点击显示 收起
1 From the Department of Medicine, Kansas University Medical Center, Kansas City; the Department of Food Science and Human Nutrition, Iowa State University, Ames; and the Laboratory for Human Nutrition, Swiss Federal Institute of Technology, Zurich, Rüschlikon, Switzerland.
2 Supported in part by Nestec Ltd, Lausanne, Switzerland. 3 Address reprint requests to MB Reddy, 1127 Human Nutritional Sciences Building, Iowa State University, Ames, IA 50011. E-mail: mbreddy{at}iastate.edu.
ABSTRACT
Background: Considerable data are available on the individual effects of dietary factors on nonheme-iron absorption, but their combined effect when they are present in the same meal is not known.
Objective: Our objective was to predict the bioavailability of iron from complex meals that are consumed commonly in the United States on the basis of the contents of factors that are known to promote or inhibit food iron absorption.
Design: Radioisotopic measurements of nonheme-iron absorption from 25 meals were made in 86 volunteer subjects by using extrinsic radioiron labeling. The meal contents of nonheme iron, calcium, ascorbic acid, polyphenols, and phytic acid were determined by biochemical analysis; energy and protein contents were estimated from food-composition tables. Animal tissue content was based on weight or was obtained from the manufacturer.
Results: After adjusting iron absorption for differences in iron status, the significant biochemical predictors of iron absorption as determined by multiple regression analysis were the contents of animal tissue (P = 0.0001), phytic acid (P = 0.0001), and ascorbic acid (P = 0.0441). Collectively, these 3 variables accounted for 16.4% of the variation in absorption. On the basis of the multiple regression analysis, we developed the following equation to estimate iron absorption: Ln absorption, % (adjusted to serum ferritin concentration of 30 µg/L) = 1.9786 + (0.0123 x animal tissue in g) - (0.0034 x phytic acid in mg) + (0.0065 x ascorbic acid in mg).
Conclusion: For the 25 meals evaluated, only the contents of animal tissue, phytic acid, and ascorbic acid were useful for estimating nonheme-iron absorption. 2000; 71:937–43.
Key Words: Iron • bioavailability • absorption • biochemical composition • nonheme iron • phytic acid • phytate • ascorbic acid • vitamin C • animal tissue content
INTRODUCTION
The large and expanding list of dietary factors that affect the bioavailability of food iron has made it increasingly difficult to predict iron absorption from complex meals (1–3). Because of their cost and complexity, it is impractical to rely on radioisotopic measurements of iron absorption in humans to evaluate the unlimited number of possible food combinations in meals. One approach to addressing this problem has been the development of in vitro measurements of food iron bioavailability, such as dialyzable iron (4) or iron uptake in the Caco-2 cell culture model (5). However, these techniques have been more useful for assessing the effects of isolated determinants than for evaluating the overall iron bioavailability of a complex meal.
Considerable data have accumulated over the years from studies that reported the effects of individual dietary factors on iron absorption. The facilitating effects of animal tissues and ascorbic acid have been known for several decades and form the basis for a widely used model for estimating nonheme-iron absorption (6). Recently, vitamin A and ß-carotene have also been shown to enhance nonheme-iron absorption in humans (7). A variety of factors that inhibit food iron absorption have been identified; these include phytic acid (8–11), polyphenols (12, 13), phosphorus (14), and calcium (14–16). Earlier studies indicated that dietary fiber has an inhibiting effect on iron absorption (17, 18), but this finding was later attributed to its phytate content (19). Most of the studies concerning the effects of these inhibitors have been performed with single food items and there is scant information about their influence in complex meals or their relative potency in the presence of other facilitating or inhibiting factors.
The purpose of the present investigation was to determine whether analyses of the primary determinants of food iron availability could be used as a practical guide for estimating the amount of nonheme iron absorbed from meals containing several food items. In human subjects, we measured nonheme-iron absorption from 25 widely consumed convenience meals that contained a wide spectrum of ingredients. The results were correlated with 6 meal constituents measured in fresh or lyophilized portions of each meal.
