See corresponding editorial on page 181.

Transcription

1 Blood lipid concentrations of docosahexaenoic and arachidonic acids at birth determine their relative postnatal changes in term infants fed breast milk or formula 1 3 Philippe Guesnet, Prity Pugo-Gunsam, Chantal Maurage, Michèle Pinault, Bruno Giraudeau, Jean-Marc Alessandri, Georges Durand, Jean-Michel Antoine, and Charles Couet ABSTRACT Background: Factors other than dietary fatty acids could be involved in the variability observed in blood docosahexaenoate (22:6n 3) and arachidonate (20:4n 6) status in formula-fed infants. Objective: We considered the 22:6n 3 and 20:4n 6 status at birth to be one of these factors and studied its influence on postnatal changes in term infants fed 4 different diets. Design: The blood phospholipid composition was determined at birth and on day 42 of feeding in 83 term infants fed breast milk, nonsupplemented formula, or 2 different 22:6n 3-supplemented formulas. Relations between 22:6n 3 and 20:4n 6 status at birth and their relative postnatal changes, calculated by the difference between status at the end of the feeding period (6 wk of age) and at birth, were assessed. Results: Postnatal changes in the plasma and erythrocyte phospholipids 22:6n 3 and 20:4n 6 were negatively related to their respective concentrations at birth (P < 0.01) and the slopes of the regression lines were not significantly affected by the type of milk ingested. Adjusted mean values for phospholipid 22:6n 3 in nonsupplemented-formula fed infants and for 20:4n 6 in formula-fed infants decreased significantly more than they did in the other infant groups (P < 0.02). The status at birth and the type of milk ingested explained 33 64% and 7 47%, respectively, of the variability in postnatal changes. Conclusions: The status of 22:6n 3 and 20:4n 6 at birth in term infants is one of the major determinants of postnatal changes in these fatty acids. This finding indicates that research is required to characterize environmental, genetic, or both factors, which, in addition to maternal diet, could influence fatty acid status at birth. Am J Clin Nutr 1999;70: KEY WORDS Arachidonic acid, birth status, blood phospholipids, docosahexaenoic acid, postnatal changes, breast milk, fatty acids, term infants, France See corresponding editorial on page 181. INTRODUCTION Part of the rationale for supplementing infant formulas with n 6 and n 3 long-chain polyunsaturated fatty acids (PUFAs) is based on comparisons of changes in blood fatty acid composition over time between breast- and formula-fed infants. For instance, concentrations of the principal n 3 long-chain PUFA docosahexaenoic acid (22:6n 3) in plasma and erythrocyte phospholipids (ie, an indicator of 22:6n 3 status) remain the same, decrease slightly, or even increase during the course of early development in breast-fed infants, whereas a decrease is always reported in infants fed formula devoid of 22:6n 3 (reviewed in 1 3). Such differences over time are still observed, although the content of the n 3 precursor -linolenic acid (18:3n 3) in formula increased to up to 5% of total fatty acids (4, 5). Conversely, no significant differences were found when comparisons were made between infants fed breast milk and those fed 22:6n 3-supplemented formulas (1 3). Similar conclusions have been drawn for n 6 fatty acids and their main long-chain PUFA arachidonic acid (20:4n 6). In our recent study conducted in healthy term infants fed breast milk or formulas supplemented with or without 22:6n 3 (6), postnatal changes over time in the 22:6n 3 content of blood phospholipids were highly variable among infants fed the same type of