Measurement of iron absorption and iron gains from birth to 6 months in breastfed and formula-fed infants using iron isotope dilution

Abstract

Little is known about iron kinetics in early infancy. We administered stable iron isotopes to pregnant women and used maternal-fetal iron transfer to enrich newborn body iron. Dilution of enriched body iron by dietary iron with natural isotopic composition was used to assess iron kinetics from birth to 6 months. In breastfed (BF, n = 8), formula-fed (FF, n = 7), or mixed feeding (MF, n = 8) infants, median (interquartile range) iron intake was 0.27, 11.19 (10.46–15.55), and 4.13 (2.33–6.95) mg/day; iron absorbed was 0.128 (0.095–0.180), 0.457 (0.374–0.617), and 0.391 (0.283–0.473) mg/day (BF versus FF, P < 0.01); and total iron gains were 0.027 (−0.002–0.055), 0.349 (0.260–0.498), and 0.276 (0.175–0.368) mg/day (BF versus FF, P < 0.001; BF versus MF, P < 0.05). Isotope dilution can quantify long-term iron absorption and describe the trajectory of iron depletion during early infancy.

Isotope dilution of ⁵⁷Fe-enriched body iron quantifies iron absorption in breast- and formula-fed infants from birth to 6 months.

INTRODUCTION

Iron deficiency (ID) is common in young children (1) and has been linked to impaired neuromotor development (2) even in early infancy (3). At 6 months of age, ID affects up to 15% of breastfed infants in high-income countries (4) and up to 70% of infants in low-income countries (5, 6). Because of the challenges of performing iron studies in early infancy, little is known about iron absorption or gains from birth to 6 months. In humans, iron absorption is responsible for control of body iron because iron cannot be actively excreted (7). However, physiological (basal) iron losses occur, principally from cell desquamation (8, 9). In healthy infants, iron stores endowed at birth together with dietary iron absorption need to cover not only basal losses but also rapid expansion of the red cell mass and myoglobin in tissues (10).

After oral administration of a stable iron isotope tracer (e.g., ⁵⁷Fe and ⁵⁸Fe) there is an equilibration period in which the tracer concentration becomes enriched and equilibrated in all body iron compartments (11). Thereafter, a decrease (dilution) of enriched tracer concentration in the body can only occur through the addition of dietary iron with natural isotopic composition. The decrease in tracer abundance is therefore proportional to the addition to body iron by iron absorption (12). The principle of isotope dilution has been used to quantify long-term iron absorption in toddlers (12) and school-age children (11) but, to our knowledge, has not been applied in infancy.

Isotope dilution may be particularly useful in early infancy when dynamic shifts between body iron compartments make interpretation of usual iron biomarkers difficult (13). However, a challenge to applying isotope dilution in early infancy is that, after tracer administration, an equilibration period is needed to uniformly label body iron before measurements can be made. This equilibration period is ~7 months in toddlers (12). Here, we administered stable iron isotopes to pregnant women in their second and third trimester and used maternal-fetal iron transfer to label the newborn. Infants were studied at three postnatal visits, at ~2 days and at 3 and 6 months of age. This allowed us to use isotope dilution to measure and compare iron absorption from birth to 6 months in infants who were fully breastfed, fully formula-fed, or receiving mixed feeding.

RESULTS

The study took place between February 2016 and April 2019. Thirty-two mother-infant pairs entered the study, two lost interest and left the study after the 3-month visit, and 30 completed all three study visits (Fig. 1). We excluded data from seven infants because the quantity and/or quality of a blood sample was inadequate for complete analysis of all parameters. Three of the infants were in Switzerland, and 20 were in Mexico. Eight infants (Switzerland: n = 2; Mexico: n = 6) were exclusively or nearly exclusively breastfed until 6 months of age. Seven of these infants were exclusively breastfed until 6 months of age, and one of these infants received, in addition to breast milk, very small amounts of low-iron formula between 5 and 6 months of age. Seven infants (Mexico: n = 7) were exclusively formula-fed with high-iron formula containing 1.2 mg of iron/100 ml as ferrous sulfate until 6 months. The remaining eight infants (Switzerland: n = 1; Mexico: n = 7) were fed a mixture of breast milk and high- or low-iron formula until 6 months. On the basis of the pattern of infant feeding, we divided the infants into three feeding groups: fully breast feeding (BF) (n = 8), fully formula feeding (FF) (n = 7), and mixed feeding (MF) (n = 8).

Fig. 1. Study design.

Table 1 shows gender, gestational age, and birthweight, and at the three postnatal study visits, age and weight, by feeding group and overall. There were 23 full-term newborns, consisting of 12 girls and 11 boys. There were no significant differences in gestational age or birthweight among the feeding groups. At 6 months of age, median body weight in the FF infants was ~500 g higher than that in BF infants, but this difference was not significant. Table 2 shows, at the three study visits, hemoglobin (Hb), iron status, and calculated total body and compartmental iron, by feeding group and overall. In all three groups, Hb concentrations were lower at 3 and 6 months compared to day 2 (P < 0.05 for all). At 6 months, Hb concentrations were greater in the FF group compared to the BF (P < 0.01) and MF groups (P < 0.05). At 6 months, four infants were anemic: three in the BF group and one in the MF group. However, none of these four anemic infants had an abnormal plasma ferritin (PF) or soluble transferrin receptor (TfR). Reflecting their higher body weight (Table 1) and greater Hb concentrations, FF infants had greater total body iron and circulating iron than BF (P < 0.01) or MF infants (P < 0.05) at 6 months. Table 3 shows, at the three study visits, inflammation biomarkers, by feeding group and overall.

