, 43:29

Soy-Based Infant Formula Supplemented with DHA and ARA Supports Growth and Increases Circulating Levels of these Fatty Acids in Infants


  • Dennis Hoffman
    • Retina Foundation of the Southwest
  • Ekhard Ziegler
    • Department of PediatricsUniversity of Iowa Hospital and Clinic
    • Mead Johnson Nutritionals
  • Cheryl L. Harris
    • Mead Johnson Nutritionals
  • Deborah A. Diersen-Schade
    • Mead Johnson Nutritionals
Original Article

DOI: 10.1007/s11745-007-3116-7

Cite this article as:
Hoffman, D., Ziegler, E., Mitmesser, S.H. et al. Lipids (2008) 43: 29. doi:10.1007/s11745-007-3116-7


Healthy term infants (n = 244) were randomized to receive: (1) control, soy-based formula without supplementation or (2) docosahexaenoic acid−arachidonic acid (DHA + ARA), soy-based formula supplemented with at least 17 mg DHA/100 kcal (from algal oil) and 34 mg ARA/100 kcal (from fungal oil) in a double-blind, parallel group trial to evaluate safety, benefits, and growth from 14 to 120 days of age. Anthropometric measurements were taken at 14, 30, 60, 90, and 120 days of age and 24-h dietary and tolerance recall were recorded at 30, 60, 90, and 120 days of age. Adverse events were recorded throughout the study. Blood samples were drawn from subsets of 25 infants in each group. Capillary column gas chromatography was used to analyze the percentages of fatty acids in red blood cell (RBC) lipids and plasma phospholipids. Compared with the control group, percentages of fatty acids such as DHA and ARA in total RBC and plasma phospholipids were significantly higher in infants in the DHA + ARA group at 120 days of age (P < 0.001). Growth rates did not differ significantly between feeding groups at any assessed time point. Supplementation did not affect the tolerance of formula or the incidence of adverse events. Feeding healthy term infants soy-based formula supplemented with DHA and ARA from single cell oil sources at concentrations similar to human milk significantly increased circulating levels of DHA and ARA when compared with the control group. Both formulas supported normal growth and were well tolerated.


Soy-based formulaTerm infantsInfant formulaDocosahexaenoic acidArachidonic acidLCPUFAInfant growth


Docosahexaenoic acid (22:6ω3; DHA) and arachidonic acid (20:4ω6; ARA) are long-chain polyunsaturated fatty acids (LCPUFAs) that accumulate rapidly in the brain from the last trimester of fetal gestation until at least 2 years of age [1, 2]. Human breast milk provides DHA and ARA for growth as well as retinal and central nervous system development and function [37]. Infants unable to breastfeed or weaned from breastfeeding now benefit from supplementation of DHA and ARA in many commercial infant formulas. Supplementation of DHA and ARA from single cell algal and fungal sources at median worldwide human milk levels [810] in both premature and term infants has resulted in red blood cell (RBC) concentrations of the LCPUFAs comparable to those of breast-fed infants [1113]. Clinical evidence of improved visual acuity and cognitive development in term infants up to 18 months of age [12, 1418] and enhanced growth in premature infants [19, 20] has resulted from inclusion of DHA and ARA at worldwide human milk levels in milk-based formulas. In addition, a follow-up study of 4-year-olds who received formula supplemented with DHA and ARA at those levels as infants, demonstrated visual and IQ maturation similar to that of breast-fed subjects [7].

Not all infants can tolerate a cow’s milk-based formula, and some parents prefer a vegetarian product. Primary clinical indications for discontinued use of cow’s milk-based formula include immunoglobulin E-mediated milk allergy (commonly associated with eczema), documented lactose intolerance following infectious diarrhea, lactase deficiency, and galactosemia [21, 22]. In such instances, soy-based formulas have been endorsed by The American Academy of Pediatrics as a safe and effective alternative [21]. An estimated 25−36% of formula-fed infants in the United States receive a soy-based infant formula at some time in the first year of life [21, 22].

