Introduction

Insulin-like growth factor-I (IGF-I) and insulin-like growth factor binding protein-3 (IGFBP-3) are important growth modulators [1]. In vitro studies demonstrate that IGF-I has strong mitogenic and anti-apoptotic effects on various cancer cells, including breast cancer [2–4]. Animal experiments suggest that IGF-I may promote tumorigenesis in the mammary gland [5, 6]. Several epidemiological studies show a positive correlation between circulating levels of IGF-I and breast cancer risk [7–13]. Genetic polymorphisms in the IGF-I and IGFBP-3 genes are also evaluated for their associations with breast cancer risk. The investigation initially focused on a limited number of genetic variations located mainly in the gene promoters or coding regions [14–17]. Recently, high-throughput analyses were used to interrogate the entire genes and their surrounding regions. A large collaborative study, the Breast and Prostate Cancer Cohort Consortium (BPC3), recently reported important influences of several single nucleotide polymorphisms (SNP) in the IGF-I and IGFBP-3 genes on circulating levels of the IGF-I and IGFBP-3, although no association was found with breast cancer risk [18].

Despite a large body of evidence indicating the possible involvement of IGF-I in breast cancer, a causal link between the growth factor and breast cancer remains to be established [15, 18–24]. Furthermore, most of the human studies are conducted among Caucasians; very few have focused on other races and ethnicities. The relationship between IGF genotype and phenotype is also investigated mainly in Caucasians; data from other racial groups are scarce. To further elucidate the role of IGF-I in breast cancer, we conducted a case–control study of breast cancer in Chinese women. In the study, we analyzed the associations of breast cancer risk with four IGF-I and two IGFBP-3 SNPs which were known to have influences on their phenotypes in the circulation, and examined the correlation between the genotypes of IGF-I and IGFBP-3 and their phenotypes in breast tumor tissues.

Methods

Patients and controls

Patients who underwent surgery at Tianjin Medical University Cancer Hospital between January 2007 and December 2007 with newly diagnosed and histologically confirmed primary breast cancer were recruited for this study. Clinical information collected for the study included histology, tumor size, lymph node involvement, disease stage, and status of estrogen receptor and progesterone receptor. Genetically unrelated control subjects with frequency-match to their cases by age (±5 years) were selected from women living in the neighboring communities where the patients resided.

The study was approved by an ethical review committee at Tianjin Medical University Cancer Hospital. All of the study participants signed an informed consent and completed a structured questionnaire which elicited information on demographic features and several risk factors of breast cancer including age at diagnosis or interview, age at menarche, age at menopause, smoking status, physical activity, numbers of pregnancy and live birth, breast feeding, body mass index, use of oral contraceptives, a history of benign breast diseases, and a family history of breast cancer in the first-degree relatives. Individuals who smoked at least one cigarette per day for half a year were considered smokers. For physical activity, less than one time or 1 h per week of any moderate--to-vigorous-intensity aerobic activity in the past 6 months was defined as “never”; otherwise it was “ever.” A blood sample (10 ml) was collected from each study subject, and buffy coats were isolated from the blood for DNA extraction and genotyping. Tumor samples from the patients were also obtained from a tissue bank at Tianjin Medical University Cancer Hospital, which routinely collects and stores tumor specimens from the operated patients upon approval by the Institution Review Board at Tianjin Medical University Cancer Hospital.

Analysis of IGF genotypes

Genomic DNA was extracted from buffy coats using the QIAGEN DNA Blood Mini Kit (QIAGEN Inc., Valencia, CA), and the extraction was performed according to the manufacturer’s instructions. SNPs with possible influences on their phenotypes in the circulation reported by a recent GWAS [18] were selected for investigation, which included four in the IGF-I gene (rs1520220, rs2195239, rs2946834, and rs7965399) and two in the IGFBP-3 gene (rs2854746 and rs2960436).

All of the chosen SNPs were genotyped with the TaqMan assays which were purchased from ABI (Applied Biosystems, Foster City, CA). In each genotyping assay, 10 ng genomic DNA was mixed with a PCR cocktail that contained two fluorescence-labeled allele specific probes (200 nM each), forward and reverse primers (1 μM each), and 5 μl of 2 × TaqMan Universal PCR Master Mix including ROX passive dye as internal control, dNTPs, PCR buffer, and Taq polymerase (Applied Biosystems, Foster City, CA) in a final volume of 10 μl. PCR amplification was achieved under the condition of one cycle of denaturing at 95°C for 10 min and 45 cycles of denaturing at 92°C for 15 s and annealing and elongation at 60°C for 1 min, followed by storage at 4°C. The fluorescence of PCR products was then plotted, and the genotypes were determined according to the allelic discrimination software. Water control and previously genotyped samples were included in each plate to monitor genotyping quality. Five percent of the samples were randomly selected for retesting, and the results of repeated tests were in complete concordance. All TaqMan assays were performed with the ABI 7500 thermal cycler (Applied Biosystems, Foster City, CA).

