Dietary Correlates of Plasma Insulin-like Growth Factor I and Insulin-like Growth Factor Binding Protein 3 Concentrations

Introduction

IGFs3 are important in the growth and function of many organs (1) . Circulating IGF-I concentration is a complex trait influenced by both genetic and nongenetic factors, but these have not yet been characterized in detail (2) . There is considerable interindividual variation in plasma levels of the peptide hormone IGF-I, and differences in circulating IGF-I levels have been associated with risk of diseases. There is some evidence that higher plasma levels of IGF-I are associated with lower risk of osteoporosis (3) and heart disease (4) . Several studies provide evidence that higher circulating levels of IGF-I and/or lower levels of the principal IGF binding protein, IGFBP-3, are associated with increased risk of prostate cancer (5, 6, 7) , colon cancer (8, 9, 10, 11) , lung cancer (12) , and premenopausal breast cancer (13, 14, 15) . Several of these studies provide prospective evidence for cancer risk (5 , 7, 8, 9 , 13 , 15) . The biological plausibility of the IGF-cancer link has been reviewed recently (16) .

It is well established that protein-calorie malnutrition lowers IGF levels (17) ; however, little is known about how diet affects IGF levels in well-nourished humans. One study reported that IGF-I levels were positively associated with intake of red meat and total fat and inversely associated with intake of carbohydrates (18) . In other studies, positive associations have been seen between IGF-I levels and alcohol (19) and zinc (20) .

Knowledge of dietary factors associated with IGF levels may elucidate disease mechanisms and offer methods of disease prevention. In the present study, we examined the cross-sectional association of IGF-I and IGFBP-3 levels with intakes of various dietary factors in 1037 healthy women enrolled in the NHS. In another report, we address the association of IGFs with age, BMI, smoking, physical activity, and reproductive and menopausal factors (21) .

Materials and Methods

The methods for the collection of blood samples and laboratory assays are described in detail elsewhere (13 , 21) . Briefly, the study participants were women in the NHS from whom blood samples were collected in 1989 or 1990. Participants were 43–69 years of age at the time. Samples used in this analysis were frozen and stored and later assayed in two batches in 1996 and 1998. This study was approved by the Institutional Review Board of the Brigham and Women’s Hospital. IGF-I and IGFBP-3 were assayed by ELISA with reagents from Diagnostic Systems Laboratory (Webster, TX). Details of the assays for IGF-I and IGFBP-3 levels have been reported previously (13 , 21) .

Women included in this analysis were controls in a nested case-control study of plasma hormone levels and breast cancer risk (13) . In addition, 49 women who had reported very low fat intake (<25 and <19% of energy on the 1986 and 1990 FFQs, respectively) were added to increase the range of fat intake. These cutpoints were chosen to maximize the number of women with low fat intakes on both questionnaires, and all women meeting these criteria were included in this analysis. Because fat intake decreased in the entire cohort over time, we were able to use a more stringent criteria for the later (1990) FFQ. These samples were interspersed with the control and were assayed at the same time. Women defined as postmenopausal had no menses for at least 12 months before blood sampling. To increase the number of premenopausal women, additional women were chosen according to the following criteria: 50 years of age or less at time of phlebotomy, premenopausal at time of phlebotomy (still having cycles and not taking HRT), not previously in any case-control data set, had sufficient plasma available, and had answered the 1990 diet questionnaire. Of women fitting these criteria, 400 were chosen randomly, and 88 additional premenopausal women with specific characteristics were oversampled: 25 women with a family history of breast cancer, 25 women reporting ≥15 g of alcohol daily in 1990, and 38 women with a reported fat intake <25% of calories in 1986 and ≤20% in 1990, and ≤2 drinks of alcohol/day and total daily energy intake between 1000 and 3000 kilocalories and BMI between 18 and 30 in both 1986 and 1990. Samples from these women were analyzed in the second batch. Participants had no previously diagnosed cancer (except nonmelanoma skin cancer) and could not have implausible scores for total energy intake (<500 kcal or >3500 kcal/day).

In each case, the factor that was oversampled is either hypothesized or known to affect cancer risk. If IGF-I or IGFBP-3 levels are correlated with these factors, then IGF-I or IGFBP-3 might be the mechanism by which these factors affect cancer risk. However, the distribution of these factors in the NHS population is such that a random sample would not provide enough women in the extremes of the distribution to test these associations, and thus we oversampled women in these extreme categories.

