Regurgitation among infants occurs commonly and is estimated to affect up to 30% of infants from birth to 2 months of age. Recently ESPGHAN published guidelines for the treatment of regurgitation due to gastroesophageal reflux. Parental reassurance and dietary management by feeding thickened formulas are often the only steps needed (1). Several thickened infant formulas are therefore commercialized. Most of them contain locust bean gum (E410) or pregelatinized high amylopectin rice starch (2). Locust bean gum, an indigestible carbohydrate, is generally considered to have a negative impact on the availability of essential elements such as calcium, iron, and zinc (3,4). An adequate calcium, iron, and zinc intake is of paramount importance, especially during growth (5,6). Data available on the impact of these thickened formulas on the bioavailability of calcium, iron, and zinc in the infant are confusing and very limited. The net(t)-intake depends on the concentration of the element in foodstuffs and the availability of the element for absorptive processes. This in turn is largely influenced by the physical and chemical properties of the different food components—for example, the presence of enhancers or inhibitors (7–9).
The purpose of this study was to evaluate the effect of addition of thickening agents on the availability of calcium, iron, and zinc in infant formulas after simulated digestion in vitro. Human milk is used as the reference standard.
MATERIALS AND METHODS
Materials and Reagents
Pooled mature human milk was studied. The infant formulas were kindly provided by the manufacturers. During the first months of infancy, first-age infant formulas are used. From the age of 4 to 6 months on, infants are switched to second-age infant formulas. The high-viscosity, anti-regurgitation (AR) formulas, are thickened with locust bean gum (AR-lb) or pregelatinized rice starch (AR-rs). The energy and nutrient compositions of the samples are shown in Table 1. The energy, protein, fat, carbohydrate, and fiber content of the infant formulas were taken from producers' information. Values of energy, protein, fat, and carbohydrate data on proximate composition of mature human milk were taken from Macy and Kelly (10).
All chemicals (Merck, Darmstadt, Germany) were of analytical grade and deionased water (MilliQ; Millipore, Bedford, MA, U.S.A.) was used throughout the study. Pepsin (P-7000, from porcine stomach mucosa) and bile salts (B-8631, porcine) were purchased from Sigma (St. Louis, MO, U.S.A.) and pancreatin (107133 0500, porcine) from Merck. To simulate the gastrointestinal conditions of children less than 6 months of age, a pepsin solution was prepared by dissolving 10 g pepsin (P-7000, porcine) in 100 ml of 0.1 mol/l HCl. The pancreatin-bile mixture contained 3 g pancreatin (107133 0500, porcine) and 7 g bile (B-8631, porcine) in 1 l of 0.1 mol/l NaHCO3. Simulation of the gastrointestinal tract of children aged 6 months and more, was obtained by preparing a 20% pepsin solution (P-7000, porcine) in 0.1 mol/l HCl. The pancreatin-bile mixture contained 56 g pancreatin (107133 0500, porcine) and 21 g bile (B-8631, porcine) in 1 l of 0.1 mol/l NaHCO3.
Simulated Gastrointestinal Digestion Procedure
The entire digestion procedure was undertaken four times for human milk and for each of the infant formulas tested.
The method used is a modification of the continuous-flow dialysis in vitro model developed by Minihane et al. (7) and modified by Shen et al. (8) adapted to the upper gastrointestinal tract of infants less than 6 months of age and children aged 6 months and more. The method consists of two phases: a gastric stage and an intestinal stage.
Intraluminal Digestive Phase
Before the gastric stage, hydrochloric acid is added to lower the pH of the food sample. To simulate the gastric conditions of infants less than 6 months of age the pH is set at 4, whereas for children aged 6 months and more gastric pH is set at 2. Pepsin is added, and the sample is placed in a shaking water bath for 2 hours at 37°C. Titratable acidity is the number of equivalents of NaOH required to titrate the amount of gastric digest to 7.5 after addition of the pancreatin-bile mixture and is measured in an aliquot of the gastric digest.
Continuous-Flow Dialysis In Vitro Method
The intestinal stage is performed in a stirred cell (Amicon, Beverly, MA, U.S.A.) and takes 2.5 hours. During the first 30 minutes of the intestinal stage, a gradual pH adjustment from acid to neutral occurs (8). After 30 minutes, a pancreatin-bile mixture is added to the neutralized sample, and dialysis is continued for another 2 hours.
Calcium, iron, and zinc content of the reagent, samples, and dialysate fractions was determined in triplicate by flame atomic absorption spectrometry (AAnalyst 300; Perkin–Elmer, Norwalk, CT, U.S.A.) or electrothermal atomic absorption spectrometry (model 4100 ZL; Perkin–Elmer) for the dialysate fractions of iron. Blanks for calcium and zinc were determined in pepsin and pancreatin-bile extract and subtracted from the results. Before analysis, four portions of approximately 0.4 g of the samples were subjected to a destruction process, as described by Hendrix et al. (11).
