|
|
||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

* Department of Food Science, Cornell University, Ithaca, New York 14850; and
Department of Animal Science, Cornell University, Ithaca, New York 14850
1 To whom requests for reprints should be addressed at 119 Stocking Hall, Cornell University, Ithaca, NY, 14853. E-mail: ddm2{at}cornell.edu
| Abstract |
|---|
|
|
|---|
Key Words: pig model human nutrition diet
| Introduction |
|---|
|
|
|---|
Currently, much attention is being paid to elucidating the exact mechanisms by which the diet can initiate the consequences outlined above, as well as investigating means of preventing or alleviating these conditions using various diet regimes and dietary supplements. This could be achieved through fortifying commonly consumed food products with the necessary vitamins, minerals, and other nutrients, or through the consumption of dietary supplements such as probiotics, prebiotics, and organic acids (16–20). To elucidate the mechanisms involved in dietary effects on health, particularly with regard to mineral absorption across the intestinal epithelium, it is sometimes necessary to access the different compartments of the gastrointestinal tract (GIT). While access to the human GIT can sometimes be obtained in a hospital environment using patients with ileostomies and colonoscopies, or patients that have been intubated, such procedures are costly and laborious, and may be hampered by the inability to locate enough willing participants and ensure their full compliance (21). Often, the more suitable approach is the use of animal models, with the option of using cannulas and catheters to access the GIT of live animals, or euthanasia to allow for excision of different GIT compartments.
Traditionally, rats have been the animal model of choice when performing nutritional studies. However, the rat model has a number of limitations which makes extrapolation back to a human situation questionable, including a significantly different food intake and energy expenditure for body size, a different lifespan and body proportion, differences in intestinal morphology and enteric microbiota, as well as other distinct physiological differences (21, 24). Another major problem with using rat models for mineral studies is their propensity for practicing coprophagy. While this is an effective way for the animals to recycle nutrients and maximize nutrient absorption, it may have a dramatic impact on the results of a nutritional study.
Although no animal model will ever perfectly mimic the human condition, the pig has emerged as a superior non-primate experimental animal model because of its much closer resemblance to humans. This review highlights the similarities between pigs and humans and thereby the value of the porcine human nutrition model, and reviews some of the more recent applications of this model for nutritional research.
| Comparative Gastrointestinal Tract Anatomy of Pigs and Humans |
|---|
|
|
|---|
|
Anatomically, the gastrointestinal tract of pigs is similar to that of humans, although the division between the duodenum, jejunum, and ileum are not as distinct in the porcine small intestine as they are in humans (29, 50). In addition, the stomach of pigs has a muscular outpouching of uncertain function termed the torus pyloricus, in the pyloric region near the gastro-duodenal junction (29, 52–53). This outpouching is not present in the human stomach. However, by far the most distinct difference between humans and pigs is the spatial arrangement of the intestine within the abdominal cavity (Fig. 1
), particularly that of the large intestine. The small intestine in humans is, for the most part, situated behind the large intestine in the abdominal cavity; whilst the small intestine of pigs is arranged in the right side of the abdomen. In humans, the large intestine is arranged in a square-like configuration. The ascending colon extends upwards from the ileo-cecal junction, where it turns left and becomes the transverse colon which stretches laterally from right to left in the abdomen. Once there, it makes a downward turn to become the descending colon, which continues downwards and becomes S-shaped to form the sigmoid colon that lies posterior to the urinary bladder and empties into the rectum (26, 50). In contrast, the greater proportion of the pig large intestine, consisting of the cecum, proximal (ascending), mid (transverse), and the majority of the distal (descending) colon, is found in a spiral conformation beginning mid-abdomen and spiralling toward the left upper quadrant of the abdomen in a series of clockwise and anticlockwise coils (26, 29, 54). The remainder of the descending colon passes posteriorly along the left abdominal wall to the rectum. Incidentally, pigs also do not possess an appendix at the terminal end of the cecum (54).
