KIR Genes and KIR Ligands Affect Occurrence of Acute GVHD After Unrelated, 12/12 HLA Matched, Hematopoietic Stem Cell Transplantation

ORIGINAL ARTICLE KIR genes and KIR ligands affect occurrence of acute GVHD after unrelated, 12/12 HLA matched, hematopoietic stem cell transplantation

K Ludajic1, Y Balavarca2, H Bickebo¨ ller2, A Rosenmayr1, I Fae1, GF Fischer1, M Kouba3, D Pohlreich3, P Kalhs3 and HT Greinix3

1Division of Blood Group Serology, Medical University of Vienna, Vienna, Austria; 2Department of Genetic Epidemiology, University of Goettingen, Goettingen, Germany and 3Department of Internal Medicine I, Bone Marrow Transplantation, Medical University of Vienna, Vienna, Austria

Interactions of polymorphic killer Ig-like receptor (KIR) Introduction receptors with KIR ligands have been shown to modify the outcome of hematopoietic SCT (HSCT). The association The killer Ig-like receptor (KIR) family includes 16 genes of these genetic factors with different transplantation coding for inhibitory (L) and activatory (S) receptors endpoints, however, varies substantially, depending on expressed on natural killer cells (NK) and subpopulations clinical and study setup variables. We aimed to assess of gd and ab T cells.1 The genomic diversity in the KIR whether KIR ligands, KIR genes and KIR haplotypes are region is attributed to the variability in both gene numbers associated with HSCT outcome of 124 patients with and allelic polymorphism. Population studies revealed various hematological malignancies, transplanted with 12/ more than 100 different KIR genotypes,2–5 whereas in 12 HLA matched grafts from unrelated donors. For this family studies over 30 distinct KIR haplotypes were purpose, patient and donor KIR gene and KIR ligand identified.6 polymorphisms were determined and correlated with KIR haplotypes fall into two broad categories: group A clinical data in simple and multiple models. We found haplotypes have a restricted gene content of 2DL1, 2DL3, that a missing HLA-C2 ligand for donor inhibitory 2DL4, 3DL1, 3DL2, 3DL3, 2DS4, 2DP1 and 3DP1 genes, KIR2DL1 was significantly associated with an increased whereas group B haplotypes may contain other inhibitory risk of acute GVHD (aGVHD) (II–IV) (hazard ratio and activatory KIR genes. (HR) ¼ 2.23, 95% confidence interval (95% CI): 1.21– Some inhibitory KIR receptors interact with HLA class I 4.10, P ¼ 0.010), as were the AA KIR haplotypes in ligands. Inhibitory receptor KIR2DL1 interacts with HLA- patients and donors in HLA-C1CX (HR ¼ 2.37, 95% CI: Cw group 2, and KIR2DL2/2DL3 withHLA-Cw group 1.16–4.84, P ¼ 0.018) and in HLA-Bw4À (HR ¼ 3.20, 1 epitopes defined by the amino-acid lysine and asparagine 95% CI: 1.35–7.60, P ¼ 0.008) patients. On the contrary, at position 80, respectively. KIR3DL1 specifically recog- transplantation of HLA-C1C2 patients with KIR2DS2 nizes the HLA-Bw4 and KIR3DL2 HLA-A3/A11 epitopes. positive grafts were associated with a decreased risk of The strength of these interactions varies, with aGVHD (II–IV) (HR ¼ 0.24, 95% CI: 0.07–0.85, HLA-C2/KIR2DL1 being the strongest, followed by P ¼ 0.027). Thus, our single center study provides HLA-C1/KIR2DL2, HLA-C1/KIR2DL3 and HLA-Bw4/ evidence for the modification of aGVHD risk by KIRs KIR3DL1.7 and their ligands. Recognition of self-HLA ligands by inhibitory KIR Bone Marrow Transplantation (2009) 44, 97–103; receptors ensures unresponsiveness of NK cells. Inability of doi:10.1038/bmt.2008.432; published online 26 January 2009 NK cells to recognize self-HLA class I leads to increased Keywords: KIR receptors; KIR ligands; KIR haplotypes; cytokine production or cytotoxicity.8 acute GVHD The role of NK cells in the treatment of hematological malignancies is subject to discussion. Ruggeri et al.,9 described clinically beneficial NK cell alloreactivity in HLA-haploidentical hematopoietic SCT (HSCT) where donors’ inhibitory NK cell receptors were not engaged by recipients’ ligands. This ‘missing ligand’ effect proved beneficial for AML patients even after HLA matched transplantation,10 and after autologous transplantation for Correspondence: Dr GF Fischer, Department of Blood Group Serology, treatment of solid tumors and lymphomas,11 but different Medical University of Vienna, Waehringer Guertel 18–20, Vienna outcomes have also been reported (reviewed by Witt and A-1090, Austria. 