Relationship of circulating insulin-like growth factor-I and binding proteins 1–7 with mammographic density among women undergoing image-guided diagnostic breast biopsy

Increased mammographic breast density (MD) is an established risk factor for breast cancer [1, 2], which reflects the relative proportions of fibroglandular and adipose tissue content within the breast [3, 4]. However, most women with high MD neither have prevalent tumor nor will develop one in future. The determinants of MD and the mechanisms by which it increases the risk of breast cancer are poorly understood. Identifying factors associated with MD could elucidate the mechanisms by which breast density contributes to the risk of breast cancer and may improve risk prediction. Several lines of evidence have demonstrated the role of the insulin-like growth factor (IGF) system in breast cancer as either a tumor suppressor or promoter [5‐7]. IGF proteins may be an underlying link between MD and the risk of breast cancer.

The insulin-like growth factor (IGF) system comprises transmembrane growth factor receptors, growth factor ligands (IGFs), IGF binding proteins (IGFBPs), proteases, and IGFBP-related proteins (IGFBP-rP) [8‐10] (Fig. 1). IGF-I is produced by the liver in response to growth hormone produced by the pituitary gland. Bioavailability of IGF-I is regulated by numerous IGFBPs. IGFBPs have high affinity to IGFs and modulate their access to IGF receptors. IGF proteases dissociate the IGFBP-IGF complex and the freed IGFs can then bind to the IGF receptors, activating downstream signaling pathways: the phosphatidylinositol-3-kinase and RAS-extracellular signal-regulated kinase pathways [11]. Activation of these tyrosine kinase pathways plays a significant role in cell proliferation, differentiation, and metabolism. Indeed, dysregulated IGF-axis proteins have been shown to play a critical role in carcinogenesis of several human cancers including breast cancer [12].

Laboratory studies suggest that IGF-I may promote breast tumorigenesis by stimulating cell proliferation and inhibiting apoptosis [10, 13]. A pooled analysis of 17 prospective epidemiologic studies showed that circulating IGF-I was associated with a 25% increased risk of breast cancer comparing women from the highest quintile to the lowest quintile, and this association did not vary by menopausal status [14]. Higher circulating concentrations of IGF-I as well as the ratio of IGF-I:IGFBP-3 have been shown to be associated with higher MD in premenopausal women in some studies [15‐18], whereas others have not observed an association [16, 19‐22]. Among postmenopausal women, there is little evidence for a relationship between IGF-I, IGF-I:IGFBP-3, and MD [16‐19, 21‐24].

While prior studies have largely focused on the role of IGF-I, IGFBP-3, and the ratio of IGF-I:IGFBP-3 in MD, the relationship of other IGFBPs (IGFBP-1, 2, 4, 5, 6, and 7 and their ratio to IGF-I) with MD has not been elucidated. Epidemiological studies have not extensively explored the associations of other IGFBPs (IGFBP 4, 5, 6, and 7) and breast cancer risk, with the exception of IGFBP-2, where findings have been inconsistent [25‐27]. Laboratory and clinical data also suggest a role for IGFBPs in mammary gland development and breast carcinogenesis [28‐34]. Although IGFBPs have significant sequence homology, they have different structures, binding patterns, and affinities [35]. IGFBPs are also known to exert IGF-dependent and independent actions [36, 37], with distinct functions in the breast [38]. Thus, in order to fully understand the role of the IGF system in MD, a panel of IGBPs may need to be investigated. We therefore assessed the relationship between serum levels of IGF-I and seven of its binding proteins with area and volume measures of MD in a cross-sectional study of women referred for image-guided diagnostic breast biopsy.

In this study of women undergoing diagnostic breast biopsy, we undertook an analysis of circulating IGF-I and seven of its binding proteins to examine relationships with MD. We did not observe clear associations between IGF-I, IGFBP-3, and MD after BMI adjustment. While prior studies have largely focused on IGF-I and IGFBP-3, we identified novel positive associations between IGFBP-2 and percent MD that withstood rigorous adjustment for BMI, suggesting that IGFBP-2 may play a key role in MD. In concert with in vitro studies demonstrating a dual role of IGFBP-2 on breast tissue (i.e., promoting cell proliferation and inhibiting adipogenesis), our findings suggest that further studies investigating the role of IGFBP-2 in breast tissue composition and the carcinogenic process are warranted.

Prior work indicates that the relation of IGF-I:IGFBP-3 with MD differs by menopausal status, with most studies demonstrating positive associations for premenopausal women, but null associations for postmenopausal women [15‐19, 21‐24]. We too observed similar associations by menopausal status that were attenuated upon adjusting fully for BMI. In addition to the widely studied IGF proteins (IGF-I and IGFBP-3), this study assessed for the first time the relationship of other IGF binding proteins with MD.

