Celiac Disease

Donald D Kasarda. Cambridge World History of Food. Editor: Kenneth F Kiple & Kriemhild Conee Ornelas. Volume 1. Cambridge, UK: Cambridge University Press, 2000.

Celiac disease has been recognized for centuries (Dowd and Walker-Smith 1974) by physicians aware of its major symptoms of diarrhea and gastrointestinal distress accompanied by a wasting away in adults and a failure to grow in children. The Greek physician Aretaeus (first century A.D.) called the condition coeliac diathesis—coeliac deriving from the Greek word koeliakos, or abdominal cavity. The British physician Samuel Gee provided what is generally considered the first modern, detailed description of the condition, which he termed the coeliac affection in deference to Aretaeus, in a lecture presented at St. Bartholomew’s Hospital in London (Gee 1888). At present, celiac disease (or, especially in Britain, coeliac disease) is the most commonly used term for the condition, although various others may be encountered, including celiac syndrome, celiac sprue, nontropical sprue, and gluten-sensitive enteropathy.

There were perceptions, certainly since Gee’s time, that celiac disease was a consequence of, or at least affected by, diet. Gee (1888: 20) noted that “[a] child, who was fed upon a quart of the best Dutch mussels daily, throve wonderfully, but relapsed when the season for mussels was over.” Such associations with diet led to wide-ranging dietary prescriptions and proscriptions (Haas 1924; Sheldon 1955; Weijers, Van de Kamer, and Dicke 1957; Anderson 1992). Some physicians recommended exclusion of fats—others, exclusion of complex carbohydrates. At times, so many restrictions were applied simultaneously that it became impossible to maintain a satisfactory intake of calories.

Because dietary treatments of celiac disease were of limited effectiveness, the food connection remained a puzzle until the end of the 1940s when a Dutch physician, W. K. Dicke, observed that removal of wheat from the diet of celiac patients led to dramatic improvement (Dicke 1950). It eventually became clear that the efficacy of the various diets recommended prior to Dicke’s discovery was in proportion to the extent that wheat was excluded from them. Initial fractionation studies pointed to the gliadin protein fraction as being most harmful to celiac patients (Van de Kamer, Weijers, and Dicke 1953).

Soon after Dicke (1950) reported the harmful effects of wheat on celiac patients, a series of investigations indicated that rye, barley, and oats were also harmful, whereas rice and maize (corn) were not (Dicke, Weijers, and Van de Kamer 1953; Van de Kamer et al. 1953; Weijers et al. 1957). With children, and most adults, the exclusion of wheat, rye, barley, and oats from the diet usually brought about a complete, or largely complete, recovery.

During the 1950s, the development of intestinal biopsy techniques (Shiner 1956; Crosby and Kugler 1957; Brandborg, Rubin, and Quinton 1959) enabled pieces of tissue to be recovered from the intestine for examination and testing, and it was recognized that ingestion of wheat and related grains often resulted in damage to the intestinal mucosa, including the absorptive cells, or enterocytes, lining the interior surface of the intestine. The enterocytes are responsible for the absorption of almost all nutrients; the damage to them provided a basis for the gastrointestinal symptoms and malabsorption.

Familial associations, as well as its rarity among the Chinese, the Japanese, and blacks in sub-Saharan Africa (McNeish et al. 1974), indicated a likely genetic basis for celiac disease. Subsequent findings of strong associations with particular histocompatibility antigens supported this possibility (Kagnoff 1992) and, along with the presence of circulating antibodies to wheat gliadin proteins in patients on a gluten-containing diet, suggested that an abnormal immune response initiated intestinal damage in susceptible individuals.

A detailed understanding of the basis for celiac disease remains to be achieved. Neither the initiating event triggered by wheat gliadin proteins or products derived from them by digestion, nor the mechanisms leading to tissue damage in the small intestine following ingestion of wheat are completely understood. On the basis of current knowledge, it appears that celiac disease has resulted from the convergence of various evolutionary developments: These include human evolution (especially evolution of the immune system), the evolution of wheat and related grasses, the evolution of the protein structures apparently unique to wheat and closely related species, and the evolution of culture—specifically the development of agriculture and the spread of wheat farming.

Geographical Distribution of Celiac Disease

Celiac disease is commonly thought of as largely afflicting people of European ancestry, and reports of cases among the Chinese, the Japanese (Kihara, Kukida, and Ichikawa 1977), and black Africans are sufficiently rare as to make it unlikely that this assumption is incorrect (also see McNeish et al. 1974; Simoons 1981). However, studies throughout much of the world either are inadequate or have yet to be done.

There are some parts of Asia where rice, a harmless grain (see ahead for definitions of harmful grains), is the predominant cereal grain in the diet, and in some parts of Africa, teff and millet—likely to be harmless cereal grains—predominate. To what extent a low intake of wheat, rye, barley, and oats in Asia and Africa contributes to the apparently low incidence of celiac disease is unknown. Furthermore, in some parts of the world—the United States, for example—medical personnel may have only minimal knowledge of celiac disease and fail to recognize it, whereas in Britain and much of Western Europe, physicians are much more attuned to its signs, which may be highly varied. The likelihood of failure to recognize celiac disease is doubtless substantially greater in some Asian and African countries and may contribute to the perception that it is rare in these places. The difficulties in diagnosing celiac disease and in developing statistical information about its incidence and prevalence have been discussed by R. F. A. Logan (1992a, 1992b).

