Main

The membranes of the CNS and the retina are enriched in the LCP fatty acids AA (20:4n-6) and DHA (22:6n-3). Human milk contains these n-6 and n-3 LCP fatty acids and their essential fatty acid precursors, linoleic acid (18:2n-6) and α-linolenic acid(18:3n-3), respectively, but levels vary depending on the maternal diet(1, 2). Similarly, infant formulas differ in their linoleic acid and α-linolenic acid levels. Formulas marketed in North America do not contain the preformed LCP fatty acids. With few exceptions, healthy, term infants fed commerical infant formulas without LCP fatty acids have lower AA and DHA levels in plasma and RBC phospholipids than infants fed most human milks(3–7).

Studies with primates(8, 9) and rodents(10, 11) fed diets nearly devoid of n-3 fatty acids were fundamental to understanding their physiologic role in development. Compared with animals fed α-linolenic acid from soy oil, animals fed a n-3 fatty acid-deficient diet had lower levels of DHA in brain and retina(9–11), impairments in electrophysiologic responses of the retina(9, 11), poorer visual acuity(8), and poorer performance in tests of discrimination learning(10, 11). These data not only added to the growing evidence that α-linolenic acid should be considered a dietary essential fatty acid, but also showed an association between tissue levels of its longer chain derivative, DHA, and visual and behavioral functions.

In previous studies with preterm infants, formulas containing DHA from fish oil resulted in higher visual acuity at 2 and/or 4 mo corrected age(12, 13), but not thereafter(13), when compared with infants fed unsupplemented formulas. The fish oil-containing formulas had 0.2-0.4% DHA, 0.35-0.65% EPA(20:5n-3), and no AA. AA is the major n-6 fatty acid in phospholipid membranes of most tissues, including the retina and CNS, and is a precursor of the biologically active series 2 prostanoids. Infants fed the supplemented formulas had higher plasma and RBC membrane levels of DHA and EPA(12, 13) and lower levels of AA(13), which may be explained by reduced synthesis of AA or competition between AA and EPA for acylation in phospholipids. In one study, infants fed formula containing DHA grew more slowly than those fed the control formula(14, 15). Similar studies with infants fed formula containing both AA and DHA have not been available.

Lower RBC and plasma phospholipid levels of AA and DHA have been consistently reported for term infants fed unsupplemented formula compared with breast-fed infants(3–7, 16–19), but the visual acuity results have been inconsistent(6, 7, 12, 16–19). Comparisons of breast-fed and formula-fed infants are confounded by the fact that these groups are self-selected, not randomized. Whether the disparate results in visual function were due to differences in parenting practices between mothers who chose to breast feed and those who chose to formula feed, in methods of assessing the visual acuity threshold, or in the compositional differences among the formulas fed in each study is not known. In addition to other differences, the α-linolenic acid levels varied from 0.25 to 1.0% kcal. Studies with rodents(11) and piglets(20) suggest that at least 0.7% kcal as α-linolenic acid is needed for normal accretion of DHA in synaptic membranes and retinal phospholipids. Studies in piglets have also suggested that diet-induced changes in plasma and RBC levels of AA and DHA are not reliable indicators of the levels of these fatty acids in brain(20, 21).

The present longitudinal, prospective study (the Ross Pediatric Lipid Study) of healthy, term infants evaluated (1) whether formula containing ≈10% kcal as linoleic acid and ≈1% kcal as α-linolenic could support development of visual acuity comparable with that of breast-fed infants, (2) whether formula containing DHA and AA or DHA alone would result in increased levels of these fatty acids in RBC phospholipids, and (3) whether increasing RBC levels of these fatty acids would enhance development of visual function in the 1st y of life. Sources of LCP fatty acids that can be used to supplement infant formula include fish oils, egg yolk-derived phospholipid, egg yolk oils, or novel fungal and/or algal(single cell) oils. There is limited experience in feeding LCP fatty acids from any sources to infants. In one study, term infants fed a formula supplemented with DHA from a high EPA fish oil and γ-linolenic acid(18:3n-6) from evening primrose oil to 7 mo of age had higher visual acuity than those fed a placebo formula(17). Others have fed formulas supplemented with AA and DHA from egg lipid to 4 mo of age(22, 23). It is vitally important to evaluate safety as well as possible benefits with sufficiently large studies.

