Main

The last intrauterine trimester is a critical period for intrauterine accumulation of very long chain polyunsaturated fatty acids of then-3 and n-6 families, especially DHA (22:6n-3) and AA (20:4n-6). Together these fatty acids account for a very high percentage of phospholipid fatty acids in the retina and the gray matter of the brain(1–3). Accumulation of DHA in these tissues is associated with enhanced retinal response to light(4) and visual acuity(5) and shorter look duration (evidence of more rapid visual processing or faster disengagement of attention)(6) in nonhuman primates.

In preterm infants, several randomized trials have compared the effects of infant formulas with and without DHA on development of the visual system. In these trials 0.2 to 0.35% of total fatty acids as marine oil DHA were added to formulas containing 1.4 to 5% linolenic acid (18:3n-3) equivalent to 0.7 to 2.5% of energy. Compared with infants fed formulas containing 2.7 to 5% linolenic acid, those fed the DHA-supplemented formulas had higher circulating phospholipid DHA(7), transiently better visual acuity at 2 mo(8–10) and 4 mo(9, 11), and a more mature pattern of visual attention (shorter look duration) at 6, 9, and 12 mo (all ages from expected term)(12). The effects of feeding preformed DHA compared with linolenic acid alone on these functions is similar to the effects seen in monkeys fed n-3-sufficient compared withn-3-deficient diets(5, 6). These similarities suggest that performed DHA is conditionally essential for optimal visual and neural development of preterm infants.

On the other hand, there is mixed evidence for a relationship between DHA status and acuity in term infants. Several nonrandomized studies of term infants receiving DHA from human milk(13–15) have provided suggestive evidence for such a relationship, whereas another has not(16). Of two recent randomized trials of DHA supplementation, one found that a DHA-supplemented diet enhanced visual acuity of term infants(17), whereas a larger, multisite trial found that it did not(18, 19). In the present study, the primary goal was to determine whether infants randomly assigned to formula with 0.1% DHA would have better visual acuity than infants randomly assigned to formula without 0.1% DHA. The source of DHA used (egg yolk lecithin) also provided by design 0.43% of total fatty acids as AA to prevent any possible imbalance of long chain n-3 and n-6 fatty acids. The amounts of AA and DHA in the experimental formula were typical of those that have been reported in milk of women in the United States delivering at term(20, 21).

METHODS

Subject selection. Infants eligible for this study were cared for in the Crump Women's Hospital at The University of Tennessee, Memphis, Memphis, TN. Infants born at term (37 to 43 wk PMA) were eligible for this study if they were not growth-retarded in utero and had no medical problems likely to influence long-term growth and development. A total of 94 infants were enrolled. Fifty-eight remained in the study through at least 4 mo(19 breast fed, 20 fed the control formula, and 19 fed the experimental formula). This group constituted the final study group whose neonatal and perinatal characteristics are provided in Table 1. Thirty-six infants were lost from the study before 4 mo (16 breast fed, 11 control, and 9 experimental). Of these, 21 infants did not return for even the first scheduled screening visit at 3-wk postnatal age. Mothers of 14 infants changed their minds about being in the study before the second scheduled visit at 2 mo or decided to stop breast feeding before 3 mo, the minimum agreed-upon duration of breast feeding. One infant was dropped because the mother restricted her feeding, and the infant grew poorly.

Table 1 Neonatal and perinatal characteristics of study infants
Full size table

Experimental design. This was a randomized, double-blind study of the effects of infant formula with and without DHA and AA on visual acuity and blood phospholipid fatty acid composition. A third diet group (breast fed) was not randomized but was followed and tested without knowledge of the diet fed. The primary objective of the study was to determine how these diets influenced visual acuity and circulating phospholipid fatty acids. This objective is the basis for this report. A second objective was to obtain normative data on growth and development of term infants in our population, which could be compared with equivalent data from preterm infants. Because our previous studies have shown that more infants are required to determine a statistically significant effect on visual acuity than for phospholipid fatty acids, the former was used to determine the n/group. The power analysis was based on our previous study of visual acuity in preterm infants(10), which demonstrated that 15 infants per group was sufficient to reject the null hypothesis at the p < 0.05 level and a power of 0.80 for a difference in acuity between groups of 0.50 octave.