Table 1. Gender, gestational age and birthweight, and postnatal age and weight at the three study visits, in fully breastfed infants, fully formula-fed infants, and infants on mixed feeding with breast milk and formula, from birth to 6 months of age.

Data are expressed as medians (IQR). Compared by one-way ANOVA or linear mixed model (LMM) analysis with Bonferroni correction; no significant between-group differences.

	Visit(s)	All	Breast feeding	Formula feeding	Mixed feeding	P value
						Group	Time	Group × time
n		23	8	7	8	–	–	–
Gender (female/male)		12/11	3/5	6/1	3/5	–	–	–
Gestational age, weeks	Birth	39.4 (38.2–40.1)	39.5 (39.2–40.0)	39.2 (37.0–39.9)	39.0 (38.2–40.4)	–	–	–
Birthweight, kg	Birth	3.24 (3.06–3.47)	3.04 (3.01–3.36)	3.30 (3.20–3.66)	3.28 (3.06–3.35)	–	–	–
Age, days	2 days	2.1 (2.1–3.5)	3.0 (2.1–3.3)	2.1 (1.5–6.4)	2.1 (1.8–2.4)	0.4855	<0.001	0.5848
	3 months	105 (100–118)	106 (103–117)	105 (100–118)	104 (98–121)
	6 months	187 (181–187)	203 (191–213)	187 (181–187)	188 (183–204)
Weight, kg	2 days	3.07 (2.91–3.29)	2.91 (2.83–3.15)	3.25 (3.03–3.46)	3.08 (2.87–3.12)	0.0187	<0.001	0.6410
	3 months	6.20 (5.70–6.68)	6.35 (5.92–6.61)	6.10 (5.90–6.50)	6.00 (5.65–6.95)
	6 months	7.80 (7.33–8.26)	7.45 (6.99–8.40)	8.10 (7.65–8.45)	7.70 (7.55–8.00)

Table 2. Hemoglobin, plasma ferritin, soluble transferrin receptor, and calculated total body iron and compartmental iron (circulating, tissue, and stores) in fully breastfed infants, fully formula-fed infants, and infants on mixed feeding with breast milk and formula, from birth to 6 months of age.

Data are expressed as medians (IQR). Linear mixed model (LMM) analysis with Bonferroni correction, with gender and age as covariates; if group × time interaction was significant, post hoc tests were performed.

	Postnatal visit	All	Breast feeding	Formula feeding	Mixed feeding	P value
						Group	Time	Group × time
n		23	8	7	8
Hemoglobin, g/dl	2 days	18.5 (16.5–19.5)	19.6 (18.6–21.5) *,†	18.5 (16.1–19.3)*,‡	17.1 (16.1–18.7) *,§	0.0851	0.0147	0.0094
	3 months	11.8 (11.3–12.4)	12.3 (11.6–13.2)	11.7 (10.7–12.1)	11.6 (11.3–12.0)
	6 months	12.0 (11.1–13.4)	11.3 (10.8–12.0) ¶	13.4 (13.1–13.7) #	11.2 (11.0–12.1)
Plasma ferritin, μg/liter	2 days	202 (183–209)	203 (188–211)	202 (178–206)	200 (180–2010)	0.0969	0.0013	0.5551
	3 months	121 (90–162)	160 (119–162)	106 (49–145)	105 (95–158)
	6 months	73 (44–113)	92 (63–112)	45 (39–95)	80 (48–117)
Soluble transferrin receptor, mg/liter	2 days	7.2 (5.6–9.8)	8.1 (7.5–11.2)	8.5 (6.0–11.7)	6.1 (5.3–6.9)	0.3088	0.1033	0.0198
	3 months	5.4 (5.0–5.8)	5.3 (5.0–5.8) ‡	5.6 (5.1–5.7)	5.4 (5.0–6.4)
	6 months	5.6 (4.9–6.5)	6.1 (5.5–6.7)	5.6 (4.9–5.7)	5.2 (4.6–6.4)
Fe total, mg	2 days	231 (218–247)	218 (212–237) **,§	244 (227–259) §	231 (216–234) §	0.0007	<0.0001	0.0031
	3 months	293 (277–321)	295 (293–335) ‡	290 (283–303) §	282 (270–329) ‡
	6 months	358 (340–392)	345 (309–364) ¶	409 (387–431) #	344 (337–355)
Fe circulating, mg	3 months	210 (192–232)	208 (202–247) ‡	219 (203–223) †	203 (181–234) ‡	0.0011	0.9545	<0.0001
	6 months	263 (241–301)	247 (223–286) ¶	327 (295–345) #	249 (239–263)
Fe tissue, mg	3 months	37 (34–40)	38 (36–40) †	37 (35–39) †	36 (34–42) §	0.0170	0.7794	0.0055
	6 months	47 (44–50)	45 (42–50)	49 (46–51)	46 (45–48)
Fe stores, mg	3 months	51 (45–55)	52 (50–61)	50 (29–55)	49 (45–52)	0.1129	0.3706	0.4894
	6 months	50 (36–56)	50 (48–55)	36 (30–54)	55 (44–58)

For between-group post hoc tests: ¶different from formula feeding at 6 months (P < 0.01); #different from mixed feeding at 6 months (P < 0.05). For within group post hoc tests: **different from 3 months (P < 0.05); *different from 3 months (P < 0.01); ‡different from 6 months (P < 0.05); §different from 6 months (P < 0.01); †different from 6 months (P < 0.001).

