Acute Leukemias

Rapid natural killer cell recovery determines outcome after T-cell-depleted HLA-identical stem cell transplantation in patients with myeloid leukemias but not with acute lymphoblastic leukemia

Article metrics


Natural killer (NK) cells are the first lymphocytes to recover after allogeneic stem cell transplantation (SCT) and can exert powerful graft-versus-leukemia (GVL) effects determining transplant outcome. Conditions governing NK cell alloreactivity and the role of NK recovery in sibling SCT are not well defined. NK cells on day 30 post-transplant (NK30) were measured in 54 SCT recipients with leukemia and donor and recipient killer immunoglobulin-like receptor (KIR) genotype determined. In univariate analysis, donor KIR genes 2DL5A, 2DS1, 3DS1 (positive in 46%) and higher numbers of inhibitory donor KIR correlated with higher NK30 counts and were associated with improved transplant outcome. NK30 counts also correlated directly with the transplant CD34 cell dose and inversely with the CD3+ cell dose. In multivariate analysis, the NK30 emerged as the single independent determinant of transplant outcome. Patients with NK30 >150/μl had less relapse (HR 18.3, P=0.039), acute graft-versus-host disease (HR 3.2, P=0.03), non-relapse mortality (HR 10.7, P=0.028) and improved survival (HR 11.4, P=0.03). Results suggest that T cell-depleted SCT might be improved and the GVL effect enhanced by selecting donors with favorable KIR genotype, and by optimizing CD34 and CD3 doses.


Natural killer (NK) cells may affect the transplant outcome by exerting direct cytotoxicity against leukemic cells and by targeting host-derived antigen-presenting cells (APCs) diminishing both the risk of relapse and acute graft-versus-host disease (aGVHD).1, 2, 3, 4, 5, 6 It is believed that NK alloreactivity following allogeneic stem cell transplantation (SCT) is regulated by quantitative differences in activating and inhibitory signals (mediated by activating and inhibitory killer cell immunoglobulin-like-receptors (KIR)).7, 8 Although NK-mediated effects after SCT have largely been ascribed to KIR-MHC ligand mismatching in the context of haploidentical transplants (‘missing-self’ hypothesis),9 there is also widespread evidence in vitro and in vivo for NK-related effects in HLA-matched SCT.3, 10, 11, 12, 13, 14, 15 Thus, the conditions required for donor NK alloreactivity cannot be entirely explained by the missing KIR hypothesis. Several models have been proposed to explain NK alloreactivity in HLA-mismatched and -matched SCT.1, 2, 14, 15, 16, 17, 18, 19, 20

In a previous study,3 we showed that early NK cell recovery was the main predictor for molecular remission in 20 patients with chronic myeloid leukemia (CML) undergoing HLA-matched T-cell-depleted transplants. Here, we studied 34 additional patients with acute myelogenous leukemia (AML) and acute lymphoblastic leukemia (ALL) and supplemented our studies by KIR analyses providing a surrogate for NK cell alloreactivity. We hypothesized that the pace of NK cell recovery early post-SCT might itself be controlled by NK cell alloreactivity, and by the quantity of stem cells and competing T lymphocytes transplanted.

Patients and methods

Study group

From a cohort of 157 consecutive patients with leukemia receiving SCT between September 1993 and July 2005, post-transplant cryopreserved cell samples around day +30 were available in 54 transplant recipients (20 patients with lymphocyte subset analysis data have been previously reported3), and pre-transplant samples were available for KIR genotyping in 50 of these donor–recipient pairs. Nineteen patients had AML or myelodysplastic syndrome (MDS), which had progressed to AML, 15 had ALL and 20 had CML. Patients and their HLA-identical sibling donors were treated on National Institutes of Health protocols, approved by the National Heart, Lung and Blood Institute Review Board. All patients and donors provided written informed consent before enrollment.

