| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Carman and Ann Adams Department of Pediatrics (W.W.K.K.)
Statistical Consultant (E.M.H.), Wayne State University
Detroit Medical Center (W.W.K.K., M.D.), Detroit, Michigan
Address correspondence to: Dr. Winston Koo, Hutzel Hospital, Department of Pediatrics, 3980 John R, Detroit MI 48201. E-mail: wkoo{at}wayne.edu
 |
ABSTRACT |
|---|
 TOPFOOTNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
Design: A systematic search of Medline and the Cochrane Database of Systematic Reviews was performed to retrieve studies comparing infants who received infant formulas containing PO with those who received infant formulas without PO or which contained synthetic triacylglyceride as a source of palmitic acid. Outcomes of interest include intestinal fractional absorption of fat, palmitic acid and calcium; and bone mass. The effect size for each dependent variable in each published study was obtained by standardizing based on the difference in means between non-PO and PO group with respect to the standard deviation of the PO group. Trend analysis of the outcome of interest was performed when 3 or more between group comparisons were available. The comparison of effect size across different studies was based on all available data and includes results that showed no significant difference between infants fed PO or non-PO study formulas in the outcomes of interest.
Results: Nine publications were identified with non-PO and PO comparison groups. The gestational ages of infants in the published studies were between 28 to 42 weeks and postnatal ages were birth to 192 days at study onset. Within each published study, there was some variability in the effect size between non-PO and PO groups. The standardized results were consistently significantly (p < 0.05) positive in favor of the feeding with non-PO formulas with respect to increased intestinal fractional absorption of fat, palmitic acid and calcium. The latter two variables were significantly different by at least 0.6 SD. Bone mass measured as total body bone mineral content was significantly higher in the non-PO group by at least 0.3 SD.
Conclusion: The use of PO in infant formulas to match the human milk content of palmitic acid has unintended physiological consequences including diminished intestinal absorption of fat, palmitic acid and calcium and lower bone mass. The avoidance of PO or its substitution with synthetic triacylglyceride in infant formulas can prevent this detrimental effect.
Abbreviations: PO = palm oil and its low melting fraction, palm olein • BMC = total body bone mineral content • BMD = total body bone mineral density
|
INTRODUCTION |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
Palm oil and its low melting fraction, palm olein (PO), a relatively inexpensive and excellent source of palmitic acid, are added to many infant formulas in amounts that mimic the palmitic acid content of human milk. Digestion of PO results in a high proportion of free palmitic acid. The free palmitic acid is poorly absorbed and the unabsorbed fraction readily forms esters with divalent cations, primarily unabsorbable calcium-fatty acid-soap complexes. Thus, biochemical basis exists for the loss of energy and minerals in particular calcium from ingestion of infant formulas with PO-dominant fat blend.
Study designs to resolve the concerns for detrimental physiological consequences of this biochemical basis for lowered digestion and absorption of fat and mineral in infants fed PO-dominant formulas have focused on mass balance studies, use of stable isotopes and bone mass measurements, each of which addresses a different aspect of the essential question. However, mass balance studies are time consuming, require trained personnel and are difficult to perform accurately; stable isotope studies have major technical and cost issues, and the skeletal effect of diminished fat and calcium absorption as determined by bone mass measurements require trained personnel, as well as instrumentation with specifically designed software. Thus the effect of plant source of palmitic acid in infant formulas is unlikely to be fully addressed by a single study and requires data from multiple studies. However, direct comparison among different clinical studies is difficult because of differences in the subject population including gestational and postnatal age at study, type of formulation used for feeding, and extent of the effect on various outcome measures.
The aim of this report is to standardize specific physiological outcome measures from published reports to increase understanding of the effects of PO as a major component in the fat blend of infant formulas. Specifically, we aim to test the hypothesis that consumption of non-PO infant formulas will result in higher fractional absorption of fat, palmitic acid and calcium, and higher bone mass.
|
MATERIALS AND METHODS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
Standardization of Published Results
The infants from each report were classified into PO or non-PO group. The PO group included infants who received formulas with fat blend containing PO and a resultant amount of palmitic acid generally in excess of 20% of total fatty acid, which is quantitatively similar to human milk. Molecular distribution of palmitic acid in PO is predominantly at the Sn-1 and Sn-3 positions of the triacylglyceride molecule with a near random distribution of palmitate in the Sn-2 position at generally <25%. The non-PO group included infants who received formulas without added palm olein i.e., a relatively low palmitate content at usually <10% of total fatty acid, or with fat blend containing synthetic triacylglyceride generally with >70% of palmitic acid chemically isomerized to the Sn-2 position. Outcome measures of interest included intestinal fractional absorption of fat, palmitic acid and calcium, and total body bone mineral content (BMC) and bone mineral density (BMD). Data from human milk-fed reference group were treated as a separate non-PO group if sufficient number of studies with the same outcome variables were available for statistical analysis. Otherwise, the human milk reference data were reported for descriptive purposes only.
