This is the old United Nations University website. Visit the new site at

Contents - Previous - Next

4. Evaluation of protein quality in experimental animals

Limitations of Biological Procedures
Evaluation and Choice of Animal Assay Procedures
Methods Based on Changes in Body Weight
Methods Based on Nitrogen Retention
In Vivo Amino Acid Availability in Food and Feedstuffs
Other Assays for Protein Quality
Conclusions: Relevance of Animal Assays to Human Protein Nutrition

Limitations of Biological Procedures

The evaluation of a protein food normally proceeds from the simple to the more complex. The evaluation starts with nitrogen and amino acid analysis, moves through a series of specific chemical measurements, and ends with biological tests. Because animal assays have been widely used to evaluate protein quality, they have tended to gain such esteem that the results obtained are often regarded as providing all the information required. There are, however, limitations in the amount and type of information that can be derived from animal assay procedures which should be emphasized.

  1. The result obtained depends on the limiting amino acids and their availability and balance, and the presence or absence of other interfering materials, and provides no information about the amounts of other essential and non-essential amino acids in the protein. Other essential amino acids may be present in almost as low a concentration as the limiting one or in relative surplus. Their importance becomes nutritionally obvious only when they are part of mixtures of dietary protein sources. For example, it is possible to damage milk protein by destroying or reducing the availability of Iysine, but because Iysine is not the limiting amino acid, such damage is not revealed by a rat bioassay until it is severe enough to make Iysine limiting. However, when the milk is consumed in combination with a food such as cereal in which Iysine is limiting, even moderate damage may be of practical importance.
  2. A single index figure is obtained and applied to a variety of protein functions. Dietary protein is used for many different purposes, some of which, such as maintenance of existing tissues or the formation of new tissues, may require different proportions of amino acids. A protein that is of high quality for one purpose may not be so for another, although the evidence for this is not definitive. An amino acid mixture formulated to be ideal for maintenance purposes was evaluated by rat net protein utilization value at a ''protein" level close to maintenance and at 9 per cent "protein"; the NPU value at maintenance level was twice that found for the higher level (1). Similarly, in studies on the limiting amino acids in maize, Kies and Fox (2) found that the level of total nitrogen intake determined the order in which amino acids become limiting. As suggested in chapter 1, the current view is that the essential amino acid needs, per unit of protein requirement, are higher for infants than for adults. Although there is no conclusive evidence that the required pattern of essential amino acids differs between infants and adults, it is unlikely that the quality of a protein will always be the same for both age groups. In many instances reported in the literature (3, 4), however, the values for the two are quite comparable.
  3. The laboratory assay is carried out under standardized conditions that may differ from those encountered in non-experimental situations. It has been established that the utilization of a protein varies with the amount of protein and previous dietary history.
  4. Results are extrapolated from experimental animals to humans. This is a well-recognized problem of nutritional work. Although there is evidence that the results of many rat assays parallel those obtained in humans, as discussed later in this report, extrapolation is not necessarily valid.
  5. Different types of assay may give different absolute values. Chow et al. (5) found differences in ranking order of proteins when using different assay procedures. More recent studies, however, have found similar ranking orders for different proteins when a series of different assay procedures were used (6-8).
  6. There is evidence from studies with rats that a surplus of one amino acid may affect the usefulness of a protein. Thus there could be a difference between the value found in a laboratory assay and that attained in practical feeding of humans (9). Moreover, the value of a supplementary protein will vary with the amino acid composition of the remainder of the diet (10).

One or more of these drawbacks applies to all of the animal assay procedures for evaluating proteins, and some methods have additional inherent defects. It is essential that researchers using these methods realize the limitations. The ultimate selection of a method depends upon the precision and reproducibility required, and the use to which the results will be applied.

Evaluation and Choice of Animal Assay Procedures

The criteria use for animal assays can be grouped into several categories, which are shown in table 8. The major subdivisions are single-level and multi-level assays, with the single-level assays being further divided into techniques depending on nitrogen balance and techniques dependent on growth.

Table Rat Assaya for Protein Quantity





A. Single-level assays*
1. Nitrogen retention
a. Difference between intake and output Mitchell 1924 11
b. Carcass analysis, final of initial Shukers and McCoilum 1929 12
c. Carcass analysis,final of non-protein control Bender and Miller 1953 13
2. Weight gain
a. Weight gain per gram protein intake Osborne, Mendel, and Ferry 1919 14
b. Weight gain and weight loss on a non      
protein diet per gram protein intake Bender and Doell 1957 15
c. Rat repletion Cannon 1945 16
d. Weight maintenance Tomarelli and Bernhart 1947 17
B. MuIti -level assays* *
1. Slope of line relating N balance to N absorbed Allison and Anderson 1945 18
2. Slope of line relating growth to nitrogen intake Allison and Anderson 1945 18
3. Slope of line relating response {growth, body Hegsted and Chang 1965 19
water, or body N) to nitrogen intake Hegsted, Neff, and Worcester 1968 20
C. Other assays
1. Specific responses
a. Plasma protein regeneration Robscheit-Robbins and Whipple 1949 21
b. Liver protein regeneration Harrison and Long 1945 22
c. Liver protein utilization Campbell and Kosterlitz 1948 23
d. Liver xanthine oxidase Mokady, Viola, and Zimmermann 1969 24  
e, Liver enzymes Litwack et al. 1952 25
  Wirthen and Bergner 1964 26
2. Abbreviated indices
a. Urinary creatinine/total urine N Marlin et al. 1948 27
b. Blood urea levels Munchow and Bergner 1967 28
  Eggum 1973 29

