Insulin-Like Growth Factor (IGF)-I and IGF-II Serum Concentrations in Patients with Benign and Malignant Breast Lesions: Free IGF-II Is Correlated with Breast Cancer Size

The insulin-like growth factors (IGFs) I and II are among the most potent stimulators of cell proliferation in humans (1). Expressed in most organs, they are systemic hormones, as well as locally acting growth factors with paracrine and autocrine functions (2). Their bioavailability is regulated by a family of at least six IGF binding proteins (IGFBPs), which can bind and inactivate IGFs but also potentiate their activity under specific conditions (3, 4). In serum, the majority of systemic IGF-I and IGF-II is bound to IGFBP3 in a ternary complex with an acid-labile subunit, which serves as a physiological transporter and restricts the bioavailability of IGFs (5). Proteolytic cleavage of IGFBPs by cathepsins and matrix metalloproteinases results in decreased affinity and releases IGFs out of the binding complex (6). There is now substantial evidence that IGFs are involved in malignant transformation and tumor cell growth. The association between serum IGF-I and breast cancer has been postulated for some time. Recent control-matched epidemiological studies have demonstrated that high IGF-I and low IGFBP3 levels are associated with an increased risk of breast cancer in premenopausal women (7, 8). Both IGF-I and IGF-II are released by tumor fibroblasts and by some breast cancer cell lines, and there seems to be a shift from fibroblastic IGF-I to IGF-II expression in the immediate vicinity of invasive tumor cells as compared with surrounding breast stroma (9, 10). Furthermore, in vitro and in vivo studies have suggested that the overexpression of IGF-II in MCF-7 tumor cells results in the conversion to a more malignant phenotype and an increased efficiency of tumor formation in nude mice (11). Conversely, tumor growth in IGF-I-deficient mice is reduced relative to control animals (12). Several studies have demonstrated elevated IGF-I and reduced IGFBP3 tissue and plasma concentrations in breast cancer patients, and some have even suggested their potential use as prognostic factors (13, 14, 15).

Although a lot of clinical data regarding the role of IGF-I and IGFBPs in human malignancies have emerged over the last years, little is still known about the role of IGF-II in malignant breast tumors. We have therefore investigated IGF-I and IGF-II serum concentrations in women with benign and malignant breast lesions and compared them to age-matched healthy controls. Because in human serum > 95% of IGFs in the circulation are bound to IGFBPs and thus biologically inactive, we measured free IGF-I and IGF-II concentrations, as well as total (= protein bound + free) IGF concentrations after acid-ethanol extraction from binding proteins. The ratio-free IGF/total IGF was used to describe the fraction of bioavailable IGF.

Serum samples were collected from 65 women undergoing surgery for suspicious breast lesions and from 38 women without evidence of benign or malignant breast disease (Table 1). None of the patients had undergone cancer treatment previously. All patients had given written informed consent before blood sample collection. The use of the collected material for serum analyses had been approved by the Institutional Review Board of the Medical School of the University of Vienna. The control group consisted of age-matched women who were either consenting healthy staff nurses or women undergoing routine preoperative blood analysis for minor surgery for nongynecological reasons (Table 1). All patients and control subjects were of Caucasian race. None of the women in the control group had a history of breast cancer or any other breast disease. Controls were within a range of 1–2 years of age with the breast cancer group. To avoid any significant influence of the nutritional status on the pattern of circulating IGF levels, all samples were obtained between 9 a.m. and 11 a.m. and in fasting condition on the day of admission to the hospital. After 2 h of incubation at room temperature for clotting, serum was separated by centrifugation at 2000 × g for 30 min and frozen at −80°C until analysis. The breast cancer group (n = 42) was comprised of ductal (n = 34), lobular (n = 7), and tubular (n = 1) carcinomas, and tumor sizes ranged from pT1 (n = 22), pT2 (n = 13), and pT3 (n = 2) to pT4 (n = 5). Benign breast lesions (n = 23) were either fibroadenomas (n = 9) or fibrous cystic mastopathy with or without calcifications (n = 14).

