Abstract
Background
Benign breast disease (BBD), particularly proliferative BBD, is an established breast cancer risk factor. However, there has been no systematic attempt to compare the hormonal profiles of the two conditions. In a case–control investigation in Athens, Greece, we compared levels of estrogens, testosterone and insulin-like growth factor-1 (IGF-1), as well as their principal binding proteins, between breast cancer patients, women with BBD by histological type (proliferative and nonproliferative) and women with no breast pathology.
Patients and methods
We studied 466 women with incident breast cancer, 704 women with BBD and 244 healthy women. We used multiple regression to compare log-transformed serum hormone levels of breast cancer patients with those of healthy women and women with BBD by histological type (proliferative and nonproliferative BBD).
Results
The hormonal profile of breast cancer in our study was in line with the generally accepted hormonal profile of this disease, as reported from large cohort studies. Compared with healthy women, breast cancer patients tended to have higher levels of steroid hormones. The evidence was strong for estrone (difference 21.5%, P < 0.001), weaker for testosterone (difference 15.8%, P = 0.07) and weaker still for estradiol (difference 12.0%, P = 0.18). Also compared with healthy women, breast cancer patients had barely higher levels of IGF-1 (difference 2.0%, P = 0.51), but had significantly lower levels of IGF binding protein 3 (IGFBP-3) (difference −6.7%, P = 0.001). Compared with women with BBD, breast cancer patients had nonstatistically significantly lower levels of steroid hormones, but they had higher levels of IGF-1 [difference 5.5%, 95% confidence interval (CI) 0.7% to 10.6%] and lower levels of IGFBP-3 (difference −3.7%, 95% CI −6.7% to −0.7%). Differences were more pronounced when breast cancer patients were contrasted to women with proliferative BBD.
Conclusions
Our findings suggest that high levels of IGF-1 may be an important factor toward the evolution of BBD to breast cancer.
Keywords: benign breast disease, breast cancer, epidemiology, IGF-1, proliferative, steroid hormones
introduction
Benign breast disease (BBD) is an established risk factor for breast cancer, the risk for developing breast cancer gradually increases with proliferation and atypia to more than 4-fold [1–3]. Although BBD is intimately entangled in the differential diagnosis of breast cancer and may be involved in its pathogenesis, there has not been a systematic attempt to compare the hormonal profile of the two conditions. A possible reason is that, unlike incident breast cancer, BBD can only be investigated as a prevalent condition through case–control or cross-sectional designs [4].
In the context of a case–control investigation in Greece, we studied women with incident breast cancer, women diagnosed with BBD by histological type (proliferative versus nonproliferative disease) and women with no breast pathology, from the same study base [4]. The objective was to compare levels of estrogens, testosterone and insulin-like growth factor-1 (IGF-1), as well as their principal binding proteins, between breast cancer and BBD (by histological type). Given the case–control nature of our study, a precondition for the external validity of our results would be that the hormonal profile of breast cancer in our study would be in line with the generally accepted hormonal profile of this disease, as reported from cohort studies and summarized in meta-analyses, for steroids (i.e. substantially higher steroid levels for breast cancer patients compared with healthy women, particularly after menopause) [5–7] and IGF-1 (i.e. higher levels among breast cancer patients compared with healthy women) [8].
materials and methods
subjects
From March 2001 to May 2005, women who had undergone mammary biopsy with a diagnosis of BBD or breast cancer in two major breast clinics in Athens, Greece, were asked to participate in the study [9]. In breast clinic I, women who underwent a breast biopsy during the duration of the study, as well as women who had undergone a biopsy with a diagnosis of BBD up to 4 years before the study initiation (but were interviewed and provided a blood sample during the study period) were included. In breast clinic II, all women underwent biopsy during the study period and were included in the present study. Thus, for the present investigation, we studied only incident breast cancer cases from both centers, for whom treatment had not been initiated. For BBD, all available cases from both centers were studied, irrespectively of the time of biopsy, as medication is generally not prescribed for BBD, and in all BBD studies in the literature, prevalent cases have been used [4]. Tissue samples and the pathology reports were reviewed for accurate diagnostic classification, blindly as to the results of the hormone determinations. Also invited to participate in the study were women who visited the breast clinics but were diagnosed as healthy and did not require a breast biopsy. The study was approved by the Bioethics Committee of the University of Athens and those who agreed to participate provided informed consent.
Among eligible women, we estimate that about 75% agreed to participate in the study (in several instances, women refused to allow any recording of information concerning agreement to participate in the study; thus, the refusal proportion cannot be accurately calculated). In the present investigation, 466 women with breast cancer, 704 women with BBD and 244 healthy women were included.