Written informed consent was obtained from the infants' parents ≈24 h after birth according to a protocol approved by the Institutional Review Board of The University of Tennessee, Memphis. Only women who expressed a desire to feed formula on admission to the hospital were approached for permission to randomize their infants to formula with or without egg phospholipid. Women who had expressed a desire to breast feed their infants were approached for assent for the nonrandomized, breast-fed group. The purpose of the study and its attendant obligations were explained to the mother. If consent was obtained, the infant was enrolled and randomly assigned to formula with or without added egg phospholipid.

The gestational age at birth (dates/ultrasound) was used to determine the dates for planned follow-up visits during the first year of life. All visual acuity assessments were obtained within windows of 2 wk time at 48, 57, 66, 79, and 92 wk PMA with the intent of decreasing the mean error in the grating acuity assessment at these ages. Because it is conventional to express the age of term infants in months, these PMA are referred to as 2, 4, 6, 9, and 12 mo, respectively, throughout this report. Venous blood was obtained for fatty acid analysis at birth from cord blood and at 2, 4, 6, and 12 mo by venipuncture.

Table 2 presents the fatty acid composition of the formulas. Both the formulas provided ≈11% of energy from linoleic acid and≈1% of energy from linolenic acid. The protein in the formulas was from cow milk protein and whey protein, and the fat was a combination of high oleic safflower, coconut and soybean oils with or without egg phospholipid to provide DHA and AA. The vitamin and mineral content of the formulas was identical with Similac with Iron (Ross Products Division, Columbus, OH).

Table 2 Fatty acid composition of study infant formulas
Full size table

Visual acuity assessment. Binocular visual acuity was determined with the Teller Acuity Cards(22, 23) at 48.2 ± 0.1, 57.2 ± 0.1, 66.3 ± 0.1, 79.4 ± 0.1, and 92.1 ± 0.1 wk PMA (LS means ± LS errors). These cards have been used previously to assess visual acuity in infants and children(24, 25). During the procedure, the investigator must determine the location of the black and white grating on the right or the left side of a card by observing the infant's preferential looking behavior from the back of the card through a center peephole. The finest grating at which the tester can locate the grating based on the infant's behavior is recorded as the infant's threshold visual acuity. A conversion table is used to determine acuity (cycles/degree) from the finest grating recognized(cycles/cm) and the distance of the subject from the visual stimulus. Because acuity develops exponentially until approximately 6 mo past term and log acuity in infants is normally distributed(26), individual acuities were transformed to log10 cycles/degree before any analyses were conducted. The mean and SD of the log acuity values were determined for each group, and then the means were transformed to cycles/degree and the SD expressed in octaves (SD of log acuity scores/0.301). A difference between groups of 1.0 octave means that the smallest stripe detected by one group of infants is twice as large as the smallest stripe detected by the other.

There are several acceptable ways to administer the procedure. The procedure we used to test for acuity was the one we used and described earlier(9), except that in this study 4-mo-old infants were presented first with the card having a grating size of 0.32 cycles/cm, and 12-mo-old infants were presented first with the card having a grating size of 1.6 cycles/cm. The procedure used differed from the procedure described in the 1990 version of the manual in that the 9- and 12-mo infants were tested at 38 cm rather than 55 cm, just as in our previous trial(9).

The investigator (S.E.C.) was unaware of the infants' dietary treatments and the results of earlier acuity tests. A second investigator (S.H.W.), also unaware of the dietary assignment and earlier results, occasionally tested in the absence of S.E.C. Interrater reliability of these investigators has been reported to be high(9). The infants received a routine examination by a pediatric ophthalmologist during the first year of life to assure that there were not visual problems that could have a detrimental effect on grating visual acuity. All infants included in the analysis had<4.7 diopters of myopia, <6.0 diopters of hyperopia, <3.0 diopters of astigmatism, and <20 diopters of esotropia. A single infant with 30 diopters of esotropia was not included in the analysis but was retained in the study of growth outcomes.

