Allogeneic bone-marrow (BM) transplantation has greatly improved the treatment of leukaemia. Conventionally, BM transplants are performed between individuals that are matched at their MHC loci (but differ with respect to 'minor' antigens). Donor T cells that recognize recipient minor antigens mediate a graft-versus-leukaemia (GVL) effect, but can also cause graft-versus-host disease (GVHD). However, an MHC-matched donor can be found for only a proportion of patients. Full MHC-haplotype-mismatched BM grafts have been made possible by depleting T cells from the graft (which reduces the otherwise lethal GVH reactions across the MHC barrier), but this leads to a higher incidence of relapse due to a decrease in the GVL effect, and a greater risk of graft rejection. Now, Ruggeri and colleagues report in Science that certain MHC mismatches between BM donors and recipients might, in fact, be advantageous — donor natural killer (NK)-cell alloreactivity correlates with the elimination of leukaemia relapses and graft rejection, and protects patients against GVHD.

NK cells, unlike T cells, are activated when they recognize an absence of MHC class I molecules. When MHC class I ligands are present, killer inhibitory receptors (KIRs) on the NK cells are engaged and the cells are not activated. Ruggeri et al. reasoned that mismatched BM transplants might, therefore, trigger NK-cell alloreactivity. To assess the effect of this, they analysed the correlation of donor NK-cell alloreactivity with rejection, relapse and GVHD in patients with acute myeloid leukaemia (AML). Because NK cells distinguish between groups of MHC molecules, rather than single variants, Ruggeri et al. were able to divide the donor–recipient pairs into two groups on the basis of the presence or absence of KIR–ligand incompatibility in the GVH direction. KIR–ligand incompatibility correlated with protection of patients against graft rejection, leukaemia relapse and GVHD. The survival rate after five years for AML patients with KIR-mismatched donors was 60%, which is a much better outcome than for matched, un-related donor transplants.

To explore this effect further, the group then performed experiments in mouse model systems. The transfer of human alloreactive NK-cell clones to non-obese diabetic (NOD) or severe combined immunodeficient (SCID) mice — which lack B cells and T cells — eradicated previously transplanted AML cells and prevented the death of the mice. In mismatched mouse transplants, the infusion of donor alloreactive NK cells also permitted the use of less drastic pre-transplant conditioning regimens. Infusion of NK cells after the BM transplant was able to convert mixed chimeras to stable full-donor chimerism. Importantly, the use of alloreactive NK-cell infusions permitted the use of otherwise lethal doses of allogeneic T cells for immune reconstitution — this NK-cell-mediated protection against GVHD seems to be due to the elimination of recipient antigen-presenting cells by the donor NK cells.

This new work has important implications for transplantation therapy. The most readily applicable concept is that choosing donors with KIR–ligand incompatibility in the GVH direction offers a striking advantage for survival. In the future, it will be interesting to see how the mouse models, in which donor alloreactive NK cells were infused as a supplement to pre-transplant conditioning, will translate to a clinical setting.