Total IGF-I and free IGF-I were quantified by ELISA with reagents from Diagnostic Systems Laboratory (Webster, TX). This method is considered to be more reproducible than the RIAs used previously (16). For total IGF-I measurements, serum was subjected to acid-ethanol extraction with 12.5% 2 n HCl and 87.5% ethanol before the assay to release IGFBP-bound IGF-I. IGF-I was then measured in an enzymatically amplified two-step sandwich-type immunoassay. In the assay, 50 μl of standards, controls, and 1:96 diluted samples were incubated with anti-IGF-I-coated microtiter wells at 25°C for 2 h. After subsequent washing, the microtiter wells were treated with 100 μl of a horseradish peroxidase-conjugated anti-IGF-I antibody solution. After an additional 30 min of incubation and subsequent washing step, the wells were incubated with tetramethylbenzidine chromogen for 10 min. The enzymatic reaction was stopped by the addition of 100 μl of 0.2 m sulfuric acid, and the extent of enzymatic conversion of tetramethylbenzidine to its colored metabolite was determined by a dual wavelength absorbance measurement at 450 and 630 nm.

Free IGF-II and total IGF-II serum concentrations were determined by using the Active-free IGF2 ELISA and the Active IGF2 ELISA from Diagnostic Systems Laboratory. After a preassay acid-ethanol extraction step to liberate protein-bound IGF-II in the Active IGF2 ELISA, IGF-II was measured in an enzymatically amplified two-step sandwich-type immunoassay as described for IGF-I.

All statistical analyses were performed with the SPSS 10.0.1 software (SPSS, Inc., Chicago, IL). Possible differences in total and free IGF concentrations between populations were analyzed using the Mann-Whitney U test. Correlations between tumor diameter and IGF serum concentrations were evaluated using the Spearman rank correlation analysis. Multiple linear regression analysis was performed to analyze the influence of single parameters on tumor size after neutralization of confounding variables. Because logistic regression analysis relies less on distributional assumptions than linear regression analysis, correlations between the ratio variables free IGF-I/total IGF-I and free IGF-II/total IGF-II and tumor size were investigated by a multiple logistic regression model.

Serum concentrations of total IGF-I, free IGF-I, total IGF-II, and free IGF-II were measured by sandwich-type immunoassays in 38 healthy women, in 42 women suffering from breast cancer, and in 23 women with benign breast disease. Results of both free (i.e., unbound fraction) and total (i.e., unbound plus protein-bound/inactive fraction) IGF concentrations, as well as the ratios free IGF/total IGF, are depicted in Fig. 1.

We then performed Mann-Whitney U test analysis to detect potential differences in IGF serum concentrations between healthy controls and women with breast tumors (Table 2). Total IGF-I serum levels were highly significantly lower in healthy women (median, 160.7 ng/ml) than in women with breast tumors (median, 221.6 ng/ml; P < 0.001), whereas free IGF-I levels did not differ significantly between the two groups (median, 5.3 and 4.9 ng/ml, respectively; P = 0.838). Levels of total IGF-II did not differ significantly in both groups with median values of 482.8 and 539.4 ng/ml (P = 0.14), but free IGF-II concentrations differed very significantly between healthy controls (median, 15.9 ng/ml) and tumor patients (median, 12.9 ng/ml; P = 0.003). Furthermore, lower ratios of both free IGF-I/total IGF-I and free IGF-II/total IGF-II were highly significantly associated with a breast tumor (P = 0.003 and P = 0.001, respectively).

When tumor patients were subgrouped into women with benign and women with malignant breast neoplasms, total IGF-I serum levels were highly significantly lower in healthy women (median, 160.7) than in women suffering from benign (median, 218.0; P = 0.010) or malignant (median, 226.9; P < 0.001) breast tumors (Table 2). However, free IGF-I concentrations were comparable in all three groups (median values of 5.2 ng/ml in benign and 4.9 ng/ml in malignant tumors and 5.3 ng/ml in healthy controls). An inverse pattern was seen for IGF-II, where healthy women had higher serum levels of free IGF-II (median, 15.9 ng/ml) than women with benign (median, 11.4 ng/ml; P = 0.016) and malignant breast tumors (median, 12.9 ng/ml; P = 0.010), albeit at a lower level of significance. Total IGF-II serum concentrations did not differ significantly between the three groups.