In the present study, infants from three U.S. cities were fed human milk or formulas containing no LCP fatty acids, both AA and DHA, or DHA alone to 12 mo of age. All of the formulas contained linoleic acid at 21% of total fatty acids (≈10% kcal) and α-linolenic acid at 2% of total fatty acids(≈1% kcal). Growth, RBC fatty acid composition, and visual function were evaluated throughout the 1st y of life.

METHODS

Study design and subjects. This longitudinal, prospective, randomized study of healthy, term infants fed one of three formulas from <7 d to 12 mo of age was conducted at Children's Mercy Hospital, Kansas City, MO; Oregon Health Sciences University, Portland, OR; and the University of Washington and Children's Hospital, Seattle, WA. Randomization of infants in the formula groups was determined centrally, and an independent randomization schedule was used at each center(24, 25). A nonrandomized group of breast-fed infants was enrolled concurrently at each center. Feeding group assignments were not disclosed to the investigators. The testing and data collection procedures were standardized across centers to minimize intersite variability. Analyses of the data were performed centrally.

Enrollment criteria required infants to be ≥37 wk of gestation with weight appropriate for gestational age. Infants with Apgar scores <7 at 5 min, those with physical or metabolic defects, or those who received i.v. lipid infusion or blood transfusion were excluded. Infants born to mothers with a history of diabetes, hyperlipidemia, or perinatal infection were also ineligible for participation. Institutional review board approval for these studies was obtained at each site, and written informed consent was obtained from parents of all infants before enrollment. For the formula-fed infants, permission for study entry was sought only if the mothers had previously decided not to breast feed their infants.

All formulas were liquid ready-to-feed formulas with (per liter): 14.3-15.0 g of protein, 72.4-74.8 g of carbohydrate, 35.9-37.2 g of fat, and 670-694 kcal. All formulas met or exceeded the levels of nutrients as recommended by the American Academy of Pediatrics(26). The protein was from nonfat milk and whey protein concentrate, and the oil blend consisted of high oleic safflower, coconut, and soy oils with or without phospholipid or triglyceride sources of LCP fatty acids. The feeding groups differed only in the amounts and sources of dietary LCP fatty acids (Table 1). The control formula contained no added LCP fatty acids, one formula contained AA (0.43 wt% total fatty acids) and DHA (0.12 wt%) from egg yolk phospholipid (AA + DHA formula), and a third formula (DHA formula) provided DHA (0.2 wt%) from a high DHA, low EPA fish (tuna) oil with a ratio of DHA to EPA of ≈4:1. The levels of AA and DHA in the supplemented formulas were within the ranges reported for human milk(1, 2). Human milk samples that were collected in Portland, OR, within the time frame of the study also contained similar levels of AA and DHA to those in the supplemented formula (our unpublished data) and provided a comparison with a contemporary population. A total of 139 samples were collected from 29 women; 64 samples were from 14 women in the breast-fed group of the present study. These samples contained 0.48 ± 0.10 (mean ± SD) wt% fatty acids as AA and 0.15 ± 0.09 wt% as DHA.

Table 1 Fatty acid composition* of human milk and formulas
Full size table

The LCP fatty acids in the AA + DHA formula were provided as phospholipids(≈9% of the total fat blend) with minimal amounts of added cholesterol(<2 mg/L additional cholesterol) and in the DHA formula as triglycerides(≈1% of the fat blend). The amount of coconut oil was reduced by 9 or 1%, respectively. There was no evidence of fatty acid oxidation as determined by analyses of the formula levels of ascorbate and α-tocopherol, vitamins that are susceptible to oxidative degradation.

Formula was provided ad libitum as the sole source of nutrition for a minimum of 4 mo in the formula groups. Breast milk was fed exclusively for a minimum of 3 mo in the breast-fed group, after which supplementation with commercial ready-to-feed SWI (60% soy oil, 40% coconut oil) was permitted. Exclusive breast feeding, however, was encouraged, and it was acceptable that some mothers may have occasionally provided expressed breast-milk by bottle. Supplementation with solid foods was permitted for all infants beginning at 4 mo of age.

Growth, formula intake, and tolerance. Body weight, crown-heel length, and head circumference were obtained at birth, study entrance, and 1, 2, 4, 6, 9, and 12 mo of age during clinic visits using standardized equipment and procedures. Parents recorded the formula intake on dietary records for 3 consecutive days immediately preceding each study visit. The occurrence of spit-up and vomiting and the frequency, color, and consistency of stools were recorded for the 3 consecutive days before the 1-, 2-, and 4-mo visits.