Computations and Statistical Analysis
Standard error (SE) and standard deviations (SD) were obtained for all reported outcome measures. The formula SD = SE * (n)
0.5 was used to estimate either the SD or SE when that statistic was not reported. The assumption was made that the references cited employed n-1 to compute the SD. From summary statistics provided in each publication, t tests were re-computed using the measures for analysis in the present study to confirm the reported statistically significant differences comparing means for each outcome measure of interest (intestinal fractional absorption of fat, palmitic acid and calcium, and, BMC and BMD), thus serving as quality assurance of the data in the present study.
The effect size for each dependent variable in each published study was standardized based on the difference in means between non-PO and PO groups with respect to the standard deviation of the PO group using the equation (Meannon-PO – MeanPO)/SDPO [4]. In our calculations, covariance between the two groups in cross-over designs was ignored as this was not available from the published data. This affected the calculation of t, as the covariance would not have been subtracted from the pooled variance estimated. In addition, subjects were treated as independent samples even with cross-over intervention, i.e., PO to non-PO and vice versa. This allowed the ability to calculate t values in the absence of the intra-individual correlations in the reports. Consequently, standard error estimates used in the computation of t values are larger than would have been obtained from the actual data since the covariance (that includes the intra-individual correlation) values were not subtracted from the pooled variance estimates, i.e., this would have biased against finding an effect.
If there were at least 3 paired comparisons between non-PO and PO groups for any of the outcomes of interest, the standardized difference between groups was further analyzed to determine the effect on each outcome measure across multiple studies. One sample t test was used to test for significance using 0, 0.3 and 0.6 SD difference as the test values. One tailed t test was used to assess for a positive trend from formulas without PO. To confirm the validity and consistency of the effect size, the calculations were repeated using the formula with respect to the standard deviation of the non-PO group using the equation (Meannon-PO – MeanPO)/SDNon-PO.
|
RESULTS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
The study design and clinical parameters of the non-PO versus PO group, i.e., absence or presence of palm olein in the fat blend are shown in Table 1A and 1B. The effect size for each reported study, standardized to the standard deviation of the PO group, is shown in Table 2. Within each published study, there was some variability in the effect size between non-PO and PO groups.
|
|
|
Human milk feeding was reported in two studies [8, 13] and bone mass measurement was the only variable of interest reported. Furthermore, only the former report quantified the amount of infant formula supplemented to these infants. The amount of infant formula supplemented averaged 480 mL at 3 months and 690 mL at 6 months indicating that infant formula was the predominant milk intake in the "breastfed" group at these ages [8]. Thus, no statistical comparison with the human milk group was possible for the measured variables although the bone mass of human milk fed infants was similar to infants fed formula with synthetic triacylglyceride [13] and became lowered to the level of infants fed formula with PO as the intake of PO study formula increased [8].
|
DISCUSSION |
|---|
There are numerous differences between the composition of human milk and that of infant formulas. Even if some nutrients in infant formulas are quantitatively similar to human milk, qualitative differences, often based on nutrient source, may result in demonstrable physiological differences. The use of palm oil and its low melting fraction, palm olein, in many infant formulas to match the high palmitic acid content (
20% by weight of total fatty acids) of human milk [15, 16] is an example of similar quantitative content but with significant qualitative differences between infant formula and human milk. During fat digestion, pancreatic lipase specifically hydrolyzes the fatty acids in the Sn-1 and Sn-3, leaving the remaining fatty acid in the Sn-2 as a 2-monoglyceride. Saturated fatty acids such as palmitic acid and stearic acid are well absorbed as 2-monoglyceride [17–19]. As most of the palmitic acid from plant sources, including palm oil and palm olein, is at the Sn-1 and Sn-3 positions of the triacylglyceride molecule [19–21], the use of PO in the fat blend of infant formulas results in a high proportion of free palmitic acid. The latter readily forms esters with divalent cations, primarily unabsorbable calcium-fatty acid-soap complexes. This qualitative difference in molecular distribution of palmitic acid from PO directly causes a decrease in fat and calcium absorption.