milk, without any obvious relation to the 22:6n 3 content of the milk ingested. Such observations were also highlighted by others (7) and it was shown recently that infants synthesize 20:4n 6 and 22:6n 3 at different rates. These observations suggest that other environmental factors, genetic factors, or both, along with the fatty acid composition of the milk, affect postnatal variations in blood 22:6n 3 status (8). In preterm infants, it was shown recently that postnatal 22:6n 3 status was partly related to 22:6n 3 status at birth (9). A similar relation was also observed for 20:4n 6 (9). In the present study, we sought to determine whether the status of 22:6n 3 and 20:4n 6 in term infants at birth could also influence postnatal changes in 22:6n 3 1 From the Equipe Associée Lipides et Croissance, Université de Tours, Tours and Institut National de la Recherche Agronomique, Jouy-en-Josas; Centre de Recherche Clinique, Université de Tours; and Direction Scientifique, Groupe Danone, Paris. 2 Supported by a grant from Blédina-sa, Groupe Danone, Paris. PP-G was sponsored by the French Ministry of Cooperation in Mauritius and the University of Mauritius. 3 Address reprint requests to P Guesnet, Institut National de la Recherche Agronomique, Laboratoire de Nutrition et Sécurité Alimentaire, CRJ, Jouy-en-Josas Cedex, France. guesnet@jouy.inra.fr. Received October 2, Accepted for publication March 16, Am J Clin Nutr 1999;70: Printed in USA American Society for Clinical Nutrition

2 TABLE 1 Major fatty acid composition of infant formulas and breast milk LONG-CHAIN PUFAS AT BIRTH AND POSTNATAL CHANGES 293 Formulas 22:6n 3-supplemented Fatty acid Nonsupplemented High in 20:5n 3 Low in 20:5n 3 Breast milk 1 % by wt % by wt % by wt % by wt Total saturated ± 2.1 Total monounsaturated ± :2n ± :4n 6 ND ± :3n ± :5n 3 ND ± :6n 3 ND ± x ± SD of breast milk collected on day 42 of lactation (n = 15). 2 ND, not detected. and 20:4n 6, taking into consideration the type of diet. With this in mind, we studied the relation between blood concentrations of 22:6n 3 and 20:4n 6 at birth and their relative postnatal changes, ie, differences from birth to the end of the feeding period, making the most of a large and complete database obtained for term infants fed either human milk, nonsupplemented formula, or 2 different 22:6n 3-supplemented formulas (6). SUBJECTS AND METHODS Subjects and experimental design Ninety-eight singleton, healthy term infants born between the 37th and 42nd week of gestation were eligible for this randomized study if they had an appropriate weight-for-gestational age after a normal pregnancy (6). Their mean (±SD) weights, lengths, and head and arm circumferences were appropriate for their gestational age and were as follows: 3349 ± 409 g (interquartile range, 25th 75th percentile: g), 50 ± 2 cm (49 51 cm), 35 ± 1 cm (34 36 cm), and 10 ± 1 cm (10 10 cm), respectively. Subjects followed a 6-wk feeding protocol, which corresponded to the usual duration of exclusive breast-feeding in France, with either breast milk or one of 3 formulas if their mothers chose not to breast-feed. Eighty-three (43 boys and 40 girls) infants successfully completed the protocol. Fifteen infants were excluded because of inadequate blood sampling (n = 3) or follow-up (n =9) and intercurrent antibiotic treatment (n = 3). Of the 83 infants, 15 were breast-fed and 68 were formula-fed. Twenty-one subjects received a nonsupplemented formula and 23 and 24 subjects received 22:6n 3-supplemented formula with a high and low amount of eicosapentaenoic acid (20:5n 3), respectively (Table 1). Blood samples were collected from the umbilical cord at birth and by venipuncture