12 E-mail: gottfried.fi[email protected] Christiansen ). Concordant withthe‘missing ligand’ Received 31 July 2008; revised 28 November 2008; accepted 2 December model, decreased survival of HLA-C2C2 CML patients 2008; published online 26 January 2009 was attributed to the delayed reconstitution of inhibitory KIRs and aGVHD in HSCT K Ludajic et al 98 KIR2DL1 receptors on NK cells post transplant, and hence Table 1 Patient and donor characteristics and transplantation their inability to kill cells with downregulated HLA-C2C2 variables 13 ligands in a timely manner. Also, the analyses of the Number of patients 124 activating KIR receptors showed donors’ KIR2DS2 to be a Age of patients; years, mean (range) 42 (18–63) significant risk factor for overall survival (OS) of myeloid Age of donors; years, mean (range)a 36 (18–56) leukemia HLA-C2C2 patients.14 In addition, not only individual genes but also KIR haplotypes of donors were Sex (patient/donor); no. (%) F/F 28 (23) associated withan increased risk of acute GVHD F/M 29 (25) (aGVHD), higher relapse rates and worse OS of HLA- M/F 16 (13) C1C1 patients diagnosed withmyeloid diseases. 15 However, M/M 47 (39) all described observations apply to distinct patient groups, CMV (patient/donor); no. (%) and differ substantially depending on transplant protocols, Pos/Pos 32 (26) 16 as reviewed recently by Verheyden and Demanet. Pos/Neg 41 (33) The purpose of this study was to analyze the association Neg/Pos 20 (16) of patient and donor KIR ligands, KIR genes and Neg/Neg 31 (25) haplotypes on acute and chronic GVHD (cGVHD), Diagnosis; no. (%) transplant-related mortality (TRM), relapse and OS of Myeloid diseases 72 (58) 124 adult patients withhematologicaldiseases transplanted Lymphoid diseases 32 (26) at our center. ‘Other’ diseases 20 (16)

Disease status at transplantation; no. (%) Standard risk 67 (54) Patients, materials and methods Highrisk 57 (46) Conditioning regime; no. (%) Study population Myeloablative 90 (73) The study population consisted of 124 consecutive patients Reduced intensity 34 (27) who had undergone HSCT from 12/12 HLA (HLA-A, -B, -C, -DRB1, -DRB3/4/5, -DQB1) allele matched unrelated ATG; no. (%) Yes 30 (24) donors between September 1995 and December 2005. The No 94 (76) median follow up time of the patients alive was 80 months (range 35–158 months). Data were analyzed as of 20 Stem cell source; no. (%) September 2008. BM 54 (43.5) Patients’ and donors’ characteristics are shown in PBSC 70 (56.5) Table 1. Disease groups analyzed in statistical models were Cell dose infused: CD34+ Â 106/kg median (range)b 5.5 (1–19) ‘myeloid’ withAML ( n ¼ 37) and CML (n ¼ 35), ‘lymphoid’ with ALL (n ¼ 19), non-Hodgkin’s lymphoma Post transplant immunosuppressive therapy; no. (%) (NHL; n ¼ 11), Hodgkin’s lymphoma (n ¼ 1) and CLL Cyclosporine-methotrexate 89 (71.8) Cyclosporine-mycophenolate mofetil 33 (26.6) (n ¼ 1), and ‘other’ with myelodysplastic syndrome (n ¼ 14), Cyclosporine 2 (1.6) severe aplastic anemia (n ¼ 1), multiple myeloma (n ¼ 2) and myeloproliferative disorders (n ¼ 3). The disease status Abbreviations: ATG ¼ anti-thymocyte globulin; F ¼ female; M ¼ male; was either ‘standard risk’, comprising CML in the first Neg ¼ negative; no. ¼ number; Pos ¼ positive. chronic phase, acute leukemias and NHL in CR or ‘high aAge and sex of 4 donors not known. b risk’ containing all other disease stages. In vivo T-cell 31 entries missing. depletion (TCD) was performed by anti-thymocyte globulin (Fresenius, Bad Hamburg, Germany) administration in DNA was isolated either by the salting-out method or by 24% of patients. The source and numbers of transplanted using an automated DNA extraction device (GenoM-6; stem cells, types of post transplant immunosuppressive GenoVision VertriebsgesmbH, Vienna, Austria). HLA therapy and conditioning therapies used are also shown in class I and class II genes were typed to the high-resolution 19 Table 1. GVHD was defined as acute if occurring within the level by sequencing based typing. first 79 days, and as chronic if occurring at a later time Sequence specific KIR primers (SSP-KIR) and amplifi- point. Due to three missing onset dates, 121 patients were cation conditions used for typing of KIR genes 2DL4, evaluated for aGVHD. As 22 patients died before day 79, 3DL1, 3DL2, 3DL3, 2DL3, 2DS1, 2DS4a, 1D, 2DS5, 3 102 patients were evaluated for cGVHD. TRM was defined 3DS1, 2DP1 and 3DP1 were published by Hsu et al., for 4 as death due to causes other than relapse. Criteria for genes 2DL2, 2DL5, 2DS2, 2DS3, 2DS4 by Uhrberg et al. 5 clinical diagnoses of acute and cGVHD were previously and for KIR2DL1 by Velickovic et al.. Eachtyping described.17,18 included positive and negative controls for eachgene. No phenotype analyses were performed.

