Candidate genetic associations with acute GVHD (aGVHD) were evaluated with the use of genotyped and imputed single-nucleotide polymorphism data from genome-wide scans of 1298 allogeneic hematopoietic cell transplantation (HCT) donors and recipients. Of 40 previously reported candidate SNPs, 6 were successfully genotyped, and 10 were imputed and passed criteria for analysis. Patient and donor genotypes were assessed for association with grades IIb-IV and III-IV aGVHD, stratified by donor type, in univariate and multivariate allelic, recessive and dominant models. Use of imputed genotypes to replicate previous IL10 associations was validated. Similar to previous publications, the IL6 donor genotype for rs1800795 was associated with a 20%-50% increased risk for grade IIb-IV aGVHD after unrelated HCT in the allelic (adjusted P = .011) and recessive (adjusted P = .0013) models. The donor genotype was associated with a 60% increase in risk for grade III-IV aGVHD after related HCT (adjusted P = .028). Other associations were found for IL2, CTLA4, HPSE, and MTHFR but were inconsistent with original publications. These results illustrate the advantages of using imputed single-nucleotide polymorphism data in genetic analyses and demonstrate the importance of validation in genetic association studies.

Introduction

During the past decade, many reports have identified genetic variants such as single-nucleotide polymorphisms (SNPs) and other polymorphisms that influence the risk of acute GVHD (aGVHD) after allogeneic hematopoietic cell transplantation (HCT). Many of these variants regulate the function of immune cells, their receptors, effector molecules, cytokines, and chemokines. Some results have led to the suggestion that pretransplantation assessment of these variants might help to assess the risk of adverse outcomes for each patient, guide the clinical management of patients who are at high risk, and ultimately serve as potential biologic targets for novel therapeutics.

Although these concepts have been greeted with some enthusiasm, the consistency of results leading to such suggestions has been questioned. Many of the reported results have been difficult to replicate in independent cohorts, and in no published studies have authors comprehensively evaluated these variants in the same cohort simultaneously. We recently had the opportunity to perform genome-wide scans of a cohort of patients and their donors with the use of an array that is informative for > 500 000 SNPs. In conjunction with the vast HapMap resource for SNP genotypes across the entire genome, these data provided a unique opportunity to use either the genotyped or imputed SNP data to determine how many of the previously published associations could be replicated.

Methods

Literature search

We performed a comprehensive PubMed search using the terms “acute GVHD” and “polymorphism” to identify all studies published by April 30, 2011, that reported an association between aGVHD and a genetic polymorphism at an α level < 0.05. Studies that did not meet this threshold were not included. Studies that reported associations with alternative genetic variants such as deletions, microsatellites, or VNTRs were excluded because our SNP genotyping array is not informative for such variants. Because our group previously published associations between SNPs in the IL10 and IL10RB genes and aGVHD using a cohort that overlaps with the current cohort,1,2  we used imputed genotypes for these SNPs as a quality control measure to evaluate whether previously identified associations could be replicated with imputed genotypes.

Study population

The source population included 1424 donor-recipient pairs randomly selected from among 3177 patients who received allogeneic HCT after myeloablative conditioning at the Fred Hutchinson Cancer Research Center and Seattle Cancer Care Alliance between 1992 and 2004. All recipient and donor samples were collected before HCT according to approved research protocols. Project-specific institutional research board approval from the Fred Hutchinson Cancer Research Center was obtained for the use of these samples. Recipient and donor demographic information collected at the time of pretransplantation evaluation, including sex, XY karyotype, ABO blood group, and race, was available through the FHCRC Clinical Research Division patient database (“Clinical Research Database”). All clinical data were prospectively collected and retrospectively reviewed.

Clinical characteristics of the patients and aGVHD prevalence data are summarized in Table 1. The cohort included recipients with either HLA-matched related donor (n = 612) or unrelated donors (n = 686) who underwent transplantation for a hematologic malignancy or myelodysplasia. Patients with related HLA-mismatched donors (n = 126) were excluded because of the limited number for meaningful stratified analysis. The pretransplantation disease risk category was determined on the basis of disease type and stage or remission/relapse status and classified as low, intermediate, or high. Patient-donor HLA matching was determined on the basis of genotyping for HLA-A, B, C, HLA-DRB1, and DQB1. All patients received T cell–replete BM or growth factor-mobilized blood cell grafts, and cyclosporine or tacrolimus plus methotrexate or mycophenylate for aGVHD prophylaxis. Conditioning regimens were categorized according to the use of total body irradiation. Peak severity of aGVHD after HCT, defined according to grade (0, I, IIa, IIb, III, IV), was recorded.