|
| Application of the Porcine Model in Nutritional Research |
|---|
|
|
|---|
A. Mineral Bioavailability Studies.
Mineral absorption has been a primary concern of much nutritional research over the last few decades, with a lot of attention being paid to understanding the "normal" absorptive capacity of the intestine for different trace minerals, and the homeostatic controls that regulate their uptake. A variety of techniques have been developed for assessing nutrient retention and absorption, each with their own advantages and disadvantages. One of the simplest methods for indirectly measuring absorption of an ingested nutrient is to assess fecal and urinary excretion levels, and then differentially calculating the absorption/retention based on the dose ingested. Another option involves the use of radioisotopes. Whole body counting can then be performed to determine retention levels, while individual tissues can also be assayed to determine distribution patterns throughout the body. To avoid the hazards associated with the use of radioactive materials, the radioisotope could be replaced with a stable isotope. Isotope absorption can then be determined by measuring the changing isotopic ratios in tissue, blood, and/or urine against the more abundant, natural isotopic form (8). The availability of cannulation procedures also provides a means of assessing nutrient absorption on a compartmental basis. Other nutrient specific techniques may also be available, such as the hemoglobin repletion assay reported by Perks & Miller (40) which can be used to obtain a relative measure of iron absorption. Blood is sampled at the start of the feeding period for measurement of initial hemoglobin concentrations. The animals are then fed the experimental diets for a period of 2 to 5 weeks, after which a second blood sample is drawn for determination of final hemoglobin concentrations. Feed intakes are measured throughout this entire period to allow calculation of iron intake. Blood volume is estimated from body weight and the following formulas are used to calculate "Hemoglobin Repletion Efficiency" (HRE):
![]() |
![]() |
Where, Hb Fe = total body hemoglobin iron, and Hb = hemoglobin concentration in blood.
This technique is particularly useful in anemic pigs who would respond rapidly to iron increases in the diet (19, 40). Table 2
summarizes some of the more recent studies which investigated mineral absorption and retention using the porcine model.
|
Along with calcium, zinc is another whose deficiency is emerging as a widespread problem (62–63). Owing to the important consequences of zinc deficiency on growth, immunity, and everyday metabolic and physiological processes, a lot of research is being directed toward alleviating this problem, and the porcine model is increasingly being utilized for this research. Poulson & Larsen (64) studied zinc absorption and retention in swine by measuring fecal and urinary output following consumption of increasing levels of dietary zinc. They found that increasing the zinc content of the diet from the natural level of 42 mg zinc per kg diet up to 162 mg per kg could increase zinc absorption from 19% to 22–26%. However, homeostatic mechanisms ensured excessive zinc absorption did not occur, as further increases in dietary zinc concentration beyond 162 mg per kg did not yield further increases in percentage absorption. A similar result was also noted by Carlson and colleagues (65) who fed increasing dietary levels (0, 125, 250, 375 or 500 ppm Zn) of a proprietary zinc polysaccharide (Sea-Questra Min Zinc, Quali Tech, MN) to swine, and compared the absorption rates to that of control pigs administered a corn-soybean meal diet supplemented with 165 ppm Zn as ZnSO4. They discovered that zinc absorption in pigs administered the control diet (165 ppm) averaged 20.9%, while that of the zinc polysaccharide averaged 25.0% and 25.9%, at concentrations of 125 ppm and 250 ppm in the diet, respectively. No further increases in percentage absorption of this zinc polysaccharide were attained with further increases in concentration beyond 250 ppm in the diet; in fact, percentage absorption was seen to decrease at higher Zn-polysaccharide concentrations. The zinc absorption values reported in the literature using the pig model are remarkably similar to those reported in human studies, which generally average between 20–40%, depending on other diet components and the zinc status of the individual (66–67).
The use of the pig as an experimental model for measuring iron absorption and retention has only recently become common. Previously, rodent models have been utilized for such determinations, in spite of their limitations. Nevertheless, in a recent study, Zinn and colleagues (68) utilized the piglet as a model for human infants to study the absorption of radioactive iron (59Fe), administered as either elemental iron or ferrous sulfate, from a rice-meal cereal diet. The apparent absorption of elemental iron in the piglets averaged 13%; as expected, the absorption of ferrous sulfate was significantly higher than this, averaging 26%. In contrast to this result, Apgar & Kornegay (69) saw much lower iron absorption values in their pig study. Absorption values averaged between 5 and 9%, based on an analysis of fecal and urinary iron excretion, in pigs administered a corn-soybean meal diet supplemented with an excess level (350 ppm) of iron (iron source not given). It is worth noting, however, that in the study of Zinn iron injections at birth were withheld, which would have affected the iron status of the piglets during the study period. It would be expected that anemic and/or iron deficient animals would absorb more dietary iron, in an attempt to rectify their mineral deficiency, than would pigs with ample body stores of iron. Similar variations in iron absorption levels have also been observed in humans, and can be the result of a number of interactions, including iron status and diet composition.
