Abstract
Colony stimulating factor-1 (CSF1) and its receptor (CSF1-R) are important in mammary gland development and have been implicated in breast carcinogenesis. In a nested case-control study in the Nurses' Heath Study of 726 breast cancer cases diagnosed between June 1, 1992, and June 1, 1998, and 734 matched controls, we prospectively evaluated whether circulating levels of CSF1 (assessed in 1989–1990) are associated with breast cancer risk. The association varied by menopausal status (Pheterogeneity=0.009). CSF1 levels in the highest quartile (versus lowest) were associated with an 85% reduced risk of premenopausal breast cancer (RR=0.15, 95%CI 0.03–0.85, Ptrend=0.02). In contrast, CSF1 levels in the highest quartile conferred a 33% increased risk of postmenopausal breast cancer (RR=1.33, 95%CI 0.96–1.86; Ptrend =0.11), with greatest risk for invasive (RR=1.45, 95%CI 1.02–2.07, Ptrend=0.06) and ER+/PR+ tumors (RR=1.72, 95%CI 1.11–2.66; Ptrend =0.04). Thus, the association of circulating CSF1 levels and breast cancer varies by menopausal status.
Keywords: Colony stimulating factor 1, breast cancer, CSF1
Introduction
Macrophage colony-stimulating factor (CSF1), originally identified as a hematopoietic growth factor, stimulates proliferation, differentiation, and survival of monocytes and macrophages (1). More recently, CSF1 and its receptor (CSF1R) have exhibited an important role in mammary gland development (2, 3) and are implicated in breast and ovarian carcinogenesis (3). Many cell types, including epithelial and mesenchymal cells and osteoblasts, produce CSF1 (4). While expressed at low levels in normal resting breast epithelium (5), CSF1 is expressed at high levels in the mammary gland during pregnancy and lactation and in breast tumors (6–8).
CSF1's proliferative effects are mediated by its binding to CSF1-R and the induction of signal transduction pathways. Autocrine CSF1R activation induces hyperproliferation and loss of basement membrane integrity in human mammary epithelial cells (9). Levels of circulating CSF1 are higher in patients with ovarian, breast, and endometrial cancer than in healthy individuals (8, 10). CSF1 expression in breast tumors also correlates with grade and progression (3, 8). No studies of prospectively measured levels of circulating CSF1 and subsequent risk of breast cancer have been published. Thus, it is unclear whether circulating CSF1 is a tumor marker for breast cancer or predictive of breast cancer risk. We conducted a study within the Nurses' Health Study to determine whether circulating levels of CSF1 predict breast cancer risk.
Materials and Methods
Study Population
The Nurses' Health Study began in 1976, when 121,700 US registered nurses age 30–55 returned a questionnaire. Information on body mass index (BMI), reproductive history, age at menopause, postmenopausal hormone (PMH) use, and diagnosis of cancer and other diseases is updated every two years through questionnaires. During 1989 and 1990, blood samples were collected from 32,826 women (11), with 99% follow-up through 1998.
In a nested case-control study among women who provided blood samples, we included 726 reporting breast cancer diagnosis between June 1, 1994 and June 1, 1998 and 734 matched controls with no cancer history. Median time from blood draw to breast cancer diagnosis was 5.5 years (interquartile range: 4–7.1 years). Breast cancer cases were confirmed by medical record review; estrogen receptor (ER) and progesterone receptor (PR) status were obtained from pathology reports. This study was approved by the Committee on the Use of Human Subjects in Research at Brigham and Women's Hospital.
CSF1 Measurements
CSF1 was measured by ELISA in Dr. Nader Rifai's laboratory at Children's Hospital, Boston, MA. The assay coefficient of variation was 8.2%. A reproducibility study among 51 participants who provided three blood samples over 3 years demonstrated that one blood measure is well correlated with longer-term measures (intraclass correlation coefficient=0.65, 95%CI 0.51–0.76). Using the extreme Studentized deviate Many-Outlier procedure to determine outlying CSF1 values (12), one control was excluded with a CSF1 value of 2654.1 pg/ml.