Table 3. Inflammation biomarkers in fully breastfed infants, fully formula-fed infants, and infants on mixed feeding with breast milk and formula, from birth to 6 months of age.

Data are expressed as medians (IQR). Linear mixed model (LMM) analysis with Bonferroni correction, with gender and age as covariates.

	Visit(s)	All	Breast feeding	Formula feeding	Mixed feeding	P value
						Group	Time	Group × time
C-reactive protein, mg/liter	2 days	1.03 (0.40–1.61)	1.32 (0.84–1.91)	0.96 (0.35–2.55)	0.82 (0.47–1.12)	0.9129	0.6687	0.9868
	3 months	0.31 (0.20–0.75)	0.32 (0.19–1.18)	0.22 (0.17–0.80)	0.32 (0.25–0.74)
	6 months	0.28 (0.15–1.04)	0.24 (0.14–5.25)	0.43 (0.21–0.71)	0.24(0.15–0.92)
α-1-glycoprotein, g/liter	2 days	0.18 (0.15–0.21)	0.18 (0.14–0.23)	0.19 (0.16–0.21)	0.17 (0.14–0.23)	0.9756	0.6131	0.6819
	3 months	0.38 (0.28–0.53)	0.27 (0.26–0.51)	0.42 (0.37–0.48)	0.45 (0.33–0.54)
	6 months	0.45 (0.34–0.73)	0.47 (0.32–0.82)	0.47 (0.43–0.57)	0.39 (0.33–0.71)

For the 23 women whose infants were included in this study, median [interquartile range (IQR)] fractional iron absorption from the 12 mg of ⁵⁷Fe or ⁵⁸Fe added to the labeled test meals consumed during pregnancy was 7.81% (4.31%, 13.62%) (14). That the isotopes had uniformly labeled newborn body iron is indicated by (i) the high correlation between enrichment of the ⁵⁷Fe and ⁵⁸Fe tracers in maternal erythrocytes at delivery and in newborn erythrocytes at the first postnatal visit [r_s = 0.998 (P < 0.001) and r_s = 0.990 (P < 0.001), respectively]; (ii) the high correlation between k_abs in the 2-day to 3-month period and in the 3- to 6-month period (r_s = 0.847); and (iii) the high correlation coefficient for the regression of k_abs (mean R² = 0.974) measured over the study period. As shown in Table 4, isotope administration to the pregnant women resulted in a distinct change in the iron isotope composition of newborn. Compared to naturally occurring ratios of ⁵⁶Fe/⁵⁷Fe and ⁵⁶Fe/⁵⁸Fe of 43.35 and 325.4, respectively, newborn erythrocytes had enriched ratios that gradually increased over the subsequent 6 months as the infants absorbed dietary iron with a natural isotopic ratio. Accordingly, the absorption of dietary iron was reflected in the decreasing abundance (concentration, k_abs) of the tracers in blood over the 6 months. The rate of change of tracer concentration constant, k_abs, indicates the fraction of total body iron absorbed per unit of time. Median (IQR) k_abs over the 6-month study in all infants was −0.00112 (−0.00160,−0.00051) and differed by feeding group: in BF infants, it was −0.00046 (−0.00055,−0.00037); in FF infants, it was −0.00157 (−0.00191,−0.00122); and in MF infants, it was −0.00133 (−0.00170,−0.00100) (BF versus FF: P < 0.05; BF versus MF: P < 0.05). The significantly steeper k_abs in the FF and MF groups compared to the BF group reflected greater dilution of enriched body iron by greater amounts of absorbed iron.

Table 4. Blood isotopic ratios and tracer concentrations in infants (n = 23) at each postnatal study visit.

Data are expressed as medians (IQR).

Postnatal age	2 days	3 months	6 months
n	23	23	23
56Fe/57Fe isotopic ratio in blood	42.88 (42.40–43.10) *	42.89 (42.47–43.10) *	43.00 (42.53–43.10)
56Fe/58Fe isotopic ratio in blood	253.71 (237.07–277.96) †,‡	260.72 (243.09–282.56) †	264.71 (247.96–286.25)
Concentration of tracer in blood [‰ = mmol/mol]	0.853 (0.705–1.110) †,‡	0.756 (0.558–1.042) †,§	0.677 (0.459–0.964)

By repeated-measures ANOVA: *different from 6 months (P < 0.01); †different from 6 months (P < 0.0001); ‡different from 3 months (P < 0.0001); §different from 6 months (P < 0.001).

