| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,1
* School of Veterinary Medicine, University of Pennsylvania, Kennett Square 19348
School of Veterinary Science, University of Sydney, New South Wales 2006, Australia
1 Corresponding author: moate{at}vet.upenn.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Key Words: milk • fatty acid • concentration
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
98%) composed of triglycerides (Taylor and MacGibbon, 2002). Triglycerides are composed of a glycerol backbone and 3 fatty acids. In the 1980s and early 1990s, a number of excellent reviews were published describing how cow-level factors and feeding practices influenced the fatty acid profiles of milk fat (Sutton, 1989; Grummer, 1991; Jensen et al., 1991; Palmquist et al., 1993). Milk fat is now known to contain more than 400 individual fatty acids (Jensen, et al., 1991). In recent years, improvements in analytical techniques have facilitated the routine reporting of a large number of the less common geometric and positional isomers of octadecenoic and octadecadienoic acids. The discoveries that cis-9, trans-11 linoleic acid has anticarcinogenic properties (Parodi, 2002), that milk fat depression is specifically associated with an increase in the concentration in milk of the trans-10 18:1 isomer, rather than trans 18:1 isomers in general (Griinari et al., 1998), and that trans-10, cis-12 18:2 has potent effects on lowering milk fat concentration when injected or infused postruminally (Parodi, 2002) triggered an explosion of research in this area. During the last 10 yr, more than 100 articles have been published that describe the influence of a wide range of nutritional treatments on milk fat yield and composition. These publications have often tabulated fatty acid profiles from 4:0 to 22:6. Fat supplements fed in these experiments have involved a diverse range of traditional fat supplements, including tallow, oilseeds, and fish oil, and some novel or commercial products, including calcium salts of palm oil fatty acids, prilled tallow fatty acids, bioengineered soybeans high in oleic acid, and new varieties of flaxseed high in linolenic acid. Despite these advances, to our knowledge no publication has documented the average concentration or the variability in concentrations of major fatty acids or various isomers of octadecenoic acid and octadecadienoic acid in milk fat. In addition, to our knowledge no statistical or dynamic mathematical model has been developed to predict the influence of dietary and cow-level factors on the profile of major fatty acids in milk fat. Documentation and examination of the means and standard deviations, and correlations of the concentrations of the major fatty acids in milk fat is a necessary starting point for the development of such a model. With respect to the development of a mathematical model, examination of the available literature data enables the adequacy of data to be assessed in terms of its scope, reliability, and accuracy.
The first objective of this paper was to describe how data from the scientific literature were collated into a data set relating how dietary and cow-level factors influence milk fatty acid profiles. It is intended that this data set will be used in a companion paper to describe the development of regression equations to predict the production of 26 major fatty acids in milk fat. A second objective of this paper was to document the mean and variability in concentrations of major individual fatty acids that occur in bovine milk and to examine how concentrations of individual fatty acids in milk fat may vary in relation to the concentrations of other fatty acids.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
) on the influence of diet on milk fat yield and the profile of individual fatty acids were collected and collated. Articles were published between 1992 and 2006 and reported trials were performed in North America (n = 20), Europe (n = 6), and Oceania (Australia and New Zealand; n = 2). Only papers published since 1992 were included in this analysis, because only in the last 15 yr have many authors included in their articles comprehensive listings of a large number of milk fatty acids (from 4:0 to 22:6). The criteria used for inclusion of an article were that the article had to report the following data: total milk yield (MY, kg/d), total milk fat yield (MF, g/d) or milk fat percentage (g/100 g of milk), profile (g/100 g) of individual milk fatty acids, including at least 8:0 to 18:3, DIM, total DMI (kg/d), total concentration of fatty acids in the diet (FA, g/kg DM), NDF content of the diet (g/kg of DM), CP content of the diet (g/ kg of DM), and proportions (by DM) of all the major feeds in the diet. Although more than 100 articles were found that contained extensive tabulation of milk fatty acid profiles, many lacked either some or all critical descriptions of dietary factors needed for the development of a predictive model, and these were therefore not included in the database. Articles describing experiments in which milk fat percentage and milk fatty acid composition were influenced by the oral, abomasal, or intravenous administration of mixtures of pharmacological quantities of conjugated linoleic acid (CLA) were excluded from the analysis. Four articles were not included in the database because they listed milk fatty acid profiles that contained obvious and substantial errors.