Statistical Analysis
We used conditional and unconditional logistic regression models to estimate the relative risk (RR) of breast cancer, with 95% confidence intervals (CI) adjusted for known risk factors. Quartiles of circulating CSF1 were based on the distribution among controls. In analyses stratified by menopausal status, categories were based on premenopausal and postmenopausal distributions separately. The simple conditional model was based on 725 matched case-control pairs with circulating CSF1 data. Multivariate models were adjusted for prior benign breast disease (yes/no), BMI at age 18 (continuous), weight gain (<5, 5−<20, 20+, missing), parity/age at first birth(nulliparous,1–2 children first birth le 24, 1–2 children parous first birth after 24, 3+ children first birth le 24, 3+ children parous first birth after 24), alcohol (none, <3, 3–6, 7–13 drinks/week), family history of breast cancer (yes/no), age at menarche (<12, 12, 13, >13), age at menopause (<=45, 46–50, 51+), duration of PMH use (premenopausal, never, past <60, past60+, current <60, current 60+). Menopausal status and use of postmenopausal hormones at blood collection were assessed through a supplemental questionnaire administered at that time. All other covariates were assessed from biennial questionnaires. Tests for trend were based on the Wald test when the log transformed continuous measure of circulating CSF1 was included as an independent variable. Polytomous logistic regression (13) was used to test for differences in trend across CSF1 levels according to ER+/PR+ to ER−/PR− tumors. There were too few cases of ER−/PR+ breast cancers for these groups to be considered separately.
Results and Discussion
Mean circulating levels of CSF1 were similar for cases (584.7 pg/ml, range 163.1–2170.5 pg/ml) and controls (583.7 pg/ml, range 169.7–2078.7 pg/ml). Among controls, circulating CSF1 was positively associated with BMI and weight gain since age 18 and inversely associated with alcohol consumption (Table 1). Women with the highest levels of circulating CSF1 were more likely to be postmenopausal, have a family history of breast cancer, and be parous than women with lower levels.
Table 1.
Age and age-adjusted characteristics among controls (N=734) according to CSF1 levels, Nurses' Health Study (1992–1998).
| Q1 (N=184) | Q2 (N=183) | Q3 (N=184) | Q4 (N=183) | |
|---|---|---|---|---|
| Median (pg/ml) | 359.5 | 487.9 | 602.9 | 832.2 |
| Range (pg/ml) | (169.7–424.8) | (425.1–540.5) | (540.9–689.4) | (692.6–2078.7) |
| Means | ||||
| Age, y | 56.3 | 56.9 | 58.3 | 56.2 |
| Age at menarche, y | 12.6 | 12.8 | 12.6 | 12.6 |
| Age at first birth1, y | 25.0 | 24.8 | 25.1 | 24.6 |
| Age at menopause2, y | 48.2 | 47.7 | 46.8 | 47.1 |
| Parity1 | 2.8 | 3.1 | 2.9 | 3.0 |
| BMI, kg/m2 | 24.6 | 24.8 | 25.5 | 26.4 |
| BMI at 18, kg/m2 | 21.2 | 21.1 | 21.5 | 21.4 |
| Weight gain since 18, kg | 8.9 | 10.0 | 11.1 | 13.5 |
| Alcohol, g | 6.4 | 5.8 | 4.7 | 3.6 |
| Frequency, % | ||||
| Premenopausal | 24.5 | 23.2 | 22.0 | 20.7 |
| Family history of BC | 13.8 | 14.5 | 15.2 | 15.7 |
| Prior benign breast disease | 56.6 | 51.2 | 51.2 | 53.9 |
| Nulliparous | 10.1 | 6.7 | 9.1 | 4.1 |
| Current PMH users2 | 59.5 | 53.2 | 59.1 | 61.1 |
| Not current PMH users2 | 40.5 | 46.8 | 40.9 | 38.9 |
| Never PMH users2 | 28.7 | 29.0 | 29.2 | 23.0 |
Among parous women only.
