OBJECTIVES: The diagnosis of coeliac disease (CD) remains sometimes difficult because the histological criteria are not fully met. The aim of this study was to refine histological diagnostic criteria of CD. METHODS: One hundred seventy-five duodenal bulb D1 (n = 79) and duodenal D2 (n = 96) biopsies of 96 patients with CD (58 girls, mean age 7 years), 135 normal D2 biopsies (69 girls, mean age 12 years), and 64 D2 biopsies of other digestive disorders (DDs) (39 girls, mean age 13 years) obtained from children during a period of 4 years were reviewed. RESULTS: Interobserver agreement was greater for the classification of Corazza-Villanacci than for Marsh-Oberhuber (κ = 0.812 vs κ = 0.409, respectively). Between 40 and 70 intraepithelial lymphocytes (IELs) per 100 epithelial cells (ECs), 32% of patients were CD, whereas 50% had other DD. Above 70 IELs per 100 EC, 53% were CD, and only 6% had other DD. In CD, IELs were significantly located above EC nuclei compared with other DD, (12 IELs/100 EC vs 2 IELs/100 EC, respectively). In 21% of CD cases, D2 were normal and the diagnosis could only be made on D1. Finally, 6% of CD cases showed isolated increase of IELs in D1 without architectural modification. CONCLUSIONS: D1 allowed diagnosis of CD in 21% of cases and IEL >70 per 100 EC correlated strongly with CD. Between 40 and 70 IELs per 100 EC, CD is very likely but other DD must be considered. Finally, the preferential localisation of IELs above EC nuclei favours CD, and increased IEL in D1 may be the sole abnormality.
OBJECTIVES: The diagnosis of coeliac disease (CD) remains sometimes difficult because the histological criteria are not fully met. The aim of this study was to refine histological diagnostic criteria of CD. METHODS: One hundred seventy-five duodenal bulb D1 (n = 79) and duodenal D2 (n = 96) biopsies of 96 patients with CD (58 girls, mean age 7 years), 135 normal D2 biopsies (69 girls, mean age 12 years), and 64 D2 biopsies of other digestive disorders (DDs) (39 girls, mean age 13 years) obtained from children during a period of 4 years were reviewed. RESULTS: Interobserver agreement was greater for the classification of Corazza-Villanacci than for Marsh-Oberhuber (κ = 0.812 vs κ = 0.409, respectively). Between 40 and 70 intraepithelial lymphocytes (IELs) per 100 epithelial cells (ECs), 32% of patients were CD, whereas 50% had other DD. Above 70 IELs per 100 EC, 53% were CD, and only 6% had other DD. In CD, IELs were significantly located above EC nuclei compared with other DD, (12 IELs/100 EC vs 2 IELs/100 EC, respectively). In 21% of CD cases, D2 were normal and the diagnosis could only be made on D1. Finally, 6% of CD cases showed isolated increase of IELs in D1 without architectural modification. CONCLUSIONS: D1 allowed diagnosis of CD in 21% of cases and IEL >70 per 100 EC correlated strongly with CD. Between 40 and 70 IELs per 100 EC, CD is very likely but other DD must be considered. Finally, the preferential localisation of IELs above EC nuclei favours CD, and increased IEL in D1 may be the sole abnormality.