J Clin Gastroenterol. 2021 Mar 1;55(3):227-232. doi: 10.1097/MCG.0000000000001349.

Alina Popp ^1 2 3, Taina Arvola ^4 5, Juha Taavela ^1 2 6, Laura Kivelä ^1 2, Adina Ene ³, Kaija Laurila ^1 2, Päivi Saavalainen ⁷, Markku Mäki ^1 2, Kalle Kurppa ^1 2 8 3

Author information

• ¹Centre for Child Health Research, Tampere University.
• ²Department of Pediatrics, Tampere University Hospital.
• ³National Institute for Mother and Child Health, University of Medicine and Pharmacy "Carol Davila", Bucharest, Romania.
• ⁴Allergy Centre, Tampere University Hospital, Tampere.
• ⁵Hämeenlinna Central Hospital, Hämeenlinna.
• ⁶Central Finland Central Hospital, Jyväskylä.
• ⁷Research Programs Unit, Immunobiology and Department of Medical Genetics, Haartman Institute, University of Helsinki, Helsinki.
• ⁸The University Consortium of Seinäjoki, Seinäjoki, Finland.

Abstract

Goals: To test the accuracy of serology-based criteria for diagnosing celiac disease utilizing quantitative histomorphometry.

Background: The revised European pediatric guidelines allow noninvasive celiac disease diagnosis for a subgroup of children. However, in some of the studies on this issue, the positive predictive value (PPV) of serology has remained suboptimal, possibly because of challenges of histopathology as the reference standard.

Study: Prospectively enrolled children with transglutaminase 2 antibodies (TGA) above the upper limit of normal (ULN) underwent blood sampling and duodenal biopsy in Finland and Romania. Those with TGA ≥10× ULN, positive endomysium antibodies (EmA), and disease-associated genetics were considered to fulfill triple criteria for celiac disease. Initial histopathologic analysis was conducted using grouped classification, whereupon centralized morphometry was performed.

Results: Altogether 88 (54%) children were triple positive. In local evaluation, 99% of triple-positive children and 73% of children with TGA <10× ULN had celiac disease. These figures increased to 100% and 85% after more precise morphometric analysis. Triple-positive children had more anemia and higher median EmA and liver enzyme values than those with TGA<10× ULN; the groups were comparable in other clinical features and laboratory parameters.

Conclusions: When applied as recommended, the nonbiopsy strategy had already yielded excellent PPV regardless of the site of diagnosis or clinical presentation in the local analysis. PPV further increased to 100% with standardized duodenal morphometry.