Background: It is unknown how many endoscopic biopsy specimens are needed to diagnose celiac disease (CD).
Objective: To determine the numbers of duodenal biopsy specimens needed to diagnose CD.
Design: Retrospective medical record audit, histology slide analysis, and chart review.
Setting: A tertiary-care university hospital.
Patients: Adults who underwent EGD to diagnose CD.
Interventions: Our pathology database was searched for the keywords "consistent with celiac disease," "not consistent with celiac disease," "villous atrophy," and "intraepithelial lymphocytes" from January 2001 to May 2006. The number of biopsy specimens was determined and graded for a modified Marsh classification, and charts were reviewed for diagnosis verification. CD was confirmed if Marsh grade 3A was found, even on one biopsy specimen.
Main outcome measurements: The number of biopsy specimens needed to make the diagnosis of CD.
Results: Of 247 cases, 102 patients were diagnosed with biopsy specimen-confirmed CD. In 9 patients, CD could not be confirmed on the basis of histology alone (highest Marsh lesion was grade 1 or 2), but a clinical diagnosis was made on the basis of presentation and serology. CD could be confirmed if only 2 biopsy specimens were obtained in 84 patients (90%), if only 3 biopsy specimens were obtained in an additional 5 patients (95%), and if at least 4 biopsy specimens were taken in the remainder. CD was ruled out in 145 patients. In 142 patients, biopsy specimens were uniformly negative; 3 patients had Marsh grade 1 lesions but negative serology.
Limitations: A retrospective design.
Conclusions: Only 2 biopsy specimens will lead to a confirmed diagnosis of CD in 90%, and a suspected diagnosis in all. For 100% confidence in diagnosis of CD, 4 duodenal biopsy specimens should be taken.