As outlined earlier, the ability of a food to fulfil protein needs is a function of both the amount of protein in the food and its nutritional quality. The purpose of animal assays or other methods of assessing protein quality is to quantify nutritional quality as a characteristic of a test protein. There are obvious advantages if the method of assessment provides values that vary linearly from O to 1 (or 0 to 100 per cent) rather than arbitrary values. If this is possible, then:

amount x quality = utilizable protein

Conventional methods of calculating protein requirements or the capability of diets to fulfil protein requirements, as expressed by the joint FAO/WHO ad hoc expert committee (30) or the US National Research Council Committee on Dietary Allowances (31), assume explicitly or implicitly that the measurement of BV or NPU does indeed fulfil these requirements, though the limitations were recognized by one expert group ( 32).

All of the commonly used methods for evaluating protein quality in experimental animals attempt to measure change in body protein associated with the ingestion of a specific protein. Although the body composition of young rats fed various diets for a limited period of time cannot be shown to be constant, it is certain that the percentage of body protein is not subject to wide variation, and thus significant changes in body weight generally reflect changes in body protein. The ratio of body water to body protein is probably more constant than is body protein expressed as a percentage of body weight. However, estimations of body protein, body water, and body weight can generally be considered to be measures of the same changes and can be used more or less interchangeably, although there is argument over the validity or utility of these three measures. In adult animals, the change in body protein is usually, and often necessarily, estimated by changes in nitrogen balance.

The validity of the various methods of evaluating protein quality may be best discussed by considering the hypothetical dose-response curves of two proteins (fig 1). Protein A represents approximately the situation found with high-quality proteins, and protein B that of a poor-quality, Iysine-limiting protein. The shaded area represents the deviation around the regression line and is drawn so that this is similar for both proteins. The slope of the regression line for protein A over the linear portion is 4, and that for protein B is 1. Thus, protein B has a slope that is only 25 per cent of that for protein A. The characteristics of these lines are:

1. The dose-response lines are, within statistical limits, linear over a substantial range of intakes above and below the maintenance level of protein. It is important to note that the range over which the response is linearly related to protein intake is a function of the protein quality. This range will inevitably be small with high-quality proteins because curvature will occur as the protein intake approaches the maximal capacity of the test animal to utilize protein with maximum efficiency. With poorer-quality proteins, the range over which linearity in response is observed will be substantially greater (33). The range of intakes over which the linear response is found will also be a function of the characteristics of the test animal. With adult animals, for example, in which body protein content is relatively stable, the response will be linear over a more limited range above and below maintenance than with young animals that have a large growth potential. In both young and adult animals, however, response will be affected by the prior degree of depletion of the animals.

FIG. 1. Hypothetical Examples of the Dose-Response Lines Obtained with Two Proteins

2. Projection of the linear portion of the dose-response line to zero intake often, or usually, yields an intercept on the Y axis that is above the point experimentally defined by feeding a protein-free diet. The regression line of protein A in figure 1 yields a Y intercept of -17 compared to the expected value of -20, a modest and statistically insignificant difference. With protein B, the intercept is -10 and clearly different from -20. This demonstrates that curvature in the dose-response line occurs at very low intakes. Because of the inherent variability in the test animals, it may be difficult to demonstrate that the Y intercept is indeed above the expected point in any particular assay, but there is no doubt that this generally occurs and may jeopardize the validity of all assay procedures.

A number of considerations are relevant in attempting to evaluate the suitability of the various assay techniques. Some of the more important are:

  1. Validity. It is desirable to have some internal test of the validity of the assay. The result obtained must be relatively independent of the protein content of the diet, food consumed, or the kinds of rats used, and should depend exclusively on changes in protein quality. If the assay assumes a straight-line relationship between the selected dependent variable and protein consumption, there should be a test for linearity incorporated into the assay. This is only possible with multi-level tests.
  2. Precision. The ability of an estimate to discriminate quality among proteins is a function of both how different the estimates are and the random error or coefficient of variation of the estimate.
  3. Reproducibility. Adequate methods should yield reproducible results when applied in different laboratories and in repeated assays within a laboratory.
  4. Proportionality. The estimate should be proportional. A material with half of the potency of another should yield estimates that are half of this value.
  5. Cost. Other factors being equal, low cost is obviously desirable. In animal assays the major costs are the numbers of animals, amount of time needed to complete the assay, and manpower.
  6. Simplicity. Methods that are simple in design and easily analysed are preferred for routine application. Rigid specifications as to diet formulation (assuming all other nutrients are present in adequate amounts) and age and size of the animals should be avoided unless they are required for satisfactory results. Complex statistical analyses are also a relative disadvantage.

