Volume 16, Issue 9 , Pages 1257-1264, September 2010
Donor KIR Genes 2DL5A, 2DS1 and 3DS1 Are Associated with a Reduced Rate of Leukemia Relapse After HLA-Identical Sibling Stem Cell Transplantation for Acute Myeloid Leukemia but Not Other Hematologic Malignancies
Article Outline
- Abstract
- Introduction
- Materials and Methods
- Results
- Discussion
- Acknowledgment
- Supplementary Data
- References
- Copyright
Stem cell transplantation (SCT) from a healthy donor can be curative for patients with hematologic malignancies resistant to other treatments. Elimination of malignant cells through a graft-versus-leukemia (GVL) effect involves donor T and natural killer (NK) cells, but their relative contribution to this process is poorly defined. NK cell alloreactivity and GVL effects are controlled by the nature of the interaction of NK activation receptors and killer-immunoglobulin-like-receptors (KIR) with major histocompatibility locus class I antigens on the target cell. We performed KIR-genotyping of HLA-identical sibling donors in 246 T cell-depleted SCTs to identify genetic factors affecting transplant outcome (treatment-related mortality [TRM], leukemic relapse, and survival). Univariate and multivariate analysis of transplant-related risk factors and KIR genotyping was performed to identify independent variables predictive of outcome for different forms of leukemia. Further to confirming known predictive factors for TRM and survival (CD34 cell dose, patient age, disease stage), statistical analysis revealed that 3 donor B haplotype KIR genes, 2DL5A, 2DS1, and 3DS1, were associated with significantly less relapse in patients with acute myelogenous leukemia (AML) (13% versus 57%) but not in patients with other myelogenous or lymphoid malignancies. AML patients receiving SCT from donors with these KIR genes relapsed 4 times less frequently than patients transplanted from donors with other KIR genotypes. These findings suggest specific, genetically determined, interactions between NK cells and AML cells that facilitate the GVL effect, and have implications for donor selection for AML patients.
Key Words: NK cells, Stem cell transplantation, Activatory KIR, Relapse, Acute myelogenous leukemia
Introduction
Although outcomes for allogeneic stem cell transplantation (SCT) are improving, better survival is largely related to reduced treatment-related mortality (TRM) and not to reduction in leukemic relapse, which still accounts for about one-third of treatment failures. Relapse represents a failure of the conditioning regimen or the alloimmune graft-versus-leukemia (GVL) effect. Both T-lymphocytes and natural killer (NK) cells have GVL effects, but the exact mechanism by which these 2 cell types contribute to leukemia cure after SCT is not known, and probably varies according to disease type, transplant conditions, and donor-recipient match. Several studies have drawn attention to the predictive value for relapse and survival of the lymphocyte count approximately 1 month after HLA identical SCT for leukemia 1, 2, 3, 4, 5, 6. In a previous small series of 54 patients receiving T cell-depleted allogeneic SCT from HLA identical donors, we showed that outcome, and especially relapse risk, was linked to NK cell and not T cell recovery [7]. Because NK alloreactivity is determined by mismatches between recipient HLA class I group and NK cell killer immunoglobulin-like (KIR) receptor expression [8], we examined patterns of NK-donor mismatching and KIR genetics in the recipient and donor to determine whether allo-disparity or specific KIR genotypes were predictive of relapse. In the multivariate analysis, only NK count at day 30 posttransplant emerged as a significant factor affecting outcome, but, in the univariate analysis, a number of donor KIR factors were significant. In particular, donor KIR genes 2DL5A, 2DS1, and 3DS1 (which are highly correlated, indicating that they are usually inherited together), were associated with a significantly reduced relapse rate and improved survival [7]. Similar results have been reported in recipients of unrelated donor SCT: in a group of acute myelogenous leukemia (AML) recipients, donors with KIR haplotype B (which includes 2DL5A/2DS1/3DS1) had less relapse and more favorable outcome after SCT than recipients of donors without these genes [9]. It therefore appears that donor KIR genetics may influence the risk of relapse, at least in patients with myelogenous leukemia.