All women completed an extensive interviewer-administered questionnaire with information on sociodemographic and lifestyle factors, as well as on gynecological history, general medical history and dietary habits. In breast clinic I, histological samples were available in the form of paraffin-embedded tissue (PET) blocks, whereas in breast clinic II, samples obtained during biopsy were frozen in liquid nitrogen before being stored in −80°C.
hormone measurements
Blood samples for hormone measurements were collected, at the time of the in-person interview, into sterile tubes, were centrifuged, and then aliquoted and stored for hormonal assays at −80°C at the University of Athens Medical School. Hormonal analyses were blindly done in the ‘Biomedicine’ laboratories, which are accredited with an ISO 15189 for a wide range of analyses.
Estradiol-17b and sex hormone binding globulin (SHBG) were measured with an electrochemiluminescence immunoassay (ECLIA, ElecsysEstradiol II kit, Roche, Mannheim, Germany). Estrone was measured with a radioimmunoassay (Estrone RIA DSL-8700 kit, Diagnostic Systems Laboratories, Prague, Czech Republic). Testosterone was measured using an ECLIA (Elecsys Testosterone kit for Elecsys 1010/2010, Roche). IGF-1 and IGF binding protein 3 (IGFBP-3) were measured by chemical luminescence immunoassay kits (Immulite 2500 kit, Diagnostic Products Corporation, Siemens, Llanberis, Gwynedd, UK). Details on the hormone determinations have been previously reported [4].
statistical analysis
Analyses were conducted using SPSS (IBM Statistical Package for Social Sciences v. 19.0, Chicago, IL). We used multiple regressions to compare log-transformed serum hormone levels of breast cancer patients with those of healthy women, women with BBD, and women with the high-breast-cancer-risk proliferative BBD. Regression analyses were carried out both overall, controlling for menopausal status, as well as separately among pre- and peri/postmenopausal women. In the initial models, we controlled for age (continuously) and external hormonal use (oral contraceptives among premenopausal women or hormone replacement treatment (HRT) among peri/postmenopausal women), while in the analysis of premenopausal women, we further controlled for day of the menstrual cycle (1 = blood sample collected in days 2–11 and 0 = else). When we studied steroids, we additionally controlled for SHBG (continuously), while IGF-1 and IGF-BP3 were adjusted for each other (continuously). Finally, we also applied regression models that excluded women who had used exogenous hormones and also controlled for age at menarche (continuously), parity (parous versus nulliparous), body mass index (BMI, continuously), smoking habits (current smokers versus other), educational level (ordered, with 1 = up to 6 years of schooling, 2 = 7–12 years and 3 = >12 years) and lactation (women who had breast-fed versus those who had not).
results
We compared the hormonal profiles of 466 women with incident breast cancer, 704 women with BBD, 396 of whom had proliferative lesions, and 244 women without evident breast disease. Basic characteristics of the 1414 women in the study are shown in Table 1. In comparison to women with breast cancer, more women with BBD were premenopausal and in the youngest age category (<49 years), and these two variables were adjusted for in all comparisons. Differences were also evident with respect to parity, BMI, smoking and educational level. Although these differences were largely accounted for by the age differences, all these variables were controlled for in the more comprehensive analytical models.
Table 1.
Distribution of 1414a women by diagnosis and demographic, lifestyle and reproductive characteristics
| Characteristics | Healthy (n = 244) | All BBD (n = 704) | Proliferative BBD (n = 396) | Cancer (n = 466) |
|---|---|---|---|---|
| Age (years) | ||||
| −49 | 99 (40.6) | 512 (72.7) | 283 (71.5) | 133 (28.5) |
| 50–59 | 86 (35.2) | 108 (15.3) | 62 (15.7) | 139 (29.8) |
| 60+ | 59 (24.2) | 84 (11.9) | 51 (12.9) | 194 (41.6) |
| Menopausal status | ||||
| Premenopausal | 89 (36.5) | 492 (69.9) | 272 (68.7) | 133 (28.5) |
| Peri/postmenopausal | 155 (63.5) | 212 (30.1) | 124 (31.3) | 333 (71.5) |