at 6 wk of age. Infant growth was monitored and found normal in all 4 dietary groups and the amounts of formula ingested over the 6-wk period were similar in the 3 formula-fed groups (6). Written, informed parental consent was obtained before enrollment of the infants. The experimental design was approved by the Ethics Committee of the Medical School of Tours, France. Formulas The nutrient composition of the 3 formulas was that of a commercial infant formula marketed in France (Gallia 1; Blédina-sa, Groupe Danone, Villefranche-sur-saône, France), except that the fat blends were different in the two 22:6n 3-supplemented formulas. The 3 formulas contained approximately the same amounts of total saturated fatty acids, total monounsaturated fatty acids, n 6 fatty acids, and n 3 fatty acids (Table 1). The nonsupplemented formula contained only 18:3n 3 (1.65% by wt) and no 22:6n 3. The 2 supplemented formulas contained 0.45% by wt of 22:6n 3 after the addition of 2 different fish oils, 1 high (0.35%) in and 1 low (0.10%) in 20:5n 3. The predominant fatty acids of mature breast milk are shown in Table 1, of which the mean 22:6n 3 content was 0.14% by wt. Blood separation and fatty acid analysis After plasma and erythrocytes were separated by centrifugation, total lipids were extracted as described previously (6). For plasma, a phospholipid internal standard was added (6), which made it possible to quantify fatty acids and then adjust for differences in concentrations of circulating phospholipids among infants and as a function of age (10). Total plasma phospholipids were then isolated by thin-layer chromatography, and erythrocyte phosphatidylcholine (PC) and phosphatidylethanolamine (PE) were isolated by HPLC (6). Phospholipid fatty acid methyl esters were prepared and analyzed by capillary gas-liquid chromatography (11). The fatty acid composition was expressed as a percentage of total fatty acids (% by wt) for all blood lipid classes studied and as mol/l for total plasma phospholipids. Data analyses PUFA contents in total plasma phospholipids and in erythrocyte PC and PE of all infants at birth were determined as medians and interquartile ranges (25th 75th percentiles; n = 83) by using the STATVIEW SE+GRAPHICS program (Abacus Concepts Inc, Meylan, France). Relations between the contents of these fatty acids at birth and anthropometric measurements at birth (weight, length, and head and arm circumferences) were assessed by simple linear regression. Because 22:6n 3 and 20:4n 6 were the main focus of this study, we calculated differences between concentrations of 22:6n 3 and 20:4n 6 at birth and at the end of the 6-wk feeding period (ie, postnatal changes: 22:6n 3 and 20:4n 6, respectively) in each dietary group and for total plasma phospholipids and erythrocyte PC and PE. Comparisons between the 4 dietary groups were conducted by analysis of variance (ANOVA) with PROC GLM (version 6; SAS

3 294 GUESNET ET AL TABLE 2 Polyunsaturated fatty acid composition of total plasma phospholipids and of erythrocyte phosphatidylcholine (PC) and phosphatidylethanolamine (PE) in term infants at birth 1 Plasma phospholipids Erythrocyte phospholipids Fatty acid Percentage Concentration PC PE % by wt mol/l % by wt 18:2n ( ) 210 ( ) 8.7 ( ) 2.2 ( ) 20:4n ( ) 430 ( ) 11.7 ( ) 21.2 ( ) 22:4n ( ) 13 (11 15) 0.5 ( ) 6.2 ( ) 22:5n ( ) 16 (12 21) 0.3 ( ) 1.6 ( ) 20:5n ( ) 9 (6 13) 0.2 ( ) 0.3 ( ) 22:5n ( ) 10 (7 13) 0.3 ( ) 0.8 ( ) 22:6n ( ) 164 ( ) 3.6 ( ) 8.5 ( ) 20:3n ( ) 8 (6 10) 0.2 ( ) 0.6 ( ) Ratio of 20:3n 6 to 20:4n ( ) 0.03 ( ) 0.03 ( ) Ratio of 22:5n 3 to 22:6n ( ) 0.11 ( ) 0.19 ( ) 1 Median; interquartile range (25th 75th percentile) in