点击显示 收起
1 From the Department of Clinical Sciences, Pediatrics, Umeå University, Sweden (OH), and the Department of Nutrition, University of California, Davis (BL).
2 Supported by Semper Foods AB, which donated the formulas used, and the Swedish Medical Research Council (05708). 3 Reprints not available. Address correspondence to B Lönnerdal, Department of Nutrition, University of California, Davis, Davis, CA 95616. E-mail: bllonnerdal{at}ucdavis.edu.
ABSTRACT
Background: The appropriate level of iron fortification in infant formula remains undetermined.
Objectives: We compared hematologic indexes and iron-status indicators in infants who were either breast-fed or fed formula with concentrations of 2 or 4 mg Fe/L and evaluated the effects of providing part of the iron as bovine lactoferrin and of adding nucleotides.
Design: Healthy term infants were exclusively breast-fed (n = 16) or fed formula (n = 10–12) from age 4 ± 2 wk to 6 mo. Anthropometric measures were taken monthly, and blood samples were taken at 1, 4, and 6 mo. Hematologic indexes; indicators of iron, zinc, and copper status; and erythrocyte fatty acids were assessed.
Results: No significant differences in hematology or iron status were observed between groups at 4 and 6 mo of age. Although 34% of all infants had a hemoglobin concentration <110 g/L at 6 mo, the absence of iron deficiency or defective erythropoiesis suggests that this hemoglobin cutoff is too high for this age group. Neither the source or the concentration of iron in formula nor fortification with nucleotides had any significant effect on serum zinc or copper, and nucleotide fortification did not affect erythrocyte fatty acids.
Conclusions: A concentration of 1.6 mg Fe/L formula meets the iron requirement of healthy term infants aged 6 mo, and providing more iron does not benefit iron stores. Fortification with bovine lactoferrin or nucleotides did not benefit either iron status or erythrocyte fatty acids. Additional studies are needed to establish age-appropriate cutoffs for iron deficiency and iron deficiency anemia in infancy.
Key Words: Infant formula • human milk • fatty acid composition • iron fortification • nucleotides • lactoferrin
INTRODUCTION
The appropriate level of iron to use in infant formula is controversial. In several countries, such as the United States, formula is available without iron fortification. Although a progressive increase in the iron concentration of unfortified formula has been noted during the past decade, the concentration is usually <0.8–1.2 mg Fe/L. It is evident, however, that long-term use of such formula leads to an increased risk of iron deficiency anemia (1–3). In the United States, iron-fortified formula conventionally contains 12 mg Fe/L, whereas European formulas contain 7 mg Fe/L. These levels of iron fortification have been shown to result in a low incidence of anemia (4, 5). We showed that infants exclusively fed formula with 4 mg Fe/L have an iron status similar to that of infants fed formula with 7 (5) or 12 (B Lönnerdal, O Hernell, unpublished observations, 1999) mg Fe/L. This finding is supported by the study of Bradley et al (6), who found similar iron status in infants fed formula with 4 or 8 mg Fe/L.
Several components have been proposed to increase the absorption of iron from infant formula. Ascorbic acid is well known to enhance iron absorption, and all infant formulas are already generously fortified with ascorbic acid. Lactoferrin has also been suggested to enhance iron absorption. It was proposed that human lactoferrin is involved in the absorption of iron from breast milk, and, as bovine lactoferrin has become commercially available, that form of lactoferrin has also been implied to facilitate iron absorption (7). Nucleotides have been added to formulas, primarily because of positive effects on the immune function of infants fed such formula (8). However, nucleotides have also been suggested to improve the iron status of infants, primarily because of findings in rats (9, 10). In the current study, we explored the effects of lowering the iron concentration of infant formula that contained potential enhancers of iron absorption from 4 to 2 mg Fe/L. The primary goal was to evaluate whether a formula with a concentration of 2 mg Fe/L would result in hematologic indexes and iron status in term infants at 6 mo of age that were satisfactorily comparable with those values in breast-fed infants and in infants fed a formula with 4 mg Fe/L. Because the lower level of iron fortification used can be expected to come nearer to meeting the iron requirements of infants during the first 6 mo of life, we also hypothesized that the effects on iron status of adding iron in the form of bovine lactoferrin or as ferrous sulfate with nucleotides would be more noticeable at this lower iron concentration. Breast-fed infants were included as a reference group. Because both nucleotides (11) and iron status (12) have been suggested to affect the fatty acid composition, particularly of the long-chain polyunsaturated fatty acids, we analyzed the fatty acid composition of the red blood cell membrane in the different groups of infants.