HLA and KIR genotyping Blood samples for tissue typing were obtained upon written Genotype and haplotype analysis consent from patients and donors, and approved by the We have compiled a list of over 100 previously published ethical committee of the Medical University of Vienna. genotypes.2–5,20 KIR genotype assignment of individuals,

Bone Marrow Transplantation KIRs and aGVHD in HSCT K Ludajic et al 99 based on the presence or absence of 14 activatory and ligands) were considered as potential risk factors for the inhibitory KIR genes was done by specially developed occurrence of acute and cGVHD, relapse, TRM and OS. software. Only one genotype, positive for 2DL1, 2DL3, Acute and cGVHD were also considered as potential 2DL4, 2DL5, 3DL1, 3DL2, 3DL3, 2DS4, 2DS5, 3DS1, factors impacting relapse, TRM and OS. Clinical factors 2DP1 and 3DP1 and negative for other genes had not been, with P values less than 0.20 by simple log-rank tests were to our knowledge, previously described. KIR genotyping of included in a multiple Cox-regression for model selection that specific donor was confirmed using Olerup SSP KIR purposes. Then, the set of clinical variables with significant Genotyping Kit (Qiagen, Vienna, Austria). Following influence on the outcome in the final Cox-regression model genotype assignment, 23 haplotypes (1–23) published by was identified by the stepwise selection. Before these Hsu et al.3 were used to identify haplotype pairs for each of analyses, age was dichotomized using Classification and our patients and donors. Due to the lack of pseudogene Regression Trees (CART).21,22 The effect of each genetic variant typing and/or incomplete lists of haplotypes, ambi- factor was assessed by simple and by multiple Cox- guities were not resolved for 19 individuals. However, broader regression models. In the latter case, the multiple model assignment of either A (1–2) or B (3–23) haplotypes was included the significant clinical factors previously obtained, possible in each case. The associations of KIR haplotypes plus one genetic factor, so that the effect of each genetic withtransplantation outcome were analyzed by distinguishing factor was adjusted by the effects of the significant clinical between AA or BX haplotypes, where X stands for either A factors. or B haplotype. Similarly, in the genotype HLA-C1CX, CX The competing risk method for survival analysis23 was stands for either C1 or C2 allelic groups. carried out to assess the single effect of the genetic factors. Relapse and TRM are competing risks for acute and Study design cGVHD, and TRM is a competing risk for relapse and vice To assess the effect of KIR and KIR ligands on versa. transplantation endpoints, four models (listed in Table 2) To address the question whether dAA/pAA HLA-C1CX were independently tested: contributes to the effect on aGVHD (II–IV) independently of dAA/pAA HLA-Bw4À, the risk of aGVHD (II–IV) in the subgroup dAA/pAA, HLA-C1CX, -Bw4 þ patients was (1) Association of KIR ligands of patients withtransplant- assessed. ation endpoints, regardless of donors’ KIR type. Clinical variables are reported only if reaching signifi- (2) Association of ‘missing ligand’, that is, the absence of cance in the final multiple models. Hazard ratios (HR) and patients’ HLA ligands for the corresponding donor 95% confidence intervals (95% CI) were calculated in each inhibitory KIR genes, with transplantation endpoints. analysis. The significance level was set to 0.05 without (3) Association of donor activatory KIR genes with adjustment for multiple testing. transplantation endpoints, independently and in combination withKIR ligands of patients. (4) Association of KIR haplotypes with transplantation endpoints, independently and in combination withKIR Results ligands of patients. Overall aGVHD was diagnosed in 62 (51%) patients, including 44 (36%) patients withaGVHD (II–IV). Median Statistical analysis time to onset of aGVHD was 16 days (range 7–52 days). All patient and donor characteristics, and genetic variables Limited cGVHD was diagnosed in 9 (9%), and extensive in (patient and donor KIR genes and haplotypes, patient KIR 27 (26%) patients. Median time to onset of cGVHD was 176 days (range 80–840 days). Of the patients, 40 (32%) Table 2 Genetic variables associated withaGvHD (II–IV), as experienced relapse by the end of the follow up, whereas 25 assessed by multiple analyses in four different models patients (20%) died of transplant related causes. By the end Model/variable P-value HR 95% CI of the follow-up period, 67 patients (54%) were alive. Multiple analyses of genetic and clinical variables of KIR Ligands patients and donors revealed their association