A 10-year Retrospective Study of Celiac Disease (CD) and other Gluten Sensitive Patients at the University of California at San Diego (UCSD)

Abstract

Purpose: To assess diagnostic testing of gluten sensitive patients seen at UCSD. Methods: We retrospectively evaluated patients seen at UCSD in the past 10 years diagnosed with CD or NCGS according to Olso criteria(1). Demographic factors, lab studies and pathology reports were reviewed. All serological testing analyzed was obtained while patients were consuming gluten. Results: Three major categories were identified; celiac disease (CD) with modified Marsh 3 histology [n=116, 73% female (F)]; non-celiac gluten sensitive (NCGS) with CD-like symptoms without histological CD while consuming gluten [n=68, 75%F]; and undiagnosed patients with gluten sensitive symptoms, without biopsies obtained prior to avoiding dietary gluten [n=32, 81%F]. Mean ages were similar in the 3 groups. Tissue transglutaminase (TTG) IgA levels were compared to modified Marsh 3 histology. Sensitivity/specificity of TTG IgA IgA>19 AU were 81%/93%, respectively. When TTG IgA was three times greater than the upper limit of normal (19 AU), sensitivity/specificity were 74%/100%. PPV and NPV were 100% and 75%, respectively. We determined how often deamidated gliadin peptide (DGP) IgA and IgG correctly predicted CD when TTG IgA levels were not elevated. When TTG IgA was <20AU in CD patients (n=12), two had DGP IgA>19, three had DGP IgG>19. Combined sensitivity of TTG IgA and DGP IgA or DGP IgG was 84% or 85%, respectively; specificity was 93% or 91%. At UCSD, a TTG IgA order set includes a reflexive measurement of endomysial antibody (EMA). We determined how often this changed management using Cohen's kappa to measure the level of agreement between the two tests. This value was 0.908 (P <0.001) indicating high agreement. We examined how often TTG IgA levels were predictive of IgA deficiency. A Spearman's rank correlation was 0.29 (P=0.009), indicating a statistically significant relationship. The average value of TTG IgA in CD patients with IgA deficiency was 1.33 AU, significantly lower than the average value of TTG IgA in those with normal IgA levels, 54.9 AU (P< 0.001). Conclusion: Gender and mean age were similar in the three different categories of gluten sensitive patients at UCSD. Specificity of TTG IgA was comparable to previous reports, while sensitivity was lower. As expected, using TTG IgA levels three times greater than the upper limit of normal decreased sensitivity but specificity was 100%. Adding DGP IgA/G testing to TTG IgA enhanced the detection of CD only slightly. We showed that reflexive EMA testing did not improve detection of CD beyond TTG IgA alone. Very low TTG IgA levels were predictive of IgA deficiency.