Detecting, Diagnosing Celiac Disease

Vol. 12 •Issue 4 • Page 74
Detecting, Diagnosing Celiac Disease

Clinical symptoms of CD are variable and often mild, resulting in significant delays in diagnosis.

Celiac Disease (CD) is an autoimmune gastrointestinal disorder that may occur in genetically susceptible individuals triggered by the ingestion of gluten-containing grains such as wheat, barley and rye. Of the many autoimmune disorders, CD represents one of the few disorders where the etiological agent is known and the disease subsides and goes in remission once the etiological agent is withdrawn from the diet.

CD is characterized by malabsorbtion resulting from inflammatory injury to the small intestinal mucosa and, when prolonged, can cause malnutrition. The classical symptoms of CD include diarrhea, weight loss and malnutrition.

Only a small percentage of patients with CD, however, presents with classical symptoms. Consequently, the clinical spectrum of CD has grown much broader than in the past to include patients that do not present with classical symptoms. It is not uncommon for the initial symptoms to be non-gastrointestinal or for gastrointestinal symptoms, if present, to be mild or intermittent.

Some of the common non-gastrointestinal manifestations include short stature, iron and folate deficiency, anemia, bone loss, aphthous stomatitis, arthralgia, dental enamel defects, etc. (Table 1). The need to examine a wider range of clinical presentation has led to greater numbers of individuals diagnosed with CD later in life than ever before. Adults may present with iron deficiency, macrocytic anemia and hypocalcaemia.

Prevalence of CD

Studies have found the prevalence of CD to be highly variable from population to population.1 The true prevalence has been difficult to ascertain. The disparate criteria used in the diagnosis of CD are often the cause. If only the clinical criteria are used in determining prevalence, the incidence of CD is much lower as compared with incidence established by serological methods (Table 2).1,2

This raises a question: How effective are the clinical criteria for the diagnosis of CD? Diagnosis of CD based on clinical criteria can be misleading and may lead to serious delays in proper diagnosis. Frequently, delays in diagnosis extend 10-13 years from the first clinical presentation of symptoms.

Failure to diagnose CD early on may predispose an individual to long-term complications such as splenic atrophy and intestinal lymphoma. The incidence of lymphoma arising in the context of CD is difficult to ascertain. One study has shown incidence of lymphoma involving the gastrointestinal (GI) tract in patients with CD to range from 3.6 percent to 40 percent.3 In another recent study,4 CD is associated with significantly elevated risk for intestinal lymphoma, especially for non-Hodgkin’s. A gluten-free diet (GFD) normalizes the mucosa and helps reduce the malignant potential.

The overall risk of malignancy in patients with CD on strict GFD for more than five years is very close to the general population. Mere reduction of dietary gluten has much more limited effect (Table 3).

The advent of serological methods for the detection of antibodies to gliadin, endomysium and tissue transglutaminase have enabled large-scale screening studies for CD both in Europe and the United States. These studies suggest that CD is far more prevalent than thought previously.2,5 Recent serological studies 2 demonstrate similar incidences of CD–between one in 130 and one in 500. Prevalence of CD is much higher still in first and second-degree relatives of patients with CD.

Laboratory Methods

Historically, the diagnosis of CD was based primarily on histological studies of the Jejunal biopsy characterized by partial to total villous atrophy, crypt hyperplasia, increased mitotic index and increased lymphocytic and plasma cell infiltrate in the lamina propria (Fig. 1).

Histological examination of the small intestinal biopsy remains the gold standard for diagnosing CD, but there are some difficulties with the procedure. Many patients with CD are small children and histological studies may be viewed by many, especially the child’s parents, to be invasive and traumatic. In addition, the histological changes are progressive initially—just the cellular infiltrate followed by crypt hyperplasia and finally villous atrophy. This may pose problems for reading and interpretation.

For example, a biopsy with an abnormally high density of intraepithelial lymphocytes and normal villous architecture may be interpreted as normal. CD might also be confused with other disorders when diagnosed histologically. Parasitic infections (e.g., Giardia lamblia), cow’s milk sensitive enteropathy and malabsorption syndrome, for example, may mimic CD histology. Finally, it has also been reported that some patients with latent or even active CD may even have normal histopathology.6

The European Society of Paediatric Gastroenterology and Nutrition (ESPGHAN) in 1969 established three biopsy criteria to confirm the diagnosis of CD. The revised ESPGHAN criteria included only a single biopsy with clear-cut remission of clinical symptoms on GFD. Positive serology at the time of diagnosis with disappearance on GFD contributes to the diagnosis. Although the revised criteria are appropriate in most cases, there are instances when examination of additional biopsies may be necessary.