To date, no published data are available for supplementation of soy-based infant formula with DHA and ARA. Given the extensive use of soy-based formulas and the developmental advantages afforded by routine supply of exogenous DHA and ARA in cow’s milk-based formulas, we felt it prudent to examine the growth, tolerance, and possible benefits of supplementation in a soy-based matrix. In this study we examined the effects of a soy-based formula supplemented with single cell oil sources of DHA and ARA at worldwide human milk levels on growth, tolerance, and the percentage of DHA and ARA (i.e., vegetarian sources) in RBCs and plasma phospholipids when fed to term infants from 14 to 120 days of age.

Experimental Procedure


Healthy term infants were recruited at 13 pediatric centers in the United States for a double-blind, randomized, controlled parallel group study. Eligible participants were 12–16 days of age, had a minimum birth weight of 2,500 g, and solely received formula at least 24 h prior to randomization. Exclusion criteria included: history of underlying disease or malformation that could interfere with growth and development; large-for-gestational-age infants whose mothers were diabetic; breastfeeding within 24 h prior to randomization; evidence of formula intolerance or poor intake at time of randomization; weight at randomization less than 98% of birth weight; enlarged liver or spleen; or plans to move outside of the study area within the study time frame (120 days).

Parents or guardians provided written informed consent prior to enrollment. The research protocol and informed consent forms adhered to the Declaration of Helsinki (including the October 1996 amendment) and were approved by the institutional review board/ethics committee of each participating institution. The study was conducted in compliance with good clinical practices.

Study Protocol

Two hundred and forty-four infants (control, n = 120; DHA + ARA, n = 124) enrolled in the study were stratified by gender and randomized to receive one of two formulas: (1) control, soy formula without supplementation (Enfamil ProSobee®, Mead Johnson & Company, Evansville, IN) or (2) DHA + ARA, soy formula supplemented with a minimum of 17 mg DHA/100 kcal from algal oil and 34 mg ARA/100 kcal from fungal oil (Enfamil ProSobee® LIPIL®, Mead Johnson & Company, Evansville, IN). Aside from the addition of DHA and ARA, the formulas were identical in all other respects. The supplemented levels of DHA and ARA were similar to reported median amounts of DHA (~0.3% by weight of fatty acids) and ARA (~0.6%) found in human milk worldwide [810].

Infants received study formulas from 14 to 120 days of age. Body weight and length, head circumference, and atopic dermatitis scores were recorded at 14, 30, 60, 90, and 120 days of age. Evidence of atopic dermatitis was determined using the SCORAD index [23] which quantifies the extent, distribution, and severity of rashes and the presence and severity of associated itching and sleep loss. Parents or guardians completed a family history of allergy and a survey of possible factors in the home and child care environments that might affect the infant’s risk of atopy, such as cigarette smoke, pets, dust, and mold. At each study visit following initial enrollment at 14 days of age, parents or care givers completed a questionnaire on dietary intake and tolerance from the previous 24 h. Adverse events were monitored throughout the study period.

Blood Sample Analysis

At five investigative sites, blood samples from a subset of infants (25 per group) were collected by venipuncture into EDTA Vacutainer tubes at 14 and 120 days of age. Only sites with experience handling clinical blood samples and appropriate sample storage facilities provided blood samples. For fatty acid analysis, plasma and RBC were separated by centrifugation and frozen. Samples were shipped on dry ice to the Retina Foundation of the Southwest (Dallas, TX) where lipids were isolated, fractionated, and analyzed by capillary column gas chromatography as previously described [12]. Fatty acid methyl esters from RBC lipids and plasma phospholipid fractions were expressed as a percentage of total fatty acids by weight. Outcomes were expressed as percentages of DHA, ARA, and other fatty acids in RBC phosphatidylethanolamine (PE) and phosphatidylcholine (PC) fractions, total RBC lipids, and plasma phospholipids. A portion of the blood sample was also used for comprehensive metabolic panel of blood chemistry analyses including glucose, cholesterol, triglycerides, electrolytes, and kidney, liver, and pancreas function markers.

Statistical Analysis

Statistical comparisons were made between control and DHA + ARA formula groups. Analyses were performed using SAS version 8 (Cary, NC). Weight gain from 14 to 120 days was the primary outcome used to establish that formulas supported adequate growth. The sample size was chosen to be able to detect a 3 g/day difference in weight gain with 80% power (α = 0.05; one-tailed). For all subsequent comparisons, P values are based on two-tailed tests. Analysis of variance was used to analyze formula intake, stool frequency and characteristics, anthropometrics (achieved growth, length, and head circumference), SCORAD grade, and blood lipid and chemistry data. Analysis of covariance was used to analyze growth rate. Race, gender, family history of allergy, home and child care environments, study discontinuation rate, and adverse events were analyzed by Fisher’s Exact test. The van Elteren test, stratified by study site, was used to analyze metabolic blood panel results.