Analysis of IGF phenotypes

Freshly frozen tumor specimens were processed to extract tissue proteins for analysis of IGF-I and IGFBP-3. Tumor tissues were first pulverized in liquid nitrogen and the tissue powders (~100 mg) were mixed with 1 ml BD TALON × Tractor buffer (Clontech, Palo Alto, CA) which was centrifuged at 14,000 rpm for 30 min at 4°C. After centrifugation, the supernatants were collected for measurement of total proteins, IGF-I and IGFBP-3. Concentrations of total proteins were determined using the bicinchoninic acid method (Pierce Inc., Rockford, IL). Levels of IGF-I and IGFBP-3 peptides were measured with enzyme-linked immunosorbent assays (ELISA) (Diagnostic Laboratories Systems, Webster, TX) following the manufacturer’s protocols, respectively. The tumor samples were tested in duplicate, and the test results with CV greater than 10% were repeated. IGF peptide concentrations in tumor samples were adjusted for total proteins.

Statistical analysis

Chi-square test was used to compare the distributions of demographic variables, known breast cancer risk factors, and IGF genotypes between the cases and controls. Analysis of the Hardy–Weinberg equilibrium for each SNP was also performed among the control subjects. Unconditional logistic regression models were developed to examine the associations between SNPs and breast cancer risk. In the logistic regression analysis, odds ratios (ORs) and their 95% confidence intervals (95% CI) were calculated without and with adjustment for other risk factors and confounding variables, including age at diagnosis or interview, number of pregnancy, a history of benign breast diseases, and a family history of breast cancer. The associations between the IGF genotypes and breast cancer risk were also analyzed after the study subjects were stratified by menopausal status, age at menopause, BMI, family history of breast cancer in the first-degree of relatives, age at menarche, use of oral contraceptives, as well as clinical and pathological features of breast cancer, including histology, tumor size, lymph node involvement, disease stage, and ER and PR statuses. Based on the molecular weight of IGF-I and IGFBP-3, a molar ratio between IGF-I and IGFBP-3 was calculated. Wilcoxon rank-sum test was used to compare levels of IGF-I and IGFBP-3 among different IGF genotypes. SAS (SAS, Cary, and NC) was used for all the analyses. In addition, pairwise linkage disequilibrium (LD) values were calculated using the expectation–maximization (EM) algorithm. Haplotype frequencies were estimated using the SAS Haplotype program and SHESIS software (http://analysis.bio-x.cn) [25].

Results

Patient characteristics

Baseline characteristics of the patients and controls are presented in Table 1. As expected, the patients reported higher percentages of known breast cancer risk factors than the controls, including fewer numbers of pregnancies and less physical activity. Breast cancer patients were also more likely than the controls to have a history of benign breast diseases or a family history of breast cancer in the first-degree relatives. Clinical and pathological features of breast cancer are also shown in Table 1. Three hundred and twenty patients (80%) had infiltrating ductal carcinoma (IDC), and 60 patients (15.4%) were diagnosed with stage I disease. Using 10% as cutoff for hormone receptor status, 213 tumors (53.5%) were estrogen receptor (ER) positive, and 234 (58.8%) were progesterone receptor (PR) positive.

Table 1 Distributions of risk factors between breast cancer cases and control subjects
Full size table