Chemical Form/Fortificant.
An understanding of the "normal" absorptive capacity of the intestine with respect to different minerals is vital to human nutrition, as is an appreciation of the changes in these absorptive functions in deficient states. The studies discussed in the previous section have provided a valuable contribution to this knowledge of human nutritional requirements. Many of these studies, in addition to others, have also highlighted the critical importance of the chemical form of fortificant utilized in the treatment of mineral deficiencies, as the bioavailability of these different chemical forms can be significantly different. For instance, Pointillart et al. (31) used a pig model to analyze bone calcium content so as to compare calcium bioavailability from milk and supplemental calcium salts. They detected significantly higher bone calcium content in pigs administered milk, as compared to the groups administered supplemental calcium salts, given as either calcium carbonate or calcium sulfate. These results suggest a significantly higher bioavailability of calcium from milk over calcium salt supplements, a result of importance for human nutrition.
In a different study of zinc supplementation, Cheng et al. (70) examined zinc levels in the liver, kidney, and ribs of pigs following administration of differing dietary zinc levels from different zinc sources, in conjunction with increasing dietary lysine concentrations. As expected, tissue zinc levels increased with the dietary zinc concentrations tested in their experiment. In addition, zinc concentrations were lower in the kidneys and ribs of pigs administered the higher dietary lysine levels. However, no significant impact of zinc source or lysine concentration on zinc absorption across the intestine was noted in this particular experiment.
In a study of iron bioavailability, Maekawa et al. (71) utilized the porcine model to compare the bioavailability of hydrogen-reduced (HR) elemental iron powder, added to bread either before or after baking. The change in hemoglobin levels over the 16-day treatment period was used to calculate the hemoglobin repletion efficiency (HRE) and relative biological value (RBV) of the two breads, as compared to FeSO4 fortified bread whose RBV was set at 100%. The HRE was found to be 8.7 ± 3.0 and 7.5 ± 1.3 for the HR Fe added to bread before and after baking, respectively; the HRE of FeSO4 fortified bread was found to be 18.7 ± 2.8. As mentioned in a previous section, the HRE gives an approximation of iron absorption. The RBV of the bread diets were 53.5% and 40.1% for bread with Fe added before and after baking, respectively (P > 0.05). These results suggested that baking does not improve the bioavailability of hydrogen-reduced elemental iron powders in unenriched, refined wheat flour; and that hydrogen-reduced elemental iron has a significantly lower bioavailability than FeSO4, which is the current "gold standard" for human iron fortification.
Impact of Dietary Mineral Content on the Bioavailability of Other Minerals.
Another factor that can significantly affect the bioavailability of specific minerals is the presence (or absence) of other minerals within the diet, and their abundance relative to the mineral of interest. In a recent study, Atkinson and colleagues (72) used a pig model to study the effect of a combination of calcium and phosphorus on zinc, copper, and iron absorption across the intestine. Piglets were fed a complete liquid diet supplemented with calcium and phosphorus, as well as one of the following: zinc alone; zinc plus copper; or a combination of zinc, copper, and iron. After a 5-day adaptation period, piglets were orally and intravenously dosed with a radioactive isotope mix containing zinc, manganese, iron, selenium, and calcium. Isotope levels in the body were measured after a 15-day period, during which time fecal excretion of any unabsorbed isotope was monitored. Dietary supplementation with calcium, phosphorus, and zinc reduced isotopic zinc uptake compared to control pigs. Increased intakes of a combination of calcium, phosphorus, zinc, and copper tended to reduce iron absorption as well. Unfortunately, due to the study design it was not possible to determine whether only one, or a combination of these elements, was causing these alterations in zinc and iron absorption. Nevertheless, this study demonstrated that the presence of certain minerals in the diet can impact upon the absorption of other minerals, although the exact mechanisms behind these interactions still remain to be clearly elucidated. Such a result has important implications for animal and human nutrition.
