Portal de Periódicos da CAPES

Serologic Indicators of Celiac Disease

1998; Lippincott Williams & Wilkins; Volume: 26; Issue: 2 Linguagem: Inglês

10.1097/00005176-199802000-00016

1536-4801

Thomas M. Rossi, A Tjota,

Microscopic Colitis

Few controversies in the discipline of pediatric gastroenterology are viewed with less emotion than the selection of the best serologic screening test for celiac disease. The current criteria for diagnosis of gluten-sensitive enteropathy, or celiac disease (CD), were set forth in 1990 by the European Society of Paediatric Gastroenterology and Nutrition (ESPGAN)(1). The requirements include the characteristic histologic appearance of the mucosa at presentation with resolution of symptoms after gluten elimination. The appearance of either antigliadin (AGA), antireticulin (ARA), or antiendomysial (EMA) antibodies at first examination and their disappearance after gluten withdrawal lends support to the diagnosis. The antibodies, although serving as indicators of active disease, are not yet universally accepted as pathognomonic for the condition. In fact, there is no telling whether we will ever find a true marker for the disease. Although the search for a serologic marker is carried out by some investigators as though it is the search for the "holy grail," we should not be disillusioned by our inability to find it. Few conditions in medicine have specific markers that satisfy the criteria of universal acceptance as a pathognomonic marker. Examples of such markers are the antibodies to the carcinoembryonic antigen, rheumatoid factor, prostate-specific antigen, neutrophil cytoplasmic antigen, and theHelicobacter antigen. All require clinical judgment in the interpretation of the test results. Regarding the celiac markers, differences in techniques for detection of these antibodies as well as variables in`patients have been responsible for the disparate results in studies investigating their effectiveness in diagnosis. Despite these issues, the serologic tests for screening may aid in diagnosis if their limitations are realized and their presence or absence is interpreted in light of the particular clinical situation. We are approaching a better understanding of their roles and interrelations. The purpose of this review is to discuss the efficacy of the existing diagnostic tests and to comment on newly discovered autologous tissue antigens that may be involved in the pathomechanism of the disease and may further augment diagnostic ability. SEROLOGIC MARKERS FOR CELIAC DISEASE Antigliadin Antibodies Circulating AGA antibodies, the earliest discovered serologic marker for CD, represent antibodies to the cereal protein, which presumably is absorbed intact across the intestinal mucosa. AGA antibodies have been extensively investigated since initial descriptions appeared in the late 1950s and the early 1960s (2,3). Techniques for detection have evolved through the years and have varied from "precipitating antibodies" to cereal proteins (2), microimmunodiffusion (3,4), radioimmunoassay(5), binding serum antibodies to wheat grains, and detection by fluorescent horse antihuman immunoglobulin (Ig) G(6). The difference in detection techniques caused problems with standardization and therefore in reproducibility between studies. The first method to improve detection, the enzyme-linked immunosorbent assay (ELISA), appeared in 1977(7,8). However, it is understandable that even techniques similar to ELISA may yield different results, because gliadins are a complex mixture of proteins that contain at least 40 components for a single variety of wheat. Several investigators have attempted to improve the sensitivity of the detection method by using gliadin fractions or peptides as antigens for ELISA (9-11). However, discordant results have been reported, even with the use of similar methods(9,10,12,13). Notwithstanding difficulties in methods, controversy exists regarding the value of the specific class of AGA antibodies in the diagnosis of celiac disease. Some investigators advocate IgG class AGA(14-17); whereas others favor IgA class AGA (18,19), and still others favored both(16,17,20). The usefulness of gliadin antibodies for diagnosis is open for criticism because their sensitivity and specificity varies so much from study to study(21-26). Figures for specificity range between 65% and 100% for IgA antibodies and 50% to 100% for IgG antibodies. Sensitivity reports for IgA antibodies range between 52% and 99.9%, and for IgG between 82% and 100%. Unfortunately, IgG antibodies can be found in normal control subjects as well as in those with other diseases, including Crohn's disease, liver disease, and other gastrointestinal disorders (10,12,27). In contrast, IgA antibodies, although more specific for celiac disease than are IgG antibodies, are not found in all patients with celiac disease. In addition, IgG increases with age in normal control subjects making it unsuitable for diagnosis in older age groups(10,12,28). Antireticulin Antibodies ARA antibodies have been investigated since initially reported in the 1970s(29). They were first described as reacting with connective tissue fibers around hepatic sinusoids and blood vessels as well as with perilobular, periglomerular, and occasionally