| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
USDA, ARS, U.S. Dairy Forage Research Center, Madison, WI 53706-1108
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsDiscussionConclusionsImplicationsLiterature Cited |
|---|
Key Words: Analytical Methods • Detergent Fiber • Dietary Fiber • Fiber Analysis • Repeatability • Reproducibility
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsDiscussionConclusionsImplicationsLiterature Cited |
|---|
; Sullivan, 1964; Van Soest, 1964; Tyler, 1975). Hipsley (1953) was the first to use the term dietary fiber to describe a nutritional property of diets. This generic term will be used to indicate the general concept of fiber as it relates to nutrition. The main challenges facing insoluble dietary fiber (IDF) methods are relevance and reproducibility. Without relevance, there is no reason for measuring dietary fiber; without reproducibility among laboratories, there is no value in reporting dietary fiber results. To be relevant, IDF methods must provide useful nutritional information and must have utility as either a quantitative description of a feed or a means of evaluating feeds or formulating rations. Reproducibility of IDF methods is crucial to have confidence that results are accurate and comparable among research, regulatory, and feed-testing laboratories. Dietary fiber is unique among feed constituents because it is defined only on a nutritional basis (that is, in terms of the digestive and physiological effects that it elicits) but must be measured chemically. Thus, the nutritional definition for dietary fiber is key to method relevance. The usefulness of dietary fiber results vary from its value as an indicator of physiological health benefits to its value as a predictor of digestibility and energy value of feeds. Furthermore, the relevance of dietary fiber data may be different between research and practical use, and vary within each use.
Numerous methods have been proposed for measuring dietary fiber, and some have become routine analyses for research and practical use. The scope of this review will be limited to the official methods of fiber analysis as described by the Association of Official Analytical Chemists (AOAC) International. These methods can be used in situations in which accuracy and precision are required and often are the ones routinely used in research and practical applications to describe feed characteristics. Objectives of this review are to define the criteria needed to evaluate IDF methods, discuss the relevance of each method in meeting nutritional needs, describe analytical problems in reproducing empirical dietary fiber results, and assess the relative merits of IDF methods.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsDiscussionConclusionsImplicationsLiterature Cited |
|---|
AOAC Official Methods for Dietary Fiber
With a few exceptions, dietary fiber is determined gravimetrically as the difference in weights of a test sample before and after extraction in a solution(s). There are two AOAC official methods for crude fiber (CF) in animal feeds: 962.09—Crude fiber in animal feed and pet foods, ceramic fiber filter method, or 978.10—Crude fiber in animal feed and pet foods, fritted glass crucible method (AOAC, 2002). In the most recent versions of Method 962.09, the precoating of the Oklahoma filter screen or California Buchner funnel with ceramic fiber when analyzing extremely fine samples was clarified. Acid detergent fiber and acid detergent lignin using sulfuric acid (ADSL) can be determined using AOAC Official Method 973.18—Fiber (acid detergent) and lignin in animal feed. Several clarifications have been included in the more recent version of Method 973.18, such as 1) description for cleaning crucibles, 2) specification of particle size for preparing samples, 3) preextraction of test samples containing >10% fat with acetone or similar solvent, 4) time of soaking of residues after acid detergent extraction to remove acid, and 5) addition of formula to report results on a as-is or as-received basis (AOAC, 2002). Amylase-treated neutral detergent fiber (aNDF) can be measured by AOAC Official Method 2002.04—Amylase-treated neutral detergent fiber in feeds using refluxing in beakers or crucibles.
There are several AOAC official methods for measuring total dietary fiber (TDF), IDF, and soluble dietary fiber (SDF). The first AOAC official method for TDF was 985.29—Total dietary fiber in food, enzymatic-gravimetric method, which did not allow separation of dietary fiber into soluble and insoluble fractions. Insoluble dietary fiber can be determined using AOAC Official Method 991.42—Insoluble dietary fiber in foods and food products, enzymatic-gravimetric method (phosphate buffer) and SDF by Method 993.16—Soluble dietary fiber in food and food products, enzymatic-gravimetric method (phosphate buffer). These methods for measuring TDF, IDF, and SDF have been superseded by Official Method 991.43—Total, soluble, and insoluble dietary fiber in foods, enzymatic-gravimetric method (MES-Tris buffer). Official Method 992.16—Total dietary fiber, enzymatic-gravimetric method, uses neutral detergent extraction with amylase treatment and measurement of SDF to determine TDF. More detailed analysis of TDF can be determined using Official Method 994.13—Total dietary fiber (determined as neutral sugar residues, uronic acid residues, and Klason lignin), gas chromatographic, colorimetric, gravimetric method, which is based on acid hydrolysis and chromatographic analysis of sugar residues.
Collaborative Studies
One of the primary purposes of the AOAC is to sponsor collaborative studies for evaluating analytical methods under actual laboratory conditions with a diversity of materials, personnel, environments, equipment, and so on (AOAC, 1993). Under these conditions, the total precision of a method (reproducibility) can be determined (Steiner, 1975), which the AOAC uses to make an informed decision about the acceptability of the method as official. The total precision of an analytical result is the sum of variability among laboratories and within laboratories. Reproducibility of a method is defined as the variation among single results for the same material when determined by different laboratories (different analyst, apparatus, environment, time, etc.). Repeatability of a method is defined as the variation among results for the same material determined in similar conditions within a laboratory (typically successive analyses within the same run: same analyst, apparatus, reagents, etc.).