In contrast to this, Zinn and colleagues (68) studied the effect of dietary iron and zinc on the retention of radioisotopes of iron, zinc, copper, and calcium in young pigs. They found no observable effect of dietary iron or zinc on the retention of orally administered zinc, copper, or calcium. A similar result was observed by Apgar & Kornegay (69), who studied the impact of increasing copper levels, administered as either copper sulfate or a copper-lysine complex, on copper, iron, and zinc absorption. While copper absorption was seen to increase with increasing copper dose, irrespective of the copper source, no significant impact of copper on zinc or iron absorption was observed.
Impact of Other Dietary Constituents on Mineral Bioavailability.
The porcine model has also been utilized to determine the impact of other dietary nutrients/factors on mineral absorption. In a recent study, South et al. (19) demonstrated that in iron deficient swine, meat consumption increased nonheme iron absorption, a result that concurred with previously obtained human bioavailability data (73–75). In a study of ascorbic acid supplementation, Pointillart et al. (76) concluded that it had no detectable impact upon calcium or phosphorus absorption measured by fecal and urinary output and bone mineral content. In a 10-day feeding trial of ascorbic acid supplementation, Perks & Miller (40) demonstrated that ascorbic acid had no effect on the bioavailability of iron from an iron-fortified milk product. This result was in agreement with a number of human studies of similar duration, but was in direct contrast to numerous single meal studies which concluded that ascorbic acid can promote iron bioavailability (77–81). This suggests that single meal studies may overestimate the impact of some dietary factors on mineral absorption, and therefore may not reflect true long-term bioavailabilities. Hence, there exists a need to confirm any bioavailability data obtained from short-term studies over a longer period of time, before extrapolation of data back to humans. The use of a porcine model would permit studies of longer duration and would overcome the difficulties in ascertaining compliance of human subjects over longer time periods.
B. Absorption Site.
The aforementioned studies utilizing pig models have provided valuable insight into the retention of minerals, and the interaction between different dietary minerals and their subsequent absorption. In most cases, however, the methods adopted do not provide more specific information about the absorptive capacity of the different GIT compartments, particularly the different regions of the small intestine. One of the major advantages of using animal models is that they can provide access to the different GIT compartments, regions which are not as readily accessible in humans. Although nutrient absorption primarily occurs in the small intestine, absorption can also occur to a limited extent in the colon. Therefore, measurement of colonic absorption rates can also be of importance, particularly when diets are geared toward altering the large intestinal environment.
Table 3
summarizes some recent studies that have investigated nutrient absorption in the large intestine. To date, most studies investigating colonic absorption of nutrients have been conducted using ex vivo approaches, such as the Ussing chamber technique. However, the availability of cannulation procedures provides a useful means of monitoring gastrointestinal uptake of nutrients on a compartmental basis, and allows for greater control over nutrient levels reaching the different intestinal regions (30, 84–86).
|
C. Regulation of Nutrient Absorption.
While adverse consequences can arise when certain nutrients are limiting in the body, there can also be significant consequences associated with nutrient overload. For instance, iron deficiency has been associated with impaired physical work performance, cognitive impairment, adverse pregnancy outcomes, and irreversible developmental delays in infants and toddlers (87). In contrast, excess iron can be lethal because it can catalyze the formation of free radicals through the fenton reaction, leading to a cascade of deleterious outcomes for the host (88–90). To prevent the adverse effects associated with nutrient limitation or overload, the body must tightly regulate absorption and excretion processes. One way of controlling nutrient status, so as to prevent deficiency or overload, is to tightly regulate absorption rather than rely on excretion processes to eliminate excess nutrients. Certainly in the case of iron, as well as some other minerals, excretion is very limited, and therefore iron status must be closely regulated at the site of absorption (6). The most economical way to modulate absorption is through substrate dependent regulation of nutrient transporter protein levels on the cell membrane and within the cytoplasm. A reduction in nutrient transporter levels therefore leads to a reduced absorption, and has the added benefit of improving cell efficiency, as any surplus protein would monopolize valuable energy and space (91). During times of nutrient limitation, the appropriate transport proteins are up-regulated, so as to enable effective nutrient scavenging from the depleted environment.