glomerular staining of the kidney, and also with fine staining of the stroma between gastric glands. They react with connective tissue of rat and human organs and appear to be directed against reticulin fibers in these tissues. The antibody is best detected by indirect immunofluorescence, using rat liver and kidney as the substrate. The ARA do not react with type III collagen, noncollagenous reticulin components, or fibronectin; and they seem to be specific for other unidentified connective tissue components (30-32). Several staining patterns by indirect immunofluorescence have been found to occur. An R-1 pattern has been exhibited by celiac and dermatitis herpetiformis patients. This pattern is characterized by staining of peritubular and periglomerular fibers in the kidney and fluorescence in portal tracts of rat liver (33). In contrast with other ARA subtypes, it also reacts with human tissues, although some investigators disagree about expression in humans. They insist that the reticulin antigen is specifically expressed in rodent but not in primate tissues (34). ARA antibodies can be of the IgG or IgA class, but IgM-ARA antibodies do not occur. Immunoglobulin G antibodies usually occur in conjunction with IgA-ARA. The specificity of IgG-ARA for gluten-sensitive enteropathy, is controversial(35). The IgA-ARA seem to be disease specific and sensitive indicators of gluten-sensitive enteropathy. The range of specificity reportedly lies between 59% and 100% and sensitivity between 30% and 95%, indicating sensitivity and specificity somewhat similar to that of IgA-ARA, however, some may argue that the R-1 ARA are probably more specific for celiac disease than the are the gliadin antibodies(24,25,35-38). Still, the R-1 ARA have been reported in patients with Crohn's disease and occasionally in those with other conditions (39). The significance of the R-1 ARA is uncertain; however, the association between it and untreated celiac disease is well established, and it disappears from the circulation after a strict gluten-free diet is imposed(40). Sensitivity and specificity reports are subject to the differences in populations studied. For example, in some studies, specific populations were examined, such as that of India, where patients with inflammatory bowel disease are not usually found. In one such study, the authors found that R-1 ARA had 100% specificity with 85% sensitivity(38). Antiendomysial Antibodies A third group of circulating tissue antibodies, the EMA, are gradually gaining acceptance as sensitive and specific markers in celiac disease. These are primarily IgA antibodies directed against the intermyofibrillar substance of the smooth muscle, which may correspond to a reticulin-like structure or to a surface component of smooth muscle fibrils (41). This antigen may also be expressed around lamina propria structures surrounding intestinal crypts, muscularis mucosa, and smooth muscle fibers, to which"human jejunal antibodies" are directed(42). Unlike R-1 ARA, which are reported to react with human and various rodent tissues, EMA are species-specific, reacting only with the endomysium in the gastrointestinal tract of primates. They are detected by indirect immunofluorescence, using monkey esophagus tissue sections. The sensitivity and specificity of the EMA approaches but does not reach 100%. Some false-positive identifications have been reported. One patient with allergy to cow's milk protein and one with Giardia lamblia have been reported (26,41,43-45). In addition, we discovered a celiac patient who was IgA- and IgG-EMA-negative at initial examination and during gluten challenge. Therefore, we believe that false-negative readings are less common than false-positive ones, but they exist and will continue to be reported (46,47). Absence of EMA may be more frequent in celiac patients younger than 2 years than in older patients (45). EVALUATING THE SCREENING TESTS In recently reported findings, three children with positive and four with weak positive results did not have celiac disease (48). Sensitivity was 100%, specificity 97%. In findings in another study that compared EMA, AGA, and ARA concentrations in an Israeli group of celiac patients, specificity was 98% and sensitivity was 97%(26). In these results, EMA appeared to be the most reliable serologic marker for the diagnosis of celiac disease. The positive predictive value of EMA and ARA were comparable (97% and 100%, respectively); however, EMA had the highest negative predictive value (98%). In addition, EMA were found to be more diet-sensitive. Three months after gluten withdrawal, more children were negative for EMA than for AGA or ARA. High EMA titers appear more quickly in response to gluten challenge than do AGA or ARA titers(49). No serologic test at present, therefore, enjoys 100% sensitivity and 100% specificity. However, of all the markers, we believe EMA is presently the best for serologic testing, based on our interpretation of available data(50-54). The occurrence of false-positive and false-negative results with these markers requires comment. Deficiencies exist in the uniformity of interpretation of the methods of detection of the antibodies when using indirect immunofluorescence. Reports of false-negative results with indirect