HLA matching between the donor and recipient improves the success of unrelated hematopoietic stem cell transplantation (HSCT). Because many patients in need of an unrelated transplant have only donors with mismatch, information is needed to evaluate the limits of HLA mismatching. We examined the association of survival, acute graft-versus-host disease (aGVHD) and relapse with HLA-A, -B, -C, -DRB, -DQB1, and -DPB1 mismatching in 334 patients coming from 12 French transplant centers and who received a non-T cell-depleted bone marrow graft from an unrelated donor. All patients were prepared with the use of myeloablative conditioning regimens. Our analyses demonstrate negative effects of HLA mismatching for either HLA-A, -B, -C, -DRB1, or -DQB1 loci on survival. Multivariate Cox analyses showed that a single mismatch was associated with a significant decrement in survival (P=.046, hazard ratio [HR]=1.41, confidence interval [CI] 95% 1.1-1.98). The presence of multiple mismatches was worse for survival (P=.003, HR=1.91, CI 95% 1.26-2.91) and severe aGVHD (grade III-IV) (P=.002, HR=2.51, CI95% 1.41-4.46). The cumulative incidences of aGVHD and relapse in those HLA-A, -B, -C, -DRB1, and -DQB1 identical pairs with 2, 1, or 0 DPB1 incompatibilities were 63%, 50%, and 51%, and 12%, 27%, and 20%, respectively, but these differences were not statistically significant. Similar differences of aGVHD and relapse, but not statistically significant, were observed in those HLA-A, -B, -C, -DRB1, and -DQB1 identical pairs with DPB1 disparities classified into permissive or nonpermissive mismatches according to Zino's classification based on a hierarchy of the immunogenicity of the HLA-DP molecules. "Missing killer cell immunoglobulin-like receptor (KIR) ligand" evaluated on the presence of HLA-C1, -C2, and Bw4 groups in the recipients was not associated with aGVHD, survival, and relapse in this cohort of non-T cell-depleted HSCT.

Polymerase chain reaction sequence-specific oligonucleotide is commonly used for HLA-typing. We replaced our LabType SSO HD (HD) kits with LabType SSO XR (XR) kits (One Lambda, Inc., Canoga Park, California) for HLA-A, -B, and -DRB1 following acquisition of a LABScan3D analyzer. The XR kits have more bead regions than the HD kits, allowing for an extended number of probes and exon coverage. They are claimed to improve typing resolution and to diminish the number of allele ambiguities, including common and well-documented (CWD) and null alleles to be resolved. We retrospectively selected patients who had their first HLA-typing performed with the HD kits and their second determination with the XR kits between 2015 and 2016. Forty-two patients were selected for HLA-A typing comparison, and 48 for HLA-B and 41 for HLA-DRB1. XR kits significantly decreased the number of allele ambiguities for HLA-A and -B. On the other hand, the improvement was limited for the HLA-DRB1 locus. The XR kits did not resolve all the CWD HLA allele ambiguities, which may be important for organ and/or hematopoietic stem cell transplantations. The XR kits eliminated 88%, 62%, and 27% of null allele ambiguities for HLA-A, -B, and -DRB1 loci, respectively. In conclusion, the XR kits allow for a significant improvement of HLA-typing resolution for HLA-A and -B loci in comparison with HD kits. In contrast, the number of oligonucleotides in the XR HLA-DRB1 kit should be extended to include exon 3 at the very least. It could also be interesting to include oligonucleotides allowing HLA-DRB3, 4, and 5 typing.

No abstract

HLA mismatching and minimization of immunosuppression are two major risk factors for the development of de novo donor specific antibodies, which is associated with reduced kidney graft survival. Antibodies do not recognize whole HLA antigens but rather individual epitopes, which are short sequences of amino acids in accessible positions. However, compatibility is still assessed by the simple count of mismatched HLA antigens. We hypothesized that the number of mismatched epitopes, or ("epitope load") would identify patients at the highest risk of developing donor specific antibodies following minimization of immunosuppression. We determined epitope load in 89 clinical trial participants who converted from cyclosporine to everolimus 3 months after kidney transplantation. Twentynine participants (32.6%) developed de novo donor specific antibodies. Compared to the number of HLA mismatches, epitope load was more strongly associated with the development of donor specific antibodies. Participants with an epitope load greater than 27 had a 12-fold relative risk of developing donor-specific antibodies compared to those with an epitope load below that threshold. Using that threshold, epitope load would have missed only one participant who subsequently developed donor specific antibodies, compared to 8 missed cases based on a 6-antigen mismatch. DQ7 was the most frequent antigenic target of donor specific antibodies in our population, and some DQ7 epitopes appeared to be more frequently involved than others. Assessing epitope load before minimizing immunosuppression may be a more efficient tool to identify patients at the highest risk of allosensitization.

Given their antileukemic activity, natural killer (NK) cells can alter the outcome of hematopoietic stem cell transplantation (HSCT). The physiologic functions of NK cells are regulated by the interaction of killer immunoglobulin-like receptors (KIR) with specific HLA class I ligands. In the literature, different models based on HLA class I and/or KIR donor (D)/recipient (R) gene disparities are considered as predictors of NK cell alloreactivity. In this retrospective and multicentric French study, we analyzed the clinical impact of the different NK-alloreactivity models in 264 patients who underwent T repleted unrelated HSCT. First, we did not observe that the "KIR ligand-ligand" model had a significant clinical impact on unrelated HSCT outcome, whereas the "missing KIR ligand" model had a significant but limited effect on unrelated HSCT, because only the absence of C1 ligand in patients with myelogenous diseases was associated with a decreased overall survival (OS) (hazard ratio=2.17, P=.005). The "KIR receptor-receptor" and the "KIR receptor-ligand" models seemed the most capable of predicting NK alloreactivity because they had a significant impact on acute graft-versus-host disease (aGVHD) occurrence, OS, and relapse incidence in D/R unrelated pairs. In particular, KIR3DL1 gene mismatches in the GVH direction (D(+)R(-)) and the D KIR3DL1(+)/3DS1(+) and R Bw4(-) combination were respectively correlated with the lowest OS in HLA identical pairs (HR=1.99, P =.02) and the highest incidence of relapse in HLA nonidentical D/R unrelated pairs (HR=4.72, P =.03). Overall, our results suggest a detrimental effect of KIR3DL1(+)/3DS1(+) donor NK cells transplanted into HLA-Bw4(-) patients in the absence of an educational process via KIR3DL1/HLA-Bw4 interactions.