Blood sampling and analysis. A 2-mL blood sample was collected from each infant at 2, 4, 6, and 12 mo of age by venipuncture and placed into tubes containing disodium EDTA. The 4- and 12-mo data are presented in this report. Plasma and RBC were separated by centrifugation, and the RBC pellet was washed with saline. All samples were stored frozen at -70°C before analyses by the Clinical Chemistry Lipid Laboratory at Ross Products Division, Columbus, OH. RBC lipids were extracted with pentane/ethyl ether/isopropanol(1:1:2 v/v/v) at a sample to solvent ratio of 1:10 using a modification of published methods(27, 28). RBC-PC and -PE were separated using thin layer chromatography(29). Silica bands corresponding to RBC-PC and -PE were extracted and saponified with methanolic potassium hydroxide at 100°C for 10 min. Fatty acids were derivatized to their methyl esters using methanolic boron trifluoride and separated by capillary column gas liquid chromatography [Hewlett-Packard Gas Chromatograph 5890 with a Supelcowax 10 column (0.52 mm × 30 cm) and a flame ionization detector]. The temperature-programmed separation included an initial temperature of 125°C for 1 min followed by an incremental increase of 5°C/min to 195°C, which was held constant for 30 min. Fatty acid methyl esters were identified and quantified by use of authentic lipid standards verified by mass spectrometry. The data are expressed as g/100 g of fatty acid (wt%).

Visual function. An ophthalmologic examination was conducted at least once between 2 and 12 mo of age to rule out visual defects or large refractive errors (greater than 4.75 diopters of myopia, 6.0 diopters of hyperopia, or 3.0 diopters of astigmatism)(30). Binocular visual acuity was determined longitudinally at 2, 4, 6, 9, and 12 mo of age. A preferential looking method, the acuity card procedure, was used for infants in Seattle and Portland and an electrophysiologic method, the swept-spatial frequency VEP, for infants in Seattle and Kansas City. In Seattle, where both procedures were carried out on all infants, the order of testing varied between infants and across visits for each infant.

The acuity card procedure provided an estimate of visual acuity by determining the spatial frequency of the smallest grating which an infant could detect, as described previously(31, 32). Stimuli were 2.5 × 55-cm rectangular gray cards (Teller Acuity Cards, Vistech, Inc., Dayton, OH). Each card had a patch of black and white stripes(12.5 × 12.5 cm) positioned with its edge 8 cm to one side of a central peephole. The gratings were matched in space-average luminance to the background (i.e. the stripes could not be discriminated from the background if the pattern was below the infant's acuity threshold). Infants were tested at a distance of 38 cm at 2 and 4 mo of age and at 55 cm at 6, 9, and 12 mo. The younger infants (2 and 4 mo of age) were held by experienced staff or a parent and the older infants by a parent. The spatial frequency of the gratings varied in 0.5-octave steps from 0.44 to 38 cycles/degree(equivalent to 20/2700 to 20/23 in Snellen acuity values) at the 38-cm test distance and from 0.63 to 38 cycles/degree (20/960 to 20/16) at the 55-cm test distance. An observer, masked to the location of the gratings, presented progressively smaller gratings and determined by observation whether the infant was able to discriminate the striped pattern from the background. The visual acuity threshold was the smallest grating judged to be detectable by the infant. Confidence ratings from 1 to 5 (low to high) were recorded for each test, and only estimates with confidence ratings of 3, 4, or 5 were included in the final analysis. Interobserver reliability (for three observers in Seattle and two observers in Portland) was tested three times during the study. Thirty-two reliability tests were conducted. Interobserver agreement was within 0.5 octave for 97% of the tests, similar to test-retest and interobserver reliabilities previously reported for the acuity card procedure(33).