Fatty acid analysis. Blood (1-1.5 mL) was removed from a small hand vein using a 25-gauge butterfly needle and antico-agulated with lithium heparin. The samples were placed on ice immediately and the plasma separated by centrifugation and stored under nitrogen at -70°C. RBC were washed three times with 0.15 M NaCl in 1 mM EDTA, resuspended in an equal volume of the saline-EDTA solution, and stored at -70°C, also in a nitrogen atmosphere for analysis within the week. Plasma and RBC were extracted according to Dodge and Phillips(27) with chromatography grade solvents and washed with 0.15 M KCl according to Folch et al.(28) to remove nonlipid contaminants. The individual phospholipids were separated on 10 × 10-cm silica gel plates prepared from 20 × 20-cm plates (0.25 mm; Analtech, Inc., Newark, DE) in chloroform-methanol-acetic acid-water (60:30:8.4:4.6, vol/vol/vol/vol), a modification of the solvent system of Zail and Pickering(29), for 20 min. The phospholipid bands were located by breaking small sections from either side of the plate, spraying them with water:sulfuric acid (50:50, vol/vol), and charring. Butylated hydroxytoluene(50 mg/L) was added to the methanol in the thin layer chromatographic separation. Fatty acids were methylated according to Morrison and Smith(30) in an oxygen-free atmosphere. The methyl esters were separated by capillary gas liquid chromatography on a 0.25 mm × 30-m fused silica column with a stationary liquid phase (SP-2330, Supelco, Inc., Bellafonte, PA) as reported previously(7). The results were expressed as g/100 g total fatty acids.

To quantify the amount of each fatty acid in plasma PE and PC, the above procedures were used for lipid extraction, thin layer chromatography, methylation and separation of fatty acid methyl esters. Total lipids were extracted from exactly 0.3 mL plasma and reconstituted to 0.3 mL in dichloromethane with a recovery of 96 to 98%. Exactly 0.15 mL (representing 0.15 mL of plasma) was spotted in one lane on a silica gel plate with additional sample spotted on either side of the lane for detection of phospholipid bands as described above. After thin layer chromatography PE and PC were removed completely from the plate with a single-edged razor blade onto glassine paper and transferred quantitatively to a methylation vial with exactly 5 μg of heptadecanoic acid (17:0) in 100% ethanol. The ethanol was evaporated under nitrogen before the addition of methanolic boron trifluoride for transesterification of fatty acids. The plasma concentrations of specific fatty acids in PE and PC were determined after capillary gas chromatography by calculating the area under the curve for 17:0 that represented 1 μg of fatty acid. Next, the number of micrograms of each fatty acid in the analysis was determined by dividing each area by the area/μg of fatty acid. Finally, the micrograms of each fatty acid in the analysis was multiplied by 6.67 (1.0 mL/0.15 mL plasma) to correct for plasma PE or PC μg/mL (mg/L). No correction was made for the small consistent losses of lipid that occurred before 17:0 could be added during transesterification.

To determine the formula fatty acid composition, total lipids were extracted(31), saponified for 10 min with methanolic 0.05 M KOH, and methylated by a modification of Morrison and Smith(30). Individual fatty acid methyl esters were separated by capillary (0.25 mm × 30-m Omegawax 320) gas liquid chromatography using a 16-min temperature program with initial and final column temperatures of 120 and 220°C, respectively.

Statistical methods. The effect of diet on the development of visual acuity was determined by using the General Linear Model procedure of SAS Version 6.0 (Cary, NC)(32) on the mainframe computer at The University of Tennessee, Memphis. The experimental design, involving repeated measures over time with each baby serving as its own control, may be called a split-plot or nested-factorial design(33). This analysis is particularly useful for clinical trials of the sort reported in this study because it retains the advantages of a repeated measures analysis of variance but does not exclude available data from subjects who are not tested at all ages.