parentheses. n = 83. To convert fatty acid concentrations from mol/l to mg/l, multiply by for 18:2n 6, by for 20:4n 6, by for 22:4n 6, by for 22:5n 6 and 22:5n 3, by for 20:5n 3, by for 22:6n 3, and by for 20:3n 9. Institute Inc, Cary, NC). Means were compared by Tukey s test with a level of significance set at P < The relation between postnatal variation and birth concentration was also analyzed in each dietary group and for total plasma phospholipids and erythrocyte PC and PE by using linear regression. The parallelism of the regression lines associated with the different dietary groups was assessed with an F test. On the statistical assumption of no inequality between slopes (ie, a nonsignificant result based on the F test), an analysis of covariance (ANCOVA) was performed to estimate the common slope and to assess the effects of the dietary groups. Pairwise comparisons of the adjusted means (ie, mean postnatal variation within a group, adjusted to the birth concentration) were thereby made. According to Bonferroni s correction, the significance level was considered to be P < 0.05/6, ie, P < These analyses were also performed by using PROC GLM. Relations between n 3 and n 6 fatty acid contents at birth and postnatal variations in anthropometric variables were explored by simple linear regression. RESULTS Fatty acid status at birth Median concentrations and percentages (by wt), as well as interquartile ranges, of major PUFAs in total plasma phospholipids and erythrocyte PC and PE of infants at birth are given in Table 2. The predominant fatty acid was 20:4n 6: 11.7% by wt in PC and >20% by wt in PE and total plasma phospholipids. The second and third most predominant fatty acids were 18:2n 6 and 22:6n 3, except in PE, for which they accounted for only 2.2% and 8.5% by wt, respectively. In general, the contents of 22:4n 6, 22:5n 6, 20:5n 3, and 22:5n 3 were low (<0.9% by wt), except for the contents of 22:4n 6 and 22:5n 6 in PE (6.2% and 1.6% by wt, respectively). The parent of the n 3 fatty acid family, 18:3n 3, was not reported because it never exceeded 0.1% of total fatty acids. The Mead acid (20:3n 9) content, an indicator of PUFA deficiency, was also low (<0.7% by wt). Ratios of 20:3n 9 to 20:4n 6 and of 22:5n 6 to 22:6n 3, indexes of n 6 and n 3 fatty acid status, were and , respectively. The interquartile ranges of the 3 predominant fatty acids were similar, with values ranging around median concentrations from 10% (20:4n 6 in erythrocyte PE) to 43% (20:4n 6 in total plasma phospholipids expressed as % by wt). There was no significant relation between any anthropometric index at birth and any blood lipid PUFA concentration at birth. Postnatal changes in 22:6n 3 and 20:4n 6 status For each infant group considered, 22:6n 3 and 20:4n 6 in total plasma phospholipids and erythrocyte PC and PE are presented in Figure 1. For 22:6n 3, the largest postnatal declines in all phospholipids were in infants fed nonsupplemented formula. In the other 3 groups, 22:6n 3 in total plasma phospholipids (% by wt) and erythrocyte PC decreased slightly, and there were no significant differences between these groups. In total plasma phospholipids ( mol/l), 22:6n 3 increased, but was not significantly different between infants fed breast milk and those fed 22:6n 3-supplemented formulas. In erythrocyte PE, 22:6n 3 increased in breast-fed infants and decreased in the supplemented formula-fed groups; the differences were significant, except in the group fed formula low in 20:5n 3. Over time, 20:4n 6 decreased in all 3 formula-fed groups and differed significantly from the changes observed in the breastfed group, except in erythrocyte PC. The magnitude