Assessment of the Analytical Performance of the Method
Initial standardization was achieved by preparing two test samples with low and high mineral content (human milk and infant formula), at concentrations of, respectively, 0.27 mg/g and 4.10 mg/g for calcium, 0.3 μg/g and 41.4 μg/g for iron, and 1.0 μg/g and 42.3 μg/g for zinc. By using these mixtures, the recovery of calcium, iron, and zinc was determined in the test system after simulated digestion. The repeatability of the simulated digestion procedure was calculated from the availability of calcium, iron, and zinc from human milk and infant formula on four occasions during a single day (intrabatch precision). The reproducibility was obtained from 16 digestions of human milk and infant formula over a 4-day period (interbatch precision).
The availability data are expressed as means ± SD. Statistical evaluation of the data was performed with commercial statistical analysis software (Prism, ver. 2.01; GraphPad Software, San Diego, CA, U.S.A.). Mean availability from nonthickened versus thickened infant food was compared by unpaired Student's t-test. For two populations with nonhomogeneous variances, a Welch's correction was performed. The difference was considered significant at P < 0.05.
The availability of the element is calculated from the amount of element that passed the dialysis membrane in proportion to the total elemental content of the original food sample, as follows EQUATION
where D is the total content of element in the dialysate after an intraluminal digestive phase (in micrograms), W is the dry weight of the food sample used for the intestinal stage (in grams), and A is the concentration of mineral and trace element of the food sample (in micrograms per gram).
Assessment of the Analytical Performance of the Method
The simulated digestion procedure indicates high recovery, approximately 90% ± 4% for calcium, 90% ± 24% for iron, 86% ± 5% for zinc, and in human milk and 86% ± 2% for calcium, 105% ± 5% for iron, and 90% ± 4% for zinc, in formula, over the range of concentrations tested. The procedure also appears to fulfill the requirement of adequate repeatability and reproducibility. Four replicate determinations of the same food sample during a single day show a variation coefficient (mean value of both test samples) of 4.1% for calcium, 14.5% for iron, and 4.0% for zinc. When four times four replicates of a food sample are processed independently and measured in one assay, the mean coefficient of variation is 6.6% for calcium, 19.2% for iron, and 13.6% for zinc.
Availability of Calcium, Iron, and Zinc From Infant Formulas
The availability data for calcium, iron, and zinc from human milk and first-and second-age infant formula powders are shown in Tables 2, 3, and 4, respectively.
The total availability per 100 ml formula is calculated by multiplication of the mean elemental content of the dialysate after digestion with the availability. Values of calcium, iron, and zinc concentration in mature human milk were similar to those found in the study performed by Fransson and Lönnerdal (12).
First-Age Infant Formulas
From Table 2 it appears that pregelatinized waxy rice starch incorporated into a first-age infant milk powder did not reduce the availability of calcium in the upper gastrointestinal tract. Moreover, calcium availability from formula AR-1-rs (16.6% ± 0.9%) was significantly higher, compared with availability of calcium from the corresponding nonthickened formula m-1 (13.7% ± 1.3%;P < 0.05). Incorporation of locust bean gum into a first-age infant formula, however, influenced the availability of calcium. The calcium availability was significantly lower from formula AR-1-lb (9.4% ± 0.7%;P < 0.01) and formula AR-1-lb` (9.9% ± 0.3%;P < 0.05), both thickened with locust bean gum, in comparison with the corresponding nonthickened formula n-1 (13.3% ± 1.2%).
Availability of iron (Table 3) was significantly lower from the thickened formula AR-1-rs (0.28% ± 0.07%) in comparison with the nonthickened first-age formula m-1 (1.01% ± 0.17%;P < 0.001). Also, iron availability from the thickened formula AR-1-lb (0.32% ± 0.08%) was significantly lower than that from the corresponding nonthickened formula n-1 (1.28% ± 0.28%;P < 0.05). However, no significant difference in iron availability could be demonstrated between the thickened formula AR-1-lb` (0.59% ± 0.03%) and the nonthickened formula n-1 (1.28% ± 0.28%;P > 0.05).
From the Student's t-test results (Table 4), it appears that zinc availability from the nonthickened formula m-1 (8.0% ± 0.5%) and formula n-1 (6.7% ± 0.6%) was significantly higher than zinc availability from the corresponding thickened infant formulas: formula AR-1-rs (4.5% ± 0.4%;P < 0.001), formula AR-1-lb (3.2% ± 0.3%;P < 0.001), and formula AR-1-lb` (4.8% ± 0.4%;P < 0.01), respectively.