In this study, we set out to better define the contribution of KIR genotype to transplant outcome in specific disease types by analyzing 261 related donor transplants for leukemia performed at the National Institutes of Health (NIH) in the 15-year period of 1993 to 2008. KIR genotype, KIR haplotype, and inheritance of genes 2DL5A, 2DS1, and 3DS1 were included in univariate and multivariate analyses of transplant variables. Our findings show that in this matched sibling transplant population, it is not the presence of the activating B haplotype, but the presence of 3 particular donor genes within haplotype B that is associated with reduced relapse. This study showed that these 3 “favorable” donor KIR genes are only beneficial in patients with de novo AML and not other forms of leukemia.
Materials and Methods
Study Population
Two hundred sixty-one consecutive patients receiving HLA identical sibling allografts at the NHLBI between 1993 and May 2008 were included in the study. Of these, 246 had donor samples available for KIR typing. This cohort of 246 was fully representative in terms of disease type and outcome. Patients received conditioning with cyclophosphamide 120 mg/kg and fractionated total body irradiation (TBI) (400-1300 cGy) with or without fludarabine 125 mg/m2 followed by T cell-depleted CD34 cell selected SCT and cyclosporine as the single posttransplant prophylaxis for graft-versus-host disease (GVHD). Donor samples for KIR DNA typing consisted of 239 cryopreserved donor cell samples and 7 buccal swabs. All patients and donors were enrolled onto NHH treatment protocols and provided written informed consent. Follow-up ranged from 6 months to 15 years. Patient characteristics are described in Table 1.
Table 1. Patient Characteristics
| Age | |
 Median | 36 years (range: 9-66) |
| Sex | |
| Male | 146 (56%) |
| Female | 115 (44%) |
| Race | |
| Asian | 35 (13%) |
| African/American | 25 (10%) |
| Hispanic | 131 (50%) |
| Caucasian | 70 (27%) |
| Disease | |
| ALL | 40 (15%) |
| AML | 70 (27%) |
| CML | 99 (38%) |
| MDS | 39 (15%) |
| CLL/NHL/MM | 13 (5%) |
| Disease risk group | |
| Standard risk | 105 (40%) |
| High risk | 156 (60%) |
| Graft type | |
| Bone marrow | 41 (16%) |
| Peripheral blood Stem Cells | 220 (84%) |
| CD34+ dose | |
| Median | 5 × 106/kg (range 0.5-16) |
| CD3+ dose | |
| Median | 0.5 × 105/kg (range 0.005-3) |
| Acute GVHD | |
| Grade 0-1 | 107 (57%) |
| Grade 2-4 | 81 (43%) |
| No GVHD data | 72 |
DNA Extraction and KIR Typing
Genomic donor DNA was extracted from frozen cell samples or buccal swabs using the Qiagen column-based method (Qiamp midi kit, Qiagen, Chatsworth, CA). Each donor was KIR typed using 0.92 μg DNA for polymerase chain reaction (PCR). Sequence-specific primers (Invitrogen KIR Genotyping SSP kit, Invitrogen, Carlsbad, CA) were used to amplify the DNA, and each KIR gene was identified by the presence or absence of a band on 1.5% agarose electrophoresis gel stained with ethidium bromide. Fifteen KIR genes and 2 pseudogenes as well as the common variants of 2DL5, 2DS4, and 3DP1 were identified.
Patterns of Donor KIR Inheritance
First, we examined the “missing KIR ligand” model 10, 11, 12 by identifying patients missing an HLA-A, -Bw, or -C ligand for which their donor had an inhibitory KIR (a situation possible in the HLA matched setting as HLA and KIR genes segregate separately on different chromosomes). We did not evaluate the reverse situation, where donors were missing a ligand for patient KIR, as no cases of graft rejection were seen. We then assessed the donor KIR genes separately to identify whether individual KIRs are associated with differences in outcome measures in univariate analysis. The presence in the donor of 3 favorable KIR genes 2DL5A, 2DS1, and 3DS1 together was then examined. The impact of donor haplotype B (defined by the presence of at least 1 of the genes 2DL5A, 2DS1, 3DS1, 2DS2, 2DS3, or 2DS5) as well as the total number of activating and inhibitory KIR [13], were also assessed.