The incidence of celiac disease varies throughout Europe and with time. For example, crude incidence rates (number of cases per 1,000 births) in various parts of Sweden currently range from about 2.2 to 3.5 per 1,000, whereas in neighboring Denmark, the rate is about 0.1 per 1,000 (Greco et al. 1992). Furthermore, the incidence varies with time, possibly indicating the contribution of some environmental factor. Prior to 1983, incidence of the disease in Sweden was much lower, but it began to increase at about that time, and it has been speculated that an increase in the amount of gluten in infant formulas was responsible for the rise (Maki et al. 1992). The suggestion has also been made (on the basis of screening of blood donors for antigliadin antibodies) that the incidence of celiac disease in Denmark may be similar to that of Sweden although largely undiagnosed (E. Grodzinsky, personal communication, cited by Ascher and Kristiansson 1994).

Ireland was reported to have a high incidence of celiac disease—about 4 per 1,000 births during an 11-year period prior to 1973, particularly in the west near Galway (Mylotte et al. 1973). Since that time, however, the incidence among children under the age of 12 has fallen by 62 percent (Stevens, Egan-Mitchell, et al. 1988), in contrast to the rise experienced in Sweden. The reasons for the decline in Ireland are unclear; they perhaps involve dietary changes such as an increase in the frequency of breast feeding or a decrease in the early introduction of grains to the diet of infants. Or some other undefined factor, perhaps a change in the type of viral infections prevalent in the population, may be involved.

The incidence in the United States has not been established, but a recent study of a Minnesota population (Talley et al. 1994) estimated the incidence as 0.01 per 1,000 person-years and the prevalence at 1 per 5,000. Because of the diversity of the U.S. population, which includes significant numbers of people having African, Chinese, or Southeast Asian ancestry, it might be expected that prevalence would be less than that of Europe as a whole, which is about 1 per 1,000. Whether the prevalence among U.S. residents of European extraction approximates that of European populations remains to be established, but it would be surprising to find that it is different.

Most studies of incidence and prevalence have focused on clearly recognizable disease resulting from significant damage to the absorptive epithelium of the intestine, often diagnosed by intestinal biopsy to show loss of villous structure and followed by removal of wheat and other harmful grains from the diet to demonstrate recovery of mucosal structure and function. It is becoming obvious, however, that there are many people who have subclinical celiac disease (Marsh 1992a), so that the true incidence of celiac disease may be at least two to three times greater than that represented by those with significant damage to the epithelium.

Origins of Agriculture and Wheat Farming

A diploid wheat and barley were likely to have been among the first crops cultivated by humans who were spearheading the Neolithic Revolution about 10,000 years ago. Wild diploid species of wheat, probably Triticum boeoticum and Triticum urartu, or possibly the wild tetraploid species Triticum dicoccoides, had most likely been harvested by hunter-gatherers for some considerable time. These species of Triticum were probably extant for perhaps 16 million years (Shewry et al. 1980) before domestication occurred (Harlan 1977) and thus those who pioneered in cultivating it probably had a reasonably good understanding of the plant’s life cycle. The origin of wheat cultivation is likely to have been somewhere in the Fertile Crescent area of the Middle East, perhaps near Jericho.

Frederick Simoons (1981) has discussed the possible effects of the spread of wheat farming (from likely centers of origin in the Middle East) on populations with members genetically susceptible to celiac disease. He used a genetic marker for celiac disease and compared the incidence of this marker in populations throughout the world with the incidence of celiac disease. The marker selected was a human leukocyte antigen, HLAB8 (a class I major histocompatibility complex [MHC] protein). HLA-B8 had been demonstrated to have approximately threefold greater occurrence in celiac patients (Falchuk, Rogentine, and Strober 1972; Stokes et al. 1972) than in normal individuals of the same populations, although it is now known that certain class II MHC proteins show even better correlations and are more likely to play a direct role in the processes of celiac disease (Kagnoff, Morzycka-Wroblewska, and Har-wood 1994). The data available to Simoons were, however, limited to HLA-B8.

Simoons reasoned that the greater rate of occurrence of celiac disease among people with the HLA-B8 antigen gave researchers a tool to use in their study of the illness. By determining the distribution of the antigen among human populations, researchers could predict which people were more at risk of developing the disease. Although there are exceptions, such that populations without HLA-B8 may yet have high levels of celiac disease (Logan 1992a), the assumption appears to have general validity.

It turned out that there were two geographic centers in Old World populations with noticeably higher frequencies of HLA-B8, one in western Europe, centered about the British Isles, and one in northwestern India, centered about the Punjab region. These regions also fell within the major areas of wheat cultivation, and in both there had been reports of relatively high incidences of celiac disease. They also represented, to a considerable degree, the boundaries of the spread of wheat farming throughout its geographical range up to relatively recent times (in this analysis, about 1000 B.C.) from its point of origin.

There appeared to be a gradient in the incidence of HLA-B8 throughout the range of wheat cultivation, such that low levels of the antigen were found at the likely origin of wheat cultivation and higher levels at the periphery. Simoons (1981) assumed that high levels of HLA-B8 were once typical of peoples at the origin and suggested that the gradient observed from the center to the periphery might well reflect the effects of a selective genetic disadvantage to members of the population carrying HLA-B8—along with other genes for susceptibility to celiac disease. Thus, the high levels of HLA-B8 in populations that took up wheat farming late (in the west of Ireland, for example) in its spread from the center of origin, along with high incidences of celiac disease, would reflect the lesser time available for the selective disadvantage to have diminished marker levels.