To assess the reproducibility and repeatability of a method requires replicated analyses of multiple materials within multiple laboratories. Youden (1975) suggested that the absolute minimum design for a collaborative study would be five laboratories analyzing three pairs of materials (low, medium, and high concentrations of the analyte). He also suggested that matched pairs of materials (Youden pairs) provide more statistical information than blind duplicates for the same number of analyses. The minimum design provides 30 observations, which is the minimum number needed to obtain an acceptable estimate of standard errors (Wernimont and Spendley, 1985). Typically, the AOAC requests at least eight laboratories and five materials (duplicate analyses) for most collaborative studies (AOAC, 1993). The materials should represent the full range of composition matrices to which the method will be applied. A study with more than one material would result in the following statistical model to obtain estimates of variation pooled across materials (Wernimont and Spendley, 1985):
 |
where
An acceptable method must have a nonsignificant Li•Mij interaction and variations within and among laboratories that are small relative to the variation among materials. Assuming the sources of variation are independent, the variances can be partitioned so that the variation for reproducibility of the method can be calculated:
 |
where
This model applies to study designs in which replicates are run successively or within the same run or day. Because replicates are analyzed successively, this model assumes that repeatability within a laboratory is primarily a function of random variation among replicate test samples and that techniques, equipment, reagents, and so on, within laboratories do not vary among runs or days.
However, it is reasonable to assume that within-laboratory repeatability has two sources of variation: replication within a run in which conditions are relatively uniform with respect to environments, apparatus, reagents, and so on, and variation among runs, which is associated with uncontrollable laboratory conditions over longer periods of time. These sources of repeatability within laboratories can be assessed by designing a collaborative study in which laboratories replicate analyses in different days or runs.
Horwitz (1982) compared the results of numerous AOAC collaborative studies and observed that the relative standard deviation of reproducibility (RSDR = SDR/mean) was related to the concentration of the analyte across a wide variety of methods and generated a formula for the Horwitz relative standard deviation of reproducibility (HRSDR):
 |
where C is the fractional concentration of the analyte. Thompson and Lowthian (1997) confirmed that the HRSDR provides an expected standard deviation of reproducibility for any method, which varies from 4% of the mean at 1% concentration to 2% of the mean at >90% concentration. The HRSDR indicates that the expected standard deviation of most feed analysis methods, which have means of 10 to 60% of DM, would be 2 to 3% of the mean. Dividing the RSDR observed for a specific method by the HRSDR generates the Horwitz ratio (HORRAT), which permits the comparison of reproducibility among diverse methods (Horwitz et al., 1990). A HORRAT of 1.0 or less indicates that a method has reproducibility similar to other methods approved by AOAC. With some exceptions, a HORRAT of >2.0 suggests that a method is unacceptable with respect to the reproducibility of other official methods.
Proficiency Testing of Laboratories
The ongoing validity of each individual laboratory’s ability to generate reliable results is accomplished by proficiency testing or performance check programs. In these programs, carefully prepared, homogeneous samples are analyzed by participating laboratories to compare results. The Association of American Feed Quality Control Officials operates a check sample program in which results of participating laboratories are summarized and reported back to the participants for use in monitoring their results. The National Forage Testing Association (NFTA) was established under the auspices of the American Forage and Grassland Council, the National Hay Association, and commercial forage analysis laboratories to certify the proficiency of participating laboratories. The results of each laboratory are compared to a consensus reference value for each material and, if they fall within a specified range, the proficiency of the laboratory is certified by the NFTA (Mertens, 1998b). Results for other analyses are monitored by NFTA, but only DM, CP, ADF, and aNDF are used for certification of proficiency.
Ideally, materials used for proficiency testing would have a known composition for each analyte. However, there are no primary standards for dietary fiber or dry matter, and it is necessary to establish reference values for each material used in a proficiency-testing program. The NFTA program allows each laboratory to select the method used to measure each of the analytes, and requires only that laboratories analyze the proficiency-testing samples with the same method they use in routine practice. The results of each laboratory are compared to a consensus value, the reference method average (RMA) for each analyte (Mertens et al., 1994). The reference method for each analyte is either an AOAC official method or a method accepted by the NFTA (Undersander et al., 1993). The results used to calculate the RMA are selected based on each laboratory’s answers to a questionnaire about the specific details of their routine methods to determine whether they followed the reference method. Results of laboratories using reference methods are censured by selecting only those within one standard deviation of the median; that is, results are ranked and the top and bottom 15.8% are discarded. Typically, 10 to 30 laboratory results are used to generate the RMA. Censuring ensures that anomalous results are not used to determine the RMA to which all laboratories are compared for certification of proficiency. Six materials are analyzed each year, and laboratories are certified as proficient if their results fall within ± 3•HRSDR of the RMA for CP, ADF, and aNDF and within a modified HRSD for DM.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsDiscussionConclusionsImplicationsLiterature Cited |
|---|
and Trowell (1974) to describe plant cell wall components in the diet that were resistant to hydrolysis by mammalian digestive enzymes. Later, Trowell et al. (1976) broadened the definition of dietary fiber to include all indigestible polysaccharides, such as gums and mucilages, whether or not they originate from plant cell walls. This physiological-chemical definition of dietary fiber as "polysaccharides and remnants of plant materials that are resistant to hydrolysis (digestion) by human alimentary enzymes" is the basis for AOAC official methods for TDF that have been accepted for human food labeling (Cho et al., 1997). Because TDF is based solely on resistance to digestion, it contains both SDF and IDF. Partitioning TDF into SDF and IDF may provide important nutritional information for nonruminants because their recovery in the feces and impact on the physiological processes of digestion (fermentability, viscosity, water-holding capacity, distension, etc.) may be quite different.