Second-Age Infant Formulas
According to the results of the calcium availability studies on first-age formulas, with respect to the availability of calcium, iron, and zinc from second-age infant formulas, similar data were obtained. The availability of calcium (Table 2) from nonthickened second-age formula m-2 (13.2% ± 0.4%) was significantly lower than from thickened formula AR-2-rs (14.6% ± 0.6%;P < 0.01). On the contrary, calcium availability from thickened formula AR-2-lb (11.9% ± 0.8%) was significantly lower than from nonthickened formula n-2 (13.5% ± 0.3%;P < 0.05).
The availability of iron (Table 3) from thickened formula AR-2-rs (0.88% ± 0.19%) was significantly higher than with the nonthickened formula m-2 (0.61% ± 0.11%;P < 0.05). The availability of iron from formula n-2 (0.37% ± 0.02%) was significantly higher than from the corresponding formula AR-2-lb thickened with locust bean gum (0.22% ± 0.05;P < 0.01).
Zinc availability (Table 4) from thickened formula AR-2-rs (13.5% ± 1.2%) was significantly higher than from nonthickened formula m-2 (11.4% ± 0.8%;P < 0.05) but was significantly lower from thickened formula AR-2-lb (6.8% ± 0.4%) than from nonthickened formula n-2 (9.3% ± 0.5%;P < 0.001).
From the results in this in vitro model, it appears that mature human milk provides optimal conditions for the availability of calcium, iron, and zinc. Incorporation of digestible carbohydrate (pregelatinized waxy rice starch) as a thickening agent in infant formulas does not influence or, in some cases, increases availability of calcium, iron, and zinc in comparison with infant formulas of normal viscosity. On the contrary, thickening of infant formula powders with indigestible carbohydrate (locust bean gum) seems to reduce availability of calcium, iron, and zinc.
The term bioavailability can be used in a very large concept to include digestion, absorption, and incorporation into metabolic processes. It can also be used in a narrow sense to mean that any potentially available part of a nutrient after gastrointestinal digestion should contribute to its bioavailability (13,14). Bioavailability should preferentially be determined by in vivo tests; however, these are time consuming, labor intensive, and often unethical. As an alternative, in vitro methods can be used to predict bioavailability from foods. The in vitro digestion model used in this study to simulate the upper gastrointestinal tract provided results correlating well with in vivo tests in adults and animals (7,9,15–17). The technique covers an intraluminal digestion phase and supposes that the dialysate would be completely absorbed in the upper gastrointestinal tract when mucosal absorption is usually complete. To describe the term in this narrow way, we propose to use the short term availability. The method was recently modified to simulate the intraluminal conditions (pH and pepsin, bile, and pancreatin output) of infants less than 6 months of age and those 6 months of age and more (D. Bosscher, unpublished results, 1999). This explains largely the different results seen for first-and second-age formula.
Numerous clinical reports have shown that infants are clearly able to digest moderate amounts of starch in spite of the infants' lower enzyme levels (18,19). In term infants Davidsson et al. (19) found that a weaning diet with a low fiber content of 1.8% to 8.0% (wheat flour) does not have an inhibitory effect on calcium, iron, and zinc absorption.
It is well known that fibers have the ability to exchange cations. The cation exchange capacity of fiber serves as a reservoir, exchanging polyvalent metal ions for hydrogen at low pH and being replaced with new cations as they become available when saliva and ingesta are mixed (20). Rendleman (21) demonstrated that in aqueous media, neutral polysaccharides have some, although little, affinity for alkali metal and alkali earth metal ions. Ha et al. (3) showed definite affinity of locust bean gum for calcium in aqueous media. Our data on the dialyzability of iron and zinc confirm the results obtained by Zemel and Zemel (4). They reported that locust bean gum causes a reduced solubility of both iron and zinc, although the nature of the metal–fiber interaction is not clear. However, comparing an AR-formula with locust bean gum in vivo, (whey-casein ratio of 20:80) with a normal whey-predominant formula, Levtchenko et al. (22) reported no difference in the calcium, iron, and zinc parameters tested in infants between birth and 13 weeks of age. Similar results were obtained by Behall et al. (23) who found no negative effect on mineral balance of calcium, iron, and zinc after adding locust bean gum in the basal diet of adult men. That there was no agreement between in vivo and in vitro results may be because of the method used or because of colonic absorption of minerals and trace elements to soluble fibers after fermentation of these fibers in the colon (24).
From these in vitro results we can conclude that human milk provides optimal conditions for the availability of calcium, iron, and zinc. Thickening infant formulas with locust bean gum (E410) can negatively influence the availability of calcium, iron, and zinc. This effect could be compensated, however, by increasing the amount of essential elements in the infant food sample.
The authors thank Dr. Tanja Mahler for kindly providing some samples of pooled mature human milk to be analyzed.