| Age at menarche (years) | ||||
| −12 | 107 (44.0) | 282 (40.4) | 157 (39.9) | 198 (43.0) |
| 13 | 77 (31.7) | 198 (28.4) | 111 (28.2) | 118 (25.6) |
| 14+ | 59 (24.3) | 218 (31.2) | 125 (31.8) | 145 (31.5) |
| Age at menopause (among peri/postmenopausal) | ||||
| −44 | 29 (18.7) | 41 (19.3) | 20 (16.1) | 48 (14.4) |
| 45–49 | 50 (32.3) | 61 (28.8) | 37 (29.8) | 110 (33.0) |
| 50–54 | 64 (41.3) | 94 (44.3) | 59 (47.6) | 137 (41.1) |
| 55+ | 12 (7.7) | 16 (7.5) | 8 (6.5) | 38 (11.4) |
| Parous | ||||
| Yes | 209 (85.7) | 484 (68.8) | 284 (71.7) | 411 (88.2) |
| No | 35 (14.3) | 220 (31.3) | 112 (28.3) | 55 (11.8) |
| BMI (kg/m2) | ||||
| −25 | 99 (40.6) | 382 (54.6) | 210 (53.6) | 155 (33.7) |
| 26–29.99 | 88 (36.1) | 188 (26.9) | 106 (27.0) | 160 (34.8) |
| 30+ | 57 (23.4) | 130 (18.6) | 76 (19.4) | 145 (31.5) |
| Current smoker | ||||
| Yes | 81 (33.3) | 305 (43.3) | 174 (43.9) | 149 (32.0) |
| No | 162 (66.7) | 399 (56.7) | 222 (56.1) | 316 (68.0) |
| Education (years) | ||||
| −6 | 55 (22.5) | 122 (17.4) | 64 (16.2) | 189 (40.6) |
| 7–12 | 108 (44.3) | 294 (41.8) | 173 (43.8) | 181 (38.8) |
| 13+ | 81 (33.2) | 287 (40.8) | 158 (40.0) | 96 (20.6) |
| Lactation (among parous) | ||||
| Yes | 173 (82.8) | 409 (84.5) | 241 (84.9) | 347 (84.4) |
| No | 36 (17.2) | 75 (15.5) | 43 (15.1) | 64 (15.6) |
| Exogenous hormone use OC use (among premenopausal) | ||||
| Yes | 33 (37.1) | 100 (20.3) | 58 (21.3) | 24 (18.0) |
| No | 56 (62.9) | 392 (79.7) | 214 (78.7) | 109 (82.0) |
| HRT (among peri/postmenopausal) | ||||
| Yes | 33 (21.3) | 26 (12.3) | 14 (11.3) | 29 (8.7) |
| No | 122 (78.7) | 186 (87.7) | 110 (88.7) | 304 (91.3) |
aThere were a few missing values in some of the variables in the table.
OC, oral contraceptives; HRT, hormone replacement treatment.
In Table 2, the quartiles of the measured hormones, by diagnosis and menopausal status, are shown. Estrogens, and to a lesser extent testosterone and IGF-1, were higher among premenopausal than among peri/postmenopausal women, whereas there was little variability by menopausal status in the levels of the two binding proteins. The data in Table 2 provide background values of the studied hormones, but contrasts across the diagnostic categories are not directly interpretable, because of substantial age differences.
Table 2.
Median (25th–75th percentile) of measured hormones by diagnosis and menopausal status
| Healthy | All BBD | Proliferative BBD | Cancer | |
|---|---|---|---|---|
| Premenopausal women | n = 89 | n = 492 | n = 272 | n = 133 |
| Estrone(pg/ml) | 52.5 (42.2–71.1) | 67.1 (50.0–95.0) | 71.0 (51.0–97.5) | 61.0 (45.1–94.5) |
| Estradiol (pg/ml) | 79.0 (51.0–139.0) | 102.5 (53.0–188.5) | 106.5 (56.0–185.0) | 94.0 (50.5–167.0) |
| Testosterone (ng/dl) | 29.0 (18.0–40.5) | 38.0 (26.0–52.0) | 38.0 (26.0–51.0) | 31.0 (20.5–47.0) |
| SHBG (nmol/l) | 59.0 (43.5–84.5) | 64.3 (45.0–90.0) | 65.5 (46.0–90.0) | 60.0 (45.0–81.5) |
| IGF-1 (ng/dl) | 166.0 (123.5–202.0) | 167.5 (131.0–213.0) | 167.0 (137.0–216.8) | 163.0 (130.0–188.5) |
| IGFBP3 (μg/ml) | 3.9 (3.5–4.8) | 3.9 (3.4–4.5) | 4.0 (3.4–4.5) | 3.8 (3.3–4.4) |
| Peri/postmenopausal women | n = 155 | n = 212 | n = 124 | n = 333 |
| Estrone (pg/ml) | 28.7 (23.9–34.0) | 36.0 (26.7–49.0) | 36.7 (28.0–54.0) | 33.0 (26.0–44.0) |
| Estradiol (pg/ml) | 10.0 (5.0–15.0) | 12.0 (5.1–23.8) | 13.0 (5.1–28.8) | 9.0 (5.1–15.5) |
| Testosterone (ng/dl) | 19.0 (10.0–29.0) | 22.0 (12.3–37.0) | 22.0 (13.0–36.0) | 22.0 (13.0–35.0) |
| SHBG (nmol/l) | 55.0 (38.0–78.0) | 55.0 (39.0–78.0) | 52.5 (37.3–75.8) | 56.0 (38.0–75.0) |
| IGF-1 (ng/dl) | 133.0 (99.0–162.0) | 113.5 (89.0–147.0) | 112.0 (88.0–142.0) | 109.0 (85.0–147.0) |
| IGFBP3 (μg/ml) | 3.9 (3.4–4.6) | 3.7 (3.2–4.5) | 3.8 (3.1–4.6) | 3.5 (2.8–4.1) |
In Table 3, the log-transformed levels of the measured hormones are compared, first between women with breast cancer and women with no evident breast pathology, and secondly between women with breast cancer and women with BBD. Log-transformation allows interpretation of the results of the comparisons in terms of percent (%) differences. An additional comparison for women overall, controlling for menopausal status, was undertaken.