Of the original 244 infants enrolled, three infants (control, n = 2; DHA + ARA, n = 1) did not receive study formula and were excluded from subsequent analyses. At study entry (14 days of age) infants in the control group were significantly heavier and longer than those in the DHA + ARA group (Table 1), although head circumference did not differ significantly between groups. Groups were similar for race and gender distribution, family history of allergy, and home and child care environments (data not shown). A total of 182 infants (control, n = 86; DHA + ARA, n = 96) completed the study.
Table 1

Infant characteristics at study entry




Total number of subjects



Number of male/female



Weight (g)a

3,677.8 ± 45

3,575.4 ± 44*

Length (cm)a

52.3 ± 0.2

51.8 ± 0.2*

Head circumference (cm)a

36.1 ± 0.1

35.8 ± 0.1

aMean ± standard error

* Significant difference between groups, P < 0.05

Growth and Tolerance

Growth rate, as assessed by mean daily gains of weight, length, and head circumference, did not differ statistically between groups from 14 to 120 days of age (Table 2). The mean achieved weight was significantly higher for infants in the control versus the DHA + ARA group at 30 and 90 days of age (P < 0.05; data not shown). The mean achieved weight for males (Fig. 1, upper panel) and females (lower panel) in both formula groups, however, fell between the 25th and 75th percentile of normal infant weight-for-age when plotted on the US Centers for Disease Control (CDC) standard growth chart [24]. Mean achieved length was higher for infants in the control group at all time points (P < 0.05; data not shown) and mean achieved head circumference did not differ significantly between groups at any assessed time point (data not shown).
Table 2

Mean weight, length, and head circumference growth rates from 14 to 120 days of age




Total number of subjects



Weight (g/day)a

27.8 ± 0.8

27.3 ± 0.7

Length (cm/day)a

0.10 ± 0.002

0.10 ± 0.002

Head circumference (cm/day)a

0.05 ± 0.001

0.05 ± 0.001

aMean ± standard error
Fig. 1

Mean achieved weights of male participants (upper panel) and female participants (lower panel) with CDC reference percentiles (3–97) from 14 to 120 days of age. Control formula group, diamonds; DHA + ARA formula group, squares

Formula intake, stool frequency, and stool characteristics were similar for infants in both groups (data not shown). Both formulas were well tolerated as reported by parental assessment of fussiness, diarrhea, and constipation, with the only difference between groups being a higher incidence of excessive gas in the control group than the DHA + ARA group at 60 days of age (15% vs. 5%; P = 0.026). Minimal incidence of atopic dermatitis, as assessed by mean SCORAD indices, was noted for infants in each study group at all time points. For example, at 120 days of age mean SCORAD values were 2.9 ± 0.76 for the control group (n = 85) versus 2.3 ± 0.72 for the DHA + ARA group (n = 93, P = 0.374). Given the SCORAD index range of 0–103 with disease severity indicated by a higher score, mean values from this study signified a very low occurrence of atopic dermatitis.

There were no significant differences between groups for total discontinuation rates (control, n = 32 [27%] vs. DHA + ARA, n = 27 [22%]; P = 0.37) or for discontinuation due to feeding-related issues (control, n = 23 [19%] vs. DHA + ARA, n = 19 [15%]; P = 0.615). The most common reasons for discontinuation were diarrhea (control, DHA + ARA: n = 8, n = 5), vomiting (n = 8, n = 4), and fussiness (n = 6, n = 6). Significant differences between groups in adverse events included gastroesophageal reflux (control, n = 13 vs. DHA + ARA, n = 3; P = 0.009) and incidence of metabolic or nutritional difficulties (control: weight loss, n = 3; poor weight gain, n = 2; Type 1 Glutaric Acidemia, n = 1 vs. DHA + ARA: n = 0 for any category; P = 0.013). Serious adverse events were reported for 12 infants (control, n = 6; DHA + ARA, n = 6). In each of these cases, the investigator stated that the serious adverse event was unrelated to the study product.