SUBJECTS AND METHODS
Subjects
Healthy term infants were recruited from 3 well-baby clinics in Umeå, Sweden. All infants were initially breast-fed and then allocated to a breast-fed group (n = 16) or to one of the formula-fed groups (n = 10–12/group) according to the parents’ choice. To be recruited to one of the formula groups, the infant had to be completely weaned from the breast by age 4 ± 2 wk. The research nurse and all laboratory personnel were blinded to the formula group to which the infants belonged. Written, informed consent was obtained from the parents of each child. The protocol for the study was approved by the Ethics Committee on Research Involving Human Subjects of the Medical Faculty, Umeå University.
Methods
Anthropometric measurements were taken monthly, and a venous blood sample (2–3 h after feeding) was taken at the start of the study and then at 4 and 6 mo of age when the study was discontinued. Length was measured to the nearest 0.5 cm by use of a Harpenden measuring device. Weight (naked) was measured on an electronic scale to the nearest gram. Measurements and blood samplings were performed by a research nurse during home visits. The nurse had weekly contact by telephone with each family. All but one of the recruited infants (whose family moved from the area) completed the study. Hematologic indexes were analyzed immediately. Plasma or serum were separated within a few hours, frozen, and stored at 220°C until they were analyzed. Red blood cells were washed with saline and frozen at 220°C until they were analyzed.
Diets
The formulas were whey-predominant (60:40), powdered formulas containing 13 g protein/L, and they were prepared according to the manufacturer’s instructions. They were fortified with zinc to a concentration of 4 mg/L and with copper to a concentration of 0.45 mg/L. The control formula (Baby Semp 2; Semper Foods AB, Stockholm) contained 4 mg Fe as FeSO4 (Fe4). The 3 experimental formulas contained 2 mg Fe/L: 2 contained the iron as FeSO4, of which one (Fe2+N) also was fortified with nucleotides as monophosphates (Yamasa Corp, Tokyo) to concentrations similar to those in human milk (5.5 mg 5'AMP/L, 20.1 mg 5'CMP/L, 3.2 mg 5'GMP/L, 3.2 mg 5'IMP/L, and 7.5 mg 5'KMP/L). In the second of these 2 formulas with iron as FeSO4 (Fe2+Lf), bovine lactoferrin (SMR, Malmö, Sweden) provided 1.3 mg of the total iron. The pyrogen-free preparation had a protein content of 95% (wt:wt), of which lactoferrin constituted >95%. The lactoferrin was saturated with iron and the iron content of the preparation was 1.24 mg/g protein. These iron concentrations (4 and 2 mg, respectively) were targets; the analyzed concentrations were slightly different. Note that the third 2-mg formula (Fe2) contained only 1.6 mg Fe/L (Table 1). No iron drops or solid foods were allowed during the study. However, limited quantities of fruit purées (without iron) were allowed at 4–6 mo of age. These were provided by the investigators and chosen to minimize interference with trace element status. The formulas and purées used were manufactured by Semper Foods AB. The parents and the research nurse were blinded to the formula composition, as were the personnel who carried out the analyses. The composition of the formulas is shown in Table 1 and the number of infants in each group in Table 2.
View this table:
TABLE 1 . Composition of study formulas1
View this table:
TABLE 2 . Anthropometric measures at 1, 4, and 6 mo of age1
Hematologic indexes
Hematologic indexes and iron status were analyzed at the Department of Clinical Chemistry, Umeå University Hospital, with the use of the Sysmex SE 9000 autoanalyzer (Tillqvist, Kista, Sweden). The hemoglobin concentrations were analyzed with the Sysmex Sulfolyser automated hemoglobin reagent (Toa Medical Electronics Co, Los Alamitos, CA), mean corpuscular volume (MCV) was automatically calculated from erythrocyte particle concentration, and erythrocyte volume fraction was analyzed with the use of reagents provided by the manufacturer. Serum iron and serum total-iron-binding capacity (TIBC) were analyzed by the ferrozine method (Iron kit 1553712 and UIBC kit 1030600; Boehringer Mannheim Scandinavia AB, Bromma, Sweden), and serum ferritin was analyzed by the immunoturbidometric technique (BM/Hitachi 704/717/911; Boehringer Mannheim Scandinavia AB) calibrated against World Health Organization standard 80-602. Serum transferrin receptor (TfR) was assayed with an enzyme-linked immunosorbent assay technique (Ramco, Houston).