of the decrease in 20:4n 6 did not differ between formula-fed groups when erythrocyte PC and PE were considered. In total plasma phospholipids, the postnatal decrease in 20:4n 6 was significantly greater in the 2 supplemented formula-fed groups than in the nonsupplemented-formula group. Relation between long-chain PUFA status at birth and its postnatal variation The relation between the long-chain PUFA concentrations 22:6n 3 and 20:4n 6 at birth and their respective postnatal changes in total plasma phospholipids and erythrocyte PC and PE in each dietary group are depicted in Figures 2 and 3. There was a significant, negative correlation between 22:6n 3 and 20:4n 6 and their respective concentrations at birth (P < 0.01). The F test showed that the slopes of the regression lines were not significantly different among the 4 diet groups (0.91 > P > 0.14) and the common slopes as well as the adjusted means calculated

4 LONG-CHAIN PUFAS AT BIRTH AND POSTNATAL CHANGES 295 FIGURE 1. Mean (± SEM) postnatal changes ( ) in 22:6n 3 and 20:4n 6 contents in total plasma phospholipids and erythrocyte phosphatidylcholine (PC) and phosphatidylethanolamine (PE) in term infants fed nonsupplemented formula ( ; n = 21), 22:6n 3-supplemented formula with a high ( ; n = 23) or low ( ; n = 24) content of 20:5n 3, or breast milk ( ; n = 15). 22:6n 3 and 20:4n 6 were calculated as the difference between the concentration of each at birth and that at 6 wk of age. Values with different letters are significantly different, P < 0.05 (ANOVA). FIGURE 2. Relations between postnatal changes ( ) in 22:6n 3 and 22:6n 3 contents at birth in total plasma phospholipids and erythrocyte phosphatidylcholine (PC) and phosphatidylethanolamine (PE) in term infants fed nonsupplemented formula ( ; n = 21), 22:6n 3-supplemented formula with a high ( ; n = 23) or low ( ; n = 24) content of 20:5n 3, or breast milk ( ; n = 15). 22:6 n 3 was calculated as the difference between the concentration of 22:6n 3 at birth and that at 6 wk of age. Relations were as follows for total plasma phospholipids (% by wt and mol/l) and erythrocyte PC and PE, respectively: y = 0.94x (r = 0.98, P < ), y = 1.04x (r = 0.86, P < ), y = 1.06x (r = 0.93, P < 0.001), and y = 0.90x (r = 0.65, P < 0.01) in ; y = 1.00x (r = 0.89, P < ), y = 1.11x (r = 0.73, P < 0.001), y = 0.73x (r = 0.79, P < 0.001), and y = 0.52x (r = 0.60, P < 0.01) in ; y = 0.80x (r = 0.83, P < 0.001), y = 0.83x (r = 0.72, P < 0.001), y = 1.13x (r = 0.67, P < 0.001), and y = 0.67x (r = 0.63, P < 0.005) in ; and y = 0.82x (r = 0.68, P < 0.01), y = 0.88x (r = 0.68, P < 0.01), y = 0.95x (r = 0.77, P < 0.001), and y = 0.60x (r = 0.80, P < 0.001) in.

5 296 GUESNET ET AL FIGURE 3. Relations between postnatal changes ( ) in 20:4n 6 and 20:4n 6 contents at birth in total plasma phospholipids and erythrocyte phosphatidylcholine (PC) and phosphatidylethanolamine (PE) in term infants fed nonsupplemented formula ( ; n = 21), 22:6n 3-supplemented formula with a high ( ; n = 23) or low ( ; n = 24) content of 20:5n 3, or breast milk ( ; n = 15). 20:4n 6 was calculated as the difference between the concentration of 20:4n 6 at birth and that at 6 wk of age. Relations were as follows for total plasma phospholipids (% by wt and mol/l) and erythrocyte PC and PE, respectively: y = 0.87x (r = 0.93, P < 0.001), y = 1.13x (r = 0.72, P < 0.005), y = 1.21x (r = 0.85, P < 0.001), and y = 0.70x (r = 0.53, P < 0.01) in ; y = 0.95x (r = 0.81, P < 0.001), y = 0.96x (r = 0.91, P < 0.001), y = 0.71x (r = 0.81, P < 0.001), and y = 0.72x (r = 0.67, P < 0.005) in ; y = 0.96x (r = 0.89, P < 0.001), y = 1.69x (r = 0.72, P < 0.005), y = 1.10x (r = 0.80, P < 