Blood Lipids and Blood Chemistry

No significant differences in mean fatty acid percentages were detected between groups for total RBC lipids or plasma phospholipids at 14 days of age in a subset of infants (control, n = 25; DHA + ARA, n = 22; data not shown). Results from infants at 120 days of age are reported in Table 3. In the ω3 fatty acid family, the percentage of DHA in total fatty acids was significantly higher in total RBC lipids from infants in the DHA + ARA group versus the control (6.23 vs. 2.47%, P < 0.001; Table 3). DHA was also significantly higher in plasma phospholipids in infants from the DHA + ARA group (5.04 vs. 1.67%, P < 0.001). Percentages of α-linolenic acid (18:3ω3; ALA) and docosapentaenoic acid (22:5ω3) in total RBC lipids (P = 0.019 and P < 0.001, respectively) and plasma phospholipids (P < 0.001 for both fatty acids) were significantly lower in the DHA + ARA group.
Table 3

Fatty acid levels (percent of total fatty acids) in infant blood lipid fractions at 120 days of age

Fatty acid

Total RBC

Plasma phospholipids





n-3 Fatty acids


0.23 ± 0.02

0.18 ± 0.01***

0.19 ± 0.01

0.15 ± 0.01*


0.10 ± 0.01

0.10 ± 0.01

0.03 ± 0.01

0.04 ± 0.01


0.01 ± 0.00

0.00 ± 0.00

0.02 ± 0.00

0.02 ± 0.00


0.79 ± 0.04

0.75 ± 0.04

0.40 ± 0.14

0.68 ± 0.13


1.12 ± 0.05

0.66 ± 0.04*

0.44 ± 0.02

0.21 ± 0.02*


2.47 ± 0.18

6.23 ± 0.16*

1.67 ± 0.14

5.04 ± 0.13*

n-6 Fatty acids


15.28 ± 0.42

12.58 ± 0.39*

27.48 ± 0.52

21.49 ± 0.48*


0.10 ± 0.01

0.07 ± 0.01**

0.12 ± 0.01

0.08 ± 0.01*


0.45 ± 0.01

0.39 ± 0.01*

0.40 ± 0.01

0.36 ± 0.01***


1.97 ± 0.08

1.26 ± 0.07*

2.50 ± 0.10

1.62 ± 0.10*


13.97 ± 0.33

16.27 ± 0.30*

7.33 ± 0.32

13.46 ± 0.29*


0.14 ± 0.02

0.11 ± 0.02

0.04 ± 0.00

0.04 ± 0.00


3.88 ± 0.12

3.61 ± 0.11

0.45 ± 0.01

0.43 ± 0.01


1.23 ± 0.5

0.75 ± 0.05*

1.14 ± 0.06

0.66 ± 0.06*

Least square mean ± SE Control group, n = 15; DHA + ARA group, n = 16

* Significant difference between groups, P ≤ 0.001

** Significant difference between groups, P = 0.01

*** Significant difference between groups, P = 0.019

In the ω6 fatty acid family, ARA as a percentage of total fatty acids was significantly higher in total RBC lipids from infants in the DHA + ARA group versus the control (16.27 vs. 13.97%, P < 0.001). ARA was also significantly higher in plasma phospholipids in infants in the DHA + ARA group (13.46 vs.7.33%, P < 0.001). Mean percentages of linoleic acid (18:2ω6), γ-linolenic acid (18:3ω6), eicosadienoic acid (20:2ω6), dihomo-γ-linolenic acid (20:3ω6), and docosapentaenoic acid (22:5ω6) were significantly lower in total RBC lipids and plasma phospholipids from infants in the DHA + ARA group as compared with the control (see individual P values in Table 3). Fatty acid profiles of RBC PE and PC fractions were not reportable due to technical problems during analysis. There were no statistical differences in blood chemistry profiles between groups at 14 or 120 days of age (data not shown).


