Permissive, nonpermissive HLA-DPB1 epitope disparities and the specificity of T cells infiltrating the skin during acute graft-versus-host disease

To the editor:

Human leukocyte antigen (HLA)–DPB1 functions as a classic transplantation antigen.¹  In the context of hematopoietic stem cell transplantation (HSCT), when donor T cells recognize host HLA-DP, they can induce a graft-versus-host disease (GVHD) and/or a graft-versus-leukemia (GVL) effect, whereas in the opposite direction, host T cells recognizing the donor can induce rejection (HVG). Accordingly, any possibility to anticipate the nature and strength of an anti-DP T-cell response is crucial in this context.

Based on the HLA-DP recognition pattern of several HLA-DPB1*0901-specific T-cell clones, Crocchiolo et al classified HLA-DPB1 alleles according to their predicted “immunogenicity,”²  and using an algorithm deduced from this classification (Figure 1), they showed that the presence of “nonpermissive” HLA-DPB1 mismatches correlated with a significantly increased hazard of acute grade 2-IV GVHD. Unexpectedly, the authors reported that the increased risk of aGVHD was detectable independently of the predicted direction (GVH/GVL or HVG) of the T-cell response. To reconcile the statistical observation with the immunologic hypothesis (the increased risk of GVH when the algorithm predicted the recognition of donor HLA-DP by host T cells), these authors considered the possibility of an indirect pathway for GVH because of cytokine release by host T cells recognizing “immunogenic” HLA-DPB1 on donor antigen-presenting cells (APC).

In 4 successive studies,^3-6  we have in the past assessed the specificities of T-cell clones infiltrating skin biopsies during aGVHD (Table 1). In each situation, T-cell clones specific for host HLA-DP were isolated from the skin at the onset of aGVHD. These studies demonstrated that no mismatch could theoretically be considered as “permissive,” as confirmed in vitro by Rutten et al.⁷  The finding of up to a 10-fold difference observed between frequencies of T cells directed at a “permissive” versus “nonpermissive” DP mismatch, as argued by Sizzano et al,⁸  can hardly make a difference in the context of transplantation. T-cell frequency depends on the kinetics of the immune response and is a key physiologic factor in the race against an infection. In the present context of transplantation, the immune target (HLA-DP on host APC) remains present for weeks after the graft, and a 10-fold difference in frequency represents only 3-4 divisions for a T cell, which would take 1-2 days at most.

Nevertheless, a detailed analysis of anti-DP specificities remains particularly interesting and potentially useful (for example, to drive a GVL effect with DP-specific T-cell clones directed against an HLA-DP mismatch in the GVH direction, as we have previously proposed^3,4,9 ). In line with HSCT, considering that “alloreactive TCR neither avoid contacting the bound peptide nor focus on the polymorphic residues that are exposed on the outer surface of the allo-MHC a-helices,”¹⁰  it would be of great interest to learn more about the set of endogenous peptides presented by HLA-DP alleles in hematopoietic and nonhematopoietic tissues. This may help to improve the targeting of anti-DP allogeneic reaction against the residual disease while avoiding the healthy tissues as much as possible.