The FFQ used has been described in detail, and its validity and reproducibility have been documented elsewhere (22) . In this analysis, nutrient intakes from the 1986 and the 1990 FFQs were averaged for each woman. In a previous analysis of dietary intake and steroid hormone levels, we found that associations were stronger for the average of the two dietary assessments than with the dietary assessment closest to blood collection (23) . This is presumably because averaging two measurements of dietary intake reduces random error. Alcohol intake was categorized as 0, <5, 5–14.9, and at least 15 g/day. Nutrient and food intakes were energy-adjusted using the residual method (24) . Nutrient intakes were categorized into quintiles. Some nutrients, particularly vitamins, are frequently found in nutritional supplements. We report results for the total nutrient, which includes supplement intake, and for the nutrient intake from food only.

Because the FFQ allows only nine choices of intake for foods, the distribution of food intakes is more discontinuous than that of nutrients because nutrients can come from many foods. Therefore, foods were divided into approximate quartiles of intake. We calculated a the glycemic load of a food by multiplying its carbohydrate content by its glycemic index value (25, 26, 27) ; we then multiplied this by the frequency of consumption and summed overall food items to produce the dietary glycemic load. Dietary glycemic load thus represents the quality and quantity of carbohydrates and their interaction, with a higher glycemic index having a greater impact at higher carbohydrate intakes. Each unit of dietary glycemic load represents the equivalent of 1 g of carbohydrate from white bread.

Statistical analyses were done with SAS software (SAS Institute, Cary NC). We excluded 3 women whose hormone values were greater than the absolute value of the 75th percentile plus three times the interquartile range. Adjusted least-square mean hormone levels across categories of factors were calculated by linear regression (28) . Ps are reported for the linear trend across categories. Multivariate models were adjusted for laboratory batch and those factors previously found to be associated with IGF levels: age, menopausal status, type and duration of hormone replacement use, smoking status, physical activity, BMI, parity, and history of ever having breastfed (21) .

Results

There were 1037 women included in the study. In 1990, participants in the study had a median (10th–90th percentile) age of 50.5 (45–66) years and a median BMI of 24.1 (21–34) kg/m². Median energy intake was 1745 (1205–2398) kcal/day. The distributions of hormone levels and other covariates are described elsewhere (21) .

We calculated the age- and energy-adjusted associations of IGF-I and IGFBP-3 with intake of alcohol, energy, macronutrients, vitamins A, C, D, and E, calcium, zinc, caffeine, and seven food groups. The results of multivariate analyses were similar, and only the multivariate results are shown here.

Table 1⇓ shows the multivariate association of alcohol, energy, and nutrients with IGF-I and IGFBP-3. The highest category of alcohol intake was associated with a statistically significant higher IGFBP-3 level compared with no alcohol intake (4042 ng/ml versus 3910 ng/ml; P = 0.02 for comparison between extreme categories). Higher total energy intake (as measured in kilocalories/day) was associated with higher IGF-I (quintile 5 = 195 ng/ml versus 181 ng/ml for quintile 1; P for the linear trend = 0.006) and higher IGFBP-3 (quintile 5 = 4092 ng/ml versus quintile 1 = 3935 ng/ml; P = 0.02).

Higher fat intake was associated with lower IGFBP-3; this association appeared to be primarily attributable to saturated fat. There was little other relation between specific types of fat (animal, vegetable, polyunsaturated, monounsaturated, saturated, trans-unsaturated and ω-3 fats from marine sources) and IGF levels. Dietary carbohydrate, glycemic load, and fiber intake were also not associated with IGF-I or IGFBP-3 levels. Higher intake of protein was associated with higher levels of IGF-I. This association was limited to animal protein and did not apply to vegetable protein, possibly because the amount and range of vegetable protein intake was much lower.

We also examined the association of vitamins and other micronutrients with IGF-I and IGFBP-3. Several of these nutrients (vitamin D, calcium, and vitamin A) had a positive association with IGF-I levels only when examining the nutrient intake that came from food sources and not when considering total nutrient intake, which included intake from supplements. For vitamin A, the strongest association was with retinol (preformed vitamin A) from food sources only. Lycopene intake was positively associated with IGFBP-3.

