Effects of weaning cereals with different phytate contents on

Effects of weaning cereals with different phytate contents on hemoglobin, iron stores, and serum zinc: a randomized intervention in infants from 6 to 12 mo of age1–3 Torbjörn Lind, Bo Lönnerdal, Lars-Åke Persson, Hans Stenlund, Catharina Tennefors, and Olle Hernell ABSTRACT Background: Weaning foods frequently contain phytate, an inhibitor of iron and zinc absorption, which may contribute to the high prevalence of iron and zinc deficiency seen in infancy. Objective: The objective was to investigate whether either an extensive reduction in the phytate content of infant cereals or the use of milk-based, iron-fortified infant formula would improve iron and zinc status in infants. Design: In a double-blind design, infants (n = 300) were randomly assigned to 3 cereal groups from 6 to 12 mo of age: commercial milk-based cereal drink (MCD) and porridge (CC group), phytatereduced MCD and phytate-reduced porridge (PR group), or milkbased infant formula and porridge with the usual phytate content (IF group). Venous blood samples were collected at 6 and 12 mo. Dietary intake was recorded monthly. After the intervention, 267 infants remained in the analysis. Results: Hemoglobin concentrations of < 110 g/L, serum ferritin concentrations of < 12 g/L, and serum zinc concentrations of < 10.7 mol/L had overall prevalences at baseline and 12 mo of 28% and 15%, 9% and 18%, and 22% and 27%, respectively. After the intervention, there were no significant differences in any measure of iron or zinc status between the CC and the PR groups. However, hemoglobin was significantly higher (120 g/L compared with 117 g/L; P = 0.012) and the prevalence of anemia was lower (13% compared with 23%; P = 0.06) in the PR group than in the IF group, which could be explained by differences in daily iron intake between the 2 groups. Conclusion: Extensive reduction in the phytate content of weaning cereals had little long-term effect on the iron and zinc status of Swedish infants. Am J Clin Nutr 2003;78:168–75. KEY WORDS Infants, cereals, iron, zinc, phytate, randomized controlled trial, weaning, Sweden

INTRODUCTION Weaning foods in developing countries are usually based on cereals, which contain phytate (myo-inositol hexaphosphate and other inositol phosphates), a known inhibitor of iron and zinc absorption (1, 2). These phytate-containing foods may therefore be a strong contributing factor to poor iron and zinc status, which is commonly seen after 6 mo of age, primarily in low-income countries (3, 4) but also in high-income countries (5, 6). In a study from Malawi, a high intake of phytate was correlated with poor 168

iron and zinc status in preschool children (7). Infant diets are usually more varied in industrialized countries, but commercial weaning foods commonly contain cereals with a high phytate content. Infant cereals often are iron fortified, but the phytate content may reduce mineral bioavailability, which would counteract the effects of the fortification. Few studies have evaluated the effects of the phytate content of weaning cereals on the iron and zinc status of infants, especially after long-term consumption. Possible effects are likely to vary with weaning practices, which differ considerably between cultures and countries. In Sweden, the intake of commercially manufactured, cereal-based products is high from 6 mo of age, as both milk-based cereal drinks (MCDs) and porridges are consumed, but the intakes of infant formula, conventional follow-on formula, and unmodified cow milk are low (8). The MCDs and porridges are composed of cooked cereals, skim milk powder, and vegetable fat, and they are fortified with minerals (ie, iron and calcium) and vitamins (ie, vitamins A, D, E, C, B-6, and B-12, thiamine, niacin, folic acid, and pantothenic acid). The composition of the products complies with European Union legislation (9). The commercially available MCDs and porridges contribute substantially to the intake of iron during infancy (8, 10, 11). Anemia and iron deficiency are present in a significant proportion of infants, even those fed iron-fortified foods. We previously observed a 13% prevalence of anemia [hemoglobin (Hb) < 110 g/L] among otherwise healthy and well-developed 12-mo-old Swedish

1 From the Department of Clinical Sciences, Pediatrics (TL and OH), and the Department of Public Health and Clinical Medicine, Epidemiology (TL, L-AP, and HS), Umeå University, Umeå, Sweden; the Department of Nutrition, University of California, Davis (BL); the International Centre for Diarrhoeal Disease Research, Bangaladesh, Centre for Health and Population Research, Dhaka, Bangladesh (L-AP); the Department of Biomedicine and Surgery, Linköping University, Linköping, Sweden (CT); and Semper AB, Stockholm (CT). 2 Supported by grants from the Swedish Council for Forestry and Agricultural Research, the Swedish Nutrition Foundation, the Sven Jerring Foundation, the Samariten Foundation, the Oskar Foundation, the Swedish Medical Research Council, and Semper AB. 3 Address reprint requests to T Lind, Department of Public Health and Clinical Medicine, Epidemiology, Umeå University, SE-901 87 Umeå, Sweden. E-mail: [email protected]. Received January 15, 2003. Accepted for publication February 10, 2003.