The present study showed that a soy-based infant formula supplemented with DHA and ARA supported growth rates and tolerance of term infants as well as commercially available soy-based formula without these fatty acids, and increased DHA and ARA status similar to milk-based formulas with similar levels of added DHA and ARA. The American Academy of Pediatrics Task Force on Clinical Testing of Infant Formulas [25] stated that rate of weight gain (g/day) is the most beneficial component of the clinical evaluation of infant formula and recommended that a difference in weight gain of more than 3 g/day over 3–4 months be considered nutritionally significant. Our study found no differences in rate of weight gain over the 4-month study. Daily gains in length and head circumference (cm/day) also did not differ between the formula groups. In addition, weights in both groups, when compared with CDC infant reference data [24], fell between the 25th and 75th percentiles of both male and female weight-for-age growth curves. Although achieved weight means were significantly higher at 30 and 90 days of age and achieved length means were higher in the control group at all time points, these differences were consistent with those upon study entry, when the control group was significantly heavier and longer on average than the DHA + ARA group.

Questions have been raised as to how some other components of soy formulas (e.g., phytoestrogens) affect growth and long-term health [21, 22]. Previous clinical reports, however, have demonstrated that soy-based formulas promoted normal growth in term infants [21, 22, 2628]. In addition, a study of adults found that long-term growth, health, and reproductive outcomes are comparable between individuals who received soy- or cow’s milk-based formula as infants [29]. The American Academy of Pediatrics has concluded that soy based formulas are safe and effective feedings for infants whose nutritional needs are not met by human milk or cow’s milk formula [21]. Our study adds further support to this conclusion and demonstrates that the addition of DHA and ARA at levels similar to those found in worldwide breast milk does not negatively influence growth [29].

Both the marketed control and the DHA + ARA supplemented soy-based formulas were well tolerated by infants in this study as evidenced by similarities in parental assessment of tolerance, formula intake, stool frequency and characteristics, and low occurrence of atopic dermatitis throughout the study. Serum chemistry values for a wide variety of metabolic parameters were similar for infants in both groups at 14 and 120 days of age. Rates of study discontinuation and adverse events were similar between formula groups and incidence of serious adverse events was low and deemed unrelated to the study products by participating physicians. Previous clinical evidence also indicates infant tolerance of soy-based formulas, with favorable comparison to cow’s milk-based formula [22, 30, 31]. Our study demonstrated that the soy-based control formula and soy-based formula supplemented with DHA and ARA were equally well tolerated by infants, with very low incidence of allergic manifestations.

Concentrations of DHA and ARA in RBC and plasma phospholipids have been widely evaluated as outcomes in studies of milk-based formula-fed and breast-fed infants [3, 6, 1117, 3234]. In the current study, mean fatty acid percentages were similar between control and DHA + ARA groups at 14 days of age. Similar to trials with term infants fed milk-based formulas [12, 3234], infants fed a soy-based formula supplemented with DHA and ARA had significantly higher percentages of fatty acids as DHA and ARA in RBC lipids and plasma phospholipids at 120 days of age than infants fed an unsupplemented soy-based formula. Previously published levels of RBC and plasma phospholipid DHA and ARA from a clinical trial of term infants fed a control milk-based formula without DHA or ARA [12] were similar to those of infants fed the control soy-based formula in this study. For example, mean total RBC DHA and ARA percentages for infants fed control soy-based formula in the current study were 2.47 and 13.97%, respectively, compared with 2.52 and 13.9%, respectively, for control milk-based formula in the earlier study [12]. Likewise, the RBC and plasma phospholipid DHA and ARA values for infants fed milk-based formulas supplemented with 0.32–0.36% of fatty acids as DHA and 0.64–0.72% as ARA [12, 13] were similar to concentrations measured in those blood fractions in the DHA + ARA group in the current study. Blood levels of DHA and ARA measured in infants in the DHA + ARA group at 120 days of age also were similar to those demonstrated in breast-fed infants where representative breast milk contained 0.29% DHA and 0.56% ARA [12]. For example, total RBC DHA and ARA were 6.23% and 16.27%, respectively, for the DHA + ARA group in the current study, compared with 4.97% and 15.9%, respectively, for the breast-fed group and 6.77% and 17.1%, respectively, for the supplemented milk-based formula in the earlier study [12]. Thus, the addition of DHA and ARA to soy-based formula produces similar responses in blood DHA and ARA as addition of similar levels to milk-based formula, as well as similar responses as comparable levels of breast milk DHA and ARA. The significant increase at 120 days of age in the DHA + ARA group further suggests that feeding term infants a soy-based formula supplemented with preformed DHA and ARA increases the levels of circulating DHA and ARA available for brain and retinal development and functions similarly to breast-fed infants [12, 18].