One obvious discrepancy was northwestern China, where wheat farming was introduced quite late through trade. This major wheat-growing area, although somewhat beyond the contiguous area of the earlier spread of wheat farming, might reasonably be considered its periphery.

There was, however, neither evidence of celiac disease nor high levels of HLA-B8 in the wheat-growing regions of China. A possible explanation for this was the likelihood that immune systems evolved differently in different populations to protect against relatively specific infectious diseases. Geographically distant populations would have been exposed to different stresses, thereby resulting in different complements of histocompatibility antigens. Furthermore, histocompatibility antigens that provided resistance to one disease might, quite coincidentally, enhance susceptibility to another, and this may well be true of genes for susceptibility to celiac disease (Strober 1992). Thus, the absence of celiac disease in the wheat-growing areas of China may well reflect the absence of the susceptibility genes in the Chinese population. Although hampered by a lack of adequate information, the analysis by Simoons (1981) is at least a highly interesting attempt to deal with what must have been selective pressures on populations containing the genes for susceptibility to celiac disease as wheat farming spread from the Middle East throughout Western Europe.

Definition of Harmful Cereal Grains and Components

Early conclusions regarding the toxicity of cereal grains and their constituents were based mainly on the ability of a grain to produce malabsorption of fats in celiac patients. Despite the lack of the more sophisticated approaches available today, early test results generally seem convincing. Currently, examination of mucosal biopsy specimens has become fairly common. When a flattened mucosa is found, with loss of villous structure, wheat and other harmful grains are removed from the diet to see if improvement follows. If it does, a subsequent challenge and biopsy to test for relapse may follow, although subsequent challenge has come to be reserved for special circumstances (Walker-Smith et al. 1990) because supporting antibody tests (tests for circulating antigliadin and antiendomysium antibodies) have lessened the likelihood of misdiagnosis (McMillan et al. 1991). In recent years, however, the situation has become more complicated through recognition that a flattened mucosa may actually be an extreme response and that there are many gluten-sensitive people who show less obvious evidence of the disease (Marsh 1992a). Circulating antigliadin antibodies may be indicative of celiac disease or at least a related gluten-sensitive condition in the absence of any evidence of mucosal changes, particularly when symptoms are present (O’Farrelly 1994).

Relatively few subjects were used in the testing of some grains or grain fractions in early work. This has been a continuing problem, resulting largely from the difficulties inherent in the requirement for human subjects in celiac disease research. It has become fairly clear that response to challenge may vary greatly from one patient to another and for a single patient over time. Furthermore, a considerably delayed response to challenge is not unusual, even when the challenge is with wheat, presumably the most toxic of the cereal grains (Egan-Mitchell, Fottrell, and McNicholl 1978; Walker-Smith, Kilby, and France 1978).

The variations in response known to occur might explain the opposing conclusions arrived at by various investigators regarding the harmful effects of oats. For example, Dicke, H. A. Weijers, and J. H. Van de Kamer (1953) asserted that oats are toxic, whereas W. Sheldon (1955) concluded that they are not. Part of the problem apparently arises from the relatively small proportion of avenins, the likely toxic fraction, in oat grain proteins (Peterson and Brinegar 1986): Avenins may make up only about 10 percent of the total protein in oats. Furthermore, when small numbers of subjects are involved, the average response of one group can be quite different from that of another. Because a negative response in one or a few patients might indicate only that the feeding time was too short, a clear positive response, as in the studies of oat protein toxicity by P. G. Baker and A. E. Read (1976), should perhaps be given more weight than a negative response, as in the study by A. S. Dissanayake, S. C. Truelove, and R. Whitehead (1974). The latter effort, however, included testing by intestinal biopsy—the more rigorous test—whereas the former did not.

A recent feeding trial using oats with 10 confirmed celiac patients produced no harmful effects as indicated by changes in mucosal architecture, endomysial antibody development, and infiltration of intraepithelial lymphocytes (Srinivasan et al. 1996). The relatively low percentage of avenins in oats may complicate this study in that a daily dosage of 50 grams (g) of oats was fed for 3 months. This would correspond to about 5 g of avenins per day. If 50 g of wheat had been fed instead, the patients would be eating at least 40 g of gliadins per day. Nevertheless, subsequent challenge of two of the patients with only 0.5 g of gluten per day produced evidence of intestinal damage. Thus, the toxicity of oats must be considered questionable on the basis of this latest study, as there are now two careful studies that showed no evidence of toxicity and only one study that was positive. It should also be emphasized that the supposedly positive study did not include biopsies.

In other cases, the test materials may not have been well defined. It is possible that the maize (corn) flour used in early testing was actually maize starch and relatively free of protein (Simoons 1981); the conclusion that maize proteins are not harmful may not have as rigorous a scientific base as is generally thought. There is, however, no obvious reason to question the apparent safety of maize.

A lack of sensitivity of the available testing methods, combined with inadequate test length, may have been responsible for the early conclusions that wheat starch is safe for celiac patients (Dicke et al. 1953). These conclusions, however, were challenged in later work (Ciclitira, Ellis, and Flagg 1984; Freedman et al. 1987a, 1987b; Skerritt and Hill 1990, 1991) as a consequence of the development of sensitive monoclonal antibody tests for gliadins (but also see Booth et al. 1991). Such tests have demonstrated the presence of small amounts of gliadins in wheat starch preparations, although the question of how harmful these small amounts may be to celiac patients is controversial (Ejderhamn, Veress, and Strandvik 1988; Hekkens and van Twist-de Graaf 1990; Campbell 1992).