While the regulation of nutrient absorption in this manner sounds quite simple, in reality this is a very complex process, and there are many different pathways and intermediate products involved in this regulation. Recent investigations using murine and rat models have begun to shed light on the complex processes involved in these regulatory pathways for various minerals, including calcium, iron, and zinc; as well as the various interactions these pathways have with other bodily processes, including the immune system. The ease of inducing gene knockouts and nutrient disorders in these particular animal models, as compared to the porcine model, is a significant factor to consider when performing pioneer studies of this type. However, in light of the limitations of murine and rodent animal models it is important to confirm any findings with a more relevant model such as pigs to ensure those regulatory pathways, and their agonists and/or antagonists, are maintained between models. This can be of particular importance when examining expression of genes coding for nutrient transporters present on intestinal enterocytes, such as the divalent metal transporter, DMT-1, that is involved in transport of iron and other divalent metals. To compare the expression of the DMT-1 gene in different GIT compartments, or to measure functional protein levels, it may be necessary to isolate intestinal enterocytes from these regions. This can be a complicated process if using human subjects, and therefore the use of a porcine model in place of a human model for these analyses can supplement the evidence obtained using rat and mouse models.
Until now, the pig model has not been utilized to a great extent for studying nutrient regulatory pathways, as many of these have only recently been elucidated using rodent and/or murine models. Current studies are under way in our laboratories using a porcine anemia model to compare iron regulatory pathways that have recently been identified in mice and rats, and to assess their corresponding function during iron limitation in pigs.
D. Prebiotics, Probiotics and the Enteric Microbiota.
An important and often overlooked effector of nutrient bioavailability is the enteric microbiota. Intestinal bacteria possess specific nutrient requirements for growth and largely meet these needs through metabolism of host diet components and exploitation of host resources. Whilst the maintenance of what is considered a beneficial balance of enteric microbes provides crucial protection against pathogen incursions, and thereby maintenance of intestinal health, the balance of enteric bacteria can also potentially affect nutrient bioavailability. For instance, most pathogenic bacteria are highly dependent on iron for survival, and the host immune system takes advantage of this as a means of combating infection. By sequestering iron away from the pathogens their growth can be impeded, allowing other components of the immune system to overwhelm them (92–93). Unfortunately, this critical immune response can also have adverse implications for the host in times of chronic infection and lead to what is termed the "anemia of inflammation", as iron availability to the host is also limited (94–95). While a large number of GIT pathogens are exogenous to the intestine, it is important to recognize that some are actually endemic to the GIT in limited numbers. These opportunistic pathogens are generally kept in check by the host immune system and by other resident commensal bacterial populations; that is until a situation arises (e.g. antibiotic administration; considerable change in diet) which provides them the opportunity to overwhelm this protection and proliferate to greater numbers, thereby causing infection. These pathogenic populations have an intrinsic iron requirement which is obtained either from the host diet or from host cells. In contrast, non-pathogenic microbes, such as lactobacilli and bifidobacteria, tend to have low iron requirements (96–97). For that reason, modulation of the enteric microbiota can potentially have a significant impact on mineral bioavailability to the host. For instance, a microflora with high proportions of lactobacilli and bifidobacteria should have a lower iron requirement, and this may lead to more available iron for the host.
Up to this point, the enteric microflora has not received significant attention relating to its impact on nutrient bioavailability, particularly with respect to minerals such as iron, which are believed to be primarily absorbed in the upper small intestine. However, the small intestine also possesses a commensal microflora, albeit in much smaller numbers than that residing in the large intestine. Nevertheless, it may be likely that this microflora plays an important role in nutrient bioavailability, and the promotion of a beneficial microflora may provide a valuable treatment option for many micronutrient deficiencies, either alone or in conjunction with a supplementation regime. Prebiotics and probiotics represent such a way of modulating the enteric microbiota.
Probiotics are defined as: "Live microorganisms, which when administered in adequate amounts confer a health benefit on the host"; while a prebiotic is: "A nondigestible food ingredient that beneficially affects the host by selectively stimulating the growth and/or activity of one or a limited number of bacteria that can improve host health" (98–99). These dietary supplements are currently being advocated as a therapeutic/preventative measure for many intestinal and extra intestinal diseases and disorders including inflammatory bowel disease, diarrhea and metabolic syndrome, but may also have applications in mineral nutrition (20, 100–102). Prebiotics in particular perform a dual role in the gastrointestinal tract. These carbohydrates resist digestion by pancreatic and salivary amylases in the small intestine, thereby minimizing postprandial glucose availability and providing relief from metabolic syndrome (or insulin resistance) and other related extra-intestinal disorders, including type II diabetes mellitus and cardiovascular disease (5, 7, 103). They also fulfil a secondary function in the large intestine, as the undigested material accumulates in this region and promotes a favorable enteric microbiota through the promotion of "friendly" bacterial populations such as bifidobacteria and lactobacilli, at the expense of pathogenic or opportunistic populations such as clostridia, enterobacteria and proteolytic bacteroides species (104–107). Another benefit derived from the beneficial modulation of the colonic flora is increased production of short chain fatty acids (SCFA), particularly butyrate, which is an important energy source for colonocytes (22, 108–109). Some better known prebiotics include inulin, fructose oligosaccharides (FOS), and galactose oligosaccharides (GOS).