Statistical Method
Baseline patient characteristics and transplant outcomes were described using summary statistics, including means, medians, percents, standard deviations, and 95% confidence intervals (CIs). Univariate and multivariate Cox Proportional Hazard Models were used to evaluate the effects of KIR genes, disease, and other factors on the cumulative distribution functions of time to TRM, overall survival (OS) time, and time to relapse (TTR) with right censoring. Because there were multiple events per patient, our analyses included both single-event Cox Proportional Hazard Models and multiple-event competing risk models. For the single-event analysis, we considered TRM, OS, and TTR separately, and used 3 right censoring schemes: (1) for the analysis of time to TRM, patients who did not die from TRM were treated as censored; (2) for the analysis of time to OS, patients who were alive at the last visits were treated as censored; (3) for the analysis of time to relapse, patients who did not relapse before death or last visit were treated as censored. Statistical inferences for the effects of KIR genes and other covariates on the distribution functions of TRM, OS, and TTR were summarized using the likelihood ratio tests, the log-rank tests, and their corresponding P-values. Cumulative distributions of TRM, OS, and TTR over days since transplant were estimated by the Kaplan-Meier method for patients stratified by KIR genotype, disease, and other factors. For the multiple-event competing risk analysis (Supplementary Table 1), we included relapse, TRM, and non-TRM death as competing risks, and considered the marginal competing risk models with event time defined to be the “time to first event” (Terry M. Therneau and Patricia M. Grambsch, Section 8.4, “Modeling Survival Data, Extending the Cox Model,” 2000, New York: Springer-Verlag). These models were selected because of their simple and clinically meaningful interpretations. Statistical inferences for the effects of KIR genes and covariates were summarized using the likelihood ratio tests and their P-values. The correlation between KIR genes was analyzed using Pearson's correlation coefficient. Numerical computations were carried out using the S-Plus 8 (TIBCO Software Inc., Palo Alto, CA) and SPSS (SPSS Inc., Chicago, IL) statistical packages.
To evaluate the effect of donor KIR genotype and identify KIR genes or groups of genes associated with distinct outcomes, we did the following analyses comparing all outcome variables with (1) individual KIR genes; (2) A versus B haplotype; (3) within the haplotype, 3 KIR genes identified as favorable from previous study [7] versus Haplotype B genes not including 3 favorable KIR; (4) favorable KIR with versus without other B-haplotype-defining genes; (5) with favorable KIR (all B haplotype) versus without favorable KIR (only 44% being B haplotype).
Definitions
Donors possessing all 3 of the KIR genes 2DL5A, 2DS1, and 3DS1 (coinherited with correlation coefficients of 0.86-0.96), were termed to have favorable KIR. Of the 246 donors KIR typed, 33% (87) had all three 2DL5a, 2DS1, and 3DS1 genes, and only these 87 are defined as having favorable KIR. When we widened the definition of “favorable” KIR to include those with just 1 or 2 of the 3 genes, our results were less significant. Standard patient risk group was defined as AML or acute lymphoblastic leukemia (ALL) in first complete remission (CR1), chronic mylogenous leukemia (CML) in first chronic phase, and myelodysplastic syndrome (MDS) intermediate-1 by WHO classification; all other patients were categorized as high risk. Relapse was defined as hematologic relapse. In considering missing KIR ligands, we looked at C1 (Cw∗01,03,07,08,12,13,14), C2 (Cw∗02,04,05,06,15,1602, 17,18), HLA G, HLA Bw4 (B∗51,52,53,57,58, 08,13,27,37,44 and A∗23,24,32), and A∗03 and A∗11. For example, if the patient was homozygous for HLA-C1, -C2, or -Bw6 (and therefore had no C2, C1, or Bw4 ligands, respectively) and the corresponding inhibitory KIR receptor 2DL3, 2DL1 or 3DL1 was present in the donor (as they are in the majority of individuals), this combination would be classified as missing a KIR ligand.
Results
Outcome after SCT
The cohort consisted of 261 patients. The majority of patients (85%) had myelogenous diseases, of whom 70 (27%) had AML. There were significant differences between diseases in relapse and OS with the highest relapse rates in AML and ALL (see Table 2.)