The definition of TDF for humans, which limits dietary fiber to components that cannot be digested by mammalian enzymes, may be unduly restrictive for ruminants and herbivores, which have a symbiotic relationship with microorganisms and other adaptations of digestive physiology that enable significant digestion of dietary fiber. In the most general terms, dietary fiber is the coarse-textured portion of edible materials that is difficult to digest and adds bulk to digesta and feces. Mertens (1985) proposed that dietary fiber for herbivores be defined as the "indigestible or slowly digesting portion of feeds that occupies space in the gastrointestinal tract." Perhaps to distinguish dietary fiber from indigestible ash, this definition should be modified to include only "indigestible or slowly digesting organic matter of feeds that occupies space in the gastrointestinal tract." These definitions of dietary fiber exclude rapidly fermenting polysaccharides of plant cell walls (such as pectin) and soluble polysaccharides that do not occupy space in a liquid environment (such as fructans and gums), but would include slowly fermented, complex polysaccharides that are digested by fermentation in the alimentary tract of herbivores (such as cellulose and hemicellulose). Essentially, this more restricted definition of dietary fiber describes IDF, which is the feed component that is variable in digestibility and affects the total DM or OM digestibility of feeds or diets by ruminants. It excludes the rapidly fermentable SDF because they have true digestibilities similar to plant cell contents. Although SDF may alter ruminal fermentation, its effect on the health and performance of ruminants are unknown. Insoluble dietary fiber affects the digestibility and passage rate of feeds and diets in all animals. Due to their high intakes of dietary fiber, the space-occupying characteristics of IDF and its requirement for chewing to reduce particle size for passage through the alimentary tract may be factors that make IDF more important to herbivorous animals than SDF.
For a dietary fiber method to be practicable, it must apply to all potential feed ingredients and compound mixtures of feeds. Therefore, the restriction that dietary fiber comes only from plant sources is practically inappropriate and nutritionally inconsistent with the definition of dietary fiber. The strictly physiological definition does not require that dietary fiber originates from plants or their cell walls, and even TDF as defined for humans contains compounds that do not occur naturally in plants. Although fiber has been linked to plant cell walls because they contain similar chemical components in forages, fiber and cell walls are not synonymous terms. Insoluble dietary fiber is not cell walls because analytical methods often isolate insoluble components in feeds other than plant cell walls, and cell walls are not IDF because some plant cell wall components, such as pectin, are rapidly fermented and are solubilized by many fiber methods.
The goal of dietary fiber methodology is to accurately evaluate nutritive value and ultimately be useful in improving the nutritional quality of animal diets. Theoretically, dietary fiber methods should be developed to fit a nutritional definition and not vice versa. However, it is unlikely that any chemical or enzymatic measurement will mimic all of the nutritional effects of fiber in the animal. Although dietary fiber should be defined by nutritional concepts and not analytical methodology, practically, dietary fiber is an empirical measurement that is defined by the method per se. In this situation, consensus dietary fiber values can be generated for reference materials by averaging the results of several laboratories that follow exactly a prescribed method. To ensure the accuracy of results, efforts are needed to generate reference materials that can be used to document the accuracy of dietary fiber analyses among research, regulatory, and feed-testing laboratories.
A practical method should not require exotic instruments, reagents, equipment, laboratory environment, and the like so that it is suitable for routine analyses by feed-testing and regulatory laboratories. In addition, the method should be rapid, convenient, and economical to allow multiple samples of materials to be analyzed in a timely manner. Given the variability in materials and their sampling before being sent to the laboratory, routine methods should be designed to allow multiple determinations, rather than be so expensive or time consuming that nutritional decisions must be based on a single sample. Specificity is required to ensure that the method measures fiber accurately without interference or artifacts that alter the fractions being measured. Finally, a method must have a clearly defined limit of reliability that determines the method’s ability to discriminate among analyte levels in materials and detect concentrations different from zero.
For research, specific methods that require sophisticated equipment and techniques may be needed to provide detailed analyses for comparisons among treatments or to assess molecular or structural characteristics of dietary fiber components. However, if their conclusions are to be relevant to field applications, researchers have a burden to provide routine analytical information as well as their detailed analyses to establish a connection between research findings and field applications. In addition, researchers have a responsibility to ensure that routine methods used in their laboratory correspond to those used in the field. Therefore, research laboratories should participate in proficiency-testing programs with regulatory and feed-testing laboratories, and provide their performance statistics in reports and manuscripts.