with HLA-C1C1 0.055 3.26 0.97–10.91 aGVHD (II–IV). Missing ligands Missing C2 for donors’ KIR2DL1 0.010 2.23 1.21–4.10 Association of patients’ KIR ligands with acute GVHD aKIR/KIR ligands (II–IV) dKIR2DS2, HLA-C1C2 0.027 0.24 0.07–0.85 Among 121 patients, 51 (42%) had HLA-C1C1, 20 (17%) KIR haplotypes/KIR ligands HLA-C2C2 and 50 (41%) had HLA-C1C2 genotypes; 67 dAA/pAA 0.050 2.18 1.00–4.75 (55%) patients had HLA-Bw4 and 45 (37%) HLA-A3/A11. dAA/pAA, HLA-Bw4À 0.018 2.37 1.16–4.84 The simple Cox analysis showed a significant association dAA/pAA, HLA-C1CX 0.008 3.20 1.35–7.60 of HLA-C1C2 (HR ¼ 0.50, 95% CI: 0.27–0.95, P ¼ 0.033), HLA-C2C2 (HR ¼ 0.22, 95% CI: 0.07–0.74, P ¼ 0.015) and Abbreviations: CI ¼ confidence interval; d ¼ donor; HR ¼ hazard ratio; HLA-Bw4À genotypes (HR 0.45, 95% CI: 0.24–0.82, KIR ¼ killer Ig-like receptor; p ¼ patient. ¼ EachKIR/KIR ligand combination was compared to all others. P-values, P ¼ 0.010) withdecreased risk of aGVHD (II–IV). Adjust- HR and CI are shown. ing patient losses due to deathor relapse by competing risk

Bone Marrow Transplantation KIRs and aGVHD in HSCT K Ludajic et al 100 analyses confirmed this finding: HLA-C1C2 patients Association of donors’ activatory KIR genes with acute (HR ¼ 0.41, 95% CI: 0.23–0.72, P ¼ 0.002) and HLA- GVHD (II–IV) C2C2 patients (HR ¼ 0.34, 95% CI: 0.16–0.74, P ¼ 0.006) The frequencies of activatory KIR genes in our donor and were at significantly lower risk for developing aGVHD patient populations were similar, and comparable to those (II-IV) than HLA-C1C1 patients. of other European populations (www.allelefrequencies.net). Consistent withthesefindings, multiple Cox regression Of the donors, 42 (35%) were positive for KIR2DS1, 52% analyses showed that HLA-C1C1 patients were at higher for KIR2DS2, 28% for KIR2DS3, 89% for KIR2DS4, risk for developing aGVHD (II-IV) (HR ¼ 3.26, 95% CI: 29% for KIR2DS5 and 38% for KIR3DS1. 0.97–10.91, P ¼ 0.055), this effect, however, is not reaching Simple Cox analysis showed a significant association of statistical significance (Table 2). In addition to KIR donors’ KIR2DS2 withaGVHD (II–IV) (HR ¼ 0.54, 95% ligands, type of diagnosis, stage of the disease, TCD, stem CI: 0.29–0.99, P ¼ 0.047). When taking into account cell source and age of donor were analyzed in multiple patients’ KIR ligands in the simple analyses, we observed models for an association withaGVHD (II–IV). Only TCD that the presence of donor’s KIR2DS2 in HLA-C1C2 (HR ¼ 0.26, 95% CI: 0.09–0.72, P ¼ 0.010) and donor age patients (HR ¼ 0.21, 95% CI: 0.06–0.75, P ¼ 0.016) was above 37 years (HR ¼ 0.52, 95% CI: 0.28–0.98, P ¼ 0.042) associated witha reduced risk of aGVHD (II–IV). were significantly associated withless aGVHD (II–IV). Competing risk analysis of the same parameters showed a TCD and donor age were also significantly associated with similar result (HR ¼ 0.24, 95% CI: 0.09–0.63 P ¼ 0.003). aGVHD outcomes in all other multiple analyses. After the clinical covariables were included in the multiple analyses, the presence of KIR2DS2 gene was associated witha reduced incidence of aGVHD (II-IV) in Association of ‘missing ligands’ with acute GVHD (II–IV) HLA-C1C2 patients (HR ¼ 0.24, 95% CI: 0.07–0.85, In 113 of 121 patients (91%), at least one ligand was P ¼ 0.027; Figure 1b, Table 2). missing for the corresponding inhibitory KIR of the donor. Of the patients, 20 (17%) missed HLA-C1, 48 (40%) Association of KIR haplotypes with acute GVHD (II–IV) missed HLA-C2 and 49 (41%) missed HLA-Bw4 ligands Of 121 patients, 38 (32%) and 39% of donors (47 of 121) for corresponding donors’ KIRs. Further on, when the had the AA KIR haplotypes whereas the rest of our combination of two missing ligands was analyzed, 3 (2%) population had AB or BB haplotypes (BX haplotypes). The patients missed HLA-C1 and HLA-Bw4 ligands, and 31 distribution of haplotypes is therefore comparable with (26%) patients missed bothHLA-C2 and HLA-Bw4 those reported elsewhere.4,15 combinations. We first analyzed the association of patient/donor In simple Cox analyses, missing HLA-C2 ligands for haplotype combinations—dBX/pAA (n ¼ 21, 17%), dBX/ donors’ KIR2DL1 (HR ¼ 2.80, 95% CI: 1.53, 5.10, pBX (n ¼ 53, 44%), dAA/pAA (n ¼ 17, 14%) and dAA/ P ¼ 0.0008) and missing HLA-Bw4 ligands for donors’ pBX (n ¼ 31, 25%)—withtransplantation endpoints. None KIR3DL1 (HR ¼ 1.99, 95% CI: 1.10–3.60, P ¼ 0.023) were of the patient and donor haplotype combinations was significantly associated with a higher incidence of aGVHD significantly associated withtransplantation endpoints in (II–IV). Competing risk analyses, however, confirmed only simple or in competing risk analyses. Nevertheless, after the significant