Poster Presentation International Celiac Disease Symposium 2026

Assessing histologic change in serial celiac duodenal biopsies using routine pathology practices: Factors affecting accuracy and interobserver reliability. (#267)

Marie Robert 1 , Eric Chen 2 , Cristina Pacheco 3 , Dipti Karamchandani 4 , Raul Gonzalez 5 , Jeffrey Goldsmith 6 , Stephen Lagana 7 , Michael Arnold 8 , Rish Pai 9 , Nicole Panarelli 10 , Daniel Leffler 11 , Benjamin Wilkins 12
  1. Yale University School of Medicine, Woodbridge, CT, United States
  2. Pathology, Mayo Clinic, Rochester, Minnesota, USA
  3. Pathology, Cincinnati Childrens Hospital, Cincinnati , Ohio, USA
  4. Pathology, UT Southwestern, Dallas, Texas, USA
  5. Pathology, Emory University, Atlanta, Georgia, USA
  6. Pathology, Boston Children's Hospital, Boston, Massachusetts, USA
  7. Pathology, Columbia University Medical Center, New York, USA
  8. Pathology, Stanford University, Palo Alto, California, USA
  9. Pathology, Mount Sinai Health Center, Toronto, Ontario, Canada
  10. Pathology, Cornell University Medical Center, New York, USA
  11. Chugai Pharmaceuticals, Boston , Massachusetts, USA
  12. Pathology, University of Washington Medical Center, Seattle, Washington, USA

Premise:Once celiac disease(CeD) therapeutics exist, payors may require demonstration of persistent histopathology before approving payment. Therefore, pathologists may be asked to compare injury in diagnostic/follow-up biopsies, with the result determining coverage. Interobserver pathologist agreement on grade of CeD injury is modest in the literature. This study sought to establish the minimal perceptible difference in villous-crypt ratio(VhCd) between CeD biopsies that can be perceived by pathologists without software tools.

Methods:10 GI pathologists compared 50 pairs of digitally scanned hematoxylin&eosin-stained CeD duodenal biopsies. Previously acquired VhCd measurements served as ground truth (GT). Differences in VhCd between two paired biopsies ranged from 0.19-2.79um. Based on VhCd, pathologists were instructed to designate which biopsy in the pair showed more enteropathy. After a wash out period, the exercise was repeated to include both VhCd and intraepithelial lymphocyte(IEL) density in the comparison assessment.


Results:Consensus(defined as 8/10 pathologists) agreement on which biopsy showed more enteropathy was achieved in 42/50(84%) biopsy pairs. 40/42(95%) consensus decisions were aligned with ground truth VhCd measurements.  80% agreement with GT was achieved at a threshold VhCd difference of 0.84um between the two biopsies in a pair (Figure 1). Smaller differences in VhCd were significantly associated with fewer correct responses(p<0.01).  Adding IEL density to the assessment did not significantly affect accuracy.  Variable VhCd within one biopsy/paucity of oriented VhCd units were cited as challenges.


Discussion:Using routine H&E evaluation without quantitation, the minimal detectible VhCd difference pathologists can detect between duodenal biopsies was 0.84um. Disagreements between pathologists’ assessment and software measured VhCd suggest that variability in VhCd in single biopsies/poor biopsy orientation may confound the ability to determine which of two biopsies shows more enteropathy.  Additional studies are underway to clarify this challenge in an era in which pathologists may play a role in drug eligibility.