Reproducibility of Dietary Fiber Methods
After it has been established that a dietary fiber method is relevant because it matches the definition of fiber, it must be established that it is reproducible. Reproducibility is related to how well analytical results are measured (Steiner, 1975). It is a deceptively simple term that represents the sum of variation associated with accuracy and precision. Accuracy is the ability to measure the "true" value of a primary standard with known composition or of the consensus concentration of a reference material that is determined by a group of analysts exactly following a defined method. Precision is the ability to repeat a measurement, or more quantitatively, the variation among repeated results. As illustrated in Figure 1, accuracy and precision are independent characteristics of a method’s reproducibility. The error associated with precision (scatter about the average value) tells us nothing about the error associated with accuracy (how close the average result matches the true value). Yet we often hear that the results of a method used in a laboratory must be correct because duplicate variation (only a measure of precision) was small. Precision does not guarantee accuracy because it is possible to precisely determine the wrong result (Figure 1c). No one would argue that an inaccurate and imprecise method (Figure 1a) is useful; however, it could be argued that precision is not particularly relevant because the true value can be detected using an imprecise method simply by averaging a large number of replications (Figure 1b). Conversely, researchers sometimes argue that precision is more important than accuracy because the true value is unattainable or irrelevant when we are only concerned about detecting differences among treatments or feeds. However, if data are compared among institutions or laboratories, if new knowledge is built on one another’s results, or if research conclusions are to be applicable in the field, then research analytical results must be both accurate and precise (Figure 1d) to allow others to reproduce and use the information that is generated.
|
Unlike research laboratories, commercial feed analysis laboratories typically analyze only one test sample for each material received. Thus, the reproducibility (approximate 95% confidence interval) among single analyses performed by two laboratories on representative samples of the same material is 2.8•SDR. The reproducibility of a method is a quantitative measure of its robustness, or its power and sturdiness in measuring the analyte in all types of materials using generally accepted practices within laboratories. Several performance characteristics of a method determine its reproducibility or robustness: ruggedness, practicality, specificity, and limit of reliability (Wernimont and Spendley, 1985). Ruggedness refers to a method’s ability to generate acceptable results when small, uncontrolled changes in operating conditions occur. Ruggedness testing of a method (Youden, 1975) involves evaluating the impact of making small perturbations in the reagents (concentrations, sources, etc.), conditions (temperature, time, etc.), equipment (settings, models, etc.), and steps (skipping or modifying). Ruggedness testing can be a daunting task when methods are complex and involve sophisticated equipment, and typically these methods are less thoroughly tested. Thus, complex methods that are less rugged are more demanding in expertise and in exactness of following procedures than are simple solubility methods.
Types of Dietary Fiber Methods and Sources of Variation
Fiber methods are typically categorized into three types (chemical-gravimetric, enzymatic-gravimetric, or enzymatic-chemical) based on the ways fibrous residues are isolated and measured. Isolation of dietary fiber residues is done by extraction in chemical solutions, enzymatic hydrolysis of nonfibrous constituents, or a combination of the two. After the fibrous residue is isolated, it is measured either gravimetrically (weighing the residue) or chemically (hydrolyzing the residue and measuring individual components, such as sugars and lignin).
A primary factor affecting gravimetric reproducibility is the accuracy and precision of the balance. Accuracy of a balance depends on its ability to report the true value when tested with a known weight and its smallest weight of detection. It is clear that balances should be routinely maintained and standardized, and should be calibrated or checked for accuracy at each use. Even with daily calibration of balances, we have observed an unexplained systematic bias that is consistent for all weights taken within weighing sessions. Correcting for blanks accounts for this systematic bias in weighing and has greatly improved the precision of replicates and the accuracy of results for test samples that have small residue weights (such as lignin or dietary fibers <10% of DM) in our laboratory. The improvement in the accuracy of results occurs because the systematic bias is often a significant proportion of the residue weight (Mertens, 2002). The problem of the lowest weight of detection may be less obvious because laboratories occasionally weigh test samples or residues only to the nearest 0.01 g without recognizing that this negatively affects results. If test samples of 0.50 g are used, results can be reported legitimately to only two significant digits (nearest 1 percentage unit) because this is the limit of information in the original weight regardless of the number of digits generated during calculation.
The reproducibility of results is also determined by the precision of weighing the test sample, such as, if the precision of this balance is ± 0.01, then the 95% confidence interval for the test sample weight is 0.48 to 0.52 and the potential variation in weighing overwhelms the remaining factors associated with method variability. However, the converse of this situation is not true. If the balance used weighs to the nearest 0.0001 g, this does not guarantee four significant digits of precision because other characteristics of the method can affect the precision of results. Sokal and Rohlf (1981) suggest that, in general, the number of decimal places for reporting results should be based on the standard error of the mean using the following guideline: divide the standard error by 3 and use the decimal place of the first nonzero digit to determine the significant digits to report. Because the standard error of most dietary fiber methods is less than ± 2.5, results should typically be reported to the nearest 0.1%.
Cherney et al. (1985) demonstrated that the variation in fiber results is also affected by the amount of test sample. The effects of weighing error increased as the test sample amount decreased, especially when <0.3 g (for alfalfa containing about 26% NDF and 18% ADF). Goering and Van Soest (1970) reported that weighing materials hot directly from the oven instead of transferring oven-dried materials to a desiccator before weighing is not only faster and more labor efficient, but also more accurate. The accuracy of the hot-weighing technique is better than when using desiccators because any change in the zero value of the balance is recorded and subtracted from the hot weight to arrive at the true weight of the material. In effect, this corrects each weight for slight changes in the balance between weighings. In addition, hot-weighing eliminates the variation due to the use or inadequate maintenance of desiccators. Both Horwitz et al. (1990) and Mertens (2002) reported that the minimum SDR for gravimetric methods is about 0.3 to 0.4 percentage units when the concentration of the analyte approaches zero. Thus, the HORRAT may be an inadequate estimate of the expected reproducibility of gravimetric methods when factional concentrations are <0.1. In gravimetric methods, the error associated with weighing does not approach zero as the analyte concentration approaches zero as predicted by the HRSDR equation, which is based on all methods including instrumental methods that have small limits of detection (approximately parts per billion or less).
