Comparison of anti-transglutaminase ELISAs and an anti-endomysial antibody assay in the diagnosis of celiac disease: a prospective study.

BACKGROUND Most studies of anti-transglutaminase (anti-tTG) assays have considered preselected groups of patients. This study compared the sensitivity, specificity, and predictive value of an immunofluorescence method for anti-endomysial antibodies (EmAs) and two anti-tTG ELISAs, one using guinea pig tTG (gp-tTG) and the other human tTG (h-tTG) as antigen, in consecutive patients investigated for suspected celiac disease (CD). METHODS We studied 207 consecutive patients (99 men, 108 women; age range, 17-84 years) who underwent intestinal biopsy for suspected CD. Patients presented with one or more of the following: weight loss, anemia, chronic diarrhea, abdominal pain, dyspepsia, alternating bowel habits, constipation, pain in the joints, and dermatitis. At entry to the study, an intestinal biopsy was performed and a serum sample was taken for IgA EmAs, anti-gp-tTG, and anti-h-tTG. RESULTS Intestinal histology showed that 24 patients had partial or total villous atrophy; in these patients the diagnosis of CD was confirmed by follow-up. The remaining 183 patients had villous/crypt ratios that were within our laboratory's reference values and were considered controls. Serum EmAs, anti-gp-tTG, and anti-h-tTG were positive in all 24 CD patients; in the control group, none were positive for serum EmAs, but 15 of 183 (8.2%) were positive for anti-gp-tTG, and 6 of 183 (3.3%) were positive for anti-h-tTG. Sensitivity was 100% for all assays, whereas specificity was 100% for the EmA, 92% for the anti-gp-tTG, and 97% for the anti-h-tTG assay. The negative predictive value was 100% for all assays; the positive predictive value was 100% for the EmA, 80% [95% confidence interval (CI), 65-95%] for the anti-h-tTG (P = 0.03 vs EmA) and 60% (95% CI, 44-76%) for the anti-gp-tTG assay (P = 0.0002 vs EmA). Areas (95% CIs) under the ROC curves were 0.987 (0.97-1.0) for anti-h-tTG and 0.965 (0.94-0.99) for anti-gp-tTG. Most of the patients testing false positive for anti-tTG had Crohn disease or chronic liver disease. CONCLUSIONS Although both anti-tTG ELISAs showed optimum sensitivity, their lack of specificity yielded positive predictive values significantly lower than those for the EmA assay.