Vitamin E, whether from food sources only or including supplements, had no association with either IGF-I or IGFBP-3. Although total vitamin C had little association with either IGF-I or IGFBP-3, vitamin C from food sources had a modest positive association with IGFBP-3. Zinc intake, whether from food sources only or including supplements, had a modest positive association with IGF-I. Caffeine intake was not associated with IGF-I or IGFBP-3.

Table 2⇓ shows the multivariate associations of food groups and foods with IGF-I and IGFBP-3. Fruit, vegetable, and poultry intake had little association with either IGF-I or IGFBP-3 levels. Red meat intake had a borderline inverse association with IGFBP-3.

IGF-I levels were higher with greater intake of whole or skim milk, even after adjustment for multiple factors, but not with other dairy foods. In addition, higher intakes of cooked or cold breakfast cereal and pasta were associated with higher levels of IGF-I. Other grain-containing foods such as bread and rice did not have this association.

The positive association between energy intake and IGFs was confined to the leaner women. For the 611 women with a BMI <25, the multivariate adjusted mean levels of IGF-I (ng/ml) across increasing quintiles of energy intake were 178, 189, 193, 205, and 202 (P for the linear trend = 0.0007). The corresponding values for the 426 women with BMI ≥25 were 187, 179, 188, 188, and 184 (P = 0.75). However, the P for the interaction term (energy intake × overweight) was not statistically significant (P = 0.11).

Among dietary factors, the association of higher dairy intake with higher IGF-I levels was particularly consistent. This association was similar among women with and without obesity, a history of benign breast disease, or a family history of breast cancer or colon cancer. The positive association between dairy intake and IGF-I was seen in the 526 premenopausal women (lowest versus highest quartile of intake; P for the linear trend, 199 versus 226 ng/ml; P = 0.001), with little relationship in the 456 postmenopausal women (174 versus 174 ng/ml; P = 0.46). However, again, the P for the interaction term (dairy intake × menopausal status) was not statistically significant (P = 0.54). Results were similar for the comparison between premenopausal women and the 306 postmenopausal women not currently using HRT.

The inverse association between IGFBP-3 and saturated fat appeared to be confined to postmenopausal women only. The mean IGFBP-3 levels (ng/ml) across increasing quintiles of saturated fat intake among premenopausal women were 3844, 3855, 3823, 3952, 3872 (P for linear trend = 0.55). The corresponding IGFBP-3 levels for postmenopausal women, not using HRT, were 4493, 4255, 3970, 4115, and 4024 (P for linear trend = 0.02). The P for the interaction term (menopausal status × saturated fat intake) was 0.06. There was no significant interaction between menopausal status and other foods or nutrients.

Methods proposed to estimate the biologically active fraction of IGF-I did not substantially change the results. The molar ratio of IGF-I:IGFBP-3 has been proposed to reflect tissue bioactivity (29) . We calculated it as (0.130 × IGF-I concentration in ng/ml) / (0.036 × IGFBP-3 concentration in ng/ml) for each of the factors in Tables 1⇓ and 2⇓ . The positive association of total energy intake with IGF-I was attenuated with the molar ratio (1st versus 5th quintile, 0.167 versus 0.173; P = 0.09).

We also calculated mean levels of IGF-I across categories of predictors, controlling for levels of IGFBP-3 (and other covariates) and likewise, levels of IGFBP-3 controlling for levels of IGF-I. The positive association of total energy with IGF-I was also attenuated when controlled for levels of IGFBP-3 (1st versus 5th quintile, 184 versus 192 ng/ml; P = 0.07). Other findings were essentially unchanged.

Discussion

In this cross-sectional analysis among >1000 women, we found associations between IGF-I and IGFBP-3 and several foods and nutrients. In particular, we found that total energy and protein intake were positively associated with plasma IGF-I levels. The association with protein intake was largely attributable to higher IGF-I levels among women who consumed higher amounts of milk. We had the ability to average two dietary assessments collected 3–4 years prior and near the time of blood collection, which should reduce random error in dietary measurement. However, we examined many foods and nutrients; associations could occur by chance and should be replicated in other studies.