Conditioning regimens and transplant approach

Patients received a T-cell-depleted HLA-identical sibling donor SCT with a total body irradiation (TBI)-based conditioning regimen and post-transplant cyclosporine as GVHD prophylaxis as described previously.3, 21 Details of stem cell source and dose, conditioning regimen and cyclosporine dosing are depicted in Table 1. Transplant protocols including post-transplant T-cell add back, monitoring of minimal residual disease, chimerism, infection prophylaxis and cytomegalovirus monitoring have been described previously.3, 21

Table 1 Patient and transplant characteristics

Flow cytometry and NK cell enumeration

NK cell fractions within the lymphocyte populations were determined by immunophenotyping of cryopreserved peripheral mononuclear cells. The absolute NK cell count was calculated from absolute lymphocyte counts from routine leukocyte differential counts. Cells were analyzed by five-color flow cytometry using a panel of monoclonal antibodies to CD3, CD4, CD8, CD56 and CD16 directly conjugated with fluorescein isothiocyanate (FITC), phycoerythrin (PE), allophycocyanin (APC), peridinin chlorophyll protein (PerCP) or APC-Cy7 (Becton Dickinson, San Jose, CA, USA). Flow cytometry analysis was performed on the LSR II flow cytometer (BD Biosciences, San Diego, CA, USA) using BD FacsDiva software (BD Biosciences, San Diego, CA, USA). A total of 50 000–1 000 000 cells were acquired. NK cells were defined as CD3−CD56+ cells and their respective fraction was calculated from the forward-sideward-scatter lymphocyte gate.

KIR genotyping

KIR genotype was performed by polymerase chain reaction using sequence-specific primers (PCR-SSP) using the KIR Genotyping Kit (Dynal Biotech, Pel-Freez Clinical Systems, Brown Deer, WI, USA) to detect the presence or absence of KIR genes (2DL1 001-006, 2DL2 001-005, 2DL3 001-006, 2DL4 00101/00102/0010301/0010302/00104/00201/00202/003- 005/00601/007/0080101/0080103/0080201/0080202/009-011, 2DL5A 001/005, 2DL5B 002-004, 2DS1 001-004, 2DS2 00101/00102/00103/002-005, 2DS3 00101-00103, 2DS4** 0010101/0010102/0010103/00102/002, 2DS4* 003/004/006, 2DS5 001-005, 3DL1 00101/00102/002/00401/00402/005-009/01501/01502/016-019, 3DL2 001-008/00901/010-013, 3DL3 001/00201/00202/003/00401, 3DS1 010-014) and two pseudogenes (2DP1 and 3DP1). The analysis of all distinct KIR alleles was performed at the low-resolution molecular level including alleles that are known not to be expressed to provide a complete genotypic analysis. All alleles were taken into consideration. KIR 2DL5A, 2DL5B, 2DS4* and 2DS** were analyzed separately. Genomic DNA was isolated from peripheral blood using standard methods and 0.02 μg DNA was used per PCR. Manufacturer's instructions were utilized in regard to master mix and volumes. The cycling parameters were as follows: 1 min at 95°C for one cycle followed by 30 cycles of 20 s at 94°C, 20 s at 63°C and 90 s at 72°C, 4°C hold. The final PCR product was visualized on a 1.5% agarose gel stained with ethidium bromide, and the result was determined by the presence or absence of an appropriately sized band.

Analysis of KIR data

Several models were used to describe NK alloreactivity as determined by KIR genotyping. (1) The missing KIR ligand model: since KIR and HLA genes segregate independently, HLA-identical siblings may inherit different KIR genes. The donor KIR may therefore not have a corresponding HLA ligand in the recipient. To identify whether a specific missing KIR ligand might have contributed to survival, acute GVHD and relapse, donor–recipient pairs were grouped according to specific missing ligands. These groups were defined respectively by the lack of recipient HLA-C group 1, group 2 and HLA Bw4 for a donor KIR. A group representing patients lacking two ligands (one or other HLA-C ligand in addition to HLA-Bw4) for donor KIR was also included in the analysis.14, 15, 17, 22 (2) The KIR genotype difference between recipients and donors was compared, and four different pairs were analyzed, as described by Gange et al.23 (3) As the ability of NK cells to detect HLA class I epitopes might be related to the number of KIR present in the donor, we examined the relationship between the number of total, inhibitory or activating KIR in the donor and transplant outcome as described by De Santis et al.16 (4) The effect of each individual KIR gene on transplant outcome was analyzed.


Patients with ALL and AML in first complete remission and patients with CML in chronic phase were considered to have standard risk (SR) disease. All other patients, including those who underwent transplantation in second or subsequent remission, those with primary refractory or relapsed disease, and those with secondary AML were defined as having high-risk (HR) disease. Overall survival (OS) was calculated from the interval between the date of transplantation and death, or the last follow-up visit. Relapsed disease for AML and ALL was defined by morphologic or cytogenetic evidence, either in peripheral blood or in bone marrow. Relapsed disease for CML was defined by hematological, cytogenetic or molecular evidence of recurrence. Non-relapse mortality was defined as the time from transplantation until death from any cause other than leukemia. Engraftment was defined as an absolute neutrophil count of >500/μl, and an unsupported platelet count of >20 000/μl for 3 consecutive days or detection of donor DNA by PCR short tandem repeat (STR). KIR genes associated with a superior transplant outcome (less acute GVHD, relapse and higher survival) were defined as ‘favorable’ KIR.