Multiple independent investigators have now documented that infant formulas with PO at levels needed to match the palmitic acid content of human milk have lowered absorption of calcium and/or fat [5–13] with increased stool hardness [12, 13], and lowered bone mass [8, 9, 13]. In this study, standardization of the results of outcome variables allows the comparison of data generated from a wide variety of subject populations and study design and methodology to gain an understanding of the overall benefit or detriment from the use of a particular ingredient in infant formula. The hypothesis that consumption of non-PO infant formulas will result in higher fractional absorption of fat, palmitic acid and calcium, and higher bone mass was supported. The consistent trend of a beneficial effect from absence of PO in the fat blend of infant formulas was applicable to the intestinal fractional absorption of fat, palmitic acid and calcium, i.e., maximizing the absorption of these nutrients. Alternately, presence of PO lowers the absorption of the same nutrients.
Within each published report, the different extent of effect size for each dependent variable most likely reflects biological variability. However, the comparison of effect sizes across different studies was based on all available data. The significance of our findings is strengthened by the inclusion of outcome measures with no statistically significant differences and without taking into account the correction factors such as intra-individual correlation in cross-over studies. The persistence in statistical significance even at 0.6 SD difference, the greatest level of difference tested, for the fractional absorption of palmitic acid and calcium is consistent with the biochemical basis of impaired digestion and absorption of palmitic acid from plant sources and its consequences on calcium absorption. Our results on the effect of PO or its absence in infant formulas are applicable to a wide range of ages during which the consumption of infant formula or human milk as exclusive or dominant source of nutrient intake is expected. The effect was demonstrated in milk- [5, 6, 8–13], soy- [7] and casein hydrolysate- [7] based infant formulas, and included preterm [5, 6, 10, 11] and term infants [5–9, 12, 13].
Lower palmitic acid absorption from PO formula results in lower weight gain in piglets [19]. But in infants, there is a lack of significant weight difference between infants fed PO and non-PO infant formulas presumably due to the compensatory increase by an average of about 7% (0.4 SD above the mean intake) in the volume of milk ingestion [13]. However, apparently higher milk intake of infants fed PO formulas could not compensate for the lower calcium absorption compared with infants fed non-PO formulas, with the latter group having higher bone mass despite the ingestion of lower volume of non-PO formula [9, 13]. The lack of apparent difference in body weight may mask the ineffectiveness to compensate for impaired calcium absorption and bone mineralization since the latter two parameters are not routinely measured. Thus the increased consumption of PO-dominant infant formula can mask its negative impact on skeletal health and potentially add an economic burden to the family and society.
The magnitude of effect size of 0.3 SD lower in bone mass from PO-dominant formula is comparable to the lowered bone mass in osteoporotic patients without effective treatment. For osteoporotic patients, effective pharmacologic treatment results in an average gain of 2% to 10% (up to 0.5 SD of the mean) in bone mass depending on the region of interest. This difference in bone mass is associated with 30% to 50% reduction of new fractures at the vertebrae [24, 25] or at any site [25]. Since bone mass at all levels strongly predicts bone strength [22, 23], the lower bone mass caused by PO formulas should be considered clinically important.
The high prevalence of clinical complications of osteoporosis is related to insufficient bone mass [23–25], and the association of childhood fractures with low bone mass [26] or calcium intake [27], indicate higher, rather than lower bone mass to be of likely clinical benefit. To this end, lifecycle strategies to encourage initial accretion and later maintenance of higher bone mass are becoming an important nutrition goal of primary practice [24]. The data from this analysis strongly support that the avoidance of palm-olein oil containing formulas in infancy can contribute to achieving this goal of optimizing bone mass accretion.
The overall conclusion from the clinical studies analyzed in this report is that the use of PO to match the content of palmitic acid in infant formulas to that of the human milk has unintended physiological consequences and its avoidance or substitution with synthetic triacylglyceride can prevent this detrimental effect. In this era of evidence based medicine, the development of human milk substitutes must be based on achieving optimal physiological outcome measures rather than superficially replicating the nutrient content of human milk [28].
|
FOOTNOTES |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
Received December 20, 2005. Accepted February 13, 2006.
|
REFERENCES |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
