Gluten challenge in the diagnosis of celiac disease
N van Rheenen1, RHJ Houwen1, FJW ten Kate2, A Mubarak1
Departments of Paediatric Gastroenterology (1) and Pathology (2), University Medical Center
Utrecht, The Netherlands
Patients suspected to have celiac disease (CD) because of symptoms or positive serology, but
with apparently normal duodenal histology, may still be suffering from CD but lack classical
histology due to insufficient gluten intake. Therefore it is common practice to perform a gluten
challenge (GC) in such patients, although studies supporting this approach are lacking. The
purpose of this study was therefore to define the diagnostic yield of a GC in paediatric patients
suspected to have CD but without histological evidence of the disease on duodenal biopsy.
All patients with CD-like symptoms and/or positive CD serology who had undergone a GC
between 1995 and 2011 due to the absence of classical CD histology upon duodenal microscopy
were included in the study. The GC comprised of >200 mg/kg/day of gluten-powder added to the
diet of the patients for a period of 6-12 weeks. All initial biopsies were revised by a single
experienced pathologist, who was blinded to the clinical and serological data, using the Marsh
classification. The biopsies were further classified into 4 groups: no CD (Marsh 0) but positive CD
serology, borderline enteropathy (Marsh I-II), CD (Marsh III), and inconclusive histology (CD
cannot be confirmed nor rejected). Finally, biopsy results before and after GC were compared.
A total of 34 patients met the inclusion criteria of the study. Of those, 16 patients (47.1%) were
diagnosed with CD at revision of the histology, including 1 patient with negative serology. The
diagnosis was rejected in 2 patients who both had negative serology, making a GC unnecessary in
retrospect. Of the 16 remaining patients, 8 had inconclusive biopsies initially (all with positive
serology), 3 a Marsh I lesion (2 with negative serology) and 5 patients a Marsh 0 lesion but
positive serology. After GC, no histological abnormalities were found in all patients with initially
Marsh I lesions and in 3 of the 5 patients with a Marsh 0 lesion after revision. In the remaining 2
patients with Marsh 0 lesions the GC resulted in inconclusive biopsies. Finally, of the patients
with initially inconclusive biopsies the diagnosis CD was confirmed in half of the patients,
rejected in 3 patients and remained inconclusive in 1 patient.
In conclusion, in children suspected to have CD, but lacking histological confirmation, the first
step is to revise the biopsies as this leads to confirmation of CD in almost half of the patients.
In patients who have an inconclusive biopsy, a GC reveals CD in half of the cases. By contrast, in
patients with a Marsh 0 or Marsh I lesion on initial biopsies, the yield of a GC is low.