Among postmenopausal women only.
Overall, circulating CSF1 levels and risk of breast cancer (Ptrend =0.37; Table 2) were not associated. However, the association varied by menopausal status (Pheterogeneity =0.009). Women with CSF1 levels in the highest (vs lowest) quartile had an 85% reduced risk of premenopausal breast cancer (RR=0.15, 95%CI 0.03–0.85, Ptrend =0.02) and a contrasting nonsignificant 33% increased risk of postmenopausal breast cancer (RR=1.33, 95%CI 0.96–1.86; Ptrend =0.11). The association between CSF1 and postmenopausal breast cancer appeared stronger when limited to invasive cancers (RR=1.45, 95%CI 1.02–2.07; Ptrend =0.06; Table 3). CSF1 was associated with ER+/PR+ breast cancers (RR=1.72, 95%CI 1.11–2.66; Table 3), but not ER−/PR− tumors (RR=0.70, 95%CI 0.34–1.44; Pheterogeneity=0.03; Table 3).
Table 2.
Relative risk (RR) and 95% confidence interval of breast cancer associated with quartiles of circulating CSF1 in the Nurses' Health Study (1992–1998).
| Q1 | Q2 | Q3 | Q4 | P-trend1 | P-Het | |
|---|---|---|---|---|---|---|
| All Women | ||||||
| Median (pg/ml)2 | 359.4 | 487.2 | 602.9 | 833.3 | ||
| Cases/controls | 183/181 | 173/180 | 174/182 | 195/182 | ||
| Conditional3 | 1.0(Ref) | 0.95(0.71–1.28) | 0.95(0.70–1.28) | 1.06(0.79–1.43) | 0.61 | |
| MV4 Conditional RR | 1.0(Ref) | 1.00(0.73–1.38) | 0.96(0.69–1.33) | 1.15(0.82–1.60) | 0.37 | |
| Premenopausal | 0.009 | |||||
| Median (pg/ml)2 | 365.8 | 467.7 | 648.6 | 882.7 | ||
| Cases/controls | 18/15 | 20/15 | 16/15 | 4/14 | ||
| Matching factors | 1.0(Ref) | 0.96(0.34–2.67) | 0.80(0.27–2.39) | 0.17(0.04–0.69) | 0.03 | |
| MV5 RR | 1.0(Ref) | 0.97(0.27–3.56) | 0.55(0.12–2.44) | 0.15(0.03–0.85) | 0.02 | |
| Postmenopausal | ||||||
| Median (pg/ml)2 | 356.3 | 488.3 | 595.0 | 817.2 | ||
| Cases/controls | 156/155 | 141/155 | 138/157 | 185/155 | ||
| Matching factors | 1.0(Ref) | 0.93(0.67–1.28) | 0.90(0.65–1.24) | 1.21(0.89–1.66) | 0.20 | |
| MV6 RR | 1.0(Ref) | 0.98(0.70–1.36) | 0.90(0.65–1.26) | 1.33(0.96–1.86) | 0.11 |
P for trend based on the log transformed continuous measure of circulating CSF1.
Median of the quartiles is based on the distribution among controls.
Simple conditional model based on 725 matched case-control pairs with circulating CSF1 data.
Conditional logistic model adjusted for: Benign breast disease, BMI at age 18, weight gain , parity/age at first birth, alcohol, family history of breast cancer, age at menarche, age at menopause, duration of PMH use
Unconditional model adjusted for matching factors and benign breast disease, BMI at age 18, weight gain, parity/age at first birth, alcohol, family history of breast cancer, age at menarche.
Unconditional model adjusted for matching factors and benign breast disease, BMI at age 18, weight gain, parity/age at first birth, alcohol, family history of breast cancer, age at menarche, age at menopause, duration of PMH use.
