Is There Any Effect of Killer Cell Immunoglobulin-like Receptor (KIR) on Outcomes after Single Unrelated Cord Blood Transplantation?

The effect of killer cell immunoglobulin-like receptor (KIR) ligand incompatibility on outcomes after unrelated cord blood transplantation (UCBT) has been controversial. Eurocord found that KIR ligand mismatching was associated with decreased relapse incidence (RI) and improved overall (OS) and leukemia-free survival (LFS) for patients with acute lymphoblastic leukemia (ALL) and for those with acute myeloid leukemia (AML). Recently, the Japanese registry found no association between KIR ligand matching and LFS or OS in 643 UCBT recipients with acute leukemia. However, both studies have analysed the KIR ligand matching effect using low resolution typing of HLA-A,-B,-C and HLA-DRB1 high resolution.

With the aim to clarify the KIR effect on outcomes (mainly RI and OS) in a larger series of single-UCBT recipients in the era of HLA-allele typing, we have analysed 1098 patients with AML and ALL reported to Eurocord and CIBMTR. All patients received single UCBT and myeloablative conditioning regimen. HLA matching was defined using high resolution typing or imputation for HLA-A,-B,-C and DRB1. Patients and donors were categorized by their KIR-ligand expression for HLA C group 1 or 2 and Bw4 as KIR ligand matched or mismatched. Patients included in the earlier Eurocord analysis were excluded. Univariate and multivariate models were built to analyse the effect of KIR ligand matching on outcomes. Results: None of the 8/8 HLA-matched transplants were KIR ligand mismatched, therefore they were excluded. Since HLA-match and KIR ligand mismatch were confounded, we conducted 2 separate analyses: a) 6-7/8 HLA-matched (n=501) and b) 3-5/8 HLA-matched transplants (n=586). In the group of 6-7/8 HLA matched, 291 recipients (58%) were KIR ligand-matched and 210 (42%) were mismatched. There were no statistically significant differences, between these two groups for gender, age, CMV serostatus, type and remission disease status, cell dose, conditioning regimen, in vivo T-cell depletion, transplant period and follow-up (around 40 months). In the group of 3-5/8 HLA matched, 176 recipients (30%) were KIR ligand-matched and 410 (70%) were mismatched. In this HLA group (i.e 3-5/8), KIR ligand-matched patients were younger (<16y, p=0.02) compared to the KIR ligand-mismatched patients. Other characteristics were similar between the KIR ligand matched and mismatched groups. In multivariate models for outcomes in both HLA groups (6-7/8 HLA and 3-5/8 HLA), KIR-match status was not associated with non-relapse mortality (NRM), RI, LFS and OS. However, KIR ligand-mismatch was associated with lower risks of grade 2-4 acute GVHD for transplants that were 3-5/8, HLA-matched (Hazard ratio, HR= 0.63, p=0.001) but not for transplants that were 6-7/8 HLA-matched. KIR ligand-matching was not associated with chronic GVHD. The effect of KIR ligand-mismatching in the GVH direction, HVG direction and bi-directional were also examined and findings were consistent with the main analysis. Lower acute GVHD rates in the worst HLA-match group (3-5/8) with KIR ligand mismatch did not translate into a survival advantage when compared to similarly HLA-matched KIR ligand matched group. Disease-specific analysis were undertaken to further explore the effects of KIR ligand matching for ALL and AML separately. For patients with ALL, KIR match status was not associated with NRM, RI, LFS or OS in either HLA group. For patients with AML KIR ligand match status was not associated with any transplantation outcome after 6-7/8 HLA-matched transplants. However, KIR ligand mismatching was associated with higher NRM (HR1.94, p=0.02) and lower OS (HR 1.51, p=0.03) after 3-5/8 HLA-matched transplants. In conclusion, in the setting of 6-7/8 HLA-matched transplants KIR ligand match status was not associated with leukemia recurrence or survival. Our observation for AML, in the 3-5/8 HLA-matched group with KIR ligand mismatching is contradictory to the earlier Eurocord report. We hypothesize the observed differences may be attributable to better HLA-matching in the current analysis. Taken together, the data does not support selecting units based on KIR ligand match status as categorized in this analysis, when donor-recipient HLA-match considers allele-level HLA typing. It remains to be seen whether assignment of KIR ligand match status by genotyping will offer additional information.