Crude Fiber
Crude fiber was supposed to measure the indigestible ballast of feeds. The method was based on chemical extraction with alkali and acid solutions, which were the known characteristics of the digestion process before enzymes were discovered. The organic matter lost during ashing is calculated as CF, which was initially called wood fiber or crude woody fiber (Dougall, 1956). It was discovered that CF was digested as it passed through the alimentary tract of ruminants (Henneberg and Stohman, 1860), but their method became a standard part of the proximate analysis scheme. The CF method was first approved as an official method for the AOAC in 1890 (Wiley, 1890), and it became the de facto definition of dietary fiber for over 100 yr. The current AOAC Official Method 920.86 was adopted for flour in 1920 and Official Method 962.09 was adopted for animal feeds in 1962. The CF method is extremely robust in that it can be easily measured in all types of feeds and foods and can be reproduced within and among laboratories. Although the method is empirical (defined solely by the method used to measure it), CF has been useful, historically, in estimating digestibility or energy value within feed types.
The major deficiency of CF is its lack of relationship to any acceptable nutritional definition of dietary fiber and its inability to advance understanding about physiological responses to dietary fiber or its impact on digestibility. The proximate system partitions carbohydrates and allied compounds into two fractions: CF, which is measured analytically, and nitrogen-free extract (NFE), which is calculated by difference (NFE = 100 - CP - EE - CF - ash). Neither of these fractions meets criteria for uniform nutritional availability or for the definition of dietary fiber (Van Soest, 1967). Depending on the feed material, CF may contain only 40 to 100% of the cellulose, 15 to 20% of the pentosans from hemicellulose, and 5 to 90% of the lignin. Lignin is dissolved, especially in grasses, by the alkali extraction step in the CF method that is used to remove protein. Because much of the hemicellulose and lignin is included in the NFE fraction, the digestibility of NFE, which is supposed to contain the easily digested carbohydrates, is less than the digestibility of CF for 25% of the feeds listed by Morrison (1956). Currently, CF is used only for quality control and specification of feeds (minimum CF) by regulatory agencies. Its lack of accuracy in measuring dietary fiber and abandonment by researchers and practicing nutritionists suggests that its use for feed regulation should be abolished.
Acid Detergent Fiber and Acid Detergent Sulfuric Lignin
Like CF, ADF (AOAC Official Method 973.18) is an empirical method that was designed to be a preparatory step for the determination of ADSL. The reason for specifying that lignin was determined using sulfuric acid is that ADF can also be the preparatory step for determining lignin using permanganate (ADPL); therefore, the term acid detergent lignin is not adequate to differentiate between the two correlated but different measures of lignin (ADPL typically has higher values than ADSL). In the ADF method, protein is removed from the fibrous residue by the cationic detergent cetyl trimethylammonium bromide to minimize the nitrogen contamination of lignin. Acid extraction is used to remove nonfibrous compounds while minimizing the losses of alkali-labile lignin. Acid detergent fiber does not meet the nutritional definition of dietary fiber or IDF for ruminants because acid-soluble hemicelluloses are removed and some rapidly fermented pectin is not. The precipitation of pectins in strong acid may be the reason that some feeds containing high proportions of pectin (e.g., immature alfalfa, citrus pulp) may have ADF results that are higher than NDF.
The reproducibility of the ADF determined during an AOAC collaborative study (Van Soest, 1973) was good, with a SDR of 1.13 and a HORRAT of 1.2 (Table 1). The SDr was approximately one-third of the total SDR of individual analyses among laboratories. This indicates that the variation among laboratories is twice the variation within laboratories, which falls within the typical range for most methods (Horwitz, 1982). The SDR for ADSL was 0.62, but the HORRAT was 3.1, primarily because the small mean of ADSL results in an unrealistically small HRSDR for a gravimetric method (Mertens, 2002).
|
Although it is not an AOAC official method, there are circumstances when it may be desirable to measure ADF sequentially (sADF) after neutral detergent extraction. Sequentially determined ADF is almost always less than ADF determined by the official method because neutral detergent removes some components that are not removed as well by acid detergent, such as pectins and tannin or phenolic acid complexes. Hintz et al. (1996) determined ADF sequentially on NDF residues that were isolated using heat-stable 
-amylase with (aNDF) or without sulfite (neutral detergent residue, NDR). When sADF was determined on NDR, values were 1 to 3 percentage units lower than ADF measured using the official method and this difference increased to 2 to 4 percentage units when sADF was determined on aNDF (Table 2).
|
). The significant nutritional attribute of neutral detergent extraction is that it separates feeds into two major fractions that are distinctly different in their digestibility and intake by ruminants and herbivores, and in many cases by nonruminants as well (Mertens, 1993). Whereas NDF has variable digestibility, occupies space in the alimentary tract that can be a physical constraint on intake, and requires significant chewing to reduce particle size, neutral detergent solubles (NDS), which represent the inverse of NDF (NDS = 100 - NDF), have nearly constant true digestibilities near 100%, occupy little space because they are rapidly solubilized, and require minimal chewing. Van Soest (1967) reported that NDS have a true digestibility of about 98% and a relatively constant endogenous loss of 12.9% when consumed by ruminants at maintenance levels of intake. The level of intake is important because, at maintenance levels of intake, herbivores, especially sheep, chew feeds adequately, which allows complete digestion of NDS.
