Donor NK Cell Education By KIR3DL1 and HLA-B Associates with Reduced Relapse in Unrelated Bone Marrow Transplantation for Adult T-Cell Leukemia

Background: Adult T-cell leukemia/lymphoma (ATL) derives from human T-cell leukemia virus-1 (HTLV-1)-infected T cells, and its prognosis is still dismal. Whereas allogeneic stem cell transplant (allo-HSCT) may lead to long-term remission, 1-year relapse rate is still high at 35-47% (Muranushi et al, Bone Marrow Transplant. 2021). As for NK cell modulation against ATL, polymorphisms of killer-cell immunoglobulin-like receptors (KIRs) are of interest. NK cell inhibition by donor KIR/patient HLA interaction and NK cell education status have been reported to affect the prognosis of allo-HSCT, but no consensus has been reached. Furthermore, the effects of entire KIR haplotypes have not been extensively studied.

Aim: Using high-resolution genotyping of all 15 functional KIR genes based on long-range PCR and long-read sequencer, correlations between KIR/HLA polymorphisms and relapses in unrelated bone marrow transplantation (UR-BMT) for ATL were evaluated.

Methods: DNAs and clinical data of 264 transplants performed between 2007 and 2016 were obtained. Alleles of 15 KIRs, HLA-A, B, C, DRB1, DPB1 and DQB1 were genotyped for patients and donors. Prognostic factors for relapse were evaluated using the Fine-Gray proportional-hazards models.

Results: Donor KIR3DL1 and patient HLA-B combination predictive of strong interaction (n=74) showed a trend for a lower relapse rate than weak or non-interactive combinations (n=145) (HR, 0.64; p=0.100), raising the beneficial impact of donor NK cell education at KIR3DL1/HLA-B interaction. Comprehensive analysis focusing on amino acids of donors' KIRs/HLAs specified combinations of HLA-B isoleucine 80 (80I)/ KIR3DL2 aspartic acid 42 and HLA-B threonine 80 (80T)/ KIR3DL2 glutamic acid 42 (n=76) associated with decreased relapse (HR, 0.35; 95%CI, 0.19-0.65; p=0.00087). On the other hand, combinations of HLA-B glutamic acid 152 (152E)/ KIR2DL4 cysteine 30 and HLA-B valine 152 (152V)/ KIR2DL4 tyrosine 30 (n=27) were associated with increased relapse (HR, 11.79; 95%CI, 1.6-86.58; p=0.015). Given that KIR3DL2/ KIR2DL4 do not bind HLA-B and are closely linked to KIR3DL1/S1 in KIR haplotypes, specific KIR3DL1/S1 alleles were further explored by subgroup analyses. As a result, combinations of HLA-B 80I/ KIR3DL1*015& KIR3DL2*002 and HLA-B 80T/ KIR3DL1∗001, ∗005, ∗007, ∗020, ∗038, ∗097 (n=60) were defined as the group at low risk for relapse (adjusted HR, 0.21; 95%CI, 0.09-0.48; p=0.00026) (Figure 1A), and combinations of HLA-B 152E/ KIR3DS1 and HLA-B 152V/ KIR3DL1∗001, ∗007, ∗015, ∗020, ∗029, ∗038 (n=176) were defined as the group at high-risk for relapse (adjusted HR, 5.31; 95%CI, 1.72-16.39; p=0.0037) (Figure 1B). The cases in the low-risk group and/or not in the high-risk group (n=84) had a lower relapse rate (adjusted HR, 0.20; 95%CI, 0.10-0.40; p=0.0000059) (Figure 1C) and a higher overall survival (adjusted HR, 0.66; 95%CI, 0.46-0.95; p=0.0265389) (Figure 1D) than the remaining cases (n=130). In spite of reduced relapse rate, the former showed a lower rate of grades II-IV acute GVHD (HR 0.65; p=0.049). Finally, CD107a degranulation of NK cells was measured following coincubation of HLA-null K562 cells and PBMCs from healthy donors with or without genotypes of the low-risk group. KIR3DL1(+) NK cells showed increased CD107a degranulation compared with KIR3DL1(−) NK cells in healthy donors with the low-risk group genotype (p=0.031), but this difference was not apparent in healthy donors without the low-risk group genotype (p=0.28) (Figure 2).

Conclusions: Allelic genotyping of KIRs enabled to specify donors associated with reduced relapses. Donor NK cell education status at HLA-B/KIR3DL1 interaction might confer enhanced GVT effects and better prognosis in UR-BMT against ATL.