Table 1

Summary demographic and outcome data for the GWAS cohort

Variables Patient 
Number of patients 1298 
Age, y 38.2 ± 14 
Sex match (patient/donor), n (%)  
    Male/male 446 (34.3) 
    Male/female 309 (23.8) 
    Female/male 288 (22.2) 
    Female/female 255 (19.7) 
Race, n (%)  
    White 1116 (86.0) 
    Other 182 (14.0) 
Diagnosis, n (%)  
    Acute leukemia 465 (35.82) 
    CML 486 (37.44) 
    MDS 229 (17.64) 
    CLL, HD, and NHL 79 (6.1) 
    Multiple myeloma 39 (3.0) 
Disease risk, n (%)  
    Low 361 (27.8) 
    Intermediate 594 (45.8) 
    High 343 (26.4) 
Donor type, n (%)  
    Unrelated donor 686 (52.8) 
    Matched related donor 612 (47.2) 
Stem cell source  
    BM, n (%) 921 (71.0) 
    Peripheral blood 377 (29.0) 
TBI dose  
    None, n (%) 531 (40.9) 
    ≤ 1200 Gy 366 (28.2) 
    > 1200 Gy 401 (30.9) 
aGVHD IIb-IV, n (%) 829 (58.5) 
aGVHD III-IV, n (%) 323 (25.0) 
Variables Patient 
Number of patients 1298 
Age, y 38.2 ± 14 
Sex match (patient/donor), n (%)  
    Male/male 446 (34.3) 
    Male/female 309 (23.8) 
    Female/male 288 (22.2) 
    Female/female 255 (19.7) 
Race, n (%)  
    White 1116 (86.0) 
    Other 182 (14.0) 
Diagnosis, n (%)  
    Acute leukemia 465 (35.82) 
    CML 486 (37.44) 
    MDS 229 (17.64) 
    CLL, HD, and NHL 79 (6.1) 
    Multiple myeloma 39 (3.0) 
Disease risk, n (%)  
    Low 361 (27.8) 
    Intermediate 594 (45.8) 
    High 343 (26.4) 
Donor type, n (%)  
    Unrelated donor 686 (52.8) 
    Matched related donor 612 (47.2) 
Stem cell source  
    BM, n (%) 921 (71.0) 
    Peripheral blood 377 (29.0) 
TBI dose  
    None, n (%) 531 (40.9) 
    ≤ 1200 Gy 366 (28.2) 
    > 1200 Gy 401 (30.9) 
aGVHD IIb-IV, n (%) 829 (58.5) 
aGVHD III-IV, n (%) 323 (25.0) 

aGVHD indicates acute graft versus host disease; CML, chronic myelogenous leukemia; CLL, chronic lymphocytic leukemia; GWAS, genome-wide association study; HD, Hodgkin's disease; HLA, human leukocyte antigen; MDS, myelodysplastic syndrome; NHL, non-Hodgkin's lymphoma; and TBI, total body irradiation.

aGVHD was considered as 2 different phenotypes: grades 0-IIa versus grades IIb-IV, and grades 0-II versus grades III-IV. Grade IIa GVHD is a subset of grade II aGVHD that includes cases with upper gastrointestinal symptoms generally documented as aGVHD by biopsy, without diarrhea or with stool volume < 1000 mL/d, without rash or with rash involving < 50% of the body surface, and without liver involvement.3  Grade IIb aGVHD is a subset of grade II GVHD that includes cases with rash involving > 50% of the body surface (stage III skin disease) or with total serum bilirubin concentration between 2.0 and 2.9 mg/dL (stage I liver disease), with or without stage I gastrointestinal involvement. As previously reported, the overall incidence of grades II-IV GVHD is greater at our center than reported at other centers, primarily because of high sensitivity for the diagnosis of upper gastrointestinal GVHD.3  The incidence of grades IIb-IV GVHD at our center is similar to the overall incidence of grades II-IV GVHD at other centers.

Sample preparation, genotyping, and imputations

DNA specimens originated from blood mononuclear cells or EBV-transformed B-lymphocyte cell lines. Genomic DNA was extracted by a standard salting-out method with the use of the Puregene kit (QIAGEN). The quantity and purity of DNA in each sample was determined by UV absorption in a spectrophotometer (OD260) and by OD260/280 ratio, respectively. After DNA quantitation and initial screening for verification of sample identity, DNA samples were distributed into 96-well plates (92 samples per plate) at a targeted concentration of 50-100 ng/μL. Each plate of 92 samples contained 4 duplicate samples carried forward from a previously prepared plate as a quality control indicator. The plates were then sealed, frozen at −20°C, and shipped on dry ice to the Affymetrix Service Laboratory (ASL) in Santa Clara, CA, for amplification and hybridization.