It may be that rigorous scientific studies have proved toxicity in celiac disease only for wheat (and wheat proteins). To a considerable extent, conclusions regarding the toxicity of rye and barley—and, conversely, the lack of toxicity for rice and maize—are not based on an adequate amount of rigorous scientific testing. (Rigorous studies of oats are beginning to appear, and these support an absence of toxicity for this grain.)

Wheat Evolution

Wheat is a member of the grass family (Gramineae), as are the other grains—rye, barley, and oats—that are suspected of harming people with celiac disease. Of course, triticale, a cross between wheat and rye, is toxic, as would be expected for any other similar crosses that included genetic material from one or more of the toxic grains.

The grasses are a relatively recent evolutionary development. They are angiosperms, flowering plants, that developed somewhere between 100 and 200 million years ago (Cronquist 1968). Fossil evidence for grasses goes back only about 65 million years. They became widespread during the Oligocene epoch about 25 to 40 million years ago (Thomasson 1980), which, when it is considered that life has been evolving on earth for about 4 billion years, is a relatively short time. Cereal grains presumably evolved more recently within the time frame of grass evolution.

The ancestors of wheat, barley, and rye were diploid species with 7 chromosomes in the haploid state, or 14 chromosomes in vegetative cells. Barley first diverged from the common ancestral line, followed by rye. The line eventually gave rise to a series of diploid species that may be classed as Triticum species (Morris and Sears 1967). A natural cross occurred at some unknown time between two slightly diverged Triticum species, presumably Triticum urartu and Triticum speltoides, that gave rise to a new species through a process of polyploid formation in which chromosomes become doubled.

Without chromosome doubling, a cross between related species, such as those of Triticum, can take place, but the offspring is infertile because of a failure by the chromosomes to pair during meiosis. Polyploid formation occurs naturally, albeit rarely, perhaps through a process involving unreduced gametes (Harlan and de Wet 1975). This process can also be achieved in the laboratory through the use of chemicals (such as colchicine) that interfere with spindle formation to block chromosomal separation at anaphase. Either way, the result is a doubled set of chromosomes, a condition that allows for each original chromosome set to pair with its identical replicated set, thus avoiding the chromosome pairing problems that occur in crosses between species. The result of allopolyploid formation between T. urartu and T. speltoides was a new fertile species, called a tetraploid because it incorporated four sets of chromosomes in vegetative cells (designated AABB). Durum wheats (Triticum turgidum var. durum), used for pasta making, belong to one of the varietal groups of the AABB tetraploids.

Some time after wheat cultivation began, a tetraploid wheat crossed accidentally with a weed growing in the same field (Triticum tauschii var. strangulata), followed by chromosome doubling to give rise to a hexaploid species, which, because the genome of T. tauschii was designated D, had the composition AABBDD. Bread wheats (Triticum aestivumvar. aestivum) produce a more elastic dough than durum wheats and have properties that lend themselves to the retention of gases produced during yeast fermentation (leavening) of doughs. It was in the context of the evolution of wheat and closely related species that proteins arose with amino acid sequences capable of initiating damage in persons with celiac disease.

Gluten Protein Evolution

The gluten proteins of wheat (a monocotyledonous plant) and certain closely related species—including rye, barley, and, with qualifications, oats—are unique among plant storage proteins in having exceptionally high proportions of the amino acids glutamine and proline. It was suggested (Kasarda 1980, 1981) that because of their unique composition, the extensive occurrence of repeating sequences based largely on glutamine and proline, and their occurrence only in recently evolved grass species, wheat prolamins are a late evolutionary development. Subsequently, with the development of molecular biological techniques, gene sequencing provided evidence of short amino acid sequences in most wheat prolamins, homologous to sequences found in storage globulins of more distantly related dicotyledonous plant species (Kreis et al. 1985). This discovery pushes the possible age of the ancestral genes back to within the period during which the flowering plants have existed—perhaps 100 million years.

However, repetitive sequences make up major parts of all gluten proteins, whereas the homologous sequences just mentioned occurred only in nonrepetitive regions. These repetitive sequences, having major proportions of glutamine and proline, apparently do not have counterparts in proteins of species outside the grass family. Accordingly, it seems likely that at least the repetitive domains of gluten proteins, which have slightly differing repeat motifs according to type, are of more recent (<~65 million years) origin. All of the various repeating sequences include glutamine and proline residues, and although the repeats are often imperfect, comparison allows a consensus repeating sequence to be recognized. It is at least possible that the most active (in celiac disease) peptides may result from variations on the themes represented by the consensus sequences, as a consequence of these imperfections, rather than from the consensus sequences themselves, but that remains to be established.

There is reasonably strong evidence that the peptides with sequences found in the repeat domain of alpha-gliadin are capable of triggering the intestinal damage characteristic of celiac disease. Furthermore, significant sequence similarities between gluten proteins and other proteins are rare. The first important similarity found was between alpha-gliadin and the E1b protein (Kagnoff et al. 1984), produced in conjunction with infection by adenovirus 12, which infects the human gastrointestinal tract. Recently, similarities have been found for peptides produced in conjunction with infection by other types of aden-ovirus (Lähdeaho et al. 1993).

The recent evolution of repeating sequence domains in gluten proteins through extensive duplication of the DNA codons (for glutamine and proline, along with a few other amino acids) corresponding to the repeat motifs may be the basis for the lack of homologies or similarities with other proteins. Most proteins do not have large amounts of glutamine and proline. Hence, the sequences active in celiac disease are likely to be confined to the grass family.