Consequently, when compared with age-matched healthy individuals, lower ratios of free IGF-I/total IGF-I and of free IGF-II/total IGF-II were both characteristic of malignant disease (free IGF-I/total IGF-I: P = 0.007; free IGF-II/IGF-II: P = 0.009) and of benign breast tumors (free IGF-I/IGF-I: P = 0.019; free IGF-II/IGF-II: P = 0.002). No significant differences in any of the IGF serum levels analyzed were found when samples of patients with malignant and benign breast tumors were compared (data not shown). Furthermore, a comparison between IGF concentrations of women who either had no breast tumor at all or who had benign breast disease (nonmalignant) and of women with invasive breast cancer (malignant) yielded no significant difference with the exception of total IGF-I (226.9 versus 175.9 ng/ml; P = 0.011).

In Table 3, Spearman’s analysis was used to investigate possible correlations between free and total IGF-I or IGF-II to the size of 41 malignant tumors as assessed by histopathological analysis. One patient with a pT1 tumor was omitted from this analysis because the exact tumor size was not known. Although the parameters free IGF-I, free IGF-I/total IGF-I, and total IGF-II were not correlated to tumor diameter, total IGF-I was found to be negatively correlated to breast cancer size (r = −0.317, P = 0.044). Interestingly, free IGF-II was positively correlated to tumor size (r = 0.350, P = 0.025). The ratio-free IGF-II/total IGF-II was even very significantly correlated to tumor size (r = 0.469, P = 0.002).

We then used multiple linear regression analysis to investigate the effect of the potential predictors total IGF-I, free IGF-I, total IGF-II, free IGF-II, and age on the tumor diameter (Table 4). Within the breast cancer population, total IGF-II concentrations at the percentile 25 (= 400.79 ng/ml) were found to be associated with a tumor diameter that was 5.89 mm smaller than that associated with concentrations at the percentile 75 (= 645.93 ng/ml). Conversely, a raise of free IGF-II from percentile 25 (= 10.96 ng/ml) to percentile 75 (= 15.26 ng/ml) would be associated with an increase of tumor size by 5.36 mm. Furthermore, the influences of total and free IGF-II were independent of age, which also provided a significant increase of 0.5 mm for each additional year of age (P < 0.001).

The ratios free IGF-I/total IGF-I and free IGF-II/total IGF-II and age as an additional confounder were tested by the multiple logistic regression model for their potential influence on tumor size. Because some surgeons consider 20 mm to be the critical tumor size for performing a mastectomy rather than a simple lumpectomy, we used multiple logistic regression analysis to investigate whether women with IGF ratios ranked at the percentile 75 within our breast cancer population would have a higher risk of harboring a tumor > 20 mm than women with serum concentrations at the percentile 25. Whereas in our model patient age and the free IGF-I/total IGF-I ratio had no significant impact on whether a tumor was >20 mm, the free IGF-II/total IGF-II ratio was positively correlated with tumor size. A free IGF-II/total IGF-II ratio at the percentile 75 (= 0.033) of serum concentrations would indicate a 3.2 times higher chance of a tumor size of >20 mm than a ratio of 0.021 (= percentile 25; P = 0.025; data not shown).

IGF-I and IGF-II are among the most potent growth stimulatory factors known to date, and both growth factors have been associated with malignant breast disease (17). Their biological activity is regulated by a number of IGFBPs, which are believed to bind and inactivate IGFs but which can also act as modulators of tumorigenesis through IGF-independent mechanisms (18). These IGFBPs are thought to exert both positive and negative regulatory effects on IGF activity. On the one hand, binding of IGFs to their binding proteins protects them from degradation, thus increasing the total amount of IGFs in the circulation; on the other hand, IGFBP-bound IGFs cannot simultaneously bind to their cognate receptor, which constitutes the major negative regulatory role of IGFBPs on the bioavailability and biological activity of IGF-I and IGF-II (19). Furthermore, free IGF-I was demonstrated to cross the endothelial cell barrier of blood vessels more readily than IGFBP-bound IGF-I (20). These complex, both positive and negative, regulatory effects of IGFBPs on the activity of IGF-I and IGF-II are reflected by the complex relationship between serum levels of IGFBP3, the major IGFBP in the blood, and breast cancer risk. Although most studies have found a correlation between low serum levels of IGFBP3 and increased breast cancer risk, others have found the opposite (7, 19, 21, 22). Presumably, a key role in development of breast cancer is linked not only to the absolute levels of IGFBPs but rather to an imbalance in the concentration of IGFs and their binding proteins (21).