The sweep VEP was recorded using the DIVA-i system as previously described(34). Recordings were obtained from the following electrode derivations: Oz, active; Cz, reference; and Pz, ground. Gold cup electrodes (Grass Instrument Co.) were placed on the scalp and secured with conductive paste and paper tape. Electrode impedance was<5 kohms and equal across electrodes. The infant, held on a parent's lap in a darkened room, was positioned in front of a video monitor that displayed black and white (80% contrast) vertical square wave gratings. The gratings were counter-phase reversed at a frequency of 6 Hz (12 reversals/s) and the mean space-average luminance of the screen was 80 cd/m2. During each 10-s presentation, the spatial frequency of the grating was swept from low to high (large to small stripes) in 19 0.5-s steps. The test distance was 72 cm at 2 and 4 mo and 114 cm at 6, 9, and 12 mo, with a range of spatial frequencies of 0.5-11.5 cycles/degree and 1-23 cycles/degree, respectively. Fixation of the infant's gaze to the screen was encouraged with the help of dangling toys and was carefully monitored. If the infant's fixation deviated from the screen, the trial was interrupted until the infant's attention was drawn back to the screen. EEG responses were amplified (gain 100), filtered(band pass 0.1 Hz to 0.1 kHz), digitized, and analyzed on line. A discrete Fourier transform was used to evaluate the second harmonic of the response (12 Hz component). Adjacent frequency bands at 10 and 14 Hz were used to estimate the level of noise for each trial. A minimum of three 10-s presentations were recorded and vector-averaged; acuity estimates were based on a single presentation only when the vector average was not scorable. The function relating the amplitude of the second harmonic to spatial frequency was determined, and the acuity threshold was estimated by linear extrapolation from the peak of this function to the noise level. Only trials meeting a 3:1 signal-to-noise and phase coherence criteria were used in the estimates.

Statistical methods. Discrete variables were analyzed byχ2 analysis. Growth and visual acuity measures were analyzed by repeated measures ANOVA. Growth was evaluated by overall ANOVA, comparisons at each age, and as interval growth. RBC-PC and -PE fatty acid levels were analyzed by one-way ANOVA. Tukey's multiple comparison procedure was used to test for pairwise differences. The paired t test was used to compare values for RBC phospholipid fatty acid levels between 4 and 12 mo. Data for the Teller acuity and sweep VEP acuity measures were log transformed before statistical evaluations according to convention(35). All statistical analyses included a random block for site (i.e. site was used as a covariate) to minimize intersite variability, and all statistical tests were two-sided using α = 0.05 to determine statistical significance; p values are reported where appropriate.

RESULTS

Enrollment characteristics, growth, and tolerance to formulas. One hundred ninety-seven subjects completed the 12-mo feeding protocol with 63, 45, 46, and 43 infants in the breast-fed, control, AA + DHA, and DHA groups, respectively. With the exception of ethnicity, the birth and enrollment characteristics did not differ among the feeding groups(Table 2). The median age at enrollment was 3 d for breast-fed infants and 2 d for formula-fed infants. Sixty-two infants (23% of the 274 infants enrolled) were withdrawn from the study early because of poor compliance with study procedures, early cessation of breast-feeding, or occurrence of an illness that compromised study participation but was not considered associated with the formula by the attending physician (breast-fed, 13; control, 18; AA + DHA, 13; DHA, 18). The illnesses included one case of each of the following: viral meningitis (control), pyloric stenosis (control), cataracts (AA + DHA), phenylketonuria (breast-fed), anisometropia(breast-fed), and sudden infant death syndrome (DHA group). This latter case was reviewed by a panel of physicians not associated with the study and was considered unrelated to study participation. An additional 15 infants (5% of infants enrolled) were withdrawn from the study due to reported formula intolerance: two in the control group (both from Kansas City), nine in the AA+ DHA group (three from Seattle, six from Kansas City), and four in the DHA group (one from Seattle, three from Kansas City). Exit status was analyzed by the χ2 test according to these criteria; i.e. completed protocol (n = 197); poor compliance or illness not associated with the formula (n = 62), or reported formula intolerance (n = 15). No significant effect by formula group was found. Two infants were excluded from the acuity card procedure analyses because the criteria for inclusion based on the ophthalmologic examination were not met [one in the AA+ DHA group (Portland); one in the breast-fed group (Seattle)]. In addition, one value from each of seven infants was excluded from the acuity card procedure analyses because tester confidence was low (<2 on a scale of 1 to 5).

Table 2 Birth and enrollment characteristics
Full size table

A subset of infants in the breast-fed group were fed human milk without any formula (SWI) supplementation. This subset included 40, 27, 21, and 17 infants at 4, 6, 9, and 12 mo, respectively.