Table 5 shows daily iron intakes from breast milk and formula, and daily amounts of iron absorbed, lost, and gained from birth to 6 months, by feeding group. Overall, the median (IQR) amounts of iron absorbed, lost, and gained were as follows: 0.333 (0.151–0.457), 0.111 (0.105–0.118), and 0.218 (0.041–0.349), mg/day, respectively, but there were marked differences among the feeding groups. FF infants had median total iron intakes that were ~40-fold higher than BF infants (P < 0.001) and 2.7-fold higher than MF infants (P < 0.05). FF infants absorbed 3.6-fold higher amounts of iron than BF infants (median, 0.457 versus 0.128 mg/day; P < 0.01). Estimated iron bioavailability (that is, the fraction of the available iron absorbed) from breast milk was 13-fold higher than from formula (median, 42.3% versus 3.2%; P < 0.001). There was a large range in iron absorption among the infants, ranging from 0.022 mg/day in one BF infant to 1.188 mg/day in one FF infant. Notably, in BF infants, the median amount of iron absorbed just covered the median amount lost, resulting in minimal gains (Fig. 2). In contrast, FF infants and MF infants had iron gains of 0.349 and 0.276 mg/day, respectively (versus BF infants, P < 0.001 and P < 0.05, respectively). With the exception of three FF infants, who absorbed 0.690, 0.978, and 1.188 mg/day, iron absorbed in the remaining 20 infants was well below the recommended requirement for absorbed iron (~0.7 mg/day) (15) for 6-month-olds.

Table 5. Daily iron intake, iron absorbed, iron lost, and iron gained from birth to 6 months in fully breastfed infants, fully formula-fed infants, and infants on mixed feeding with breast milk and formula.

Data are expressed as medians (IQR). Iron intake from breast milk estimated using consensus values for mean iron concentration (0.35 mg/liter) and milk intake (780 ml/day) for the first 6 months of lactation (15). Iron bioavailability calculated as the fraction of iron intake that was absorbed.

	All	Breast feeding	Formula feeding	Mixed feeding
n	23	8	7	8
Iron intake from breast milk, mg/day	0.14 (0–0.27)	0.27	0	0.13 (0.08–0.20)
Iron intake from formula, mg/day	2.89 (0–10.46)	0	11.19 (10.46–15.55)*	3.97 (2.14–6.86)
Iron intake, mg/day	3.08 (0.27–10.46)	0.27*,†	11.19 (10.46–15.55)‡	4.13 (2.33–6.95)
Iron absorbed, mg/day	0.333 (0.151–0.457)	0.128 (0.095–0.180)§	0.457 (0.374–0.617)	0.391 (0.283–0.473)
Iron bioavailability, %	10.4 (5.5–24.9)	42.3 (20.4–52.7)¶,‡	3.2 (2.5–7.4)	7.3 (6.4–11.3)
Iron loss, mg/day	0.111 (0.105–0.118)	0.108 (0.102–0.116)	0.118 (0.109–0.120)	0.110 (0.106–0.116)
Iron gain, mg/day	0.218 (0.041–0.349)	0.027 (−0.002–0.055)*	0.349 (0.260–0.498)	0.276 (0.175–0.368)

By independent t test (intake from formula) or one-way ANOVA (all others): ‡different from mixed feeding (P < 0.01); *different from formula feeding (P < 0.0001); †different from mixed feeding (P < 0.05); §different from formula feeding (P < 0.01); ¶different from formula feeding (P < 0.001).

Fig. 2. Daily iron absorbed, iron lost, and iron gained (mg/day) from birth to 6 months in Swiss and Mexican infants (n = 23).

(A) All infants (n = 23); (B) infants fed breast milk (n = 8); (C) infants fed formula (n = 7); or (D) infants fed mixed breast milk and formula (n = 8). Formula-fed infants absorbed 3.5-fold more iron and had 13-fold higher iron gains compared to breastfed infants (one-way ANOVA, for both, P < 0.01).

To identify predictors of iron absorption, we performed linear regression analysis on k_abs, the fraction of total body iron absorbed over the 6-month study. Gender and infant weight gain were not significant predictors of k_abs. Feeding practice (coded with increasing numbers from BF to FF) was the strongest predictor of k_abs (β = −0.6396, P = 0.0139), indicating that higher iron absorption (as reflected in a more negative k_abs) was associated with FF. Maternal Hb (data not shown) measured at the 36th week of pregnancy was a significant positive predictor (β = 0.4359, P = 0.0411), indicating that higher iron absorption (as reflected in a more negative k_abs) was associated with lower maternal Hb concentration. These variables explained 59.0% of the variability in k_abs.

DISCUSSION

This study uses isotope dilution to directly measure iron absorption from birth to 6 months and provides long-term estimates of iron absorption and gains during this critical period of iron nutrition. We show that FF infants (receiving high-iron formula) had median dietary iron intakes ~40-fold higher than BF infants, but iron bioavailability from breast milk was >10-fold higher than from formula. As a result, FF infants absorbed nearly fourfold higher amounts of dietary iron than BF infants and significantly more than MF infants. In FF and MF infants, the amount of iron absorbed exceeded the amount lost resulting in iron gains of 0.28 to 0.35 mg/day. In BF infants, the amount of iron absorbed just covered basal iron losses, but resulted in minimal gains, resulting in greater depletion of iron stores. Except for three FF infants, absorbed iron was well below the recommended absorbed iron requirement of ~0.7 mg/day for 6-month-olds.

