Temporal discordance between graft-versus-leukemia and graft-versus-host responses: a strategy for the separation of graft-versus-leukemia/graft-versus-host reactivity?

The graft-versus-leukemia (GVL) effect is often coexpressed with graft-versus-host disease (GVHD), although the temporal kinetics of these responses have not been critically examined. To evaluate this question in the absence of the confounding effects of the conditioning regimen, 23 patients who received donor lymphocyte infusions from HLA-identical siblings and subsequently developed GVHD and/or a GVL response were studied to determine whether these were temporally synchronous events. The GVL effect occurred significantly earlier than GVHD, being that 19 of 23 patients had a sustained GVL response that antedated the onset of clinical GVHD. The median difference between time to GVL and graft-versus-host (GVH) reactivity in the entire cohort was 14 days. There was no correlation between total T-cell dose and the relative onset of GVL versus GVH reactivity, indicating that temporal dissociation of GVL and GVH responses was not a function of the absolute number of infused donor T cells. These data support existing murine bone marrow transplantation studies indicating that GVL and GVH responses are not temporally synchronous events and raise the possibility that targeted elimination of alloreactive donor T cells after bone marrow transplantation may be an effective strategy for the separation of GVL/GVH reactivity.