Circulating Insulin-like Growth Factor-I in Pregnancy and Maternal Risk of Breast Cancer

Previously, in a study nested within the Northern Sweden Maternity Cohort (NSMC), we observed that insulin-like growth factor (IGF)-I measured mostly during the first trimester of a primiparous pregnancy were positively associated with maternal risk of breast cancer (1), consistent with observations in nonpregnant women (2). To confirm our initial findings and explore the association in greater detail, we conducted a study with a very similar design, nested in the Finnish Maternity Cohort (FMC), the world's largest biorepository of serum samples from pregnant women.

Study design has been described in detail previously (3). In brief, FMC members who donated serum samples between the 6th and 14th gestational weeks of a primiparous, singleton, full-term pregnancy, younger than 40 years of age, and with no history of in situ breast or any other cancer (except nonmelanoma of the skin) were eligible. Case subjects were 535 women with breast cancer identified through linkage with the Finnish Cancer Registry. Because IGF-I concentration does not vary with gestational age during early pregnancy, 184 cases with no data on exact gestational age were also included. For each case, 2 controls were matched on age at sampling (±6 months) and date of sampling (±3 months), for a total of 1,434 controls. Ninety-two percent of the cases were younger than 50 years at the time of diagnosis; thus, the vast majority of cases were likely had a diagnosis during fertile.

The study was approved by the ethical committee of the National Institute for Health and Welfare (Oulu, Finland).

IGF-I assay was quantified by immunometric assays on an Immulite 2000 Siemens analyzer. The inter-run coefficients of variation (CV) of the laboratory quality controls were 9.3% and 3.2% at concentrations of 81.5 and 229 ng/mL, respectively. The inter- and intrarun CVs for a blinded pool of controls (mean concentration of 177 ng/mL) were 3.9% and 8.3%, respectively.

Prior to analysis, IGF-I values were log₂-transformed to normalize their distributions. The correlation of IGF-I with gestational age was assessed by Pearson's partial correlation (r = −0.03). Subjects were categorized into quintiles on the basis of IGF-I distribution among the controls. Conditional logistic regression was used to calculate the OR and corresponding 95% CI of breast cancer across quintiles of IGF-I. The associations were also explored by histologic subtypes, median ages at first full-term pregnancy (29 years), diagnosis (41 years), and lag time to cancer diagnosis (11 years), as well as in finer subgroups of the latter 3 variables. Analyses limited to women with information on gestational age and by tertiles of time in storage were also conducted. Adjustment for potential confounders (gravidity, parity by index date, family history of breast cancer, smoking, and gestational day) sporadically changed risk estimates, but with less than 5%, and were not retained in the final model. Similarly, adjustment for estradiol (available for 534 case–control sets) had negligible effect on risk estimates. All statistical tests were 2-sided, and values of P < 0.05 were considered statistically significant.

Selected characteristics of the study population and IGF-I concentrations are presented in Table 1. Cases and controls were comparable in all characteristics except for family history of breast cancer.

There was no association of breast cancer with IGF-I concentrations overall and in all the subgroup analyses (by histology, age at sampling, age at diagnosis, lag time to diagnosis, storage time; Table 2). Similarly, in analysis limited to case–control sets with information on gestational age (n = 535), there was no association between breast cancer and IGF-I, and adjustment for gestational age did not alter the risk estimates.

In contrast to our previous findings in the NSMC (1), in the FMC, IGF-I during early pregnancy was not associated with maternal risk of developing breast cancer. Both studies had very similar design and were nested in population-based maternity cohorts in neighboring countries. The samples were stored at comparative temperature (−25°C), and the mean IGF-I concentrations in the FMC controls (134.7 ng/mL) were comparable with those from the NSMC (133.6 ng/mL). IGF-I was analyzed in the same laboratory with the same assay kits. The current study is 3 times larger and had 87% statistical power to detect an OR of 1.50. Nevertheless, we cannot exclude the possibility that some analyte degradation has occurred and reduced our ability to find an existing association.

Another limitation of our study is the lack of information on estrogen receptor (ER) status of the tumors, as this is not collected centrally in Finland. Most of the cases (92%) had a diagnosis before 50 years of age and thus were more likely to be ER negative (4). The analysis by the Endogenous Hormones and Breast Cancer Collaborative Group suggested that the association of IGF-I with breast cancer is confined to hormone receptor–positive tumors (2). Thus, a relatively large proportion of receptor-negative tumors in our data could have obscured an association with hormone receptor–positive disease.

In summary, no association between IGF-I concentrations during early pregnancy and maternal breast cancer risk was observed in the FMC.

The funding source had no role in the study design, interpretation of data, and publication of results.

We appreciate the excellent technical assistance provided by Pirjo Kontiokari, Annika Uimonen, and Sara Kuusiniemi in the conduct of the study.

This work was supported by research grants from the U.S. National Cancer Institute (CA114329 and CA120061). A.T Toriola was supported by a European Association for Cancer Research (EACR) Travel Fellowship Award to visit the Division of Cancer Epidemiology, German Cancer Research Center, Heidelberg.