Previous estimates of iron absorption from breast milk vary and depend on the method used and the age of the infants studied. Early radioisotope studies used ⁵⁹Fe and whole-body counting to measure iron absorption (16–19). Götze et al. (17) reported ⁵⁹Fe retention given with breast milk was 18% during the first month, 30% from 1 to 3 months, and 37% from 3 to 6 months. In 6-month-old infants, Saarinen et al. (19) reported that iron retention from ⁵⁹Fe given with breast milk was 49%, but the potential lack of equilibration of the extrinsic tracer in breast milk iron may have favored high absorption. Saarinen and Siimes (20) and Garry et al. (21) estimated that iron absorption from breast milk was as high as 80% using an indirect method based on calculated changes in total body iron over time but included calculation errors (e.g., breast milk iron content was assumed to be 1 mg/liter). Several early studies measured iron absorption from breast milk in adults and reported fractional absorption of 15 to 48% (22–24), but it is uncertain whether these values can be extrapolated to infants.

In contrast to radioisotope studies, stable isotope studies using an extrinsic tag report lower fractional iron absorption from breast milk, ranging from 7 to 20% (25–28). In BF Peruvian infants at 2 to 3 months and 5 to 6 months of age, mean iron absorption from breast milk was 7.1 and 13.9%, respectively (28). In BF infants (n = 8) at a mean age of 5 months who were fed 700 to 1000 g of ⁵⁸Fe-labeled breast milk over several feeds, iron absorption was 11.8% (25). In 6-month-old BF Swedish (n = 25) and 5- to 7-month-old U.S. infants (n = 14), mean iron absorption from breast milk was 16.4 and 20.7%, respectively (26, 27). Several factors may have contributed to the lower absorption values in these stable isotope studies. Compared to radioisotopes, a much higher amount of stable iron isotope needs to be added to breast milk, which has very low native iron content. This increases the iron content of the labeled breast milk up to 10-fold [e.g., (25)], and fractional iron absorption decreases with increasing dose (29). Also, most stable isotope studies assumed 80 to 90% incorporation of iron into Hb (30) but measured incorporation is <80% in infants (31, 32) and use of higher values likely underestimated iron absorption. Our estimate of ~40% iron bioavailability (Table 3) from breast milk using long-term isotope dilution is within the range (18 to 49%) reported using radioisotopes and whole-body counting (17, 19).

Formulas contain considerably more iron than breast milk, and it is generally assumed that despite lower fractional absorption, they provide higher amounts of absorbed iron than breast milk (15). Our data (Table 3) support this assumption. The bioavailability of iron in formula likely varies (15) but is usually estimated to be 10% by expert groups making intake recommendations (15, 33). However, iron absorption from formula directly measured using isotopic methods is generally lower than 10%. Absorption from low-iron infant formula given in multiple feeds over 24 hours to 4- to 5-month-old infants (n = 10) was 8% (34). Rios et al. (35) reported 4% absorption from iron-fortified formulas containing 1.2 mg of iron/100 ml and 6% absorption from formula containing 0.6 mg of iron/100 ml (18). Saarinen and Siimes (18) reported that 11- to 13-month-old infants absorb <10% of iron from formula, regardless of its iron concentration. Heinrich et al. (36) reported a mean iron absorption of 4% from a 5-mg dose of iron given with 50 ml of cow milk. In our study, all the FF infants were consuming formula containing 1.2 mg/100 ml, and fractional iron absorption was only 3.2% (Table 3) despite the formula being fortified with iron as ferrous sulfate together with ascorbic acid. Iron bioavailability might have been expected to be higher, considering that addition of 100 mg of ascorbic acid can double iron absorption from formula (37). Our data support efforts to design improved formulas with higher bioavailability, e.g., through addition of prebiotics (38).