Statistical methods

Descriptive statistics were used to describe the patient characteristics, pre-transplant variables and post-transplant outcomes. Standard techniques in survival analysis, including Kaplan–Meier estimates and the Cox proportional hazard models, were used to estimate the time-to-event distributions of OS, DFS or current LFS, relapse, GVHD and transplant-related mortality (TRM). Statistical associations between pre-transplant variables were investigated using correlation analysis, including Pearson's correlation coefficients and Spearman's rank correlation coefficients, and multiple regression analysis. Statistical tests based on t-tests, χ2 tests and F-tests were used to evaluate the statistical significance of covariates in multiple regression models or the Cox proportional hazard models. Variables included in univariate analysis were as follows: age (continuous, < vs median); gender; donor–patient sex match (female to male vs others); disease risk (HR vs SR); type of transplant (BMT vs PBSCT); missing KIR ligand (yes vs no, >1 vs 1 or no); number of total, activating, inhibitory KIRs in donors (cutoff model < vs median or in a model using KIR as a continuous variable); individual KIR in donor (present vs not present); donor/recipient KIR pairs23—four types (1, recipient KIR genotypes included in the donor KIR genotypes; 2, donor KIR genotypes included in the recipient; 3, identical donor and recipient KIR genotypes; 4, different donor and recipient KIR genotypes); CD34 dose (continuous, < vs median); CD3 dose (0.2 vs >0.2 × 105/kg CD3+ cells, NK30 (continuous, < vs median); T-cell sub-populations (T30) (C3+CD4+, CD3+CD8+) (continuous, < vs median) and CSA dose (STD vs LD). In view of multiple comparisons that have been made in the study and relatively small sample size, P-value should be considered preliminary. Multivariate analysis was performed using the Cox models. Data analysis was performed using SPSS 14 for Windows (SPSS Inc., Chicago, IL, USA) software.



Patient characteristics are shown in Table 1. The median absolute NK cell count 30 days after transplant (NK30) was 149/μl (6–1005) for all patients, 217/μl (6–577) for patients with AML, 110/μl (18–1005) for patients with CML and 199/μl (22–975) for patients with ALL. For further analysis, the group of all patients was dichotomized at 150/μl.

Donor KIR gene frequencies

Donor KIR gene frequency is shown in Figure 1. Frequencies of KIR types corresponded with published series.14, 20, 24, 25 Five KIR genes were almost universally expressed: 2DL4, 3DL2 and 3DL3 were present in 100% of the donors; all except one was positive for 2DL1 and all except two were positive for 2DL3 (Figure 1). The most frequent activating KIR genes were 2DS4*00101/00102/002 and 2DS4*003-005 found in 29 (58%) of the donors. The activating KIR gene 2DS3 occurred in 9 (18%). In 23 (46%) of donors, KIR 2DL5A, 2DS1 and 3DS1 were co-inherited and 2DS5 was also inherited in 21 of these, owing to a strong linkage disequilibrium. Patients receiving transplants from 23 donors positive for the gene 2DL5A/2DS1/3DS1 (favorable KIR) had a lower incidence of acute GVHD and relapse but improved survival (Table 2). Therefore, we considered this KIR as ‘favorable’.

Figure 1

KIR genotype frequencies in donors. KIR, killer immunoglobulin-like receptor.

Table 2 Univariatea analysis for acute GVHD, NRM, chronic GVHD, relapse and survival

The median number of total KIR genes in the donor was 10 (range, 7–14) (activating 4 (range, 1–6) and inhibitory 7 (range, 5–8)). HLA and KIR typing showed that 30 (60%) donor–recipient pairs had missing KIR ligands for the corresponding HLA type. Among 30 pairs with missing KIR ligands, 24 (48%) pairs had no C1 or C2 group allele in the recipient for donor KIR 2DL2/2DL3 or 2DL1 respectively and 15 (30%) pairs had no HLA Bw4-associated HLA alleles in the recipient for donor KIR 3DL1 and 9 (18%) had both (group C1 or C2 and Bw4) missing KIR ligands. When comparing donor and recipient KIR genotypes, the frequency of the donor KIR genotype included in the recipient KIR genotype in 8 (16%), recipient KIR genotype included in the donor in 4 (8%), identical donor and recipient KIR genotypes reached 22 (44%) while different donor and recipient KIR genotype was seen in 16 (32%).

