B-062 A Self-Imposed Gray Area? Analysis of Anti-Tissue Transglutaminase and Anti-Endomysial Antibody Discordance in a Celiac Disease Screening Serology Testing Algorithm

Abstract

Abstract Background Celiac disease (CD) is an autoimmune enteropathy affecting around 1% of the population. While occasionally asymptomatic, the disease can present in a variety of ways and can be triggered by dietary gluten at any age. CD serology, including anti-tissue transglutaminase (anti-tTG) IgA and/or IgG, anti-endomysial antibody IgA and/or IgG, and anti-deamidated gliadin peptide IgA and/or IgG, is a key tool for CD screening, diagnosis, and monitoring. Of these tests, the clinical utility of anti-EMA testing has been questioned due to its high cost and limited sensitivity: 5–10% of CD patients do not test positive for anti-EMA and a negative anti-EMA does not rule out CD. The objective of this study is to: 1) evaluate the concordance between anti-tTG and anti-EMA serology test results, and 2) assess the clinical validity of our current CD reflex testing algorithm in which every positive anti-tTG test result goes on for anti-EMA testing. Methods Using our laboratory’s information system, we conducted a retroactive study on patients who underwent CD serology testing at our institution. Query was performed for all patient data [pediatric (<18 years) and adult (>18 years)] collected between April 2020 and August 2022 for anti-tTG IgA performed on the BioPlex 2200 (BioRad Laboratories Inc., Hercules, CA) and anti-EMA IgA (Euroimmun, Germany) (N = 124 308). Vendor-supplied clinical sensitivity and specificity are as follows: anti-tTG IgA 94.3% sensitivity, 98.8% specificity; anti-EMA IgA 95.3% sensitivity and 98.0% specificity. Results From our data, we calculated a mean positive agreement between our anti-tTG IgA and anti-EMA test results of 53.2% (95% confidence interval (CI) 47.5–58.8%) for all anti-tTG IgA positive cases (N = 6691, 5.4% of total anti-tTG IgA test results). The agreement between anti-tTG IgA and anti-EMA test results is improved if multiples of the upper limit of normal (ULN) for anti-tTG IgA is applied. The mean positive agreement between anti-tTG IgA and anti-EMA test results when the anti-tTG result is >10x ULN is 98.4% (95% CI 97.7–99.2%). However, if an anti-tTG IgA cutoff of <10x ULN is employed a mean positive agreement of 32.3% (95% CI 25.8–38.7%) is observed. At an anti-tTG IgA cutoff of <3x ULN, the mean positive agreement is just 11.8% (95% CI 8.2–15.4%). These findings are consistent in both pediatric (N = 29 381) and adult (N = 94 961) populations and are problematic as 45.4% of our positive anti-tTG IgA test results are accompanied by a negative anti-EMA. Conclusion The agreement between anti-tTG IgA and anti-EMA is dependent on the value of anti-tTG IgA with greater agreement observed with increasing anti-tTG IgA values. As all specimens at our institution with positive anti-tTG IgA test results are automatically reflexed for anti-EMA testing, has led to physician confusion as the majority of patients are falling into a gray area with conflicting CD serology test results. Our algorithm is also out of line with most CD screening guidelines that suggest an initial positive anti-tTG test result is followed up with biopsy. Finally, discontinuing anti-EMA reflex testing could result in cost savings of approximately $120 000 CAD per year.