KIR Ligand Absence in Recipients of Unrelated Donor (URD) Allogeneic Hematopoietic Cell Transplantation (HCT) Is Associated with Less Relapse and Increased Graft Versus Host Disease (GVHD).

NK cell alloreactivity in the GVH direction, defined by donor expression of inhibitory KIR for which the recipient lacks the corresponding class I MHC KIR ligand, can alter the outcome of HCT. All humans express at least one KIR ligand (C1, C2 and Bw4), and most genotypes frequently include the corresponding inhibitory KIR (e.g. in Caucasoid populations KIR2DL2, 49–60% and KIR2DL3, 85–93% (C1), KIR2DL1, 91–100% (C2); and KIR3DL1, 87–98% (Bw4)). Consequently, most donor-derived NK cells express inhibitory KIR recognizing all three MHC KIR ligands. Therefore NK cell alloreactivity after HCT may be operationally defined by the number and type of KIR ligands a recipient lacks. To evaluate the clinical effects of KIR ligand absence we studied 2062 patients with myeloid malignancies (556 AML, 1224 CML and 282 MDS) receiving URD HCT through the NMDP between 1988 and 2000, all part of the high-resolution typing project. Presence or absence of the C1, C2 and Bw4 KIR ligands was determined for each recipient from high resolution HLA class I typing. The median age and performance status of patients expressing all three KIR ligands (n=719) were not significantly different from those missing at least 1 KIR ligand (n= 1343) (p=0.82 and p=0.71, respectively). A preliminary multivariate analysis of all patients showed no differences in the risk of relapse based on KIR ligand content. Significant differences emerged using stratification by disease stage. For patients with early disease (AML in CR1, 1^st chronic phase(CP) CML <1 year from diagnosis, or MDS with refractory anemia) the absence of ≥1 KIR ligand was protective against relapse (RR 0.54 [0.30–0.95], n=534, P=0.03), independent of HLA-matching and consistent with reports of potentially beneficial NK cell mediated anti-tumor effects following HCT from HLA-matched sibling donors. In these early disease patients, 3 year relapse rates were 11% [CI: 7–17] with all KIR ligands present vs. 6% [CI: 4–9] with ≥1 ligand absent. In contrast, KIR ligand absence did not affect clinical outcomes for patients with more advanced myeloid disease. NK cells may also play an indirect role in GVHD through their effects on dendritic cells. Therefore we analyzed the impact on acute GVHD. HLA-matching was protective against acute GVHD, and generally, missing ≥1 KIR ligand had no effect on acute GVHD. Unexpectedly, in patients with late CP CML > 1 year from diagnosis missing ≥1 ligand was associated with more frequent grade 3–4 acute GVHD (RR 1.6 [1.1–2.2], n=481, P=0.008): 30% [CI: 23–37] with all KIR ligands present vs. 44% [CI: 39–50%] missing ≥1 KIR ligand. We noted more frequent pre-HCT interferon use late vs. early CP CML (63% vs. 41%, P=<0.001) but prior interferon use did not independently affect GVHD. These data support the premise that NK cells have a genetically determined potential to affect outcomes following unrelated HCT in myeloid leukemia. A better understanding of how to manipulate KIR repertoires after HCT, by attenuation of inhibitory signals in patients who express all KIR ligands, for example, may allow extension of this clinical benefit to limit GVHD and relapse.