Neutral detergent fiber is measured using a chemical solubility-gravimetric method. Proteins are extracted using anionic detergent and sodium sulfite. Fats are removed using hot detergent and acetone. Soluble dietary fiber is removed primarily by hot detergent extraction, and the solubility of easily fermented pectin is enhanced by chelating calcium bound in pectin complexes using EDTA. In the original NDF method (Van Soest and Wine, 1967), soluble carbohydrates and starch were extracted by hot solutions. It was discovered that the original NDF method inadequately removed starch from some feeds and foods. Numerous modifications of the NDF method have been proposed since the original publication of the method (McQueen and Nicholson, 1979; Robertson and Van Soest, 1980; Mascarenhas Ferreira et al., 1983; Van Soest et al., 1991). Of these modifications, the NDR method of Robertson and Van Soest (1980), which uses a heat-stable -amylase to remove starch during detergent extraction and eliminated the use of sodium sulfite, became the de facto method for measuring NDF.
The original NDF method of Van Soest and Wine (1967) was never evaluated by a collaborative study. However, a method for measuring IDF based on the NDR modification of Roberson and Van Soest (1980) was evaluated as a method for measuring the IDF portion of TDF (Mongeau and Brassard, 1990). Although most of the materials used in the collaborative study were human foods, the comparison for wheat bran may represent results for high-fiber by-product feeds (Table 3). The SDR for wheat brans (±1.92) was slightly higher than that observed for ADF (±1.13), and the RSDR was also slightly higher (4.7 vs. 2.9%). The SDR was higher for TDF compared to IDF in wheat brans and foods, suggesting that the SDF contained in TDF may be less reproducible than IDF.
|
). The method is a modification of the original NDF method, which included the use of sodium sulfite, but added the use of a heat-stable -amylase standardized to remove starch during neutral detergent extraction and with specific modifications (sand and other filter aids) that solve filtering difficulties for all types of materials. The aNDF method allows results to be reported as either aNDF with fiber-associated ash or as ash-free aNDF organic matter (aNDFom), and each of these results can be reported with or without blank correction. It is anticipated that aNDF will be reported for routine feed analyses because it does not require an ashing step before reporting results. Although crucibles are routinely cleaned by ashing, that step represents an additional time delay in reporting results and most, if not all, commercial feed testing laboratories currently report aNDF.
For the most accurate estimate of insoluble dietary fiber, blank-corrected aNDFom is recommended. Blank correction is especially important when fiber results are <25% aNDF because systematic weighing variation can have substantial impact on these small residue weights (Mertens, 2002). Reporting results as aNDFom more accurately matches the definition of insoluble dietary fiber as organic matter and improves the accuracy of calculating nonfibrous carbohydrates because the ash in fiber is not subtracted twice. It is often unclear whether NDF results reported in the literature are ash-free organic matter, but neither the original method of Van Soest and Wine (1967) or the handbook of Goering and Van Soest (1970), which are often cited as sources of methods, indicate that NDF should be determined as ash-free organic matter.
Determination of aNDF has SDR among laboratories (Table 4) similar to that reported for ADF (Table 1
). It is surprising that the SDr within laboratories, which is due primarily to random variation in test samples, was a much larger proportion of SDR for aNDF (79%) than for ADF (34%). This suggests that either the variability in aNDF among test samples is much larger than for ADF or that other sources of variation within laboratories contributed to repeatability differences for aNDF. The repeatability value in Table 4 indicates that analyses should be rerun if duplicates differ by more than about 2.9 percentage units. The reproducibility value indicates that results for 19 out of 20 laboratories performing a single analysis on a well-mixed material should be within 3.7 percentage units of each other.
|
) differs from the aNDF method (Mertens, 2002) in the use of sodium sulfite and type of amylase. However, current implementations of the NDR method use amylases similar to the aNDF method because the original amylase is no longer available. Sulfite was eliminated when the NDR modification was developed because it may extract lignin and phenolic complexes. Removing sodium sulfite from the NDF procedure increases the protein contamination of fibrous residues, which has been used to define a slowly degrading protein fraction in feeds (Sniffen et al., 1992). However, protein contamination can greatly inflate the IDF values of feeds, especially heated feeds (Table 5). Using sodium sulfite reduces aNDF values for forages and oilseed meals by only 1 to 4 percentage units; however, it reduces the fiber values of heated by-product feeds, such as brewer’s and distiller’s grains, by about 11 percentage units. Thus, the use of sodium sulfite in the aNDF method is a compromise between losing a small amount of phenolic compounds in some feeds or having IDF contaminated by a large amount of protein that is apparently digested in other feeds. With the exception of grass forages, which have the smallest difference in fiber due to the use of sodium sulfite, most of the material extracted from feeds by sulfite is crude protein equivalent (Table 5). Although NDF, NDR, and aNDF can be corrected for protein contamination by measuring the nitrogen content of the respective fiber residues, this additional step requires time and expense, which are large disadvantages for a routine method and have little impact on results in comparison to variation within and among laboratories for IDF results. The aNDF method was designed to improve the routine determination of IDF in all feeds; therefore, sulfite was retained in the method to remove most protein contamination. Sodium sulfite has the additional benefit that it improves filtration (and precision) during the analysis of some problem materials.
|
and Hintz et al. (1996) reported the variation in dietary fiber concentration among the three major modifications of the NDF method. In general, the NDR modification (Robertson and Van Soest, 1980), which uses heat-stable -amylase, but not sulfite, results in slightly higher values (approximately 0 to 1 percentage unit) for feeds with little starch and moderate protein and much lower values for starch-containing feeds (approximately 1 to 5 percentage units) compared to the original NDF method (Van Soest and Wine, 1967). The aNDF modification (Mertens, 2002), which uses both sodium sulfite and heat-stable -amylase, generates lower results than either NDF or NDR.