Methods Based on Changes in Body Weight

The simplest method for determining nutritive value is to measure the growth rate of young animals fed a test food. Osborne, Mendel, and Ferry (14) put this on a quantitative basis by relating weight gain to the amount of protein eaten; the index obtained was termed protein efficiency ratio (PER). They showed that the PER varied with the level of protein in the diet and recommended that each protein be assayed at its optimum level. This recommendation was not adopted by subsequent workers, and the conventional level of 10 per cent dietary protein was in general use until the AOAC standardized procedure, which recommended feeding at 9.09 per cent protein (34), was established. Canadian authorities selected a standardized version of PER for legal standardization of protein advertising claims (35, 36). This method was later adopted by the AOAC (37), and is currently used for labelling regulations in the United States. As a consequence of its role in legislation, the method has been the one most widely used in recent years.

The most serious fault of the PER assay is that it makes no allowance for protein used for maintenance, and consequently values are not proportional; i.e., a PER value of 2 is not twice as good as a PER value of 1. Thus PER is inappropriate as a quality estimate in a protein-rating system where the multiple of protein quality x protein quantity is considered as utilizable protein. The result is known to be dose-dependent, but no correction can be applied because it is a single-point assay and no internal test of validity can be used. It may be noted that it can also be treated as a "two-point assay," i.e., the slope between two points, one of which is the starting point (fig. 2). Factors that influence total food intake increase the variability of PER estimates, reducing the capability to discriminate between proteins. The assay is not always reproducible in different laboratories, and attempts to eliminate laboratory variation by correcting to an assumed PER value of 2.5 for casein (an internal standard) were not always successful in a collaborative assay (Samonds and Hegsted, unpublished).

Net protein ratio (NPR) (15) is an improvement over PER in that a zero protein control group is used. In practice, NPR is comparable to NPU, the well-known and much-used procedure based upon nitrogen retention, which is discussed later. It differs in that it is estimated from body-weight changes rather than body nitrogen changes. The similarity of these procedures will be seen by examination of figure 2. NPR is based on the supposition that the dose-dependent lines are linear from zero dosage up to some point at which curvature begins to occur. For those proteins that show characteristics meeting this assumption (a small deviation of the calculated Y intercept from that defined by feeding a zero-protein diet), these assays will provide a reasonable estimate of the slope of the true dose-response line. However, for Iysine-limiting proteins the NPR will indicate a higher nutritive value than do slope assays discussed below, and will depend, in part, on the actual level of the test protein chosen. Use of an internal standard in the NPR assay attempts to reduce variability and correct values to a scale of 100 (0-100, or 0-1). This modified assay is then termed relative NPR or RNPR. This standardization, however, does not completely eliminate the difficulties encountered with Iysine-limiting proteins. It has recently been claimed (39) that, while RNPR may indeed overestimate the protein value of low-lysine proteins for the rat, it may in fact be a better predictor of protein value for human infants than are some other procedures.

Fig.. 2. Some Protein Bioassay Procedures (From ref. 38)

Note: Actual intake of protein or N at point T is dependent, among other things, on length of assay and dietary protein concentration.

PER: protein efficiency ratio = weight gain per gram of protein consumed: slope of line relating weight gain and protein consumption.

NPR: net protein ratio = weight gain plus weight loss on non-protein diet per gram of protein consumed: slope of line relating weight gain and protein consumption, weight gain considered in relation to non-protein group. If slope is expressed relative to slope obtained with 8 per cent lactalbumin, NPR becomes RNPR.

NPUst: net protein utilization (standardized} = body N gain plus body N loss on non-protein die, per gram N consumed: slope of line relating N gain and N consumption, N gain considered in relation to non-protein group. Protein level either very low at maintenance or about 10 per cent.

NPUop: net protein utilization (operative)-defined as for NPUst but with no limit to protein level fed.

PV: protein value = multi-point assay relating response (weight or body N} to protein intake. Protein dietary levels selected so that points fall on straight-line portion of the response. When slope is expressed relative to lactalbumin slope, it becomes RPV, relative protein value. Nonprotein data not included.

Nitrogen growth index (NGI) is similar to PV except that regression lines are calculated for each protein but with the inclusion of the non-protein data. Relative nutritive value (RNV) uses the same data as NGI but is calculated using multiple regression analysis and is then expressed relativ to lactalbumin. Neither NGI nor RNV is shown in the diagram, but the regression lines would lie between the lines shown for PV and for NPUst, NPR and would cut the vertical axis above the zero intake point.

Relative nutritive value (RNV) is, in some ways, similar to RNPR, but includes several doses of both the test and reference protein. This method, based upon the classical slope-ratio procedure of Finney, utilizes a multipleregression model to force the regression lines through a common intercept (40). The ratio of the slope of the line for the test protein to that of the standard protein is calculated in an attempt to improve reproducibility of the assay among laboratories. Like RNPR, this method also suffer' from its assumption of a Common intercept for ail regression lines, and therefore yields overestimates for Iysine-limiting proteins.

Nitrogen growth index (NGl) is also a multi-level procedure and is one of the earliest slope-assay procedures for protein quality (18). It uses several (three or four) levels of a test protein and a non-protein control group; the same non-protein control data can serve for use with several sets of test-protein data. The procedure is almost identical to that for RNV and differs only in the statistical method used for evaluation of the data. As indicated above, a rigorous multiple-regression model is used in RNV to force the regression lines through a common intercept. In the NGI procedure, by contrast, separate linear regression lines are calculated for each test protein, the slopes obtained being the NGl. If a standard high-quality protein is tested at the same time, relative NGI can be obtained as a ratio of the slopes. The method is described in chapter 9 together with a worked example of the calculation. The method suffers the same limitations found with RNV for Iysine-limited proteins because it includes the zero-protein data in the regression analysis.