Table 3.
Percent (%) differencesa (and 95% confidence intervals) of the measured hormone levels for women with breast cancer versus women with no evident breast pathology (healthy) or versus women with BBD
| Women with breast cancer versus healthy women |
Women with breast cancer versus women with BBD |
|||||
|---|---|---|---|---|---|---|
| Premenopausal (133 versus 89) | Peri/postmenopausal (333 versus 155) | All women (466 versus 244) | Premenopausal (133 versus 492) | Peri/postmenopausal (333 versus 212) | All women (466 versus 704) | |
| Estrone (E1)b | ||||||
| Model I | 18.2 (1.8, 37.2) P = 0.03 |
22.8 (11.5, 35.1) P < 0.001 |
20.9 (11.6, 31.0) P < 0.001 |
−7.7 (−16.5, 2.0) P = 0.12 |
−3.5 (−12.0, 5.8) P = 0.45 |
−5.4 (−11.6, 1.4) P = 0.12 |
| Model II | 15.4 (−2.7, 36.8) P = 0.10 |
24.5 (12.2, 38.1) P < 0.001 |
21.7 (11.2, 33.1) P < 0.001 |
−8.4 (−18.3, 2.6) P = 0.137 |
−3.6 (−13.0, 6.7) P = 0.47 |
−5.4 (−12.4, 2.1) P = 0.15 |
| Model III | 18.2 (−0.5, 40.4) P = 0.06 |
21.7 (9.4, 35.2) P < 0.001 |
21.5 (11.1, 33.0) P < 0.001 |
−7.4 (−17.4, 3.7) P = 0.18 |
−4.8 (−14.3, 5.8) P = 0.36 |
−5.1 (−12.2, 2.7) P = 0.19 |
| Estradiol (E2)b | ||||||
| Model I | 6.0 (−17.2, 35.7) P = 0.65 |
7.5 (−9.4, 27.4) P = 0.41 |
6.4 (−7.6, 22.5) P = 0.39 |
−10.4 (−25.2, 7.3) P = 0.23 |
−10.9 (−24.8, 5.7) P = 0.19 |
−10.6 (−21.1, 1.4) P = 0.08 |
| Model II | 12.7 (−15.8, 51.0) P = 0.42 |
10.1 (−9.0, 33.1) P = 0.32 |
10.8 (−5.6, 30.2) P = 0.21 |
−9.2 (−25.3, 10.5) P = 0.33 |
−11.4 (−26.3, 6.5) P = 0.20 |
−9.8 (−21.4, 3.5) P = 0.14 |
| Model III | 15.4 (−14.4, 55.4) P = 0.35 |
8.4 (−11.2, 32.4) P = 0.43 |
12.0 (−5.2, 32.3) P = 0.18 |
−5.3 (−22.3, 15.5) P = 0.59 |
−10.1 (−26.1, 9.3) P = 0.28 |
−7.1 (−19.4, 6.9) P = 0.31 |
| Testosteroneb | ||||||
| Model I | 19.1 (−0.1, 42.1) P = 0.05 |
27.5 (6.7, 52.4) P = 0.008 |
24.9 (9.3, 42.7) P = 0.001 |
−2.9 (−13.5, 9.1) P = 0.62 |
4.8 (−10.2, 22.4) P = 0.55 |
1.3 (−8.1, 11.7) P = 0.80 |
| Model II | 13.0 (−7.5, 38.0) P = 0.23 |
22.0 (0.1, 48.7) P = 0.05 |
19.1 (2.2, 38.8) P = 0.02 |
−6.1 (−16.8, 6.0) P = 0.32 |
4.9 (−11.4, 24.2) P = 0.58 |
−0.1 (−10.3, 11.3) P = 0.98 |
| Model III | 12.0 (−8.7, 37.3) P = 0.28 |
17.5 (−4.8, 44.9) P = 0.14 |
15.8 (−1.2, 35.8) P = 0.07 |
−7.1 (−18.1, 5.3) P = 0.25 |
−0.7 (−17.1, 18.9) P = 0.94 |
−3.4 (−13.6, 8.0) P = 0.54 |
| SHBG | ||||||
| Model I | −1.0 (−13.7, 13.6) P = 0.89 |
−0.2 (−9.2, 9.6) P = 0.96 |
−0.5 (−7.8, 7.4) P = 0.90 |
−1.1 (−10.5, 9.3) P = 0.83 |
−0.6 (−9.0, 8.6) P = 0.89 |
−0.8 (−7.2, 6.0) P = 0.81 |
| Model II | 7.5 (−8.0, 25.5) P = 0.36 |
0.8 (−9.0, 11.6) P = 0.88 |
2.7 (−5.6, 11.8) P = 0.54 |
0.5 (−10.1, 12.4) P = 0.93 |
−1.2 (−9.9, 8.3) P = 0.81 |
−0.5 (−7.3, 6.8) P = 0.90 |
| Model III | 6.7 (−8.8, 24.8) P = 0.42 |
4.0 (−5.4, 14.2) P = 0.42 |
2.6 (−5.8, 11.9) P = 0.55 |
0.5 (−10.3, 12.6) P = 0.94 |
2.1 (−6.3, 11.3) P = 0.64 |
0.4 (−6.6, 8.0) P = 0.91 |
| IGF-1c | ||||||
| Model I | 0.3 (−7.8, 9.1) P = 0.94 |
2.3 (−3.9, 9.0) P = 0.47 |
1.3 (−3.7, 6.6) P = 0.61 |
4.8 (−1.6, 11.6) P = 0.15 |
7.1 (1.0, 13.6) P = 0.02 |
5.8 (1.3, 10.4) P = 0.01 |
| Model II | 0.6 (−9.1, 11.4) P = 0.92 |
3.1 (−3.5, 10.3) P = 0.36 |
2.1 (−3.5, 8.1) P = 0.47 |
4.6 (−2.3, 12.0) P = 0.20 |