immunofluorescence raises the question of whether the technique and interpretation of the test system are adequate. What titer should be considered a positive result? Should weakly stained sections at low serum dilution be considered positive? Whether the patient is IgA-deficient must also be considered when IgA-AGA, IgA-ARA, or IgA-EMA are sought. Indeed, some investigators advocate that IgA-AGA do not offer an advantage over IgG-AGA and that celiac disease in patients with isolated selective IgA deficiency would be better detected by screening for both IgG-AGA and IgG-ARA, because IgG-AGA are less specific and are commonly found in patients with no celiac disease who have selective IgA deficiency(55-58). False-positive results obtained in testing for EMA or for AGA or ARA raise the question of whether these patients may be have latent disease. The antibodies are detected in these patients, but their presence remains an enigma. Perhaps we should not consider this to be a false-positive result. Only follow-up and further studies will determine whether celiac disease develops in these patients. The question of "false"-positive results of serologic testing has recently been investigated(59). In 7 of 25 patients exhibiting ARA or AGA and displaying normal small bowel mucosal villous architecture, villus atrophy subsequently developed after a follow-up period of 1 to 7 years(60,61). In patients who are"false"-positive for EMA, the histopathologic features of celiac disease may also appear in the future. Results of future studies will elucidate this issue. Regarding false-negative cases, however, there are very few published reports of patients with celiac disease who do not possess EMA(62). Because it is becoming generally accepted that subclinical celiac disease is common in the general population, the use of screening tests in clinical studies is becoming increasingly important. Many patients who are free of major symptoms exhibit the typical histologic features of celiac disease. The disease is discovered in these patients during studies investigating the incidence of celiac disease and in studies using serologic screening for markers in such high-risk populations as family members of patients with the disease, patients with diabetes, those of short stature, and those with Down's syndrome, or in the general population(55,63-68). Considering those with silent celiac disease, the true incidence of the disease in the United States and Europe may be higher than reported. Using EMA, we found that the rate in family members was 8%, in those with diabetes was 4%, and in those of short stature 1.7%. These figures are similar to those in reported results of European studies. However, the incidence in patients with major symptoms was only 1.29 per 10,000 live births, much lower than the rate of 1:1000 reported in European centers(63,69-71). We await with great interest the results of additional studies investigating the incidence of silent celiac disease in both populations(72-74). Certainly, the incidence of those with major symptoms in the United States is much less than it is in Europe. THE PATHOGENESIS OF CELIAC DISEASE Efforts to discover the best screening test for celiac disease and efforts to diagnose disease in every last patient affected have vastly overshadowed and outnumbered studies of the pathophysiology of the disease. Although gliadin antibodies were reported approximately 30 years ago, work on characterizing the toxic component(s), both in vivo and in vitro, has not been completed. Many hypotheses have been proposed to explain the pathomechanism of the observed gliadin-induced enteropathy. Studies have been limited to investigations of lymphocyte populations and cytokine production or of toxic peptides acting as lectins that induce cell death(75-78). Results of these studies seem promising and serve to direct future studies toward investigating immunologic reactions associated with tissue injury, but definitive statements regarding pathophysiology cannot yet be made. Furthermore, the mechanism of generation of the AGA antibodies is uncertain. There are even fewer reports of investigations of the reticulin and endomysium antibodies. The molecular mechanism behind the pathogenesis of CD is still unsettled, but the immunologic aspects of the disease have attracted great attention, both in the proposal of pathogenetic mechanism and in the efforts to established a serodiagnostic test for CD. Operating within the frustration of this situation, it is recognized that the sera of patients with CD contain antibodies that bind to a variety of tissues, including human umbilicus and fetal lung tissue. Using indirect immunofluorescence, antibodies to umbilicus and fetal lung have been detected in the sera of celiac patients(79-85). In addition, Maki et al. (86-88) found six polypeptides with isoelectric points ranging from PI 7.49 to 7.75, and molecular weights from 18.5 to 37 kDa that were found by inhibition assay to bind to ARA and EMA. Fibroblasts in culture were determined to secrete four similar single polypeptides with molecular weights of 17 to 39.5 kDa, which bind to ARA and EMA. These low molecular weight proteins can be considered secretory antigens. Picarelli et al. (89) found EMA in the culture