All assays used the Affymetrix GeneChip Genome-Wide Human SNP Array 5.0, with hybridization and scanning at the ASL. After scanning of hybridized GeneChip arrays, the resulting raw probe intensities were evaluated with the Bayesian Robust Linear Model with Mahalanobis distance classifier (BRLMM) analysis tool adapted for the 5.0 Array (BRLMM-P). Data quality was assessed via 3 different methods: the Affymetrix “QC call rate,” the clustering call rate, and the sex call rate. Once the array data were cleared by internal ASL quality assessment, the genome-wide SNP data were transmitted to FHCRC, where they were subjected to further QC analysis. This analysis included assessment of overall sample quality using the DM and BRLMM-P call rate statistics generated by Affymetrix, adjustment of threshold standards to optimize genotyping accuracy, if necessary, and assessment and resolution of sample identity questions with the use of calculations for estimating genomic similarity between samples. Because sex and ABO type was known for all samples, PCR-based ABO and XY genotyping assays were implemented to determine sample quality and verify sample identity before submitting the DNA samples to the ASL for hybridization and scanning. A further evaluation of sample identity was performed via use of the Affymetrix-supplied genotypes obtained by clustering with the BRLMM-P algorithm for those samples passing the QC call rate threshold. The final SNP-based genotype data generated is referred to as the “GWAS-HCT” dataset.

The candidate SNP genotype determination algorithm is summarized in Figure 1. If the original candidate SNP was not genotyped on our array, we evaluated the imputed genotype, which was obtained by imputing all the HapMap release 22 SNPs (combined panel of ethnic groups: CEU [Utah residents with Northern and Western European ancestry from the CEPH collection], YRI [Yoruba in Ibadan], CHB [Han Chinese in Beijing], and JPT [Japanese in Tokyo]) that were not genotyped on the Affymetrix 5.0 array using IMPUTE software (https://mathgen.stats.ox.ac.uk/impute/impute.html). At each unobserved HapMap SNP locus, 1 or more genotypes may be predicted with a positive posterior probability, which measures the probability of observing that genotype at the imputed locus. We assign the genotype with the maximum posterior probability as the imputed SNP genotype only if the maximum posterior probability exceeded 0.8. All genotyped and imputed SNPs that violated Hardy-Weinberg equilibrium with P < .001, had a minor allele frequency < 0.05, or had a < 90% call rate were excluded from the analyses.

Figure 1

Candidate SNP genotype determination algorithm. HWE indicates Hardy-Weinberg equilibrium; MAF, minor allele frequency; and SNP, single-nucleotide polymorphism.

Figure 1

Candidate SNP genotype determination algorithm. HWE indicates Hardy-Weinberg equilibrium; MAF, minor allele frequency; and SNP, single-nucleotide polymorphism.

Statistical analysis

All analyses were performed with Matlab v.2010b (MathWorks). All analyses were stratified by donor type (matched related donor [MRD] vs unrelated donor [URD]) and were examined for recipients and donors in separate univariate and multivariate Cox regression models, censored at the time of death or onset of recurrent malignancy. Multivariate Cox models adjusted for clinical factors recently found to be associated with the risk of aGVHD female donor and male recipient status and total body irradiation.4  For URDs, an additional in-model adjustment was made for HLA match. We also incorporated as covariates principal components derived from a principal component analysis of population stratification,5  where the first 4 components were included as covariates in the final adjusted analysis.

Each SNP was assessed for allelic and genotypic (recessive and dominant models) modes of transmission, stratified according to donor type (HLA-matched related versus unrelated). For a SNP with a major allele “a” and a minor allele “b,” the recessive model tests the hypothesis that the genotype “bb” is associated with an increased or decreased risk compared with the collective genotypes “ab and “aa.” The dominant model tests the hypothesis that the collective genotypes “bb” and “ab” are associated with an increased or decreased risk compared with the genotype “aa.” The allelic model tests the hypothesis that the minor allele b is associated with an increased or decreased risk compared with the major allele a, and the number of copies of the minor allele is modeled as an additive effect. Because the main goal of this analysis was to replicate published associations with candidate SNPs, a 2-sided P ≤ .05 was selected as the threshold of significance, despite the multiple comparisons made in this analysis.