There were no significant differences in growth among the four study groups. Mean values for weight, length, and head circumference (Fig. 1), and their respective standardized z scores (Table 3) were within age-appropriate reference ranges(36). In addition, no significant growth differences were found when infants in the formula groups were compared with the subset of infants fed human milk without formula supplementation (data not shown).

Figure 1
figure 1

Weight, length, and head circumference (±SD) in healthy, term infants fed human milk or formulas with or without LCP fatty acids. Circles, breast-fed (n = 63); squares, control formula (n = 45); triangles, AA + DHA supplemented formula (n = 46); inverted triangles, DHA-supplemented formula (n = 43). No significant differences among groups were found.

Full size image
Table 3 Standardized Z scores of infants fed formula with or without sources of LCP fatty acids or breast milk
Full size table

Formula intake (kcal/d) and tolerance to the formulas, based on the occurrence of spit-up and/or vomiting, did not differ among the formula groups at any time point. Infants in the breast-fed, control, and DHA groups did not differ in their average stool consistencies; in comparison with these groups, infants in the AA + DHA supplemented group had firmer stools at 2 and 4 mo of age (p < 0.05).

Fatty acid composition in RBC-PC and RBC-PE. The fatty acid composition of RBC phospholipids is shown in Table 4 for PC and Table 5 for PE. Comparisons among groups indicated no differences in the levels of the saturated fatty acids. All of the formula groups had levels of oleic acid and total monounsaturated fatty acids that were 10-20% higher than infants in the breast-fed group (p < 0.01). The n-6 and n-3 fatty acid levels, on the other hand, differed considerably among all of the groups.

Table 4 RBC-PC fatty acids in term infants fed formula with or without sources of LCP fatty acids or breast milk
Full size table
Table 5 RBC-PE fatty acids in term infants fed formula with or without sources of LCP fatty acids or breast milk
Full size table

Compared with the breast-fed group, infants fed the control formula had higher linoleic acid levels in both RBC-PC (15% greater) and RBC-PE (40% greater) at 4 mo of age (p < 0.001). However, despite higher levels of linoleic acid, the precursor to AA, infants fed the control formula had AA levels that were about 30% lower in RBC-PC than those in the breast-fed group and about 10% lower in RBC-PE at 4 mo (p < 0.001). These differences were no longer present at 12 mo. The levels of DHA, however, were about 40% lower in RBC-PC and RBC-PE at both 4 and 12 mo than in the breast-fed group (p < 0.001), and levels of 22:5n-3 in RBC-PE were about 25% lower at both ages (p < 0.001).

Infants fed the AA + DHA formula, which contained AA and DHA at levels within the ranges reported for human milk, had levels of AA and DHA in RBC-PC and -PE that in most cases were not different from infants in the breast-fed group and were greater than for infants in the control group. Linoleic acid levels also differed by less than 15%, if at all, from values for breast-fed infants and were about 10-15% lower than for infants in the control group(p < 0.001). The levels of the 20 and 22 carbon fatty acids intermediates resulting from the elongation and desaturation of linoleic acid and α-linolenic acid, however, were notably different from those in both the breast-fed and the control groups. Specifically, in comparison with the breast-fed infants, dihomo-γ-linolenic acid (20:3n-6) levels were ≈35% lower in RBC-PC at 4 mo and 10-15% lower in RBC-PE at both ages, and the levels of docosapentaenoic acid (22:5n-6) were about 50-110% greater in RBC-PE (p < 0.001). The total n-6 fatty acids in the AA + DHA group, however, were generally not different from those in the breast-fed group. The levels of the n-3 docosapentaenoic acid(22:5n-3) in RBC-PE were 40-55% lower in the AA + DHA group than in the breast-fed and control groups (p < 0.001). The total n-3 fatty acid levels were slightly higher in RBC-PC at 4 mo but 20-30% lower in RBC-PC at 12 mo and in RBC-PE at both 4 and 12 mo than in the breast-fed group (p < 0.001).