0.001), and y = 0.76x (r = 0.46, P < 0.05) in ; and y = 0.77x (r = 0.67, P < 0.01), y = 0.74x (r = 0.63, P < 0.01), y = 1.05x (r = 0.91, P < 0.001), and y = 0.91x (r = 0.77, P < 0.01) in. from ANCOVA are displayed in Tables 3 and 4, respectively. For 22:6n 3, the adjusted means were negative and significantly lower in the nonsupplemented-formula group than in the other 3 groups (P < ; Table 4). Such differences indicated a greater postnatal decline of 22:6n 3 in the nonsupplementedformula group, although the power of the relation between 22:6n 3 and 22:6n 3 at birth was similar among the 4 groups (same slope). In contrast, the adjusted means did not differ TABLE 3 Common slope (±SD) of the 4 infant groups, calculated by ANCOVA, for the relations between postnatal changes ( ) in 22:6n 3 and 20:4n 6 and their respective contents in total plasma phospholipids and erythrocyte phosphatidylcholine (PC) and phosphatidylethanolamine (PE) at birth 1 22:6n 3 20:4n 6 Plasma phospholipids (% by wt) 0.90 ± ± 0.07 ( mol/l) 0.93 ± ± 0.09 Erythrocyte PC (% by wt) 0.94 ± ± 0.08 Erythrocyte PE (% by wt) 0.74 ± ± n = :6n 3 and 20:4n 6 were calculated as the difference between the concentration of each at birth and that at 6 wk of age. significantly between the 3 infant groups receiving dietary 22:6n 3 (ie, the breast milk group and the 22:6n 3-supplemented formula groups). ANCOVA showed that 34 52% of the variability in 22:6n 3 was explained by the concentration of 22:6n 3 at birth, and 13 32% by the type of milk ingested (Table 5). The adjusted means for 20:4n 6 were negative and lower in infants fed formula than in infants fed breast milk (P < 0.011; Table 4), except for total plasma phospholipids, for which adjusted means were most negative in infants fed the 22:6n 3- supplemented formulas (P < 0.001). From 33% to 64% of the variability in postnatal changes in 20:4n 6 was explained by the concentration of 20:4n 6 at birth and from 7% to 47% by the type of diet (Table 5). Our results showed that there was a negative relation between postnatal gains in length and 20:5n 3 concentrations at birth in total plasma phospholipids, expressed as both % by wt (r = 0.23) and mol/l (r = 0.22), and in erythrocyte PE (r = 0.24). DISCUSSION The aim of the present study was to look at factors that, in addition to milk fatty acid composition, could contribute to the

6 LONG-CHAIN PUFAS AT BIRTH AND POSTNATAL CHANGES 297 TABLE 4 Relations between postnatal changes ( ) in 22:6n 3 and 20:4n 6 and their respective contents in total plasma phospholipids and erythrocyte phosphatidylcholine (PC) and phosphatidylethanolamine (PE) at birth in the breast-fed and 3 formula-fed infant groups 1 Formulas 22:6n 3-supplemented Nonsupplemented High in 20:5n 3 Low in 20:5n 3 Breast milk Fatty acid (n = 21) (n = 23) (n = 24) (n = 15) Plasma phospholipids (% by wt) 22:6n ± 0.94 a 2.22 ± 1.01 b 1.87 ± 0.93 b 2.02 ± 0.93 b 20:4n ± 1.60 a ± 1.63 b ± 1.52 b 7.26 ± 1.51 c ( mol/l) 22:6n ± 47.0 a 36.6 ± 47.9 b 25.4 ± 48.2 b 48.5 ± 44.3 b 20:4n ± a ± b ± b 63.6 ± c Erythrocyte PC (% by wt) 22:6n ± 1.10 a 0.81 ± 1.06 b 0.86 ± 1.08 b 0.82 ± 1.05 b 20:4n ± 2.34 a 6.36 ± 2.25 a 5.47 ± 2.30 a 2.60 ± 2.29 b Erythrocyte PE (% by wt) 22:6n ± 1.33 a 0.52 ± 1.25 b 0.39 ± 0.27 b 0.14 ± 1.32 b 20:4n ± 2.80 a 3.54 ± 2.73 a 1.89 ± 0.58 a 0.47 ± 2.67 b 1 Adjusted means ± SDs. 22:6n 3 and 20:4n 6 were calculated as the difference between the concentration of each at birth and that at 6 wk of age. Values in the same row with different superscript letters are significantly different, P < (ANOVA with Bonferroni s correction). To convert fatty acid concentrations from mol/l to g/l, multiply by for 20:4n 6 