association of missing ligands for donors’ adjustment by clinical covariables, the group of patients KIR2DL1 withaGVHD (II–IV) (HR ¼ 3.00, 95% CI: withKIR AA haplotypesreceiving grafts from donors with 1.83–4.93, P ¼ 0.00001). KIR AA haplotypes was at higher risk of aGVHD (II-IV) In multiple Cox regression analyses, only the absence of (HR ¼ 2.18, 95% CI: 1.00–4.75, P ¼ 0.050), withborderline ligands for donors’ KIR2DL1 was a significant risk factor significance (Table 2). for occurrence of aGVHD (II–IV) (HR ¼ 2.23, CI: 1.21– In simple Cox analyses, KIR haplotypes were analyzed 4.10, P ¼ 0.010; Figure 1a; Table 2). When the association together with HLA-C2C2, -C1CX, -Bw4 þ and -Bw4À types of two or more missing ligands withtheaGVHD (II–IV) of patients; the combinations dAA/pAA and HLA-C1CX was analyzed in simple or multiple models, an additive (HR ¼ 2.13, 95% CI: 1.05–4.31, P ¼ 0.036) and dAA/pAA effect was not observed and hazard ratios did not and HLA-Bw4À (HR ¼ 3.50, 95% CI: 1.48–8.32, P ¼ 0.005) significantly change (data not shown). were significantly associated withan increased risk of

1.0 1.0 dK1R2DS2–C1C2–, ref. 1.0 dAA/pAA, P=0.05 C1C1. P=0.01 + dK1R2DS2–C1C2 , P=0.99 dBX/pAA, P=0.85 C2C2. C1C2. ref. + 0.8 0.8 dK1R2DS2 C1C2–, P=0.51 0.8 dBX/pBx, ref + + dK1R2DS2 C1C2 , P=0.027 dAA/pBX, P=0.22 n=48 n=17 0.6 0.6 n=31 0.6 n=27 n=30 0.4 0.4 n=40 0.4 n=53 n=73 n=21 0.2 0.2 n=23 0.2 Cumulative incidence 0.0 0.0 0.0 0 20406080 020406080 0 20406080 Days after transplantation Figure 1 Cumulative incidence of aGVHD (II–IV) according to (a) the presence or absence of C2 ligand for KIR2DL1, (b) the presence or absence of donors’ KIR2DS2 genes and patients’ C1C2 genotypes and (c) the combination of AA and BX KIR haplotypes in donors (d) and patients (p).

Bone Marrow Transplantation KIRs and aGVHD in HSCT K Ludajic et al 101 aGVHD (II–IV), when compared to all other combinations risk of aGVHD (III–IV). Other reports, however, showed of KIR and HLA haplotypes. no association of missing ligands withaGVHD after HLA Adjustment by the clinical variables in Cox-regression identical sibling transplantation, neither with unmanipu- models showed that only KIR AA haplotypes in patients lated27,28 nor withTCD grafts. 10 and donors were associated withan increased risk of When analyzing activatory genes only donors’ KIR2DS2 aGVHD (II–IV) in HLA-C1CX (HR ¼ 2.37, 95% CI: 1.16– was associated withlower risk for aGVHD (II–IV), and 4.84, P ¼ 0.018) and in HLA-Bw4À (HR ¼ 3.20, 95% CI: after including KIR ligands in multiple models, the same 1.35–7.60, P ¼ 0.008), but not in HLA-C2C2 or -Bw4 þ association remained only in HLA-C1C2 patients. This is, patients (Figure 1c and Table 2). to our knowledge, the first observation of a protective effect Other KIR/HLA haplotype combinations analyzed in of the KIR2DS2 on aGVHD (II–IV). Regarding the role of the multiple models were not significantly associated with activatory KIR genes in transplantation, Gagne et al.29 aGVHD (II–IV). have reported that KIR2DS3 in donors is a potential high Considering that all 7 dAA/pAA, HLA-Bw4À patients risk marker for aGVHD. Also, donor’s KIR2DS2 was are included in the group of 17 dAA/pAA, HLA-C1CX shown to be associated with worse OS in HLA-C2C2 patients, we further explored whether the association of patients withmyeloid leukemia. 14 Although the ligands dAA/pAA, HLA-C1CX withaGVHD (II–IV) is influenced for activatory receptors are not known, our finding by the group of dAA/pAA, HLA-Bw4À patients. The concerning KIR2DS2 indicates there might be different analyses of dAA/pAA, HLA-C1CX (HLA-Bw4 þ ) haplo- molecular pathways of activation of individual activatory types did not show evidence of an association with KIR genes. increased aGVHD (II–IV) (HR ¼ 1.96, 95% CI: 0.62– The analyses of KIR haplotypes showed pAA/dAA to be 6.12, P ¼ 0.250), indicating that the significant effect significantly associated withincreased aGVHD (II–IV), initially found for dAA/pAA, HLA-C1CX was influenced and after including KIR ligands in multiple analyses this by HLA-Bw4À. association was significant only in HLA-C1 þ and in HLA- Bw4À patients. Haplotypes A lack most of the activatory genes and, similarly, NK cells of HLA-C1C1 and HLA- Chronic GVHD, relapse, transplant-related mortality Bw4À patients have a low activatory potential due to the and overall survival weak inhibitory signals during their development.30 There- In any of the four models of KIR gene/KIR ligand fore, there might be a link between those NK cells and interaction analyzed we did not find significant associations increased T-cell alloreactivity, but its exact nature still has withcGVHD, relapse, TRM or OS in