Babies were nested within dietary treatments and repeatedly tested at 2, 4, 6, 9, and 12 mo. Because not every baby could be evaluated at each age, LS means (also called population marginal means or adjusted treatment means) were estimated for each treatment time subclass(34). The number of infants tested for each treatment time subclass are provided in Table 4.

Table 4 Ages at assessment, time required for measuring acuity at each age, and visual acuity of term infants fed human milk, or formulas with and without egg yolk lecithin DHA and AA
Full size table

The specific hypotheses tested in this study were the following.1) There is no difference in visual acuity between the dietary treatments at 2, 4, 6, 9, or 12 mo. 2) There is no change in visual acuity over time for the three dietary treatments. To test these hypotheses, we subdivided the treatment-time sum of squares and degrees of freedom into separate contrasts and constructed F tests. The pooled within-baby variance was used as the error term (or denominator) for F tests of specific hypotheses at each of the time intervals and to assess changes within each dietary treatment group over time. A p value of <0.05 was considered evidence of a significant difference between groups.

RESULTS

Effect of DHA and AA supplementation on plasma and RBC PE DHA and AA. The effects of diet on RBC PE DHA and AA and the concentration of DHA and AA in plasma PC are shown in Table 3. At birth the mean concentrations of plasma PC AA and DHA were high, and both declined significantly in infants randomly assigned to be fed infant formula without egg phospholipid AA and DHA but not in infants assigned to the formula with egg phospholipid AA and DHA (Table 3). At 2, 4, and 6 mo, the supplemented group had significantly higher concentrations of plasma AA and DHA compared with the unsupplemented group, but at 12 mo the groups no longer differed in plasma AA and DHA concentrations. At 2 mo, when all infants in the group were still being breast fed, their AA and DHA concentrations were equivalent to those of the group fed egg phospholipid-supplemented formula. After this time infants fed human milk had concentrations of AA and DHA that were intermediate between the two formula groups at all ages, which was not surprising considering that the breast-fed group was receiving human milk and formula after 3 mo.

Table 3 Select RBC (g/100 g total fatty acids) and plasma(mg/L) phospholipid DHA and AA* in breast-fed infants and those fed the standard and experimental formulas
Full size table

At birth, RBC PE DHA (g/100 g total fatty acids) was similar in infants randomly assigned to the formulas but was significantly higher in the supplemented compared with the unsupplemented infants at 4, 6, and 12 mo. At 2 mo when all breast-fed infants were still receiving human milk, the breast-fed and supplemented infants had similar RBC PE DHA and AA. Thereafter, when the breast-fed group was receiving mixed human milk and formula, they had values for DHA and AA that were intermediate between those of the supplemented and unsupplemented infants.

Effect of DHA and AA supplementation on visual acuity. The effects of diet, diet × age and age on grating acuity are shown in Table 4. There was no overall effect of diet on visual acuity, but at 2 mo breast-fed infants and those fed formula supplemented with egg phospholipid had better visual acuity compared with infants fed the unsupplemented formula. This effect of diet was transient; i.e. it was not observed at 4, 6, 9, and 12 mo. Age had a highly significant effect on visual acuity. Visual acuity of each diet group improved between 2 and 4 mo and between 4 and 6 mo before plateauing during the last 6 mo of infancy.

The amount of time required for the actual assessment of visual acuity was not influenced by diet, but it was significantly influenced by the age of the infant at the time of assessment (Table 4). At 2 mo just over 3 min was typically required for assessment. At 4, 6, 9, and 12 mo, approximately 2 min were required to determine grating acuity.