Table 3.
Relative risk (RR) and 95% confidence interval of postmenopausal breast cancer associated with quartiles of circulating CSF1 in the Nurses' Health Study (1992–1998) according to tumor characteristics.
| Q1 | Q2 | Q3 | Q4 | P-trend1 | P-Het | |
|---|---|---|---|---|---|---|
| Postmenopausal only | ||||||
| In Situ | ||||||
| Cases/controls | 33/155 | 25/155 | 20/157 | 28/155 | ||
| Matching factors | 1.0(Ref) | 0.75(0.42–1.34) | 0.65(0.35–1.20) | 0.88(0.50–1.55) | 0.86 | |
| MV2 RR | 1.0(Ref) | 0.79(0.43–1.45) | 0.67(0.35–1.28) | 1.05(0.56–1.97) | 0.82 | |
| Invasive | ||||||
| Cases/controls | 117/155 | 111/155 | 109/157 | 151/155 | ||
| Matching factors | 1.0(Ref) | 0.97(0.69–1.37) | 0.94(0.66–1.33) | 1.31(0.94–1.83) | 0.12 | |
| MV2 RR | 1.0(Ref) | 1.02(0.71–1.47) | 0.96(0.67–1.38) | 1.45(1.02–2.07) | 0.06 | |
| ER+/PR+ | 0.03 | |||||
| Cases/controls | 60/155 | 64/155 | 63/157 | 90/155 | ||
| Matching factors | 1.0(Ref) | 1.10(0.72–1.68) | 1.15(0.76–1.74) | 1.45(0.96–2.18) | 0.10 | |
| MV2 RR | 1.0(Ref) | 1.21(0.78–1.88) | 1.21(0.78–1.87) | 1.72(1.11–2.66) | 0.04 | |
| ER+/PR− | ||||||
| Cases/controls | 8/155 | 15/155 | 9/157 | 11/155 | ||
| Matching factors | 1.0(Ref) | 1.85(0.75–4.58) | 1.24(0.45–3.40) | 1.38(0.53–3.60) | 0.85 | |
| MV2 RR | 1.0(Ref) | 2.17(0.83–5.72) | 1.48(0.51–4.28) | 1.52(0.53–4.34) | 0.97 | |
| ER−/PR− | ||||||
| Cases/controls | 25/155 | 18/155 | 11/157 | 21/155 | ||
| Matching factors | 1.0(Ref) | 0.83(0.43–1.58) | 0.56(0.27–1.16) | 0.81(0.42–1.56) | 0.32 | |
| MV2 RR | 1.0(Ref) | 0.80(0.41–1.58) | 0.50(0.24–1.06) | 0.70(0.34–1.44) | 0.16 |
P for trend based on the log transformed continuous measure of circulating CSF1.
Unconditional model adjusted for matching factors and benign breast disease, BMI at age 18, weight gain, parity/age at first birth, alcohol, family history of breast cancer, age at menarche, age at menopause, duration of PMH use.
Because we observed a strong positive association between circulating CSF1 with both weight gain since age 18 and current BMI (Table 1), and adiposity is a risk factor for breast cancer we also ran multivariate models adjusted for both weight gain since age 18 and current BMI (continuous). The results were essentially unchanged from those presented in Tables 2 and 3.
To understand the divergence of associations by menopausal status, we conducted stratified analyses among postmenopausal women with higher estrogen environments. However, the CSF1-breast cancer association was not modified by postmenopausal hormone use (Pheterogeneity=0.68) or BMI (Pheterogeneity=0.29).
To our knowledge, this is the first prospective study to examine circulating levels of CSF1 and subsequent risk of breast cancer. CSF1 was inversely related to premenopausal and positively associated with postmenopausal breast cancer. Although a priori we would not have predicted differential effects for CSF1 by menopausal status, the association of other exposures (e.g., BMI, circulating IGF1 levels) with breast cancer depends on menopausal status. The mechanism by which CSF1 may differentially influence premenopausal and postmenopausal breast cancer risk is unclear. Additional studies in premenopausal women are necessary to confirm these results.
