Characterization of Hepatic Dysfunction in Subjects Diagnosed With Chronic GVHD by NIH Consensus Criteria

Abstract

Hepatic chronic graft-versus-host disease (cGVHD) causes morbidity and current diagnostic criteria are nonspecific. An accurate diagnosis is imperative because overdiagnosis can lead to unnecessary treatment with immunosuppressive agents and raising the risk of opportunistic infections. We aim to characterize different patterns of liver injury and cytokine profiles associated with hepatic dysfunction in cGVHD, to evaluate the accuracy of the NIH Consensus Criteria (NCC) for hepatic cGVHD and to explore predictors for hepatic cGHVD. Patients were evaluated in this prospective cross-sectional study of patients with cGVHD recruited under a natural history protocol. Laboratory tests and cytokines were measured. The cGVHD were diagnosed and scored based on NCC. Clinically indicated liver biopsy specimens or autopsies were reviewed by an expert hepatopathologist (D.E.K.). Comparisons were made between groups, and univariable and multivariable logistic regression were calculated. Of the 302 patients enrolled, 151 fulfilled hepatic cGVHD based on NCC; however, 69% had at least 1 abnormal liver test result. Abnormal alanine aminotransferase (ALT) and aspartate aminotransferase were associated with lower platelets, higher total bilirubin (TB), total cholesterol, serum amyloid A, and IL 15. Abnormal ALP and gamma-glutamyl transpeptidase were associated with higher cholesterol, and IL7. Lower platelet count was associated with higher ALT, TB, and triglycerides and lower albumin. Of the 27 with liver tissue, 16 had histologic features of GVHD, only eight met clinical criteria for hepatic GVHD. Sensitivity and specificity of NCC in identifying hepatic GVHD were 50% and 27% (Kappa=-0.23). Only 6 had only hepatic GVHD, whereas 10 had hepatic GVHD with either iron overload, nodular regenerative hyperplasia, or steatosis. Multivariable logistic regression showed that ALP and total cholesterol were associated with hepatic GVHD and total cholesterol >220 mg/dL increased the sensitivity for histologic hepatic GVHD. In conclusion, abnormal liver enzymes in cGVHD are nonspecific and have poor correlation with histologic evidence for hepatic GVHD, highlighting the importance of histology. Cytokines provide insight into the pathogenesis of hepatic cGVHD. Decreased platelet count was associated with factors associated with liver disease including portal vein diameter, which may suggest progression of liver disease. This highlights the need of incorporating these factors in natural history study and using liver biopsy to understand the development of liver dysfunction in hematopoietic stem cell transplantation and to develop better instruments to decreased hepatic cGVHD related morbidity and mortality. The study was registered with a ClinicalTrials.gov identifier NCT00092235.