Autoantibodies to human tissue transglutaminase: superior predictors of coeliac disease

Abstract

Background: Using selected sample populations, we compared sensitivity and specificity of autoantibodies to guinea pig and human tissue transglutaminase to assess if the human antigen is superior for predicting coeliac disease.

Methods: Four commercial enzyme-linked immunoassay kits using human tissue transglutaminase as antigen were used to measure autoantibody levels in serum samples from untreated adult coeliacs (n = 32). They were from a series of 130 cases diagnosed between 1997 and 1999 and chosen to bias the group towards subjects with negative autoantibodies when measured with guinea pig tissue transglutaminase as antigen. Samples from 38 control subjects (biased towards false-positive levels with guinea pig antigen) were used to compare specificity. We also assessed if human antigen kits could differentiate between levels in normal subjects and in selective IgA deficiency.

Results: Sensitivity for coeliac disease in this selected group using the human antigen kits ranged from 88% to 100%. Three kits showed significantly higher specificity (82%-97%, P < 0.05) than the guinea pig antigen kit (71%) for the samples studied. No kit achieved complete separation between normal autoantibody levels and lower levels in selective IgA deficiency.

Conclusions: All human antigen kits showed significantly higher sensitivity for coeliac disease compared to guinea pig antigen (P < 0.001). Receiver operating characteristic curves confirmed the superior diagnostic accuracy of the human antigen kits.