As a consequence of their multi-faceted role in the gastrointestinal tract, prebiotics can potentially affect mineral bioavailability by a variety of mechanisms. This can include a reduction in intestinal pH through promoting the production of SCFA, which can in turn increase mineral solubility; the promotion of a reductive environment in the intestine, which may prevent mineral loss due to precipitate formation; promotion of epithelial cell proliferation in response to SCFA, thereby increasing the available surface area for mineral absorption; or by affecting the expression of mineral transport proteins or regulatory genes involved in the absorptive process (110–112). A prebiotic effect on mineral availability may also arise through modulation of the enteric microbiota. Each microbial population has differing requirements for specific minerals and the provision of these nutrients in the intestinal lumen can potentially be influenced by changing enteric microbial population dynamics. Currently, there is much research being directed toward understanding the mechanisms by which prebiotics can influence mineral availability (113–115).
Several studies utilizing rat models have demonstrated that prebiotic supplementation can influence mineral bioavailability (113, 116–119). At present, however, there is limited data on mineral availability in pigs arising from prebiotic supplementation, in spite of the abundance of literature pertaining to the impact of prebiotics on the porcine enteric microflora and intestinal environment (SCFA production etc). In one recent study, Yasuda et al. (120) showed that supplementation of a corn-soybean meal diet with 4% dietary inulin increased iron bioavailability to pigs measured by the hemoglobin repletion efficiency assay. In another study by Houdijk and colleagues (121), no effect of FOS or TOS on absorption of calcium, magnesium, iron, phosphorus, copper, or zinc in weaner and grower pigs was found. In contrast, the infant pig model adopted by Morais et al. (122) showed a promotion of intestinal calcium and iron absorption in response to dietary supplementation with resistant starch.
Evidently, more research utilizing the pig model is required to elucidate the impact of dietary supplements such as prebiotics on mineral bioavailability. The use of the porcine model for these studies is of particular importance because of the similarity between their enteric microbiota and that of humans, in addition to the other various similarities presented earlier. While it is possible to utilize a rodent model with a human associated enteric flora, the many variances between human and rodent digestive physiology and function still significantly limit the conclusions which can be made from such studies and their applicability to humans.
| Limitations of the Porcine Model |
|---|
|
|
|---|
As is the case in humans, body fat content and distribution in pigs varies markedly depending on age, energy balance, and genotype (128–129). Newborn humans have much higher levels of body fat (16%) compared to pigs (1%) (130). Body fat peaks at about 26% at 4 months in humans, and then gradually declines to 18% at 36 months (131). Pigs become severely obese when given ad libitum access to feed, and pigs fed high fat diets have higher body fat content than pigs fed low fat diets (132). There is little evidence that body fat content affects nutrient absorption in the intestine. However, Bekri and colleagues (133) reported a high prevalence of anemia in severely obese human patients. They showed elevated expression of hepcidin (both mRNA and protein) in the liver and adipose tissue of these patients, and suggested that this may be due to the chronic inflammation that is common in obese subjects. Hepcidin is known to inhibit iron absorption by blocking the baso-lateral export of iron from enterocytes (134). Therefore, it is possible that differences in body fat content between pigs and humans could translate into differences in nutrient absorption but this seems unlikely except in cases of severe obesity.
Incidentally, pigs have also been known to practice coprophagy which can be another confounding experimental factor. Albeit this practice is quite rare in pigs, as compared to rats that frequently practice coprophagy.
| Summary |
|---|
|
|
|---|
| Footnotes |
|---|
| References |
|---|
|
|
|---|
-cellulose. Am J Clin Nutr 40:1071–1077, 1984.