An Analysis of the Application and Reproducibility of the NIH Consensus Guidelines for the Histologic Diagnosis of Gastrointestinal Acute Graft Versus Host Disease
Diana M Cardona, Michael J Shealy, Rex C Bentley, Emanuela Veras. Duke University Medical Center, Durham, NC
Background: Following bone marrow transplant (BMT), acute graft-versus-host disease (aGVHD) of the gastrointestinal (GI) tract is a common complication associated with significant morbidity and mortality. Clinical symptoms often raise concerns for aGVHD; however, histologic confirmation and exclusion of other disease entities is critical as presentations can overlap. The histologic diagnosis of GI aGVHD can be difficult. As a result, in 2006, the NIH published consensus guidelines for the histologic diagnosis of aGVHD.
Design: A retrospective search of our LIS beginning in 2004 was performed. We identified 25 gastric and 25 colon biopsies that, based on the interpretation of the prior pathologic diagnoses, represented an even distribution of potential NIH diagnostic categories (not aGVHD, possible aGVHD, consistent with aGVHD or unequivocal aGVHD). These biopsies represent 40 unique BMT patients (pts). Following review of the NIH Consensus guidelines, H&E and CMV immunohistochemical slides were blindly and independently reviewed by 3 pathologists and classified into one of the 4 NIH diagnostic categories. A chart review was performed to determine if there was a clinical diagnosis of aGVHD at the time of biopsy (based on the presence of skin or liver involvement or empiric treatment for GI symptoms without other known etiology).
Results: Of the 40 pts, 28 had a clinical diagnosis of aGVHD. These 28 pts had 37 biopsies (21 colon and 16 gastric). Twenty-five of the 37 biopsies were classified as either consistent with or unequivocal aGVHD by at least 2 of the 3 pathologists yielding a positive predictive value (PPV) of 96%, sensitivity (Sen) of 68% and specificity (Spec) of 92%. By site, 15 of 21 colon cases (PPV 100%, Sen 71%, spec 100%) and 10 of 16 gastric cases (PPV 91%, Sen 63%, Spec 89%) were diagnosed within the same categories. Of the clinically non-aGVHD pts, 6 were diagnosed with CMV or other infection, 3 with a potential drug toxicity, 1 with gastroesophageal reflux disease and 2 had resolution of symptoms without intervention. There was a single histologic false positive case within this group (presumed drug toxicity). The interobserver agreement for the classification of all 50 biopsies using the NIH guidelines was considered excellent (kappa 0.70-0.75) amongst the 3 pathologists.
Conclusions: Although the histologic confirmation of clinically diagnosed aGVHD can be elusive, the NIH diagnostic categories may aid in creating uniformity and diagnostic clarity.
Wednesday, March 21, 2012 9:30 AM
Poster Session V # 79, Wednesday Morning