8.1 (1.5, 15.1) P = 0.01 |
6.3 (1.4, 11.4) P = 0.01 |
| Model III | −0.5 (−10.0, 10.0) P = 0.92 |
3.1 (−3.7, 10.5) P = 0.38 |
2.0 (−3.6, 8.0) P = 0.51 |
5.1 (−2.0, 12.8) P = 0.17 |
7.1 (0.4, 14.3) P = 0.04 |
5.5 (0.7, 10.6) P = 0.03 |
| IGFBP−3d | ||||||
| Model I | −2.4 (−7.9, 3.5) P = 0.42 |
−8.0 (−12.0, −3.7) P < 0.001 |
−6.4 (−9.8, −2.8) P < 0.001 |
−0.7 (−4.1, 2.9) P = 0.71 |
−4.9 (−9.1, −0.5) P = 0.03 |
−3.1 (−6.0, −0.3) P = 0.03 |
| Model II | −4.8 (−11.1, 2.0) P = 0.16 |
−8.6 (−12.8, −4.2) P < 0.001 |
−7.7 (−11.2, −4.0) P < 0.001 |
−2.5 (−6.2, 1.4) P = 0.21 |
−5.3 (−9.4, −0.9) P = 0.02 |
−4.1 (−7.1, −1.1) P = 0.01 |
| Model III | −4.0 (−10.4, 2.8) P = 0.24 |
−7.2 (−11.5, −2.8) P = 0.002 |
−6.7 (−10.3, −2.9) P = 0.001 |
−3.1 (−6.9, 0.7) P = 0.11 |
−4.2 (−8.6, 0.4) P = 0.08 |
−3.7 (−6.7, −0.7) P = 0.02 |
aResults from multiple log-linear models by menopausal status; for all women controlling for menopausal status. Model I: controlling for age and exogenous hormone use. Model II: as model I, excluding women with exogenous hormone use. Model III: as Model II, controlling also for age at menarche, parity, BMI, smoking status, education and lactation. Among premenopausal women controlling also for day of menstrual cycle.
bControlling for SHBG.
cControlling for IGFBP-3.
dControlling for IGF-1.
When comparing women with breast cancer to women with no evident breast pathology, we found evidence that levels of steroid hormones tended to be higher in the former group. The evidence in the full model III was strong for estrone [difference 21.5% with 95% confidence interval (CI) 11.1% to 33.0%], weaker for testosterone (difference 15.8%, with 95% CI −1.2% to 35.8%) and weaker still for estradiol (difference 12.0%, with 95% CI −5.2% to 32.3%). There was little evidence for a difference in SHBG levels. With respect to IGF-1, women with breast cancer, compared with women with no evident breast pathology, had barely higher levels of this hormone (difference 2.0%, with 95% CI −3.6% to 8.0%), but had significantly lower levels of its main binding protein, IGFBP-3 (difference -6.7%, with 95% CI −10.3% to −2.9%).
When comparing women with breast cancer to women with BBD, there was some evidence, albeit not statistically significant, that levels of steroid hormones tended to be lower in the former group, point estimates of percent differences being −5.1% for estrone, −7.1% for estradiol and −3.4% for testosterone; there was no evidence for an important difference in SHBG levels between the two groups of women. With respect to the evaluated components of the IGF axis, women with breast cancer, compared with women with BBD, had significantly lower levels of IGFBP-3 (difference −3.7%, with 95% CI −6.7% to −0.7%) and significantly higher levels of IGF-1 (difference 5.5%, with 95% CI 0.7% to 10.6%). Because the results were mainly driven by the associations among postmenopausal women, we tested for interaction by menopausal status of the IGF-1 association with breast cancer rather than BBD. The P values for interaction were not significant for any of the three models fitted (model I, P = 0.13; model II, P = 0.09; model III, P = 0.43).