Here, most infants absorbed dietary iron in amounts that fell well below the estimated requirements for their age, requiring the utilization of iron from their birth stores to fill the gap. Absorbed iron in a nonpregnant adult only needs to replace basal iron losses, whereas infants must also expand their body iron to meet the demands of growth. Our healthy full-term newborns had a substantial iron endowment at birth, reflected in high SF and Hb concentrations (Table 2). After birth, there are rapid shifts in iron compartments, as fetal Hb concentration falls and adult Hb formation begins (13). To support blood volume expansion and somatic growth, iron is recycled from fetal Hb and is mobilized from stores, decreasing SF (Table 2). Our estimates of infant body iron and compartmental iron are similar to previous estimates in infancy (39, 40). Expert estimates of the mean or median requirement for absorbed iron (mg/day) at 6 to 7 months are as follows: 0.69 mg/day (15), 0.72 mg/day (41), and 0.79 mg/day (33). Median absorbed iron in our BF, FF, and MF infants (Table 3) was only 19, 66, and 57%, respectively, of the lowest estimate, 0.69 mg/day (15). Our isotopic findings are consistent with those of Abrams et al. (42) who analyzed dietary records in older, 6- to 12-month-old U.S. infants and found that estimated daily iron absorption was below the recommended amount in 54% of infants. Because the infants in the FF and MF groups were growing normally and were not anemic or iron deficient at 6 months, our findings suggest that about one-half to one-third of their iron requirement from birth to 6 months was successfully met from iron stores present at birth. Fully BF infants needed to draw substantially more iron from body iron stores (~0.56 mg/day) to meet their iron requirement, and three of the eight BF infants in our study were unable to do this and were anemic at 6 months. Although these anemic infants did not have abnormal PF of TfR values, defining iron status in early infancy is challenging and cutoffs for iron biomarkers are less well-established than for older age groups. Therefore, we believe that the anemia was most likely caused by mild ID that was not detected by iron biomarkers. Our data, showing the low amount of absorbed iron from breast milk from birth to 6 months, support the assumption that the depletion of birth iron stores is physiologically normal (15, 43). Our findings support the recommendation to introduce iron-rich complementary foods between 4 and 6 months of age to avoid exhaustion of birth iron stores and the development of ID.

Our study has several strengths. We used maternal-fetal iron transfer of a stable isotope to label the newborn. We studied healthy term infants with normal birth weight in two centers using standardized methods. The isotope dilution method is minimally invasive (requiring only three 2-ml heel-stick blood samples) and safe (no radioactivity) but provides long-term assessment of iron homeostasis from birth to 6 months, rather than “point’ estimates from single absorption studies. It has additional advantages compared to radio- and stable isotopic methods previously used in infancy, as described above. We included infants who were fully BF or fully FF from birth to 6 months, allowing comparisons of these two feeding methods. We performed dietary assessment of infant feeding at three time points, allowing us to calculate iron intakes and bioavailability. Our study also has limitations. Our small sample size may not cover possible wide variations in physiological needs for iron in early infancy. We did not measure breast milk intake or breast milk iron concentrations but used well-established consensus values. We did not measure the iron isotopic composition of breast milk, which would have required large volumes of breast milk due to its very low iron concentration; thus, we could not estimate additional ⁵⁷Fe or ⁵⁸Fe provided in breast milk. We could not measure k_loss isotopically because of rapid shifts in circulating and storage iron (and tracer) characteristic of early infancy, but assumed losses per unit body weight based on data using a similar method in toddlers (12). Our calculations of total body iron included several assumptions derived from studies in older children.

The concept of a measurement of tracer concentration after tracer administration has commonly been used in the determination of fractional absorption. Here, we apply an inversion of this concept, with iron of natural composition acting as a tracer diluting an ad hoc modified isotopic signature obtained via stable isotope administration and equilibration with body iron. Future studies using this method in infancy could (i) provide more precise estimates of iron requirements; (ii) be applied in low-birthweight infants to estimate their iron requirement; (iii) inform development of better-absorbed infant formula; and (iv) describe the time course of depletion of birth iron stores with different feeding regimens and to inform recommendations for the timing of introduction of iron-rich complementary foods (44).

METHODS

This two-center study had a prospective, observational design. The infants in this study were born to women from the prenatal clinics of the University Hospital Zurich, Switzerland and the Hospital Regional de Alta Especialidad Materno Infantil in Monterrey, Mexico who participated in our previous study of iron kinetics in pregnancy (14). At pregnancy week 20 ± 2 and 30 ± 2, women consumed test meals labeled with 12 mg of the stable iron isotopes ⁵⁷Fe or ⁵⁸Fe as ferrous sulfate under standardized conditions and close supervision (14). Written informed consent was obtained from all women. The protocols for the original pregnancy study and for this infant study were approved by the ethics committees of the Canton of Zurich and ETH Zurich, Switzerland and the Hospital Regional de Alta Especialidad Materno Infantil in Monterrey, Mexico (HRMI:238/2017). This trial was registered at clinicaltrials.gov as NCT02747316.

Seventy-four women completed the original pregnancy study to delivery. The inclusion criteria for this study were as follows: (i) singleton term birth to a mother who previously received ⁵⁷Fe or ⁵⁸Fe during pregnancy; (ii) parental informed written consent; (iii) birth weight ≥ 2500 g); (iv) considered healthy by their physicians and the investigators; and (v) able to present for the first postnatal study visit at ~2 days after birth. Thirty-two mother-infant pairs were enrolled. There were no restrictions regarding infant feeding practice. Infants were studied at three postnatal visits, at ~2 days and at 3 and 6 months of age (study design, Fig. 1). Carefully trained staff measured infant body weight using an infant digital scale and collected a pendant capillary blood sample (~2 ml) using a disposable spring-loaded lancet. At each visit, mothers answered a standardized infant health questionnaire and a dietary questionnaire. The dietary questionnaire assessed the quantity and quality of the formula, food, and/or supplements that were being consumed by the infant.

