Introduction Counting intraepithelial lymphocytes (IEL) has been a prominent diagnostic histological approach for distinguishing normal disease-control (DC) mucosae from celiac disease (CD) patients for over five decades. Most surprisingly, during that period, no definitive definition of a ”normal” count has ever appeared: indeed, the existing literature on this point is extremely fragile. In this new, multi-centre study, ROC-curve analysis was used to determine the optimal cut-off between normal and CD (Marsh III lesion) mucosae, based on IEL counts on over 400 mucosal biopsy specimens.
Method The study was designed during the International Meeting on Digestive Pathology, Bucharest 2015 in which investigators from 19 centres from eight countries and involving three continents recruited patients to measure numbers of IEL per 100 enterocytes in well-oriented duodenal biopsies. All centres worked to an agreed protocol to ensure uniformity of approach. Demographic and serological data were also collected, wherever feasible, to compare cases and controls.
Results The mean age of DC and CD groups were 38.3 and 45.5 years respectively, of which 59% and 71% were female. The mean IEL count was 54±18/100 enterocytes for CD compared with 13±8 for DC (p=0.0001). ROC-curve analysis indicated an optimal cut-off point of 25 IEL/100 enterocytes, with a sensitivity and specificity of 99% and 92%: the area under the curve was 0.0995. Using this technique we showed that other proposals in the literature, between 20–40 IEL, were far less discriminatory. Additionally, there was a sufficiently high number of biopsies to explore IEL counts across the sub-classification of the Marsh III lesion.
Conclusion The IEL count is not normally-distributed in CD. The numerical overlapping confirmed that IEL do not represent a bi-modally distributed population[s], thus being indicative of a continuously graded dose-response by IEL to environmental (gluten) antigenic influence. In the largest study of this type, we showed using ROC-curve analysis that for Marsh III lesions, a cut-off value of 25 IEL/100 enterocytes is the optimal dividing line between DC and CD biopsies. Finally, we were unable to demonstrate any difference in IEL counts between Marsh Stage III a,b,c, subdivisions, with histology or immunocytochemisty. These results are consistent with a growing literature that finds this sub-classification of little value.
Disclosure of Interest None Declared
- Coeliac disease
- Intraepithelial lymphocytes