In this study, body size in early life was inversely associated with circulating IGF-1 and IGFBP-3 concentrations in adulthood. Women in the highest categories of body fatness had approximately 10% lower IGF-1 levels than women who were leanest in youth and adolescence. After adjustment for BMI at blood draw, these associations were somewhat attenuated but persisted. Associations were more modest for IGFBP-3. We observed no strong evidence that the association between early-life body size and IGF-1 or IGFBP-3 differed by menopausal status, age, or study cohort.
Several groups have examined the relation between birth weight and adult IGF-1 levels (40
). Although most reported no association between birth weight and circulating IGF-1, Jernström and Olsson (44
) reported an inverse association, in agreement with our findings. However, this study was small (n
= 40) and the women were aged 19–25 years. Notably, in our previous analysis of early-life body size and adult hormone levels (23
), we observed a suggestion of a positive association between birth weight and adult IGF-1, a finding which was not replicated with the larger sample size here. It is possible that a large sample size is needed to detect the modest inverse association between birth weight and IGF-1 that we observed here. To our knowledge, fewer inquiries into the association between early-life body size and adult IGF-1 and IGFBP-3 levels have been conducted (40
). One study found no association with adiposity at age 1 year or 5 years (40
), while another found an inverse association only among persons who were lean as adults (43
). In a study of 728 participants, nonsignificant inverse associations between childhood BMI and adult IGF-1 (P
-trend = 0.09) and IGFBP-3 (P
-trend = 0.13) were observed (45
). In this study, the adults were, on average, about 20 years older (mean age = 71 years) than the adults in the present study when their IGF-1 and IGFBP-3 levels were measured, suggesting that the inverse relation between childhood size and adult IGF-1 levels persists into old age.
In utero and early-life exposures have been the subjects of much investigation. Birth weight, a proxy for in utero exposures, has been linked consistently to increased premenopausal breast cancer risk. In a recent meta-analysis, Michels and Xue (1
) estimated the relative risk in the highest birth weight category (vs. the lowest) to be 1.25 (95% confidence interval: 1.14, 1.38) for premenopausal women, but they found no association among postmenopausal women. By contrast, increasing body fatness in childhood and adolescence has been associated with decreased breast cancer risk in a number of studies (3
) (reviewed by Ruder et al. (2
)). In the NHS and NHSII, women in the highest body fatness categories (somatotypes 6–9) at age 5 years had a statistically significant decrease in breast cancer risk compared with those in the lowest category (3
). Similar relations were observed for body fatness at ages 10 and 20 years (for all associations, P
-trend < 0.0001). These results suggest that early-life body size plays an important role in future breast cancer risk; however, the mechanisms through which early-life body size may act are unclear.
One possible mechanism is that early-life exposures may be involved in establishing set points of hormone levels later in life (reviewed by Michels and Xue (1
) and Schernhammer et al. (23
)). Although greater adiposity in childhood has been associated with increased IGF-1 measured concurrently (22
), the inverse association with circulating IGF-1 in adulthood suggests that the association between body size and IGF-1 levels may change over time. This suggests that early-life exposures could have long-term effects on the IGF pathway, which, in turn, could influence later cancer risk. For example, it has been hypothesized that decreased nutrition to the fetus causes permanent reprogramming of IGF signaling (47
). However, the mechanism driving the change in the direction of the association between childhood size and IGF-1 levels measured in childhood and adulthood is unclear. One hypothesis may be that increased adiposity in childhood causes negative feedback via the pituitary gland, resulting in lower long-term growth hormone/IGF levels (23
Early-life body size and the IGF pathway also may act via mammographic density. Mammographic density is associated with greatly increased breast cancer risk (48
), but the mechanisms that link mammographic density and breast cancer are not well understood (49
). In the NHS, increased childhood body fatness was inversely associated with mammographic density (P
< 0.001) (50
). In addition, increased adult IGF-1 levels have been associated with increased breast density, particularly among premenopausal women (51
). Although it is plausible that early-life exposures have independent associations with IGF-1 and mammographic density in adulthood, our findings suggest that childhood body size may act through the IGF pathway to influence mammographic density, which in turn is associated with increased breast cancer risk.
Other mechanisms through which early-life body size may act on breast cancer risk are also plausible. Although early-life exposures may affect sex steroid hormone levels in adults, few associations between early-life body size and adult circulating sex steroid hormone levels were observed in NHSII (55
). Increasing body fatness at ages 5 and 10 years was associated with decreased plasma prolactin levels in postmenopausal women (56
) but not premenopausal women (55
). Prolactin, a growth hormone, has been linked to breast cancer risk (57
), indicating that altering prolactin concentrations may be another pathway through which early-life body size may affect breast cancer.
The major strength of this study was its large sample size, with adequate power to detect modest associations and to evaluate effect modification; however, there were several limitations. First, IGF-1 and IGFBP-3 were measured in multiple batches over several years, with substantially different mean levels across batches, even when adjusted for age. However, the associations were consistent across batches, indicating that batch-to-batch variability did not influence the results. Further, the single measurements of IGF-1 and IGFBP-3 limited our ability to evaluate associations over time within the same woman. However, the intraclass correlation coefficient over 2–3 years in premenopausal women was high (the range of intraclass correlation coefficients was 0.69–0.83 for IGF-1 and 0.69–0.76 for IGFBP-3 over the course of the menstrual cycle) (28
), indicating that a single measure is a good approximation of long-term IGF-1 and IGFBP-3 levels. In addition, IGF-1 and IGFBP-3 values were correlated, so results from the mutually adjusted analyses may have been unstable because of collinearity. However, the correlation was moderate and the sample size was large, so collinearity is unlikely to have had a substantial effect on the observed associations. The association between IGF-1 and early-life body size changed very little when adjusting for IGFBP-3, although the association between IGFBP-3 and early-life body size was attenuated after adjustment for IGF-1. As such, it is possible that the relation with IGF-1 is the more important biologic association.
Self-reported childhood body fatness may be subject to misclassification because of the amount of time elapsed between childhood and response to the questionnaire. In addition, self-reported size may be influenced by current body size; women may report a size more closely related to their current size than their childhood size. However, in a validation study of self-reported childhood body fatness, the correlation between reported and measured childhood body fatness ranged from 0.48 to 0.68 and did not change substantially when adjusted for current BMI (37
), indicating only modest misclassification due to poor recall or current BMI. Any error due to misclassification would likely be nondifferential with respect to IGF-1 and IGFBP-3 and would therefore underestimate the true associations. Nevertheless, misclassification may be an important source of error in this study. Additionally, we could not account for many important factors in our analysis—namely, genetic variability (58
), glucose levels (60
), and stress (61
). These factors should be accounted for in future studies.
In summary, our results suggest that birth weight and body size during childhood and adolescence are associated with circulating IGF-1 and IGFBP-3 concentrations in adulthood. This may help to explain the mechanism underlying the association between childhood body fatness and breast cancer risk. However, the relation between birth weight and adult IGF-1 suggests that the biologic mechanisms between in utero exposures and cancer risk in later life are complex. Future studies, with repeated samples taken at defined intervals in childhood, adolescence, and adulthood, will help to elucidate the biologic pathways through which body size, particularly in early life, may affect cancer risk.