Am J Clin Nutr 2003;78:168–75. Printed in USA. © 2003 American Society for Clinical Nutrition

PHYTATE-REDUCED INFANT CEREAL

169

TABLE 1 Energy and nutrient content of the infant cereals and formula1

CC Energy (kcal) Protein (g) Fat (g) Carbohydrate (g) Iron (mg) Zinc (mg) Calcium (mg) Vitamin C (mg) Phytate (mmol) IP6 (% of total phytate) Phytate:iron5 Phytate:zinc5

66 2.3 2.4 8.6 1.22 0.3 93 17 30.0 65 1.4:1 6.5:1

Milk-based cereal drink or formula 6–7 mo 8–11 mo PR IF CC PR 66 2.3 2.4 8.6 1.22 0.3 93 17 9.9 87 1:2.2 2.2:1

65 1.3 3.4 7.3 0.43 0.5 44 13 0 — 0 0

65 2.4 2.5 8.0 1.22 0.4 106 17 30.0 75 1.4:1 4.9:1

65 2.4 2.5 8.0 1.42 0.3 106 17 3.4 83 1:7.4 1:1.3

Porridge IF

CC

6–7 mo PR

65 1.3 3.4 7.3 0.43 0.5 44 13 0 — 0 0

101 2.5 2.8 16.5 1.64 0.4 149 28 23.1 88 1:1.2 3.8:1

101 2.5 2.8 16.5 1.64 0.4 149 28 23.1 88 1:1.2 3.8:1

IF

CC

8–11 mo PR

IF

101 2.5 2.8 16.5 1.64 0.4 149 28 23.1 88 1:1.2 3.8:1

96 2.8 2.7 15.0 1.74 0.5 149 28 78.5 86 2.6:1 10.3:1

98 3.2 2.9 14.7 1.34 0.5 149 28 9.7 86 1:2.4 1.3:1

96 2.8 2.7 15.0 1.74 0.5 149 28 78.5 86 2.6:1 10.3:1

1 Per 100 g ready-to-feed product. CC, commercial milk-based cereal drink and porridge; PR, phytate-reduced milk-based cereal drink and phytatereduced porridge; IF, infant formula and porridge with the usual phytate content; IP6, myo-inositol hexaphosphate. 2 Iron pyrophosphate. 3 Iron sulfate. 4 Eka-Fe. 5 Molar ratio.

infants, born at term and fed according to current recommendations including iron-fortified infant cereals (6). Twenty-six percent had low concentrations of serum ferritin (< 12 g/L), and 36% had low concentrations of serum zinc (< 10.7 mol/L). We speculated that the high phytate content of weaning foods contributed to these high numbers, especially because iron and zinc intakes appeared adequate to meet requirements, as was found in earlier Swedish studies in similar age groups (8, 10). Our hypothesis was that a diet with extensively phytatereduced infant cereals (MCDs and porridge) or milk-based infant formula (instead of MCD) and regular porridge with the usual phytate content would result in a lower prevalence of depleted iron stores (ie, serum ferritin: < 12 g/L) and of low serum zinc (< 10.7 mol/L) than does the currently used feeding regimen of commercial MCD and porridge. SUBJECTS AND METHODS Participants We recruited 300 infants aged < 6 mo from 6 well-baby clinics in Umeå, a Swedish university town with a population of 105 000, after obtaining written informed consent from the parents. The well-baby clinics have an attendance of > 95% of the children in the area. Recruited infants were all term (gestational age at birth, 38–42 wk) with birth weights > 2500 g and with no known chronic illness, previous treatment with iron supplements, developmental delay, or feeding problems. Infants who at age 6 mo had Hb < 100 g/L, serum ferritin < 12 g/L, and mean corpuscular volume (MCV) < 70 fL were considered to have iron deficiency anemia (IDA), and they were excluded from the study and recommended for appropriate treatment. Infants who at age 9 mo had Hb < 105 g/L, serum ferritin < 12 g/L, and MCV < 70 fL were also considered to have IDA; they were removed from the study and recommended for appropriate treatment, and their data were not used for the study analyses. Other exclusion criteria were severe and protracted illness and allergy or intolerance to the study products. The study

was approved by the Research Ethics Committee, Faculty of Medicine and Odontology, Umeå University, Umeå, Sweden. Outcomes and sample size The major outcomes of the study were Hb < 110 g/L, serum ferritin < 12 g/L, and serum zinc < 10.7 mol/L at 12 mo of age. Sample size calculations were based on a low serum ferritin (< 12 g/L) prevalence of 25% and a low serum zinc (< 10.7 mol/L) prevalence of 35% in a high-phytate group (6) and an estimated low serum ferritin prevalence of 5% and a low serum zinc prevalence of 15% in a low-phytate group (12). With 95% CI and a power of 80%, we needed 58 infants per group to show a difference in the prevalence of low serum ferritin and 82 infants per group to show a difference in the prevalence of low serum zinc. To allow for a dropout rate of 20%, we included 100 infants per group. This sample size would allow us to detect statistical group differences of 2.0 g/L in Hb and of 0.45 mol/L in serum zinc. Randomization and intervention In a double-blind intervention, infants were randomized in blocks of 12 to 1 of the 3 dietary intervention groups from 6 to 12 mo of age (Table 1). All 3 diets contributed adequate amounts of all macronutrients and micronutrients when consumed as part of an age-appropriate, diversified diet. Both breastfed and nonbreastfed infants were recruited. If the participating infants were breastfed when entering the study, the mothers were encouraged to breastfeed as long as they wished. At the mothers’ own discretion, the study products were introduced into the infants’ diets from 6 mo of age, with no other interventions being done. The study nurses allocated each recruited infant to one of the study groups according to a predetermined, randomly generated list of numbers. The nurses supplied the families with the predetermined MCD or formula and porridge ad libitum and free of charge. The research nurses, investigators, and families were all blinded as to the feeding group to which the individual infant belonged. The code was broken only after data collection was complete and the intentionto-treat analyses had been performed.

