Antibodies to Human Tissue Transglutaminase for the Diagnosis of Celiac Disease

Over the last few years, it has become evident that celiac disease (CD) occurs more frequently than previously thought (1 in 1,000–1,500 people). Screening studies in healthy blood donors as well as in the general population support a much higher prevalence. It is now commonly estimated that prevalence of CD may be as high as 1 in 200 people. A wide heterogeneous array of clinical presentations has long been noted, and many patients remain symptomless for a long time (1). Because early diagnosis decreases the risk of complications, laboratory tests as accurate as possible are urgently needed to select patients in whom to perform an intestinal biopsy—so far the “gold standard” diagnostic test—or, should they prove so reliable, perhaps enabling the elimination biopsies in the future. Immunoglobulin A antiendomysial antibodies (EMA) generally have been considered the best immunologic markers of CD, with almost 100% sensitivity and specificity. This test, however, has some important flaws: there is interobserver variability in the appraisal of immunofluorescence, it may prove cumbersome for mass screening studies, and there are some difficulties in standardization of results among different laboratories. Although conservationists have objected to the use of esophageal cuts from monkeys, nowadays umbilical cord cuts can be used instead. Dieterich et al. (2) recently reported that tissue transglutaminase (tTG) is the main, if not the sole, autoantigen for EMAs. After this breakthrough, a logical step forward was the development of an enzyme-linked immunosorbent assay (ELISA) technique to determine serum anti-tTG antibodies (tTG-ab) (3–5). We previously investigated, with a commercially available ELISA kit that uses a guinea pig tTG as substrate, the value of these antibodies for the diagnosis of CD, and we found that immunoglobulin A antibodies have 100% sensitivity and 94% specificity with a very good correlation with traditional EMA (6). The nonabsolute concordance between EMA and tTG-ab may have a number of explanations (5). Recent work supports the assertion that the use of human, instead of guinea pig, tTG may improve further the reliability of the test (7). In our previous study (6), we suggested that the use of human recombinant tTG as antigen may avoid many of the shortcomings of the technique and could even enable to detect every case of untreated CD. We have now studied the value of the determination of antibodies to tTG for the diagnosis of active CD with a commercially available ELISA using human recombinant tTG as substrate. PATIENTS AND METHODS We studied 70 serum samples from patients seen at the Paediatric Gastroenterology Clinic. Forty-two samples belonged to patients with symptoms of CD at presentation. They were following a gliadin-containing diet, and all had subtotal atrophy of jejunal mucosa. Mean age was 4.9 ± 4.2 years. We also studied 28 serum samples from patients with a number of disturbances other than CD (chronic diarrhea, growth retardation, persistent and isolated antigliadin antibodies increase, etc.) and normal jejunal biopsy results. Their mean age was 6.1 ± 5.2 years. Antiendomysial antibodies were measured by indirect immunofluorescence, using the distal esophageal portion of the African green monkey (Cercopithecus aethiops) as substrate, according to the method described by Chorzelski et al. (8) (Byosistems, Barcelona, Spain). Results were considered positive when a reticular pattern of immunofluorescence was observed in the muscularis mucosa at a dilution of serum of 1:5 or more. The performance of this test in our lab is as follows: sensitivity, 97%; specificity, 97%; positive predictive value, 98%; and negative predictive value, 93%. Antibodies to tissue transglutaminase were assessed by a commercially available ELISA (CelikeyT; Pharmacia & Upjohn, Freiburg, Germany) as described by Dieterich et al. in 1997 (3), which uses human recombinant tTG as substrate. Values of tTG-ab of 9 U/mL or more were considered positive. Values between 5 and 8 were considered doubtful. RESULTS All 42 serum samples obtained from patients with CD at diagnosis had positive EMA, and 40 also had positive tTG-ab (concordance, 95.2%). One sample had a doubtful result (6.5 U/mL), and another clearly was negative. Antiendomysial antibodies and tTG-ab were negative in all 28 serum samples from non-CD control participants with normal intestinal biopsy results (concordance, 100%). The median serum tTG-ab in CD patients was 176.2 U/mL (range, 2–1000 U/mL). All values in the control group were less than 4 U/mL (mean ± standard deviation): 2.13 ± 0.5 U/mL (range, 2–3.7 U/mL). Considering the doubtful result as negative, the sensitivity of the test was 95%, the specificity was 100%, the positive predictive value was 100%, and the negative predictive value was 93%. DISCUSSION The diagnosis of CD classically has relied on the pathologic confirmation of a gluten-sensitive enteropathy. However, it is now recognized that subtotal atrophy of the intestinal villi, the classic hallmark of the disease, is not a universal finding, because some patients have subtler lesions that may cause some difficulties in diagnosis. Indisputable evidence that the lesion is gluten mediated would require obtaining a biopsy specimen before the treatment, another after withdrawal of gluten from the diet, and a third after gluten challenge. However, in clinical practice, obtaining serial biopsies is only reasonable in doubtful cases or in very young children (9). Characteristic changes in a first biopsy, together with a good clinical response to gluten withdrawal from the diet, is considered enough for the diagnosis when serologic markers are consistent with CD. After the development of a highly sensitive specific laboratory test, the EMA, serologic diagnosis of CD has come of age, and the feasibility of screening large series of patients with serologic tests and thus selecting those candidates to have an intestinal biopsy has provided further insights into CD diagnosis. The recent development of an ELISA technique for the determination of antibodies to tTG that correlate closely with EMA makes it possible to avoid some drawbacks in the determination of the latter by immunofluorescence. In our previous study, guinea pig tTG-ab had 94% specificity and 100% sensitivity. In the present study using human recombinant tTG, specificity was 100%, but sensitivity was 95%, lower than that in our previous study. We found two patients with CD who, on a gluten-containing diet, had positive EMA but doubtful or negative tTG-ab. Sera positive for EMA but negative for tTG-ab may prove to be of great help for the investigation of other endomysial antigens, because tTG is not the unique endomysial antigen involved in CD. We have not compared titers of tTG-ab with those of antigliadin antibodies, because responses of patients to these two antigens may be different. The determination of serum antibodies to tTG by ELISA are at the moment the laboratory test with better sensitivity and specificity for the diagnosis of CD, and so could replace the determination of antigliadin antibodies advantageously. Studies are needed to monitor changes in these antibodies in relationship with gluten withdrawal and intake in patients with CD and to determine if they eventually could replace antigliadin antibodies in the monitoring of the adherence to the gluten-free diet. In doubtful cases, EMA would be of great value. Antibodies to tTG are also invaluable for mass screening; in studies of silent CD requiring the performance of huge numbers of tests, an ELISA is more convenient for this purpose than immunofluorescence. Nevertheless, intestinal biopsy remains fundamental for the diagnosis of CD for the time being, despite all research efforts to find a noninvasive diagnostic test. TABLETABLE 1: Distribution of values between the three groups of patients