Results

Candidate gene and SNP selection

A search of the PubMed database identified 41 publications that reported associations of donor or recipient genotypes for 40 candidate SNPs in 22 genes with the risk of aGVHD: CD31,6,7 CTLA4,8 ESR1,9 FAS,10 HMGB1,11 HPSE,12 HSPA1L,13,14 IL1α,9,15 IL1β,16 IL2,17 IL6,10,18,19 IL10,1,16,18,20,21,22,23,2425 IL10RB,2,24 IL23R,26,2728 MADCAM1,29 MTHFR,30,31,3233 NOD2,34,35,3637 RFC1,38 TGFβ1,39,40 TNF,10,16,20,21,23,41,4243 TNFRII,23,42  and VEGFα.44  The median cohort size in these studies was 89 (range, 24-536). The median number of genetic variations examined in each study was 3.5 (range, 1-19); in none of the studies did authors adjust for multiple comparisons in the analyses.

Of the 40 published candidate SNP associations, usable genotype or imputed data were available for 16 (40%; Table 2). Six candidate SNPs were genotyped: CD31 (rs1131012), CTLA4 (rs3087243), HPSE (rs4364254), IL23R (rs11209026), IL6 (rs1800795), and TNF (rs1799964). Ten candidate SNPs were imputed and passed the criteria for analysis: CD31 (rs668), IL1β (rs16944), IL2 (rs2069762), IL10 (rs1800871, rs1800872), MTHFR (rs1801131), RFC1 (rs1057807), TNF (rs1800629, rs1800630), and TNFRII (rs1061622). Of the remaining 24 candidate SNPs, 10 candidate SNPs were imputed but did not pass the criteria for analysis: ESR1 (rs2234693), FAS (rs1800682), HPSE (rs4693608), IL10 (rs1800896), IL10RB (rs2834167), MTHFR (rs1801133), RFC1 (rs4975003, rs6833176), and VEGFα (rs699947, rs833061). Fourteen SNPs were not genotyped and could not be imputed: CD31 (rs12953), FAS (rs2234767), HMGB1 (rs41376448), HSPA1L (rs2075800), MADCAM1 (rs2302217), NOD2 (rs2066844, rs2066845, rs2066847), TGFβ1 (rs1800470), TNF (−488, rs1799724, rs361525), VEGFα (rs2010963, rs3025039).

Table 2

Summary of genotyped and imputed SNP data generated from the GWAS-HCT dataset using the Affymetrix 5.0 Array and the HapMap release 22 SNPs