Trace element status
Serum zinc and copper were determined with atomic absorption spectrophotometry as described previously (13).
Fatty acid composition
Venous blood, collected in tubes containing EDTA, was immediately put on ice and centrifuged at 3000 x g for 15 min at room temperature within a few hours after sampling. The plasma was removed, and the erythrocyte pellet was washed 3 times with cold 0.15 mol NaCl/L and 1 mmol EDTA/L (pH 7.4). The erythrocytes were resuspended in the same solution and, after the addition of 10 mL all-rac--tocopherol, samples were stored at 270°C until they were analyzed. Erythrocyte lipids were extracted according to the method of Folch et al (14) as modified by Dodge and Phillips (15), with the use of 20 volumes of chloroform and methanol at a ratio of 1 to 1 (vol:vol); the methanol contained butylated hydroxytoluene at a concentration of 50 mg/L. Transmethylation of fatty acids was carried out with boron trifluoride methanol according to the procedure of Morrison and Smith (16). Methyl esters were then extracted with light petroleum, evaporated under nitrogen, and dissolved in dichloromethane. Fatty acid methyl esters were separated and measured with a gas chromatograph (AutoSystem GC; Perkin-Elmer, Norwalk, CT) attached to a Perkin-Elmer integrator (model 1020) with the use of a fused silica capillary column (30 m x 0.32 mm internal diameter, 0.25-mm film thickness) (Omegawax 320; Supelco Inc, Bellefonte, PA). The injector temperature was set at 250°C and the flame ionization detector at 260 or 280°C. After 1 min, the initial column temperature of 130°C was increased by 58°C/min to 170°C and then by 38°C/min to 190°C. After 5 min at 190°C, the temperature was increased by 38°C/min to 240°C, and that temperature was maintained for 10 min. Helium at a pressure of 12 psi was used as carrier gas. Individual fatty acids were identified by comparing their retention times with reference standards (Larodan Fine Chemicals, Malmö, Sweden).
Statistical analysis
Variables were examined for conformance to the normal distribution and transformed as appropriate. For each outcome variable and subject, the slope of the relation between the variable and the subject’s age was calculated with regression analysis, and that slope was then compared between groups with the use of analysis of covariance, with the baseline value of the variable as the covariate. When the analysis of covariance indicated significant group differences (P < 0.05), multiple comparisons were performed by Tukey’s method to identify the groups that differed (P < 0.05). Values in the tables are given as means ± SDs.
RESULTS
Anthropometric measures
All formulas were well tolerated. No significant differences in weight or length at birth were observed among the groups. The weight and height of the groups at 1, 4, and 6 mo of age are shown in Table 2. After the adjustment for initial weight and height (1 mo), height was significantly greater in the Fe2+Lf group than in the Fe2+N group at 4 and 6 mo. At 6 mo of age, the weight of infants in the Fe2+Lf group was significantly greater than that of infants in the Fe2+N group.
Hematologic and iron status indexes
Hematologic variables and other indexes of iron status are shown in Table 3. No significant differences in hemoglobin, MCV, serum iron, TIBC, log serum ferritin, and serum TfR were observed at the start of the study (at age 4 ± 2 wk), but the serum iron concentration was significantly lower in the Fe2+N group than in the breast-fed group. At 4 and 6 mo of age, no significant differences in any of the hematologic indexes were observed between groups when there was correction for the initial difference in serum iron concentration. Although several indexes indicated that the iron status of the Fe4 group was greater than that of the Fe2 group, adjustment of the values by the initial means resulted in no significant differences in any of the variables between the groups at either 4 or 6 mo of age (Table 3). The sample size in each group was limited by the expense of producing these unique formulas and the costs of the clinical study and the extensive analyses. We did, however, have the power to detect differences of 1.55 SD between group means. For hemoglobin, for example, this corresponded to 10 g/L, which we would consider a biologically meaningful difference.