Dietary assessment and estimation of iron intakes

At each study visit, the women reported whether they were breast feeding, feeding formula, or feeding both breast milk and formula, or feeding solid foods. If feeding formula, the specific formula type, form (ready-to-feed liquid, concentrated liquid, or powder), and quantity of formula (frequency and volume of feedings provided to the infant) over the previous week were recorded. Iron contents of the formulas were obtained from manufacturer websites. Daily iron intakes from formula were calculated based on these data. We did not quantify breast milk intake or measure breast milk iron concentration; intake of iron from breast milk was estimated using consensus values for mean iron concentration (0.35 mg/liter) and milk intake (780 ml/day) for the first 6 months of lactation (15). For infants in the mixed feeding group who consumed both breast milk and formula, the volume of formula given to the infant each day was used to calculate iron intake from formula. This volume was then subtracted from 780 ml to estimate the quantity of breast milk consumed each day and the corresponding iron intake from breast milk (45).

Laboratory analysis

On the day of blood collection, we measured Hb by using an automated hematology analyzer and separated and froze whole blood and serum aliquots. Samples from Mexico were transferred on dry ice to ETH Zurich. We measured TfR, PF, high-sensitivity C-reactive protein (CRP), and α-1 glycoprotein (AGP) by using a multiplex immunoassay (46). We defined ID as either (i) PF <30 μg/liter at 3 months (47) and <12 μg/liter at 6 months of age (48), or (ii) TfR >11 mg/liter for all ages (49). We defined anemia as Hb <8.9 g/100 ml at 3 months (50) and <11 g/100 ml at 6 months (48). PF was not adjusted for inflammation because there were very few elevated CRP and AGP concentrations and we were uncertain if these measures should be used to adjust SF during early infancy. Isotopic composition in erythrocytes was determined in duplicate under chemical blank monitoring by using inductively coupled plasma mass spectrometry (NEPTUNE, Thermo Finnigan, Fisher Scientific, Hampton, USA) at ETH Zürich as described previously (51).

Calculation of body iron

Widdowson and Spray (51) performed autopsy studies on term infants and directly measured total body iron; they reported mean values of 74 to 75 μg/g body weight in term infants weighing 2500 to 3500 g. Similar values for total body iron at birth, 75 μg/g body weight, were estimated by Oski (40). Thus, we calculated total body iron (Fe_total) at 2 days of age in our infants, all with birthweight ≥2500 g, as

$${\text{Fe}}_{\text{total}}=75 \mathrm{\mu }\mathrm{g} \text{iron}/\mathrm{g}\times \text{birthweight} (\mathrm{g})$$

At 3 and 6 months of age, Fe_total was considered to have three compartments (15) and was calculated as the sum of circulating iron, tissue iron, and storage iron according to the equation

$${\text{Fe}}_{\text{total}}={\text{Fe}}_{\text{circulating}}+{\text{Fe}}_{\text{tissue}}+{\text{Fe}}_{\text{store}}$$

Fe_circulating reflects iron in erythrocytes and disregards small amounts of iron in circulating enzymes and proteins (12). Blood volume in 1- to 6-month-old infants is estimated to be 84 ml/kg (53); we used the value of 84 ml/kg at 3 and 6 months. Fe_circulating is then calculated according to the equation

Fe_circulating = Hb concentration in g/liter × 3.47 [Hb iron concentration (mg/g)] × (0.84 × body weight).

Fe_tissue is active tissue iron in myoglobin and iron-containing enzymes. In adults and children, noncirculating active iron is estimated to be 6 to 7 mg/kg body weight (40, 54). A value of 6 mg/kg body weight was used by Fomon et al. (12) for toddlers, and we assumed this value for infancy.

Fe_store reflects storage iron consisting mainly of iron in liver ferritin. Storage iron was calculated from PF on the assumption that, after correction for body size, PF bears the same relation to storage iron in 3- and 6-month-old infants as in adults. This assumption is supported by recent data showing that erythrocyte incorporation of an isotopic tracer in infants is inversely correlated with PF in a relationship very similar to adults (10). Thus, Fe_store was calculated according to the equation

$$\begin{array}{c}{\text{Fe}}_{\text{store}}=\{9380\times \text{log}[\text{PF} (\mathrm{\mu }\mathrm{g}/\text{liter})]–11,260\}/1000\times \text{body weight}\hfill \end{array}$$(55)

Calculation of iron absorbed, lost, and gained, and iron bioavailability

Absorption of ⁵⁷Fe or ⁵⁸Fe and incorporation into new erythrocytes changes the iron isotope composition of blood. Isotope concentration in blood then sharply decreases after an initial enrichment shift due to recycling of iron from senescent erythrocytes into other tissues. After approximately 12 months in adults (56, 57) and 7 months in toddlers (12), complete isotopic equilibration of all body iron is established, indicated by a linear slope of the natural logarithm of the tracer concentration in blood plotted against time. From this point onward, the tracer concentration, e.g., ⁵⁷Fe, can only decrease due to iron with natural isotopic composition entering the body, and the decrease in the concentration of ⁵⁷Fe tracer in circulation (expressed as the slope of [⁵⁷Fe] over time) is proportional to iron absorption. The factor describing this relationship is the slope of the logarithmical ⁵⁷Fe concentration plotted over time (k_abs), the fraction of total body iron absorbed per unit of time.

