A major area of concern in infant lipid nutrition is the effect of the dietary fatty acid composition of arachidonic acid [20:4(n-6)] and docosahexaenoic acid [22:6(n-3)] on the brain and retina (Innis 1991). The pathways by which the brain normally acquires 20:4(n-6) and 22:6(n-3) are not well understood. Potential pathways include uptake of 20:4(n-6) and 22:6(n-3) from plasma, possibly from lipoprotein phospholipid, triglyceride or cholesteryl esters, or albumin-bound unesterified fatty acids or lysophospholipids (Innis 1991). Alternatively, or in addition, 20:4(n-6) and 22:6(n-3) could be derived by synthesis in the brain and retina by uptake, then desaturation and elongation of linoleic acid [18:2(n-6)] and -linolenic acid [18:3(n-3), respectively (Moore et al. 1991). Dietary deficiency of 18:2(n-6) and 18:3(n-3) is known to result in reduced levels of 20:4(n-6) and 22:6(n-3), respectively, in the developing brain, as well as in plasma, red blood cell and other tissue phospholipids (Bourre et al. 1989, Carlson et al. 1986, Hrboticky et al. 1990, Neuringer et al. 1984). Studies in animals have shown increasing brain and retina levels of 20:4(n-6) and 22:6(n-3) with increasing dietary intakes of 18:2(n-6) and 18:3(n-3), respectively, until a plateau is reached at about 2.4% dietary energy 18:2(n-6) and about 0.7% dietary energy 18:3(n-3) (Arbuckle et al. 1992 and 1994, Bourre et al. 1989 and 1990, Hrboticky et al. 1990 and 1991). Further increases in the intake of 18:2(n-6) and 18:3(n-3) do not seem to result in increased plasma or tissue levels of 20:4(n-6) or 22:6(n-3), respectively. Increasing intakes of 20:4(n-6) and 22:6(n-3), on the other hand, are accompanied by a progressive increase in plasma phospholipid 20:4(n-6) and 22:6(n-3), respectively (Arbuckle et al. 1991, Innis and Hansen 1996). Because of this, it is difficult to interpret the physiological importance of differences in plasma phospholipid (n-6) and (n-3) fatty acids between infants or animals receiving diets differing in 20:4(n-6) and 22:6(n-3) content.

Infants fed formula with 18:2(n-6) and 18:3(n-3) but no 20:4(n-6) or 22:6(n-3) have lower plasma phospholipid levels of 20:4(n-6) and 22:6(n-3) than infants who are breast-fed (Innis et al. 1994, Ponder et al. 1992). This seems to be explained, at least in part, by the small amounts of 20:4(n-6) and 22:6(n-3) in human milk. This explanation is supported by results of studies showing that plasma phospholipid levels of 20:4(n-6) and 22:6(n-3) are higher in infants fed formula containing 20:4(n-6) and 22:6(n-3) than in infants fed formula without these fatty acids (Carlson et al. 1991, Decsi and Koletzko 1995, Makrides et al. 1995). In contrast to plasma, the brain and retina are very resistant to changes in fatty acid composition under conditions of adequate dietary 18:2(n-6) and 18:3(n-3) intake (Innis 1991). Despite this, the analysis of plasma phospholipid fatty acids remains an integral part of many studies on dietary (n-6) and (n-3) fatty acid requirements for growth and development.

Investigation of the effects of diet on the composition of plasma phospholipid fatty acids in relation to the effects on brain, liver or other organs necessitates the feeding of milk or formula with a range of different fatty acid compositions as well as methods of tissue sampling impossible in humans. This paper describes analyses of results from studies with piglets fed with milk or one of seven formulas, differing only in fat composition, undertaken to determine the usefulness of plasma phospholipid fatty acids as predictors of diet-related changes in brain and liver phospholipid fatty acids. Some studies have indicated that dietary fat influences the composition of bile lipids (Berr et al. 1992, Rioux and Innis 1993), but the effects of particular dietary fatty acids are poorly understood. These studies, therefore, also included collection and analysis of bile from piglets fed milk or formula. Piglets were used because of the similarities in milk and changes in plasma phospholipid fatty acids in response to formula and milk feeding between pigs and humans. Results concerning the effects of some of these formulas on plasma lipids and growth have been published (Innis et al. 1993).