Plasma IGF-I levels are decreased by malnutrition, in both animals and humans, and protein is particularly important for increasing IGF-I levels during refeeding (17 , 30) . However, few studies have been examined linking dietary components in relation to IGF levels in well-nourished humans. Kaklamani et al. (18) examined the associations of dietary data collected by a FFQ with plasma IGF-I and IGFBP-3 levels in 130 healthy Greek subjects between 30 and 84 years of age. They found that IGF-I levels were positively associated with intake of red meat and total fat and inversely associated with intake of carbohydrates. In addition, they found that levels of IGFBP-3 were inversely associated with intake of saturated fat (18) .

Goodman-Gruen and Barrett-Connor (19) reported a positive association between alcohol use and IGF-I levels among men and women in Rancho Bernardo, and the linear relationship was statistically significant among men. We found a nonlinear association between alcohol intake and IGF-I that was somewhat positive for low intakes but inverse for the highest category of intake. However, the distribution of alcohol intake in the Rancho Bernardo study, which was conducted in an older population (mean age of the women, 77 years), overlapped only the low end of the intake distribution in our study. Therefore, the results of the two studies were compatible. We also found a positive association between alcohol intake and IGFBP-3 levels. The significance of this finding is not clear, but it would tend to make bioavailable levels of IGF-I even lower in the highest category of alcohol intake.

We found a positive association between IGF-I levels and intake of total energy, after controlling for BMI and physical activity, with an 8% increase between the lowest and highest intake. We also found a positive association between protein intake and IGF-I levels (13% increase between extremes of intake). Turnover of epithelial cells requires substantial energy and protein and is increased by IGF-I (31) . It is possible that at times of protein-calorie malnutrition, down-regulation of cell turnover may be advantageous to conserve protein and energy. Thus, the decline of IGF-I levels in the setting of malnourishment could serve a physiological purpose. On the other hand, we speculate that there may be a tendency in some individuals for excess nutrition to be associated with elevation of IGF-I levels and increased cellular turnover rate that over prolonged periods of time could lead to increased cancer risk. In an animal carcinogenesis model, supplementation of IGF-I attenuated the protective effect that dietary restriction had on tumor incidence (32) . However, in this cross-sectional study, we cannot distinguish which comes first: whether high intake of energy leads to higher IGF-I levels or whether having high IGF-I levels causes an individual to consume more energy.

Total energy intake has been consistently associated with cancer development in animals (33, 34, 35, 36) , although less consistently in humans (37, 38, 39) . The positive association of IGF-I levels with intake of total energy in our study was somewhat stronger among the leanest women. This association could also be a reflection of residual confounding with physical activity (21) . It may be important in the future to assess the association between energy intake and human cancer risk stratified by obesity.

The association between protein and IGF-I appeared to be limited to animal and not vegetable protein intake. Among animal protein-containing food groups, we found no association with red meat or poultry intake and a moderate positive association with fish intake. The relationship between protein or animal protein intake and IGF-I was not completely attenuated when controlled for dairy intake. However, the lack of association between IGF-I and other sources of animal protein makes dairy food the most likely cause of the relationship between IGF-I and protein.

Devine et al. (20) reported that zinc intake, as measured by weighed diet records, was positively associated with IGF-I levels in >100 postmenopausal women enrolled in a study of calcium supplementation for bone density. Although the major source of zinc in the diet is animal products, in their study this association persisted after control for protein intake. The authors postulated that zinc is necessary for hepatic IGF-I gene synthesis (20) . We also found that intake of zinc, whether from food sources only or including supplements, was associated with higher levels of IGF-I.

Among other micronutrients, we found a positive association with calcium from foods but not from total calcium including supplements. Likewise, we found a stronger association with vitamin A from foods, retinol from foods, and vitamin D from foods than from total vitamin A, retinol, and vitamin D including supplements. These findings cast doubt on calcium, vitamin A, retinol, and vitamin D being a cause of higher IGF-I levels. In fact, in models containing both dairy intake and calcium, vitamin A, retinol, or vitamin D from foods, the association with IGF-I is attenuated, except for vitamin A from foods, where the association is approximately the same. In 1990, the largest (nonsupplement) dietary source of calcium and vitamin D was dairy food, and dairy foods were among the highest contributors to the intake of vitamin A and retinol from foods as well. The positive association we observed between lycopene intake and IGFBP-3 is consistent with the inverse association of lycopene with prostate cancer risk (40) .