View this table:
TABLE 3 . Hematologic indexes and measures of iron, zinc, and copper status in all groups at 1, 4, and 6 mo of age1
Serum ferritin, serum iron, TIBC, the percentage of serum iron, MCV, serum TfR, log TfR/ferritin, serum zinc, and serum copper all showed strong tracking during the study period. Although there was no association between weight and hemoglobin, there was a strong negative correlation (r = -0.42, P = 0.0065) between the change in weight from 1 to 6 mo and the serum ferritin concentration.
Several indexes of iron status have been used to assess iron deficiency. With the use of common cutoff values for older groups (17), 20 (34%) of 59 infants were found to be anemic (hemoglobin <110 g/L) at 6 mo of age, whereas only 3 infants were iron deficient when a serum ferritin cutoff of 12 µg/L was used (16). Only one infant was found to have an MCV < 70 fL. In contrast, 36 infants (61%) were found to have a serum iron concentration <10 mmol/L. Because serum ferritin is also an acute phase reactant and therefore may be a poor indicator of iron deficiency (ie, depleted iron stores), we also evaluated the serum TfR concentration, an indicator of the cellular need for iron, which is not affected by infection and which may be a more reliable indicator of iron deficiency. Seven infants (12%) were found to have elevated serum TfR (>9 mg/L) at 6 mo of age, which indicates iron depletion (18).
Trace element status
No significant differences in serum zinc and copper concentrations were observed between groups at 4 or 6 mo of age when adjusted means were used (Table 3).
Fatty acid composition of the erythrocyte membrane
The fatty acid composition of the erythrocyte membrane in various groups is shown in Table 4. As expected, breast-fed infants had significantly higher concentrations of docosahexaenoic acid (22:6n-3) than did all formula-fed groups at both 4 and 6 mo of age, when adjusted means were used. For arachidonic acid (20:4n-6), the only significant difference was a higher mean value in the breast-fed group than in the Fe2 group at both 4 and 6 mo of age. With respect to the effect of nucleotides, no significant difference in the composition of long-chain polyunsaturated fatty acids was found between the Fe2+N group and the other formula groups or between the breast-fed group and the Fe2+N group on the one hand and the other formula groups on the other. No significant correlation between either 20:4n-6 or 22:6n-3 and weight or height at 4 or 6 mo of age was observed.
View this table:
TABLE 4 . Fatty acid composition of the erythrocyte membrane in all groups at 1, 4, and 6 mo of age1
DISCUSSION
In a previous study, we found no significant differences in any hematologic variable at 6 mo of age when exclusively breast-fed infants and infants fed formula containing 4 or 7 mg Fe/L were compared (5). This strongly suggests that a modern infant formula with a concentration of 4 mg Fe/L meets the iron requirement of healthy term infants for at least the first 6 mo of life. In the present study, we showed no significant differences in the hematologic variables between infants fed formula with 1.6 mg Fe/L (Fe2 group) and those fed formula containing 4 mg Fe/L (Fe4 group). It is therefore reasonable to believe that this lower iron concentration also meets the iron requirement of healthy term infants during the first 0.5 y of life. After 6 mo of age, iron requirements increase and this amount of iron fortification may not be sufficient to meet these needs if formula is the only diet fed. However, iron-fortified complementary foods are usually introduced at 6 mo, and iron drops are also used, which makes the appropriate level of iron fortification in formula more difficult to assess.
Several reports in the literature support the notion that the iron status of exclusively breast-fed, healthy term infants is satisfactory at 6 mo of age (5, 19–21). The concentration of iron needed in infant formula to result in the absorption of an amount of iron similar to that absorbed from breast milk can be estimated from bioavailability data. Iron absorption from breast milk was estimated to be 50% and that from infant formula to be 10% (22). Because mature human milk contains 0.2–0.3 mg Fe/L, formulas would need to contain 5 times as much iron, ie, 1.0–1.5 mg Fe/L, to provide the amount of iron absorbed from breast milk. More recent studies using stable isotope methods have shown that the iron absorbed from breast milk represents 20% of the iron content (23), and that from infant formula represents only 6% (24). According to these more recent data, formulas would need to contain 3.5 times more iron than mature breast milk to result in the absorption of similar amounts of iron. Thus, both data sets lead to the same conclusion, ie, a concentration of 1.6 mg Fe/L formula should result in the absorption of an amount of iron similar to that absorbed from breast milk. This agrees with our observations on iron status in infants at 6 mo of age. It is apparent from our combined studies that feeding infants formula with an iron concentration higher than that needed to meet the iron requirement (in our opinion, 1.6 mg/L) does not lead to any advantages with regard to hemoglobin concentrations, circulating serum iron, or iron stores (as assayed by serum ferritin or serum TfR concentrations). This is most likely explained by the well-known fact that the amount of iron absorbed is negatively proportional to the amount of iron in the diet (25). In fact, Fomon et al (24) showed that the amount of iron absorbed from infant formula containing 8 mg Fe/L is similar to that absorbed from formula containing 12 mg Fe/L.
