Background: The clinical prognosis of gastrointestinal (GI) graft-versus-host disease (GvHD) is based on either a clinical staging system or histological or endoscopic findings. How these different scores correlate with each other and which have the greater impact on transplantation outcomes is, however, not clear.
Methods: Clinical, pathological, and endoscopic findings of the upper GI tract on 201 patients who underwent allogeneic hematopoietic stem cell transplantation were reviewed. The association between clinical, histological, and endoscopy grading was assessed by Kendall correlation coefficient. The agreement between grading systems was evaluated by kappa statistics. Factors associated with survival or steroid resistance were analyzed by proportional hazard models.
Results: At disease onset, no strong association was found between pathological and clinical grade at disease onset (τ=0.034, P=0.6). In contrast, endoscopy score and clinical grades were strongly associated (τ=0.37, P<0.001). The Kappa concordance coefficient (0.20) between histological and endoscopy scores was poor. However, by multivariate analysis all three scores significantly predicted survival rates
Conclusions: Clinical, histological, and endoscopic scores poorly correlated with each other when estimated at the GI-GvHD onset. However, all three severe initial scores independently predict poor outcome. Of interest, clinical and endoscopic scores predict resistance to steroids while pathological does not.