and by for 22:6n 3. interindividual variability in long-chain PUFA concentrations in plasma and erythrocyte phospholipids in infants fed the same milk. We found that, in term infants from all groups showing no biochemical signs of essential fatty acid deficiency, postnatal changes in 22:6n 3 and 20:4n 6 in plasma and erythrocyte phospholipids were negatively correlated with their respective concentrations at birth, whereas dietary intakes of n 3 and n 6 fatty acids were variable. These data, obtained from a large number of subjects (n = 83), indicate that the long-chain PUFA status at birth in term infants is one of the major determinants of postnatal change in long-chain PUFA status. Together, birth status and diet explained 50 85% of the early postnatal changes in 22:6n 3 and 20:4n 6 blood concentrations. In preterm infants fed breast milk or a standard formula, Foreman-van Drongelen et al (9) showed that the 22:6n 3 status in preterm infants (32 wk postconceptional) and 8 wk after birth (ie, the age equivalent to term) were related. More precisely, they noted that the concentration of 22:6n 3 in umbilical artery wall phospholipids determined its concentration in erythrocyte phospholipids at term, indicating that a higher 22:6n 3 status at birth predetermines a higher 22:6n 3 status at term. On the other hand, the type of feeding was the sole determining factor for total plasma phospholipids at the equivalent gestational age at term. A similar influence of the birth status was also shown for 20:4n 6 in preterm infants (9) and recently in term infants (12). On the basis of these data, we speculated that the fetal accretion of high amounts of long-chain PUFAs up until birth helps maintain a high long-chain PUFA concentration postnatally. However, our results showed clearly that the higher the long-chain PUFA status at birth, the greater the postnatal changes. Together, these 2 studies highlight the importance and complexity of the influence that long-chain PUFA status at birth has on its evolution during the early stages of life. Gestational age at birth, growth indexes at birth, birth order, and length of labor were reported previously in the literature to be a few of the environmental factors that could modulate the longchain PUFA status at birth and explain its large variability. For instance, gestational age at birth is positively related to longchain PUFA concentrations at birth (9, 13). This agrees with increasing long-chain PUFA concentrations in fetal circulation during the third trimester of pregnancy (14). In a study conducted in preterm infants (9), some relations were reported at birth between infants birth weights, head circumferences, and concentrations of 22:6n 3 in umbilical artery wall phospholipids, but not in plasma and erythrocyte phospholipids. Similarly, in our study of term infants, there was no relation between anthropometric indexes and long-chain PUFA concentrations in plasma or in erythrocyte phospholipids. To date, the functional relevance of TABLE 5 Relative contribution of the type of diet and of the fatty acid status at birth to the total variability in postnatal changes ( ) in 22:6n 3 and 20:4n 6 and their respective contents in total plasma phospholipids and erythrocyte phosphatidylcholine (PC) and phosphatidylethanolamine (PE) in term infants in the 4 groups 1 Fatty acid status at birth Type of diet Plasma phospholipids (% by wt) 22:6n :4n ( mol/l) 22:6n :4n Erythrocyte PC (% by wt) 22:6n :4n Erythrocyte PE (% by wt) 22:6n :4n n = :6n 3 and 20:4n 6 were calculated as the difference between the concentration of each at birth and that at 6 wk of age. ANCOVA was used to determine variability.