MATERIAL AND METHODS

Table 1. Fatty acid composition of sow milk and formulas [View Table]

Bile phospholipid fatty acids may originate from a pre-existing or a newly formed pool in the liver, or from plasma lipoproteins. An underlying relation between the composition of liver and plasma phospholipid fatty acids could, therefore, result in a significant, although indirect relation between the composition of bile and plasma fatty acids. To explore this further, the relation between the proportion of each fatty acid in the bile with that in the liver and plasma phospholipids was assessed by fitting a regression plane to the bile fatty acids using the liver and plasma as independent variables. The relation between the level of a particular fatty acid in the bile and liver, in the presence of the relation to the plasma phospholipid, was then examined by determining the statistical significance of the slope of the regression of bile with respect to the liver axis. Similarly, the relation of the level of a particular fatty acid in the bile to the plasma phospholipid was examined from the slope of the regression of the bile with respect to the plasma. A partial F-test, which is equivalent to the significance of the respective coefficient in the linear regression model, was used to determine the significance of the associations. The results for piglets fed Formula R with synthesized triacylglycerols and piglets fed sow milk were not included in this analysis because of the possibility that triacylglycerol fatty acid distribution, cholesterol or other factors in milk not present in the formula might alter the secretion of bile lipids. The statistical computer package using the GLM procedure in the Number Cruncher Statistical System, version 5.1 (Kaysville, UT) was used for all regression calculations and the ANOVA. The plotting package SIGMAPLOT, DOS version (Jandel Scientific, San Rafael, CA) was used to construct the regression plane figures.

RESULTS

The sow milk and formula fatty acids contained only 2.3-5.6% 18:0. Despite this, 18:0 represented 23-30% of the fatty acids in the piglet plasma phospholipids (Fig. 1). Piglets fed the five formulas low in 16:0 (Formula A-E) had significantly higher plasma phospholipid levels of 18:0 than piglets fed sow milk or a formula high in 16:0 (Formulas F and R). Piglets fed Formulas F and R, which had levels of 16:0 similar to milk, had plasma phospholipid levels of 18:0 which were not different from those of piglets fed sow milk (Fig. 1). These results suggest that the dietary intake of 16:0 influences plasma phospholipid, and possibly endogenous synthesis of 18:0.

The piglet plasma phospholipid levels of 18:1 and 18:2(n-6) were also influenced by the amount of the respective fatty acid, and by the composition of the other fatty acids in the formula. Piglets fed Formula A or B with about 11% 18:1 had significantly lower plasma phospholipid levels of 18:1 than piglets fed sow milk or Formula C, D, E, F or R with 35-41% 18:1 (Fig. 1). However, piglets fed Formula C, D, E, F or R had significantly higher plasma phospholipid levels of 18:1 than piglets fed sow milk, even though the milk fatty acids also had about 40% 18:1. The significantly higher plasma phospholipid 18:2(n-6) in the piglets fed formula than in those fed sow milk may be due in part to the higher 18:2(n-6) in the formula (15.6-23.7% fatty acids) than in milk (8.2% fatty acids). As for 18:1, however, the differences in plasma phospholipid 18:2(n-6) among the groups fed formula show that the relation between dietary 18:2(n-6) intake and plasma phospholipid 18:2(n-6) is not simple. For example, piglets fed Formula C with 23.7% 18:2(n-6) had a significantly lower plasma phospholipid 18:2(n-6) than piglets fed Formula A with 21.7% 18:2(n-6), but a similar plasma phospholipid 18:2(n-6) to piglets fed Formula R which had 16.4% 18:2(n-6).