Engraftment parameters

All 54 patients engrafted. Mean time to neutrophil and platelet engraftment was 17 (range, 11–26) and 20 (range, 14–38) days respectively. Patients with NK30>150/μl had faster platelet engraftment (mean±s.e.m. 17.5±0.6 vs 22.1±1.4 days for NK30<150/μl; P=0.003) and faster neutrophil engraftment (15.6±0.4 vs 17.8±0.8 days for NK30<150/μl; P=0.02). Mean number of febrile days post-transplant was 7 (range, 0–36); patients with NK30>150/μl had fewer febrile days (mean±s.e.m. 5.0±0.95 days) compared to NK30<150/μl (9.1±1.8 days; P=0.04). STR chimerism (data available for 26 patients receiving SCT after 11/2001) showed a median time to full donor lymphoid chimerism of 90 days (range, 30–616). There was no correlation between NK30 (either as a continuous or a categorical variable) and time to full donor T-cell chimerism (P=0.45 and 0.55 respectively).


Nineteen of 54 patients (AML and CML 12/39, ALL 7/15) relapsed with a cumulative incidence of 40.8±7.5%. Median time to relapse was 106 days (range, 28–1134).

Based on clinical studies showing that NK cell alloreactivity only affected relapse risk in myeloid leukemia, relapse rates were analyzed separately in myeloid leukemias (AML and CML). In univariate analysis (Table 2), decreased risk of relapse was associated with SR disease, NK30>150/μl (Figure 2a), higher donor total KIR, activating KIR, inhibitory KIR and occurrence in the donor of the favorable co-inherited KIR groups 2DL5A/2DS1/3DS1 and also KIR 2DS5. Higher total number of KIR genes (total, activating, inhibitory) was associated with protection against relapse in both continuous (total KIR, P=0.002, 95% CI 0.4–0.8; activating KIR, P=0.006, 95% CI 0.4–0.8; inhibitory KIR, P=0.01, 95% CI 0.15–0.8) and cutoff ( vs Table 2) models. As shown in Figure 2a, relapse was significantly higher when patients had NK30 <150/μl (cumulative incidence of 70.5±12.8 vs 5.3±5.1% for patients with higher NK 30; P=0.0001). We confirmed that the impact of NK30 occurred only in myeloid leukemias and not in ALL (Figure 2b), while the benefits of a higher NK30 were seen in myeloid leukemias in both STD and HR disease. In multivariate analysis, the only independent factor associated with higher relapse in myeloid leukemias was NK30<150/μl (HR 18.3, P=0.039) (Table 3).

Figure 2

NK30 and transplant outcome (a) relapse in myeloid leukemia (AML/MDS), (b) relapse in ALL, (c) acute GVHD (grade II–IV) (all patients), (d) non-relapse mortality (all patients), (e) survival (all patients), (f) survival in ALL. ALL, acute lymphoblastic leukemia; AML, acute myelogenous leukemia; GVHD, graft-versus-host disease; NK30, absolute natural killer cell counts on day 30 post transplant.

Table 3 Multivariate Cox regression models for relapse in myeloid leukemia

Acute GVHD

Twenty-seven of 54 patients developed acute GVHD grade (II–IV), with a cumulative incidence of 50.4±6.8%. In univariate analysis (Table 2), factors significantly associated with increased acute GVHD were HR disease, NK30<150/μl (Figure 2c), lower than median number of total and activating KIR in the donor, and there was trend toward more acute GVHD in patients with low numbers of inhibitory and absence of favorable KIR in the donor (Figure 2a). Multivariate analysis showed that the only independent factor associated with more acute GVHD was NK30<150/μl (HR 3.2, P=0.032) (Table 4).

Table 4 Multivariate Cox regression models for acute GVHD (all patients)

Chronic GVHD

Twenty-six of 41 (63.4%) evaluable patients surviving more than 100 days developed chronic GVHD (limited 20 (48.7%); extensive 6 (14.6%)) with a cumulative incidence of 82.6±12.8%. In univariate analysis, factors significantly associated with increased chronic GVHD were NK30>150/μl and more inhibitory KIR (Table 2). Multivariate analysis showed that more chronic GVHD was associated with NK30>150/μl (HR 0.30, 95% CI 0.9–1.0; P=0.057) and 7 inhibitory KIR in donor (HR 3.6, 95% CI 0.96–13.7; P=0.058).