As the limitations of the histology have been recognized, serum antibody tests have gained acceptance in screening for CD and in follow-up of patients with CD to determine their compliance with the GFD. The various serological tests employed in the work-up of patients suspected to have CD include anti-gliadin antibody (AGA), anti-endomysial antibody (EMA), anti-reticulin antibody (ARA) and anti-tissue transglutaminase (tTG) antibody tests. Antibodies to gliadin and tTG are detected by ELISA, whereas endomysium and reticulin antibodies are detected by indirect immunofluorescence. Of the serum antibody tests, EMA and tTG antibody primarily detect antibodies of IgA immunoglobulin isotype, whereas the AGA test detects both IgG and IgA isotypes. No IgM class antibodies to these antigens are detected in patients with CD; hence, there is no need to test for IgM class antibodies in the laboratory work-up of patients with CD.

Of these tests, AGA was the first to be described in the literature and has been evaluated most extensively.7 AGA of IgG is more sensitive but less specific than IgA-AGA. The major utility of IgG-AGA is in CD patients who are IgA deficient. In a study conducted recently in our laboratory,7 all of the 15 IgA-deficient CD patients were found positive for IgG-AGA and negative for IgA-AGA and other autoantibodies.

EMA and ARA are very specific indicators of CD. These assays are immuno-histochemical methods and require experience in reading immunofluorescence reactions (Fig. 2). Some investigators suggest that they are less sensitive. However, in all the studies conducted since our laboratory first described EMA test in 1983, we find EMA to be 100 percent specific and sensitive for CD. Other investigators may find EMA to be less sensitive due to the selection of the substrate, fixation of tissue sections, specificity of the conjugate employed or serum screening dilution. Internally, we find that testing for EMA at dilutions of 1:2.5 or 1:5 yield 5 percent of patients positive for EMA yet negative at 1:10 or 1:20. It could be that some of the investigators who have reported low sensitivity might be screening the patients at high serum dilutions.

In addition, proper selection of the substrate is very important. If one is using the primate distal esophagus, the quality control should be such to make certain that the substrate has appropriate antigenic smooth muscle against which the EMA react. Without proper quality control, one may read skeletal muscle rather than smooth muscle on the tissue section or find minimal antigenic substrate. Either condition will hinder inappropriate reading and interpretation.

Some laboratories use the umbilical cord as the substrate for detecting EMA. We find umbilical cord to be somewhat less sensitive than either distal esophagus or primate intestinal smooth muscle. In addition, it can be difficult to differentiate smooth muscle antibody reactions from the EMA reactions on the umbilical cord. In proficiency studies8 and recent CAP proficiency there was a significant variability in the sensitivity of the assay. Because of the variability of the assays for CD, Rosario et al9 concludes that “EmA-IgA is 100 percent sensitive and specific in active, untreated IgA-sufficient CD patients when performed by an established laboratory.”

Since identification of tTG as the endomysial antigen, ELISA methods have been described for detecting antibodies in the sera of patients with CD. In a study conducted by Conrad (Dresden Symposium, 1998), there is a good correlation of the tTG antibody assay to the EMA when the EMA titers are strong (>1:40). The sensitivity of the assay is somewhat compromised when the EMA titers are low. Similar observations have been made by our laboratory and others. We have also observed some patients with high EMA titers that are consistently negative for tTG antibodies. This raises the question that there may be some other antigens in addition to the tTG or that some of the epitopes on tTG may not be accessible to react to the EMA antibodies.

The advantage of the anti-tTG antibody assay is that it is automatable and less subjective than EMA. For this reason, many laboratories have opted to use the tTG antibody method as the screening method. In these laboratories, it may be the only assay used for detection of CD cases. In various studies on the efficacy of the tTG antibody method for screening for CD, the specificity and sensitivity of this method has been found to range from 90 percent to 95 percent.10 Table 4 summarizes the specificity of the AGA, EMA and tTG antibody methods most commonly employed by laboratories performing tests for CD.