170

LIND ET AL

The commercial phytate group (CC group) received MCD and porridge, both with their usual phytate content. The phytatereduced group (PR group) received extensively phytate-reduced MCD and phytate-reduced porridge, and the infant formula group (IF group) received milk-based infant formula and porridge with the usual phytate content (Table 1). In the CC and PR groups, 2 different MCDs were used, one from 6 mo of age and another from 8 mo of age, to reflect the common feeding pattern. In the IF group, the same infant formula (Baby Semp 2; Semper AB, Stockholm) was used throughout the study. Because the contribution of iron from porridge in the first 2 mo of the study was considered small, because of low intake, and thus the contribution of phytate also relatively low, the 6-mo porridge was the same in all 3 groups. From age 8 mo, identical regular porridges were used in the CC and IF groups, whereas the PR group used an extensively phytatereduced product. Semper AB manufactured all study products. Phytate reduction In the commercial MCD (consumed from age 6 mo) used in the CC group, the cereal part was made up of oat and wheat flours, which were mixed with milk powder and fortified with iron. The cereal in the commercial MCD had an already somewhat reduced phytate content through activation of endogenous phytases in the manufacturing process, whereas the porridge cereal retained its original phytate content. For the MCD in the PR group, the oat flour contents were reduced, and low-extraction white wheat flour was added. This reduced the phytate concentration from 30.0 to 9.9 mol/100 g, or 67%, in the extensively phytate-reduced MCD. In the MCD and porridge used from age 8 mo, the oat, wholemeal wheat, and rye flours were scalded and soaked at pH 4.5, which allowed endogenous phytases to further reduce the phytate in the cereal. The slurry was then drum-dried and mixed with lowextraction white wheat flour. This hydrothermal process reduced the phytate concentration from 30.0 to 3.4 mol/100 g, or by 89% in the extensively phytate-reduced MCD, and from 78.5 to 9.7 mol/100 g (88%) in the phytate-reduced porridge. All products were iron-fortified. The phytate content of the MCD and porridge was analyzed as individual inositol phosphates by HPLC according to a method of Sandberg and Ahderinne (13).

infant during 5 consecutive days. Household measures were used for quantities, and the participating family was encouraged to serve the study infant all meals from a standardized plate. The families were also given a booklet with photos of different portion sizes of common infant foods, using the standardized plate, to facilitate consistency of recording. Breastfeeding was recorded as a “meal,” ie, equivalent to a full meal, or as a “snack,” ie, a short feed mainly for comfort or other nonnutritive purposes, according to the mother’s own perception. Daily energy and nutrient intakes were calculated with the use of MATs software (Rudans Lättdata, Västerås, Sweden), which uses the food-composition tables of the Swedish National Food Administration (15). The database was supplemented with baby foods, formulas, and recipes not originally included, according to information from the participating families and the manufacturers. Intakes of breast milk were set at 134 g per meal ≤ 8 mo of age, at 102 g per meal > 8 mo, and at 25 g per snack at all ages (16). Nutrient intake from breast milk was calculated according to Jensen (17) and Tsang et al (18). Nutrient and energy intakes from breast milk were added to the total daily nutrient and energy intakes of the study subjects for the period from 6 mo through 8 mo. Statistical analysis For statistical computations, SPSS software (version 10.0; SPSS, Inc, Chicago) was used. Skewed variables, eg, serum ferritin, were logtransformed. Main outcomes are shown as means ± SDs, geometric means (when applicable), and proportions. In the intention-to-treat analysis for the continuous variables, analysis of variance (ANOVA) was used. The intention-to-treat analysis was repeated after stratification for sex. To adjust for effects over time, we used analysis of covariance (ANCOVA), using the initial values at 6 mo as covariates for results at 12 mo. Significant differences between study groups underwent Bonferroni correction to allow for multiple comparisons. In comparisons of the proportions, the chi-square test, Fisher’s exact test, or both were used. Factorial analysis was used on the monthly diet registrations. The dietary data are shown as mean or geometric mean daily intakes during the periods of 6–8 mo and 9–11 mo, respectively. Intakes of energy and nutrients include both breast milk and complementary food. Statistically significant results were those with values of P < 0.05.

Biochemical assays Venous blood was collected from the participating infants at inclusion at 6 mo and again at 12 mo with the use of a zinc-free vacuum system (Vacutainer; Becton Dickinson, Plymouth, United Kingdom). Hematologic indexes and iron status were analyzed at the Department of Clinical Chemistry, Umeå University Hospital, with the use of a Sysmex SE 9000 Autoanalyzer (Tillqvist, Kista, Sweden). The Hb concentration was analyzed by using a Sysmex Sulfolyser automated hemoglobin reagent (Toa Medical Electronics Co, Los Alamitos, CA), and MCV was automatically calculated from the erythrocyte particle concentration. Serum ferritin was analyzed by using an immunoturbidometric technique (BM/Hitachi 704/717/911; Boehringer Mannheim, Mannheim, Germany) calibrated against World Health Organization standard 80–602. Serum zinc and serum copper were analyzed by atomic absorption spectrophotometry as described previously (14). Dietary intake Each month, starting from baseline, parents or caregivers recorded the type and amount of each food item consumed by the

RESULTS Of the 300 infants randomly assigned to the 3 dietary regimens, 267 (89%) completed the 6-mo intervention period, provided complete blood samples at baseline and 12 mo of age, and had 3 or more diet registrations (Table 2). Altogether, 33 infants were excluded from analysis: 24 left the study before 12 mo of age, 7 were excluded because of incomplete dietary information (ie, 3 complete registrations), and 2 were excluded because of missing blood samples (Figure 1). Of the 24 who left the study, 16 refused to continue because of the blood sampling, the diet registrations, or other reasons; 5 moved from the area; and 3 were diagnosed with cow milk protein allergy. No child was excluded because of IDA according to the stated criteria (Hb < 105 g/L, serum ferritin < 12 g/L, and MCV < 70 fL). The proportion of infants who were excluded from the analysis (total n = 33) was significantly (P = 0.042) larger for the IF group than for the CC and PR groups (17% compared with 6% and 10%, respectively), but the proportion of infants who left the study (n = 24) did not differ significantly (P = 0.17) between the 3 groups (12%, 5%, and 7% for the IF, CC, and PR groups, respectively). There were no significant differences