Significant differences in percentages of other PUFA in RBC lipids and plasma phospholipids seen in the DHA + ARA group compared with control also parallel, for the most part, changes seen in previous studies of milk-based formulas [12, 13, 16, 17, 3234]. Lower blood levels of DHA precursors ALA (18:3ω3) and docosapentaenoic (22:5ω3), and ARA precursors linoleic (18:2ω6), γ-linolenic (18:3ω6), and dihomo-γ-linolenic (20:3ω6), as well as eicosadienoic (20:2ω6) and docosapentaenoic (22:5ω6) acids in the DHA + ARA group were consistent with supplemental feeding or feeding higher levels of DHA and ARA, as previously noted [13, 16, 17, 34]. Breast-fed infants have also been shown to have lower percentages of linoleic acid and ALA in total RBC and plasma phospholipids [12] and lower linoleic acid in RBC PC and PE [32] than infants fed control milk-based formula without added DHA and ARA. Blood levels of linoleic acid and ALA were comparable among infants fed human milk with 0.29% DHA and 0.56% ARA and infants fed milk-based formula with 0.36% DHA and 0.72% ARA in the earlier study [12], and infants fed soy-based formula with 0.32% DHA and 0.64% ARA in the current study; mean total RBC and plasma phospholipid linoleic acid across the three groups ranged from 10.9% to 12.58% and 20.2% to 21.49%, respectively, and mean ALA values for both blood fractions across all three groups were less than 0.2%. Our data demonstrate that the addition of DHA and ARA to soy-based formula results in similar responses in blood PUFA levels at similar levels of DHA and ARA in milk-based formulas and in breast milk [12], and provide further evidence that the conversion of ALA and linoleic acid to DHA and ARA, respectively, does not compensate for lack of dietary DHA and ARA [13].

Supplementation of formulas with DHA and ARA is now a widely accepted method of attempting to reproduce concentrations typical of human breast milk and to parallel developmental benefits. The current study is unique in comparing growth, tolerance, and blood chemistry outcomes between infants who received commercially available soy-based formulas with or without DHA and ARA supplementation. Both formulas supported normal growth of infants, and tolerance for each formula was similar between groups. Supplementation of soy-based formula with preformed DHA and ARA from single cell oil sources also resulted in circulating levels of DHA and ARA comparable to previous clinical data from breast-fed infants and infants who received cow’s milk-based formula supplemented with median worldwide levels of DHA and ARA. These results demonstrate that feeding soy-based formulas supplemented with DHA and ARA supports infant growth and achieves increased levels of circulating DHA and ARA.


This study was supported by a grant from Mead Johnson & Company. The authors wish to thank study investigators and their staff for their cooperation, including Barbara Alexander, M.D. (Pediatric Associates, Hollywood, FL), Dean Antonson, M.D. (The Center for Human Nutrition, Omaha, NE), Janet Barnes, M.D. (New Orleans, LA), Azzam Baker, M.D. (Riverside Pediatric Group, Secaucus, NJ), Wesley Burks, M.D. (Arkansas Children’s Hospital, Little Rock, AR), William M. Crecelius, M.D. (Welborn Clinic, Evansville, IN); James Hubbard, M.D. (Medical Associates Clinic, Dubuque, IA); William H. Johnson, Jr., M.D. (Birmingham, AL), Barry Kroll, M.D. (Margiotti & Kroll Pediatrics, Newtown, PA); Matthew Sadof, M.D. (Baystate Pediatric Group, Springfield, MA); Charles Sheaffer, M.D. (Chapel Hill, NC); and Eric Slosberg, M.D. (ProMed Pediatrics, Kalamazoo, MI). The participation of parents and infants in this study is gratefully acknowledged. We also acknowledge valuable contributions from Julia Boettcher, R.D., M.S. and Suzanne I. Stolz in the execution and completion of this study. We are grateful to Dianna K.H. Wheaton, K.J. James, and L.E. Wiedemann for fatty acid analysis conducted at the Retina Foundation. Additionally, we would like to acknowledge Jennifer Wampler, Ph.D. for her contributions to the manuscript development.

Copyright information

© AOCS 2007