Our most consistent dietary finding was the positive association of IGF-I levels with total dairy and milk intake. This association is the most logical explanation for our finding a stronger association with calcium from foods but not total calcium including supplements, and the similar results with vitamin A and retinol. Supporting this finding was a positive association with grain-containing foods, most strongly with breakfast cereal. We also found no association with dietary fiber. In our data, the correlation between milk intake and cereal intake was 0.29. It is possible that these findings are attributable to the joint consumption of cereal and milk.

Dairy foods other than milk (ice cream, yogurt, and cheese) did not have an association with IGF-I. It is possible that either IGF-I in cows’ milk or a substance in milk that stimulates endogenous production of IGF-I is inactivated during the processing of milk to ice cream, yogurt, or cheese. There was an inverse association between ice cream intake and levels of IGFBP-3. This may have to do with the saturated fat content of ice cream. However, cheese also has a high saturated fat content, and no such association was seen with cheese.

Two other studies have supported an effect of milk intake on IGF-I levels. A randomized trial of 204 men and women where the intervention was to encourage consumption of three servings/day of nonfat milk to affect bone remodeling found that the 101 subjects in the intervention group had a statistically significant 10% average increase in serum IGF-I levels, whereas the control group had no change in levels (41) . In addition, Ma et al. (42) observed a positive association between intake of dairy food and IGF-I levels among 318 men enrolled in the Physicians’ Health Study.

IGF-I occurs naturally in cows’ milk, but orally ingested IGF-I has been thought not to be absorbed. Supporting this conclusion is an unpublished study done by the Monsanto Agricultural Co. and communicated to the Food and Drug Administration in 1989 (43) . Because IGF-I mediates the actions of GH, this study was conducted in anticipation of the Food and Drug Administration’s approval of the use of rBGH in dairy cows to improve milk production. In this study, rats were orally administered high doses of recombinant IGF-I. No changes were seen in the serum IGF-I levels, blood and urine chemistries, and organ weights of the treated rats, as were seen in rats administered IGF-I or GH parenterally. Thus, it was considered unlikely to have any effect when ingested orally by humans. However, a recent study has reported that radioactively labeled IGF-I in milk fed to neonatal rats could be detected in low levels in their systemic circulation (44) . In addition, neonatal rats fed excess amounts of IGF-I had increased growth of specific organs (45) .

It is uncertain whether results of absorption studies in neonatal rats are relevant to the adult humans in the present study. In any case, the Food and Drug Administration’s approval of the use of rBGH for dairy cows did not occur until 1994, and thus, the blood samples in the present study, collected in 1989 and 1990, were not affected by rBGH. However, bovine IGF-I (which is identical to human IGF-I), normally occurs in the milk of untreated cows, which is identical to human IGF-I and is increased in rBGH-treated cows (43) . The possibilities that IGF-I in milk could be absorbed by the human gut, or that a component of milk, a food ideally suited for the growth of the infant mammal, may stimulate endogenous production of IGF-I, deserve investigation.

The primary limitation of this study is its cross-sectional nature; there is no way to know whether factors associated with IGF levels determine those levels or are in fact determined by them. Because subjects were sampled to increase the range of fat and alcohol intake, the distribution of dietary factors in the study population does not necessarily reflect the distribution of these factors in the underlying population. The strengths of this study include its large size and the extensive information on diet and lifestyle factors collected over 25 years.

We examined many dietary factors and found associations with several. However, the clearest findings in this study of healthy women were a positive association between intake of total energy and milk and circulating IGF-I levels. The milk association is supported by several pieces of evidence. This includes positive associations between IGF-I and nutrients from food sources but not supplements where the most frequent food source of those nutrients is milk. The milk association is consistent with similar findings in a cross-sectional study of men (42) and a randomized trial of increasing milk intake (41) . These results raise the possibility that milk consumption could influence cancer risk by a mechanism involving IGF-I. In fact, positive associations between milk intake and risk of prostate cancer have been reported (46, 47, 48, 49, 50, 51) . In the NHS, one or more servings of milk/day was associated with a higher risk of serous ovarian cancer (relative risk, 1.66; 95% confidence interval, 1.10–2.51) compared with three or fewer servings/month (52) .

In conclusion, we found that women consuming higher levels of alcohol and lower levels of saturated fat had higher levels of IGFBP-3. In addition, those consuming higher levels of protein, energy, and milk had higher circulating IGF-I levels. These results raise the possibility that cancer risk could be reduced by dietary manipulation of IGF levels.