Recently, there has been a lot of interest in the measurement of free IGFs in the plasma, which are considered to be the biologically available and thus active fraction. Although the existence of a truly free IGF plasma compartment is controversially discussed, pharmacokinetic studies indicate that a small percentage of IGFs is not associated with IGFBPs (23). It has been demonstrated that exogenously administered IGFs almost immediately associate with low molecular weight IGFBPs, then quickly move into a high molecular weight ternary complex together with a liver-derived glycoprotein known as the acid-labile subunit. Ternary complexes do not appear to be easily dissociated and do not re-equilibrate with exogenously added IGF or IGFBP3 to a significant degree. Several investigators have therefore assayed specific IGFBPs to assess the protein-bound and presumably biologically inactive IGF fraction. Because IGFBP3 is considered the principal binding protein in the circulation, it was most commonly used as a surrogate parameter for protein-bound IGFs (24). Women with relatively high levels of total IGF and comparably low levels of IGFBP3 were found to be at higher risk for breast cancer than women with the same levels of IGF-I but unselected for IGFBP3 (7). Although this correction for IGFBP3 did not allow to directly quantify the levels of free IGF-I, it is reasonable to assume that woman with low serum levels of IGFBP3 have a higher average fraction of free IGF-I, indicating that not only the levels of total IGF but also those of free IGF play a crucial role in breast cancer risk. However, in human serum, IGFBP3 accounts for only ∼75% of protein-bound IGFs, and recently, it has been reported that IGFBP5 can form ternary circulating complexes with IGF-I and acid-labile subunit as well (25). Furthermore, a family of IGF-specific low-affinity IGFBP-related proteins have been identified (IGFBP-related proteins 1–4). Its members possess the capacity to act on target cells independently of IGFs and can bind IGF, although to a much lesser extent than the classical IGFBPs (26).

To further address this questions, we aimed at a direct quantification of the levels of both total and free IGF-I and IGF-II in a side-by-side comparison. Accordingly, we have used a modified acid-ethanol extraction/precipitation method to liberate all protein-bound IGF for measurement of the total amount of IGF-I and IGF-II in serum samples. This method has an excellent correlation with acid-column chromatography, which is considered the gold standard for separating IGFs from their respective binding proteins (27). The free IGF/total IGF ratio was then used to describe the nonprotein-bound IGF fraction of the total (i.e., nonprotein-bound and protein-bound) IGF levels. Using this technique, we found total IGF-I levels to be elevated in both patients with benign and malignant breast lesions when compared with age-matched healthy controls (P < 0.001, Fig. 1 and Table 2). These findings are consistent with previous studies of elevated IGF-I levels in patients with breast cancer. Unfortunately, these studies did not include benign breast lesions (19, 21, 22, 28). Indeed, a comparison of both total and free IGF-I levels in benign and malignant lesions did not yield any statistically significant difference, demonstrating that malignant tumors cannot be identified by systemic IGF-I concentrations. Although our observations appear to be somewhat contrasting to epidemiological studies that found serum IGF-I to be correlated with an increased risk for breast cancer, they are well in line with findings by Ng et al. (29), who did not find significant differences in IGF-I serum levels between actual breast cancer patients and women suffering from benign breast lesions in a case-control study, and with findings by Holdaway et al. (30), who also did not observe differences in serum IGF-I levels between healthy individuals and women having benign or malignant breast lesions in RIAs.

Even more interestingly, no difference was found in free and presumably bioactive IGF-I serum concentrations between healthy women and patients with benign or malignant breast disease (Fig. 1 and Table 2). This result appears somewhat counterintuitive if one assumes that free IGF-I is the biologically active form; however, other studies have also shown that total IGF-I was more strongly associated with breast cancer risk than free IGF-I, in agreement with evidence that high rather than low levels of IGFBP3 are associated with breast cancer (21, 22). Muti et al. (21) proposed that there might be a threshold effect and that free IGF-I may exert a permissive effect toward breast cancer development only if its levels are high enough to exceed the rapid degradation process to which free but not IGFBP-bound IGF-I is subjected in blood serum.