The plasma phospholipid levels of 20:4(n-6) were significantly lower in all the piglets fed formula than in the piglets fed sow milk (Fig. 1). The plasma phospholipid levels of 22:6(n-3), on the other hand, were significantly lower in piglets fed Formulas A and C, but not in those fed the other formulas, than in piglets fed sow milk. Piglets fed Formula B containing 22:6(n-3) from fish oil had a significantly higher plasma phospholipid 22:6(n-3) than those fed sow milk or the other formulas, and a significantly lower plasma phospholipid 20:4(n-6) than piglets fed Formula E, F or R. 

Table 2. Relations between the levels of individual major fatty acids in the liver, brain or bile and plasma phospholipids of piglets [View Table]

The 95% prediction limits shown in Figure 2 give an estimate of the precision with which the equation for the regression lines (Table 2) can be used to estimate the enrichment of 16:0, 18:0, 18:1, 18:2(n-6), 20:4(n-6) and 22:6(n-3) in the liver phospholipid from analyzed levels of the same fatty acid in the plasma phospholipids. The probability that the enrichment in liver calculated from these equations would fall outside the prediction limits is <5%. It should be noted that the width of the prediction limits is much narrower for 18:1 and 18:2(n-6) than for 20:4(n-6) (Fig. 2).

Although the fatty acid composition of the brain and plasma phospholipids showed no significant relation, differences in the brain levels of some fatty acids were found among the milk- and formula-fed piglets. The brain fatty acids of piglets fed sow milk, or Formula A, B, C, D, E, F or R had (mean ± SEM, n = 6-8/group) 11.3 ± 0.1, 11.2 ± 0.1, 10.5 ± 0.1, 10.9 ± 0.2, 10.4 ± 0.2, 10.9 ± 0.2, 11.0 ± 0.2 and 10.8 ± 0.1% 20:4(n-6), and 9.2 ± 0.4, 9.7 ± 0.1, 10.7 ± 0.2, 9.7 ± 0.2, 10.5 ± 0.2, 10.3 ± 0.2, 10.3 ± 0.4 and 9.8 ± 0.4% 22:6(n-3), respectively. The proportion of 20:4(n-6) in the brain of piglets fed Formula B, which contained fish oil, was significantly lower and 22:6(n-3) was significantly higher than in piglets fed either the same formula without fish oil (Formula A), or sow milk, P < 0.05. 

DISCUSSION

Several recent studies have shown that the inclusion of small amounts of 20:4(n-6) in formula increases the plasma phospholipid 20:4(n-6) in infants fed formula (Decsi and Koletzko 1995). Based on the results in the study reported here, it is reasonable to expect that inclusion of 20:4(n-6) in formula will also increase 20:4(n-6) in liver phospholipids. Similarly, the significant decrease in the plasma phospholipid 20:4(n-6) of piglets fed formula with fish oil was accompanied by decreased liver phospholipid 20:4(n-6). Studies with premature infants fed formula containing 20:5(n-3) and 22:6(n-3) from fish oil, at levels similar to those fed to piglets here, also found lower plasma phospholipid levels of 20:4(n-6) than in infants fed formula without fish oil (Carlson et al. 1991). The studies reported here suggest that the alterations to plasma phospholipid fatty acids invoked by these small amounts of fish oil in formula are likely to extend to the liver and possibly other organs as well.