|
). Each article usually described 4 experimental treatments (range 2 to 8 treatments), and there were generally at least 4 cows in each treatment. Cows were usually blocked into groups according to BW, calving date, and preexperimental period milk yield. These groups were then randomly assigned to treatment groups and maintained in controlled conditions during the course of the experiment. Many of the experiments involved Latin square designs in which cows cycled through experimental treatments during the course of the experiment. Cows in all experiments were Holstein-Friesian. Cows in the experiment by AbuGhazaleh et al. (2003) were primiparous, those in the experiment by Morales et al. (2000) were a mixture of primiparous and multiparous, and those in all other experiments were multiparous. The mean and standard deviation of BW (kg) of cows in each treatment group was 616 ± 16, and for DIM, 103 ± 61. Cows were fed either a TMR (24 experiment, 102 diets) or were involved in grazing experiments (5 experiments, 18 diets) in which they may also have received dietary supplements. In most experiments, one group of cows received a control diet, and other groups of cows received the control diet plus a substantial amount of a fat supplement. The diets covered a wide range of different types of treatments and supplements. In this database, there were 10 dietary treatment groups that received dietary supplements containing oilseeds; 19 diets containing tallow, grease, or a similar supplement; 11 diets with calcium salts of palm oil fatty acids; 14 diets with plant oils; 6 diets with fish meal; 16 diets with fish oil; and 12 diets in which fatty acids were infused into the abomasum. The average composition of major chemical components in the diets was as follows: NDF, 35.7 ± 6.4%; CP, 17.3 ± 1.8%; total fatty acids as a percentage of DM, 4.6 ± 1.7%.
Treatment means for production data from each publication were entered into an Excel 2006 (Microsoft Corporation, Redmond, WA) spreadsheet. The production data included daily MY, MF, and the profile of milk fatty acids (4:0 to 22:6). Data on each individual milk fatty acid (FAi) was generally reported as grams of FAi per 100 g of total milk fatty acids. A summary of the main production data used in this analysis is shown in Table 1. Across all treatment groups, the production of milk was 30.4 ± 7.2 (kg/d) and milk fat was 1,011 ± 235 (g/d). The 26 major milk fatty acids that were entered into the spreadsheet included: 4:0 6:0, 8:0, 10:0, 12:0, 14:0, cis-9 14:1, 16:0, cis-9 16:1, C17, 18:0, cis-9 18:1, trans 6–8 18:1, trans-9 18:1, trans-10 18:1, trans-11 18:1, trans-12 18:1, cis-9, cis-12 18:2, cis-9, trans-11 18:2, trans-10, cis-12 18:2, other CLA, 18:3, 20:0, 20:5, 22:6, and "other." Various publications reported different fatty acids and possibly had different reporting conventions. For many fatty acids, the exact isomer description was often not reported. In tabulating reported milk fatty acid data in the database, 14:1 was assumed to be cis-9 14:1 unless otherwise indicated, and 16:1 cis was assumed to be cis-9 16:1. Fatty acids with 17 carbon atoms were generally not reported, or sometimes there was no distinction between 17:0 and 17:1 fatty acids. In this analysis, C17 represents the combination of 17:0 and 17:1 fatty acids. Frequently, 18:1 was reported with no indication of whether this represented a cis or trans isomer, or a combination of both. In this situation, 18:1 was assumed to represent cis-9 18:1. When 18:1 trans was the only description of trans octadecenoic acid, this was assumed to represent trans-11 18:1. When 18:2 was the only description of octadecadienoic acid, this was assumed to be cis-9, cis-12 18:2, and when 18:3 was reported, this was assumed to be cis-9, cis-12, cis-15 18:3. In most publications, the individual milk fatty acids were reported in units of grams per 100 g of milk fatty acids. In many publications, the total of all the reported fatty acids summed to approximately 93 g, and it was assumed that the approximately 7-g discrepancy could be described as "other" fatty acids. These "other" fatty acids possibly represented many unidentified acids, such as odd and branched-chain fatty acids, and also positional isomers of some of the more common fatty acids.