Thus, there is no recognizable benefit from a "high" concentration of iron in infant formula. There may be, however, deleterious effects associated with high concentrations of iron. First, ferrous iron is a known prooxidant, particularly in combination with high concentrations of ascorbic acid (as in infant formula), and, because the regulation of iron absorption occurs at the mucosal level (26), the enterocytes will be exposed to high concentrations of unabsorbed iron, which may exert oxidative damage before they are sloughed. Second, we showed earlier that a higher concentration of iron fortification (7 mg/L) results in significantly lower serum copper concentrations than are seen in infants fed formula with a lower iron concentration (4 mg Fe/L) (5). Similarly, Haschke et al (27) found a significantly lower copper balance in infants fed formula with a higher iron concentration (10.2 mg/L) than in infants fed formula with a lower iron concentration (2.8 mg Fe/L).We observed no effect of formula iron concentration on serum copper concentration, which suggests that a concentration of 4 mg Fe/L does not impair copper status. However, the fact that a significant effect on copper status was observed at a modestly higher iron concentration may support the value of keeping the iron at the lowest concentration needed to meet the iron requirement.
The addition of bovine lactoferrin to infant formula with an iron concentration of 1.8 mg Fe/L did not result in any significant advantages with regard to hematologic indexes or iron status. This result is similar to that of our previous study as well as to results of other investigators (5, 28–30). In the present study, we used a lower total iron concentration in the formula, and, thus, iron bound to lactoferrin contributed a larger fraction of the total iron than it contributed in previous studies. It is reasonable to believe that this would enhance the possibility of observing an effect, had there been one. It is possible that the lack of effect on iron status is explained by the observation that bovine lactoferrin is not recognized by the mucosal lactoferrin receptor present in the small intestine of human infants (31). If this is the case, human lactoferrin would be needed for an effect to be observed. A surprising finding was that the weight gain of the infants fed the Fe2+Lf formula was significantly higher than that of the infants fed the Fe2+N formula. This observation may have been due to type 2 error, but it is interesting to note that bovine lactoferrin has been shown to have a growth-promoting effect on the intestinal mucosa (32) and on cells in culture (33). Of further interest is the fact that the only effect observed in mice whose lactoferrin gene was knocked out was impaired growth (34).
The addition of nucleotides to one of the formulas studied did not result in any beneficial effects on hematologic indexes or iron status. Note that the earlier study suggesting a positive effect of nucleotides on iron metabolism was performed in adult rats (9). It is possible that the effects of nucleotide fortification could be different between species or diets fed (ie, solid rat chow and liquid infant formula).
Because nucleotides (11) as well as iron status (12) have been proposed to affect fatty acid metabolism, we analyzed the fatty acid composition of the erythrocyte membrane. Because the formulas used contained both linoleic acid and -linolenic acid (weight ratio 9.7:1, Table 1), but were devoid of long-chain polyunsaturated fatty acids, we found the expected decline (35, 36) in 22:6n-3 with age in the formula-fed groups, but not in the breast-fed group. Similarly, at 4 and 6 mo of age, total n-3 fatty acids were higher in the breast-fed group than in all other groups (data not shown). No significant effect of nucleotide supplementation or iron fortification of the formula on the fatty acid composition of the erythrocyte membrane was observed. The possibility exists that the concentrations of nucleotides used in our study were lower than those needed to observe effects. The concentrations chosen were similar to those used in commercial formulas available at the start of our study and to those reported for free nucleotides in human milk (37, 38). It was shown previously that the total amount of potentially available nucleotides in human milk is considerably greater (38, 39), because of the presence of DNA and RNA, and that newborns have the capacity to use these sources of nucleotides (39). Thus, it is possible that formulas with such nucleotide concentrations may have other biological effects. In addition, we cannot exclude from this study the possibility that nucleotides had an effect on the long-chain polyunsaturated fatty acid composition of the erythrocyte membrane before 4 mo of age.