Because women with proliferative BBD are at higher risk for breast cancer than women with nonproliferative BBD, we have examined whether the contrasts evident in the right part of Table 3, are more or less pronounced when comparing women with breast cancer to women with proliferative BBD. The results shown in Table 4 indicate that the hormonal profile of women with breast cancer tended to deviate more clearly from the hormonal profile of women with proliferative BBD. Thus, levels of estrone, estradiol and testosterone were noticeably lower in women with breast cancer than in women with proliferative BBD and with respect to estrone, significantly so. With respect to IGF-1, the contrast also became slightly more evident.
Table 4.
Percent (%) differencesa (and 95% confidence intervals) of the measured hormone levels for women with breast cancer versus women with proliferative BBD
| Women with breast cancer versus women with proliferative BBD |
|||
|---|---|---|---|
| Premenopausal (133 versus 272) | Peri/postmenopausal (333 versus 124) | All women (466 versus 396) | |
| Estrone (E1)b | |||
| Model I | −9.6 (−19.3, 1.3) P = 0.08 |
−11.0 (−20.4, −0.5) P = 0.04 |
−9.6 (−16.6, −2.0) P = 0.01 |
| Model II | −10.8 (−21.8, 1.7) P = 0.09 |
−11.4 (−21.5, 0.1) P = 0.05 |
−10.0 (−17.7, −1.5) P = 0.02 |
| Model III | −10.6 (−21.4, 1.8) P = 0.09 |
−11.5 (−21.9, 0.3) P = 0.06 |
−8.7 (−16.6, −0.1) P = 0.05 |
| Estradiol (E2)b | |||
| Model I | −10.8 (−26.5, 8.3) P = 0.25 |
−20.3 (−34.6, −2.9) P = 0.03 |
−14.1 (−25.4, −1.1) P = 0.03 |
| Model II | −9.8 (−27.3, 11.9) P = 0.35 |
−20.5 (−35.8, −1.6) P = 0.03 |
−13.3 (−25.6, 1.0) P = 0.07 |
| Model III | −6.2 (−24.1, 15.9) P = 0.55 |
−17.7 (−34.3, 3.1) P = 0.09 |
−8.9 (−22.1, 6.6) P = 0.25 |
| Testosteroneb | |||
| Model I | −5.9 (−16.5, 6.0) P = 0.32 |
2.4 (−14.6, 22.9) P = 0.79 |
−2.4 (−12.7, 9.2) P = 0.67 |
| Model II | −5.2 (−17.3, 8.8) P = 0.45 |
3.0 (−15.5, 25.6) P = 0.77 |
−1.7 (−13.3, 11.5) P = 0.79 |
| Model III | −6.6 (−18.9, 7.6) P = 0.35 |
−3.4 (−21.9, 19.3) P = 0.75 |
−5.5 (−17.0, 7.5) P = 0.39 |
| SHBG | |||
| Model I | −1.8 (−12.0, 9.6) P = 0.74 |
4.6 (−5.7, 16.1) P = 0.39 |
1.6 (−5.7, 9.5) P = 0.69 |
| Model II | −1.5 (−12.9, 11.5) P = 0.81 |
5.0 (−5.9, 17.2) P = 0.38 |
2.5 (−5.6, 11.3) P = 0.55 |
| Model III | −1.3 (−13.1, 12.1) P = 0.84 |
5.3 (−5.1, 16.9) P = 0.32 |
1.9 (−6.3, 10.9) P = 0.66 |
| IGF-1c | |||
| Model I | 3.0 (−3.2, 9.7) P = 0.34 |
10.4 (2.9, 18.5) P = 0.007 |
6.1 (1.0, 11.4) P = 0.018 |
| Model II | 2.3 (−4.5, 9.6) P = 0.52 |
12.2 (4.1, 20.9) P = 0.002 |
6.6 (1.3, 12.2) P = 0.015 |
| Model III | 2.7 (−4.3, 10.2) P = 0.45 |
11.3 (2.9, 20.4) P = 0.007 |
6.8 (1.3, 12.6) P = 0.015 |
| IGFBP-3d | |||
| Model I | −1.0 (−4.8, 3.0) P = 0.61 |
−3.7 (−8.9, 1.7) P = 0.17 |
−2.0 (−5.4, 1.5) P = 0.27 |
| Model II | −1.8 (−5.9, 2.5) P = 0.43 |
−4.8 (−10.0, 0.8) P = 0.09 |
−2.8 (−6.3, 0.9) P = 0.14 |
| Model III | −3.0 (−7.1, 1.3) P = 0.18 |
−3.4 (−9.0, 2.4) P = 0.24 |
−2.9 (−6.4, 0.8) P = 0.13 |
aResults from multiple log-linear models by menopausal status; for all women controlling for menopausal status. Model I: Controlling for age and exogenous hormone use. Model II: As model I, excluding women with exogenous hormone use. Model III: as Model II, controlling also for age at menarche, parity, BMI, smoking status, education and lactation. Among premenopausal women controlling also for day of menstrual cycle.
