In this analysis of cross-sectional data from NHANES III, a large nationally representative sample of US adults aged 20 years and older, we document significant associations between race/ethnicity and serum IGF-I and IGFBP-3 levels. The observed differences in the IGF axis are not accounted for by the effects of anthropometric variables, notably BMI. Note that the height and weight avariables in this study were obtained from standardized measurements in the examination center, not from self report. Mexican-Americans had lower levels of IGF-I and NHWs had higher levels of IGFBP-3, resulting in the lowest IGF-I to IGFBP-3 ratio values in male and female Mexican Americans. Among men, non-Hispanic whites had the highest levels of igf-I and IGFBP-3, but Non-Hispanic blacks had the highest molar ration of IGG-I:IGFBP-3 in both men and women. These results are important because the ratio of IGF-I to IGFBP-3 is a measure of bio-available IGF-I and a potential link to variation in cancer incidence at some sites [15
]. As in past studies, serum IGF-I and IGFBP-3 levels were lower in older respondents [4
]. All three analytes showed curvilinear relationships with age and values for one gender are not uniformly higher than the other.
Because these results are based on a large nationally representative sample of US adults, they should prove invaluable as a guide to calculating reference values for the design and evaluation of future epidemiological studies of the IGF axis in many parts of the world. Details of the distributions for both analytes by age, gender, race/ethnicity, and many other classification variables can be obtained by downloading the public access data (http://www.cdc.gov/nchs/about/major/nhanes/nh3data.htm
) and calculating values for IGF-I and IGFBP-3 for demographic categories of interest. In the remainder of the paper we focus on race/ethnic differences in serum IGF levels, potential confounders of these differences, and the possibility that population level differences in IGFs may help explain variation in cancer incidence.
Race/ethnic differences in serum IGFs could occur because of genetic differences in IGF regulation among different race/ethnic groups. One study of Finnish twins reports that genetic variation accounted for 38% of the variance in serum IGF-I levels and 60% of the variance in IGFBP-3 [36
]. Second, lifestyle, anthropometric, or other variables associated with the IGF axis could account for differences in IGF levels among racial/ethnic groups. For example, diverse anthropometric variables [4
] and regional fat deposition [37
] have been associated with IGF-I levels. In our study, after adjustment for age and race/ethnicity, height was positively associated with IGF-I in men (p = 0.021) and waist circumference was negatively associated with IGFBP-3 in women (p = 0.002). Height, BMI, and waist circumference were not significantly associated with IGFBP-3 in men or women. However, adjustment for height, BMI, and waist:hip ratio did not significantly alter the associations between race/ethnicity and the IGF variables measured. Thus there appear to be associations between race/ethnicity and the IGF axis, independent of anthropometric characteristics.
It is possible that dietary intakes explain some of the observed relationships. We found little evidence for associations between food and alcohol consumption estimated from limited dietary assessment instruments available in this study and aspects of the IGF axis measured here (Potischman et al. Personal Communication), despite past studies suggesting that some dietary factors may be associated with the IGF axis [39
]. Nor did we detect differences between race/ethnic groups and dietary variables. In general, the literature relating dietary factors and IGF-I and IGFBP3 is inconsistent. For example, carbohydrate intake has been shown to be negatively [24
], positively [39
] and not associated [41
] with IGF-I or IGF-IGFBP3 concentrations. In addition, there are null intervention and cross-sectional studies relating common dietary factors to IGF-I and IGFBP3 [20
] and further work would need to address associations within race/ethnic groups.