One study suggests that CSF1 can influence breast cell proliferation either positively or negatively depending on the estrogen environment (14). In ER+ cell lines, estradiol induced a three- to fivefold increase in growth, while CSF1 alone did not (14). However, in combination, CSF1 inhibited the proliferative effects of estradiol by inducing G1 arrest. In contrast, earlier work in ER− cell lines showed that CSF1 induced proliferation (15, 16). These data suggest that in a high-estrogen environment CSF1 inhibits proliferative effects of mitogens such as estradiol.
Accumulating evidence supports a role for CSF1 in breast carcinogenesis. Activation of CSF1R causes uncontrolled growth (9), increases the invasive potential of epithelial cells (17), and promotes angiogenesis (18). In animal models, blockade of CSF1 through anti-sense oligonucleotides or neutralizing anti-CSF1 antibodies suppressed tumor growth and prolonged long-term survival (19, 20). Thus, CSF1R may be a target for breast cancer chemoprevention.
This initial prospective study of circulating CSF1 levels suggests that CSF1 is a biomarker of subsequent postmenopausal breast cancer risk. Additional studies will be necessary to replicate these findings. This study was limited by the number of premenopausal breast cancer cases; it remains to be seen whether CSF1 has disparate effects in premenopausal and postmenopausal women.
Acknowledgments
Funding Support: Supported by Public Health Service Grants CA087969, CA049449, and CA075016, SPORE in Breast Cancer CA089393, from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services and a grant from the Breast Cancer Research Foundation, New York, New York. Dr. Graham Colditz is supported in part by an American Cancer Society Cissy Hornung Clinical Research Professorship.
References
- 1.Roth P, Stanley ER. The biology of CSF-1 and its receptor. Curr Top Microbiol Immunol. 1992;181:141–67. doi: 10.1007/978-3-642-77377-8_5. [DOI] [PubMed] [Google Scholar]
- 2.Pollard JW, Hennighausen L. Colony stimulating factor 1 is required for mammary gland development during pregnancy. Proc Natl Acad Sci U S A. 1994;91:9312–6. doi: 10.1073/pnas.91.20.9312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sapi E. The role of CSF-1 in normal physiology of mammary gland and breast cancer: an update. Exp Biol Med (Maywood) 2004;229:1–11. doi: 10.1177/153537020422900101. [DOI] [PubMed] [Google Scholar]
- 4.Wiktor-Jedrzejczak W, Gordon S. Cytokine regulation of the macrophage (M phi) system studied using the colony stimulating factor-1-deficient op/op mouse. Physiol Rev. 1996;76:927–47. doi: 10.1152/physrev.1996.76.4.927. [DOI] [PubMed] [Google Scholar]
- 5.Ryan GR, Dai XM, Dominguez MG, et al. Rescue of the colony-stimulating factor 1 (CSF-1)-nullizygous mouse (Csf1(op)/Csf1(op)) phenotype with a CSF-1 transgene and identification of sites of local CSF-1 synthesis. Blood. 2001;98:74–84. doi: 10.1182/blood.v98.1.74. [DOI] [PubMed] [Google Scholar]
- 6.Tang R, Beuvon F, Ojeda M, et al. M-CSF (monocyte colony stimulating factor) and M-CSF receptor expression by breast tumour cells: M-CSF mediated recruitment of tumour infiltrating monocytes? J Cell Biochem. 1992;50:350–6. doi: 10.1002/jcb.240500403. [DOI] [PubMed] [Google Scholar]