Supported by Institute for the Promotion of Innovation by Science and Technology in Flanders (IWT).
1. Vandenplas Y, Belli D, Benhamou P-H, et al. Current concepts and issues in the management of regurgitation
of infants: A reappraisal. Acta Paediatr 1996; 85:531–4.
2. Vandenplas Y, Belli D, Cadranel S, et al. Dietary treatment for regurgitation
: Recommendations from a working party. Acta Paediatr 1998; 87:462–8.
3. Ha Y, Thomas R, Dyck L, Kunkel M. Calcium
binding of two microalgal polysaccharides and selected industrial hydrocolloids. J Food Sci 1989; 54:1336–40.
4. Zemel M, Zemel P. Effects of food gums on zinc
solubility following in vitro digestion. J Food Sci 1985; 50:547–50.
5. NRV: Nationale Raad voor de Voeding. Voedingsaanbevelingen voor België.
Sint-Martens-Latem: G. De Backer, 1996.
6. Mednews and mydate.
19981109 pob.tx, Bethesda, MD, Reuters Health: 1998.
7. Minihane A, Fox T, Fairweather-Tait S. A continuous flow in vitro method to predict bioavailability of Fe from foods. In: Schlemmer U, ed. Bioavailability '93: Nutritional chemical and food processing implications of nutrient availability
. Proceedings part 2; Bundesforschungsanstalt für Ernärhung, Engesserstrabe, Germany, 1993:175–9.
8. Shen L, Luten J, Robberecht H, Bindels J, Deelstra H. Modification of an in-vitro method for estimating the bioavailability of zinc
from foods. Z Lebensm Unters Forsch 1994; 199:442–5.
9. Bosscher D, Van Dyck K, Robberecht H, Van Caillie-Bertrand M, Deelstra H. Bioavailability of calcium
from cow's milk based versus soya-based infant food. Int J Food Sci Nutr 1998; 49:277–83.
10. Macy I, Kelly H. Human milk and cow's milk in infant nutrition. In: Kon SK, Cowie AT, eds. Milk: The mammary gland and its secretion . New York: Academic Press, 1961: 265.
11. Hendrix P, Van Cauwenbergh R, Robberecht H, Deelstra H. Daily dietary zinc
intake in Belgium measured using duplicate portion sampling. Z Lebensm Unters Forsch 1998; 206:222–7.
12. Fransson G, Lönnerdal B. Distribution of trace elements and minerals in human and cow's milk. Pediatr Res 1995; 17:912–5.
13. Jackson MJ. The assessment of bioavailability of micronutrients: Introduction. Eur J Clin Nutr 1997; 51:S1–S2.
14. Bender AE. Nutritional significance of bioavailability. In: Southgate DAT, Johnson IT, Fenwick GR, eds. Nutrient bioavailability: Chemical and biological aspects . London, United Kingdom: Dorset Press, 1989: 3.
15. Miller D, Schricker B, Rasmussen R, Van Campen D. An in vitro method for the estimation of iron
availability from meals. Am J Clin Nutr 1981; 34:2248–56.
16. Luten J, Crews H, Flynn A, et al. Interlaboratory trail on the determination of the in vitro iron
dialysability from food. J Sci Food Agric 1996; 72:415–24.
17. Narasinga Rao B, Prabhavathi T. An in vitro method for predicting the bioavailability of iron
from foods. Am J Clin Nutr
18. De Vizia B, Ciccimarra F, De Cicco N, Auricchio S. Digestibility of starches in infants and children. J Pediatr 1975; 86:50–5.
19. Davidsson L, Mackenzie J, Kastenmayer P, et al. Dietary fiber
in weaning cereals: A study of the effect of stool characteristics and absorption of energy, nitrogen and minerals in healthy infants. J Pediatr Gastroenterol Nutr 1992; 22:167–79.
20. Van Soest P, Jones L. Analysis and classification of dietary fibre. In: Bratter P, Schramel P, eds. Trace element analytical chemistry in medicine and biology . New York: W. De Gruyter, 1988: 351.
21. Rendleman J. Metal-polysaccharide complexes. Part 2. Food Chem Toxicol 1978; 3:127.
22. Levtchenko A, Hauser B, Vandenplas Y. Anthropometry and nutritional value of a whey adapted formula or a thickened AR-formula in term born infants (abstract). J Pediatr Gastroenterol Nutr 1997; 24:491.
23. Behall K, Scholfield D, Lee K, Powell A, Moser P. Mineral balance in adult men: Effect of four refined fibers. Am J Clin Nutr 1987; 46:307–14.
24. Torre M, Rodriguez R, Saura-Calixto F. Effects of dietary fiber
and phytic acid on mineral availability. CRC Crit Rev Food Sci Nutr 1991; 1:1–22.