Although we found no significant differences in hematologic variables or iron status between groups, several observations at 6 mo warrant further discussion. Overall, up to 34% of the infants in our study would be classified as anemic according to the cutoff value used by the World Health Organization for this age group (17). Because all of the infants were healthy, born at term, and either exclusively breast-fed or fed formulas fortified with nutrients to meet current recommendations, other causes of nutritional anemia were highly unlikely. Furthermore, none of the infants had any severe infection of long duration during the study period or had an acute infection at the time of blood sampling. Thus, were the infants anemic, iron deficiency would have been the most likely cause. However, only 3 infants at this age had a serum ferritin concentration <12 µg/L, which indicated depleted iron stores (17), and only 7 infants had elevated serum TfR (>9 mg/L), which suggests enhanced cellular iron needs (18). Because none of these infants had low hemoglobin, and because MCV values for all infants except one were satisfactory, we found it unlikely that the high proportion of low hemoglobin concentrations indicated iron deficiency anemia. We therefore believe that the current cutoff for hemoglobin is too high for this age group and should be lower. This agrees with a recent epidemiologic study from the United Kingdom (40), in which the hemoglobin concentration was examined in 8-mo-old infants. The hemoglobin values were normally distributed, with a mean of 117 g/L. On the basis of the 5th percentile, the authors of that report suggested a cutoff of hemoglobin <97 g/L for anemia. By that standard, only one of the infants in our cohort would be classified as anemic. Other possibilities would be that the cutoff for serum ferritin, indicating depleted iron stores (<12 µg/L), is too low for this age group or that serum ferritin is not a useful indicator of iron status in infants. The first of these alternatives is discredited by the fact that the use of a putative cutoff of 20 µg/L does not markedly affect the results and by the observation that the iron stores were adequate to support normal erythropoiesis as judged by MCV values.
We found that 60% of the infants at 6 mo of age had low serum iron (<10 µmol/L). Similar proportions were found in our previous study, in which we compared formulas with 7 and 4 mg Fe/L (5). To our knowledge, this cutoff has been established for adults, and few data are available for infants. We found no significant correlation between serum iron and hemoglobin or between serum iron and serum ferritin, which again indicates that different reference values may be needed for infants.
Circulating concentrations of serum TfR have been suggested to be a sensitive indicator of cellular iron needs (18, 41). In our previous study, we found no significant differences in serum TfR between breast-fed infants and infants fed formula with different concentrations of iron, although breast-fed infants (fed the lowest amount of iron) had the highest mean serum TfR concentration, and infants fed the formula with the higher concentration of iron (7 mg/L) had the lowest mean serum TfR concentration. Note, however, that serum TfR at 6 mo showed a significant negative correlation with serum ferritin. This may support the notion that serum TfR and serum ferritin are correlated with iron stores in infants. To enhance the discriminatory power of serum ferritin and serum TfR, Cook et al (42) suggested that the ratio of serum TfR to serum ferritin should be used. This is supported by the finding of a linear relation between this index and body iron stores in adults (42). Today, there are no such data available for infants. With the cutoff ratio used by Cook et al for adults, we found that only 5 infants (one in each group) had "low" iron stores. It is obvious that further studies defining the proper indicators of normal iron status in infants are needed.
In conclusion, we believe that a concentration of 1.6 mg Fe/L formula meets the requirement of healthy term infants up to 6 mo of age and that higher iron concentrations are unnecessary and may pose some risks. Furthermore, it is evident that the assessment of iron status at 6 mo of age is problematic and that further studies are needed to obtain reliable reference values and establish appropriate cutoffs for iron status in infants.
ACKNOWLEDGMENTS
We are grateful to research nurse Margareta Henriksson for work in the clinical part of the study, to Shannon Kelleher and Else-Britt Lundström for technical assistance, and to Janet Peerson for assistance with the statistical analyses.
REFERENCES