HOME
Return to Table of Contents
225 HOW MANY DUODENAL BIOPSIES ARE REQUIRED TO MAKE A DIAGNOSIS OF CELIAC DISEASE? WP Pais1, DR Duerksen1, NM Pettigrew2, CN Bernstein1 BACKGROUND: Histopathological diagnosis of celiac disease (CD) is considered the gold standard. Because the mucosal lesions are patchy in distribution, it has been our routine practice to obtain 4 random biopsies from the second duodenum to clinch or exclude the diagnosis of adult CD. We aimed to study how many duodenal biopsies are needed to clinch the diagnosis of CD in adults.
1Section of Gastroenterology, 2Department of Pathology, University of Manitoba, Winnipeg, Manitoba
METHODS: A chart audit was carried out at our tertiary care university hospital. All surgical biopsy results in our department of pathology database were searched with the keywords 'consistent with CD', 'not consistent with CD', 'villous atrophy', and 'intraepithelial lymphocytes'. All the pathology slides were retrieved, number of biopsy specimens was determined, and each specimen was graded according to Marsh classification modified by Oberhuber. We considered that Marsh Grade 3A or higher clinched a diagnosis of CD, if found even on 1 biopsy specimen.
RESULTS: We identified 248 patients who met our search criteria from 01/01 to 05/06. Of these, 102 were diagnosed with biopsy confirmed CD. The number of biopsies per case was: 1 in 1 case, 2 in 13 cases, 3 in 22 cases and at least 4 in 66 cases. All 102 cases had abnormal biopsies with at least Marsh grade 1 lesions in all biopsies. There were 9 cases where CD could not be clinched on the basis of histology alone (the highest Marsh lesion identified was grade 1 or 2). Of the remaining 93 cases, CD was clinched if at least 1 biopsy was Marsh grade 3A or higher. CD could have been clinched if only 2 biopsies were taken in 84 (90%), if only 3 biopsies were taken in an additional 5 (89/93=95%), and if at least 4 biopsies were taken in the remainder. When biopsy specimens were compared in each patient, we found 76 (75%) had identical Marsh grade, 18 (18%) had 1 grade difference, and 8 (8%) had 2 or more grade differences. We identified 145 subjects in which CD was ruled out and all biopsies were uniformly negative. In 3, Marsh grade 1 lesions were identified but antibody testing was negative.
CONCLUSIONS: While CD can be patchy, 75% of duodenal biopsies in CD cases and 100% of biopsies in non-CD cases show uniform findings. When CD is suspected, taking only 2 biopsies will lead to a clinched diagnosis in 90% and a suspected diagnosis in all. 3 biopsies led to a clinched diagnosis 95% of the time and 4 biopsies led to a clinched diagnosis in 100%. For 100% confidence in clinching the diagnosis, there is minimal extra time and a small cost. Hence we recommend taking 4 random duodenal biopsies when CD is suspected.