Gene Type rs number Chr Alleles* MAF Call rate p_HWE Average posterior probability Status 
CD31 Genotyped rs1131012 17 C/T 0.472 0.98 0.08 0.98 Pass 
CD31 Imputed rs668 17 G/C 0.494 0.966 0.7 0.978 Pass 
CTLA4 Genotyped rs3087243 A/G 0.446 0.814 Pass 
ESR1 Imputed rs2234693 C/T 0.35 0.642 0.121 0.831 Fail 
FAS Imputed rs1800682 10 G/A 0.464 0.899 0.218 0.948 Fail 
HPSE Genotyped rs4364254 C/T 0.296 0.999 0.349 0.999 Pass 
HPSE Imputed rs4693608 G/A 0.328 0.62 0.668 0.817 Fail 
IL10 Imputed rs1800871 A/G 0.238 0.951 0.352 0.972 Pass 
IL10 Imputed rs1800872 T/G 0.238 0.951 0.359 0.972 Pass 
IL10 Imputed rs1800896 C/T 0.472 0.799 0.245 0.898 Fail 
IL10RB Imputed rs2834167 21 G/A 0.162 0.353 0.027 0.722 Fail 
IL1B Imputed rs16944 A/G 0.359 0.944 0.44 0.962 Pass 
IL2 Imputed rs2069762 C/A 0.317 0.998 0.115 0.998 Pass 
IL23R Genotyped rs11209026 A/G 0.064 0.998 0.598 0.998 Pass 
IL6 Genotyped rs1800795 C/G 0.375 0.998 0.006 0.998 Pass 
MTHFR Imputed rs1801131 G/T 0.304 0.946 0.238 0.967 Pass 
MTHFR Imputed rs1801133 A/G 0.182 0.449 0.221 0.77 Fail 
RFC1 Imputed rs1057807 G/A 0.425 0.999 0.166 0.999 Pass 
RFC1 Imputed rs4975003 C/G 0.441 0.893 0.274 0.932 Fail 
RFC1 Imputed rs6844176 C/T 0.494 0.748 0.047 0.878 Fail 
TNF Genotyped rs1799964 C/T 0.205 0.063 Pass 
TNF Imputed rs1800629 A/G 0.135 0.926 0.726 0.965 Pass 
TNF Imputed rs1800630 A/C 0.152 0.996 0.153 0.997 Pass 
TNFRII Imputed rs1061622 G/T 0.227 0.964 0.12 0.971 Pass 
VEGFA Imputed rs699947 A/C 0.114 0.141 0.244 0.633 Fail 
VEGFA Imputed rs833061 C/T 0.183 0.094 0.283 0.607 Fail 
Gene Type rs number Chr Alleles* MAF Call rate p_HWE Average posterior probability Status 
CD31 Genotyped rs1131012 17 C/T 0.472 0.98 0.08 0.98 Pass 
CD31 Imputed rs668 17 G/C 0.494 0.966 0.7 0.978 Pass 
CTLA4 Genotyped rs3087243 A/G 0.446 0.814 Pass 
ESR1 Imputed rs2234693 C/T 0.35 0.642 0.121 0.831 Fail 
FAS Imputed rs1800682 10 G/A 0.464 0.899 0.218 0.948 Fail 
HPSE Genotyped rs4364254 C/T 0.296 0.999 0.349 0.999 Pass 
HPSE Imputed rs4693608 G/A 0.328 0.62 0.668 0.817 Fail 
IL10 Imputed rs1800871 A/G 0.238 0.951 0.352 0.972 Pass 
IL10 Imputed rs1800872 T/G 0.238 0.951 0.359 0.972 Pass 
IL10 Imputed rs1800896 C/T 0.472 0.799 0.245 0.898 Fail 
IL10RB Imputed rs2834167 21 G/A 0.162 0.353 0.027 0.722 Fail 
IL1B Imputed rs16944 A/G 0.359 0.944 0.44 0.962 Pass 
IL2 Imputed rs2069762 C/A 0.317 0.998 0.115 0.998 Pass 
IL23R Genotyped rs11209026 A/G 0.064 0.998 0.598 0.998 Pass 
IL6 Genotyped rs1800795 C/G 0.375 0.998 0.006 0.998 Pass 
MTHFR Imputed rs1801131 G/T 0.304 0.946 0.238 0.967 Pass 
MTHFR Imputed rs1801133 A/G 0.182 0.449 0.221 0.77 Fail 
RFC1 Imputed rs1057807 G/A 0.425 0.999 0.166 0.999 Pass 
RFC1 Imputed rs4975003 C/G 0.441 0.893 0.274 0.932 Fail 
RFC1 Imputed rs6844176 C/T 0.494 0.748 0.047 0.878 Fail 
TNF Genotyped rs1799964 C/T 0.205 0.063 Pass 
TNF Imputed rs1800629 A/G 0.135 0.926 0.726 0.965 Pass 
TNF Imputed rs1800630 A/C 0.152 0.996 0.153 0.997 Pass 
TNFRII Imputed rs1061622 G/T 0.227 0.964 0.12 0.971 Pass 
VEGFA Imputed rs699947 A/C 0.114 0.141 0.244 0.633 Fail 
VEGFA Imputed rs833061 C/T 0.183 0.094 0.283 0.607 Fail 

Chr indicates chromosome; HWE, Hardy-Weinberg equilibrium MAF, minor allele frequency; and SNP, single-nucleotide polymorphism.

*

The first allele is designated as the minor allele, and the second allele is designated as the major allele.

Validation of previous published IL10 SNP associations with aGVHD

We previously published associations between aGVHD risk and SNPs on IL10 (rs1800896, rs1800871, and rs1800872)1  and IL10RB (rs2834167)2  in a cohort that overlapped significantly with the current cohort. Because these SNPs were not directly genotyped on the array, we evaluated whether the original association could be replicated with the use of imputed genotypes. Although the genotypes of all 4 SNPs were imputable, only the IL10 SNPs rs1800871 and rs1800872 passed the selection criteria for analysis (Table 2). These 2 successfully imputed IL10 SNPs are in significant linkage disequilibrium (D′ > 0.9). Among MRD transplantations, the patient's genotypes for both SNPs were significantly associated with a 30% decrease of the risk for grade III-IV aGVHD in the allelic model (Table 3; hazard ratio [HR] 0.72, P = .048). These findings are consistent with the findings of our original publication (Table 4).1 