- 7.Scholl SM, Pallud C, Beuvon F, et al. Anti-colony-stimulating factor-1 antibody staining in primary breast adenocarcinomas correlates with marked inflammatory cell infiltrates and prognosis. J Natl Cancer Inst. 1994;86:120–6. doi: 10.1093/jnci/86.2.120. [DOI] [PubMed] [Google Scholar]
- 8.Kacinski BM. CSF-1 and its receptor in ovarian, endometrial and breast cancer. Ann Med. 1995;27:79–85. doi: 10.3109/07853899509031941. [DOI] [PubMed] [Google Scholar]
- 9.Wrobel CN, Debnath J, Lin E, et al. Autocrine CSF-1R activation promotes Src-dependent disruption of mammary epithelial architecture. J Cell Biol. 2004;165:263–73. doi: 10.1083/jcb.200309102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lawicki S, Szmitkowski M, Wojtukiewicz M. The pretreatment plasma level and diagnostic utility of M-CSF in benign breast tumor and breast cancer patients. Clin Chim Acta. 2006;371:112–6. doi: 10.1016/j.cca.2006.02.033. [DOI] [PubMed] [Google Scholar]
- 11.Hankinson SE, Willett WC, Manson JE, et al. Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women. J Natl Cancer Inst. 1998;90:1292–9. doi: 10.1093/jnci/90.17.1292. [DOI] [PubMed] [Google Scholar]
- 12.Rosner B. Percentage points for a generalized ESD Many-Outlier procedure. Technometrics. 1983;25:165–72. [Google Scholar]
- 13.Marshall RJ, Chisholm EM. Hypothesis testing in the polychotomous logistic model with an application to detecting gastrointestinal cancer. Stat Med. 1985;4:337–44. doi: 10.1002/sim.4780040313. [DOI] [PubMed] [Google Scholar]
- 14.Lee AW, Nambirajan S, Moffat JG. CSF-1 activates MAPK-dependent and p53-independent pathways to induce growth arrest of hormone-dependent human breast cancer cells. Oncogene. 1999;18:7477–94. doi: 10.1038/sj.onc.1203123. [DOI] [PubMed] [Google Scholar]
- 15.Sapi E, Flick MB, Rodov S, Kacinski BM. Ets-2 transdominant mutant abolishes anchorage-independent growth and macrophage colony-stimulating factor-stimulated invasion by BT20 breast carcinoma cells. Cancer Res. 1998;58:1027–33. [PubMed] [Google Scholar]
- 16.Filderman AE, Bruckner A, Kacinski BM, Deng N, Remold HG. Macrophage colony-stimulating factor (CSF-1) enhances invasiveness in CSF-1 receptor-positive carcinoma cell lines. Cancer Res. 1992;52:3661–6. [PubMed] [Google Scholar]
- 17.Sapi E, Flick MB, Rodov S, et al. Independent regulation of invasion and anchorage-independent growth by different autophosphorylation sites of the macrophage colony-stimulating factor 1 receptor. Cancer Res. 1996;56:5704–12. [PubMed] [Google Scholar]
- 18.Aharinejad S, Marks SC, Jr., Bock P, et al. CSF-1 treatment promotes angiogenesis in the metaphysis of osteopetrotic (toothless, tl) rats. Bone. 1995;16:315–24. doi: 10.1016/8756-3282(94)00044-1. [DOI] [PubMed] [Google Scholar]
- 19.Aharinejad S, Paulus P, Sioud M, et al. Colony-stimulating factor-1 blockade by antisense oligonucleotides and small interfering RNAs suppresses growth of human mammary tumor xenografts in mice. Cancer Res. 2004;64:5378–84. doi: 10.1158/0008-5472.CAN-04-0961. [DOI] [PubMed] [Google Scholar]
- 20.Paulus P, Stanley ER, Schafer R, Abraham D, Aharinejad S. Colony-stimulating factor-1 antibody reverses chemoresistance in human MCF-7 breast cancer xenografts. Cancer Res. 2006;66:4349–56. doi: 10.1158/0008-5472.CAN-05-3523. [DOI] [PubMed] [Google Scholar]