Isotopic calculations

Stable isotope concentration is determined based on the isotope composition (R) determined in a whole blood sample. Stable isotope tracers are not monoisotopic, and the measured isotopic ratio in the analyzed blood sample R_57/56 can be expressed as follows

$$\begin{array}{c}{R}_{57/56}=\frac{a_{\text{nat}}^{57}\times n_{\text{nat}}+a_A^{57}\times n_A}{a_{\text{nat}}^{56}\times n_{\text{nat}}+a_A^{56}\times n_A}\hfill \end{array}$$

where $a_{\text{nat}}^{57}$ represents the concentration of the isotope ⁵⁷Fe in natural iron, n_nat is the amount in moles of natural iron, $a_A^{57}$ is the concentration of the isotope ⁵⁷Fe in the enriched tracer, and n_A is the amount in moles of tracer.

The tracer concentration ${}_a{}^A{A}^t$ in circulation at time t is represented by

$${}_a{}^A{A}^t=\raisebox{1ex}{$n_A$}\!\left/ \!\raisebox{-1ex}{$n_{\text{tot}}$}\right.=\frac{a_{\text{nat}}^{57}-a_{\text{nat}}^{56}\times R_{57/56}}{(a_{\text{tracer}}^{56}-a_{\text{nat}}^{56})\times R_{57/56}+a_{\text{nat}}^{57}-a_{\text{tracer}}^{57}}$$

where n_A is the amount in moles of tracer (here, ⁵⁷Fe) and n_tot is the amount in moles of all iron in circulation.

The stable decrease in tracer concentration from 3 days to 6 months of age reflects the rate of change of body iron composition (d⁻¹) expressed as

$$\frac{d({}_a{}^{A}A)}{\mathit{dt}}=-k_{\text{abs}}\times {}_a{}^{A}A$$

where d(^AA)/dt represents the rate of change of the tracer concentration ^AA per unit of time, and k_abs the rate of change of tracer concentration constant.

Resolution of the differential equation leads to

${}_a{}^{A}A(t)={}_a{}^{A}A_0\times e^{-k_{\text{abs}}\times t}$ or $\text{ln}({}_a{}^{A}A_t)=-k_{\text{abs}}\times t+\text{ln}({}_a{}^{A}A_0)$

The amount of iron absorbed each day (mg) between 2 days and 6 months (Fe_abs) was calculated as

$${\text{Fe}}_{\text{abs}}=k_{\text{abs}}\times {\text{Fe}}_{\text{total}}$$

where Fe_total is mean total body iron (mg) between 2 days and 6 months of age.

We intended to quantify iron losses over the study period by measuring loss of circulating tracer (k_loss), as described previously in older children (11, 12). However, there were large shifts from circulating to stored iron and then back in the first four postnatal months. As a result, we could not calculate k_loss because the log change in amount of circulating tracer over time was not described by a linear slope (in contrast to the tracer concentration). Therefore, we used previous estimates of obligatory basal losses in infancy (12, 40, 58, 59). Oski (40) estimated daily iron losses in infancy to be 20 μg kg body weight⁻¹ day⁻¹, while Fomon et al. (12), using stable isotopes similar to this study, reported basal iron losses of 22 μg kg⁻¹ day⁻¹ in toddlers. We used the estimate of 20 μg kg⁻¹ day⁻¹ (40) and calculated daily losses based on body weight measured at the three study visits. Net iron balance (Fe_gain) was then determined by subtracting iron loss from Fe_abs. Dietary iron bioavailability, the fraction of iron ingested that was absorbed, was calculated by dividing Fe_abs by daily iron intake.

Statistical analysis

To estimate the required sample size, we assumed a relevant daily difference of 30% from the median absorbed iron requirement from WHO for 6-month-old infants of 0.72 mg/day or 0.22 mg/day (41). Using stable isotopes in a similar way to measure long-term iron absorption, Fomon et al. (12) reported an SD of 0.13 mg/day in 13- to 26-month-old toddlers. Thus, we estimated that 24 participants (8 participants per group) would be needed for between-group comparisons with a type I error rate of 5 and 85% power. We were uncertain about isotopic enrichment and dilution in earlier infancy and therefore invited all mother-infant pairs who had completed our previous pregnancy study (14) to participate. We also conducted a retrospective calculation, and our effect size, calculated at f = 0.866, resulted in an actual power of 94%. We conducted the statistical analyses with SPSS (IBM SPSS statistics, Version 28). Data were presented as medians (IQR). Data were checked for normality by Shapiro-Wilk tests and by visual inspection of histograms and non-normally distributed data were logarithmically (log10) transformed for statistical analyses. For parameters assessed at multiple time points, we used a random intercept linear mixed model (LMM) analysis with Bonferroni-corrected multiple comparisons to assess the effect of group (breast feeding, formula feeding, and mixed feeding) and time (2 days, 3 months, and 6 months). We added gender and age as covariates to all models. If LMM analysis showed a significant group × time interaction, post hoc tests were performed. For parameters assessed at only one time point, we conducted a one-way analysis of variance (ANOVA) across the three groups. For parameters measured over time, we used repeated-measures ANOVA to assess changes. Linear regression analyses were used to test associations. p-values < 0.05 were considered statistically significant.