The piglets fed sow milk in the studies reported here had relatively low plasma phospholipid levels of 22:6(n-3), probably as a result of the absence of 22:6(n-3) in the sows' diets and the low sow milk concentration of 22:6(n-3) (about 0.1% milk fatty acids). This probably explains the lack of consistent difference in plasma phospholipid 22:6(n-3) between the piglets fed milk and those fed formula. Previous studies have shown that inclusion of 22:6(n-3) in sow diets results in increased 22:6(n-3) in milk, and in the plasma and liver phospholipids of sow milk-fed piglets (Arbuckle and Innis 1993). Similar to this effect in pigs, Sanders and Reddy (1992) reported that the milk fatty acids of women with vegan diets had 0.1 % 22:6(n-3) compared with about 0.4 % 22:6(n-3) in the milk of women who follow mixed diets. In the same study, the erythrocyte lipids of infants breast-fed by the vegan mothers had 1.9% 22:6(n-3), those breast-fed by mothers with mixed diets had 6.2% 22:6n-3, and infants fed a formula with 0.24% energy 18:3(n-3) and no 22:6(n-3) had about 3.7% 22:6(n-3) (Sanders and Reddy 1992). This information suggests that differences in blood lipid 22:6(n-3) between infants fed milk and formula depend in part on the amount of 22:6(n-3) in the milk fed.

Interest in the relation between the (n-6) and (n-3) composition of the diet and that of the growing brain has come in part from studies showing reduced visual function and altered behaviors in animals fed diets deficient in 18:3(n-3) (Bourre et al. 1989, Innis 1991, Neuringer et al. 1984). Functional changes in animals fed diets deficient in 18:3(n-3) are generally considered to be explained by the decrease in brain and/or retina levels of 22:6(n-3) which result from dietary 18:3(n-3) deficiency. In contrast to the liver, no significant relation was found between the plasma phospholipid levels of 16:0, 18:0, 18:1, 18:2(n-6), 20:4(n-6) or 22:6(n-3) and the level of the same fatty acid in the brain in these studies with piglets fed milk or formula containing 15-24 g/100 g fatty acids as 18:2(n-6) and 2.9-3.4% as 18:3(n-3). This finding suggests that with these dietary intakes, brain fatty acid uptake and/or synthesis and incorporation were largely independent of the fatty acid compositions of circulating plasma phospholipids. Autopsy studies have found similar or higher levels of 20:4(n-6) in brain phospholipids of infants fed formula without 20:4(n-6) than in breast-fed infants (Farquharson et al. 1992 and 1995, Makrides et al. 1994). At the same time, plasma phospholipid levels of 20:4(n-6) are lower in infants fed formula than in breast-fed infants (Innis et al. 1994, Ponder et al. 1992). The results of the studies with piglets reported here, which show no difference in brain 20:4(n-6) despite a more than threefold variation in plasma phospholipid 20:4(n-6), confirm this lack of relation between brain and plasma phospholipid 20:4(n-6). It seems reasonable that several pathways, including uptake of preformed 20:4(n-6) and 22:6(n-3) and uptake, desaturation and elongation of 18:2(n-6) and 18:3(n-3) derived from different plasma lipid sources may be available to help ensure the supply of 20:4(n-6) and 22:6(n-3) to the developing brain. Such a situation is teleologically reasonable for metabolites of crucial importance to optimal brain function. However, this would limit the usefulness of analysis of plasma fatty acids as an index of brain fatty acids except in situations of marked or prolonged dietary deficiency, or supplementations which alter normal pathways of 20:4(n-6) and 22:6(n-3) uptake.

