NK cell recovery after haploidentical HSCT with posttransplant cyclophosphamide: dynamics and clinical implications

Abstract

The use of posttransplant cyclophosphamide (PT-Cy) as graft-versus-host disease (GVHD) prophylaxis has revolutionized haploidentical hematopoietic stem cell transplantation (HSCT), allowing safe infusion of unmanipulated T cell-replete grafts. PT-Cy selectively eliminates proliferating alloreactive T cells, but whether and how it affects natural killer (NK) cells and their alloreactivity is largely unknown. Here we characterized NK cell dynamics in 17 patients who received unmanipulated haploidentical grafts, containing high numbers of mature NK cells, according to PT-Cy-based protocols in 2 independent centers. In both series, we documented robust proliferation of donor-derived NK cells immediately after HSCT. After infusion of Cy, a marked reduction of proliferating NK cells was evident, suggesting selective purging of dividing cells. Supporting this hypothesis, proliferating NK cells did not express aldehyde dehydrogenase and were killed by Cy in vitro. After ablation of mature NK cells, starting from day 15 after HSCT and favored by the high levels of interleukin-15 present in patients' sera, immature NK cells (CD62L⁺NKG2A⁺KIR^-) became highly prevalent, possibly directly stemming from infused hematopoietic stem cells. Importantly, also putatively alloreactive single KIR⁺ NK cells were eliminated by PT-Cy and were thus decreased in numbers and antileukemic potential at day 30 after HSCT. As a consequence, in an extended series of 99 haplo-HSCT with PT-Cy, we found no significant difference in progression-free survival between patients with or without predicted NK alloreactivity (42% vs 52% at 1 year, P = NS). Our data suggest that the majority of mature NK cells infused with unmanipulated grafts are lost upon PT-Cy administration, blunting NK cell alloreactivity in this transplantation setting.

Figure 1.

NK cell counts and chimerism early after haploidentical HSCT and PT-Cy. (A) Outline of the haploidentical HSCT platforms used at OSR (left) and at JHU (right). (B) Absolute counts of NK cells (red circles) and T cells (blue squares) circulating in the PB of patients (n = 17) receiving haploidentical HSCT followed by PT-Cy (days of Cy administration are shaded in yellow). Data are displayed as mean values ± standard error of the mean (SEM). (C) Absolute counts of PB NK cells in patients treated in the 2 transplantation centers (OSR, red circles: n = 10; JHU, white circles: n = 7). Data are displayed as mean values with SEM. (D) Chimerism within T- and NK-cell compartment after HSCT. HLA-A*02 mismatches allowed to discriminate by multiparametric flow cytometry donor-derived cells (colored) and residual host lymphocytes (white). Pies depict NK cell (upper line) and T cell (bottom lines) chimerism measured at different time points after HSCT in 3 donor-recipient pairs. Numbers indicate mean values with standard deviation (SD). MMF, mycophenolate mofetil; PBSC, peripheral blood cell transplantation; TBI, total body irradiation.

Figure 2.

Cyclophosphamide administration induces killing of proliferating NK cells. (A) Flow cytometry histograms depicting Ki-67 expression measured on T (blue) and NK cells (red) from the leukapheresis (LP) graft and in the PB of a representative patient (OSR #1) at different time points after HSCT, as indicated. Percentages indicate frequencies of Ki-67–positive cells. (B) Ki-67 positivity in NK (red circles) and T (blue squares) cells from the graft (day 0), or circulating in the PB of patients after HSCT with PT-Cy (n = 17). (C) Ki-67 positivity in NK cells from the graft (day 0) or in the PB of patient transplanted at OSR (red circles: n = 10) or JHU (white circles: n = 7). (D) Level of proliferation of donor-derived NK cells (red circles) and residual host NK cells (gray circles) measured with Ki-67 intracellular staining in 3 representative patients. Donor- or host-derived NK cells were discriminated by immunophenotypic analysis using differential expression of the mismatched HLA-A*02 allele. (E) Scatter plot depicting the mean percentage of ALDH⁺ cells detected by flow cytometry among NK (red circles), T (blue squares), and stem (orange diamonds) cells present within the infused graft, or in the patient PB 3 days after transplant, in 4 representative patients. Note that for 2 of 4 patients, CD34⁺ stem cells were no more detectable in the patient PB at day 3. (F) In vitro assay of mafosfamide-induced cell death. The LP product of 6 patients was stimulated with IL-15 at day 0 and treated with different doses of mafosfamide at day 3. The graphs report the number of viable (AnnexinV⁻) NK cells, T cells, and HSCs detected by flow cytometry before the treatment (day 3) and after administration of different doses of the drug (day 5; light pink bars: untreated; dark pink bars: 2.5 μg/mL; light red bars: 7.5 μg/mL) and after irradiation as positive control (day 5, red bars). Unless otherwise specified, shown in all panels are average values ± SEM.

Figure 3.

IL-15 serum levels sharply increase after HSCT and cyclophosphamide administration and correlate with NK cell dynamics. (A) IL-15 concentration (green triangles) measured in the sera of 7 patients before and after HSCT with PT-Cy is shown with the number of NK cells (red circles) and T cells (blue squares) in the PB of the same patients. Mean values with SEM are shown. (B) Correlation between IL-15 serum concentration at 3 days after HSCT (x-axis) and NK cell counts at day 15 after HSCT (y-axis). The red line denotes the best-fit line of the linear regression analysis. (C) Correlation between IL-15 serum concentration at 3 days after HSCT (x-axis) and the percentage of NK cells proliferating at the same time point, measured by Ki-67 intracellular staining (y-axis). The red line denotes the best-fit line of the linear regression analysis.

