The Predictive Value of Gene Expression Profiles for Acute Graft-Versus-Host Disease after Hematopoietic Cell Transplantation with Nonmyeloablative Conditioning for Hematological Malignancy.

Purpose: To test the hypothesis that global gene expression profiles of peripheral blood mononuclear cells (PBMNC) day +14 after hematopoietic cell transplantation with nonmyeloablative conditioning could predict the later occurrence of acute graft-versus-host disease (GVHD) grade II–IV.

Material: Between March 2000 and Marts 2006, 100 patients with hematological malignancies received peripheral blood stem cells from an human leukocyte antigen identical sibling/mother donor or from a matched unrelated donor following nonmyeloablative conditioning with low dose fludarabine and 2 Gy of total body irradiation. Post-transplant immunosuppression consisted of cyclosporine and mycophenolate mofetil. Only patients with sustained engraftment, who did not experience late-onset acute GVHD after day +100 were included; eight patients were excluded due to graft rejection, three patients due to suboptimal RNA or lacking PBMNC samples, and further 15 patients due to late-onset acute GVHD. Seventy-four patients were then eligible for microarray analysis.

Methods: RNA was precipitated from frozen PBMNC from day +14 post-transplant and gene profiling analyses were performed using Human Genome U133 Plus 2.0 GeneChip Array. The array data were normalized, RMA modelled and asinh transformed in R. The differentially regulated gene expression between the group of patients developing acute GVHD before day +40, +56 and +84 post-transplant compared to the patients never experiencing acute GVHD was identified and formed the basis for the subsequent principal component analysis (PCA) and classifying models. No patients experienced acute GVHD between day +85 and +100 post-transplant.

Results: The patients experiencing acute GVHD by different time points were separated from the patients never experiencing acute GVHD by the PCA plot. Furthermore the classifying models could separate the groups correctly in up to 93% of cases in the best of the classifying model. In addition, differentially regulated genes between the two groups were identified.

Conclusion: These data suggest that the pattern of gene expression profiles early post-transplant is able to predict patients with a high risk of later occurrence of acute GVHD from those never experiencing acute GVHD. This knowledge could be exploited to increase the immunosupression and thus prevent acute GVHD in patients at risk. Furthermore, candidate genes of interest for the pathogenesis of acute GVHD have been identified.