Table 2. Transplant Outcome
| No. pts (%) | TRM | Overall Survival | Relapse | GVHD Grade 2-4 | |
|---|---|---|---|---|---|
| Disease | |||||
| AML | n = 70 (27%) | 16 (23%) | 30 (43%) | 24 (34%) | 31 (53%) |
| MDS | n = 39 (15%) | 6 (16%) | 21 (54%) | 12 (31%) | 4 (36%) |
| CML | n = 99 (38%) | 23 (23%) | 61 (61%) | 21 (21%) | 33 (40%) |
| CLL/NHL | n = 13 (5%) | 2 (15%) | 6 (46%) | 4 (31%) | 1 (25%) |
| ALL | n = 40 (15%) | 8 (20%) | 16 (40%) | 17 (43%) | 17 (43%) |
| Cox model | P = .8 | P = .022∗ | P = .03∗ | P = .520 | |
| All patients | n = 261 | 55 (21%) | 134 (51%) | 78 (30%) | 85 (44%) |
Univariate analysis of variables affecting transplant outcome (Supplementary Table 2) showed that, as expected, CD34 dose above 5 × 106/kg was associated with less relapse, less TRM, and better survival. High-risk patients had more relapse, more TRM, and lower survival rates. Peripheral blood graft type, higher lymphocyte count at day 30 posttransplant (LC30), and later transplant era, were all associated with reduced TRM, but they did not affect survival or relapse rates. The only KIR-related variable to affect outcome was the possession by the donor of the 3 favorable KIR genes that were associated with reduced relapse and improved OS. No other KIR-related variable affected transplant outcome, including any single KIR gene, missing donor KIR ligand, presence of donor haplotype B, total number of inhibitory KIR, and total number of activatory KIR.
In multivariate analysis (Table 3), high-risk disease was associated with more relapse, more TRM, and lower survival rates. Higher LC30 was associated with less TRM (P = .015) (Figure 1A) but did not independently affect risk of relapse or OS. Better OS was seen in recipients of a CD34 cell dose >5 × 106/kg.
Table 3. Multivariate Analysis of Significant Risk Factors in Cox Models
| Hazard Ratio | 95% CI | P-Value | |
|---|---|---|---|
| Relapse | |||
| High-risk disease | 4.65 | 2.6-8.4 | <.001∗ |
| CD34 >5 × 106/kg | 0.64 | 0.4-1.0 | .64 |
| Donor-favorable KIR | 0.54 | 0.3-0.9 | .02∗ |
| Overall survival | |||
| High-risk disease | 3.1 | 2.0-4.7 | <.001∗ |
| CD34 >5 × 106/kg | 0.54 | 0.3-0.8 | .001∗ |
| Donor-favorable KIR | 0.64 | 0.43-0.94 | .024∗ |
| TRM | |||
| High-risk disease | 2.36 | 1.3-4.3 | .004∗ |
| CD34 >5 × 106/kg | 0.55 | 0.3-1.0 | .057 |
| PBSC graft | 0.68 | 0.3-1.4 | .276 |
| LC30 >600 × 107/L | 0.42 | 0.2-0.9 | .015∗ |
| Recent transplant era | 0.95 | 0.5-1.8 | .866 |
Figure 1
Factors affecting SCT outcome in multivariate analysis of 261 patients. (A) Lymphocyte count at day 30 posttransplant and TRM; (B) donor KIR genotype and relapse.
