This population-based cross-sectional study shows an association between a common genetic haplotype in IGF1 and absolute mammographic density in postmenopausal women after adjustment for age and BMI. Although not statistically significant, stratification by age and BMI revealed that the upper age tertiles and middle BMI tertile increased the mammographic density level. One haplotype in IGF2 was associated with the levels of both IGF1 and IGFBP3, while two haplotypes in the IGF2R gene were associated with the levels of IGF1. Within the IGFBP3 gene, two haplotypes were found associated with the IGFBP3 level indicating a regulation in cis.
The strength of our study is the large sample size and the fact that the samples were collected as part of a population-based screening project with high attendance rate. Highly experienced personnel that were blinded to the characteristics of the women performed the reading and measurements of both the mammographic density and hormone levels. Also, we had information on age, BMI and HT use and were able to both adjust and stratify for these variables when necessary.
The limitation of our study is that the associations made with the polymorphisms within IGF2R were difficult to interpret due to the lack of measurements of its ligand IGF2. The women in the study were all postmenopausal, and some were taking HT, which could influence the circulating levels of IGF1 and IGFBP3. Furthermore, HT is demonstrated to have an impact on mammographic measurements, increasing the density. However, as HT cannot have influenced genotype, it is not technically a confounder, and as in most other analyses of mammographic density, we adjusted for age and BMI. We did however, stratify for HT, predominantly because of the possibility that HT use could be an effect modifier, i.e. have modified the effect of genotype on mammographic density.
Mammographic density is reduced by successive pregnancies and menopause, as well as with advancing age. Furthermore, mammographic density may reflect the cumulative exposure to hormones and growth factors that stimulate cell division and growth in the breast. Pike and colleagues proposed a model, stating that the effects of hormone exposure throughout life and the accumulation of genetic damage may cause an increased probability of breast cancer later in life . The age-specific absolute risk of breast cancer caused by mammographic density is not yet determined, and it is unknown whether interventions that reduce cumulative exposure to density will reduce risk of breast cancer . Because of the role of the IGF pathway in breast development and cellular proliferation, genetic variation within this pathway is of interest. Similar to mammographic density, IGF1 levels are related to age , BMI, and menopause status and young women tend to have higher IGF1 levels than women in their postmenopausal years.
A positive association of IGF1 and IGFBP3 levels in relation to mammographic density in premenopausal women was found in most[17–20, 32]but not all [46–48] previous studies. In postmenopausal women however the results are less consistent [18–21, 47]. Data already published from this study were positive, suggesting an association between IGF1 and mammographic density. Our finding of a common haplotype in IGF1 associated with mammographic density is in agreement with previously published findings of an association with IGF1 levels and density. Associations of genetic variants in IGF1 and mammographic density in postmenopausal women can potentially better reflect the lifetime exposure of IGF1 compared to the IGF1 level measured at a certain time point, at a late stage, in a woman's life.
Other studies [18–20, 46, 47] such as the one of Dos Santos Silva et al. found no association between postmenopausal mammographic density and levels of IGF1, IGF2 or IGFBP3, nor the ratio of IGF1/IGFBP3, although, there was an association between the mammographic lucent area and IGFBP3 serum levels. In summary, many studies have looked at variations in the IGF genes and their relationship with IGF plasma levels and mammographic density, but the results remain inconclusive, emphasizing the need for more studies.
The incidence of breast cancer has been associated with levels of IGF1 and IGFBP3 in premenopausal women in most [22, 49–53], but not all studies [54, 55]. Hankinson et al. performed a nested case-control study and found that IGF1 levels were higher among premenopausal women who developed breast cancer before age 50 than among age matched women who remained cancer free. For the postmenopausal women in the study no such association was established. In postmenopausal women the findings are less clear and positive association of either IGF1, IGFBP3 or both with breast cancer [49, 50, 56] has been reported while other studies are negative[22, 54, 57]. The use of HT by postmenopausal women is known to lower both IGF1 and IGFBP3 levels significantly; thus the IGF1-associated increase in mammographic density seen in the non-HT users, may be difficult to observe in the HT users. Nevertheless, one study reported an increased risk of breast cancer with increasing IGF1 levels also for postmenopausal HT-users (>55 years). Despite lowering the IGF1 levels, HT increases the mammographic density for most women and the age related decrease in mammographic density around the age of 55-64 does not commence in these women . These findings support an emerging model of crosstalk between IGF1 and estrogens, suggesting that estrogens act through their receptor (ER) and affect the IGF1 expression.
Variation in mammographic density due to polymorphisms in the IGF1 gene has been reported in both pre- and postmenopausal women, but the association between genetic variants in IGF1 and mammographic density in breast tissue in postmenopausal women has been inconclusive [11, 17, 46].