IGF genotypes and breast cancer risk

Four SNPs in the IGF-I genes and two in the IGFBP-3 gene were examined in the study. All the SNPs investigated in the study were in Hardy–Weinberg equilibrium among the control subjects. Table 2 shows the genotype frequency of each polymorphism in the cases and controls, along with the corresponding ORs after adjusting for age, frequency of pregnancy, a history of benign breast diseases, and a family history of breast cancer. Three genetic models, including co-dominant, dominant, and recessive, were utilized to calculate the ORs. Of the four SNPs examined in the IGF-I gene, only one (rs7965399) was shown to have a significant association with breast cancer in a recessive model. Compared to those with the ‘T’ allele (wild genotype), either one or two, women with homozygous ‘CC’ genotype (variant genotype) had increased risk of breast cancer (OR = 1.86; 95% CI: 1.04–3.32). Subgroup analyses stratified by age at menopause, family history of breast cancer, histology, or ER status were also performed for this SNP. In the stratified analyses, we found that the risk association with rs7965399 was more evident among those who had menopause at younger ages. This polymorphism also showed stronger associations in patients with IDC (OR = 2.18; 95% CI: 1.08–4.38) and ER-negative tumors (OR = 2.48; 95% CI: 1.08–5.67), but not in those with ER-positive tumors (Table 3). None of the IGFBP-3 polymorphisms were found to be associated with breast cancer risk.

Table 2 Associations of breast cancer risk with IGF-I and IGFBP-3 polymorphisms
Full size table
Table 3 Associations of breast cancer risk with IGF-I (rs7965399) polymorphism stratified by selected variables
Full size table

Next, we evaluated the pairwise LD and haplotype distribution among these SNPs. The results indicated that three of the four IGF-I SNPs (rs1520220, rs2946834, and rs2195239) as well as two of the IGFBP-3 SNPs were in LD, and the lowest D′ was greater than 0.90; r 2 greater than 0.72 (data not shown). The most common haplotypes in the IGF-I gene were “CCG” and “GTC,” which accounted for 91.2% of the cases and 92.9% of the controls; the most common haplotypes in the IGFBP-3 gene were “GA” and “CG,” which accounted for 98.9% of the cases and 98.4% of the controls. The distributions of the common haplotypes were not significantly different between the cases and controls, either in the IGF-I or IGFBP-3 gene (P = 0.290 and 0.445, respectively).

IGF genotype and phenotype correlation

To assess the genotype and phenotype correlation, protein concentrations of IGF-I and IGFBP-3 in breast tumor samples were analyzed with ELISAs. Our analyses suggested that peptide levels of IGF-I were inversely correlated with age and menopause status (data not shown). The results also indicated that tumor levels of IGF-I were higher in women with variant alleles of IGF-I SNPs than those with the homozygous wild genotypes. For IGFBP-3 SNPs, however, the homozygous variant genotypes had lower IGF-I compared to their wild counterparts. The differences in phenotype, either in IGF-I alone or the molar ratio of IGF-I to IGFBP-3, were statistically significant or borderline significant in two IGF-I intron SNPs (rs1520220 and rs2195239) and two IGFBP-3 SNPs (rs2854746 and rs2960436) (Table 4). No significant correlation was found between IGFBP-3 genotype and phenotype in the tissue (data not shown).

Table 4 Associations of IGF-I and IGFBP3 genotypes and phenotypes in breast tumors
Full size table

Discussion

In this study, we selected four IGF-I and two IGFBP-3 SNPs, which are known to be associated with their phenotypes in the circulation among Caucasians, to investigate their associations with breast cancer risk among Chinese women. Table 5 shows the allele frequencies of the six selected SNPs by ancestry recorded by the NCBI dbSNP database as well as research findings reported by recent studies of these SNPs in association with cancer risk. The allele frequencies of these SNPs in our study were similar to those of Asians, but different from those of Europeans in the database. The IGF-I SNP rs7965399 seems to be quite different between Asians and Caucasians. The frequency of the variant allele is quite high among Asians, but virtually none in Caucasians. This racial difference may indicate the possibility that an association between the SNP rs7965399 and a disease differs by race. This possibility seems to exist when we compare our finding with other studies on SNP rs7965399. IGF-I rs7965399 was found to be associated with breast cancer risk among Chinese women in our study, and the association seemed to be more evident in women with early age at menopause, or having ER negative tumors, but the association was not seen in Caucasian women. However, other SNPs investigated showed no association with breast cancer risk, which was consistent with the finding in Caucasian women. Interestingly, most of the SNPs investigated in this study also showed a possible impact on IGF-I phenotype in breast tumors. These results seemed to be consistent with the finding of Caucasian women when blood samples were used for evaluation of phenotype.