PHYTATE-REDUCED INFANT CEREAL

171

TABLE 2 Baseline characteristics of the participating infants and families1

Persons per household (n) University education, at least one parent (%) Girls [n (%)] Birth weight (kg) Birth length (cm) Age at start of intervention (mo) Breastfed at 6 mo [n (%)] Weight at 6 mo (kg) Length at 6 mo (cm)

CC (n = 94)

Intervention group PR (n = 90)

IF (n = 83)

3.76 ± 0.882 78 44 (49) 3.67 ± 0.50 51.0 ± 2.03 6.0 ± 0.18 69 (77) 8.03 ± 0.974 68.5 ± 2.10

3.75 ± 0.90 76 46 (52) 3.57 ± 0.48 50.7 ± 2.10 6.0 ± 0.20 67 (75) 7.96 ± 0.846 67.9 ± 2.23

3.55 ± 0.73 88 44 (55) 3.61 ± 0.47 50.8 ± 2.34 6.0 ± 0.15 58 (72) 8.03 ± 0.869 67.8 ± 2.05

1 CC, commercial milk-based cereal and porridge; PR, phytate-reduced milk-based cereal and phytate-reduced porridge; IF, infant formula and porridge with the usual phytate content. There were no statistically significant differences among study groups (ANOVA). 2– x ± SD.

between the infants who completed the trial and those who did not or between the study groups with regard to birth weight, birth length, breastfeeding, parental level of education, family size, or baseline mean Hb, MCV, serum ferritin, serum zinc, serum copper, and prevalence of anemia or low serum ferritin. There was, however, a difference in the proportion of infants with low serum zinc at baseline, inasmuch as those who did not complete the trial had a significantly lower mean prevalence of low serum zinc than did the others (6% versus 22%; P = 0.04).

At baseline, there were no significant differences between the 3 study groups for any of the socioeconomic, anthropometrical, hematologic, or biochemical variables, except that low serum zinc was significantly (P = 0.04) more common in the CC group than in the other groups (31% compared with 16% and 19% in the PR and IF groups, respectively) (Table 3). Across the 3 study groups, anemia [ie, Hb < 110 g/L (19)] was observed in 74 of 267 (28%) of the infants, depleted iron stores [ie, serum ferritin < 12 g/L (19)] in 23 of 267 (9%), IDA [Hb < 110 g/L, serum ferritin < 12 g/L, and MCV < 73 fL (19)] in 6 of 267 (2%), and low serum zinc (< 10.7 mol/L) in 59 of 267 (22%). Intention-to-treat analysis At 12 mo of age, Hb was slightly but significantly (P = 0.015) lower in the IF group than in the PR group (117 compared with 120 g/L) (Table 3). Neither MCV, mean log serum ferritin, nor serum zinc differed between the study groups, whereas serum copper did differ significantly between study groups, with the CC group having a significantly (P = 0.004) higher concentration than the PR group (17.0 compared with 15.6 mol/L). Overall, from 6 to 12 mo of age, the prevalence of anemia decreased significantly (28% compared with 15%, P = 0.001), the prevalence of low serum ferritin increased significantly (9% compared with 16%, P = 0.012), and the prevalence of low serum zinc remained unchanged (23% compared with 29%, P = 0.09). The proportion of infants with anemia at 12 mo tended to be higher in the IF group than in the other 2 groups (23% compared with 11% and 13% in the CC and PR groups, respectively, P = 0.06). There were no differences between study groups in prevalence of low serum ferritin or low serum zinc at 12 mo (Table 3). Boys had significantly lower MCV at both 6 and 12 mo of age and lower serum ferritin at 6 mo of age than did girls, but there were no sex differences in the prevalences of anemia, low serum ferritin, IDA, or low serum zinc at any age. Likewise, the effect of the intervention did not differ between the sexes. Dietary intake

FIGURE 1. Trial profile. CMA, cow milk protein allergy.

At baseline, 200 of 267 (75%) of the infants in the 3 intervention groups were still being breastfed (Table 2). Median duration of breastfeeding was 8.9 mo. There were no differences in consumption of the study products between study groups (Table 4). Factorial analysis of dietary intake showed that data for energy and most nutrients—ie, energy, protein, vitamin C, zinc, and

172

LIND ET AL

TABLE 3 Main outcomes at 6 and 12 mo of age for the 3 intervention groups1

Hemoglobin (g/L) MCV (fL) Serum ferritin (g/L)5 Serum zinc (mol/L) Serum copper (mol/L) Anemia (%) Low serum ferritin (%) Low serum zinc (%)

CC (n = 94)

6 mo PR (n = 90)

IF (n = 83)

CC (n = 94)

12 mo PR (n = 90)

IF (n = 83)