Among the common haplotypes analyzed in IGF1 the least frequent haplotype was statistically significantly associated with an increase in ABDEN levels. This haplotype consists of the major allele of rs6220 and the minor allele of rs2162679. Other studies have reported an association between the minor allele of rs6220 and mammographic density in premenopausal women[17, 60] but to our knowledge, no other study has looked at this SNP in relation to postmenopausal mammographic density. Separately, the SNPs comprising this haplotype have been reported to be associated with IGF1 levels and breast cancer risk. The polymorphism rs6220 has also been correlated with elevated IGF1 levels whereas homozygosity G/G of rs2162679, has been associated with reduced breast cancer risk as well as reduced levels of IGFBP3. This is in agreement with our observation of increased mammographic density for haplotype 4, given that low levels of IGFBP3 and high levels of IGF1 have been reported to increase mammographic density.
It is surprising to find the IGF1 haplotype associated with increased mammographic density most strongly in women with higher BMI at an older age. However, since postmenopausal production of estrogens takes place predominantly in the adipose tissue, an increase in BMI would hypothetically result in increased estrogen levels. In turn estrogens may increase cellular IGF1 through crosstalk, and IGF1 may up-regulate the receptor response to estrogens. This haplotype has a frequency of 0.04947 in the population studied which equal to 45 women and could be said to have little power. However, similar results have been reported by Muti et al. who reported that heavier postmenopausal women (BMI>26) had IGF1 levels associated with breast cancer risk . Analysis of the SNPs in IGF1 did not reveal any significant associations with any of the parameters studied (IGF1, IGFBP3, IGFratio or mammographic density), and thus we were unable to verify the previous findings regarding these SNPs and association to IGF1 levels and breast cancer risk. However, this is an indication that the aforementioned association of the IGF1 haplotype 4 with mammographic density is dependent on the co-occurrence of these two SNPs.
Several associations of IGFBP3 polymorphisms and levels of IGFBP3 have been reported [25, 46, 47, 62, 63] one example is the -202(rs2854744) polymorphism associated with increased levels of IGFBP3 [46, 47, 62]. The -202 SNP has been associated with levels of IGF1 , IGFBP3 [46, 47, 60] and premenopausal mammographic density  but not with postmenopausal mammographic density [46, 47]. The present study examined SNPs in the surrounding area of the -202 polymorphism which is an area suggested to be in strong LD[25, 64].
The two IGFBP3 haplotypes analyzed here were found to be significantly associated with the levels of IGFBP3 suggesting a putative regulatory effect in cis. The C allele of SNP rs2471551 of the IGFBP3 haplotype has previously been associated with increased levels of IGFBP3 in combination with surrounding SNPs. Upon single SNP analysis we found a significant trend of SNP rs2471551 with the level of IGFBP3 indicating that having two copies of the frequent allele C increase the least square mean of IGFBP3 compared to having two copies of the rare allele G, confirming the finding of Canzian et al. 2006. In the haplotype analysis we found that the AC haplotype was associated with higher levels of IGFBP3 than the AG haplotype and can thus confirm the result published by Canzian et al. that SNP rs2471551 is associated with increased IGFBP3 levels. In addition, these haplotypes signify a trend of association with low levels (AG) and high levels (AC) of mammographic density (Table 3). These findings are consistent with the review of Fletcher et al. in which most reports agree on increased breast cancer risk with high levels of IGFBP3. This may most strongly apply to premenopausal women, in whom the IGF levels are higher than in postmenopausal, and where the IGF axis is postulated to have an increased role because of involvement of the sex hormones. The emerging belief that higher levels of IGFBP3 may decrease mammographic density and may decrease the risk of cancer[22, 25, 66] through low IGF1/IGFBP3 ratio must be further substantiated.
Even though it is known that the circulating IGF2 concentration is much higher than that of IGF1, there is limited evidence on its mitogenic activity in relation to breast cancer and disease[47, 49, 56], thus implications of IGF2 on breast cancer risk are inconclusive. Studies on genetic variants of IGF2 in relation to breast disease are few, and to our knowledge, the present study is the first to look at IGF2 polymorphisms in relation to levels of mammographic density, IGF1 and IGFBP3. The finding of a common haplotype (4, Table 3) significantly associated with higher levels of both IGF1 and IGFBP3 may be explained by a decrease in clearance of IGF1 due to potentially lower levels of IGF2. IGFBP3 is the principal carrier of both IGF1 and IGF2 and the possibility of a regulatory feedback of IGF1 and IGFBP3 through polymorphisms in IGF2 cannot be excluded. However, we did not have measurements of IGF2 and could therefore not test this. The association with haplotype 4 was no longer seen in analyses of the single IGF2 SNPs and thus there is reason to believe that the association is dependent on the combination of the two SNPs comprising this haplotype. After stratifying the analysis by HT the association was still significant for the women currently taking HT, implying that HT could be an effect modifier of this association. The sex hormones play an important role in the IGF axis and could be the reason why the association is stronger in these women.
