KIR reconstitution is altered by T cells in the graft and correlates with clinical outcomes after unrelated donor transplantation

Abstract

Although unrelated hematopoietic cell transplantation (HCT) is curative for many hematologic malignancies, complications and relapse remain challenging obstacles. Natural killer (NK) cells, which recover quickly after transplantation, produce cytokines and express killer immunoglobulin-like receptors (KIRs) that regulate their cytotoxicity. Some clinical trials based on a KIR ligand mismatch strategy are associated with less relapse and increased survival, but results are mixed. We hypothesized that T cells in the graft may affect NK cell function and KIR expression after unrelated transplantation and that these differences correlate with clinical outcomes. NK cell function was evaluated using 77 paired samples from the National Marrow Donor Program Research Repository. Recipient NK cells at 100 days after both unmanipulated bone marrow (UBM) and T-cell depleted (TCD) transplants were compared with NK cells from their healthy donors. NK cells expressed fewer KIRs and produced more interferon gamma (IFN-gamma) after UBM compared to TCD transplants. Multivariate models showed that increased NK cell IFN-gamma production correlated with more acute graft-versus-host disease (GVHD), and decreased KIR expression correlated with inferior survival. These results support the notion that T cells in the graft affect NK cell reconstitution in vivo. Understanding these mechanisms may result in strategies to improve clinical outcomes from unrelated HCT.

Figure 1.

Recipients of UBM transplants exhibit lower KIR expression compared to their donors than recipients of TCD transplants. Donor or day +100, donor-derived recipient PBMCs were stained with a cocktail of directly conjugated mAbs to KIR (DX9, EB6, GL183, and FES173, which stain CD158e, CD158a, CD158b, and CD158i, respectively). The means, 25th to 75th percentile ranges, and extreme values of the percentages of CD56⁺/CD3^- cells expressing KIR by flow cytometry are shown in the box plot. Results are presented for donors or recipients separated by graft type (UBM or TCD) or combined (ALL). P = not significant (NS) between any donor groups.

Figure 2.

UBM recipients express fewer KIRs that recognize HLA-C ligands compared to KIRs that recognize Bw4. Aliquots of donor or day +100, donor-derived recipient PBMCs were stained with 3 directly conjugated mAbs to designated KIR (DX9, KIR3DL1; EB6, KIR2DL1/2DS1; GL183, KIR2DL2/2DL3/2DS2). The means, 25th to 75th percentile ranges, and extreme values of the percentages of CD56⁺/CD3^- cells expressing KIR by flow cytometry are shown in the box plots. Results are presented for donors or recipients separated by graft type (UBM: n = 33 donor, n = 38 recipient; TCD: n = 34 donor, n = 37 recipient). P = NS between any donor groups.

Figure 3.

NKG2A expression is increased in recipients of unrelated transplants, irrespective of graft manipulation, and is inversely correlated with KIR expression. Donor or day +100, donor-derived recipient PBMCs from UBM or TCD transplants were labeled with the directly conjugated mAb Z199, which stains NKG2A. The means, 25th to 75th percentile ranges, and extreme values of the percentages of CD56⁺/CD3^- cells expressing NKG2A by flow cytometry are shown in the box plot on the left. Results are presented for donors (n = 71) or recipients (n = 70). The same cells were stained with a cocktail of directly conjugated mAbs KIR (DX9, EB6, GL183, and FES173, which stain CD158e, CD158a, CD158b, and CD158i, respectively). The percentage of NKG2A recovery was calculated as a ratio of the percent expression on recipient NK cells over that on donor NK cells (R/D) and plotted on the x-axis versus the R/D ratio of total KIR recovery (right). There was no difference between recipients of UBM or TCD transplants, so data are presented for both transplant types.

Figure 4.

Recipients of UBM transplants have more IFN-γ-producing NK cells than recipients of TCD transplants. Donor or day +100, donor-derived recipient PBMCs from UBM (n = 24) or TCD (n = 25) transplants were incubated for 24 hours in IL-12 (10 ng/mL) and IL-18 (100 ng/mL). Harvested cells were permeabilized and stained intracellularly with a directly conjugated mAb to IFN-γ. The means, 25th to 75th percentile ranges, and extreme values of the percentages of CD56⁺/CD3^- cells expressing IFN-γ by flow cytometry are shown in the box plot. Results are presented for donors or recipients.

Figure 5.

Incidence of acute GVHD correlates with increased IFN-γ production by donor-derived NK cells. Cumulative incidence estimates of acute GVHD in recipients of UBM or TCD transplants are shown. The percentage of day +100, donor-derived recipient CD3^-/CD56⁺ NK cells producing IFN-γ was analyzed as described, and the results were analyzed with a Cox multivariate proportionate hazards model that included KIR expression, percentage of CD56^+bright cells, graft type, IFN-γ production, and the incidence of acute GVHD. The recipients were further stratified by IFN-γ production by equal to or more than 50% or less than 50% of NK cells, which was the mean percentage of NK cells producing IFN-γ for all samples (left panel). The type of graft (UBM or TCD) did not independently correlate with acute GVHD but did accentuate the effect of IFN-γ expression (right panel).

Figure 6.

Survival correlates with increased KIR expression on donor-derived NK cells after transplantation. Kaplan-Meier survival estimates in recipients of UBM or TCD transplants are shown below. The ratios of day +100, donor-derived recipient CD3^-/CD56⁺ NK cells to donor cells expressing KIR were analyzed as described. The results were analyzed with a Cox multivariate proportionate hazards model that included KIR expression, percentage of CD56^+bright cells, IFN-γ production, and survival (n = 72). The recipients were further stratified by KIR expression ratio of equal to or more than 0.6 or less than 0.6, which was the mean R/D ratio of all samples (left panel). Stratification based on graft type (UBM, n = 35; TCD, n = 37) accentuates the correlation of KIR expression with survival but was not an independent predictor (right panel).