Abstract

MATERIALS AND METHODS

RESULTS

The median time between blood sampling and diagnosis was 13 years for both pre- and postmenopausal women; no cases were diagnosed within 3 years of recruitment. For women who were premenopausal at recruitment, the mean age at diagnosis was 57 years; we do not have information on menopausal status at diagnosis, but the majority of these cases would have been postmenopausal by the time breast cancer was diagnosed. The mean age at diagnosis among women who were postmenopausal at recruitment was 72 years.

Selected baseline characteristics of cases and controls are presented in Table 1. There were no significant differences in the median age at recruitment, age at menarche, age at first birth, age at menopause, height, weight or BMI between cases and controls among pre- or postmenopausal women. The proportions of women who were nulliparous, who had ever breast-fed, who had ever taken exogenous hormones or who were current smokers were also similar between cases and controls. The median concentrations of IGF-I, IGF-II and IGFBP-3 did not differ significantly between cases and controls in either pre- or postmenopausal women.

Table 1

Baseline characteristics of breast cancer cases and controls by menopausal status^a

	Premenopausal women			Postmenopausal women
	Cases (n=70)	Controls (n=209)	P-valueb	Cases (n=47)	Controls (n=141)	P-valueb
Age at recruitment (years)	41 (37–44)	41 (37–44)	0.83	59 (55–62)	58 (55–63)	0.98
Age at menarche (years)	13 (12–14)	13 (12–14)	0.35	14 (12–15)	13 (12–14)	0.49
Age at first birthc (years)	23 (21–26)	24 (22–27)	0.62	26 (22–29)	25 (23–28)	0.58
Age at menopause (years)	N/A	N/A		50 (47–52)	50 (48–53)	0.24
Height (cm)	162 (157–165)	161 (157–165)	0.95	158 (155–164)	160 (155–163)	0.52
Weight (kg)	63 (59–69)	62 (56–68)	0.20	62 (58–70)	63 (58–69)	0.96
BMI (kgm−2)	24 (23–27)	24 (22–26)	0.16	25 (23–28)	25 (23–27)	0.54
Parity
Nulliparous	6 (9%)	25 (12%)		12 (26%)	20 (14%)
1	8 (11%)	28 (13%)		4 (9%)	28 (20%)
2	35 (50%)	83 (40%)		14 (30%)	36 (26%)
3+	21(30%)	73 (35%)	0.50	17 (35%)	57 (40%)	0.13
Past OC use
Yes	54 (77%)	152 (73%)	0.47	9 (19%)	27 (19%)	1.00
Past HRT use d
Yes	N/A	N/A	N/A	11 (24%)	33 (24%)	0.98
Serum peptides
IGF-I (nmoll−1)e	22.4 (18.6–26.9)	22.2 (18.5–26.5)	0.88	16.4 (12.8–23.3)	16.8 (13.5–21.2)	0.86
IGF-II (nmoll−1)e	124 (107–144)	129 (114–148)	0.27	125 (111–164)	131 (116–151)	0.63
IGFBP-3 (nmoll−1)	170 (141–193)	172 (155–194)	0.10	167 (139–194)	164 (142–194)	0.93

Open in a separate window

OC=oral contraceptive; HRT=hormone replacement therapy; IGF=insulin-like growth factor; IGFBP=IGF-binding proteins; N/A=not applicable.

^aValues are medians (interquartile range) or numbers (percent).

^bP-value based on Kruskal–Wallis rank sum test for continuous variables and χ² test for categorical variables.

^cAmong parous women only.

^dData were missing for three postmenopausal women.

^eSerum IGF-I and IGF-II measurements were unavailable for one premenopausal case.

The correlations of the three peptide measures with each other and with other factors were examined in controls. Serum IGF-I was positively correlated with serum IGF-II (age-adjusted Spearman's rank correlation coefficient r=0.32 in premenopausal women and r=0.47 in postmenopausal women) and with IGFBP-3 (r=0.51 in premenopausal women and r=0.55 in postmenopausal women). IGF-II was also strongly positively correlated with IGFBP-3 (r=0.58 in premenopausal women and r=0.62 in postmenopausal women). IGF-I was inversely related to age, but no significant associations were found between peptide concentrations and characteristics such as BMI, smoking, parity, age at menarche, age at first birth or age at menopause in either pre- or postmenopausal women (data not shown).