simple or multiple to be elucidated. analyses. Although our results and recent findings of Cooley et al., 200831 point to a negative effect of AA donor haplotypes regarding aGVHD and OS in unrelated donor HSCT, Discussion McQueen et al.15 find B haplotypes predictive of aGVHD in sibling HSCT. Thus, the effect of the KIR This study provides evidence of an association of KIR haplotypes might depend on differences in clinical proto- genes and KIR ligands withaGVHD (II–IV), but not with cols used for sibling vs transplantations withunrelated other endpoints of HSCT, in 124 patients receiving 12/12 donors. HLA matched unrelated donors grafts. The role of NK cells in the pathophysiology of aGVHD Our analyses revealed that HLA-C1C1 tends to be a risk remains to be elucidated. aGVHD is a T-cell-mediated factor for aGVHD (II–IV). If HLA-C1C1 patients received disease, but there is evidence showing that it can be grafts from KIR2DL1 negative donors, this effect was modified by IFN-g production of NK cells and by the highly significant. Importantly, the association with the kinetics of KIR receptor reconstitution.32 It might be aGVHD cannot be attributed to HLA mismatches, as all possible that NK cells, activated through missing ligand patient/donor pairs in our study were matched at 12/12 mechanisms, create a setting in which T cells cause HLA loci. HLA-DPB1 typing, in addition, showed that this aGVHD. Sun et al.33 suggested that the combined increase of aGVHD (II–IV) is independent of the HLA- alloreactivity of T and NK cells in the graft might ‘bring DPB1 matching (data not shown). the alloimmune response beyond appropriate level and Our findings are in line withtheresults of two recent cause inferior transplantation outcome’. In addition to this studies analyzing the outcome of HLA matched sibling indirect effect, alloreactive NK cells might directly attack HSCT. McQueen et al.15 observed a trend toward a higher non-hematopoietic cells and so contribute to aGVHD. This incidence of aGVHD in HLA-C1C1 patients receiving non- hypothesis is in line with recent findings that NK cells are TCD grafts, and Verheyden et al.24 observed a significantly able to adhere to endothelial cells, which suggest they play higher incidence of aGVHD in HLA-C1C1 and ÀC2C2 a role in vascular injuries or solid organ failures (reviewed transplant patients. by Vivier et al.34). Besides, the possibility cannot be By contrast, other investigators reported either no effect excluded that the observed association of donor KIR of KIR ligands on aGVHD in AML patients,25 or receptors with increased aGVHD might be due to the decreased aGVHD (III–IV) in HLA-C1C1 CML patients.13 alloreactive KIR expressing T cells, as the signaling When the analyses were performed according to the through NK receptors on T cells was shown to affect ‘missing ligand’ concept, Miller et al.26 reported that, TCR mediated functions in B-CLL and other target similar to our findings, early CML patients were at higher cells.35–38

Bone Marrow Transplantation KIRs and aGVHD in HSCT K Ludajic et al 102 The discrepancies between studies regarding the effects of leukemia predicted by KIR and HLA genotypes. Blood 2005; KIR ligands, ‘missing ligands’, and KIR haplotypes on 105: 4878–4884. HSCT outcome could be due to the variable degrees of 11 Leung W, Handgretinger R, Iyengar R, Turner V, Holladay HLA matching. Not less important are substantial MS, Hale GA. Inhibitory KIR-HLA receptor-ligand mismatch differences in underlying disease and transplant protocols in autologous haematopoietic stem cell transplantation for Br J Cancer including conditioning regiments, composition of allograft solid tumour and lymphoma. 2007; 97: 539–542. 12 Witt CS, Christiansen FT. The relevance of natural killer cell and GVHD prophylaxis. human leucocyte antigen epitopes and killer cell immunoglo- The stringent setup of our single center study, based on bulin-like receptors in bone marrow transplantation. Vox Sang the patient/donor 12/12 HLA match and KIR gene 2006; 90: 10–20. assessment, showed a significant association of KIR/HLA 13 Fischer JC, Ottinger H, Ferencik S, Sribar M, Punzel M, genes and haplotypes with increased risk of aGVHD (II– Beelen DW et al. Relevance of C1 and C2 epitopes for IV), but not other transplant endpoints. Further studies hemopoietic stem cell transplantation: role for sequential will be necessary to understand the role of KIR genes in the acquisition of HLA-C-specific inhibitory killer Ig-like receptor. pathophysiology of aGVHD under distinct clinical condi- J Immunol 2007; 178: 3918–3923. tions. The prognostic value of our results should be 14 Cook MA, Milligan DW, Fegan