Fractionation of Wheat


The wheat kernel is made up largely of the endosperm, the interior part of the grain, which constitutes about 85 percent of the kernel. In the milling process, the crushed kernels are fractionated by sieving. The crushed endosperm is the source of white flour, and the outer layers of the kernel yield the bran and germ fractions. Endosperm cells are largely made up of starch granules (about 75 percent) surrounded by a proteinaceous matrix. The proteins of this surrounding matrix are mainly storage proteins; upon germination of the seed, they are broken down to provide a source of nitrogen for use by the new plant. The storage proteins are to a large extent the same as the proteins of gluten, which contribute elasticity, yet at the same time extensibility, to flour–water doughs.

Gluten (Beccari 1745), prepared by washing away the starch granules from dough, constitutes the resulting cohesive, elastic mass of protein. It usually retains about 15 percent albumins and globulins, which are water- and salt-soluble proteins, respectively, according to the classification system of T. B. Osborne (1907). The remainder of the protein may be considered to constitute true gluten proteins, notable for their high percentages of two amino acids, glutamine and proline, which provided the basis for the name prolamins (Osborne 1907). Glutamine usually makes up 30 to 55 percent, and proline 15 to 30 percent, of the amino acids in gluten proteins. These two amino acids seem fairly certain to be included in the amino acid sequences of gluten proteins that are active in celiac disease.

The Meaning of “Gluten-Free”

Celiac patients, the physicians who treat them, and the various organizations that represent celiac patients usually indicate that a food safe for someone with celiac disease is “gluten-free.” This use of the term “gluten” is sometimes confusing. Traditionally, gluten, as just mentioned, is the cohesive, elastic protein obtained when a wheat flour dough is kneaded in excess water to wash away the starch granules (Beccari 1745). The resultant gluten ball can only be obtained readily from wheat-flour doughs, with difficulty from rye, and probably not at all from barley or oats. Cereal chemists would not ordinarily refer to “rye gluten” or “barley gluten,” let alone “oat gluten,” but would refer specifically to the equivalent storage proteins in these grains (for example, the proteins equivalent to the gliadin proteins of wheat are called hordeins in barley, secalins in rye, and avenins in oats). The term gluten has become corrupted in recent times, however, occasionally being used in industry to refer to the protein residues from other grains. For example, when maize is separated into starch and protein fractions, the protein fraction is often referred to as “corn gluten.”

What celiac patients wish to know about a food when they ask if it contains gluten is: Does it contain wheat, rye, barley, or oat proteins or any of the harmful peptides that are derived from the storage proteins of these grains during food processing or through the action of the digestive enzymes? Use of the term gluten-free is particularly awkward when applied to seeds of rather distantly related plants, such as the dicots amaranth and quinoa. These plants are not known to have proteins at all similar to gluten proteins. However, because quite small peptides of specific amino acid sequence may be active in celiac disease, it is not beyond possibility that equivalent sequences might occur in generally quite different proteins of distantly related species. Despite its ambiguities, the term gluten-free in relation to foods for celiac patients is already well established, and finding a satisfactory and short alternative may be difficult.

Digestion and Nutrient Absorption

Digestion of proteins commences in the stomach with partial denaturation (unfolding) of the large protein polypeptide chains by hydrochloric acid, which enhances their breakdown into smaller polypeptide chains by the proteolytic enzyme pepsin. In the highly acidic conditions of the stomach, pepsin is very active, cleaving a number of different peptide bonds. Breakdown of food proteins in the stomach is incomplete (Gray 1991), and relatively large peptides pass from the stomach into the duodenal part of the small intestine. Enzymes secreted by the pancreas—trypsin and carboxypeptidases—enter the interior (lumen) of the small intestine to continue digestion of proteins until they are broken down into amino acids or small peptides. Peptides this small are unlikely to be active in celiac disease.

Breakdown of proteins to amino acids or short peptides, both of which can be absorbed, is probably complete in the distal small bowel, even in celiac patients, as evidenced by a gradient of intestinal damage from the duodenum through the jejunum in active celiac disease. The ileum may remain free of tissue damage (Rubin et al. 1962). Instillation of wheat proteins into the ileum (Rubin et al. 1962) or the rectum (Dobbins and Rubin 1964; Loft, Marsh, and Crowe 1990) produces similar damage to that seen in the small intestine, indicating that the entire epithelial surface is susceptible to the damaging effects of wheat proteins or peptides. The final breakdown of food particles to relatively small molecules, and their active or passive absorption for use by the body, takes place at the membranes of cells lining the surface of the small intestine.

Almost all nutrients, ranging through amino acids, sugars, fatty acids, vitamins, and minerals are absorbed from the small intestine so that damage to this absorptive surface may have many manifestations in celiac disease. Obviously, a deficiency of calories from carbohydrates and fats and a deficiency of amino acids needed for protein synthesis can be responsible for a loss of weight or a failure to thrive, but the effects of malabsorption are often more diverse, ranging from osteoporosis in later life as a consequence of a failure to absorb calcium adequately to nerve degeneration as a consequence of a failure to absorb vitamins. The large intestine, or colon, extracts water and electrolytes from the food residues while bacterial action on these residues reduces the bulk of indigestible materials, such as cellulose. Relatively little absorption of nutrients occurs in the colon.