One of the unappreciated sources of variation in NDF results is the effect of DM adjustment. Most chemical entities are measured on a test sample that is not completely dry; thus, the analysis is determined on an as-is, or as-received, basis. Although feeds are bought, sold, and regulated (feed tag specifications) on an as-is basis, nutrients are also reported on a DM basis to allow a more direct comparison of nutrient densities without the confounding factor of moisture in the material. However, small apparent differences in DM determination, which may be caused by poor technique or differences in methodology, can result in large artifact differences in nutrient concentration on a DM-adjusted basis when the magnitude of analyte concentration is large (Table 6). The small discrepancy between these two laboratories for EE or CP is inconsequential, but the larger difference in NDF on a DM basis between these laboratories would be disconcerting. This illustrates that nutrient concentrations should not be compared among laboratories on a DM basis because these results combine the potential errors in both DM and nutrient determinations, making it impossible to determine which is the culprit when discrepancies occur. Unfortunately, there is no AOAC official method for routine measurement of DM that is acceptable for all feeds and this situation needs to be rectified. The determination of DM is empirical and methods vary significantly among laboratories. The 100 laboratories that responded to a 1993 NFTA questionnaire about drying methods indicated that they used 47 different combinations of time and temperature (Mertens, 1994). Twenty-one temperatures (ranging from 57 to 140°C) and 16 drying times (ranging from 2 to 48 h) were used for DM determinations.
|
-amylase and amyloglucosidase to hydrolyze starch and proteases to hydrolyze proteins. Insoluble dietary fiber is the residue remaining after enzymatic hydrolysis. Soluble dietary fiber is precipitated from the hydrolyzed solution with ethyl alcohol at a final concentration of 78%. Ash and protein is determined on one of the duplicate IDF and SDF residues and used to correct IDF and SDF for protein and ash.
Although few materials were analyzed that are typical animal feeds, the results in Tables 7 and 8 suggest that enzymatic-gravimetric methods measure IDF with reproducibility and repeatability similar to those of the aNDF method (Prosky et al., 1985, 1992, 1994; Lee et al., 1992). The values for IDF using these methods are within the range of aNDF values expected for these feed sources. However, these methods have not been used on as wide a variety of feeds as the aNDF method and it is difficult to draw firm conclusions about comparisons between aNDF and enzymatic-gravimetric methods. Enzymatic-gravimetric methods are more complex and require more intermediate analyses than aNDF; therefore, it is unlikely that reproducibility would be better. The complexity and time required to conduct enzymatic fiber assays makes it difficult to envision that they will be adopted as routine methods for feed analysis. Their appeal is based on the concept that the measurement of dietary fiber by enzymes mimics the process of digestion; however, it is clear that neither the conditions nor the enzymes used for measurement approach the complexity of hydrolysis in the gastrointestinal tract. Nonetheless, SDF measured by enzymatic methods appear to affect physiological processes in humans.
|
|
; Englyst et al., 1994). The Uppsala method (Theander et al., 1995) is the only enzymatic-chemical method for measuring TDF that is an official method. Official Method 994.13 of the AOAC uses -amylase and amyloglucosidase to hydrolyze and remove starch, but no proteases are used to remove protein. After starch hydrolysis, ethyl alcohol is used to precipitate soluble polysaccharides. The combined soluble and insoluble fiber residue is hydrolyzed to neutral sugars and uronic acids using 12 M sulfuric acid at 30°C for 1 h followed by diluting the sulfuric acid to 0.4 M and autoclaving at 125°C for 1 h. Klason lignin is determined as the loss of acid insoluble residue after ashing and the acid-hydrolyzed filtrate is analyzed for neutral sugars and uronic acids. The neutral sugars are measured as alditol acetates by gas chromatography and uronic acids are measured colorimetrically.
The enzymatic-chemical chromatographic method is time consuming and expensive in terms of both labor and equipment. It requires highly skilled and trained personnel to manage sensitive chemical reactions and to operate and maintain chromatographic instruments. The method is currently used mostly in research laboratories and is not used extensively by commercial or regulatory laboratories. Although the method is chemically sophisticated, the reproducibility of both TDF and total nonstarch polysaccharides (Table 9) indicates that variability among laboratories is about twice that of gravimetric assays for dietary fiber (aNDF or enzymatic).
|
) suggests that laboratories have difficulty reproducing monomeric information, which appears to be related to the empirical conditions used to isolate enzymatic residues and compromises between maximum solubilization and minimum degradation of sugars during acid hydrolysis. The benefits of extensive carbohydrate analysis will be realized in the short term only if routine analyses are reported concomitantly to provide a bridge between current and future fiber analyses and to determine when alternative methods of analysis provide similar or complementary information.