CD, Darbyshire PJ, et al. validated in a larger study. Mahendra P, Craddock CF 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. Acknowledgements 15 McQueen KL, Dorighi KM, Guethlein LA, Wong R, Sanjanwala B, Parham P. Donor-recipient combinations of We thank Dejan Stokic at the Complex System Research Group, group A and B KIR haplotypes and HLA class I ligand affect ENT, Medical University of Vienna, for his help with the KIR the outcome of HLA-matched, sibling donor hematopoietic genotype database. This work was supported in part by the cell transplantation. Hum Immunol 2007; 68: 309–323. European Commission Grant MCRTN-CT-2004-512253 16 Verheyden S, Demanet C. NK cell receptors and their ligands (TRANSNET). in leukemia. Leukemia 2008; 22: 249–257. 17 Sullivan KM, Shulman HM, Storb R, Weiden PL, With- erspoon RP, McDonald GB et al. Chronic graft-versus-host References disease in 52 patients: adverse natural course and successful treatment withcombination immunosuppression. Blood 1981; 1 Uhrberg M, Valiante NM, Young NT, Lanier LL, Phillips JH, 57: 267–276. Parham P. The repertoire of killer cell Ig-like receptor and 18 Rowlings PA, Przepiorka D, Klein JP, Gale RP, Passweg JR, CD94:NKG2A receptors in T cells: clones sharing identical Henslee-Downey PJ et al. IBMTR Severity Index for grading alpha beta TCR rearrangement express highly diverse killer acute graft-versus-host disease: retrospective comparison with cell Ig-like receptor patterns. J Immunol 2001; 166: 3923–3932. Glucksberg grade. Br J Haematol 1997; 97: 855–864. 2 Flores AC, Marcos CY, Paladino N, Capucchio M, Theiler G, 19 Fischer GF, Fae I, Frey E, Mayr WR. HLA-A*02172* adds to Arruvito L et al. KIR genes polymorphism in Argentinean the heterogeneity of HLA-A*02 alleles. Tissue Antigens 1998; Caucasoid and Amerindian populations. Tissue Antigens 2007; 51: 312–314. 69: 568–576. 20 Du Z, Gjertson DW, Reed EF, Rajalingam R. Receptor-ligand 3 Hsu KC, Liu XR, Selvakumar A, Mickelson E, O’Reilly RJ, analyses define minimal killer cell Ig-like receptor (KIR) in Dupont B. Killer Ig-like receptor haplotype analysis by gene humans. Immunogenetics 2007; 59: 1–15. content: evidence for genomic diversity witha minimum of six 21 Atkinson EJ, Therneau TM. An Introduction to Recursive basic framework haplotypes, each with multiple subsets. Partitioning using the RPART Routines, vol. 61 Mayo Clinic: J Immunol 2002; 169: 5118–5129. Rochester, 1997. 4 Uhrberg M, Parham P, Wernet P. Definition of gene content 22 Breiman L, Friedman JH, Stone CJ, Olshen RA. Classification for nine common group B haplotypes of the Caucasoid and Regression Trees, 1 edn. Chapman & Hall/CRC: London, population: KIR haplotypes contain between seven and eleven 1984. KIR genes. Immunogenetics 2002; 54: 221–229. 23 Fine JP, Gray RJ. A proportional hazards model for the 5 Velickovic M, Velickovic Z, Dunckley H. Diversity of killer subdistribution of competing risk. J Am Stat Assoc 1999; 94: cell immunoglobulin-like receptor genes in Pacific Islands 496–509. populations. Immunogenetics 2006; 58: 523–532. 24 Verheyden S, Schots R, Duquet W, Demanet C. A defined 6 Single RM, Martin MP, Gao X, Meyer D, Yeager M, Kidd JR donor activating natural killer cell receptor genotype et al. Global diversity and evidence for coevolution of KIR and protects against leukemic relapse after related HLA-identical HLA. Nat Genet 2007; 39: 1114–1119. hematopoietic stem cell transplantation. Leukemia 2005; 19: 7 Parham P. MHC class I molecules and KIRs in human history, 1446–1451. health and survival. Nat Rev Immunol 2005; 5: 201–214. 25 Sobecks RM, Ball EJ, Maciejewski JP, Rybicki LA, Brown S, 8 Karre K, Ljunggren HG, Piontek G, Kiessling R. Selective Kalaycio M et al. Survival of AML patients receiving HLA- rejection of H-2-deficient lymphoma variants suggests alter- matched sibling donor allogeneic bone marrow transplantation native immune defence strategy. Nature 1986; 319: 675–678. correlates withHLA-Cw ligand groups for killer immuno- 9 Ruggeri L, Capanni M, Urbani E, Perruccio K, Shlomchik globulin-like receptors. Bone Marrow Transplant 2007; 39: WD, Tosti A et al. Effectiveness of donor natural killer cell 417–424. alloreactivity in mismatched hematopoietic transplants. 26 Miller JS, Cooley S, Parham P, Farag SS, Verneris MR, Science 2002; 295: 2097–2100. McQueen KL et al. Missing KIR ligands are associated with 10 Hsu KC, Keever-Taylor CA, Wilton A, Pinto C, Heller G, less relapse and increased graft-versus-host disease (GVHD) Arkun K et al. Improved outcome in HLA-identical sibling following unrelated donor allogeneic HCT. Blood 2007; 109: hematopoietic stem-cell transplantation for acute myelogenous 5058–5061.