Other potential confounders of the association between race/ethnicity and IGF levels include physical activity [42
], alcohol use [52
], and smoking [54
]. However, reports concerning these variables are contradictory and in general suggest that their effects, if present are small [4
]. It is a major challenge to determine the mechanisms responsible for race/ethnic difference in the IGF axis because of the complexity of IGF regulation [8
] and the fact that lifestyle and environmental influences on the IGF axis are still poorly understood [8
A growing literature concerning the epidemiology of IGF suggests that the IGF axis and cancer at several sites [13
] have small but significant associations. The strongest evidence to date has been found for colon, prostate and pre-menopausal breast cancer. At these sites, serum or plasma levels of IGF-I and these cancers are positively associated. Results for IGFBP-3 are mixed. A recent meta-analysis indicated that IGFBP-3 levels were associated with increased risk of pre-menopausal breast cancer but not with risk for cancer at other sites [13
]. Given that IGFBP-3 is the major binding protein of IGF-I, such associations are thought to be consistent with the regulation of IGF-I’s mitogenic effects through binding to IGFBP-3 [8
]. Thus, the elucidation of demographic and behavioral factors associated with variation in serum levels of IGF-I and its binding proteins could help explain racial variation in cancer incidence [22
Past studies have examined race/ethnic differences in serum levels of IGF and one or more of its binding proteins. A small study of middle-aged men found that reported IGF-I levels were higher in Caucasians (224 ng/ml) than in African Americans (205 ng/ml) and Asians (208 ng/ml), and IGFBP-3 levels were lowest in African Americans (3373 ng/ml) and higher in Caucasians (3868 ng/ml) and Asians (3926 ng/ml), resulting in lower ratios of IGF-I:IGFBP-3 in Asians but similar values among the other groups [25
]. Platz et al. [25
] point out that the lower level of bio-available IGF-I in Asians is consistent with their lower incidence of prostate cancer, but their results do not account for the higher prostate cancer incidence observed in African Americans. Slattery et al. report similar levels of IGF-I, lower levels of IGFBP-3 and comparable ratios of IGF-I:IGFBP-3 in Hispanic compared to non-Hispanic white women [59
]. One large study of men and women aged 45–75 years (The Multi-Ethnic Cohort (MEC)) focusing on racial/ethnic variation in cancer risk factors has examined the IGF axis. This study, of men and women aged 45–75 years, set in Hawaii and Los Angeles [20
], has reported significant associations between race/ethnicity and both IGF-I and IGFBP-3 levels [20
]. Systematic review of race/ethnic variation in the IGF axis might help identify current challenges to teasing apart environmental and genetic influences on this key molecule [4
Overall, the results of this study match some cancer data according to ethnic group incidence rates. Colorectal, pre-menopausal breast and prostate cancer risk have been linked to the IGF axis and have incidence rates that vary by race and ethnicity [63
]. Most notably, the incidence of all three cancers is lowest in Hispanics and Mexican-Americans in population-based cancer registries in the US [63
]. In our national data, the concentration of IGF-I and IGF-I:IGFBP-3 ratio is also lowest in this ethnic group in younger men and older women. Prostate cancer incidence is markedly higher in non-Hispanic Blacks (234 per 100,000) than non-Hispanic Whites or Mexican Americans (152 and 134 per 100,000, respectively) [63
]. The IGF-I:IGFBP-3 ratio is also higher in non-Hispanic Blacks than the other two groups. Compared to Non-Hispanic Whites and Mexican Americans, colorectal cancer is higher in non-Hispanic Black males and females and the IGF-I:IGFBP-3 ratio is also highest in non-Hispanic Blacks at all ages, except ages ~40–50 in men.
These considerations suggest that the IGF axis could be related to variation in cancer incidence among these race/ethnic groups. Our estimates of serum IGF levels are based on NHANES data whereas these estimates of race/ethnic variation incidence from SEER, nevertheless, both surveys are population-based with highly standardized data collection and processing systems necessary for research purposes. Only a few longitudinal studies of variation in the IGF axis have been conducted. Thus, it is difficult to determine whether cross-sectional studies reflect long-term exposures. One study indicates that IGF-I levels show little intra-individual variation over the short term (about 40 days) in relative values despite significant declines with over the long term (as subjects aged) [54
]. The Cardia Male Hormone study has examined longitudinal change in IGF-I and IGFBP-3 [21
]. However, they have not yet reported an analysis of repeatability over this time period.
The results presented here were obtained from a cross-sectional study and the serum samples analyzed had been stored for 10–16 years before analysis. Quality control and sample degradation characteristics are discussed in some detail in an earlier publication [33
]. Mean levels of IGF-I and IGFBP-3 observed in this study are comparable to those seen in many past studies [4
], but we cannot rule out differential levels of sample degradation in specific demographic subgroups.
In sum, these results provide strong evidence for race/ethnic variation in serum levels of IGF-I and IGFBP-3 in a large nationally representative sample of US adults. Furthermore, they offer an introduction to an easily accessible public use data resource that allows calculation of reference values for IGF-I and IGFBP-3 stratified by gender, age and race ethnicity. This should be of some utility for evaluating data quality in future epidemiological studies. Our results provide strong support for further efforts to explore variations in the IGF axis related to race and ethnicity and their association with cancer incidence at several sites. NHANES III also contains data concerning many more serum components and diverse measures of health status and health behavior and these data are freely available for analysis ((http://www.cdc.gov/nchs/about/major/nhanes/nh3data.htm