Table 5 Allele frequencies of IGF-1, IGFBP3 genetic polymorphisms, and the studies of these SNPs in association with cancer risk
Full size table

IGF-I rs7965399 was the only SNP that was found to be associated with breast cancer risk in our study. This association, however, was shown only in a recessive model; no relationship was indicated in other genetic models. Besides our study, there was another investigation which found the same SNP to be associated with prostate cancer risk [26]. Both that and our studies indicated that the individuals with variant rs7965399 had elevated cancer risk. The SNP is located in the 5′-noncoding region of the IGF-I gene, which is near the transcription site. Based on the location, it is plausible that SNP rs7965399 may have a potential influence on the regulation of IGF-I activity through its impact on IGF-I transcription. In addition, it is also possible that this polymorphism may be associated with breast cancer risk through its LD with a functional SNP in the IGF-I gene. Further evaluation of the functional relevance of IGF-I rs7965399 appears to be warranted.

Other IGF-I SNPs investigated in this study (rs1520220, rs2946834, and rs2195239) are located in different regions of the IGF-I gene, including introns and 3’ UTR. So far, only one study has reported an association between rs1520220 and breast cancer risk [20]. Although two recent studies indicated possible associations of rs1520220 and rs2946834 with breast mammographic density, which could be indirectly linked to breast cancer risk [27, 28], most genotype studies failed to demonstrate any associations between genetic polymorphisms in the IGF-I gene and breast cancer risk [18, 21, 22]. For the polymorphisms in IGFBP-3, one SNP (rs2854744, A-202C) in the promoter region has been studied extensively, but its association with cancer risk remains to be inconclusive. Other IGFBP-3 SNPs have not been found to be associated with breast cancer risk [13, 16, 22, 28–30]. A large consortium study of 6,912 breast cancer cases and 8,891 matched controls also found no associations between breast cancer risk and SNPs in the IGF-I and IGFBP-3 genes among Caucasian women [18]. Moreover, a recent meta-analysis of 96 studies concluded no apparent influences of several well-studied polymorphisms in the IGF-I and IGFBP-3 genes on breast cancer risk [13]. Similar to these results, we found no associations between breast cancer risk and the IGF-I and IGFBP-3 SNPs among Chinese women, either as a single SNP or in their haplotypes.

Numerous studies have shown that high IGF-I and low IGFBP-3 in the circulation are associated with increased risk of breast cancer, but these findings are challenged by the speculation that circulating IGF-I and IGFBP-3 may differ from those in local tissues either in quantity or quality. IGF-I and IGFBP-3 in breast tissues are derived from both local expression and systemic regulation [31]. Recent studies indicated that IGF-I polymorphisms were associated with breast mammographic density, especially the SNP rs1520220 which was reported by two independent studies to be strongly associated with breast mammographic density [27, 28]. This SNP is located in the third intron with unclear biological implications. Our study showed that this SNP was associated with the phenotype of IGF-I in breast tissue. Another intron SNP rs2195239, which was in LD with this one, was also shown to be significantly associated with the IGF phenotype in breast tissue. Moreover, two IGFBP-3 polymorphisms investigated were correlated with the phenotype of IGF-I, both in IGF-I peptide levels and the molar ratio of IGF-I to IGFBP-3. The latter reflects the bioavailability of IGF-I in local tissues. The correlation between IGFBP-3 genotype and local phenotype, however, was not observed in our study. Our finding seems to be different from the results of two previous studies which analyzed IGFBP-3 phenotype in the circulation [18, 28]. Two clinical studies found that IGFBP-3 polymorphisms or IGFBP-3 mRNA expression, but not its protein concentrations, had some influences on breast cancer survival [31, 32].

IGFBP-3 binds to approximately 90% of IGF-I in the circulation. This binding inhibits IGF-I interacting with its receptor IGF-IR, and reduces the IGF signal or activity [22]. Based on these understandings, we speculate that our findings of correlations between IGFBP-3 genotypes and IGF-I phenotypes in local tissues may be mediated through the effects of IGFBP-3 genotypes on its phenotype in the circulation. Our results also indicate that SNPs in the IGF-I and IGFBP-3 genes may affect levels of IGF-I expression not only in the circulation but also in local tissues, which provides new insights into the relationship of IGF genotype and phenotype.

To our knowledge, this is the first study that evaluates the relationship between IGF genotype and phenotype in breast tumors among Chinese women. Our finding suggests that genetic polymorphisms in the IGF-I and IGFBP-3 genes may affect IGF-I expression in local tissues. Although there were possible influences of IGF-I genotypes on its phenotype in breast tumors, we found no strong evidence that genetic polymorphisms in the IGF-I and IGFBP-3 genes were associated with breast cancer risk. One IGF-I SNP rs7965399, however, was shown to be associated with breast cancer risk, and this association seemed to be more evident in women with early menopause or having ER-negative tumors.