Knudsen (1997) reported detailed analysis of nonstarch polysaccharides and Klason lignin in feeds using modifications of the Uppsala method, that were reported as soluble, insoluble, and total nonstarch polysaccharide and lignin, which were summed to obtain TDF (Table 10). Although NDF was not measured by Knudsen (1997), several samples of each feed were analyzed, which can be compared to typical NDF values reported by NRC (2001). The IDF of feeds based on insoluble nonstarch polysaccharides plus lignin are generally similar to NDF.
|
describes empirical methods as Type I Defining Methods because the values obtained can be generated only in terms of the specific method—that is, defined by the method. This classification is not a negative reflection on the quality of a method or its repeatability within or reproducibility among laboratories. Because there are no primary reference standards for these methods, they cannot be validated for accuracy in determining the "true" value for the constituent. To minimize systematic errors (bias) among laboratories, empirical methods must be followed exactly because even slight variations in methodology can result in the measurement of a different constituent. Systematic bias among laboratories for empirical methods can be determined only using consensus values, which are the average results of laboratories that follow the method exactly.
It might seem that dietary fiber can be defined more accurately and precisely by specifying and quantifying the chemical monomers in dietary fiber carbohydrates. However, this assumes that sugar analysis is exact and that detailed knowledge of polysaccharide analysis will lead to nutritional insights. Although the accuracy of chromatographic measurement of sugars can be determined using primary standards, the preparation of enzymatic residues is relatively empirical and can affect the quantities of sugars and lignin recovered (Marlett et al., 1989). In addition, polysaccharides are converted to monosaccharides, and monosaccharides are degraded at different rates depending on acid hydrolysis conditions and characteristics of the enzymatic residues. Because there is incomplete recovery of sugars, correction factors (based on typical analyses) are needed to correct results and generate adequate quantification.
The nonuniform digestibility of fibrous carbohydrates suggests that knowing their monomeric composition, no matter how accurately, provides little information about their availability to the animal. The physicochemical nature of fiber and its relationship to noncarbohydrate components, such as lignin, may have greater nutritional significance than its intrinsic monosaccharide composition. The consequence of this speculation is that even the most elegant analysis of carbohydrate composition may add little to our ability to evaluate feeds nutritionally. The rationale for dietary fiber analysis is derived from its nutritional consequences. Caution should be exercised to ensure that research on dietary fiber is not shifted from an approach that uses the nutritional definition of dietary fiber to develop methods for measuring it to an approach that develops a method that measures chemical constituents and attempts to define fiber on the basis of its composition.
Dietary fiber methods are now, and may always be, empirical because the result is dependent on the reagents and conditions used in each method to isolate fibrous residues. Horwitz et al. (1990) reevaluated the collaborative studies of all methods used for nutritional labeling and concluded that all fiber methods had poor reproducibility among laboratories when compared to the measurement of crude protein. They suggested that the lack of reproducibility was related to the empirical nature of these methods. The relatively poor reproducibility of dietary fiber methods compared with other methods may be related to several factors.
Proficiency of Laboratories in Using Empirical Methods
Although the standard deviation from the RMA among laboratories (SDL) eligible for NFTA certification for each method is not the same statistical parameter as the SDR determined in AOAC collaborative studies, it provides a similar indication of variation among laboratories. The SDL differs from the SDR because it does not include variation among replicates within laboratories and is based on the average of triplicates rather than single analyses; however, it includes systematic bias because it is determined in relation to the RMA. Observed SDL (Table 11) of ADF and NDF are larger than the HSDR (ranging from 144 to 173%), which may indicate that laboratories participating in the NFTA proficiency-testing program are not required to follow any specific method for their analyses (although they are encouraged to do so). It may also indicate the variability of empirical fiber methods because the HORRAT for all of the dietary fiber methods are >1.0. Surprisingly, the relative SDL for CP, ADF, and NDF were similar (3.3 to 4.2% of the mean), which disagrees with the conclusion of Horwitz et al. (1990) that fiber results are less reproducible than protein results. The small relative SDL (<5%) for the routine methods of feed analysis suggests that analyses among laboratories certified as proficient by the NFTA are comparable when replicated results are reported.
|
summarized a large number of published reports that provided independent regression equations for predicting digestibility based on various fiber fractions, which were averaged to obtain the mean and standard deviation of intercepts, slopes, R2, and standard deviations of regression (Table 12). The highest average and smallest SD of R2 indicates that ADSL had the best average relationship with digestibility. However, the large SD for the slope of the regression for ADSL suggests that its quantitative relationship with digestibility is variable among populations of materials. Although it has the highest and most variable average standard error of regression, NDF obtained the most robust relationship with digestibility as evidenced by the smallest variation in the slope. The average regression parameters for NDF suggest that about half of NDF is typically digested and that the digestibility of NDS is about 97.7% (i.e., the digestibility of a feed containing no NDF and therefore 100% NDS).
|
|
Conclusions |
|---|
|
TopAbstractIntroductionMethodsDiscussionConclusionsImplicationsLiterature Cited |
|---|
|
Implications |
|---|
|
TopAbstractIntroductionMethodsDiscussionConclusionsImplicationsLiterature Cited |
|---|
1 Correspondence: 1925 Linden Drive, West (phone: 608-264-5228; fax: 608-264-5147; E-mail: davem{at}dfrc.wisc.edu).
Received for publication December 5, 2002. Accepted for publication September 2, 2003.
|
Literature Cited |
|---|
|
TopAbstractIntroductionMethodsDiscussionConclusionsImplicationsLiterature Cited |
|---|
-amylase. J. AOAC 62:676–680.
This article has been cited by other articles:
|
|
 |
|
  L. O. Tedeschi, P. J. Kononoff, K. Karges, and M. L. Gibson Effects of chemical composition variation on the dynamics of ruminal fermentation and biological value of corn milling (co)products J Dairy Sci, January 1, 2009; 92(1): 401 - 413. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