Eight of 54 patients died from non-relapse causes, with a cumulative incidence of NRM of 17.7.±6%. Causes of death were: fungal infection (1), viral infection (2), GVHD-related (2) and acute respiratory distress syndrome/idiopathic pneumonitis (3).

In univariate analysis (Table 2), the only risk factor associated with increased NRM was NK30<150/μl (cumulative incidence 40.5±13.9 vs 3.6±3.5%; P=0.007) (Figure 2d).

The adverse effect of NK30<150/μl on NRM was confirmed in multivariate analysis (HR 10.7, 95% CI 1.3–88.1; P=0.028).


Currently, 28 of 54 patients are alive and disease free, with an actuarial survival of 49.3±7.3% at a median follow-up of 57, range 14–157 months. In univariate analysis (Table 2), increased survival was associated with SR disease, NK30>150/μl (Figure 2e), higher numbers of KIR, donor with favorable KIR and occurrence of chronic GVHD. As shown in Figure 2e, higher NK30 (>150/μl) was associated with improved overall survival (74.8±8.2 vs 22.9±9% for patients with NK30<150/μl; P=0.0002). When analyzed separately, the impact of NK30 on survival was found only in myeloid leukemias (both HR and SR disease), but not in ALL (Figure 2f). Multivariate analysis showed that the independent factors associated with improved survival in myeloid leukemias were NK30>150/μl (HR 11.4, P=0.03) and chronic GVHD (HR 5.3, P=0.043) (Table 5).

Table 5 Multivariate Cox regression models for survival in myeloid leukemia

Interrelationship between NK30, KIR genotype and transplant variables

Since NK30 was an independent factor associated with improved transplant outcome with less acute GVHD, relapse and better survival, we sought for factors contributing to NK30 recovery and for relationships between NK30, KIR genotype characteristics and other transplant variables. We found that favorable KIR genotype and median inhibitory KIR numbers in the donor associated with NK30 above the median of 150/μl (16 of 23 with favorable KIR had NK30>150/μl compared to only 10 of 27 patients who received transplant from donor without favorable KIRs; P=0.027; 21 of 32 with inhibitory KIR 7 had NK30>150/μl compared to only 5 of 18 patients who received transplant from donor with inhibitory KIR numbers <7; P=0.018).

When studied as continuous variables, there was a significant correlation between NK30 count and numbers of donor inhibitory KIRs (correlation coefficient 0.364, P=0.009), favorable donor KIR (correlation coefficient 0.324, P=0.021), CD34 cell dose (correlation coefficient 0.483, P=0.0002), and an inverse correlation with CD3 cell dose of greater or less than 2 × 104/kg (correlation coefficient −0.296, P=0.029).


Recent experimental and clinical data have elucidated the role of NK cells in allogeneic SCT. MHC-mismatched transplants in mice and humans demonstrate that donor NK cells target hematopoietic tissues of the host, eliminating host antigen-presenting cells, host hematopoiesis and host leukemia. These effects translate into better engraftment, diminished risk from acute GVHD, reduced relapse from an NK-mediated graft-versus-leukemia (GVL) effect and lower NRM.1, 2, 3, 4, 5, 6, 9, 14, 15

Previously, we observed that NK30, our surrogate marker for early NK cell recovery, affected the rapidity of molecular remission in 20 CML patients.3 In the present study, we could confirm these favorable effects of early NK cell recovery for patients with AML but not with ALL, coinciding with experimental observations that ALL cells may carry resistance against NK-mediated lysis.1, 2, 9 Our analysis supports the contention that alloreacting NK cells can affect acute GVHD and NRM and exert powerful GVL effects in myeloid malignancies. Multivariate analysis showed that NK30 above >150/μl independently associated with less aGVHD (HR 3.2; P=0.03), NRM (HR 10.7; P=0.028) for all patients less relapse (HR 18.3; P=0.039) and better survival (HR 11.4; P=0.029) in myeloid leukemia. Interestingly, our series also showed a higher incidence of chronic GVHD in patients with high NK30. A possible explanation is that more patients with higher NK30 survived to develop chronic GVHD (which was not associated with higher mortality because the majority had only limited chronic GVHD). Chronic GVHD contributed an additional GVL effect, as both NK30 and chronic GVHD were independently associated with better survival in multivariate analysis (Table 5).