Impact of KIR Genes on Transplant Outcome in Multivariate Analysis
KIR gene frequency among the donor population studied was similar to that of published data 14, 15, 16 and is shown in Table 4. Briefly, 90 (37%) donors were haplotype A and 156 (63%) had at least 1 B haplotype. Within the group with haplotype B, 87 (56%) had the 3 favorable KIR genes and 69 (44%) did not. Within those of B haplotype who had favorable KIR, 41 had 1, 30 had 2, and 16 had 3 other B-defining KIR genes in addition to the favorable 3. This frequency distribution allowed us to make robust statistical comparisons between haplotype A versus B and within haplotype B between donors who had “favorable” KIR versus those who did not. In multivariate analysis there was no significant difference between A and B haplotype on outcome. In this sample of 246 patients for whom donor KIR results were available, the presence of the 3 favorable donor KIR genes 2DL5A, 2DS1, and 3DS1 occurred in 87 cases (35% of donors). In multivariate analysis, patients transplanted from donors with the 3 favorable KIR genes had significantly less relapse (P = .02) (Figure 1B). When we compared outcome following transplants from donors with the 3 favorable KIR genes, 1 or 2 favorable KIR genes, or the 3 favorable genes plus 1, 2, or 3 additional B haplotype genes, versus the remainder population, possession of the 3 favorable KIR genes discriminated best for the patient group who had least relapse. In fact, of the 68 AML recipients of transplants, only 3 of 23 patients from donors possessing the 3 favorable KIR relapsed, whereas 20 relapses occured in the 45 AML patients transplanted from donors with other KIR combinations. Therefore, the favorable component of the B haplotype was best described as the 3 closely inherited KIR genes 2DL5a, 2DS1, and 3DS1, previously identified as favorable genes.
Table 4. Distribution of KIR Groups Among Donors
| Different Ways of Grouping Donors | No. (%) in Each Group |
|---|---|
| Haplotype A/A | 90 (37% of 246) |
| At least one B Haplotype | 156 (63% of 246) |
| Within Haplotype B: with 2DL5a, 2DS1,3DS1 (Fav KIR) | 87 (56% of 156) |
| Within Haplotype B: without 2DL5a, 2DS1,3DS1 | 69 (44% of 156) |
| Within Haplotype B with Fav KIR: with 1 other B-defining gene | 41(47% of 87) |
| Within Haplotype B with Fav KIR: with 2 other B-defining genes | 30 (34% of 87) |
| Within B with Fav KIR: with 3 other B defining genes | 16 (18% of 87) |
| With “favorable” KIR genes 2DL5a/2DS1/3DS1 | 89 (35% of 246) |
| No “favorable” KIR genes | 157 (65% of 246) |
| Within no fav KIR genes: with other B defining genes | 69 (44% of 157) |
| Within no fav KIR genes: without other B defining genes | 90 (56% of 157) |
Impact of Donor KIR Genes in Individual Diseases
RelapseBecause relapse rates were different in different diseases, we selected factors that significantly affected relapse rates in multivariate analysis, and studied them in disease subsets (Table 5). The protective effect of favorable donor KIR genes on relapse was restricted to patients with AML. Patients receiving transplants from donors with favorable KIR were 4 times less likely to relapse than recipients of donors not possessing favorable KIR (P = .02, hazard ratio [HR] = 0.24) (Figure 2A). The protective effect of favorable KIR on relapse was observed in both standard and high-risk AML patients (Figure 2B). None of the 13 patients with standard-risk AML receiving SCT from favorable KIR donors relapsed, whereas 19 standard risk patients receiving SCT from donors without favorable KIR had a 32% relapse rate. Similarly, patients with high-risk AML receiving SCT from favorable KIR donors had a relapse rate of 30%, whereas high-risk patient recipients of SCT from donors without favorable KIR had a 70% relapse rate. In diseases other than AML, neither favorable KIR, nor any other KIR-related variable, affected relapse, survival, or TRM (data not shown).
