Selection of Donors with Favorable KIR B Genotypes for Unrelated Hematopoietic Cell Transplantation Results in Superior Relapse Protection and Better Relapse-Free Survival for Patients with AML.

Abstract

Abstract 665Unrelated donor (URD) transplants from donors with KIR group B/x genotypes (vs. A/A) confer a significant relapse-free survival (RFS) benefit for patients with acute myeloid leukemia (AML) (RR: 0.70 [95% CI: 0.55-0.88]; P = .002; Blood 2009; 113[3]). This new analysis was designed to investigate the beneficial effect of KIR B donors and to develop a donor selection strategy to improve clinical outcomes after hematopoietic cell transplantation (HCT) for leukemia. Based on an analysis of 27 unique KIR haplotype sequences we identified common centromeric and telomeric gene content motifs. KIR A haplotypes contain a Cen-A motif (defined by the presence of the inhibitory KIR gene 2DL3) and a Tel-A motif (defined by the presence of the activating gene 2DS4). The B haplotypes were defined as containing Cen-B (presence of 2DS2 and/or 2DL2) and/or Tel-B (presence of 2DS1), with further subdivisions possible at the allelic level. Thus, based on gene content alone, donor KIR genotypes can be classified as homogyzous A/A or defined by the type (Cen-B or Tel-B) or number (0, 1, 2 or ≥3; B domain content score) of B domains. Multivariate models were used to evaluate the effect of donor KIR genotypes on clinical outcomes after URD transplants facilitated by the National Marrow Donor Program for AML (n=1086) and acute lymphoblastic leukemia (ALL: n=334) between 1988 and 2006. The improved RFS associated with donor KIR B genotypes (Cen-A/B, Cen-B/B, Tel-A/B or Tel-B/B) vs. KIR A genotypes (Cen-A/A or Tel-A/A) in AML was most evident in the 115 donors (10.6%) who were homogyzous for the Cen-B motif (RR 0.72 [95% CI 0.55-0.94]; p=0.014). Likewise, Cen-B/B donors conferred significant protection against relapse (RR 0.34 [95% CI 0.2-0.57]; p < 0.0001); with absolute relapse rates of only 10% (Cen-B/B) vs. ∼31% (A/A). Similarly, multivariate models demonstrated that compared to KIR A/A donors, donors with higher KIR B domain content scores resulted in improved RFS (2B motifs: (RR 0.78 [95% CI 0.63-0.95]; p=0.013; n=244) or ≥3 B motifs: (RR 0.76 [95% CI 0.57-1.02]; p=0.07; n=84) and less relapse (2B motifs: (RR 0.54 [95% CI 0.39-0.74]; p=0.0001) or ≥3 B motifs: (RR 0.45 [95% CI 0.27-0.74]; p=0.0017). Donor KIR genotype had no effect on rates of graft vs. host disease or treatment related mortality. Importantly, the use of KIR B donors of any type was not associated with any improvement in clinical outcomes for patients with ALL. These data suggest that AML blasts may be particularly sensitive to killing by NK cells and raises the possibility that activating genes present in the donor KIR B haplotypes may uniquely recognize ligands on AML blasts. The KIR B genotype effects were not affected by the degree of HLA matching. Therefore, these data support the consideration of KIR genotyping with HLA typing into the unrelated donor search criteria for patients with AML. To capture the benefit of Cen-B and/or higher B domain content scores we propose that the ∼30% of donors who have ≥2 B domains (includes all Cen-B/B donors) be given preference over donors with 0 or 1 KIR B domains. In this large retrospective dataset, assignment by that rule resulted in significant improvements in RFS (RR 0.80 [95% CI 0.68-0.94]; p=0.0063) and relapse (RR 0.53 [95% CI 0.41-0.69]; p<0.0001). A prospective trial to test this strategy is planned. Disclosures:No relevant conflicts of interest to declare.