Infants fed the DHA formula, which contained DHA within the ranges reported for human milk but no added AA, had higher levels of DHA than infants in the breast-fed as well as the other formula groups. The DHA group, however, had lower levels of AA and its 22-carbon elongation and desaturation products, docosatetraenoic acid (22:4n-6) and docosapentaenoic acid(22:5n-6), than the other groups (p < 0.001). Specifically, infants fed the DHA formula had DHA levels that were 20-55% greater than in the breast-fed group and more than twice that of infants fed the control formula (p < 0.001). In RBC-PE the level of the intermediate in the pathway for DHA biosynthesis, docosapentaenoic acid(22:5n-3), was about 30% lower than that for the breast-fed group(p < 0.001) and slightly lower than that for the control group at both ages. As a consequence, total n-3 fatty acid levels in the DHA group were 20-50% higher in RBC-PC (p < 0.001) but were not different or only slightly higher in RBC-PE when compared with the breast-fed group. The AA levels in RBC-PC and RBC-PE, on the other hand, were about 15-35% lower than in the breast-fed group and about 10-15% lower than in the control group. In addition, infants fed the DHA formula had levels of docosatetraenoic acid (22:4n-6) and docosapentaenoic acid(22:5n-6) in RBC-PE that were 30-45% lower than infants in either the breast-fed or control groups at both 4 and 12 mo (p < 0.001).

The subset of infants who were fed human milk without formula (SWI) supplementation had AA and DHA levels that were generally within ≈10% of those for the breast-fed group as a whole. Specifically, in the exclusively breast-fed group, AA was 10.2 ± 0.3 and 9.8 ± 0.4 wt% total fatty acids in RBC-PC and 26.0 ± 0.7 and 27.1 ± 0.6 wt% in RBC-PE at 4 and 12 mo, respectively; DHA levels were 2.4 ± 0.1 and 2.4± 0.2 wt% in RBC-PC and 6.6 ± 0.3 and 6.0 ± 0.4 wt% in RBC-PE at 4 and 12 mo, respectively. At 12 mo, however, DHA levels in RBC-PC were about 25% greater in the exclusively breast-fed group (2.4 ± 0.2 wt%) than for the breast-fed group as a whole (1.9 ± 0.1 wt%).

RBC fatty acid levels in the exclusively breast-fed infants were compared with those in the formula-fed infants. In infants fed the AA + DHA formula, RBC-PC and -PE AA levels were not different from those in infants fed exclusively breast milk at both 4 and 12 mo. DHA levels were 20-25% lower in RBC-PC at both 4 and 12 mo and 15% lower in RBC-PE at 4 mo only (p< 0.001). In infants fed the control and DHA formulas, RBC-PC AA levels were about 40% lower at 4 mo and 20-30% lower at 12 mo. RBC-PE AA levels were 10-15% lower in the DHA group at 4 and 12 mo and in the control group at 4 mo only (p < 0.001). DHA levels in both RBC-PC and -PE, however, were consistently 40-45% lower in infants fed the control formula when compared with the exclusively breast-fed infants (p < 0.001). Infants fed the DHA formula, on the other hand, had RBC-PE DHA levels that were 25% higher at 4 mo and 40% higher at 12 mo than those fed breast milk only (p < 0.001). RBC-PC DHA levels were 20% higher at 4 mo(p < 0.001), but were not different at 12 mo.

Visual function. There were no significant differences in acuity thresholds among the four infant groups at 2, 4, 6, 9, or 12 mo of age using either the acuity card procedure or the sweep VEP method (Fig. 2). Acuity values for the three formula groups also did not differ from those of infants in the subset fed human milk without formula supplementation (Fig. 2, inset).

Figure 2
figure 2

The visual acuity thresholds measured by the acuity card procedure (A) and the sweep VEP method (B). Data are shown on the left axis as mean (cycles/degree; cy/deg) ± SD(octaves) and on the right axis as the equivalent Snellen values. The inset shows the acuity thresholds for infants in the formula groups compared with breast-fed infants given no formula (SWI) supplementation. The number of infants by group are shown; no more than three values were missing at any time point because of low tester confidence or missed visits.(A) Acuity card procedure. Open circles, breast-fed(n = 38); squares, control formula (n = 28);triangles, AA + DHA-supplemented formula (n = 26);inverted triangles, DHA-supplemented formula (n = 28);closed circles (inset only), breast-fed with no formula supplementation. The number of infants at 2, 4, 6, 9, and 12 mo was 37, 27, 17, 15, and 11, respectively. (B) Visual evoked potential.Open circles, breast-fed (n = 41); squares, control formula (n = 26); triangles, AA + DHA-supplemented formula (n = 23); inverted triangles, DHA supplemented formula (n = 28); closed circles (inset only), breast-fed with no formula supplementation. The number of infants at 2, 4, 6, 9, and 12 mo was 37, 22, 16, 12, and 11, respectively. No significant differences among groups were found.

