands), The Regional Committee for Medical and Health Research Ethics, North Norway (REK nord) (Norway), Scotland A Research Ethics Committee (Oxford; UK), Ethikkommission der Landesärztekammer Brandenburg Cottbus, (Potsdam; Germany), CEIC Comité de Ética de Investigación Clínica (Asturias, Barcelona, Granada, Murcia, Navarre, San Sebastian; Spain), Human Genetics Foundation Torino: Ethics Committee (Torino; Italy), Umea Regional Ethical Review Board (Umea, Sweden). 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); Cancer Research UK, and the Medical Research Council (UK).
Funding This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
KT, DS and RK contributed to the conception of the current analysis and SR, IR, AT, KO, FC-C, HB, AT, DT, VK, RT, M-JS, M-DC, AB, HBB, EM, CO-M, AA, EW, K-TK, TK, RT, MM, ER, LD, and RK were involved in the design and acquisition of data from the EPIC cohort. TJ carried out the immunoassays. KT, DS and RK contributed to the analysis and KT, DS, RF, SR, IR, AT, AO, KO, FC-C, LB, HB, AT, PL, DT, GM, VK, RT, FR, AM, AA, VM, PA, M-JS, M-DC, AB, HBB, EM, CO-M, AA, MS, EW, K-TK, TK, RT, MM, ER, LD, and RK contributed to the interpretation of the data. KT, RF and RK drafted the manuscript and DS, TJ, SR, IR, AT, AO, KO, FC-C, LB, HB, AT, PL, DT, GM, VK, RT, FR, AM, AA, VM, PA, M-JS, M-DC, AB, HBB, EM, CO-M, AA, MS, EW, K-TK, TK, RT, MM, ER, and LD revised the final draft critically for important critical content. All authors read and approved the final manuscript.