Milk fat production (g/d) was either obtained directly from each publication or calculated from MY and milk fat percentage:
 |
Milk fat was assumed to contain 93.3% milk fatty acids (Glasser et al. 2007):
 |
Production of individual fatty acids (PFAi, g/d) was calculated as:
 |
where FAi is the percentage (wt/wt) of each of the following individual milk fatty acids in the total milk fatty acids: 4:0, 6:0, 8:0, 10:0, 12:0, 14:0, cis-9 14:1, 15:0, 16:0, cis-9 16:1, C17, 18:0, trans-6–8 18:1, trans-9 18:1, trans-10 18:1, trans-11 18:1, trans-12 18:1, cis-9 18:1, cis-9, cis-12 18:2, cis-9, trans-11 18:2, trans-10, cis-12 18:2, other CLA, 18:3, 20:0, 20:5, 22:6, and other. The concentration (g/kg) of individual fatty acids in milk was calculated as:
 |
Fatty acids in bovine milk are considered to be either produced de novo in the mammary gland or derived from plasma lipids. Generally, 4:0 to 14:0 and some 16:0 are thought to be produced de novo in the mammary gland (Grummer, 1991). In this investigation, the production of total de novo fatty acids is defined as the sum of the production of 4:0 to 15:0:
 |
Milk fatty acids that are not produced de novo in the mammary gland are often termed "preformed" fatty acids (McGuire and Bauman, 2002). Preformed fatty acids are generally considered to contain some C16 milk fatty acids and milk fatty acids with more than 16 carbon atoms. In this analysis, the term "preformed fatty acids" will be used to signify only fatty acids with more than 16 carbon atoms:
 |
Statistical Analysis
Simple statistics (means, standard deviations, medians, minima, and maxima) were calculated to describe the concentrations (mg/g) of the major individual fatty acids in milk fat. Least squares means and standard deviations of the individual fatty acids in milk from cows fed TMR- and pasture-based diets were obtained by means of categorical regressions clustered on experiment. Multiple pairwise correlations using the Bonferroni adjustment were made between concentrations of the major individual fatty acids, groupings of fatty acids, and total milk fat concentration. All statistical analyses were performed with Stata software, version 9 (Stata, 2007).
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Concentrations of Total and Individual Fatty Acids in Milk
In carrying out this investigation, we had to decide on the most appropriate method of estimating the concentrations of total and individual milk fatty acids in milk fat. Glasser et al. (2007) recently presented a technical note in which they explained that if milk fat is assumed to be composed entirely of triglycerides, then according to their data, milk fat contains 94.4 ± 0.2% fatty acids. We repeated their calculations with this data set and obtained a value of 94.6 ± 0.2%, thus confirming the accuracy of this coefficient and its very small standard deviation. However, Glasser et al. (2007) further refined their calculations to take into account the presence in milk fat of other fatty constituents besides triglycerides. They found that a constant coefficient of 93.3% was most appropriate to describe the concentration of fatty acids in milk fat; therefore, we adopted the latter coefficient.
A summary of the proportions of 26 major milk fatty acids in the total milk fatty acids (mg/g of total milk fatty acids) from the 120 experimental diets is shown in Table 2. To our knowledge, Table 2 presents the first summarized tabulation of literature data on the concentrations of many of the recently discovered, or only recently routinely reported, positional isomers of many unsaturated fatty acids.
|
are generally consistent with those reported in previous reviews (Jensen et al., 1991; Palmquist et al., 1993). However, 2 interesting discrepancies are the concentrations of 14:1 and 16:1 shown in Table 2, which are approximately 50% lower than concentrations previously reported by Palmquist et al. (1993) for these fatty acids. This difference may reflect the improved accuracy of analytical methods since 1993. Difficulties in resolving closely related fatty acids determined by using older analytic methods may have led researchers to aggregate a number of fatty acids. Thus, in the older publications C15 was frequently not reported, and during analysis this fatty acid may have been aggregated with cis-9 14:1. We speculate that a similar situation may have occurred with cis-9 16:1 and 17:0. A major finding of this analysis is that for many of the individual fatty acids listed in Table 2, concentrations had standard deviations that ranged from less than 25% to more than 100% of the mean values. This high degree of variability highlights the need for equations that predict the influence of diet and other factors on the fatty acid profile of milk fat.