Table 3

Summary of significant associations (P < .05) between aGVHD and candidate SNPs

Gene Phenotype Genome Donor type SNP Alleles MAF* Model HR Unadjusted CI P HR Adjusted CI† P 
IL2 III-IV DNR URD rs2069762 C/A 0.34 Allelic 1.16 (0.96-1.40) .1293 1.15 (0.94-1.40) .1811 
       Recessive 1.04 (0.70-1.55) .8466 0.98 (0.64-1.50) .9310 
       Dominant 1.32 (1.00-1.74) .0487 1.33 (0.99-1.78) .0557 
IL6 IIb-IV DNR URD rs1800795 C/G 0.39 Allelic 1.21 (1.07-1.36) .0028 1.19 (1.04-1.36) .0106 
       Recessive 1.48 (1.19-1.84) .0004 1.44 (1.15-1.81) .0013 
       Dominant 1.17 (0.97-1.40) .0966 1.13 (0.93-1.38) .2231 
IL6 IIb-IV PT MRD rs1800795 C/G 0.34 Allelic 0.87 (0.73-1.04) .1362 0.8 (0.66-0.97) .0240 
       Recessive 0.84 (0.57-1.25) .3983 0.81 (0.54-1.20) .2923 
       Dominant 0.84 (0.66-1.06) .1444 0.74 (0.57-0.95) .0195 
IL6 III-IV DNR MRD rs1800795 C/G 0.36 Allelic 1.26 (0.98-1.63) .0719 1.27 (0.97-1.67) .0864 
       Recessive 1.09 (0.65-1.82) .7360 1.06 (0.63-1.79) .8177 
       Dominant 1.54 (1.05-2.26) .0257 1.59 (1.05-2.41) .0279 
IL10 III-IV PT MRD rs1800872 T/G 0.24 Allelic 0.72 (0.52-1.00) .0481 0.73 (0.52-1.04) .0796 
       Recessive (0.00,Inf) .9817 (0.00,Inf) .9815 
       Dominant 0.82 (0.56-1.19) .2933 0.85 (0.58-1.25) .4035 
IL10 III-IV PT MRD rs1800871 A/G 0.24 Allelic 0.72 (0.52-1.00) .0481 0.73 (0.52-1.04) .0796 
       Recessive (0.00,Inf) .9817 (0.00,Inf) .9815 
       Dominant 0.82 (0.56-1.19) .2933 0.85 (0.58-1.25) .4035 
CTLA4 III-IV PT URD rs3087243 A/G 0.45 Allelic 1.17 (0.97-1.42) .1042 1.23 (1.01-1.50) .0374 
       Recessive (0.72-1.41) .9782 1.05 (0.74-1.49) .7870 
       Dominant 1.5 (1.09-2.07) .0138 1.65 (1.18-2.30) .0035 
HPSE IIb-IV DNR URD rs4364254 C/T 0.29 Allelic 1.07 (0.93-1.22) .3377 1.11 (0.97-1.27) .1453 
       Recessive 0.91 (0.66-1.26) .5736 0.95 (0.68-1.33) .7767 
       Dominant 1.15 (0.97-1.37) .1158 1.21 (1.01-1.45) .0415 
MTHFR III-IV PT MRD rs1801131 G/T 0.29 Allelic 0.88 (0.67-1.17) .3870 0.84 (0.63-1.11) .2157 
       Recessive 1.29 (0.76-2.19) .3408 1.22 (0.71-2.08) .4679 
       Dominant 0.71 (0.49-1.04) .0764 0.66 (0.45-0.96) .0309 
Gene Phenotype Genome Donor type SNP Alleles MAF* Model HR Unadjusted CI P HR Adjusted CI† P 
IL2 III-IV DNR URD rs2069762 C/A 0.34 Allelic 1.16 (0.96-1.40) .1293 1.15 (0.94-1.40) .1811 
       Recessive 1.04 (0.70-1.55) .8466 0.98 (0.64-1.50) .9310 
       Dominant 1.32 (1.00-1.74) .0487 1.33 (0.99-1.78) .0557 
IL6 IIb-IV DNR URD rs1800795 C/G 0.39 Allelic 1.21 (1.07-1.36) .0028 1.19 (1.04-1.36) .0106 
       Recessive 1.48 (1.19-1.84) .0004 1.44 (1.15-1.81) .0013 
       Dominant 1.17 (0.97-1.40) .0966 1.13 (0.93-1.38) .2231 
IL6 IIb-IV PT MRD rs1800795 C/G 0.34 Allelic 0.87 (0.73-1.04) .1362 0.8 (0.66-0.97) .0240 
       Recessive 0.84 (0.57-1.25) .3983 0.81 (0.54-1.20) .2923 
       Dominant 0.84 (0.66-1.06) .1444 0.74 (0.57-0.95) .0195 
IL6 III-IV DNR MRD rs1800795 C/G 0.36 Allelic 1.26 (0.98-1.63) .0719 1.27 (0.97-1.67) .0864 
       Recessive 1.09 (0.65-1.82) .7360 1.06 (0.63-1.79) .8177 
       Dominant 1.54 (1.05-2.26) .0257 1.59 (1.05-2.41) .0279 
IL10 III-IV PT MRD rs1800872 T/G 0.24 Allelic 0.72 (0.52-1.00) .0481 0.73 (0.52-1.04) .0796 
       Recessive (0.00,Inf) .9817 (0.00,Inf) .9815 
       Dominant 0.82 (0.56-1.19) .2933 0.85 (0.58-1.25) .4035 
IL10 III-IV PT MRD rs1800871 A/G 0.24 Allelic 0.72 (0.52-1.00) .0481 0.73 (0.52-1.04) .0796 
       Recessive (0.00,Inf) .9817 (0.00,Inf) .9815 
       Dominant 0.82 (0.56-1.19) .2933 0.85 (0.58-1.25) .4035 
CTLA4 III-IV PT URD rs3087243 A/G 0.45 Allelic 1.17 (0.97-1.42) .1042 1.23 (1.01-1.50) .0374 
       Recessive (0.72-1.41) .9782 1.05 (0.74-1.49) .7870 
       Dominant 1.5 (1.09-2.07) .0138 1.65 (1.18-2.30) .0035 
HPSE IIb-IV DNR URD rs4364254 C/T 0.29 Allelic 1.07 (0.93-1.22) .3377 1.11 (0.97-1.27) .1453 
       Recessive 0.91 (0.66-1.26) .5736 0.95 (0.68-1.33) .7767 
       Dominant 1.15 (0.97-1.37) .1158 1.21 (1.01-1.45) .0415 
MTHFR III-IV PT MRD rs1801131 G/T 0.29 Allelic 0.88 (0.67-1.17) .3870 0.84 (0.63-1.11) .2157 
       Recessive 1.29 (0.76-2.19) .3408 1.22 (0.71-2.08) .4679 
       Dominant 0.71 (0.49-1.04) .0764 0.66 (0.45-0.96) .0309 