A modification of RNV, the relative protein value (RPV) procedure, alleviates the intercept problem mentioned above by calculating regression equations for the individual proteins without including the zero-protein response and without forcing the lines through a common intercept. The slope of the test protein is compared to that of the reference protein, as in RNV. It does not, however, as is indicated elsewhere, eliminate the problems of Iysine-limited proteins. This calculation reduces the variability among laboratories analysing the same proteins (40). Selection of the appropriate protein levels for the diets is determined, in part, by the amino acid composition of the proteins under test. More detailed consideration of protein ranges is included in chapter 9.

Concerns have been raised regarding "parallelism"-lines with similar slopes but different intercepts on the weight-gain axis-particularly with threonine-limiting protein sources (41). The multi-level assays require more animals and diet preparation, but little work has been reported to indicate the extent to which the procedure may be simplified while still maintaining reproducibility and discriminating ability.

A summary of the results of a collaborative assay involving seven laboratories analysing the same seven proteins by PER,NPR, and RPV is given in table 9. Values shown are means for the seven laboratories. The PER and NPR for lactalbumin were arbitrarily set at 100 so that these data would be directly comparable with the RPV values. PER gave lower values than given by the other methods for all samples, and the PER value for wheat gluten was particularly low. Agreement between NPR and RPV was reasonably good for all proteins except white flour and wheat gluten, which are Iysine-deficient.

Analysis-of-variance data from the same collaborative assay are shown in table 10 It can be seen that PER and NPR demonstrated significant variation among laboratories analysing the same proteins. Only RPV resulted in a significant reduction in laboratory-to-laboratory variability, presumably by reducing the effects of such variables as the use of different strains of animals and differing environmental conditions. The laboratory times protein interaction was also reduced, but remained significant. RNPR was not reported in the collaborative assay, but it may be assumed that a similar correction to a reference protein would improve the interlaboratory reproducibility of this method as well.

TABLE 9. Summary of the Results of a Collaborative Study Involving Seven Laboratories Analysing the Same Seven Proteins by Three Standardized Rat Bioassay Methods

Sample PER* NPR* RPV
Lactalbumin 100 100 100
Casein (HNRC) 54 75 83
Meat 57 80 77
Soy flour 65 76 70
Soy concentrate 47 74 70
White flour 21 48 31
Wheat gluten 5 43 22

Data of Samonds and Hegsted, unpublished e Both PER and NPR are expressed in relation to lactalbumin abitrarily set at 100. All according to standardizad procedure for length of assay.

TABLE 10. Summary of Analysis of Variance Results for a Collaborative Assay Involving Seven Laboratories Analysing Seven Proteins by a Variety of Methods

  Laboratories Proteins Laboratory x
    protein interaction  
PER 6.7* 68.8* 9.8*
NPR 5.5* 65.5* 7.4*
RPV 2.3NS 108.9* 2.1 *

Food samples analysed as in table 9.


It will be noted that the collaborative assay is referenced to an unpublished communication from K. Samonds and D.M. Hegsted. Full details of this study, with some 20 tables of data, were made available to the working group. It was originally intended that this would be published as an integral part of this document; however, for reasons of space, only two summary tables have been included together with a brief discussion of some of the results obtained.

Methods Based on Nitrogen Retention

Several methods use nitrogen retention as the dependent variable in a protein-quality assay. The simplest of these is net protein utilization (NPU), which measures the difference in carcass nitrogen between rats fed a test protein (NPU, carcass) and those fed a protein-free diet (13). The carcass method has been abbreviated by determination of body water content and derivation of nitrogen from the predetermined nitrogen/water ratio of the animals (42). The question of the constancy of this ratio has been explored by several workers, and was reviewed in 1973 (43). As with other approaches in which the zero intercept serves as a component of the estimation, the protein-quality estimate will depend upon the level of protein fed.

Assays may also be based upon nitrogen-balance methods in which nitrogen intake and excretion are determined for rats fed diets containing the test protein or a protein-free diet, and nitrogen retention is estimated indirectly. This allows determination of faecal and urinary nitrogen excretion of metabolic and endogenous origin. This provides for estimations of apparent digestibility (AD), true digestibility (TD), net protein utilization (NPU), and biological value (BV). Conventionally, protein is fed at a level of 100 g/kg (10 per cent) of the diet, and the result is designated NPU10. This conventional level permits comparisons among different proteins, although it is accepted that utilization is higher at lower levels of feeding and decreases as the dietary level is increased. On the other hand, this procedure provides a measure of TD that may be independent of the dietary protein level (44).