Total IGF-II did not appear to be associated with benign or malignant breast lesions, in agreement with previous studies (19, 22). However, we found that free IGF-II serum levels were significantly lower in breast tumor patients than in healthy controls (Fig. 1 and Table 2). Consequently, the free IGF-II/total IGF-II ratio was decreased in women with breast lesions (median values of 0.030 versus 0.025, P < 0.001; Table 2). It appears paradoxical that in breast tumor patients serum concentrations of a mitogenic growth factor are lower than in healthy women; however, two possibilities can be discussed: (a) breast lesions contain larger numbers of IGF-II/mannose 6 phosphate receptor than normal tissues and can thus bind and internalize more free IGF-II, thereby lowering free IGF-II serum concentrations (31). However, as malignant tumors grow and become more aggressive, they usually spread into local lymph nodes—an advanced state that has already been associated with a decreased ability to produce IGF-II/mannose 6 phosphate receptor (32). A decrease of IGF-II binding sites in larger tumors might well serve as an explanation why, despite an overall reduction of total IGF-II, levels of free IGF-II actually increase significantly (Table 4). (b) Alternatively, tumor cells or their stromal environment could respond to tumor-derived cytokines with an overcompensating expression of IGF-II-neutralizing proteins such as IGFBPs or soluble IGF-II receptors (33, 34). Considering that total IGF-II levels are indeed similar in women with and without breast tumors (Table 2), increased binding of IGF-II by IGFBPs or soluble IGF-II receptors could decrease the fraction of free IGF-II.

The observation that in breast cancer patients tumor size is significantly correlated to free IGF-II levels appears contradictory to the increased levels of free IGF-II in healthy women (Table 2). It is, however, in analogy to findings by Helle et al. (35, 36), who detected increased IGFBP3 proteolysis most frequently in patients harboring large tumors and metastatic disease and who saw an inverse correlation between serum levels of IGF-I/II and IGFBP3 proteolysis. An increased tumor diameter would thus result in an elevation of the fraction of non-IGFBP3-bound IGFs and could eventually cause the elevated serum levels of free IGF-II that we observed.

It also fits well our previous findings that breast cancer cells can influence the IGF expression pattern in neighboring stromal fibroblasts: fibroblasts derived from a malignant tumor express mostly IGF-II mRNA and protein, whereas fibroblasts from adjacent normal and unaffected tissue of the same patients tend to express more often IGF-I than IGF-II mRNA. Furthermore, in coculture with MCF-7 breast cancer cells, breast fibroblasts increased IGF-II mRNA expression most likely through the clonal selection of IGF-II-expressing subpopulations (10). The intratumoral shift from IGF-I to IGF-II could thus find its reflection in systemic IGF levels and help to explain why free IGF-II concentrations and the free IGF-II/total IGF-II ratio are directly and total IGF-1 concentrations are indirectly correlated to the size of a malignant breast tumor (Table 3). The relevance of our findings in vivo is supported by a recently published article by Holdaway et al. (37), who found that in breast cancer patients, tumorectomy leads to a transient but significant fall in IGF-II but not in IGF-I and IGFBP3 serum concentrations. Even more interestingly, the fall in IGF-II is significantly related to the size of the removed tumor, thus indicating a significant intratumoral IGF-II production (37).

Taken together, we have shown for the first time that in breast cancer patients, free IGF-II serum concentrations and the free IGF-II/total IGF-II ratio are well correlated with tumor size, thus suggesting a role of this growth factor in the growth control of malignant tumors. Furthermore, we have demonstrated that total IGF-I serum concentrations are increased and that free IGF-II serum concentrations are decreased in women with breast tumors. However, benign and malignant tumors cannot be distinguished by their systemic IGF concentrations, in disagreement with previous studies, which have implied that IGF-I expression is somewhat specific for malignant tumor growth (13, 28).

We thank Ernst Ruecklinger for help with statistical analysis.