SUBJECTS AND METHODS
Subjects
Nine separate studies were performed with a total of 86 subjects. Each study group included 8–10 subjects, with the exception of one study that involved 15 subjects. The composite group included 52 men and 34 women ranging in age from 19 to 40 y, with an overall mean age of 26 y. The geometric mean serum ferritin concentrations in the study groups ranged from 28 to 77 µg/L, with a composite mean of 41 µg/L. Eleven of the subjects had iron deficiency, as defined by a serum ferritin value <12 µg/L, but none were anemic. All subjects were healthy and denied a history of disorders known to influence the gastrointestinal absorption of iron. Written, informed consent was obtained from each volunteer before the study. All experimental procedures were approved by the Human Subjects Committee of the University of Kansas Medical Center.
Absorption measurements
Four separate iron absorption tests were performed in each subject by using dual 59Fe and 55Fe radioisotopes sequentially. Because iron absorption tests were performed for unrelated investigations in some studies, absorption data reported here were obtained from 1–4 meals in any particular group of subjects. All meals were consumed between 0700 and 0900 after subjects had fasted for 10 h. Subjects were permitted to drink only water for 3 h after consuming the test meal, after which they were allowed to resume their regular diets. On the day before they consumed the first test meal, we obtained 30 mL blood for the measurement of background blood radioactivity, packed cell volume, and serum ferritin concentration. On the next 2 d, test meals tagged with either 74 kBq 55Fe or 37 kBq 59Fe were consumed. Fourteen days after the second test dose, the subjects returned to the laboratory and a fasting blood sample was obtained for measurement of incorporated red cell radioactivity. The third and fourth test meals, tagged with 59Fe or 55Fe, were consumed on 2 successive days. Subjects returned 14 d after the final test meal so we could obtain blood to measure the increase in red cell radioactivity.
All test meals were labeled extrinsically by adding 1 mL 0.01 mol HCl/L containing 0.1 mg Fe as FeCl3 plus the required amount of radioactive iron. The extrinsic label was mixed with the entire meal when studying frozen meals and was added to the bread for meals that included a sandwich. All measurements of blood radioactivity were performed on duplicate 10-mL samples of whole blood by using a modification of the method of Eakins and Brown (20). Absorption was calculated by using blood volume estimated from the height and weight of each subject (21, 22). Incorporation of absorbed radioactivity was assumed to be 80% in all subjects (23).
Because of the marked influence of iron stores on iron absorption and the appreciable differences in iron status among the subjects, a method was required to adjust absorption data obtained for each subject to a common amount of iron stores. Dietary iron absorption by each subject was adjusted to a value corresponding to a serum ferritin concentration of 30 µg/L by using the following equation, which was reported by Cook et al (24):
RESULTS
The 25 meals studied varied widely in composition (Table 1). The energy and protein contents ranged from 837 to 4992 kJ (
View this table:
TABLE 1.. Meal composition
Intercorrelations of meal constituents are shown in Table 2. The highest positive correlations were observed between nonheme-iron and both phytic acid and polyphenol contents and between polyphenol and ascorbic acid contents. In contrast, the highest negative correlation was observed between animal tissue and calcium contents. Less significant correlations were observed between phytic acid and both calcium and polyphenol contents.
View this table:
TABLE 2.. Intercorrelations of meal constituents1
There was striking variation in measured mean iron absorption, from low values of 1.2% and 2.8% for 2 large breakfast meals containing >15 mg Fe to a high value of 19.7% for meal 14 (Table 3). The overall mean absorption for the 25 meals was 5.4%. After correcting for differences in iron status, the adjusted mean absorption values were similar to the measured mean absorption values because the iron status of the subjects did not differ appreciably. The one exception was that the mean serum ferritin of 77 µg/L in study V resulted in a 2-fold increase in adjusted absorption. After adjusting for iron status, the composite mean absorption for all 25 meals increased from 5.4% to 7.4%. The correlations between meal constituents and adjusted iron absorption are shown in Table 4. A highly significant, positive correlation with iron absorption was observed for animal tissue content, whereas negative correlations were seen with phytic acid and calcium contents and, to a lesser extent, with polyphenol and nonheme iron contents. Ascorbic acid content was not significantly related to iron absorption.