The concepts of NPU, operative, and net dietary protein calories per cent (NDpCal%) should also be considered briefly, as they are often believed to be indices of protein quality, a belief that is not entirely true. The utilization of protein (i.e., the percentage retained) falls with increasing concentration of protein in the diet (45), and it was proposed by some workers (46, 47) that two terms be used: net protein utilization, standardized (NPUst), and net protein utilization, operative (NPUop). NPUst would be determined at a low level of protein, usually at maintenance, while NPUop would be obtained at any higher (stated) protein level. Thus, NPUst can be considered a measure of quality and, unless amino acid availability or toxic factors impinge, should be a function of amino acid composition and score; whereas in contrast NPU Op is a measure of the overall protein value of the diet at a particular level of protein, i.e., a combined measure involving both quality and the lower utilization with increasing protein concentration. Thus, NPU Op is only meaningful for a specific diet at a specific level of protein. Since NPU falls continuously with increasing protein concentration (43, 48-50), the whole relationship between response and protein intake should be reflected by a continuous curve. However, the earlier part of the curve can be considered a straight line. NPUst and NPUop can be distinguished in figure 2, and it can be seen that NPUop is measured where increasing protein intake is no longer causing increased nitrogen retention. This concept does not conflict with multi-point assays where the slope is measured on the assumed straight portion of the line before curvature begins. Two-point assays (e.g., NPR and NPUst) are also measured on this assumed straight-line portion.

NDpCal%, a multiple of NPUop and PCal% (protein calories as a percentage of total calories), is an index designed to measure utilizable protein in the diet at the level at which it is consumed. It is important to observe that this index of utilizable protein is, as with NPU Op, a combined index of both quality and decreased utilization with increasing protein concentration. It differs from utilizable protein (UP), which indicates the "potential" amount of utilizable protein present and pays no attention to the fact that at high dietary protein concentrations much would not be retained. NDpCal% can be a useful index of the complementary value of proteins in diets, but it is concerned with a wider issue than protein quality alone.

It should be noted that NPR, the various NPU methods, and BV become two-level assays if response, i.e., growth or nitrogen retention, is compared with nitrogen or protein intake rather than with protein percentage. The slope of a line connecting the data of the test group with the protein-free group is then the index concerned, a straight-line relationship being assumed (fig. 2). No test of curvilinearity is possible, however, to test this assumption. Detailed methodology is given in chapter 9 for most of the procedures that have been discussed earlier in this chapter.

In Vivo Amino Acid Availability in Food and Feedstuffs

The faecal amino acid method for determining amino acid availabilities is analogous to the determination of true digestibility (TD) of the total protein. It consists of measuring the amount of amino acid ingested in the diet, the amount excreted in the faeces, and the so-called metabolic losses in the faeces. the latter is estimated from the amount of amino acid excreted by an individual fed a protein-free diet or a diet with completely digestible protein (either extracted, freeze-dried egg protein or casein), and is adjusted for differences in the feed intake in the same way as in the determination of the TD of a protein (51):

Availability = {amino acid intake - [faecal amino acid excretion - metabolic amine acid excretion] / amino acid intake}

The presence of a population of micro-organisms in the alimentary tract of conventional laboratory animals has given rise to considerable speculation as to the extent to which microbial activity influences the course of digestion and metabolism of protein and on whether the results are beneficial or detrimental to the host. It is generally agreed that the effect of the microbial activity increases with decreases in digestibility. However, it was demonstrated in experiments with rats (52) having reduced microbial activity that the microbial effect on amino acid availability was only marginal in a diet average in crude fibre (2.5 per cent).

While the TDs of the individual amino acids are, in most cases, found to be approximately the same as the TD of total nitrogen in the diet considered, it should be noted that the amino acids in different foods and feedstuffs are not directly additive because they may not be available to the body to the same degree. By comparing casein and barley, for example, it can be shown that approximately 20 per cent more amino acids are required as barley to equalize biological availability of the amino acids in an equivalent amount of casein protein (44, 51, 52).

Other procedures for the determination of in vivo amino acid availability are based upon direct animal growth experiments using various dietary levels of proteins and amino acids. Assays have been described for available Iysine (53), sulphur amino acids (54), and tryptophan (55).

Other Assays for Protein Quality

Of the many indirect methods suggested for protein evaluation, none seems to have been generally accepted. However, protein metabolism in humans and animals can be studied by a wide variety of indirect criteria (56). The results of several studies conducted for the most part with animals have suggested that various biochemical parameters may be useful indirect indices of the protein status of an individual.

It is well established that in severe protein malnutrition in humans the total protein concentration in the plasma is greatly decreased. The decrease is mainly in the albumin fraction. However, a fall in plasma albumin concentration is a relatively late event in short-term experiments with rats. Dietary protein quality was not found to affect total protein or albumin concentration in blood plasma (57).

The plasma amino acid method may provide valuable information concerning the limiting amino acid ( 58). However, several workers have failed to establish a relationship between dietary amino acid concentrations and increases in plasma amino acid concentrations during the absorption period. Furthermore, the blood amino acid procedure will always be an expensive one (29).

Results of several experiments confirm the thesis that blood urea increases as the protein quality of the diet decreases (28) or protein content increases. The significant relationship between the biological value of the diet and blood urea levels has been shown by several workers (28, 59). It thus appears that a determination of the blood urea level might be a valuable indirect method for predicting dietary protein quality. However, the experimental conditions must be strictly standardized.

Direct relationships have not been found to exist between the biological value of protein and urinary creatinine excretion levels in human subjects. No changes are observed in the creatinine excretion of human subjects ingesting either complete purified amino acid diets or diets deficient in valine, methionine, threonine, or histidine (60).