When intestinal biopsy was introduced in the 1950s, it was demonstrated that the surface of the intestinal mucosa in patients with active celiac disease appeared flattened as a consequence of the loss of villous structure and enhanced proliferation of immature cells in the crypts. Subsequently it was recognized that the microvilli of mature enterocytes were often damaged as well. Together, these losses significantly diminish the absorptive surface area and give rise to an increased crypt layer with an immature population of enterocytes having incompletely developed enzyme and transport activities. The net result is a diminished capability to absorb nutrients, although it should be noted that malabsorption may occur even in patients with little or no obvious damage to the epithelium.

It has been noted that initial responses to gluten challenge included infiltration by lymphocytes of the epithelium and lamina propria and thickening of the basement membrane and that these changes preceded major changes in mucosal structure (Shmerling and Shiner 1970; Marsh 1992a), providing strong support for the involvement of immunological processes in the destruction of the absorptive epithelium. A highly significant association of celiac disease with certain proteins of the major histocompatibility complex (histocompatibility antigens) also provides evidence for involvement of the immune system and a genetic predisposition to the disease. The inheritance of celiac disease appears to be complex, however, involving two or more genes (Strober 1992), and an environmental contribution, such as a viral infection or stress, may be necessary before the genetic predisposition comes into play (Kagnoff et al. 1984). What remains to be clarified is how these immune processes are triggered by gliadin peptides and how they ultimately result in the loss of epithelial absorptive cells.

Mechanisms of Tissue Damage

The way in which ingested gluten protein triggers events in the body that may ultimately bring about damage to the absorptive epithelium of the small intestine is unknown. The first important hypothesis put forward postulated that a key enzyme, a protease that could degrade proteins, was absent in celiac patients. As a consequence, certain harmful gliadin peptides that would, in the normal person, have been broken down and rendered harmless, continued to exist with a consequence of either direct toxicity on the absorptive epithelial cells or initiation of immune responses that secondarily damaged the epithelium (Cornell 1988). Although it is difficult to disprove this hypothesis beyond any reasonable doubt, no missing enzyme has ever been found despite considerable research effort to locate it. Furthermore, the likelihood that limited digestion of proteins in the stomach results in exposure of at least the proximal small intestine (duodenum) to fairly large peptides raises the question of why most people do not suffer some damage from such peptides.

The missing enzyme hypothesis was largely supplanted by the proposal that binding of gluten proteins or peptides to the enterocytes targeted them specifically for destruction and that enhanced proliferation of crypt cells, in response to destruction of the villous enterocytes, resulted in flattening of the mucosal surface, one of the major characteristics of advanced disease. Unfortunately, no evidence for the binding of gluten proteins or peptides to enterocytes in vivo has been found. The finding that enterocytes were capable of expressing MHC class II antigens that might be capable of presenting small gluten peptides of perhaps 10 to 20 amino acid residues to T cells raises the question of how effective the search for the binding of peptides to enterocytes has been, insofar as the methods used (often based on antibody probes produced in response to stimulation with intact proteins) were unlikely to recognize such small gluten peptides.

The strong correlation of celiac disease with particular class II histocompatibility antigens, which present peptides to T cells (lymphocytes), thereby activating them, has resulted in a currently favored hypothesis in which this presentation to T cells and binding of gliadin/gluten peptides to T cells is the initiating process that results in damage to the mucosal tissues, especially those underlying the epithelial cells (enterocytes). Binding of peptides by T cells may, however, activate a number of pathways, and beyond the activation step, the details of the hypothesis become rather vague.

Genes and Proteins of the Major Histocompatibility Complex

It is clear that there is an increased incidence of celiac disease in certain families, providing support for the hypothesis that there is a genetic predisposition to the disease. We continue to stress, however, that this predisposition may be insufficient for development of the disease without the intervention of some environmental factor—other than gluten proteins.

Celiac disease occurs in about 14 percent of siblings, 8 percent of parents, and 15 percent of children of celiac patients (Strober 1992). Although dizygotic (fraternal) twins do not show major differences from normal siblings in their tendency to develop celiac disease, monozygotic (identical) twins are about 70 percent concordant for the disease. The fact that 30 percent of the identical twins are discordant, with one having the disease and the other being free of it, is strong support for the necessary role of an environmental factor, such as viral infection (Kagnoff et al. 1984) or some other stressful event.

It was recognized in conjunction with attempts to transplant tissues from one individual to another, that certain protein antigens (human leukocyte antigens, or HLA molecules) had to be matched to avoid rejection of the foreign tissue. These proteins, now often called the proteins of the major histocompatibility complex (MHC proteins, MHC antigens), are coded by a cluster of genes on human chromosome 6. MHC proteins are cell-surface receptors, proteins that bind peptides at specific binding sites for presentation to a receptor site on T-cells. The T-cell receptor must interact simultaneously with both the MHC molecule and the peptide presented by the MHC molecule before the presentation is recognized as a signal for the T cell to carry out some other function. An activated T cell might go on, for example, to activate B lymphocytes to produce antibodies, or it must suppress the normal immune response to antigens encountered as part of our food intake (development of oral tolerance). The latter function may be especially important because, presumably, in most people the immune response to gluten peptides has been suppressed, whereas in celiac patients, certain gluten peptides trigger an immune response capable of damaging the small intestine.

The MHC proteins of concern to us here are divided into class I and class II on the basis of a number of distinguishing characteristics, one of which is the type of cell on which they are expressed. Class I antigens appear on most cells, whereas class II antigens are found mainly on cells of the immune system, although class II antigens are expressed on the surfaces of enterocytes as well. The first demonstrations of associations between MHC proteins and celiac disease indicated that a class I antigen, designated HLA-B8, was found in nearly 90 percent of celiac patients but in only about 20 percent of controls (Falchuk, Rogentine, and Strober 1972). This antigen, HLA-B8, was the genetic marker that, as was discussed earlier by Simoons, was thought to determine the incidence of celiac disease throughout the world.