Table 3 summarizes the least squares mean concentrations (g/kg milk) of total and individual fatty acids in the milk of cows fed TMR- and pasture-based diets. Table 3 is somewhat unusual in that the unit of concentration of milk fatty acids is grams per kilogram of milk instead of the more traditional grams per 100 g of milk fatty acids. We chose to document the fatty acid concentrations in this way because it is an easy unit to conceptualize and because it removes the complicating process that occurs with the traditional unit of trying to determine whether a change in the fatty acids in the numerator or denominator resulted in a particular proportion.
|
Concentrations of 15:0 and C17 were higher (P < 0.05) in milk from cows fed pasture-based rations compared with the concentrations in milk from TMR-fed cows. These findings are consistent with recent findings by Couvreur et al. (2006), who showed a positive linear relationship between the proportion of fresh grass in the diet and the concentrations of 15:0 and C17 fatty acids in milk fat. The estimated mean concentration of total trans-18:1 fatty acids was higher in milk from pasture-fed cows compared with TMR-fed cows, principally because of higher concentrations of trans-11 18:1 in milk from pasture-fed cows. Diet type (pasture- vs. TMR-based diet) also was associated with significant differences in concentrations of a number of other acids, including trans-10 18:1, cis-9, cis-12 18:2, 20:0, and 20:5 (P < 0.05).
In Table 3, the least squares mean concentrations of cis-9, trans-11 18:2 in the milk of TMR- and pasture-fed cows were not significantly different (P > 0.05). This may, prima facie, appear surprising because previous research has shown that the concentration of cis-9, trans-11 18:2 is generally higher in milk from pasture-fed cows than in milk from TMR-fed cows (Kelly et al., 1998). There are 2 likely explanations for this anomaly. First, the concentration of cis-9, trans-11 18:2 in milk was reported for all 18 of the pasture-based rations, whereas it was reported for only 58 of the 102 TMR-based rations. We speculate that nonreporting (reporting bias) of the concentration of cis-9, trans-11 18:2 probably occurred if the concentration of cis-9, trans-11 18:2 was very low or, indeed, if cis-9, trans-11 18:2 was not detected. Second, fish oil, a supplement known to enhance cis-9, trans-11 18:2 concentrations in milk (AbuGhazaleh et al., 2003), was not included in any of the 18 pasture-based diets but was included in 20 of the 58 TMR rations for which the concentration of cis-9, trans-11 18:2 was reported. Further, fish oil was not included in any of the 44 TMR diets for which the concentration of cis-9, trans-11 18:2 was not reported. Thus, with respect to TMR diets, the mean concentrations of cis-9, trans-11 18:2 shown in Table 3 may well be overestimated. We further speculate that a similar reporting bias may influence means for many of the fatty acids in Tables 2 and 3, especially in cases in which substantially fewer than the maximal number of observations contributed to a mean. Despite the limitations of the database and differences in concentrations of some minor isomers of fatty acids, the means shown in Table 2 are consistent with data presented in previous reviews, in which it has been shown that the concentrations of major individual fatty acids in bulk milk fat from Australian cows, where grazing is the dominant dairy production system, are generally similar to those in bulk milk fat from US cows, which are predominantly fed TMR rations (Jensen et al., 1991). The data set in this investigation included some studies in which extreme quantities of fat supplements and unusual types of fat supplements were fed. Thus, in Table 2, the minima and maxima for concentrations of individual fatty acids must be regarded as being outside the limits of what could be considered normal milk.
We consider that the consistency of milk fat concentrations under different feeding conditions and the wide range of diets used in the studies that form this database strongly support the validity of models potentially derived from this study. Further, the consistency of findings between grazing and TMR diets suggests that a single model may be able to describe milk fat profiles for grazing cows and cows fed TMR diets.
The scientific literature seems to contain only limited reports on pairwise correlations of the major fatty acids in milk fat (Karijord et al., 1982; Soyeurt et al., 2006). Tables 4, 5, and 6 present pairwise correlations of the concentrations of 26 of the major individual fatty acids and groupings of fatty acids in milk. The major difference between our findings and those of earlier researchers is that we report the correlation coefficients of many more fatty acids, especially the various isomers of un-saturated C18 fatty acids. The magnitude of the correlations shown in Tables 4, 5, and 6 may differ somewhat from those estimated by Karijord et al. (1982) and Soyeurt et al. (2006), but our findings support the direction of individual associations reported by earlier researchers.