Lower levels of 22:6(n-3) have been found in the brain of infants who had been fed formula with 0.5 to <1.6 g/100 g fatty acids as 18:3(n-3) than in infants who had been breast-fed (Farquharson et al. 1992 and 1995, Makrides et al. 1994). Low brain levels of 22:6(n-3) could be explained by inadequate 18:3(n-3), a high (n-6)/(n-3) content in the formula resulting in reduced synthesis or uptake of 22:6(n-3), or infants could require a dietary source of 22:6(n-3) to meet the needs of the developing brain. The formula fed to piglets in the studies here contained 2.9-3.4% 18:3(n-3), thus providing higher amounts of 18:3(n-3) than in the formulas associated with reduced brain 22:6(n-3) in formula-fed human infants. Previous studies to show that brain 22:6(n-3) is lower in piglets fed formula with 1 rather than 4% 18:3(n-3), or sow milk (Arbuckle et al. 1994) suggest that the dietary intake of 18:3(n-3) may be important. Alternatively, the absence of a difference in brain 22:6(n-3) between piglets fed sow milk and piglets fed formula with about 2.9-3.4% 18:3(n-3) and no 22:6(n-3) could be explained by inadequate 22:6(n-3) in the sow milk (0.1 g/100 g milk fatty acids). It is not known if the range of 22:6(n-3) in human milk from about 0.1 to 0.9 g/100 g fatty acids (Innis 1992) is accompanied by physiologically important differences in brain 22:6(n-3) in breast-fed infants. As in previous studies (Arbuckle et al. 1991), piglets fed formula with 22:6(n-3) from fish oil had higher 22:6(n-3) and lower 20:4(n-6) in liver and brain than piglets fed the same formula without fish oil. Whether providing 22:6(n-3) from fish oil triglycerides results in a similar pattern of tissue 22:6(n-3) assimilation to that achieved with milk feeding is not known. However, reports of lower growth (Carlson et al. 1992) and scores on some developmental tests (Carlson et al. 1993, Janowsky et al. 1995) in infants fed formula with rather than without 22:6(n-3) from fish oil suggests that important differences may occur.

Bile phospholipids have an important role in solubilizing bile cholesterol and in providing an endogenous though not obligatory source of phospholipids and fatty acids for intestinal lipoprotein synthesis (Coleman and Rahman 1992, Jüngst et al. 1993, Nalbone et al. 1974, Pavero et al. 1992). Relatively little is known about the effect of dietary fatty acids on bile fatty acids. The study reported here provides new information to show that feeding formula similar to preterm and term infant formula alters the fatty acid composition of piglet bile lipid in a similar direction to that in the liver and plasma phospholipids. The statistical analyses undertaken suggest that the association between the fatty acid composition of the bile and plasma phospholipids is best explained by an underlying relation between the fatty acid composition of the liver and plasma phospholipid. This is consistent with a pivotal role of the liver in selecting and retaining specific phospholipid species for secretion in bile (Coleman and Rahman 1992, Robins and Brunengraber 1985). Of note, the piglet bile fatty acids contained substantial amounts of 20:4(n-6) and 22:6(n-3), (mean ± SEM) 9.4 ± 0.5% and 2.0 ± 0.1%, respectively. The piglets fed sow milk also had significantly higher concentrations of phospholipid in bile than the piglets fed formula (27 ± 3 µmol/L, compared with a range of 12-21 µmol/L phospholipid for piglets fed sow milk or the formulas, respectively), and this resulted in 1.5- to 3-fold more 20:4(n-6) in the bile of piglets fed sow milk than in those fed formula. Presumably, the higher 20:4(n-6) in bile phospholipids could contribute to the higher plasma phospholipid 20:4(n-6) associated with milk rather than formula feeding.

In summary, the results of these studies suggest that dietary fatty acid requirements for incorporation of optimum levels of 20:4(n-6) and 22:6(n-3) into the developing brain of infants are best assessed by measures of functions known to be related to 20:4(n-6) and 22:6(n-3), respectively, rather than by extrapolation from differences in blood lipid fatty acids. On the other hand, clear associations between diet-induced changes in saturated, monounsaturated and (n-6) fatty acids in plasma and liver phospholipids suggest that changes in plasma phospholipid fatty acids incurred by formula feeding are likely to extend to the liver. Whether this has any physiological relevance to normal liver function is not yet known. The studies reported here were done with piglets, a species which shares many similarities in physiology and lipid metabolism with humans. However, the applicability of equations generated to describe relations between liver and plasma phospholipid fatty acids in these studies with piglets clearly should not be extrapolated to humans or other species without caution.