Full size image

DISCUSSION

This is the first prospective, longitudinal study of healthy, term infants to evaluate whether providing AA and DHA or DHA in formula containing ≈10% kcal as linoleic acid and ≈1% kcal as α-linolenic acid would increase RBC phospholipid AA and DHA levels, enhance the development of visual function, and support normal growth. The results show that levels of RBC phospholipid DHA and AA similar to those found in infants fed human milk could be attained in formulafed infants by providing both AA and DHA in amounts within the ranges reported for human milk. Infants fed the supplemented formulas, however, showed no detectable enhancement in visual acuity development as measured by either a behavioral (acuity card) or electrophysiologic (VEP) method. Furthermore, infants fed the control formula with no LCP fatty acids exhibited visual acuity development comparable with breast-fed infants during the 1st y of life. Thus, the results do not provide support for published recommendations for LCP supplementation of term formulas(37, 38).

A similar study has also recently shown no overall difference in grating acuity between 2 and 12 mo of age in infants fed a control formula containing≈11% kcal as linoleic acid and ≈1% kcal as linolenic acid when compared with breast-fed infants or infants fed a formula supplemented with AA + DHA from egg phospholipid(19). Although there was no overall diet effect on grating acuity, there was an age by diet interaction with lower acuity card scores at 2 mo, but not at 4, 6, 9, or 12 mo, for infants fed the control formula. An explanation for the disparity in acuity card results at 2 mo between this and the present study is not readily apparent. The composition of the control formula and the sources and amounts of AA + DHA in the supplemented formula were similar in both studies. Whether other explanations, such as population characteristics, may account for the difference in grating acuity at 2 mo is unknown. Grating acuity at 2 and 4 mo, when visual acuity is increasing exponentially, was higher in the Carlson study than in the present study. It is conceivable that infants differed in age when tested, as visits for testing were scheduled by postconceptional age (postmenstrual age) in the Carlson study and postnatal age in the present study. A consistent finding between these two studies is that, between 4 and 12 mo of age, there was no association between RBC phospholipid levels of DHA and visual acuity.

These visual acuity results, however, are contrary to those reported in a study conducted in Australia with infants fed breast milk or randomized to formulas with or without a supplement that provided DHA and EPA from fish oil and γ-linolenic acid (18:3n-6) from evening primrose oil(17). Lower VEP acuity scores were reported for infants fed the control formula than those fed breast milk or the supplemented formula. In addition, infants randomized to the control formula, but not the supplemented formula, showed a decrease of about 0.5 in weight and length Z scores during the 7-mo feeding study. This is in contrast to our results showing no formula effects on growth. Possible explanations for the contradictory findings between this study and the present study include differences in the characteristics of the breast-feeding and formula-feeding populations, differences in the compositions of the control and supplemented formulas studied, and/or differences in the method used to obtain the VEP acuity estimates. The levels of DHA supplementation were comparable, although the levels of linoleic acid (≈8% kcal) and α-linolenic acid (0.8% kcal) were slightly lower in the formulas from the Australian study. Whether there are other differences in composition or nutrient stability between the liquid formulas used in our study and the powder formulas used in the Australian study is not known. The procedures to obtain the VEP acuity estimates also differed between the two studies. Differences included the temporal dimension of the stimulus (transient versus steady-state VEP), spatial features of the pattern (checkerboard versus square wave gratings), and the response component of the VEP that was analyzed. Although these two stimulus configurations have not been explicitly compared in infants, it is possible that the two techniques are tapping different mechanisms with different developmental time courses(39, 40). The difference is not likely to be in the sensitivity of the two methods. Based on calculations of the statistical power for the sweep VEP in the present study, a difference of about 0.5 octave could be detected with the number of infants included in each feeding group, well within the 1-octave difference reported in the study from Australia.

