Introduction: The gold standard in the diagnosis of gastrointestinal graft-versus-host disease (GVHD) after allogenic hematopoietic stem cell transplantation is conventional endoscopy with histopathological assessment. The role of capsule endoscopy (CE) is uncertain. The aim of the present study was to investigate the accuracy of CE in this setting, comparing the clinical, endoscopic and histological data.
Methods: This was a retrospective single-centre study that included 57 (mean age: 47.18±15.05 years, 57.89% men) patients presenting with GVHD who underwent a valid CE and histopathological analysis by conventional endoscopy between January 2004 and July 2016. The endoscopic scored findings, clinical data and histopathological diagnosis were compared using duodenal histology as the gold standard.
Results: CE detected mild (n=4, 7.02%), moderate (n=6, 10.53%) and severe (n=21, 36.84%) grades, with a higher diagnostic yield than conventional endoscopy (54.39 vs. 28.07%, P<0.001). A positive CE predicted positive histology (80.64 vs. 15.38%, P<0.001) with a sensitivity, specificity, negative predictive value and accuracy of 86.21, 78.57, 84.62 and 82.46%. This procedure detected more mild-to-moderate histological lesions than conventional endoscopy (8.77 vs. 3.51%, P=0.25). In addition, 16% of patients with a previous normal endoscopy with biopsies had a pathological CE and there were eight (25.81%) patients with positive CE images unreachable by conventional endoscopy.
Conclusion: CE is a useful device in GVHD, achieving high accuracy values and diagnostic yield. However, its results may be interpreted in conjunction with clinical and histological features, particularly in mild-to-moderate stages.