View this table:
TABLE 3.. Measurement of iron status and iron absorption in humans
View this table:
TABLE 4.. Correlations of meal constituents with iron absorption (%)1
In view of the significant intercorrelations between various meal constituents, a more meaningful assessment of their influence on iron absorption was obtained by performing multiple regression analysis (Table 5). A backward variable-elimination procedure was performed by including all the variables (listed in Table 4) at first and then eliminating the least significant variable, one at a time, until only significant variables (P 0.05) were left in the model. After the regression analysis was completed, 16.4% (P = 0.0001) of the variance in iron absorption was accounted for by animal tissue (P = 0.0001), phytic acid (P = 0.0001), and ascorbic acid (P = 0.0441) contents. A major portion of the total variance in iron absorption was explained by the animal tissue and phytic acid contents of the meals. Ascorbic acid made a much smaller, although significant, contribution than did animal tissue and phytic acid. In contrast, polyphenols, calcium, and nonheme iron were not significant in predicting iron absorption. Based on the regression analysis (Table 5), the following model was developed to estimate iron absorption:
View this table:
TABLE 5.. Regression analysis: predictors of nonheme-iron absorption1
DISCUSSION
In recent years, several dietary factors that influence the absorption of nonheme iron from foods have been identified. Most of the data were obtained in human subjects by measuring the absorption of nonheme iron from single meals tagged extrinsically with radioactive iron (30). Certain obstacles interfere with attempts to translate this information into practical dietary recommendations for increasing or reducing the absorption of dietary iron. Many previous studies were performed by adding purified ingredients to test meals rather than by attempting to define the influence of the native dietary components; this has been particularly true for phytic acid (8–11). Other food constituents, such as polyphenols, occur in numerous biochemical forms or complexes in foods, and different assay methodologies were used in each study. Difficulties arise even when the biochemical nature of the dietary compound is well defined. For example, in most studies of ascorbic acid, purified vitamin was added to the meal because the native dietary form is often partially degraded by cooking or allowing the food to stand before serving (32). Ascorbic acid may be partially degraded to dehydroascorbic acid, which is detected by some methods but not others. A similar problem exists in studies with meat, because of the variable degradation of heme depending on the method of preparation (33).
Another important limitation in applying our current knowledge of factors that affect the absorption of food iron is the interactions that occur when 2 active compounds are contained in the same meal. For example, the promoting effects of ascorbic acid and animal tissue on iron absorption are not additive when they are contained in the same meal (34, 35) and the same is undoubtedly true for a large number of dietary inhibitors of iron absorption. Ascorbic acid was shown to be more effective in increasing iron absorption from low-bioavailability foods than from those containing meat (34, 36). Another example of the unpredictable effects of multiple dietary factors is the studies with NaFeEDTA, which is a promising form of iron for fortification. When added to a meal of low bioavailability, this iron chelate enhances iron absorption, but iron absorption is unaffected when it is added to a meal of high bioavailability (37). The model that was developed many years ago to estimate iron bioavailability does not take into consideration either the amounts of the absorption inhibitors or the combined effect of the absorption enhancers that are present in the same meal (6). Recently, Tseng et al (38) refined this model by including inhibitors found in Russian diets, but the authors did not account for the combined effect of enhancers and inhibitors. Because the inhibitory or enhancing effects of various factors depend on other factors in the meal, it is important to examine the combined effect on iron absorption of all the factors.
The difficulty in predicting iron absorption from complex meals was approached in the present study by basing the estimates of bioavailability on the contents of factors known to influence nonheme-iron absorption. We attempted to include the main reported determinants of nonheme-iron absorption and it is unlikely that other major factors will be found. There are several potential limitations in using this approach. Selection of the type of meals studied could be important. Many of our meals were frozen dinners because this facilitated preparation for several participants at one time. These meals have a well-defined composition but are widely consumed only in the United States and may not represent the diets in other parts of the world. We attempted to select these meals in such a way that they contained a wide spectrum of factors that are known to influence iron absorption. The choice of the analytic methodology may also play a major role in developing a model for predicting iron absorption. We attempted to select the most widely used methods of food analysis.
