Impact of Donor KIR and HLA Genotypes on Clinical Outcomes According to Pre-Transplant Remission Status after HLA-Haploidentical Transplantation with Post-Transplantation Cyclophosphamide

INTRODUCTION

HLA-haploidentical allogeneic hematopoietic cell transplantation (allo-HCT) using post-transplantation cyclophosphamide (PT/Cy-haplo) can be a standard of care in patients suffering from poor prognostic hematological malignancies without conventional donors. Regarding killer cell immunoglobulin-like receptor (KIR) and HLA information on optimal donor selection in PT/Cy-haplo settings, the following factors associated with improved survival have been reported: HLA-DR/HLA-DP mismatch, KIR receptor-ligand mismatch, KIR B/x haplotype with KIR2DS2, and inhibitory KIR gene mismatch (Willem 2019, Solomon 2018, Symons 2010). However, the results were still inconclusive. In addition, it remains unknown whether the graft-versus-leukemia/tumor (GVL) effect of donor KIRs or HLAs is modified by residual tumor burden at PT/Cy-haplo.

METHODS

We retrospectively examined consecutive patients who received PT/Cy-haplo at our institution between June 2009 and December 2018. In both patients and donors, 16 KIR genes were genotyped using KIR SSO Genotyping Test (One Lambda, Inc.) and HLA allele typing was performed at HLA-A, -B, -C, and -DRB1. Cumulative incidence of relapse (CIR) was estimated using a cumulative incidence curve with nonrelapse mortality as a competing risk and compared using the Gray's test.

RESULTS

A total of 91 patients with available KIR typing data were eligible. Of these, HLA typing data were unavailable for 5 patients (5.5%). In this cohort, 76 HLA-C mismatched transplants (88%) were included, and the frequencies of donor KIR ligand were 78 (90.7%) in C1/C1, 8 (9.3%) in C1/C2, and 0 in C2/C2. The median age was 48 years (range, 17 - 68 years). Median follow-up time among survivors was 1,271 days (range, 242 - 3,135 days) after PT/Cy-haplo. This study included 54 AML, 6 MDS, 2 CML, 13 ALL, and 16 NHL patients. Thirty-four patients (37%) showed complete remission (CR) at PT/Cy-haplo, and 37 on the secondor third transplant (40.7%).

In CR population at PT/Cy-haplo, the patients who underwent PT/Cy-haplo from a KIR2DS1-positive donor had significantly lower rates of CIR than those from a KIR2DS1-negative donor (2-year CIR, 9.2% vs 42%; P = 0.037; Figure 1A). Due to unavailability of cases, we were unable to perform the subgroup analysis based on donor C1 or C2 status. In PT/Cy-haplo from a KIR3DS1- or KIR2DL5-positive donor, similar results were obtained, most likely due to genetic linkage disequilibrium among these genes. No other donor KIR genes were associated with CIR. Furthermore, PT/Cy-haplo from a KIR2DS1-positive donor was significantly associated with improved OS (2-year OS, 83% vs 34%; P = 0.01; Figure 1C). Also, PT/Cy-haplo from a B/x donor significantly increased OS (2-year OS, 77% vs 35%; P = 0.019), but did not decrease CIR (2-year CIR, 16% vs 40%; P = 0.122). These results suggested that donor KIR2DS1 could have a more crucial role in prevention of leukemia relapse than donor B/x haplotype.

In non-complete remission (NCR) population at PT/Cy-haplo, however, PT/Cy-haplo from a KIR2DS1-positive donor and a B/x donor did not significantly improve CIR (Figure 1B) or OS (Figure 1D).

Although we investigated the following previously reported models: KIR mismatch with ligand incompatibility model, receptor-ligand model, missing ligand model, inhibitory KIR gene model, and HLA-DRB1 disparity of graft-versus-host direction, none was found to be associated with significantly improved CIR or OS.

CONCLUSION

We found that in PT/Cy-haplo settings, with donor-recipient HLA-C mismatch in almost all cases, a KIR2DS1-positive donor significantly contributed to decreased CIR and increased OS in CR population at PT/Cy-haplo, but not in NCR population.　These results were consistent with the 2012 NEJM data by Venstrom et al in patients undergoing allo-HCT from HLA-matched or one-allele mismatched unrelated donors. Although the exact mechanism remains unclear, activating KIR2DS1, known as a player in the activation and tolerance of NK cells, could mediate NK-cell function and enhance GVL effect through NK alloreactivity in low tumor burden status at PT/Cy-haplo also in the PT/Cy-haplo setting. Our results may contribute to the establishment of an optimal donor selection algorithm. In future, elucidating the detailed mechanism of our findings could lead to the development of a novel preventive or therapeutic strategy for leukemia relapse.

View largeDownload slide

View largeDownload slide

Close modal