In conclusion, it should be stated that, in general, indirect assays of protein quality are usually not of sufficient validity for routine use. Further studies of procedures of this type would, however, appear to be warranted.

Conclusions: Relevance of Animal Assays to Human Protein Nutrition

The relevance of data obtained from animal bioassay procedures can be considered from two points of view: one related to methodology and the other related to the applicability of the animal assay values to estimate the nutritive value in human diets. In terms of methodology, standardized human assays are subject to the same limitations and provide the same type of information as do standardized animal assays. If the experimental approaches are appropriate and well standardized, the values resulting from animal assays are, with some exceptions, similar to those with human subjects (3, 61). For example, using the nitrogen growth index in rats and nitrogen balance index in children, the correlation between the two values for a series of proteins is high, and from these results it is evident that, with an appropriate choice of assay procedure, values obtained from rat bioassays have predictive significance in relation to human protein nutrition. Thus, values obtained from animal bioassays using multiple doses of the test protein can, under appropriate conditions, be used to estimate maintenance and growth in human subjects (61).

The second issue concerns the applicability of animal assay values for estimation of the absolute values for the efficiency of utilization of a protein(s) in diets consumed by humans. Values obtained from standardized animal assays such as RPV, NGI, and NPU designed to estimate the quality of a test protein are not necessarily definitive quantitative indices of the degree of utilization of proteins when consumed by humans. In addition, there are other dietary factors that influence protein utilization in practice, such as energy intake and amount of fibre. Therefore, it is essential to assay proteins in humans when they are to be used in human nutrition. Also, depending upon the protein in question, animal assays may underestimate the value of proteins as tested in humans, for example, soy isolates (62).

The application of animal assay values to human protein nutrition is perhaps more often qualitative than quantitative. The lack of relationship between values from experimental animals and human subjects is partly due to the fact that human studies have not yet reached the standardization that has been achieved in animal procedures. Furthermore, differences in amino acid requirements, for example, the relatively higher sulphur amino acid requirement of the rat, may account for part of the lack of agreement between animal and human assays.

Finally, in addition to dietary factors, the applicability of animal assay results is limited by the fact that the human diet and eating patterns differ from those of experimental animals. The latter are frequently fed diets of constant composition ad libitum, while humans consume diets that vary in amount and composition over periods of time. For example, a diet of corn and beans fed together to children gives higher protein quality values than when the two food components are fed separately (10).

For all of these reasons, the significance for human nutrition of values obtained from an animal bioassay procedure needs to be interpreted with caution.


1. J.A. Campbell and J.M. McLaughlan, "Applicability of Animal Assays to Humans," in Proceedings SOS/70, Third International Congress on Food Science and Technology Institute of Food Technology, Chicago,1971), pp. 336-343.

2. C. Kies and H.M. Fox, "Effect of Level of Total Nitrogen Intake on Second Limiting Amino Acid in Corn for Humans,"J. Nutr., 100: 1275-1285 (1970).

3. R. Bressani, "Human Assays and Applications," in C.E. Bodwell, ea., Evaluation of Protein for Humans (Avi Publishing Co.,Westport,Conn., USA, 1977),pp.81 - 118.

4. V.R. Young and N.S. Scrimshaw, "Nutritional Evaluation of Proteins and Protein Requirements," in M. Milner, N,S. Scrimshaw, and D.I.C. Wang, eds., Protein Resources and Technology (Avi Publishing Co., Westport, Conn., USA, 1978), pp. 136-173.

5. B.F. Chow, C. Alper, and S. DeBiase, "The Effects of Oral Administration of Different Proteins on the Plasma Proteins of Protein Depleted Dogs," J. Nutr., 36: 785-801 (1948).

6. J.M. McLaughlan and M.O. Keith, "Bioassays for Protein Quality," in M. Friedman, ea., Protein Nutritional Quality of Foods and Feeds (Marcel Dekker, New York, 1975), pp. 79 88.

7. L.R. Hackler, "Methods of Measuring Protein Quality: A Review of Bioassay Procedures," Cereal Chem., 54: 984-995 (1977).

8. C.E. Bodwell, "The Nutritive Value of the Same Protein Preparations as Estimated by Human, Rat and Chemical Assays," J. Am. Oil/ Chem. Soc., 56: 156-159 (1979).

9. A.E. Harper, "Effects of Disproportionate Amounts of Amino Acids," in Committee on Amino Acids, Food and Nutrition Board, National Research Council, Improvement of Protein Nutriture (National Academy of Sciences, Washington, D.C., 1974), p. 138.

10. R. Bressani, "Complementary Amino Acid Patterns," in P.L. White and D,C. Fletcher, eds., Nutrients in Processed Foods-Proteins (published for the AMA by Publishing Sciences Group, Inc., Acton, Mass., USA, 1974), pp.149-166.

11. H.H. Mitchell, "A Method of Determining the Biological Value of Protein," J. Biol. Chem., 58: 873-903 (1924).

12. N. Shukers and E.V. McCollum, The Newer Knowledge of Nutrition, 4th ed. (Macmillan, New York, 1929).

13. A.E. Bender and D.S. Miller, "A New Brief Method of Estimating Net Protein Value," Biochem. J., 53: vii-viii (1953).

14. T.B. Osborne, L.B. Mendel, and E.L. Ferry, "A Method of Expressing Numerically the Growth Promoting Value of Proteins,"J. BioL Chem., 37: 223-229 (1919).