|
|
|
). The concentration of 16:0, which is generally the fatty acid present in milk at the highest concentration, was also (P < 0.05) positively correlated with all but one of the individual de novo fatty acids, with total de novo fatty acids, with cis-9 16:1, with C17, and with total milk fat concentrations, but was not correlated with the total concentration of preformed fatty acids. The concentration of 18:0 was positively correlated (P < 0.05) with 4:0, 6:0, 8:0, 15:0, 17:0, cis-9 18:1, 20:0, total de novo fatty acids, total preformed fatty acids, and total milk fat concentration (Tables 4, 5, and 6). The positive correlation between the concentrations of 18:0 and cis-9 18:1 probably reflects the fact that cis-9 18:1 is produced in the mammary gland by the action of the 
9-desaturase enzyme working on 18:0. Another finding that supports the general conclusions made by Karijord et al. (1982) is that concentrations of individual unsaturated preformed fatty acids were generally positively correlated (P < 0.05) with each other but negatively correlated (P < 0.05) with the concentration of total de novo fatty acids (Table 5).
We note that the concentration of cis-9 18:1 was negatively correlated (P < 0.05) with the concentration of 20:5, which suggests that 20:5 might in some way inhibit the 9-desaturation reaction (Table 5). Tables 4, 5, and 6 show a number of positive and negative (P < 0.05) correlations among the various individual fatty acids. These correlations no doubt reflect many biological and chemical processes, which include absorption of fatty acids from the intestines and transformation reactions within the mammary gland (e.g., isomerization). The negative correlations of trans-10 18:1 with total milk fat concentration and of trans-10, cis-12 18:2 with total milk fat concentration are not surprising because these negative associations have been well documented previously (Bauman and Griinari, 2003).
Fish oil feed supplements are well known for their propensity to induce milk fat depression (Bauman and Griinari, 2003). As shown in Table 6, the only major fish oil fatty acid that was negatively correlated (P < 0.05) with total milk fat concentration was 20:5. Further, there were no significant positive correlations between 20:0, 20:5 and 20:6 and either trans-10 18:1 or trans-10, cis-12 18:2, 2 fatty acids that have been associated with low milk fat concentration. This observation suggests that 20:5 per se may be the principal fish oil fatty acid associated with low milk fat concentration. The multiple significant correlations shown in Tables 4, 5, and 6 provide much information that allows speculation, but causal relationships between individual milk fatty acid isomers and milk fat concentration will only be established with subsequent experimentation. Nevertheless, Tables 4, 5, and 6 show that some milk fatty acids are positively correlated both with others and with concentrations of total milk fat, whereas other fatty acids are negatively correlated with each other and with the concentration of total milk fat.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
and 3 show that the concentrations of individual fatty acids in milk can vary substantially. Critical examination of the data indicates that a number of aspects influence the usefulness of these data for model-building purposes. The primary limitation is that in many publications, the description of some fatty acids is poor and acronyms are used for groupings of fatty acids instead of specific fatty acids. Another limitation is in reporting some minor fatty acids, especially various isomers of octadecanoic and octadecadienoic acid. We consider that average literature concentrations of many minor fatty acids may be overestimated because of selective or biased reporting. Despite these limitations, the extensive data tabulated and the wide range of basal and experimental diets used in the studies suggest that sufficient data are available to enable the development of empirical equations to predict concentrations and daily production of major individual fatty acids in milk fat. The positive and negative correlations (P < 0.05) found among many individual fatty acids and groupings of fatty acids suggest that the concentrations of individual milk fatty acids are not similarly influenced by various dietary or animal factors. The fact that concentrations of all the individual de novo fatty acids are correlated with each other and with the concentration of total de novo fatty acids suggest that a common predictive equation may be able to explain much of the variation in the production of these fatty acids. The complex matrix of positive and negative correlations among the individual preformed fatty acids suggests that individual equations may be needed to predict how dietary and animal factors influence their production.
Received for publication March 23, 2007. Accepted for publication June 28, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. Glasser, A. Ferlay, and Y. Chilliard Oilseed Lipid Supplements and Fatty Acid Composition of Cow Milk: A Meta-Analysis J Dairy Sci, December 1, 2008; 91(12): 4687 - 4703. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. J. Moate, W. Chalupa, R. C. Boston, and I. J. Lean Milk Fatty Acids II: Prediction of the Production of Individual Fatty Acids in Bovine Milk J Dairy Sci, March 1, 2008; 91(3): 1175 - 1188. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