Interestingly, we identified IGFBP-2 as being positively associated with percent MD among all women, a finding which persisted after adjustment for BMI, despite a strong and inverse relation of BMI with both IGFBP-2 [26] and percent MD [39]. When we stratified analyses by BMI, we found that the observed positive association between IGFBP-2 and percent MD was most apparent among women with lower BMI, particularly among postmenopausal women. Though we cannot rule out the possibility that the association between IGFBP-2 and percent MD may be due to residual confounding by adiposity, both laboratory and epidemiologic studies have suggested an important role of IGFBP-2 in breast cancer etiology. Prior epidemiologic studies have reported inverse associations between IGFBP-2 and risk of atypical hyperplasia [26] and postmenopausal breast cancer [27]. Other studies have reported null findings for circulating IGFBP-2-associated breast cancer risk relationships [25, 49, 50]. Experimental studies have shown that IGFBP-2 can either inhibit or promote tumorigenesis [30, 51]. IGFBP-2 may inhibit tumorigenesis by inhibiting adipogenesis [52] and promoting apoptosis in an IGF-independent mechanism [53]. Conversely, IGFBP-2 is thought to induce tumorigenesis by promoting cell proliferation and invasion through suppression of PTEN activation and prolonged activity of PI3K [30, 51, 54] . Thus, with IGFBP-2’s hypothesized dual effect on breast tissue, it is biologically plausible that the proliferative effect of IGFBP-2 on breast epithelium, along with its inhibition of preadipocyte differentiation into mature adipocytes [55], may be reflected radiologically in elevated breast density.

The relationship between IGF proteins and MD measures varied by menopausal status in this study. As several of the IGF proteins as well as the MD measures vary by age and menopausal status, this finding is not necessarily surprising. Further, differences in interactions between circulating hormones, IGF proteins, and the tissue microenvironment [56, 57] may contribute to observed differences by menopausal status. For example, estrogen, a strong breast cancer risk factor, both regulates and is influenced by the IGF family [58]. To better understand the relationship between IGF proteins and MD, future larger studies assessing interrelationships between IGF, hormones, and breast tissue composition in pre-, peri-, and postmenopausal women will be important.

In this study, both volumetric and area mammographic density measures were assessed, and IGF associations with MD-A and MD-V were largely consistent. We have previously shown volumetric and area measures of MD to be highly correlated [39]. We found in multivariate analyses that IGF proteins appeared to be associated with percent MD-V/MD-A, but not absolute dense volume/area measures. Interestingly, we also observed that many of the IGFBPs were inversely associated with non-dense area, suggesting that IGFBPs may also be important correlates of adipose tissue composition in the breast. Indeed, in vitro studies have suggested a role for several IGFBPs in adipogenesis [59]. The role of breast adipose tissue reflected by the radiologic non-dense area in breast cancer development is not well understood [60, 61]. Future mechanistic studies are warranted to study interrelationships between adipocyte biology and endogenous growth factors like IGFBP, with stromal, epithelial, and adipose tissues that are reflected radiologically in MD.

A major strength of this study is that we investigated the relationship of multiple IGFBPs, in addition to the widely studied IGF-I and IGFBP-3, with quantitative, reliable measures of MD. IGF assays used were also highly reliable, with the IGF proteins demonstrating several expected relationships with breast cancer risk factors [14, 62‐67]. For example, we found that IGF-I was inversely associated with age, likely due to declines in growth hormone secretion in with aging [63, 66‐70]. As expected, IGFBP-2 was inversely associated with BMI [26]. In contrast to published studies [14, 64, 65, 67], we did not see an association of IGF-I with BMI, though some studies report that only the most extreme BMI levels are associated with the lowest IGF-I, whereas others hypothesize that IGF-I increases with body weight until a threshold is reached and then it triggers the feedback loop suppressing GH secretion. A non-linear U-shaped relationship between BMI and IGF-I has also been reported [71, 72]. We possibly did not see the association due to the low prevalence of obesity in our study population. In examining published literature on serums IGF levels in pre- and postmenopausal women, we found that the levels of IGF-I, IGFBP-3, and their ratio that we observed were within comparable ranges to those reported by the Nurses’ Health Study [19]. Likewise, levels of IGFBP-2 in our study were comparable with those previously reported from the nested case-control study within Women’s Health Initiative Clinical Trial [26] and within a general community healthcare plan setting [27], lending external validity to our results.

A limitation of this study is that the circulating IGF proteins were measured in a single serum sample, which may not be an adequate representation of future IGF levels within a woman. However, a stability study of IGF-I and IGFBP-3 among premenopausal women in NHS-II demonstrated that the 3-year intraclass correlation coefficient was 0.70 for IGF-I and 0.74 for IGFBP-3 [73]. Thus, a single measurement of IGF-I and BP-3 may be representative for at least 3 years. For other IGF proteins, stability studies should be carried out in the future. In our study, a high proportion of IGFBP-1 (N = 91) was below the LLOD, perhaps precluding our ability to detect an association between IGFBP-1 and MD. Though alcohol drinking has been associated with MD [74‐78], breast cancer risk [79], and IGF measures [80], information on alcohol consumption was not collected as part of this study. Our study also had a relatively small sample size, particularly upon stratification by menopausal status; larger studies are needed to validate our observed associations between IGF and MD measures. In addition, we measured circulating IGF proteins, and systemic IGFs may differ from those regulating tissue levels [81, 82]. Nevertheless, there is evidence demonstrating that local tissue expression of the IGF proteins may influence the systemic expression of IGF proteins [83, 84].

We demonstrated a novel positive association between circulating IGFBP-2 and percent MD, particularly among women with lower BMI. These results contribute to the rationale for evaluating relationships between serum IGFs, their multiple binding proteins, and breast cancer risk factors, including MD, to advance our understanding of the IGF proteins in breast cancer etiology.