Increased risk of lethal graft-versus-host disease-like syndrome after transplantation into NOD/SCID mice of human mobilized peripheral blood stem cells, as compared to bone marrow or cord blood

We tested the ability of human cells from different hematopoietic tissues to generate graft versus host disease-like syndrome (GVHD) in sublethally irradiated non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice. Tissue sources of human hematopoietic cells were: (1) bone marrow (BM), (2) nonmobilized peripheral blood (PB), (3) mobilized peripheral blood stem-progenitor cells (PBSC), and (4) cord blood (CB). To avoid interindividual donor variation, part of this study was done using BM, PB, and PBSC donated by a single healthy adult volunteer. A total of 179 NOD/SCID mice received graded human hematopoietic cell doses 5-500 x 10(6) mononuclear cells (MNC), containing 2-325 x 10(6) CD3(+) T cells, per mouse] from individual donors. Mice were observed for the development of GVHD and sacrificed 60 days after transplantation (earlier if ill). Mice were analyzed quantitatively by flow cytometry for human hematopoietic cell types and histologically, especially for human T lymphocytes infiltrating BM. No mouse transplanted with the tested doses of human CB or BM cells developed GVHD (experimentally defined as >10% human T lymphocytes infiltrating the mouse BM). For PB and PBSC, the frequencies of death, death with GVHD, and GVHD were directly related to the dose and source of human cells. Because PB cells contaminate harvested BM, the results from infused BM and PB were next combined for further analysis (BM/PB). The relative risks (hazard ratios estimated from the proportional hazards model) for death with GVHD, for each 10 human T cell dose increase, were 1.15 for BM/PB (p < 0.0001) and 1.47 for PBSC (p < 0.0001). In this in vivo xenogeneic model, the average T cell from human PBSC generated GVHD more potently than did the average T cell from human BM/PB, and the average CB T cell had a much lower GVHD potential. These results suggest that the potential for clinical GVHD from an HLA-disparate donor graft is likely to be quantitatively dependent both on the total number of T lymphocytes in the donor graft and the tissue source of the graft. Quantitative criteria for optimal T cell content of allogeneic donor hematopoietic grafts from different sources are discussed.