We also identified certain characteristics of the donors’ KIR genotype as influential on transplant outcomes in this HLA-matched, T-cell-depleted transplant series. In contrast to the strong impact of KIR ligand incompatibility on outcome for myeloid leukemias in haploidentical SCT, analyses of transplant outcome of HLA-identical sibling SCT have not produced consistent findings. Cook et al.14 showed that in HLA-matched (T-replete) sibling SCT for myeloid leukemia (n=112), patients homozygous for C2 alleles receiving a graft from a donor carrying KIR gene 2DS2 had a significantly reduced survival. Verheyden et al.20 observed (in mixed T-deplete and -replete populations; n=65) the presence of two KIR, 2DS1 and 2DS2, in the donor to be significantly associated with a decreased leukemia relapse after related HLA-identical SCT. Hsu et al.15 showed that after partially T-depleted HLA-identical sibling SCT (n=178), absence of recipient HLA-C or HLA-B ligands for donor-inhibitory KIR contributed to improved outcomes for patients with AML/MDS. Gagne et al.23 showed no acute GVHD after related BMT in four patients whose KIR genotypes were shared with the KIR genotypes of their donors. Chen et al.19 showed (after T-replete HLA-identical SCT; n=131) that additional activating KIR genes are associated with less TRM and a better survival.

Profound T-cell depletion has been considered as a prerequisite for detecting effects mediated by alloreactive NK cells.1, 2, 9, 26 This assumption is supported by data from Bishara et al.,18 who showed that lack of extensive T-cell depletion in haploidentical SCT was associated with high GVHD rates and diminishes the benefits of NK cell alloreactivity. Similarly, Sun et al.27 found a detrimental effect of NK alloreactivity after unrelated (HLA-mismatched and -matched) T-replete allo-SCT. Our quantitative T-cell depletion (around 2–5 × 104/kg) was comparable to that used by other groups and may be a contributing factor in our study. In fact we found an inverse relationship between T-cell dose and higher NK30 counts. This may be due to competition between NK cells (or their progenitors) and T cells for a limited reservoir of growth factors during the early lymphopenic stage post-transplant, although this hypothesis requires substantiation by further studies. Finally, T-cell depletion may very well enhance and ‘visualize’ the beneficial effects from NK cells.

We observed an effect of favorable KIR and KIR numbers (total, activating or inhibitory) on transplant outcome, but no effect of missing KIR ligands. De Santis et al.16 similarly found that higher numbers of KIR genes in donors protect against leukemia relapse following unrelated donor SCT. Because favorable KIR types and KIR numbers were so closely linked in our series, it was not possible to determine which factors were covariates and which were directly affecting outcome.

Multivariate analysis models that included the NK30 consistently showed that NK30 and not KIR group genotype features emerged as the independent variable predicting outcome. This could be explained by the fact that NK30 was also positively influenced by the CD34 dose and negatively with the infused CD3+ T-cell dose in the graft. In addition, our data showed that patients with higher NK30 have faster engraftment characteristics, and NK30 recovery correlated with infused CD34 dose. A high CD34 dose has been identified by many transplant groups, including ours, as being associated with improved outcome.28, 29 Our findings suggest that a better NK recovery may be the reason why there is a distinct advantage conferred by a high CD34 cell dose in the graft. The implication of the superior predictive value of the NK30 count over genotype is that transplant outcome is as much influenced by quantity as the quality of NK recovery. How KIR genotype might affect the pace of NK cell recovery is not known. In in vitro proliferative culture assays of NK cells derived from donors with favorable and unfavorable genotypes, we did not find any difference in the proliferative capacity (data not shown), although there are data supporting the importance of activating and inhibitory class I MHC receptors in regulating NK cell proliferation.30

In conclusion, the NK30 count appears to be a highly important and predictive factor for the outcome after T-cell-depleted HLA-identical transplantation in patients with myeloid leukemia. Our data suggest that CD34+ dose, CD3+ dose and KIR gene characteristics affect the overall outcome after transplant in a univariate fashion. All these variables may have affected NK cell recovery, explaining why NK30 was the only and very strong independent factor in our series.

Nevertheless, we should emphasize that these findings are preliminary and limited by the retrospective nature of the study, small sample size (with wide confidence interval for survival and relapse) and limited (albeit random) sample availability for lymphocyte subset analysis and KIR genotyping. Further prospective studies will be necessary to verify these findings. Taken together, we identified parameters that could ameliorate the conditions after T-cell-depleted transplants for myeloid leukemias by optimizing both T-cell and stem cell doses and thereby creating the optimal conditions for sufficient NK cell recovery required for disease control and GVHD prevention.