116 ± 7.12 76 ± 3.0 48.5 13.2 ± 4.04 15.4 ± 3.57 24 5 317

115 ± 9.0 76 ± 2.9 40.9 13.0 ± 3.64 14.5 ± 4.26 30 10 19

115 ± 8.5 75 ± 3.6 44.1 13.5 ± 3.31 14.5 ± 3.10 29 12 16

119 ± 8.03 76 ± 3.4 25.33 12.2 ± 2.563 17.1 ± 3.533 104 15 28

120 ± 8.43,4 76 ± 2.8 21.33 12.5 ± 2.80 15.6 ± 3.406 13 23 27

117 ± 9.5 75 ± 3.8 25.23 12.4 ± 2.683 16.4 ± 3.253 23 17 26

1 CC, commercial milk-based cereal and porridge; PR, phytate-reduced milk-based cereal and phytate-reduced porridge; IF, infant formula and porridge with the usual phytate content; MCV, mean corpuscular volume. 2– x ± SD. 3 Significantly different from 6 mo, P < 0.05 (ANOVA). 4 Significantly different from IF group at 12 mo, P = 0.05 (ANOVA with Bonferroni correction). 5 Geometric –x . 6 Significantly different from CC group at 12 mo, P = 0.012 (ANOVA with Bonferroni correction). 7 Significantly different from other groups at 6 mo, P < 0.05 (chi-square).

phytate—could be combined into 2 equal intervals, with the first interval encompassing the sixth, seventh, and eighth mo of the infants’ lives (6–8 mo) and the second encompassing the ninth, tenth, and eleventh mo (9–11 mo). Thus, data on energy and nutrient intakes are presented as mean daily intakes for the periods of 6–8 mo and 9–11 mo, respectively. Energy intake from the total diet was similar in all groups throughout the study, but the intakes of protein, iron, vitamin C, calcium, and phytate differed (Table 4), basically mirroring the differences in nutrient content between the infant cereals and infant formula. Zinc intake also differed, but only during the 6–8 mo period. Total duration and daily frequency of breastfeeding and the daily intake of breast milk were similar between the 3 groups (Table 4).

A significant proportion of the daily intake of iron and zinc came from MCD or infant formula and porridge. On average, 62% (4.5 mg) and 59% (5.6 mg) of the daily iron intake came from the study products during the periods of 6–8 mo and 9–11 mo, respectively. For zinc, 45% (1.5 and 2.0 mg) of the daily intake came from the study products during the periods of 6–8 mo and 9–11 mo, respectively. Of the total energy intake, the study products contributed on average 41% during the 6–8 mo period and 44% during the 9–11 mo period. The IF diet had a lower iron content than did the other diets (Table 1), and the total iron intake was significantly (P < 0.05) lower in the IF group than in the other groups (Table 4). After adjustment of the intention-to-treat analysis for total iron intake,

TABLE 4 Mean daily energy and nutrient intakes, consumption of study formula or milk-based cereal drink and porridge, frequency of breastfeeding, and intake of breast milk at 6–8 and 9–11 mo of age1 CC (n = 94)

6–8 months2 PR (n = 90)

IF (n = 83)

CC (n = 94)

9–11 mo PR (n = 90)

IF (n = 83)

Energy (kcal/kg/d) 89.9 ± 15.27 89.8 ± 18.50 87.9 ± 16.53 85.2 ± 15.27 87.4 ± 15.74 84.7 ± 17.83 Protein (g/kg/d) 2.4 ± 0.67 2.4 ± 0.82 2.0 ± 0.563 3.0 ± 0.71 3.0 ± 0.66 2.3 ± 0.574 Fat (g/d) 30.9 ± 6.03 31.5 ± 6.49 33.8 ± 6.283 29.4 ± 5.35 31.3 ± 6.22 33.0 ± 7.335 Carbohydrates (g/d) 98.0 ± 17.34 95.8 ± 22.15 91.3 ± 18.90 107.3 ± 21.71 106.5 ± 20.91 97.9 ± 18.455 Iron (mg/d) 7.5 ± 3.69 7.6 ± 4.27 4.7 ± 2.133 9.9 ± 3.40 10.3 ± 3.66 6.2 ± 2.004 Zinc (mg/d) 2.8 ± 0.78 2.8 ± 0.98 3.3 ± 1.273 4.0 ± 1.02 4.1 ± 1.02 4.3 ± 1.21 Vitamin C (mg/d) 124 ± 48.9 121 ± 54.0 104 ± 39.83 140 ± 48.7 139 ± 46.7 109 ± 34.54 Calcium (mg/d) 661 ± 228.5 649 ± 261.1 473 ± 138.73 832 ± 248.2 830 ± 225.0 556 ± 146.14 Phytate (mol/d)6 124 ± 82.4 48 ± 31.05 26 ± 18.23 189 ± 86.8 36 ± 18.95 62 ± 46.84 Formula or milk-based cereal (g/d) 329 ± 275.5 321 ± 268.2 310 ± 262.1 441 ± 226.1 455 ± 231.6 388 ± 229.5 Porridge (g/d) 94 ± 50.9 85 ± 51.4 85 ± 55.1 85 ± 59.4 80 ± 52.7 87 ± 62.0 Breastfeeding frequency (times/d)7 3.9 ± 2.5 4.1 ± 2.4 4.2 ± 2.7 2.5 ± 2.38 2.5 ± 1.88 3.0 ± 2.88 Breast milk intake (g/d)7 413 ± 276 424 ± 254 440 ± 277 201 ± 1948 215 ± 1758 240 ± 2358 – 1 x ± SD. CC, commercial milk-based cereal and porridge; PR, phytate-reduced milk-based cereal and phytate-reduced porridge; IF, infant formula and porridge with the usual phytate content. 2 Energy and nutrient intakes include both breast milk and complementary food. 3 Significantly different from other groups at 6–8 mo, P < 0.05 (ANOVA with Bonferroni correction). 4 Significantly different from other groups at 9–11 mo, P < 0.05 (ANOVA with Bonferroni correction). 5 Significantly different from CC group, P < 0.05 (ANOVA with Bonferroni correction). 6 Phytate from study products only. 7 At 6–8 mo, n = 70 for CC group, 67 for PR group, and 62 for IF group. At 9–11 mo, n = 34 for C group, 35 for PR group, and 33 for IF group. 8 Geometric –x for 9–11-mo period.