Table 5. Effect of Favorable Donor KIR Genes on Transplant Outcome in Different Diseases
| Patient Group with Donor KIR Group | TRM | P | HR | Survival | HR | P | Relapse | HR | P |
|---|---|---|---|---|---|---|---|---|---|
| AML + Fav KIR donor (n = 23) | 6 (26%) | .8 | 1.14 | 13 (57%) | 0.59 | .151 | 3 (13%) | 0.24 | .02∗ |
| Not fav (n = 45) | 9 (20%) | 17 (38%) | 20 (57%) | ||||||
| AML + haplo B donor (n = 46) | 10 (22%) | .92 | 0.95 | 20 (44%) | 1.07 | .86 | 15 (33%) | 0.86 | .73 |
| Not haplo B donor (n = 22) | 5 (23%) | 10 (46%) | 8 (36%) | ||||||
| MDS + Fav (n = 16) | 1 (6%) | .53 | 0.46 | 10 (63%) | 0.56 | .28 | 6 (37%) | 0.61 | .412 |
| Not fav (n = 17) | 2 (12%) | 8 (47%) | 6 (35%) | ||||||
| CML + Fav (n = 32) | 6 (19%) | .52 | 0.73 | 21 (66%) | 0.78 | .47 | 6 (19%) | 0.75 | .55 |
| Not fav (n = 63) | 15 (24%) | 38 (60%) | 14 (22%) | ||||||
| CLL/NHL + Fav (n = 4) | 0 (0%) | .56 | 0.03 | 3 (75%) | 0.01 | .22 | 0 (0%) | 0.02 | .37 |
| Not fav (n = 9) | 2 (22%) | 3 (33%) | 4 (44%) | ||||||
| ALL + Fav (n = 12) | 2 (17%) | .82 | 0.83 | 4 (33%) | 1.21 | .66 | 6 (50%) | .417 | |
| Not fav (n=12) | 5 (20%) | 11 (44%) | 9 (36%) | ||||||
| High risk + Fav (n = 50) | 8 (16%) | .1 | 0.51 | 24 (48%) | 0.59 | .02∗ | 19 (38%) | 0.63 | .1 |
| Not Fav (n = 94) | 25 (27%) | 31 (33%) | 41 (44%) | ||||||
| Std risk Fav (n = 37) | 7 (19%) | .43 | 1.51 | 27 (73%) | 0.88 | .74 | 2 (5%) | 0.27 | .09 |
| Not Fav (n = 65) | 8 (12%) | 46 (71%) | 12 (18%) |
Figure 2
Effect of donor KIR genes on relapse in AML patients (n = 68). (A) Stratified by donor KIR genes; (B) stratified by donor KIR and patient risk group.
In the entire cohort of 261 patients, LC30 and high-risk disease were significant for TRM in multivariate analysis. In disease subsets, LC30 below 600 × 107/L was again significant for TRM in AML (45% versus 9.5% for LC30 greater than the median; P = .04) and MDS (28% versus 4.8%, respectively; P = .047), but not in CML. In CML, TRM was affected by risk group with high risk patients having a higher TRM (39% versus 14%; P = .005) and by CD34 count with higher TRM for CD34 counts below the median (32% versus 12%; P = .022). In ALL, only bone marrow (BM) as the stem cell source was a significant risk factor for higher TRM (P = .043).
OSDisease risk, CD34 cell dose, and a donor with favorable KIR significantly affected OS in the entire cohort. In disease subsets, disease risk and CD34 cell dose were still significant in AML (for risk group: 63% survival for low risk versus 26% for high, P = .004; for CD34 dose: 21% below versus 64% above median dose, P = .001). CD34 dose was also significant for survival in CML (28% below versus 79% above median dose).
Discussion
In this study, we analyzed factors affecting transplant outcome in a cohort of 261 patients with hematologic malignancies undergoing T-depleted myeloablative SCT. Out of many single gene analyses and many combinations within and without the B haplotype, we identified 3 donor KIR genes (2DL5A, 2DS1, and 3DS1) associated with a reduced AML relapse rate post-SCT. Our findings suggest that an NK-mediated GVL effect in AML may be modulated by specific KIR genes in the donor. Studying the determinants of SCT outcome can identify risk factors that can be used to optimize transplant schedule design and donor selection. Furthermore, clinical results can point to biologic mechanisms underlying success or failure of SCT. In addition to well-recognized disease-related factors predicting outcome, we and others have identified CD34+ cell dose and lymphocyte counts 30 days after SCT as predictive factors for transplant outcome 3, 17, 18. In a previous analysis on a smaller cohort of 54 patients, we found that donor KIR genotype was also a strong predictor of survival and relapse after SCT. In particular, 3 closely linked donor genes 2DL5A, 2DS1 and 3DS1, of the B haplotype were predictive for outcome and relapse [7]. We therefore extended this observation in the present study where there were sufficient numbers of patients to assess the impact of donor KIR group on outcome in different diseases. Here we show that, in addition to known risk factors for outcome (disease risk, disease type, CD34+ cell dose, and day 30 lymphocyte count), donor KIR genes 2DL5A, 2DS1 and 3DS1 were associated with significantly reduced relapse rates after T-depleted SCT. This effect was restricted to patients with de novo AML and was not significant in CML, MDS, or lymphoid malignancies.