supernatants of biopsy samples from 16 untreated celiac patients regardless of the presence of gliadin in the media. In a similar fashion, we searched for a common celiac antigen that would serve as an inexpensive source for studying the incidence and pathomechanism of the condition. Human placenta tissue sections express an antigen that binds to the IgA contained in the sera of patients with celiac disease (90). Findings in microscopic evaluation of placenta tissue sections indicate that the antigens were expressed on the cell membrane and cytoplasm of the trophoblast cells. Placental cells grown in primary culture also express the antigen. However, the antigen is not secreted into the culture media. Proteins were extracted from lysed placental cultured cells and were identified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and western blot analysis. Nine protein molecules were bound to the IgA of CD patients. The molecular weights ranged from 17 to 110 kDa. These proteins exhibit different target inhibition activities. A 110-kDa protein inhibited binding of IgA in CD patients to rat kidney (reticulin-like antigen). In contrast, A 65-kDa protein inhibited the binding of IgA to the monkey esophagus (endomysium-like antigen). Both protein molecules had 60% cross-inhibition activity. The 110-kDa protein is a heat-labile protein that breaks down to a 55-kDa molecule without losing its activity. These protein molecules (MW, 110 kDa, 65 kDa, and 55 kDa) have been designated as embryonic celiac antigens (ECA) and show isoelectric points between PI 5.1 and 5.9 (90). Six low molecular weight proteins (MW, 17 kDa to 45 kDa) inhibited binding of celiac patient sera to gliadin (gliadin-like antigen). In IgA-deficient celiac disease patients, an IgG antibody was found to bind to the antigens. In contrast to AGA, ECA do not bind to the IgA or IgG contained in the sera of normal patients, those with inflammatory bowel disease, or those with other immunologic conditions. These findings indicate that patients with celiac disease possess a variety of antibodies directed at a variety of normal tissue antigens, even fetal antigens. The disease may be genetically encoded from birth, and gliadin may alter the signal transduction pathway. The environmental factor, in particular the toxic antigen that is contained in gliadin, is the major substance involved in turning on the signal transduction pathway, so that the immune system recognizes normal tissue antigens as foreign. For years, the antibodies in CD have been viewed as innocent bystanders that are markers of the disease but are not involved in the pathogenesis. Recently, there has been a shift in this hypothesis to implicate them as having an active role in pathogenesis of the villus atrophy. The possibility of "molecular mimicry"-that is, structural similarity of external antigens and self-component autoantigens-is receiving attention(91,92). Indeed, 28-, 62-, and 66-kDa proteins reactive with celiac disease antibodies have been isolated from rat enterocyte surface membrane (89). Also, a tissue transglutaminase protein of 85 kDa has been found to react with EMA(93). In our opinion, the fact that AGA antibodies are present in many conditions, even in normal subjects, suggests that they are probably not involved directly in pathogenesis. ARA antibodies, and EMA and probably the ECA, are confined, for the most part, to the active celiac population and may have a role in precipitating villus atrophy. Because EMA may be produced in the intestine (54) and rise early in response to gliadin challenge and therefore in the earliest phases of CD, they may have a more important role in the pathogenesis of CD. Gliadin may itself, or by acting through AGA, alter connective tissue antigens so that they are recognized as foreign and may turn on production of ARA, EMA, and ECA. As suggested by Picarelli et al., gliadin and the EMA antigens (ECA) may have a T-cell epitope with features of molecular mimicry. Thus, gliadin may make available for immunologic recognition self-antigens, normally hidden from the immune system, thus allowing recognition of cryptic epitopes (89). These antibodies, particularly EMA, may attack components of the extra-cellular matrix of the mucosa, including collagen, reticulin, and muscle. They may unmask cryptic reticulin endomysium epitopes. By their adherence, the antibodies may induce or disrupt the anchoring mechanism of fibroblasts and villus structural elements. Gliadin may also be involved as the stimulus for cell-mediated immunologic events, including proliferation of mononuclear cells and production of such cytokines as tumor necrosis factor, thereby prepetuating the immune response and tissue injury (94). It is hoped that through further investigation of ECA and other tissue antibodies we will better understand the pathophysiology of celiac disease. Diagnostic ability may also improve if preliminary results using a newly developed ELISA are confirmed. Results of preliminary experiments demonstrate that the sensitivity is higher than the indirect immunofluorescence EMA test. The specificity and the cross-reactivity, however, remain to be determined. Perhaps they will bring us closer to our goal in our never-ending search for the "holy grail."