Figure 4.

The second wave of NK cells that appear after PT-Cy displays an immature phenotype. (A) Flow cytometry histograms depicting expression of indicated maturation markers on CD56⁺ CD3⁻ NK cells from a representative patient (OSR #1) at the indicated time points (LP: red; day 3: orange; day 15: green; day 30: blue). (B) The expression of the described maturation markers measured by flow cytometry in NK cells from the graft (day 0) and from patients longitudinally sampled after HSCT with PT-Cy (n = 10). For each marker, a normal reference interval (mean ± SD) measured in NK cells from 5 healthy subjects is displayed (gray box). (C) Heat map of the average NK cell maturation marker expression at different time points, with unsupervised hierarchical clustering, for 10 OSR patients analyzed by multiparametric flow cytometry at the indicated time points. (D-E) Multidimensional single-cell analysis of the maturation status of NK cells harvested from 5 healthy controls (HC), from leukaphereses (LP, n = 10) and from 10 patients at different time points after HSCT. The panel D bidimensional maps were obtained from flow cytometric data using the bh-SNE algorithm show all analyzed NK cell events, with coloring denoting the expression of each maturation marker, as indicated. The same data are depicted in panel E, where NK cells from a selected time point are displayed separately.

Figure 5.

Despite early recovery of NK cell counts after PT-Cy–based HSCT, NK cell phenotype normalization occurs only several months after transplantation. (A) Absolute counts of NK cells (red circles) and T cells (blue squares) detected in the PB of 10 patients followed long-term after HSCT with PT-Cy (from days 30 to 360). A reference physiological cell count interval (mean ± SD) obtained from 5 healthy controls is shown for NK cells (light red box) and T cells (light blue box). (B) The proportion of NK cells out of total lymphocytes in the infused donor graft (day 0) and longitudinally after HSCT is displayed for 10 OSR patients. (C-F) Flow cytometry histograms depicting the expression of selected markers on CD56⁺CD3⁻ NK cells in the donor graft (day 0) or from patient PB obtained longitudinally after transplant in 10 OSR patients. For all panels, gray boxes show reference values (mean ± SD) obtained analyzing 5 healthy controls. (C) Expression of maturation markers. (D) Expression of KIRs and lectin-type receptors. (E) Expression of markers of exhaustion or activation. (F) Expression of natural cytotoxicity receptors. Unless otherwise specified, data are shown as mean ± SEM.

Figure 6.

PT-Cy eliminates single-KIR⁺NK cells and thus dampens NK cell–mediated alloreactivity. (A) Flow cytometry histogram depicting proliferation measured by Ki-67 expression in total NK cells (red), in single KIR⁺ NK cells predicted to be alloreactive (green), and in single-KIR⁺ NK cells predicted to be nonalloreactive (purple) in a representative patient (OSR #10) immediately before PT-Cy administration (day 3 after HSCT). (B) Time course of Ki-67 positivity in total NK cells (red circles), single-KIR⁺ NK cells predicted to be alloreactive (green triangles), or single-KIR⁺ NK cells predicted not to be alloreactive (purple triangles) from the graft (day 0) or circulating in the PB from a representative patient (OSR #10) after HSCT. (C) Percentage of Ki-67 positivity in total NK cells from 3 PBSC grafts (red), in the subset of single-KIR⁺ NK cells predicted to be alloreactive (green), or in the subset of single-KIR⁺ NK cells predicted not to be alloreactive (purple) upon 3 days of exposure to IL-15, and after subsequent addition of mafosfamide to the culture medium. (D) Percentage of proliferating cells, measured through CTV dilution, among total NK cells from 3 PBSC grafts (red), among the subset of single-KIR⁺ NK cells predicted to be alloreactive (green), and among the subset of single-KIR⁺ NK cells predicted not to be alloreactive (purple) upon 3 days of exposure to IL-15 and after subsequent addition of mafosfamide to the culture medium. (E) Frequency of predictably alloreactive single-KIR⁺ NK cells within the graft and in PB NK cells in patients 30 days after HSCT, measured in 8 donor-recipient pairs with KIR-ligand mismatches. (F) Target cell death, expressed as Annexin V positivity (AnnV⁺), measured on K562 cells, OCI/AML cells, or primary leukemic cells after incubation at a 10:1 effector:target ratio with NK cells purified from patient PB day 30 after HSCT with PT-Cy (n = 8, red dots) or from their respective donors PB (n = 8, black dots). (G-H) Progression-free survival (G) and overall survival (H) in patients who received PT-Cy–based haploidentical HSCT from donor with (green line, n = 41) or without (purple line, n = 58) predicted NK cell alloreactivity. Tick marks represent censoring for live patients. Unless otherwise specified, data are shown in all panels as mean values ± SEM.

Figure 7.

Mature KIR⁺NK cells that are spared by PT-Cy can protect against posttransplantation disease relapse. (A) Absolute counts of circulating NK cells, (B) absolute counts of CD57⁺CD16⁺KIR⁺NKG2C⁺ memory-like NK cells, (C) percentage of CD62L⁺ NK cells, and (D) percentage of KIR⁺ NK cells were determined in samples collected at day 30 after HSCT from 59 patients who received haploidentical HSCT followed by PT-Cy at OSR. For each of these parameters, dot plots to the left of the figure display the distribution in the patient cohort, discriminating between patients with values above (red dots and background) or below (blue dots and background) the median. Curves display the cumulative incidence of disease relapse (center panels) and PFS (right panels) in each subgroup. P values reported in each panel corner are relative to univariate comparisons performed using Gray's test (for cumulative incidence of relapse) or log-rank test (for PFS).