The impact of donor KIR on relapse in AML was striking: actuarial relapse was 13% in recipients of transplants from donors with favorable KIR and 57% in recipients whose donors lacked these genes (P = .02). The effect of donor KIR was seen in both standard and high-risk AML: no standard-risk patient receiving a transplant from a donor with favorable KIR relapsed, compared with 32% relapse rates for standard-risk recipients of donors with unfavorable KIR. Similarly, relapse rates for high-risk patients were affected by donor KIR group: 30% for recipients with favorable KIR donors and 70% for recipients with unfavorable KIR donors. Because the 3 favorable genes are usually inherited together, it is impossible to determine whether all 3, or only 1 KIR gene acts as a protective factor. Because 1 of the 3 genes occurs in only 5% of donors, very large patient cohorts would be required to explore the contribution of specific favorable KIR. Nevertheless, 4 donors who had only 3DS1 (grouped as “unfavorable” as they did not have all 3 favorable genes), conferred poor outcomes (3 died of relapse within 6 months of transplant and 1 from alveolar hemorrahage) indicating that perhaps 3DS1 is not the gene of interest.
Our results differ significantly from those of a recent analysis in AML patients receiving unrelated donor SCT, where SCT from donors of the B haplotype (which contains the favorable KIR genes) were associated with less relapse. As there is strong linkage disequilibrium among B haplotype genes it is difficult to be certain which genes are associated with the antileukemic effect. The statistical problems associated with multiple comparisons mean the results must be viewed cautiously. However, in this study, the strongest association with reduced AML relapse was with the KIR genes 2DL5A, 2DS1 and 3DS1 (P = .02, HR = 0.24) and analysis of the B haplotype itself was not found to be significant (P = .73, HR = 0.86). Our findings contrast with data by Hsu et al. [11], who studied the role of donor KIR in a mixed disease population of patients receiving identical sibling, T cell-depleted SCT similar to ours, and found no benefit for the presence of any particular donor KIR on outcome. Although we found that missing a ligand for donor KIR reduced TRM in AML patients, we could not replicate their findings of less relapse and improved survival in these patients. The difference may be explained by the fact that patients in Hsu et al.'s study all received BM grafts, whereas the majority of our patients received peripheral blood SCT. A separate analysis of the 39 BM recipients in our series did not show any impact of donor KIR groups, whereas the association between donor KIR and relapse was of greater significance when the peripheral blood stem cell transplants (PBSCTs) were analyzed separately (P < .029). Further analysis of larger patient series receiving different types of transplant regimens is required to clarify these differences. The disease specificity of our finding, with the impact of KIR group genetics mainly affecting AML, is consistent with findings from other centers where NK-KIR-mediated effects have been found to be strongest in AML 10, 11 as well as studies showing that alloreactive NK cells do not kill ALL cells in vitro [19].
The mechanism whereby KIR genes 2DL5A, 2DS1, and 3DS1 mediate a protective effect against relapse is unclear. Our previous small study linked these genes to higher NK cell counts post transplant suggesting a genetic control of NK cell recovery [7]. Interestingly, in our larger sample, we did not find an association of these genes with higher lymphocyte counts post transplant, but in the small group of patients who did not achieve lymphocyte counts above 0.2 × 109/L, the positive effect of these KIR genes was not seen, suggesting that a minimum level of NK cells need to be present for the effect of KIR genes to be facilitated. The association between favorable KIR and lower relapse rates was also stronger in peripheral blood stem cell transplantation, where higher circulating levels of all cells including NK occur earlier after transplant, than with BM grafts.
Our results suggest that there may be specific ligands on the surface of AML cells that make the interaction between favorable KIR receptors and AML cells particularly effective at causing NK-mediated AML cell death. Alternatively, these favorable KIR genes may simply represent a marker linked to an as yet undefined GVL pathway. Further studies are required to investigate this and determine whether the beneficial KIR-AML interaction is restricted to specific AML subtypes and reproduced in HLA matched but unrelated SCT.