The IGF1 receptor (IGF1R) sets off a complex cascade of signals upon binding of its ligands, IGF1 and IGF2. IGF1R functions as an anti-apoptotic agent by enhancing cell survival, and has been found expressed in most breast cancer cell lines and highly over-expressed in most malignant tissues. In a study on genetic variation and breast cancer survival, Deming et al.  found SNP rs951715 within the IGF1R gene associated with breast cancer survival in postmenopausal women, whereas SNP rs2229765 included in the present study, was not. SNP rs2229765 results in a silent mutation, and has thus far not been found associated with any epidemiological traits. In the haplotype analysis none of the IGF1R haplotypes were found significantly associated with any of the studied parameters. The single SNP analysis revealed significant association of SNP rs2229765 with both percent and absolute mammographic density, increased numbers of the G allele increased the least squares means of mammographic density. In addition the SNPs rs3743259 and rs2016347 were also significantly associated with percent and absolute mammographic density, and in both cases increased number of the most frequent allele increased the least square mean of mammographic density. Functional studies are needed to investigate if these SNPs influence the affinity to IGF1 and IGF2 increasing their growth promoting effects and possibly mammographic density.
IGF1 and IGF2 send their mitogenic and antiapoptotic signals through a common thyrosine kinase receptor, the IGF1R. Modulation of the mitogenic pathway occurs in part via the M6P/IGF2R, which functions in the internalization and degradation of IGF2. IGF2R is also important in the activation process of TGFβ, which amongst other properties has the ability to inhibit cell growth. Loss of heterozygosity (LOH) of the M6P/IGF2R has been linked to liver and breast cancers, whereas somatic mutations of the M6P/IGF2R have been found in cancers of the prostate, lung, endometrium, brain, stomach and colorectum. Chen et al. found that decreased ribosomal expression of the receptor leads to increased proliferation of MCF7 cells by a IGF2 related mechanism, mediated through IGF1R . These findings have led to the suggestion that IGF2R is a tumor suppressor gene. Our results show that two of the IGF2R haplotypes are significantly associated to decreased levels of IGF1. For haplotype 1 the association was still significant after stratification for HT, in women that are never or past users of HT. Postmenopausal women that are not under the influence of hormones potentially have lower IGF1 levels. Analysis performed on the individual SNPs supports this finding, with a significant association of two IGF2R SNPs and the levels of IGF1, in addition they are also significantly associated with the levels of IGFBP3. IGF1 which is produced in the liver is influenced by several factors such as growth hormone and insulin, and its bioavailability is regulated by IGF2, IGFBPs and Als (acid-labile protein subunit). It is well known that IGF2 can act through IGF1R, in contrast IGF1 does not act through IGF2R, and to our knowledge no association between IGF2R and IGF1 levels have been described. IGF2R is able to degrade IGF2 and thereby regulates the circulating concentration of IGF2, in turn IGF2 clearance has the ability to regulate the level of IGF1. Thus, a possible explanation for the association of the two haplotypes and the SNP within IGF2R with IGF1 levels could be a change in clearance of IGF2 levels leading to decreased production of IGF1 through a regulative feedback loop. Whether or not such an interaction is present between IGF2 and IGF1 levels is impossible to confirm, due to lacking measurements of circulating IGF2.
Despite being an important member in IGF regulation, few studies have looked at this protein and variations within it in regards to breast cancer[25, 64]. Canzian et al.  studied three SNPs within exon 2 of IGFALS in regards to breast cancer risk, two of which are included in the IGFALS haplotype of our study (rs3751893, rs17559), and found that homozygous carriers of SNP rs3751893 were associated with reduced circulating levels of IGF1. Deming et al. conducted a study on IGFALS promoter SNPs in relation to menopausal status but found no association .
The IGFALS haplotypes in this study were not significantly associated with neither the levels of IGF1, IGFBP3, their ratio nor mammographic density. Stratification by HT of the haplotype analysis revealed a significant association of the never/past HT group with the level of IGFBP3, in addition the SNP analysis revealed significant association of SNP rs9282731 with the level of IGFBP3. Increased number of the C allele increases the IGFBP3 level compared to the rare allele G. One hypothesis could be that the C allele modifies the IGFALS and reduces either its affinity or reduces its level causing increased level of free IGFBP3. Although the SNP analysis is based on low frequencies, functional studies could to be done to verify such a hypothesis. Further investigation into the role of this protein is needed to establish its involvement in the development of mammographic density and breast cancer.