DISCUSSION

Preterm infants fed formula with added DHA have been shown to have higher grating acuity at 2 and 4 mo corrected age compared with infants fed formulas with ≥0.7% of energy from linolenic acid in three published randomized, double-blind trials(9–11). Similarly, we found in this study that term infants fed formulas with added AA and DHA had higher grating acuity at 2 mo of age but not at 4, 6, 9, or 12 mo of age compared with infants fed an unsupplemented formula that provided ≈1% of energy from linolenic acid. In this trial, just as in our earlier trials in preterm infants(9, 10), the effect of dietary DHA on visual acuity was transient, disappearing within the first 6 mo of infancy despite higher plasma and RBC phospholipid DHA in the experimental compared with the control group(7, 10). However, despite the fact that the effect of DHA supplementation on visual acuity of preterm infants was transient, the DHA-supplemented infants had differences in visual attention later in infancy compared with controls. These differences suggested that DHA supplementation enhanced the speed of visual information processing in 12-mo preterm infants whether they were supplemented to 9 mo(12) or 2 mo(35) corrected age.

Two other randomized, double-blind trials of DHA supplementation in term infants have been reported(17, 18). The first provided term infants with a marine oil source of DHA and eicosapentaenoic acid, 0.36 and 0.58% of total fatty acids, respectively, and found higher visually evoked potential grating acuity at 4 and 7 mo compared with infants fed 0.75% of energy from linolenic acid(17). The second, a large multisite trial, used formulas identical to those fed in this study; i.e. the supplemented formula contained only 0.1% DHA and no eicosapentaenoic acid, but was conducted independently from our study. Although detailed data from this study are not yet available, an abstract of that study concluded that egg phospholipid supplementation did not enhance grating acuity measured by the Teller Acuity Card procedure at any age(18, 19).

Because this area of research is so new and several of the trials have appeared only in abstract form, it is difficult to speculate about differences among studies that might have led to different visual function outcomes. However, the possibilities include differences in 1) population characteristics that might influence DHA accumulation in utero,2) sources of DHA, 3) concentrations of DHA in the experimental formula, 4) techniques used to measure visual acuity,5) study ages, and 6) ranges of PMA at study ages when visual acuity is increasing exponentially.

Differences in human milk and formula composition other than essential fatty acid composition and the inevitable differences between parents who choose to breast compared with bottle feed add to the potential confounders in nonrandomized studies, which also show inconsistent effects on visual function(13–16). Even though breast-fed infants in this study were selected from the same hospital-based population and had family incomes and numbers of individuals in the household statistically equivalent to the randomized groups, their parents were more likely to be white and better educated than parents of formula-fed infants(Table 1). The different visual acuity outcomes among nonrandomized(13–16) and randomized(17–19) trials may be due to differences among the possibilities listed above. The need for preformed DHA by term infants may also be less universal than that of preterm infants who are removed from the maternal/placental supply early in the last intrauterine trimester(36). Given the number of possible differences among populations and studies designed to measure the effects of DHA intake upon visual function, it is likely that the need by term infants for DHA will remain controversial for some time.

An effect of DHA supplementation on visual function is predicated on higher accumulation of DHA in the retina and/or visual cortex. Although RBC PE DHA and visual acuity were not correlated in this study, this does not weaken the conclusion of this study that dietary DHA enhanced early visual development. Although RBC PE DHA has been shown previously to correlate with grating acuity in infants(9, 11, 13, 14), a number of variables, including the past and present DHA exposure (amounts, durations, and their temporal relationship), undoubtedly determine when and to what extent the influence of dietary DHA will be reflected in RBC PE DHA. We think it is remarkable that DHA supplementation increased 2-mo visual acuity in these infants despite apparently normal intrauterine DHA accumulation (based on plasma PC and RBC PE DHA at birth; Table 3) and the small amount of DHA (0.05% of energy) included in the diet.

In summary, term infants were fed infant formula containing 11% and 1% of energy as linoleic and linolenic acid, respectively, with or without DHA and AA from egg phospholipid. The amounts of DHA and AA in the supplemented formula were similar to those that have been reported in milk of women in the United States(20, 21). Infants fed the supplemented formula, as well as infants in the breast-fed group, had higher plasma phospholipid DHA and AA and visual acuity at 2 mo compared with infants fed the formula without egg phospholipid. Early visual acuity of the term infants fed the supplemented formula increased with dietary DHA compared with linolenic acid alone despite quite low amounts of DHA (0.05% of dietary energy) and apparently normal intrauterine accumulation of DHA.