15. A.E. Bender and B.H. Doell, "Note on the Determination of Net Protein Utilization by Carcass Analysis," Brit. J. Nutr, 11: 138-143 (1957).

16. P.R. Cannon, "The Relationship of Protein Metabolism to Antibody Production and Resistance to Infection," in Advances in Protein Chemistry, vol. 2 (Academic Press, New York, 1945), pp.135-154.

17. R.M. Tomarelli and F.W.J. Bernhart, "A Bioassay for Proteins and Protein Digests," J. Nutr., 34: 263-272 11947).

18. J.B. Allison and J.A. Anderson, "The Relation between Absorbed Nitrogen, Nitrogen Balance and Biological Value of Proteins in Adult Dogs,"J. Nutr., 29: 413 - 442 (1945).

19. D.M. Hegsted and Y. Chang, "Protein Utilization in Growing Rats. l. Relative Growth Index as a Bioassay Procedure,"J. Nutr.,85: 159-168 (1965).

20. D.M. Hegsted, R. Neff, and J. Worcester, "Determination of the Relative Nutritive Value of Proteins: Factors Affecting Precision and Validity," J. Agr. Fd. Chem., 16: 190-195 (1968).

21. F.S. Robscheit-Robbin and G.H. Whipple, "Dietary Effects on Anemia plus Hypoproteinemia in Dogs. I. Some Proteins Further the Production of Hemoglobin and Other Plasma Protein Production," J. Exp. Med., 89: 339358 (1949).

22. H.C. Harrison and C.N.H. Long, "Regeneration of Liver Protein in the Rat," J. Biol. Chem., 161: 545-557 (1945).

23. R.M. Campbell and H.W. Kosterlitz, "The Assay of the Nutritive Value of a Protein by Its Effect on Liver Cytoplasm," J. Physiol., 107: 383-398 (1948).

24. S. Mokady, S. Viola, and G. Zimmermann, "A New Biological Method for Estimating Food Protein Nutritive Value," Brit J. Nutr., 23: 491-495 (1969).

25. G. Litwack, J.N. Williams, L. Chen, and C.A. Elvehiem, "A Study of the Relationship of Liver Xanthine Oxidase to Quality of Dietary Protein,"J. Nutr., 47: 299 306 (1952).

26. B. Wirthen and H. Bergner, "Evaluation of Protein in Feeds. III. Enzyme Activity of GOT, GPT, OCT, LAP and Arginase in the Rat Liver as a Function of the Biological Value of Administered Nutrient Proteins," Arch. Tierernaehr., 17: 281-288 (1964).

27. J.R. Murlin, T.A. Szymanski, and E.C. Nasset, "Creatinine Nitrogen Percentage as a Check on the Biological Values of Proteins," J. Nutr., 36: 171-175 (1948).

28. H. Munchow and H. Bergner, "Untersuchungen zur Protein Bervertung von Futtermitteln. 2. Mitteilung,"Arch. Tierernashr., 17: 141 (1967).

29. B.O. Eggum, "The Levels of Blood Amino Acids and Blood Urea as Indicators of Protein Quality," in J.W.G. Porter and B.A. Rolls, eds., Proteins in tiuman Nutrition (Academic Press, New York and London, 1973), pp. 317328.

30. Joint FAO/WHO Ad Hoc Expert Committee, Energy and Protein Requirements, WHO Technical Report Series, no. 522; FAO Nutrition Meetings Report Series, no. 52 (WHO, Geneva; FAO, Rome, 1973).

31. Food and Nutrition Board, National Research Council, Recommended Dietary Allowances, 9th rev. edn. (National Academy of Sciences, Washington, D.C., 1980).

32. Committee on Amino Acids, Food and Nutrition Board, National Research Council, Improvement of Protein Nutriture, ed. A.E. Harper and D.M. Hegsted (National Academy of Sciences, Washington, D.C., 1974).

33. D.M. Hegsted and R. Neff, "Efficiency of Protein Utilization in Young Rats at Various Levels of Intake,"J. Nutr.,100: 1173-1178 (1970).

34 P.H. Derse, "Evaluation of Protein Quality (Biological Method)," J. Assoc. Offic. Agr. Chemists, 45: 418422 (1962).

35. D.G. Chapman, R. Castillo, and J.A. Campbell, "Evaluation of Protein in Foods. I. A Method for the Determination of Protein Efficiency Ratio," Can. J. Biochem. Physiol., 37: 679-686 (1959).

36. J.A. Campbell, "Evaluation of Protein in Foods for Regulatory Purposes," J. Agr. Fd. Chem., 8: 323-327 (1960).

37. AOAC, Official Methods of Analysis of the Association of Officia/ Analytical Chemists, 12th ea., ed. W. Horwitz (Washington, D.C., 1975).