7 298 GUESNET ET AL the variability in long-chain PUFA concentrations in blood lipids at birth is unknown. The observed negative relation between 20:5n 3 status at birth and postnatal changes in length could be related to the deleterious effect of marine-oil supplementation on growth and 20:4n 6 blood status in preterm infants as reported by Carlson et al (15). However no relation was noted between 20:4n 6 and growth indexes in the present study. In the present study, the effect of long-chain PUFA concentrations at birth on postnatal changes was similar regardless of the composition of the formula or milk supplied, as evidenced by the similar slopes of the regression lines. However, the adjusted means, calculated on the basis of a common slope, indicated differences between the formulas and milk. For all phospholipids, a significantly lower negative adjusted mean was noted in the group fed the nonsupplemented formula than in the other 3 groups. Thus, the absence of 22:6n 3 in the nonsupplemented formula resulted in a larger decrease in 22:6n 3 values during the early stages of life for the same initial birth concentration. The adjusted means for 20:4n 6 were negative and always lower in the nonsupplementedand supplemented-formula groups than in the breast-milk group, which was the only 1 of the 4 products providing 20:4n 6. Together, the diet and the status of 22:6n 3 and 20:4n 6 at birth explained 50 85% of the total variance in postnatal changes, depending on the phospholipid studied and the fatty acid considered. With respect to infant nutrition, complementary studies focusing on environmental factors, genetic factors, or both which, in addition to maternal diet, could influence the long-chain PUFA status at birth, would prove valuable to the extensive research devoted to designing the fatty acid composition of formulas. We are grateful to Françoise Houlier and Alain Linard for their technical assistance and to Anne-Marie Wall for revising the English in the manuscript (Service Linguistique, Unité Centrale de Documentation, Jouy-en-Josas). REFERENCES 1. Bendich A, Brock PE. Rationale for the introduction of long chain polyunsaturated fatty acids and for concomitant increase in the level of vitamin E in infants formulas. Int J Vitam Nutr Res 1997;67: Carlson S. Long-chain polyunsaturated fatty acid supplementation of preterm infants. In: Dobbing J, ed. Developing brain and behavior. The role of lipids in infant formulas. San Diego: Academic Press, 1997: Innis S. Polyunsaturated fatty acid nutrition in infants born at term. In: Dobbing J, ed. Developing brain and behavior. The role of lipids in infant formulas. San Diego: Academic Press, 1997: Innis S, Auestad N, Siegman J. Blood lipid docosahexaenoic and arachidonic acids in term gestation infants fed formulas with high docosahexaenoic acid, low eicosapentaenoic acid fish oil. Lipids 1996;31: Innis S, Akrabawi S, Diersen-Schade D, Dobson V, Guy D. Visual acuity and blood lipids in term infants fed human milk or formulae. Lipids 1997;32: Maurage C, Guesnet P, Pinault M, et al. Effect of two types of fish oil supplementation on plasma and erythrocyte phospholipids in formula-fed term infants. Biol Neonate 1998;74: Carlson S, Rhodes P, Rao V, Goldgar D. Effect of fish oil supplementation on the n 3 fatty acid content of red blood cell membranes in preterm infants. Pediatr Res 1987;21: Carlson S, Cooke R, Rhodes P, Peeples J, Werkman S. Effect of vegetable and marine oils in preterm infant formulas on blood arachidonic and docosahexaenoic acids. J Pediatr 1992;120:S Foreman-van Drongelen M, Van Houwellingen A, Kester A, Hasaart T, Blanco C, Hornstra G. Long-term polyunsaturated fatty acids in preterm infants: status at birth and its influence on postnatal levels. J Pediatr 1995;126: Foreman-van Drongelen M, Van Houwellingen A, Kester A, et al. Long-chain polyene status of preterm infants with regard to the fatty acid composition of their diet: comparison between absolute and relative fatty acid levels in plasma and erythrocyte phospholipids. Br J Nutr 1995;73: Guesnet P, Antoine JM, Rochette de Lempdes JB, Galent A, Durand G. Polyunsaturated fatty acid composition of human milk in France: changes during the course of lactation and regional differences. Eur J Clin Nutr 1993;47: Foreman-van Drongelen M, Van Houwellingen A, Koopman-Esseboom C, Kester A, Hornstra G. Does the long-chain polyunsaturated fatty acid (LCPUFA) status at birth affect the postnatal LCPUFA status? In: Galli C, Simopoulos A, Tremoli E, eds. Fatty acids and lipids: biological aspects. World Rev Nutr Diet 1994;76: Van Houwellingen A, Sorensen J, Hornstra G, et al. Essential fatty acid status in neonates after fish-oil supplementation during late pregnancy. Br J Nutr 1995;74: Leaf A, Leighfield M, Ghebremeskel K, Costeloe K, Crawford M. Long-chain polyunsaturated fatty acids and fetal growth. Early Hum Dev 1992;30: Carlson S, Werkman S, Peeples J, Cooke R, Tolley E. Arachidonic acid status correlates with the first year growth in preterm infants. Proc Natl Acad Sci U S A 1993;90: