The Effect of Natural Killer Cell Killer Ig-Like Receptor Alloreactivity on the Outcome of Bone Marrow Stem Cell Transplantation for Severe Combined Immunodeficiency (SCID)
- First Online:
- Cite this article as:
- KELLER, M.D., CHEN, D., CONDRON, S.A. et al. J Clin Immunol (2007) 27: 109. doi:10.1007/s10875-006-9058-7
- 73 Views
Natural killer (NK) cell alloreactions against recipient cells in the setting of bone marrow transplantation have been associated with decreased rates of graft-versus-host disease (GVHD) and improved survival in transplant recipients with myeloid leukemia. These alloreactions are predicted by the absence of recipient HLA class I ligands for donor inhibitory killer Ig-like receptors (KIR). We hypothesized that donor NK cell alloreactions against recipient cells may affect the development of T and B-cell functions and incidence of GVHD in infants with severe combined immunodeficiency (SCID). Of the 156 patients with SCID who had received related bone marrow transplants without pretransplant chemotherapy or posttransplant GVHD prophylaxis, 137 patient–donor pairs were evaluated for the absence of recipient HLA class I ligands for donor inhibitory KIR. Analysis showed that the absence of a KIR ligand had no effect on the incidence or severity of GVHD (R2 = 0.95, p = 0.84), development of T-cell function (R2 = 1.05, p = 0.69), production of IgA (p = 0.46) or IgM (p = 0.33), or on 5-year survival (R2 = 1.21, p = 0.10). Further, in patients possessing native NK cells, the absence of KIR ligands in donors for recipient-inhibitory KIR did not alter transplantation outcomes. This study suggests that inhibitory KIR/HLA interactions do not play a significant role in bone marrow transplantation for SCID.
KEY WORDSNatural killer cellsImmunodeficiencyKiller Ig-like receptorsHematopoietic stem cell transplantation
Severe combined immunodeficiency (SCID) is a fatal genetic disorder characterized by the presence of lymphopenia, absence of T cells, and extreme susceptibility to infection (1). Since its initial description, 12 distinct genes have been shown to result in SCID when mutated (1). The incidence of SCID has been estimated at between 1 in 30,000–70,000 live births (2), however no studies have been performed to establish the true rate. The current treatment of choice is hematopoietic bone marrow stem cell transplantation (HCT) which, if given during the first 3.5 months of life, has been shown to result in long-term survival of up to 96% of patients (3). Although HLA-identical related transplants are ideal, methods of rigorous T-cell depletion have made haploidentical parental marrow transplantation a successful alternative (1, 3, 4), with relatively low rates and severity of graft-versus-host disease (GVHD). Due to their lack of T cells, patients with SCID cannot mount an adaptive immune response to allografts, thus allowing HCT to be performed successfully without preconditioning by irradiation or chemotherapy.
Natural killer (NK) cells have been well-studied in the field of HCT (5–7). As a link between adaptive and innate immunity, NK cells are able to function as either cytokine producers or cytotoxic effectors. NK cells can mediate cytotoxicity by both antibody-dependent means (ADCC) or via NK receptor families, many of which interact with MHC class I ligands. NK receptor families include the C-type lectin-like family, natural cytotoxicity receptors (NCR), and Killer-Ig-like receptors (KIR), which make up the “NK synapse” (8). Both the KIR and C-type lectin-like families possess activating and inhibitory receptors, which are expressed variably on subsets of NK cells. These receptors are believed to trigger reactions in response to missing individual class I MHC ligands on target cells, which may occur as a result of viral infection or tumor (9). With regard to KIR inhibitory receptors, the interaction of KIR3DL1 with HLA-Bw4 has been well described (10), as has the interaction of KIR2D with HLA-C, with the specificity for KIR2DL1 versus KIR2DL2/3 dependent on the presence of either a Ser77Asn80 or Asn77Lys80 motif, respectively, in HLA-C (11).
SCID resulting from mutations in common γ chain, Janus kinase 3 (JAK3), or adenosine deaminase (ADA) is characterized by low to absent NK cells, presumably due to an inability to signal through the IL-15 receptor in the case of γc and JAK3 deficiencies (12) or to destruction by toxic metabolites in the case of ADA deficiency (13). Other forms of SCID generally have normal NK cell presence and function. Following HCT for SCID, it has been seen that NK cells are among the first lineages to appear following engraftment (14). These cells generally have normal cytotoxic function (unpublished observations), and previous studies in transplanted SCID infants showed these early NK cells to be of donor origin (15 and unpublished observations). Analyses of KIR reconstitution following HCT have given mixed results (16–18), though several investigations have noted that the KIR expression pattern reapproximates the donor's KIR repertoire within 3 months of HCT when pretransplant conditioning is given. The absence of an MHC class I ligand in the recipient for donor KIR may allow the development of potentially alloreactive subsets of engrafted NK cells, the clinical impact of which was demonstrated by Ruggeri et al. in recipients of HCT for myeloid leukemia, for whom the absence of an inhibitory KIR ligand in recipients for donor KIR resulted in significantly decreased tumor relapse, decreased rates of GVHD, and increased survival rates (19). Mouse models have further shown the power of KIR alloreactivity in preventing GVHD and facilitating successful engraftment of marrow allografts, possibly by mediating the killing of host antigen presenting cells (20). Subsequent retrospective studies have confirmed the effect in myeloid leukemia, although effects in HCT for other conditions have been less clear (21–24). No large studies have been performed examining KIR effects on bone marrow transplantation for human SCID.
Of further interest is the role of the murine Ly49 family in the “hybrid resistance” mouse model, in which lethally irradiated F1 hybrid mice will reject parental marrow allografts in an NK-dependent manner (25–27). Clinical observations suggesting increased resistance to engraftment following HCT for SCID in patients possessing normal NK cells has led to the postulate that human NK cells may similarly mediate graft rejection (28). Despite their lack of homology, the similar functions of Ly49 and KIR have caused some to hypothesize that host-versus-graft alloreactions in this context could be mediated by KIR (29).
In this study, we explored whether absence of recipient KIR ligands for donor KIR resulted in beneficial graft-versus-host alloreactions that affected clinical outcomes in transplantation for SCID. We further studied whether KIR-mediated HVG reactions may be occurring in SCID patients who possess NK cells.
A total of 156 patients with SCID received allogeneic bone marrow transplants at this institution from 5/13/82 through 9/11/06, of which only 137 could be studied due to genetic material limitations. Median age at transplantation was 165 days (range: 7–597 days). Molecular basis of SCID was γc deficiency in 64, IL-7Rα deficiency in 15, ADA deficiency in 17, JAK3 deficiency in 12, RAG1 or RAG2 deficiency in 7, CD3 chain deficiencies in 4, and Artemis deficiency in 2. Thirteen patients had unidentified defects that were inherited in autosomal recessive patterns, and the remaining 3 males had unknown defects.
Informed consent was obtained from the parents of all participants, and all protocols were approved annually by the Duke University Institutional Review Board.
Thirteen of the patients received bone marrow transplants from HLA-identical related donors, and 124 patients received T-cell depleted haploidentical transplants from related donors. Bone marrow donors included parents, siblings, grandparents, and uncles. T-cell depletion was performed via soybean lectin agglutination and sheep erythrocyte rosetting as previously described (4). The mean number of nucleated bone marrow cells given was 2.6 × 108/kg body weight. None of the patients received pretransplant chemotherapeutic conditioning nor posttransplant graft-versus-host disease (GVHD) prophylaxis. Eight of these patients died prior to 100 days posttransplantation, with viral infections (including EBV lymphoproliferative disease in two) being the final causes of death.
Of the remaining patients, 111 (86%) survived, of which 95 developed T-cell function within a year of transplantation. Twelve patients went on to receive further transplantation, while two are maintained on PEG-ADA (30) and two patients underwent successful gene therapy (31).
Genomic DNA Preparation and HLA and KIR Typing
Patient and donor DNA was obtained from peripheral blood mononuclear cells (PBMCs) and EBV-transformed B-cell lines using a DNAeasy kit (Qiagen, Valencia, CA). All posttransplant samples were tested by restriction fragment length polymorphism to ensure host origin. HLA compatibility was initially analyzed by either serologic or molecular sequence-specific oligonucleotide (SSO) typing at the HLA-A, B, C, DR, and DQ loci (One Lambda, Canoga Park, CA). All serologic data were subsequently reconfirmed via SSO typing. High resolution HLA-C data were obtained via DNA sequencing. Inhibitory and activating KIR alleles were analyzed in donors and patients by high-resolution SSO typing (One Lambda, Canoga Park, CA).
Analysis of KIR/HLA Compatibility
Patients were divided into groups dependent on the presence or absence of HLA-Bw4, HLA-C Ser77Asn80 (Group C1), and HLA-C Asn77Lys80 (Group C2). The absence of a KIR ligand in the recipient (Bw4, C1 or C2) for inhibitory KIR present in the donor (KIR3DL1, KIR2DL2/3, KIR2DL1, respectively) was defined as having the potential for a KIR alloreaction. If a donor-inhibitory KIR allele was absent, no alloreaction was defined regardless of the recipient HLA status. Conversely, in order to define possible host-versus-graft KIR alloreactions, host KIR alleles and donor HLA-Bw4 and HLA-C were compared in transplantations where the host possessed natural killer cells.
Lymphocyte proliferation was determined by levels of 3H-thymidine incorporation into patient lymphocytes following stimulation by mitogens phytohemagglutinin (PHA), concavalin A, and pokeweed mitogen. All proliferation assays were run in parallel with identical studies with PBMCs from healthy adult volunteers. Flow cytometry via fluorochrome-conjugated monoclonal antibody staining was used to classify donor and recipient PBMCs using a FACSCalibur (BD Biosciences, San Jose, CA). KIR expression was determined with phycoerythrin-conjugated monoclonal antibodies EB6 (anti-CD158a,h), GL183 (anti-CD158b1/b2,j), and Z27 (anti-CD158e1/e2) (Beckman Coulter, Fullerton, CA). Additional monoclonal antibodies against CD3, CD16, CD56, and CD45 (BD Pharmingen, San Jose, CA) were also utilized.
The primary endpoints measured in this study were T-cell function at 1 year and 5 years, 5-year survival, incidence and severity of graft-versus-host disease, and B-cell function at 1 year and 5 years. T-cell function was determined by lymphocyte proliferation assays, with a PHA stimulation index >100 considered to be normal function. The Kaplan–Meier method was used to determine 5-year survival. GVHD was defined on the 0–4 clinical grading scale (32). B-cell function was determined by measuring immunoglobulin concentrations and antibody titeres to vaccines as well as the requirement for exogenous immunoglobulin therapy.
Risk ratios were used to measure associations between missing KIR ligands and T-cell function, GVHD incidence, and 5-year survival, and were further used to subanalyze groupings by specific missing KIR ligand and by SCID molecular defect. These were created using SAS Ver. 8.2 (SAS Institute, Cary, NC) with the PROC GENMOD procedure (“log” link). Relationships between absence of KIR ligands and immunoglobulin values were calculated using ANOVA through the PROC GLM procedure in SAS Ver. 8.2. Associations between activating and inhibitory KIR pairs and clinical outcomes (GVHD incidence, T-cell function at 1 year, and 5-year survival) were analyzed via the chi-squared statistic in SAS Ver 8.2. All tests were performed at the standard significance level of p < 0.05.
Donor KIR and Recipient HLA Compatibility
Recipients with present KIR ligands, no. (%)
Recipients with missing KIR ligands, no. (%)
X-linked (γc Def)
IL-7R alpha Def
CD3 chain Def
RAG1 or RAG2 Def
T-cell function, 1 year
SI > 100 (PHA)
KIR Ligand Absence and HCT Outcomes
Analysis of HCT recipients by presence or absence of ligands for donor KIR showed no significant effect on development of T-cell function at one year posttransplantation (n = 135, R2 = 1.05, p = 0.69; Fig. 2 and Table II). Similarly, neither the incidence nor severity of GVHD was affected by host KIR ligand absence (n = 137, R2 = 0.95, p = 0.84), regardless of the specific missing ligand. Analysis of 5-year survival showed that the absence of KIR ligands was associated with a favorable survival effect (Figs. 2 and 3) which did not reach significance in the overall population (n = 111, R2 = 1.21, p = 0.11). Analysis of haploidentical transplants alone showed similar statistically insignificant improvement in 5-year survival (n = 98, R2 = 1.23, p = 0.13), although the exclusion of patients who died prior to 100 days posttransplant diminished this effect (n = 103, R2 = 1.13, p = 0.23). This improvement did not appear to be dependent on any specific missing KIR ligand, nor did any particular ligand produce a significant survival effect.
Further risk ratio analysis of the recipients by KIR ligand status and specific molecular defect similarly showed no statistical difference in clinical outcomes (Table III). Notably, patients who possess defects in γc (representing the majority of the cohort) showed virtually unchanged rates of T-cell function, survival, and GVHD regardless of the presence or absence of KIR ligands.
Missing KIR Ligand Effect on Transplantation Outcomes by Ligand Type
T-cell function (1 yr)
R2 (95% CI)
R2 (95% CI)
R2 (95% CI)
Missing HLA-Bw4 and C2
Overall missing KIR ligand
KIR Ligand Effects in Sibling Pairs Following Maternal Haploidentical HCT
In several families with more than one sibling affected with SCID, the same donor was used for bone marrow transplantation for both siblings. Within these pairs (all of whom developed T-cell function within a year of transplantation), the presence or absence of KIR ligands in the recipient did not correspond with improvements in GVHD or development of B-cell function (as determined by sufficient immunoglobulin production to allow discontinuation of IVIG, Table V). The number of mismatched HLA Class I and II antigens in a graft-versus-host direction did not correlate with incidence of GVHD, nor did the number of HLA Class I and II mismatches in a host-versus-graft direction correlate with development of T-cell or B-cell function.
Activating KIR and HCT Outcomes
Missing KIR Ligand Effects on Haploidentical Transplantation by Specific Molecular Defect
T-cell function (1 yr)
R2 (95% CI)
R2 (95% CI)
R2 (95% CI)
Effect of KIR Ligand Absence on Maximal IgA/IgM Levels (n = 122)
Present KIR ligands mean (mg/dL) ± SD
Missing KIR ligands mean (mg/dL) ± SD
Maximal IgA at 1 year posttransplantation
30.8 ± 47.1
97.1 ± 288.9
Maximal IgM at 1 year posttransplantation
103.5 ± 106.9
187.7 ± 449.5
Recipient KIR/Donor HLA Effects on T-Cell Function in NK-Positive SCID Patients
Of the 22 recipients with NK+ SCID (IL-7Rα, CD3 chain, RAG1/2, or Artemis deficiency), analysis of donor HLA and recipient KIR showed that six donors had all appropriate KIR ligands, while 18 were missing one or more ligands. No correlation was found between donor KIR ligand absence and failure of development of T-cell function in the recipient (n = 22, R2 = 0.60, p = 0.17). Donor KIR ligand absence also had no effect on incidence of GVHD (n = 20, R2 = 0.81, p = 0.79) or 5-year survival (n = 17, R2 = 1.03, p = 0.94).
Studies of KIR alloreactions have shown a beneficial effect on myeloid leukemia following HCT, although additional studies have not yet shown a similar effect in hematopoietic transplantation for other conditions. No previous studies have examined the effect of KIR on marrow transplantation in primary immunodeficiency diseases, and none have focused on transplantation without pretransplant conditioning of any type. Posttransplant GVHD prophylaxis was also used in several studies (23, 24), which may diminish alloreactions.