Allo indicates allogeneic; CI, confidence interval; DNR, donor, HR, hazard ratio MAF, minor allele frequency; MRD, matched related donor; NR, not reported; PT, patient; SNP, single-nucleotide polymorphism; and URD, unrelated donor.

*

The first allele is designated as the minor allele, and the second allele is designated as the major allele.

Models adjusted for total body irradiation (yes vs no), sex mismatch, and 4 principal components for population stratification. Matched related donor analyses are also adjusted for HLA match.

Table 4

Summary of previous publications reporting associations with candidate genes that meet the P < .05 threshold in the current validation study

Gene Reference Phenotype Genome Donor type SNP Alleles* MAF HR Unadjusted CI P 
IL2 MacMillan et al17  II-IV PT URD rs2069762 G/T 0.3 2.1 (1.0-4.5) .05 
IL6 Mullighan et al10  I-IV DNR MRD rs1800795 G/C 0.4 4.4 (1.6-11.7) .001 
IL6 Karabon e tal18  II-IV DNR NR rs1800795 G/C 0.5 3.4 (1.1-13.0) .03 
IL6 Ambruzova et al19  II-IV PT Allo rs1800795 G/C 0.5 2.2 (1.1-4.4) .03 
IL10 Lin et al1  III-IV PT MRD rs1800871† A/C 0.3    
IL10 Lin et al1  III-IV PT MRD rs1800872 A/C 0.3 0.4 (0.2-0.9) .02 
CTLA4 Perez-Garcia et al8  II-IV DNR MRD rs3087243 A/G 0.5 1.5 (1.0-2.3) .03 
HPSE Ostrovsky et al12  II-IV and III-IV PT Allo rs4364254 T/C 0.4 NR NR < .01 
MTHFR Robien et al30  I-IV PT Allo rs1801131 A/C 0.4 3.6 (1.0-12.7) < .01 
Gene Reference Phenotype Genome Donor type SNP Alleles* MAF HR Unadjusted CI P 
IL2 MacMillan et al17  II-IV PT URD rs2069762 G/T 0.3 2.1 (1.0-4.5) .05 
IL6 Mullighan et al10  I-IV DNR MRD rs1800795 G/C 0.4 4.4 (1.6-11.7) .001 
IL6 Karabon e tal18  II-IV DNR NR rs1800795 G/C 0.5 3.4 (1.1-13.0) .03 
IL6 Ambruzova et al19  II-IV PT Allo rs1800795 G/C 0.5 2.2 (1.1-4.4) .03 
IL10 Lin et al1  III-IV PT MRD rs1800871† A/C 0.3    
IL10 Lin et al1  III-IV PT MRD rs1800872 A/C 0.3 0.4 (0.2-0.9) .02 
CTLA4 Perez-Garcia et al8  II-IV DNR MRD rs3087243 A/G 0.5 1.5 (1.0-2.3) .03 
HPSE Ostrovsky et al12  II-IV and III-IV PT Allo rs4364254 T/C 0.4 NR NR < .01 
MTHFR Robien et al30  I-IV PT Allo rs1801131 A/C 0.4 3.6 (1.0-12.7) < .01 

Allo indicates allogeneic; CI, confidence interval; DNR, donor; HR, hazard ratio; MAF, minor allele frequency; MRD, matched related donor; NR, not reported; PT, patient; SNP, single-nucleotide polymorphism; and URD, unrelated donor.