However, subsequently it became clear that associations of celiac disease with class II antigens were even stronger than those of the class I antigen HLAB8. Particularly strong associations have been found for the HLA class II alleles DR3 and DQ2. The association with HLA-B8 apparently resulted from linkage disequilibrium in which certain closely linked genes tend to remain associated during genetic recombinations to a greater extent than would be expected.

The associations of celiac disease with various MHC proteins observed throughout the world are quite complex (see reviews of Kagnoff 1992; Tighe and Ciclitira 1993), but they are sufficiently strong to make it seem likely that class II antigens are directly involved in the mechanisms responsible for tissue damage in celiac disease—possibly by presenting peptides derived from gluten proteins to T cells.

In support of the possibility that gliadin peptide binding by T cells is involved in celiac disease, K. E. A. Lundin and colleagues (1993) have found that T-cell lines derived from intestinal biopsies recognized gliadin peptides, and the cell lines were mainly stimulated by antigen presentation in the context of the DQ heterodimer. In addition, H. A. Gjertson, Lundin, and colleagues (1994) found that a peripheral blood lymphocyte clone from a celiac patient specifically recognized a synthetic peptide corresponding to alpha-gliadin residues 31-47. Further interest in the role of class II MHC proteins derives from studies showing that HLA-DR molecules, which are expressed by enterocytes of normal individuals or of celiac patients on a wheat-free diet (Arnaud-Battandier et al. 1986; Ciclitira et al. 1986), became differentially expressed in the crypt cells of celiac patients (but not in those of normal controls). This occurred when gluten proteins or peptides were added to biopsied tissues in culture that had been obtained from treated patients on a wheat-free diet (Fais et al. 1992).

W. Strober (1992) has pointed out that because discordance for celiac disease is much less common in identical twins than in siblings who have apparently identical complements of histocompatibility antigens (MHC proteins), it seems likely that the difference cannot be explained by an environmental factor because the contribution of environmental effects should have been about the same for the sets of twins and siblings. Strober considers this as strong evidence for a contribution by some non-MHC-defined disease gene. Whether or not this gene turns out to lie outside the MHC complex, celiac disease is likely to have a two-locus basis. D. A. Greenberg, S. E. Hodge, and J. I. Rotter (1982) considered that the available data fit best with a recessive-recessive two-gene model, but studies in the west of Ireland (Hernandez et al. 1991) favored dominance for the gene associated with the HLA locus. Conclusions may be somewhat dependent on the population studied, as many diseases show variation in their genetic basis.

Associated Diseases

Many different diseases have been reported to occur concurrently with celiac disease, including dermatitis herpetiformis, insulin-dependent diabetes, Down syndrome, IgA nephropathy, and epilepsy associated with cerebral calcifications (Bayless, Yardley, and Hendrix 1974; Gobbi et al. 1992; Collin and Maki 1994). In only a few diseases are statistically significant data available for the establishment of an association with celiac disease.

Dermatitis herpetiformis is a disease manifested by a rash with small blisters and IgA deposits even in uninvolved skin (Fry 1992). It is quite strongly associated with celiac disease. Intestinal lesions similar to those encountered in celiac disease are found upon biopsy of about two-thirds of patients with dermatitis herpetiformis, and a gluten-free diet usually improves both the skin rash and the intestinal lesions. Furthermore, even those patients without significant damage to the intestinal epithelium usually have an increased number of intraepithelial lymphocytes, and patients without obvious mucosal damage have developed such damage when their intake of gluten was increased. Nevertheless, malabsorption is much less common in dermatitis herpetiformis than in celiac disease (Kumar 1994). Indeed, the incidence is rather less than that of celiac disease, with dermatitis herpetiformis occurring about half as frequently as celiac disease in Sweden and perhaps one-fifth as frequently in Scotland (Edinburgh).

Insulin-dependent diabetes mellitus also shows a definite association with celiac disease. Between 5 and 10 percent of children with celiac disease have diabetes mellitus, whereas about 1 to 3 percent of children with diabetes mellitus have celiac disease (Visakorpi 1969; Strober 1992). Both celiac disease and insulin-dependent diabetes mellitus share an association with the histocompatibility antigen HLADR3 (Maki et al. 1984), and the common MHC gene(s) may predispose individuals carrying them to both conditions.

Down syndrome may be weakly associated with celiac disease. J. Dias and J.Walker-Smith (1988) found an increased incidence of Down syndrome in celiac patients compared with the incidence in the general population. They considered this as supporting evidence for earlier findings (Nowak, Ghisham, and Schulze-Delrieu 1983). M. Castro and colleagues (1993) have confirmed a significant increase in celiac disease among Down syndrome patients and found that antigliadin antibodies provide a useful screening tool to look for celiac disease among these patients. In addition, there are reports that patients with schizophrenia, autism, and IgA nephropathy have apparently benefited by removing wheat (and related harmful grains) from their diets, although such reports are not universally accepted as valid.