Based upon specificity and sensitivity of the various serological markers, EMA is the only method that provides 100 percent positive and negative predictive value for CD. 10 In contrast, tTG will only have 14 percent positive predictive value. The optimal method of screening for CD depends upon the likelihood of CD in the population studied and the experience of the laboratory performing the test.

Some investigators may use only the ELISA methods (AGA or tTG). The disadvantage of this approach will be many false positives, thus necessitating unnecessary biopsies and other examinations. Because of the high degree of specificity and sensitivity of the EMA test, Murray and coworkers conclude in their study that EMA provides the least costly method of detecting CD when incidence of CD is lower then 42 percent, as is the case in most laboratories performing these tests. 8 In the case of laboratories that opt to use the ELISA methods for CD testing, our recommendation is to confirm all positive cases by EMA. This approach will identify all CD patients, including those that are IgA deficient.

CD patients are prescribed a GFD for life. Serological tests are useful in monitoring a patient’s response and adherence to the GFD. The levels of the various antibodies (AGA, EMA, ARA and tTG) decrease and eventually disappear in the majority of the patients on a complete GFD. Similarly, these antibodies either appear or rise in level when the patient is on a gluten-containing diet. Serological methods, therefore, play a significant role in both diagnosis and follow-up of CD patients.

CD has been associated with many other autoimmune disorders such as type 1 diabetes, thyroid autoimmunity and other autoimmune disorders. Approximately 5 percent of patients with type 1 diabetes have CD. 11 Similarly, approximately the same percentage of patients with CD has type 1 diabetes.12 It has been proposed that early detection of CD may be beneficial in such cases as it is believed that adherence to a GFD may delay the onset of diabetes. If true, this further emphasizes the utility of and need for serum antibody tests in the screening of population genetically susceptible for CD.


Clinical symptoms of CD are variable and often mild, resulting in significant delays in diagnosis. The use of serological tests has improved ease of detection, monitoring and continuing care of CD patients.

Dr. Kumar is president, IMMCO Diagnostics Inc., Buffalo, NY.


1. Guandalini S, Gupta P. Celiac disease- A diagnostic challenge. Clin Appl Immun Rev 2002;2:293-305.

2. Fasano A, et al. Prevalence of celiac disease in at risk and not-at-risk groups in the United States 2223; 163:286-292.

3. Wolber R, et al. Lymphocytic gastritis in patients with celiac sprue or spruelike intestinal disease. Gastroenterology 1990;98:310-315.

4. Catassi, et al. Risk of non-Hodgkin lymphoma in celiac disease. JAMA 2002;287:1413-1419.

5. Hill ID, et al. Celiac disease: Working group report of the first world congress of pediatric gastroenterology, hepatology, and nutrition. J. Pediatr Gastroenterol Nutr 2002; 35:S78-S88.

6. Arranz E, Ferguson A. Intestinal antibody pattern of celiac disease: Occurrence in patients with normal jejunal biopsy histology. Gastroenterology 1993;104:1263-72.

7. Kumar V, et al. Celiac disease and immunoglobulin a deficiency: How effective are the serological methods of diagnosis? Clin Diagnostic Lab Immunol 2002;9:1295-1300.

8. Murray, et al. Serologic testing for celiac disease in the United States: Results of a multilaboratory comparison study. Clin Diagn Lab Immunol 200; 7:584-587.

9. Rosario D, et al. Further studies of anti-endomysium and anti-gliadin antibodies in patients with suspected celiac disease. J Pediatr Gastroenterol Nutr 1998;27:191-195.

10. Vitoria JC, et al. Antibodies to gliadin, endomysium, and tissue transglutaminase for the diagnosis of celiac disease. J Pediatr Gastroenterol Nutr 1999;29:571-4.

11. Aktay AN, et al. The prevalence and clinical characteristics of celiac disease in juvenile diabetes in Wisconsin. J Pediatr Gastroenterol Nutr 2001;33:462-5.

12. Kumar V, Rajadhyaksha M, Wortsman J. Celiac disease-associated autoimmune endocrinopathies. Clin Diagn Lab Immunol 2001;8:678-85.