This study was designed to determine if KIR alloreactivity affects bone marrow stem cell transplantation in SCID infants who received both HLA-identical and haploidentical related donor marrow without pretransplant chemotherapy or posttransplant GVHD prophylaxis. Although KIR and NK reconstitution patterns resembled previous studies of HCT for other conditions, no correlation was found between KIR ligand compatibility and rates of GVHD or development of T-cell function. The lack of correlation persisted regardless of the specific molecular defect producing the SCID phenotype, and additionally was reflected in the outcomes of sibling pairs who underwent haploidentical HCT from the same donor. Although an improvement in 5-year survival approached significance in those lacking KIR ligands, the effect was lessened by exclusion of patients who died prior to 100 days posttransplant. These results strongly suggest that KIR play a very limited role in the setting of HCT for treatment of SCID.
It is possible that the lack of an effect on GVHD may be due to the differences in transplantation procedure, as several publications point out that the tissue damage and inflammation caused by preparative chemotherapy or radiation plays a large role in the initiation of GVHD by host antigen presenting cells(APC) (33, 34). Since pretransplant conditioning was not used in the SCID patients treated here, the depletion of host APCs by alloreactive NK cells may not have an appreciable effect on GVHD possibly because the role of APCs in initiation may be reduced. Further, a recent study showed that production of IFN-γ by matched unrelated donor NK cells correlated strongly with the incidence of GVHD in the posttransplantation period (16), again in a situation where pretransplant conditioning was used. It has been shown that the primary producer of IFN-γ within the NK population is the CD56bright subset, which at baseline, express little or no KIR (35).
KIR Effects on Maternal Haploidentical Transplantations into Sibling Pairs
Total HLA Ag mismatches
No missing ligands
No missing ligands
No missing ligands
No missing ligands
It has been shown in the “hybrid resistance” model that natural killer cells can cause rejection of marrow allografts in mice after lethal irradiation, and subsequent work attributed this effect in part to Ly49, a KIR homologue on murine NK cells (5). This study showed no change in GVHD or survival with missing KIR ligands in donors, while a nonsignificant decrease in development of T-cell function at 1-year posttransplantation was seen (n = 22, R2 = 0.60, p = 0.17). Though this limited power study may suggest that increased resistance to engraftment occurs with absence of KIR ligands in donors for the recipient KIR, the lack of impact on 5-year survival (n = 17, R2 = 1.03, p = 0.94) suggests that either this effect is truly insignificant, or that the resulting resistance is not severe enough to influence mortality.
This study suggests that KIR alloreactivity neither helps nor hinders HCT for SCID. Accordingly, KIR typing of donors of HCT for SCID does not appear to be clinically useful. It is possible that the inherent differences in the stem cell transplants in this cohort compared to HCT given to leukemia patients, most notably the lack of preconditioning regimens and posttransplant GVHD prophylaxis, may allow other factors to predominate over KIR alloreactivity in affecting clinical outcomes.
We believe that the risks of infection and other serious adverse effects following myeloablative conditioning and immunosuppressive agents given for GVHD prophylaxis far outweigh the risks of graft failure and the relatively low risk of high-grade GVHD following related haploidentical HCT for SCID, particularly given the success rates achieved (1, 3, 4). Whether selection of donors by both KIR genotype and KIR serologic data will yield improved transplantation outcomes in conditions other than myelogenous leukemia is not yet known. Further research, particularly in regard to the relationship between KIR genotype and phenotype, may determine if it will be possible to extend the clinical benefits of KIR alloreactivity to a broader variety of conditions treated by HCT.
This research was supported by NIH Grant 5R01-AI/HD042951-07 and the Howard Hughes Medical Institute Research Fellows Program. The authors would like to thank Ms. Adella Clark, Ms. Myriah Cooney, Ms. Donna Oliver, Ms. Roberta Parrott, and Ms. Elisa Sajaroff, for their excellent technical assistance, and Ms. Lora DeRubeis for her administrative work. We would also like to thank Dr. Eric Long, Dr. Sumi Rajagopalan, and Dr. Andreas Velardi for kindly sharing their protocols and advice. We are also grateful to Dr. Peter Parham for his generosity in providing cell lines, and Steven Grambow for his expertise in our statistical analyses.