There have been extensive attempts to show a correlation between schizophrenia and celiac disease (Dohan 1966, 1988; Lorenz 1990), but the results have not been convincing. Evidence for intestinal damage in the group of schizophrenics studied by F. M. Stevens and colleagues (1977) was no greater than that for controls, and there was no significant increase in serum antireticulin antibodies. Both intestinal damage and antireticulin antibodies are usually present in celiac patients taking a normal (wheat-containing) diet. Nevertheless, published reports from physicians indicating that removal of wheat from the diet can produce a marked reduction in psychotic symptoms for some schizophrenics were considered sufficiently convincing by K. Lorenz (1990) to indicate that in at least a subset of schizophrenics, wheat has an adverse effect on the disease. Even in studies where no positive correlation was found, investigators noted that some patients did apparently show considerable improvement on a wheat-free (“gluten-free”) diet (Rice, Ham, and Gore 1978).

Some celiac patients on a gluten-free diet, perhaps about 25 percent (Ansaldi et al. 1988), have indicated that they feel temporary psychological disturbance upon eating wheat, including symptoms such as irritability, hostility, depression, or a general feeling of mental unease. Although such symptoms may be “normal” in the face of physiological changes produced by eating wheat, investigation of a connection between such responses in celiac patients and the effects of wheat in the diet of carefully selected schizophrenics might be worthwhile. Again, even if wheat in the diet does adversely affect the course of schizophrenia, the mechanisms involved may be quite different from those involved in producing intestinal damage in celiac disease.

The role of wheat in autism is also controversial, but development of the hypothesis that wheat and casein exacerbate the symptoms of autistic patients parallels that for schizophrenia. The hypothesis is based so far on studies that indicate value for a wheat-free diet in improving behavior of patients with autism (Reichelt et al. 1991), although these studies are not well controlled because of the difficulties in carrying them out with patients who may not be able to supply informed consent. It has been discovered that amino acid sequences found in the primary structures of wheat gliadin proteins and casein proteins are similar to those of endorphins and other neuroactive peptides termed exorphins (Zioudrou, Streaty, and Klee 1979). It has been proposed that these, along with other neuroactive peptides from wheat, are responsible for the exacerbation of symptoms claimed for schizophrenic patients on a normal, wheat-containing diet (Zioudrou et al. 1979; Dohan 1988). This working hypothesis is also favored by some researchers investigating the possibility that wheat and casein proteins in the diet exacerbate symptoms in autistic patients (Reichelt et al. 1991). According to the hypothesis, food-protein-derived neuroactive peptides pass through the wall of the intestine and also pass the blood-brain barrier to affect brain function. This results in a variety of abnormal behaviors.

Abnormal peptide patterns appear in the urine of autistic subjects and schizophrenics (Cade et al. 1990; Shattock et al. 1990; Reichelt et al. 1991), and the hypothesis has been put forward that these abnormal patterns reflect to some extent the abnormal absorption and then excretion of exorphin peptides derived from wheat or milk proteins in the diet. This implies either excessive passage through the intestinal wall and/or a failure in some other way to process these peptides into harmless forms. However, that these abnormal urine peptide patterns truly represent excretion of peptides of dietary origin does not seem to have been proved.

IgA nephropathy is a kidney disease characterized by protein and blood in the urine and IgA deposits in the kidney (Stevens, Lavelle, et al. 1988). Although the severity is variable, the disease can lead to chronic renal failure. Reports of concurrence with celiac disease, the presence of circulating IgA antibodies to gluten proteins, and patient improvement (even with nonceliac patients) on a gluten-free diet have led to speculation that there may be a connection between the two diseases (Coppo et al. 1986; Sategna-Guidetti et al. 1992). It has been suggested that IgA nephropathy might be similar to dermatitis herpetiformis, with the main difference being that IgA deposits form in the kidney rather than in the skin. However, the evidence indicates that epithelial damage is uncommon in IgA nephropathy, although the activities of marker enzymes in the brush border were significantly lower (Stevens, Lavelle, 1988). Gliadin proteins or peptides have been found in complex with the IgA deposits in IgA nephropathy (Russel et al. 1986), and similar associations should be sought in dermatitis herpetiformis.

Celiac patients have been found to be at increased risk of developing certain types of cancer (Holmes et al. 1989; Logan et al. 1989; Holmes and Thompson 1992), especially small intestinal lymphoma. Nonetheless, the absolute risk of a celiac patient dying from this cancer is small because the incidence in the normal population is quite low.

G. K. T. Holmes and colleagues (1989) found in a study of 210 patients that those who had been on a strict gluten-free diet for more than 5 years did not have significantly increased risk of developing lymphoma. But the risk of developing lymphoma was reported to be high (1 in 10) for celiac patients who were diagnosed late in life (over the age of 50). The investigators did not discuss what constituted a “strict gluten-free diet,” and J.A. Campbell (1992) has pointed out that it would not have been unusual for patients who thought themselves to be on a strict gluten-free diet to be using products containing wheat starch, which has a small amount of gluten in it. It seems prudent for celiac patients to follow a strict gluten-free diet as recommended by Holmes and colleagues (1989), but whether traces of gluten in the diet, such as might result from use of wheat-starch products, contribute to the development of malignancies late in life does not appear to have been established.


Human evolution has produced a complex immune system designed to protect us from harmful parasites, bacteria, viruses, and foreign substances. Human cultural development has led to a dependence on agriculture, and in a large part of the world, a heavy dependence on wheat and related grain crops as part of that agriculture. The proteins of these grain crops have evolved in a somewhat isolated manner (in part because they are recently evolved) such that particular amino acid sequences have appeared in these proteins that can induce the immune system in a susceptible subset of people into damaging the absorptive layer of cells lining the small intestine. The mechanisms are not very well understood, but they seem to involve a failure of the suppression mechanisms by which oral tolerance to food-protein antigens is developed in normal individuals.