  1. 1

    Ruggeri L, Capanni M, Casucci M, Volpi I, Tosti A, Perruccio K et al. Role of natural killer cell alloreactivity in HLA-mismatched hematopoietic stem cell transplantation. Blood 1999; 94: 333–339.

  2. 2

    Ruggeri L, Capanni M, Urbani E, Perruccio K, Shlomchik WD, Tosti A et al. Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants. Science 2002; 295: 2097–2100.

  3. 3

    Savani BN, Rezvani K, Mielke S, Montero A, Kurlander R, Carter CS et al. Factors associated with early molecular remission after T cell-depleted allogeneic stem cell transplantation for chronic myelogenous leukemia. Blood 2006; 107: 1688–1695.

  4. 4

    Zeis M, Uharek L, Glass B, Gaska T, Steinmann J, Gassmann W et al. Allogeneic NK cells as potent antileukemic effector cells after allogeneic bone marrow transplantation in mice. Transplantation 1995; 59: 1734–1736.

  5. 5

    Hauch M, Gazzola MV, Small T, Bordignon C, Barnett L, Cunningham I et al. Anti-leukemia potential of interleukin-2 activated natural killer cells after bone marrow transplantation for chronic myelogenous leukemia. Blood 1990; 75: 2250–2262.

  6. 6

    Murphy WJ, Reynolds CW, Tiberghien P, Longo DL . Natural killer cells and bone marrow transplantation. J Natl Cancer Inst 1993; 85: 1475–1482.

  7. 7

    Brooks AG, Boyington JC, Sun PD . Natural killer cell recognition of HLA class I molecules. Rev Immunogenet 2000; 2: 433–448.

  8. 8

    Boyington JC, Motyka SA, Schuck P, Brooks AG, Sun PD . Crystal structure of an NK cell immunoglobulin-like receptor in complex with its class I MHC ligand. Nature 2000; 405: 537–543.

  9. 9

    Ruggeri L, Mancusi A, Burchielli E, Aversa F, Martelli MF, Velardi A . Natural killer cell alloreactivity and haplo-identical hematopoietic transplantation. Cytotherapy 2006; 8: 554–558.

  10. 10

    Jiang YZ, Barrett AJ, Goldman JM, Mavroudis DA . Association of natural killer cell immune recovery with a graft-versus-leukemia effect independent of graft-versus-host disease following allogeneic bone marrow transplantation. Ann Hematol 1997; 74: 1–6.

  11. 11

    Mackinnon S, Bungey J, Chase A, Paulsen W, Hows JM, Goldman JM . Origin and function of adherent lymphokine activated killer cells in patients with chronic myeloid leukaemia who relapse following bone marrow transplantation. Br J Haematol 1991; 77: 60–65.

  12. 12

    Sconocchia G, Lau M, Provenzano M, Rezvani K, Wongsena W, Fujiwara H et al. The antileukemia effect of HLA-matched NK and NK-T cells in chronic myelogenous leukemia involves NKG2D-target-cell interactions. Blood 2005; 106: 3666–3672.

  13. 13

    Sconocchia G, del PD, Barrett AJ . Non-classical antileukemia activity of early recovering NK cells after induction chemotherapy and HLA-identical stem cell transplantation in myeloid leukemias. Leukemia 2006; 20: 1632–1633.

  14. 14

    Cook MA, Milligan DW, Fegan CD, Darbyshire PJ, Mahendra P, Craddock CF et al. The impact of donor KIR and patient HLA-C genotypes on outcome following HLA-identical sibling hematopoietic stem cell transplantation for myeloid leukemia. Blood 2004; 103: 1521–1526.

  15. 15

    Hsu KC, Keever-Taylor CA, Wilton A, Pinto C, Heller G, Arkun K et al. Improved outcome in HLA-identical sibling hematopoietic stem-cell transplantation for acute myelogenous leukemia predicted by KIR and HLA genotypes. Blood 2005; 105: 4878–4884.

  16. 16

    De SD, Bishara A, Witt CS, Nagler A, Brautbar C, Slavin S et al. Natural killer cell HLA-C epitopes and killer cell immunoglobulin-like receptors both influence outcome of mismatched unrelated donor bone marrow transplants. Tissue Antigens 2005; 65: 519–528.