PHYTATE-REDUCED INFANT CEREAL the group differences in effect on Hb between the PR and IF groups became nonsignificant (P = 0.48), and the effect estimates for Hb were reduced from a difference of 3.3 g/L to 0.8 g/L.

DISCUSSION In several single-meal studies, phytate been shown to significantly inhibit iron and zinc absorption (20, 21), but few studies have explored the effects of long-term modification of phytate intake (22). We assessed the longitudinal effects of the extensive reduction of the amount of phytate in infant cereals, a common weaning food, on micronutrient status. Despite a reduction in the daily phytate intake from infant cereals of ≤ 77%, we found no greater effect on Hb, serum ferritin, or serum zinc than that with commercial weaning cereals, which are richer in phytate. Feeding infant formula, with its lower iron content but presumably higher bioavailability, resulted in a significantly lower Hb and higher prevalence of anemia than did feeding phytate- reduced infant cereal. Except for a significant difference between the study groups in the prevalence of low serum zinc at baseline, the randomization was successful. Adjusting for initial zinc status did not affect the main outcomes of serum zinc or the prevalence of low serum zinc. Intakes of MCD and formula were similar to those in other Swedish studies (11 and J Svahn, personal communication, 2002)—ie, ≤ 2 servings MCD/d and ≤ 1 serving porridge/d—which suggests that the participants used the study products as parts of a mixed diet. Although the study products were supplied without cost, they were not consumed in larger quantities than expected. However, a greater than average consumption cannot be ruled out completely when comparing the results from this study with those from a previous study in the same area of Sweden (6), in which prevalences of low Hb, low serum ferritin, and low serum zinc were somewhat higher. In this study, we made an attempt to quantify nutrient intake from breast milk, because most of the infants were still being breastfed at baseline. Even though the breastfed infants had lower nutrient intake and lower serum zinc at the end of the study than did those not breastfed, adjustment for breastfeeding status in the analyses did not change the main outcomes. Phytate, in the absence of ascorbic acid (AA), inhibits iron absorption in a dose-dependent manner above a molar ratio of phytate to iron of 1:7 (1). However, AA counteracts the effect of phytate when AA:Fe exceeds 4:1 (23). In the present study, all infant cereals except PR had phytate:Fe > 1:7, which implies that the phytate could theoretically bind all available iron, except that in the phytatereduced MCD. On the other hand, AA:Fe was high in all cereals (> 4:1). Thus, the AA content may have limited the phytate effect, ensuring similar long-term iron absorption from the commercial and phytate-reduced cereals. When Davidsson et al (24) reduced the phytate content of soy formula from phytate:Fe of 2:1 to phytate:Fe of 1:2.6 (83% reduction), Fe incorporation into red blood cells increased from 5.5% to 6.8%. An even greater increase, from 3.9% to 8.7%, was seen when all phytate was removed. However, a similar increase was achieved by increasing AA:Fe from 2.1:1 to 4.2:1 in ordinary phytate-containing soy formula. In another study, an 88% reduction of phytate in infant cereals had no effect on iron bioavailability (25), most likely because AA:Fe was already high. Studies have shown that phytate:Zn > 15:1 may be associated with increased risk of zinc deficiency (7). In the present study, phytate:Zn were generally < 10:1, and a further reduction in phytate content had no effect on serum zinc. At 12 mo of age, the PR group had significantly lower concentrations of serum copper than

173

did the CC group. A possible explanation is that zinc absorption may have been higher in the PR group, which would cause lower copper absorption. In a study on infant rhesus monkeys, zinc absorption was higher and plasma copper concentration was lower in the group fed phytate-reduced soy formula than in the group fed soy formula with the usual amount of phytate (26). In addition, infants fed a low-phytate formula had higher plasma zinc as well as coat color changes indicative of low copper status. These observations would support our interpretation of the lower serum copper observed in this study. In many industrialized countries, it is common to use infant or follow-on formulas in combination with infant cereals during the second half of infancy. We compared this regimen with the Swedish practice of using only infant cereals and found a small but significant difference in iron status. The formula-fed group had significantly lower Hb and higher prevalence of anemia than did the PR group. Although the difference in mean Hb was small (3.3 g/L) and of reasonably minor importance from a public health point of view, the difference in anemia prevalence was substantial: 23% in the IF group compared with 10% and 13% in the CC and PR groups, respectively. The IF group had a lower iron content in the diet and a lower total iron intake than did the other groups, and this most likely explains the differences in mean Hb between the groups. Despite a marginally lower mean Hb and a higher percentage of infants with anemia in the IF group at the end of the study, there were no differences in mean serum ferritin, prevalence of low serum ferritin, or prevalence of IDA, which are the other important measures of iron deficiency; this suggests that formula with 4 mg/L provides adequate amounts of iron to prevent iron deficiency in late infancy when used as part of a diversified diet. It is interesting that higher iron intake resulted in a significantly higher Hb but no effect on serum ferritin (ie, iron stores), even after control for total protein intake and weight gain. Domellöf et al (27) found that infants given supplemental iron between 4 and 6 mo of age had increased Hb regardless of their baseline iron status, which suggests that the regulation of Hb synthesis is immature at this age. The result in the present study may have a similar explanation: ie, when a surplus of iron was given as part of the normal diet, the response in Hb might have continued until 12 mo of age, but the infants would not be iron deficient. This finding needs confirmation, especially as iron supplementation of iron-replete infants and children has been associated with adverse effects (28, 29). At 6 mo of age, 28% of the infants were anemic by the current definition, 9% had low serum ferritin indicative of depleted iron stores, and 2% were classified as having IDA according to multiple criteria (Hb < 110 g/L, serum ferritin < 12 g/L, and MCV < 73 fL) (19). At 12 mo of age, the prevalence of anemia declined to 15%, and the prevalence of serum ferritin < 12 g/L doubled to 18%, but only 1% of infants had IDA according to the multiple criteria, and the prevalence of IDA was not affected by the intervention. More plausible as an explanation that insufficient iron intake causes this anemia is the fact that the current cutoffs for Hb (< 110 g/L) and serum ferritin (< 12 g/L) overestimate the problem of iron deficiency in this age group (30). Total daily zinc intake was higher in the IF group than in the other 2 groups during the 6–8 mo period, but not during the 9–11 mo period. Despite the presence of phytate, a known inhibitor of zinc absorption, no difference in serum zinc was found between the groups at the end of the study (12 mo of age). More than one-quarter of the infants had serum zinc concentrations