There is lack of agreement among previous studies that compared visual function in term infants fed breast milk with those fed infant formulas without LCP fatty acids(6, 7, 12, 16–19). In a cross-sectional study of 16 infants, VEP acuity thresholds at 5 mo of age were lower for infants fed one of three commercial formulas available in Australia compared with a breast-fed cohort(6). In another report(16), 4-mo-old infants fed a formula with ≈0.25% kcalα-linolenic acid had poorer preferential looking and VEP acuity than infants fed human milk. Another cohort of infants fed the same formula to 12 mo of age had poorer random dot stereoacuity and letter matching skills at 3 y of age. In the latter retrospective study, the formula-fed infants were given a solid food diet low in cholesterol and n-3 LCP fatty acids and a high linoleic acid supplement, whereas infants fed human milk were given a high oleic acid supplement and egg yolk (containing cholesterol and n-3 LCP fatty acids). In contrast to these reports, no differences in visual acuity assessed by the acuity card procedure were reported in a prospective study of Canadian infants fed a commercial formula or human milk to 3 mo of age(7). The higher level of α-linolenic acid (≈1% kcal) in the formula from the latter study compared with the retrospective studies (≈0.25-0.8% kcal) may explain the discrepant results. Other possible explanations for the lack of agreement among studies include additional differences among the formulas and/or differences in the methods for estimating visual acuity thresholds. Within the sensitivity of the methods used in the present study, our results are consistent with the conclusion that formula with ≈1% kcal as α-linolenic acid supports normal visual development in term infants.

There is evidence of slower growth in preterm infants fed formula with DHA from a high DHA, high EPA fish oil(14, 15). Preterm infants fed formula containing 0.2 wt% total fat as DHA and 0.3 wt% as EPA to 9 mo corrected age had reduced plasma AA and slower growth to 1 y of age, compared with infants fed a control formula. In the present study of term infants, RBC membrane AA concentrations in infants fed the DHA formula were 10-15% lower than in the control group and as much as 37% lower than in the breast-fed or the AA + DHA groups. However, growth was normal. It is not known whether the relative differences in growth between this study and the preterm infant study are due to differences in formula composition, such as the lower amount of EPA used in the present study, or to differences in the physiologic maturity of the infants. Based on the normal growth, the control formula and the formulas supplemented with fish oil (DHA alone) or egg yolk phospholipid(AA + DHA) appear suitable for feeding to healthy, term infants.

Consistent with other reports(3–7, 16–19, 22, 23), healthy, term infants fed formula without LCP fatty acids in this study had lower RBC levels of both DHA and AA than those fed human milk or supplemented formulas. The group of infants fed the AA + DHA formula had levels of AA and DHA most closely resembling those of breast-fed infants, confirming together with one other report(22) that it is possible to achieve RBC AA and DHA levels similar to those of breast-fed infants. Both studies used egg sources of AA and DHA, thus providing most of the LCP fatty acids in phospholipid form. It is of interest that the differences in AA and DHA levels among the groups generally persisted to 12 mo of age despite the introduction of solid food in the diet after 4 mo of age. If subsequent studies suggest benefits to providing LCP fatty acids in infant formula, adding AA and DHA is preferable to adding DHA alone to simulate more closely the levels of these two fatty acids in human milk.

The mean levels of RBC DHA among the three formula groups encompassed the range of DHA of values for infants fed human milk. In some(6, 16) but not all studies(7), RBC phospholipid levels of DHA(6, 16) or the ratio of DHA/22:5n-6(16) at 4 mo of age was correlated with visual or retinal function. In contrast to these reports, the absence of an apparent relationship between RBC levels of DHA or the DHA/22:5n-6 ratio and visual acuity in the present study indicates that RBC fatty acid levels may not always be correlated with visual function in healthy, term infants.

In summary, the present study showed normal growth and visual acuity development in infants fed one of three formulas containing ≈10% kcal linoleic acid and ≈1% kcal α-linolenic acid and either no added LCP, AA and DHA from egg yolk phospholipid, or DHA alone from fish oil, despite large differences in RBC membrane AA and DHA levels among groups. These data also show that providing 0.4 wt% AA and 0.12 wt% DHA from egg yolk phospholipid in formula results in RBC membrane AA and DHA levels similar to those of infants fed human milk. However, the fact that visual function was not different among any of the groups in this study does not support adding DHA or AA to infant formula. Thus, the premise that adding DHA and AA would be beneficial to infants born at term remains unproven. Further research is clearly necessary and should include studies with populations of human infants that may be nutritionally compromised at birth.