Our findings have practical implications and support the concept of selecting donors by KIR genotype to optimize GVL effects after SCT for AML. In the absence of a favorable donor, patients could be considered for experimental treatments to boost GVL, for example, using NK infusions or lower doses of immunosuppression. Recent trials using allogeneic NK cell infusions to boost immunity post transplant are promising 20, 21, 22. Selecting donors with favorable KIR could potentially increase the ability of such infusions to control leukemia and prevent relapse. This would be a practical possibility because donors with favorable KIR genes represent up to 40% of the population.
Financial disclosure: The authors have nothing to disclose.
Supplementary Data
Supplementary Tables.
References
- Effect of slow lymphocyte recovery and type of graft-versus-host disease prophylaxis on relapse after allogeneic bone marrow transplantation for acute myelogenous leukemia. Bone Marrow Transplant. 2001;28:951–956
- Lymphocyte recovery after allogeneic bone marrow transplantation predicts risk of relapse in acute lymphoblastic leukemia. Leukemia. 2003;17:1865–1870
- Clinical impact of early absolute lymphocyte count after allogeneic stem cell transplantation. Br J Haematol. 2004;125:217–224
- Absolute lymphocyte count on day 30 is a surrogate for robust hematopoietic recovery and strongly predicts outcome after T cell-depleted allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2007;13:1216–1223
- . Early lymphocyte recovery post-allogeneic hematopoietic stem cell transplantation is associated with significant graft-versus-leukemia effect without increase in graft-versus-host disease in pediatric acute lymphoblastic leukemia. Bone Marrow Transplant. 2008;41:245–252
- . The relationship between circulating natural killer cells after reduced intensity conditioning hematopoietic stem cell transplantation and relapse-free survival and graft-versus-host disease. Haematologica. 2008;93:1852–1858
- 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. 2007;21:2145–2152
- . In search of the “missing self”: MHC molecules and NK cell recognition. Immunol Today. 1990;11:237–244
- Donors with group B KIR haplotypes improve relapse-free survival after unrelated hematopoietic cell transplantation for acute myelogenous leukemia. Blood. 2009;15(113):726–732
- Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants. Science. 2002;295:2097–2100
- 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
- Determinants of antileukemia effects of allogeneic NK cells. J Immunol. 2004;172:644–650
- Low number of donor activating killer immunoglobulin-like receptors (KIR) genes but not KIR-ligand mismatch prevents relapse and improves disease-free survival in leukemia patients after in vivo T-cell depleted unrelated stem cell transplantation. Transplantation. 2006;82:1024–1030
- . Population frequencies and putative haplotypes of the killer cell immunoglobulin-like receptor sequences and evidence for recombination. Transplantation. 1999;68:1784–1789
- 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
- Middleton D, Menchaca L, Rood H, Komerofsky R. New allele frequency database: http://www.allelefrequencies.net. Tissue Antigens. 2003;61:403–407.
- 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
- Factors associated with early molecular remission after T cell-depleted allogeneic stem cell transplantation for chronic myelogenous leukemia. Blood. 2006;107:1688–1695
- Role of natural killer cell alloreactivity in HLA-mismatched hematopoietic stem cell transplantation. Blood. 1999;94:333–339
- Successful adoptive transfer and in vivo expansion of human haploidentical NK cells in patients with cancer. Blood. 2005;105:3051–3057
- Clinical-grade ex vivo-expanded human natural killer cells up-regulate activating receptors and death receptor ligands and have enhanced cytolytic activity against tumor cells. Cytotherapy. 2009;11:341–355
- Purified donor NK-lymphocyte infusion to consolidate engraftment after haploidentical stem cell transplantation. Leukemia. 2004;18:1835–1838
Financial disclosure: See Acknowledgments on page 1263.
PII: S1083-8791(10)00109-6
doi:10.1016/j.bbmt.2010.03.004
Published by Elsevier Inc.
Volume 16, Issue 9 , Pages 1257-1264, September 2010