  17. 17

    Leung W, Iyengar R, Triplett B, Turner V, Behm FG, Holladay MS et al. Comparison of killer Ig-like receptor genotyping and phenotyping for selection of allogeneic blood stem cell donors. J Immunol 2005; 174: 6540–6545.

  18. 18

    Bishara A, De SD, Witt CC, Brautbar C, Christiansen FT, Or R et al. The beneficial role of inhibitory KIR genes of HLA class I NK epitopes in haploidentically mismatched stem cell allografts may be masked by residual donor-alloreactive T cells causing GVHD. Tissue Antigens 2004; 63: 204–211.

  19. 19

    Chen C, Busson M, Rocha V, Appert ML, Lepage V, Dulphy N et al. Activating KIR genes are associated with CMV reactivation and survival after non-T-cell depleted HLA-identical sibling bone marrow transplantation for malignant disorders. Bone Marrow Transplant 2006; 38: 437–444.

  20. 20

    Verheyden S, Schots R, Duquet W, Demanet C . A defined donor activating natural killer cell receptor genotype protects against leukemic relapse after related HLA-identical hematopoietic stem cell transplantation. Leukemia 2005; 19: 1446–1451.

  21. 21

    Montero A, Savani BN, Shenoy A, Read EJ, Carter CS, Leitman SF et al. T cell depleted peripheral blood stem cell allotransplantation with T cell add back for patients with hematological malignancies: effect of chronic GVHD on outcome. Biol Blood Marrow Transplant 2006; 12: 1318–1325.

  22. 22

    Dupont B, Hsu KC . Inhibitory killer Ig-like receptor genes and human leukocyte antigen class I ligands in haematopoietic stem cell transplantation. Curr Opin Immunol 2004; 16: 634–643.

  23. 23

    Gagne K, Brizard G, Gueglio B, Milpied N, Herry P, Bonneville F et al. Relevance of KIR gene polymorphisms in bone marrow transplantation outcome. Hum Immunol 2002; 63: 271–280.

  24. 24

    Witt CS, Dewing C, Sayer DC, Uhrberg M, Parham P, Christiansen FT . Population frequencies and putative haplotypes of the killer cell immunoglobulin-like receptor sequences and evidence for recombination. Transplantation 1999; 68: 1784–1789.

  25. 25

    Crum KA, Logue SE, Curran MD, Middleton D . Development of a PCR-SSOP approach capable of defining the natural killer cell inhibitory receptor (KIR) gene sequence repertoires. Tissue Antigens 2000; 56: 313–326.

  26. 26

    Cooley S, McCullar V, Wangen R, Bergemann TL, Spellman S, Weisdorf DJ et al. KIR reconstitution is altered by T cells in the graft and correlates with clinical outcomes after unrelated donor transplantation. Blood 2005; 106: 4370–4376.

  27. 27

    Sun JY, Dagis A, Gaidulis L, Miller MM, Rodriguez R, Parker P et al. Detrimental effect of natural killer cell alloreactivity in T-replete hematopoietic cell transplantation (HCT) for leukemia patients. Biol Blood Marrow Transplant 2007; 13: 197–205.

  28. 28

    Singh AK, Savani BN, Albert PS, Barrett AJ . Efficacy of CD34+ stem cell dose in patients undergoing allogeneic peripheral blood stem cell transplantation after total body irradiation. Biol Blood Marrow Transplant 2007; 13: 339–344.

  29. 29

    Ringden O, Barrett AJ, Zhang MJ, Loberiza FR, Bolwell BJ, Cairo MS et al. Decreased treatment failure in recipients of HLA-identical bone marrow or peripheral blood stem cell transplants with high CD34 cell doses. Br J Haematol 2003; 121: 874–885.

  30. 30

    Warren HS . Using carboxyfluorescein diacetate succinimidyl ester to monitor human NK cell division: analysis of the effect of activating and inhibitory class I MHC receptors. Immunol Cell Biol 1999; 77: 544–551.

Download references

Author information

Correspondence to A J Barrett.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Savani, B., Mielke, S., Adams, S. et al. Rapid natural killer cell recovery determines outcome after T-cell-depleted HLA-identical stem cell transplantation in patients with myeloid leukemias but not with acute lymphoblastic leukemia. Leukemia 21, 2145–2152 (2007) doi:10.1038/sj.leu.2404892

Download citation


  • hematological malignancies
  • natural killer cells
  • KIR genotype
  • allogeneic stem cell transplantation

Further reading