PREDICTING GRAFT-VERSUS-HOST DISEASE IN HLA-IDENTICAL BONE-MARROW TRANSPLANTS - A COMPARISON OF T-CELL FREQUENCY-ANALYSIS AND A HUMAN SKIN EXPLANT MODEL

Background. Graft-versus-host disease (GVHD) occurring after HLA-ident ical sibling bone marrow transplantation (BMT) is considered to be mai nly caused by minor histocompatibility antigen (mHag) disparities betw een the recipient and donor. In our laboratory, a human skin explant m odel has been successfully used to predict acute GVHD in HLA-identical sibling BMT. More recently, the frequency analysis of host-reactive h elper and cytotoxic T lymphocyte precursors (HTLp and CTLp, respective ly) has been shown to have potential application for predicting GVHD. In the present study, HTLp, and CTLp frequency analysis and the skin e xplant model were directly compared for their ability to predict acute GVHD in HLA-identical sibling BMT. Methods. Host-reactive HTLp and CT Lp frequencies were determined using a combined limiting dilution assa y. A human skin explant model was used to detect graft-versus-host rea ctions in vitro. The results from the skin explant model (graft-versus -host reaction grades I-IV) and T cell frequency analysis (>/< 1:100,0 00) were correlated with posttransplant GVHD outcome, respectively. Re sults. The skin explant model correctly predicted GVHD outcome in 77% of cases (P=0.03). HTLp frequencies were very low in all patient/donor pairs tested. None of them exceeded 1:100,000, although 9/18 recipien ts developed GVHD (greater than or equal to clinical grade II) after t ransplant. In all patients tested, the relationship between either hig h (>1:100,000) or low (<1:100,000) CTLp frequency and occurrence of GV HD appeared to be random (P=1.0). Conclusions. HTLp and CTLp, frequenc y analysis did not predict the occurrence of acute GVHD after HLA-iden tical sibling BMT. The human skin explant model, however, remained an accurate indicator of acute GVHD and probably detects mHag disparities .