bControlling for SHBG.
cControlling for IGFBP-3.
dControlling for IGF-1.
discussion
From a large case–control study, comparing 466 incident cases of breast cancer with 704 women with BBD (308 with nonproliferative disease; 396 with proliferative disease) and 244 women without breast pathology, we have found that: (i) the levels of the studied steroid hormones among women with breast cancer tended to be higher than those among women with no breast pathology, but lower than those among women with BBD, particularly proliferative BBD and (ii) the levels of IGF-1 among women with breast cancer were marginally, if at all, higher than those among women without breast pathology, but they were significantly higher than those among women with BBD, particularly proliferative BBD.
We have previously reported that steroid hormones are higher among women with BBD than women with no breast pathology and higher in proliferative than in nonproliferative disease [4]. We have now found evidence that steroid hormones tend to be higher among women with BBD than among breast cancer patients. What appears to qualitatively contrast the profiles of BBD and breast cancer is that IGF-1 tends to be reduced in BBD, but elevated in breast cancer. Thus, it appears that high levels of steroid hormones are important in the natural history of breast cancer, but not sufficient for the occurrence of the disease. IGF-1 may be crucial for the completion of this causal chain. In other words, lower IGF-1 levels in BBD, particularly high breast cancer risk BBD with proliferative histology, could hinder the progression to breast cancer. If confirmed, this finding could have implications that go beyond the elucidation of the natural history of BBD-related breast cancer, because lifestyle factors have been shown to affect levels of IGF-1 [10, 11].
The literature concerning comparisons of hormone levels between breast cancer patients and women with BBD is limited. Most studies were fairly small or their hormone measurements had been conducted before the 1990s and did not include the IGF system. Their results in general were not consistent. Holdaway et al. [12], however, relying on a sample of 12 women with BBD and 31 breast cancer patients and age-matched controls, have reported that, in BBD, IGFBP-3 was higher, so that there was less free IGF available to cells, conveying protection against malignant transformation. Moreover, in 2012, Rice et al. [13] studied BBD patients and reported an inverse association of blood IGF-1 and the ratio of IGF-1 to IGFBP-3 with the predominance of type 1/no type 3 lobules, a marker of complete involution, which is associated with low breast cancer risk. The results of the latter two studies presage our own results and conclusions, which are also compatible with animal data suggesting that IGF-1 may inhibit involution [14].
Regarding comparisons of breast cancer patients with healthy women, there is strong evidence from cohort studies that steroid hormones are higher among breast cancer patients; the results are conclusive among postmenopausal and strongly suggestive among premenopausal women [5–7]. For IGF-1, early results were inconsistent [15–17], but a recent pooled analysis indicates a positive association of IGF-1 with breast cancer risk among both pre- and postmenopausal women [8]. The compatibility of our results with the results generated from cohort studies with respect to the comparisons of breast cancer patients with healthy women strengthens the confidence to the validity of our results with respect to the comparisons of breast cancer patients with women with BBD, for which the existing literature is limited.
Strengths of this investigation are the adequate sample size, the use of state of the art assays for hormone determinations, and recruitment of breast cancer cases, BBD cases and control women from the same study base. As in all studies of BBD cases, these had to be prevalent, and some of the women in breast clinic I had their biopsies before their blood collection; this, however, is unlikely to have systematically affected hormone measurements, as medication is generally not prescribed for BBD, and age and menopausal status of these women were assessed at blood draw. The case-control design raises concerns about bias and reverse causation. Selection bias is unlikely, because refusal to cooperate has no known or plausible association with hormone levels. Information bias is also unlikely, because hormone determinations were done blindly and uniformly for all study subjects. Lastly, reverse causation does not have a plausible scientific basis in the present context.
In conclusion, our findings indicate that in the ‘fertile soil’ for breast cancer occurrence, generated by high levels of steroid hormone and manifested as proliferative BBD, high or low levels of IGF-1 may be an important factor toward the evolution or not of BBD to breast cancer.
funding
The study was partially supported by the National Institutes of Health/National Cancer Institute, USA (CA89823); the European Union—European Social Fund and Greek Ministry of Development—General Secretariat for Research and Technology (03ED44 PENED project); the Academy of Athens; and the University of Athens, Greece (Kapodistrias Program).
disclosure
The authors have declared no conflicts of interest.