174

LIND ET AL

< 10.7 mol/L at 12 mo of age, which is indicative of suboptimal zinc status. Zinc deficiency has been associated with poor growth (31) as well as increased morbidity in diarrheal disease (32). The functional consequences of suboptimal zinc intake or low serum zinc concentrations have been insufficiently studied in a population such as the Swedish population and will be reported from this study elsewhere. Iron-fortified infant cereals are commonly used in the weaning diet. A concern has been that these products contain phytate and thus may contribute to poor iron and zinc status during infancy (6). In this study, we found no evidence that a substantial reduction in the phytate content of infant cereal improves iron and zinc status during infancy, at least as long as the infant cereals are sufficiently iron-fortified and contain adequate amounts of AA, ie, AA:Fe > 4:1. Single-meal and other short-term studies have established several factors that facilitate or inhibit iron and zinc absorption, eg, AA and meat as facilitators and calcium and phytate as inhibitors (33). It has been more difficult to show these associations in longterm studies or studies of absorption from complete diets. Longitudinal interventions with enhancers such as AA or meat (34, 35) or inhibitors such as calcium or phytate (22, 36, 37) have found few, if any, significant effects on iron status. The present study, in which the phytate content of infant cereal was extensively reduced to improve iron and zinc status, echoes those results. It seems that clear effects of so-called enhancers and inhibitors are less obvious in long-term follow-up of infants consuming an ordinary diet. This should have an effect on dietary guidelines for iron and zinc intakes during infancy. We acknowledge the participation of the families and the dedicated work of research nurses Margareta Henriksson and Margareta Bäckman and dieticians Inger Öhlund, Maria Sehlstedt, Anna Karlsson, Maj-Britt Nyberg, and Agneta Frängsmyr. Carina Lagerqvist at the Department of Pediatrics, Umeå; Shannon Kelleher at the Department of Nutrition, University of California, Davis; and Ann-Sofie Sandberg at the Department of Food Sciences, Chalmers University of Technology, Gothenburg, did an excellent work in handling the blood samples, analyzing serum zinc and serum copper, and analyzing phytate content, respectively. Håkan Ahlmén, Arla Foods, contributed to the development and manufacture of the study products. OH and BL contributed to the planning and analysis of the data and the writing of the manuscript. TL was the main author of the paper and participated in the planning, data collection, and analysis of results. HS took part in the data analysis and writing of the manuscript. L-ÅP and CT contributed to the planning and analysis of the data and the writing of the manuscript. CT is the director of research at Semper AB. OH is a member of the Scientific Advisory Board of Semper AB. None of the other researchers had a financial or personal interest in any of the organizations or companies sponsoring the research.

REFERENCES 1. Hurrell RF, Juillerat MA, Reddy MB, Lynch SR, Dassenko SA, Cook JD. Soy protein, phytate, and iron absorption in humans. Am J Clin Nutr 1992;56:573–8. 2. Lönnerdal B. Dietary factors influencing zinc absorption. J Nutr 2000;130(suppl):1378S–83S. 3. DeMaeyer E, Adiels-Tegman M. The prevalence of anaemia in the world. World Health Stat Q 1985;38:302–16. 4. Gibson RS. Zinc nutrition in developing countries. Nutr Res Rev 1994;7:151–73. 5. Oski FA. Iron deficiency in infancy and childhood. N Engl J Med 1993;329:190–3. 6. Persson LÅ, Lundström M, Lönnerdal B, Hernell O. Are weaning foods causing impaired iron and zinc status in 1-year-old Swedish infants? A cohort study. Acta Paediatr 1998;87:618–22.