Finally, variation in iron status, and therefore iron absorption, among different groups of subjects is a well-known methodologic problem in human radioisotopic studies. In the preliminary evaluation of our data, we used serum ferritin concentration as one of the independent variables and observed that it accounted for 32% of the overall variability in iron absorption. In the regression model reported here, we eliminated the effects of differences in iron status by adjusting each individual absorption value to a constant serum ferritin concentration of 30 µg/L. We used this approach rather than reference-dose absorption because a previous study showed that the ferritin-correction method is more reliable (24).
One of the surprising findings of the current study was that even after we adjusted each individual's absorption to their iron status, only 16% of the variation in absorption was explained by dietary factors and that animal tissue and phytic acid were the main factors that had a major influence. Inclusion of serum ferritin concentration as one of the independent variables along with the dietary factors improved the variability of unadjusted absorption to 50%. Our results indicate that dietary factors make a small contribution to determining the amount of iron absorbed and that some unknown physiologic factors contribute the majority of variation in iron absorption. Variation in iron absorption data in human studies is not uncommon even after adjustment for reference-dose absorption or iron status.
Certain dietary factors were correlated with iron absorption when examined individually (Table 4), but their influence largely disappeared in multiple regression analysis. The most notable example was calcium, which correlated negatively with iron absorption when tested separately but had no effect in multiple regression analysis, perhaps reflecting its relations with phytic acid content, animal tissue content, or both. A similar finding was the significant correlation between iron absorption and polyphenol content when examined in isolation; this correlation also disappeared with multiple correlation analysis, perhaps because of the relation between polyphenols and phytic acid. Ascorbic acid had no effect on iron absorption when tested alone, but showed a borderline significant effect in multiple regression analysis (P 0.05). The influence of ascorbic acid on iron absorption was less pronounced than that of animal tissue and phytic acid; the reason might be that most of our test meals contained animal tissue, which may have obscured the influence of ascorbic acid.
One important issue not addressed in this investigation is the relevance of using single meals to estimate food iron bioavailability. There is increasing evidence of large differences in nonheme-iron absorption with extrinsic labeling of single meals, and these differences were greatly reduced when iron absorption was measured from a complete diet (24, 39–41). For example, calcium has been shown to reduce iron absorption in single-meal studies, but the effect was dampened when absorption was measured from a 5-d varied diet (41) or from a complete diet (39). In one study, the difference in absorption between high- and low-bioavailability meals was 6-fold when the foods were consumed as single meals, but was only 2.5-fold when the total diet was evaluated (24). These differences can be attributed to differences in meals because single meals were controlled by the investigator whereas the total diet was self-selected by the subjects. However, these results parallel to some extent the findings of the present study in that a 10-fold difference in iron absorption from individual meals was observed, but only 3 meal components were found to be relevant when all meals were analyzed together. Although it is clear that the extent of enhancement or inhibition is less pronounced when iron absorption is measured over several days than when it is measured from single meals, single-meal studies are still useful for identifying the relative effects of enhancers and inhibitors of iron absorption.
In conclusion, our regression equation based on animal tissue, phytic acid, and ascorbic acid contents is useful for predicting nonheme-iron absorption from meals typical of Western diets. To use our model, one has to estimate or measure the animal tissue content of the meals. Although cooked, lean, animal tissue content is needed to predict iron absorption, a reasonable estimate can be made from the raw tissue content (1.3 g raw tissue = 1 g cooked tissue) as suggested by Monsen (42). Ascorbic acid content can be obtained easily from food-composition tables. Phytic acid content is difficult to obtain, unless it is measured in the laboratory, because of the lack of information in food-composition tables; however, reasonable estimates can be made from published values (43).
REFERENCES