references
- 1.Dupont WD, Page DL. Risk factors for breast cancer in women with proliferative breast disease. N Engl J Med. 1985;312:146–151. doi: 10.1056/NEJM198501173120303. [DOI] [PubMed] [Google Scholar]
- 2.Hartmann LC, Sellers TA, Frost MH, et al. Benign breast disease and the risk of breast cancer. N Engl J Med. 2005;353:229–237. doi: 10.1056/NEJMoa044383. [DOI] [PubMed] [Google Scholar]
- 3.Pearlman MD, Griffin JL. Benign breast disease. Obstet Gynecol. 2010;116(3):747–758. doi: 10.1097/AOG.0b013e3181ee9fc7. [DOI] [PubMed] [Google Scholar]
- 4.Samoli E, Trichopoulos D, Lagiou A, et al. The hormonal profile of benign breast disease. Br J Cancer. 2013;108:199–204. doi: 10.1038/bjc.2012.493. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Key T, Appleby P, Barnes I, et al. Endogenous Hormones and Breast Cancer Collaborative Group. Endogenous sex hormones and breast cancer in post menopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst. 2002;94:606–616. doi: 10.1093/jnci/94.8.606. [DOI] [PubMed] [Google Scholar]
- 6.Eliassen AH, Missmer SA, Tworoger SS, et al. Endogenous steroid hormone concentrations and risk of breast cancer among pre menopausal women. J Natl Cancer Inst. 2006;98:1406–1415. doi: 10.1093/jnci/djj376. [DOI] [PubMed] [Google Scholar]
- 7.James RE, Lukanova A, Dossus L, et al. Postmenopausal serum sex steroids and risk of hormone receptor-positive and -negative breast cancer: a nested case-control study. Cancer Prev Res (Phila) 2011;4:1626–1635. doi: 10.1158/1940-6207.CAPR-11-0090. [DOI] [PubMed] [Google Scholar]
- 8.Key TJ, Appleby PN, Reeves GK, et al. Endogenous Hormones and Breast Cancer Collaborative Group. Insulin-like growth factor 1 (IGF1), IGF binding protein 3 (IGFBP3), and breast cancer risk: pooled individual data analysis of 17 prospective studies. Lancet Oncol. 2010;11:530–542. doi: 10.1016/S1470-2045(10)70095-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lagiou P, Georgila C, Samoli E, et al. Estrogen alpha and progesterone receptor expression in the normal mammary epithelium in relation to breast cancer risk. Int J Cancer. 2009;124:440–442. doi: 10.1002/ijc.23899. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Johansson H, Baglietto L, Guerrieri-Gonzaga A, et al. Factors associated with circulating levels of insulin-like growth factor-I and insulin-like growth factor binding protein-3 in 740 women at risk for breast cancer. Breast Cancer Res Treat. 2004;88:63–73. doi: 10.1007/s10549-004-0746-9. [DOI] [PubMed] [Google Scholar]
- 11.Barnes BB, Chang-Claude J, Flesch-Janys D, et al. Cancer risk factors associated with insulin-like growth factor (IGF)-I and IGF-binding protein-3 levels in healthy women: effect modification by menopausal status. Cancer Causes Control. 2009;20:1985–1996. doi: 10.1007/s10552-009-9393-0. [DOI] [PubMed] [Google Scholar]
- 12.Holdaway IM, Mason BH, Lethaby AE, et al. Serum levels of insulin-like growth factor binding protein-3 in benign and malignant breast disease. Aust N Z J Surg. 1999;69:495–500. doi: 10.1046/j.1440-1622.1999.01610.x. [DOI] [PubMed] [Google Scholar]
- 13.Rice MS, Tamimi RM, Connolly JL, et al. Insulin-like growth factor-1, insulin-like growth factor binding protein-3 and lobule type in the Nurses' Health Study II. Breast Cancer Res. 2012;14:R44. doi: 10.1186/bcr3141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Neuenschwander S, Schwartz A, Wood TL, et al. Involution of the lactating mammary gland is inhibited by the IGF system in a transgenic mouse model. J Clin Invest. 1996;97:2225–2232. doi: 10.1172/JCI118663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hankinson SE, Willett WC, Colditz GA, et al. Circulating concentrations of insulin-like growth factor-I and risk of breast cancer. Lancet. 1998;351:1393–1396. doi: 10.1016/S0140-6736(97)10384-1. [DOI] [PubMed] [Google Scholar]
- 16.Schernhammer ES, Holly JM, Pollak MN, et al. Circulating levels of insulin-like growth factors, their binding proteins, and breast cancer risk. Cancer Epidemiol Biomarkers Prev. 2005;14:699–704. doi: 10.1158/1055-9965.EPI-04-0561. [DOI] [PubMed] [Google Scholar]
- 17.Schernhammer ES, Holly JM, Hunter DJ, et al. Insulin-like growth factor-I, its binding proteins (IGFBP-1 and IGFBP-3), and growth hormone and breast cancer risk in The Nurses Health Study II. Endocr Relat Cancer. 2006;13:583–592. doi: 10.1677/erc.1.01149. [DOI] [PubMed] [Google Scholar]