7. Ferguson EL, Gibson RS, Thompson LU, Ounpuu S. Dietary calcium, phytate, and zinc intakes and the calcium, phytate, and zinc molar ratios of the diets of a selected group of East African children. Am J Clin Nutr 1989;50:1450–6. 8. Persson LÅ, Johansson E, Samuelson G. Dietary intake of weaned infants in a Swedish community. Hum Nutr Appl Nutr 1984;38:247–54. 9. Commission directive 96/5/EC of 16 February 1996 on processed cereal-based foods and baby foods for infants and young children. Brussels: European Union, 1996. 10. Kylberg E, HofvanderY, Sjölin S. Diets of healthy Swedish children 4–24 months old. III. Nutrient intake. Acta Paediatr Scand 1986;75:937–46. 11. Virtanen MA, Svahn CJ, Viinikka LU, Räihä NC, Siimes MA, Axelsson IE. Iron-fortified and unfortified cow’s milk: effects on iron intakes and iron status in young children. Acta Paediatr 2001;90:724–31. 12. Fuchs GJ, Farris RP, DeWier M, et al. Iron status and intake of older infants fed formula vs cow milk with cereal. Am J Clin Nutr 1993;58:343–8. 13. Sandberg AS, Ahderinne R. HPLC method for determination of inositol tri-, penta-, and hexaphosphates in foods and intestinal contents. J Food Sci 1986;51:547–50. 14. Clegg M, Keen C, Lönnerdal B, Hurley L. Influence of ashing techniques on the analysis of trace elements in animal tissue. I. Wet washing. Biol Trace Elem Res 1981;3:107–15. 15. Swedish National Food Administration. Kodlista 1995 till Livsmedelsverkets databas PC-kost. Rapport 15/95. (List of codes 1995 for the Swedish National Food Administration database for personal computers. Report 15/95.) Uppsala, Sweden: Livsmedelsverket (Swedish National Food Administration), 1995 (in Swedish). 16. Paul AA, Black AE, Evans J, Cole TJ, Whitehead RG. Breastmilk intake and growth in infants from two to ten months. J Hum Nutr Dietet 1988;1:437–50. 17. Jensen RG. Handbook of milk composition. San Diego, CA: Academic Press Inc, 1995. 18. Tsang RC, Zlotkin S, Nichols BL, Hansen JW. Nutrition during infancy. 2nd ed. Cincinnati, OH: Digital Educational Publishing Inc, 1997. 19. World Health Organization. Iron deficiency anaemia. Assessment, prevention and control. A guide for programme managers. Geneva: World Health Organization, 2001. 20. Hallberg L, Rossander L, Skanberg AB. Phytates and the inhibitory effect of bran on iron absorption in man. Am J Clin Nutr 1987;45:988–96. 21. Sandström B, Cederblad Å, Lönnerdal B. Zinc absorption from human milk, cow’s milk, and infant formulas. Am J Dis Child 1983;137:726–9. 22. Hertrampf E, Cayazzo M, Pizarro F, Stekel A. Bioavailability of iron in soy-based formula and its effect on iron nutriture in infancy. Pediatrics 1986;78:640–5. 23. Stekel A, Olivares M, Pizarro F, Chadud P, Lopez I, Amar M. Absorption of fortification iron from milk formulas in infants. Am J Clin Nutr 1986;43:917–22. 24. Davidsson L, Galan P, Kastenmayer P, et al. Iron bioavailability studied in infants: the influence of phytic acid and ascorbic acid in infant formulas based on soy isolate. Pediatr Res 1994;36:816–22. 25. Davidsson L, Galan P, Cherouvrier F, et al. Bioavailability in infants of iron from infant cereals: effect of dephytinization. Am J Clin Nutr 1997;65:916–20. 26. Lönnerdal B, Jayawickrama L, Lien EL. Effect of reducing the phytate content and of partially hydrolyzing the protein in soy formula on zinc and copper absorption and status in infant rhesus monkeys and rat pups. Am J Clin Nutr 1999;69:490–6. 27. Domellöf M, Cohen RJ, Dewey KG, Hernell O, Rivera LL, Lönnerdal B. Iron supplementation of breast-fed Honduran and Swedish infants from 4 to 9 months of age. J Pediatr 2001;138:679–87. 28. Dewey KG, Domellof M, Cohen RJ, Landa Rivera L, Hernell O, Lonnerdal B. Iron supplementation affects growth and morbidity of breast-fed infants: results of a randomized trial in Sweden and Honduras. J Nutr 2002;132:3249–55.

PHYTATE-REDUCED INFANT CEREAL 29. Idjradinata P, Watkins WE, Pollitt E. Adverse effect of iron supplementation on weight gain of iron-replete young children. Lancet 1994;343:1252–4. 30. Aggett PJ, Agostoni C, Axelsson I, et al. Iron metabolism and requirements in early childhood: do we know enough?: a commentary by the ESPGHAN Committee on Nutrition. J Pediatr Gastroenterol Nutr 2002;34:337–45. 31. Brown KH, Peerson JM, Rivera J, Allen LH. Effect of supplemental zinc on the growth and serum zinc concentrations of prepubertal children: a meta-analysis of randomized controlled trials. Am J Clin Nutr 2002;75:1062–71. 32. Bhutta ZA, Black RE, Brown KH, et al. Prevention of diarrhea and pneumonia by zinc supplementation in children in developing countries: pooled analysis of randomized controlled trials. Zinc Investigators’ Collaborative Group. J Pediatr 1999;135:689–97.

175

33. Hallberg L, Hulthen L. Prediction of dietary iron absorption: an algorithm for calculating absorption and bioavailability of dietary iron. Am J Clin Nutr 2000;71:1147–60. 34. Cook JD, Reddy MB. Effect of ascorbic acid intake on nonheme-iron absorption from a complete diet. Am J Clin Nutr 2001;73:93–8. 35. Engelmann MD, Sandström B, Michaelsen KF. Meat intake and iron status in late infancy: an intervention study. J Pediatr Gastroenterol Nutr 1998;26:26–33. 36. Dalton MA, Sargent JD, O’Connor GT, Olmstead EM, Klein RZ. Calcium and phosphorus supplementation of iron-fortified infant formula: no effect on iron status of healthy full-term infants. Am J Clin Nutr 1997;65:921–6. 37. Ames SK, Gorham BM, Abrams SA. Effects of high compared with low calcium intake on calcium absorption and incorporation of iron by red blood cells in small children. Am J Clin Nutr 1999;70:44–8.