Table 2 shows the risk of breast cancer associated with increasing thirds of serum IGF-I, IGF-II and IGFBP-3 in premenopausal women. Serum IGF-I concentration was not associated with breast cancer risk, either before or after adjustment for potential reproductive and other confounders; however, after statistical adjustment for serum IGFBP-3 concentration, women in the highest third of IGF-I concentration had an increased risk compared with women in the lowest third, although this was not statistically significant (OR 1.71; 95% CI: 0.74–3.95; test for linear trend; P=0.21). The association between the molar ratio of IGF-I/IGFBP-3 and breast cancer risk was similar to that of IGF-I adjusted for IGFBP-3 (OR 1.61; 95% CI: 0.75–3.47 in the highest vs the lowest third of IGF-I/IGFBP-3 concentration; test for linear trend; P=0.19). Serum IGFBP-3 concentration was associated with a reduction in risk, which was strengthened after statistical adjustment for BMI, age at menarche and age at first birth (OR 0.60; 95% CI: 0.29–1.24) for the highest vs the lowest third; test for linear trend; P=0.02). This association remained but was not statistically significant after additional statistical adjustment for serum IGF-I concentration (OR 0.49; 95% CI: 0.21–1.12; test for linear trend; P=0.07). There was no association between serum IGF-II concentration and risk in premenopausal women, either before or after statistical adjustment for IGFBP-3 (Table 2), and no association was found for the molar ratio of IGF-II/IGFBP-3 (data not shown).

For postmenopausal women, serum concentrations of IGF-I, IGF-II and IGFBP-3 were not associated with breast cancer risk either before or after statistical adjustment for potential confounding factors (Table 3).

DISCUSSION

The findings from this small prospective study are consistent with a moderate increase in breast cancer risk associated with high levels of IGF-I, after adjustment for IGFBP-3. This is broadly consistent with previous prospective studies that have found IGF-I, as absolute concentrations or relative to IGFBP-3, to be predictive of breast cancer risk in premenopausal women (Hankinson et al, 1998; Toniolo et al, 2000; Krajcik et al, 2002; Muti et al, 2002), although one study found no association between serum IGF-I levels and risk in women who were under the age of 50 years at the time of diagnosis (Kaaks et al, 2002). The finding that serum IGF-I was not associated with breast cancer risk in postmenopausal women is consistent with data from other prospective studies (Hankinson et al, 1998; Kaaks et al, 2002; Krajcik et al, 2002; Toniolo et al, 2000; Keinan-Boker et al, 2003; Grønbæk et al, 2004).

Our data also show that a relatively high circulating IGFBP-3 concentration is associated with a reduction in risk in pre-, but not postmenopausal women. Although our findings are consistent with the nonsignificant inverse association reported in one previous prospective study (Hankinson et al, 1998), others have found either no association (Toniolo et al 2000; Kaaks et al, 2002) or an increased risk with increasing IGFBP-3 concentration in premenopausal women (Toniolo et al, 2000; Krajcik et al, 2002; Muti et al 2002). Nevertheless, our findings are consistent with what is currently known about the growth-inhibitory properties of IGFBP-3. It is thought that a relatively high IGFBP-3 concentration may indirectly reduce cancer risk by binding a greater proportion of circulating IGF-I, thereby reducing its bioavailability and thus inhibiting its mitogenic effects (Jones and Clemmons, 1995). In addition, IGFBP-3 has been shown to directly inhibit cell growth and induce apoptosis of breast cancer cells independently of its effect on IGF-I (Oh et al, 1993; Gill et al, 1997).

In this study, adjusting for IGFBP-3 increased the OR for the association of IGF-I with premenopausal cancers. There is considerable controversy in the literature regarding measurements of IGFBP-3. Several commentaries have noted that measurements of IGFBP-3 are still hampered by technical problems, particularly in relation to standardisation of assays. It is now evident that there is considerable heterogeneity in the forms of IGFBP-3. In addition to major differences in glycosylation and proteolytic modification, it is clear that there are many other post-translationally modified forms present in the circulation. The heterogeneity of the protein presents a considerable challenge for assay calibration. The various reports of strong and sometimes conflicting associations between IGFBP-3 concentrations and cancer risk indicate that it would be of considerable interest to measure individual forms of IGFBP-3. At present, there are no assays that can reliably measure specific characterised isoforms of IGFBP-3 and these isoforms can only really be assessed by electrophoretic techniques that are not easily applied to epidemiological studies.