Bone Marrow Transplantation KIRs and aGVHD in HSCT K Ludajic et al 103 27 Chen C, Busson M, Rocha V, Appert ML, Lepage V, Dulphy the graft and correlates with clinical outcomes after unrelated N et al. Activating KIR genes are associated withCMV donor transplantation. Blood 2005; 106: 4370–4376. reactivation and survival after non-T-cell depleted 33 Sun JY, Dagis A, Gaidulis L, Miller MM, Rodriguez R, HLA-identical sibling bone marrow transplantation for Parker P et al. Detrimental effect of natural killer cell malignant disorders. Bone Marrow Transplant 2006; 38: alloreactivity in T-replete hematopoietic cell transplantation 437–444. (HCT) for leukemia patients. Biol Blood Marrow Transplant 28 Clausen J, Wolf D, Petzer AL, Gunsilius E, Schumacher P, 2007; 13: 197–205. Kircher B et al. Impact of natural killer cell dose and donor 34 Vivier E, Tomasello E, Baratin M, Walzer T, Ugolini S. killer-cell immunoglobulin-like receptor (KIR) genotype on Functions of natural killer cells. Nat Immunol 2008; 9: outcome following human leucocyte antigen-identical haema- 503–510. topoietic stem cell transplantation. Clin Exp Immunol 2007; 35 Guerra N, Michel F, Gati A, Gaudin C, Mishal Z, Escudier B 148: 520–528. et al. Engagement of the inhibitory receptor CD158a interrupts 29 Gagne K, Brizard G, Gueglio B, Milpied N, Herry P, TCR signaling, preventing dynamic membrane reorganization Bonneville F et al. Relevance of KIR gene polymorphisms in in CTL/tumor cell interaction. Blood 2002; 100: 2874–2881. bone marrow transplantation outcome. Hum Immunol 2002; 36 Ikeda H, Lethe B, Lehmann F, van Baren N, Baurain JF, de 63: 271–280. Smet C et al. Characterization of an antigen that is recognized 30 Martin MP, Qi Y, Gao X, Yamada E, Martin JN, Pereyra F on a melanoma showing partial HLA loss by CTL expressing et al. Innate partnership of HLA-B and KIR3DL1 subtypes an NK inhibitory receptor. Immunity 1997; 6: 199–208. against HIV-1. Nat Genet 2007; 39: 733–740. 37 Junevik K, Werlenius O, Hasselblom S, Jacobsson S, Nilsson- 31 Cooley S, Trachtenberg E, Bergemann TL, Saeteurn K, Klein Ehle H, Andersson PO. The expression of NK cell inhibitory J, Le CT et al. Donors withgroup B KIR haplotypesimprove receptors on cytotoxic T cells in B-cell chronic lymphocytic relapse-free survival after unrelated hematopoietic cell trans- leukaemia (B-CLL). Ann Hematol 2007; 86: 89–94. plantation for acute myelogenous leukemia. Blood 2009; e-pub 38 Poon K, Montamat-Sicotte D, CumberbatchN, McMichael ahead of print, 22 October 2008; doi:10.1182/blood-2008-07- AJ, Callan MF. Expression of leukocyte immunoglobulin-like 171926. receptors and natural killer receptors on virus-speciﬁc CD8+ 32 Cooley S, McCullar V, Wangen R, Bergemann TL, Spellman T cells during the evolution of Epstein-Barr virus-speciﬁc S, Weisdorf DJ et al. KIR reconstitution is altered by T cells in immune responses in vivo. Viral Immunol 2005; 18: 513–522.

Bone Marrow Transplantation