38. P.L. Pellett,''Protein Quality Evaluation Revisited," Fd. Technol., 32 (5): 60-79 (1978).

39. G.R. Jansen, "Biological Evaluation of Protein Quality," Fd. Technol., 32 (12): 52-56 (1978).

40. K.W. Samonds and D.M. Hegsted, "Animal Bioassays: A Critical Evaluation with Special Reference to Assessing Nutritive Value for the Human," in C.E. Bodwell, ea., Evaluation of Proteins for Humane (Avi Publishing Co., Westport, Conn., USA, 1977), pp. 68-80.

41. J.M. McLaughlan, "Critique of Methods for Evaluation of Protein Quality," in H.L. Wilcke, D.T. Hopkins, and D.H. Waggle, eds., Soy Proteins and Human Nutrition (Academic Press, New York and London, 1979), pp. 281-298.

42. D.S. Miller and A.E. Bender, "The Determination of the Net Utilization of Proteins by a Shortened Method," Brit. J. Nutr., 9: 382-388 (1955).

43. P.L. Pellett, "Methods of Protein Evaluation with Rats," chap.15 in J.W.G. Porter and G.A. Rolls, eds., Proteins in Human Nutrition (Academic Press, New York and London, 1973), pp. 225-244.

44. B.O. Eggum, A Study of Cettain Factors Influencing Protein Utilization in Rats and Pigs, publ. no. 406 (National Institute of Animal Science, Copenhagen, 1973).

45. R.H. Barnes, M.J. Bates, and J.E. Maack, "The Growth and Maintenance Utilization of Dietary Protein," J. Nutr., 32: 535-548 (1946).

46. B.S. Platt and D.S. Miller, "The Net Dietary Protein Value (NDpV) of Mixtures of Foods-Its Definition, Determination and Application,"Proc. Nutr. Soc.,18: vii-viii (1959).

47. D.S, Miller and P.R. Payne, "Problems in the Prediction of Protein Values of Diets: The Influence of Protein Concentration," Brit. J. Nutr., 15: 11-19 (1961).

48. D.S. Miller and P.R. Payne, "Problems in the Prediction of Protein Values of Diets," J. Nutr., 74: 413-419 (1961).

49. D.S. Miller and P.R. Payne, "Nitrogen Balance Experiments: Some Theoretical Considerations,"Nature,204: 480-481 (1964).

50. P.R. Payne, "Protein Quality of Diets, Chemical Scores and Amino Acid Imbalances," in The International Encyclopaedia of Food and Nutrition, vol.11: Protein and Amino Acid Functions, ed. E.J, Bigwood (Pergamon Press, Oxford and New York, 1972), pp. 259-306.

51. K.A. Kuikan and C.M. Lyman, "Availability of Amino Acids in Some Food," J. Nutr., 36: 359-368 (1948).

52. B.O. Eggum, M. Fekadu, J. Wolstrup, W.C Sauer, and A. Just, "The Effect of Dietary Antibiotic on Protein and Energy Metabolism in Rats: Possible Significance of the Gut Microflora,"J. Sci. Fd. Agric, 30: 177-184 (1979).

53. F. Mottu and J. Mauron, "The Differential Determination of Lysine in Heated Milk, II. Comparison of the In Vitro Methods with the Biological Evaluation," J. Sci. Fd. Agric., 18 57-62 (1967).

54. K.J. Carpenter, I. McDonald, and W.S. Miller, "Protein Quality of Feeding Stuffs: Collaborative Studies on the Biological Assay of Available Methionine Using Chicks," Brit. J. Nutr., 27: 7-17 (1972).

55. J D. Gupta and C.A. Elvehjem, "Biological Availability of Tryptophan," J. Nutr., 62: 313-324 (1957).

56. G.R. Jansen, "Biochemical Parameters and Protein Quality," in C.E. Bodwell, J.S. Adkins, and D.T. Hopkins, eds., Protein Quality in Humans: Assessment and In Vitro Estimation (Avi Publishing Co., Westport, Conn., USA, 1980 [in press]).

57. B.O. Eggum and I. Jacobsen, "Blodplasmaets Protein-og Albuminkoncentration samt Blodets Haemoglobinkoncentration hos Rotter sat i relation til Foderproteinets Kvalitet," publ. no.142 (National Institute of Animal Science, Copenhagen, 1977).

58. J.M. McLaughlan, "Nutritional Significance of Alterations in Plasma Amino Acids and Serum Proteins," in Committee on Amino Acids, Food and Nutrition Board, National Research Council, Improvement of Protein Nutriture (National Academy of Sciences, Washington, D.C., 1974), p. 89.

59. B.O. Eggum, "Blood Urea Measurement as a Technique for Assessing Protein Quality," Brit. J. Nutr., 24: 983-988 (1970).

60. W.C. Rose, W.J. Haines, D.T. Warner, and J.E. Johnson, "The Role of Threonine and Histidine,"J. Biol. Chem., 188: 49-58 (1951).

61. R. Bressani, F. Viteri, and L.G. Elias, "The Usefulness of Results with Human Beings in Standardizing Rat Assays," in J.W.G. Porter and B.A. Rolls, eds., Proteins in Human Nutrition (Academic Press, New York and London, 1973), pp. 293-316.

62. V.R. Young, N.S. Scrimshaw, B. Torun, and F. Viteri, "Soybean Protein in Human Nutrition: An Overview,"J. Am. Oil. Chem. Soc., 56: 110-120 (1979).

Contents - Previous - Next