Our finding of a lack of association between postmenopausal circulating levels of IGF-I and IGFBP-3 with subsequent breast cancer risk, albeit based on only 47 breast cancer cases, are also consistent with those reported by most other prospective studies, although the recent study of Grønbæk et al (2004) reported a positive association between IGFBP-3 and breast cancer risk. The possible difference in associations of IGF-I and IGFBP-3 and risk by menopausal status suggests that high levels of IGF-I and/or low levels of IGFBP-3 may be important in breast cancer development in younger, but not older, women. If confirmed, the stronger association of IGF-I with risk in younger than in older women might be because IGF-I levels are higher, or perhaps because oestrogen levels are high. IGF-I and IGF-II may be indirectly acting as an indicator of sex hormone activity in younger women or they may be directly interacting with oestrogens to increase breast cancer risk. Indeed, oestradiol has been shown to upregulate expression of the IGF-I receptor and both hormones can interact in a synergistic manner to stimulate breast cancer cell proliferation (Yee and Lee, 2000; Hamelers and Steenbergh, 2003). However, there are limited epidemiological data on the potential joint effect of endogenous oestrogens and IGFs on breast cancer risk (Yu et al, 2003), and other, as yet unknown, age-related changes might influence this association.

This is the first study to examine prospectively the association between serum IGF-II concentration and breast cancer risk in premenopausal women. Although IGF-II has a clearly established role in foetal development (Allan et al, 2001), its function in postnatal life is less understood, despite circulating at a much higher concentration than IGF-I. Although experimental evidence suggests that IGF-II can increase the growth of breast cancer cells in vitro (Cullen et al, 1992) and is upregulated in many breast tumours (Li et al, 1998; Fichera et al, 2000), our results show that circulating IGF-II concentration is not strongly associated with subsequent breast cancer risk in pre- or postmenopausal women, which is consistent with evidence from case–control studies (Holdaway et al, 1999; Li et al, 2001; Yu et al, 2002), although Grønbæk et al (2004) reported in a prospective study an increase in risk of breast cancer among postmenopausal women with increasing IGF-II, which was of borderline statistical significance.

The main strengths of this study are the prospective design with cases and controls matched according to predefined established risk factors for breast cancer. In addition, blood samples were taken at least 3 years before cancer diagnosis, thereby reducing the possibility that differences in IGF levels are a result of an undiagnosed breast tumour at the time of blood collection. Although peptide measurements were made in a single blood sample from each woman, previous studies suggest that this reliably reflects long-term IGF-I and IGFBP-3 levels in adult women, with intraclass correlation coefficients between 0.73 and 0.87 for samples taken a year apart (Kaaks et al, 2000; Borofsky et al, 2002), and Hunt et al (2002) have reported a Spearman rank correlation of 0.81 for measures of IGF-II in two samples taken at different rounds of recruitment. Further, all assay measurements were conducted blinded and in case–control sets, thereby minimising the impact of laboratory variation on case–control comparisons. However, little is known about the correlation between serum concentrations and bioactive tissue levels. Although IGFs originate mainly in the liver, they can also be produced locally in breast cancer cells and can thus act in an autocrine/paracrine manner as well as through endocrine pathways (Sachdev and Yee, 2001). Nevertheless, the observation that a reduction in circulating IGF-I can reduce tumour growth in animals (Dunn et al, 1997; Wu et al, 2003) suggests that blood levels of IGF-I are biologically meaningful and can serve as an indicator of the total pool of IGF-I available to cells (Holly and Hughes, 1994).

In summary, the results of this prospective study are consistent with the hypothesis that premenopausal women with a relatively high circulating concentration of IGF-I and low IGFBP-3 are at an increased risk of developing breast cancer.

Acknowledgments

References