*

The first allele is designated as the minor allele, and the second allele is designated as the major allele.

Assessed as component of a haplotype.

Associations between IL6, IL2, CTLA4, HPSE, and MTHFR SNPs and aGVHD

Results for the remaining 14 evaluable candidate SNPs are summarized in Figure 2A-D. These analyses revealed that rs1800795 in IL6, rs2069762 in IL2, rs3087243 in CTLA4, rs4364254 in HPSE, and rs1801131 in MTHFR had unadjusted or adjusted models that met the P < .05 threshold (Table 3). The previously published associations for these SNPs are summarized in Table 4 for comparison. We report in the next 3 paragraphs the details of these associations in our cohort.

Figure 2

Forest plots for unadjusted and adjusted analyses of aGVHD and candidate SNPs. Forest plots for unadjusted (squares) and adjusted (circles) analyses of aGVHD and candidate SNPs among MRD (A) and URD (B) HCT. Hazard ratios are represented on the x-axis in logarithmic scale. Positive values indicate an increased risk for aGVHD, and negative values indicate a decreased risk for aGVHD. Gene/rs numbers in bold and closed squares and circles indicate the association exceeded the P = .05 threshold.

Figure 2

Forest plots for unadjusted and adjusted analyses of aGVHD and candidate SNPs. Forest plots for unadjusted (squares) and adjusted (circles) analyses of aGVHD and candidate SNPs among MRD (A) and URD (B) HCT. Hazard ratios are represented on the x-axis in logarithmic scale. Positive values indicate an increased risk for aGVHD, and negative values indicate a decreased risk for aGVHD. Gene/rs numbers in bold and closed squares and circles indicate the association exceeded the P = .05 threshold.

Of all the SNPs that met the P < .05 threshold, the associations with rs1800795 in IL6 were most consistently observed throughout the genetic models. In the unadjusted and adjusted URD analyses, the donor genotype for rs1800795 was associated with a 20% to 50% increase in the risk for grade IIb-IV aGVHD in the allelic (unadjusted P = .003; adjusted P = .011) and recessive (unadjusted P < .001; adjusted P = .001) models (Table 3). For MRD HCT, the donor genotype for rs1800795 was associated with a 60% increase in risk for grade III-IV aGVHD in both dominant unadjusted (P = .026) and adjusted (P = .028) models. We also found the patient C allele for rs1800795 in IL6 was associated with a 20% to 26% decrease in risk of grade IIb-IV aGVHD but only in the adjusted MRD allelic (P = .024) and dominant (P = .020) models (Figure 2A).

We found that the IL2 polymorphism rs2069762 in the donor genotype was associated in the dominant genetic mode with a 1.3-fold increase in risk of grade III-IV aGVHD after URD HCT (P = .049; Table 3, Figure 2B). The P value shifted slightly to .056 in multivariate analysis, but the point estimate remained unchanged at 1.3. In both unadjusted and adjusted analyses, the patient genotypes in our cohort indicated a consistent shift toward slightly greater risks of grade IIb-IV and III-IV aGVHD after URD HCT in the allelic and recessive models, but the P values did not exceed the .05 threshold. For CTLA4, the donor rs3087243 genotype was associated with an increased risk for grade III-IV aGVHD after URD HCT in both the unadjusted (HR 1.5, P = .014) and adjusted (HR 1.65, P = .004) analyses (Table 3, Figure 2B). However, unlike the published association, the point estimates in our analysis of the donor rs3087243 genotype in MRD HCT suggested a reduced risk for grade IIb-IV aGVHD in all models, with no appreciable association with risk for grade III-IV aGVHD (Figure 2A).

In the recently published HSPE analysis, the patient-donor disparity for SNP rs4693608 and rs4364254 genotypes was examined.12  Because the imputed rs4693608 genotype did not meet our quality control threshold, we were not able to perform a disparity analysis, and therefore we restricted our analysis to rs4364254, which was directly genotyped. The donor genotype for rs4364254 was significantly associated with a 1.2-fold increase (P = .042) in risk for grade IIb-IV aGVHD after URD HCT in the dominant multivariate model (Table 3, Figure 2B). For the MTHFR gene, we identified an association between rs1801131 and grade III-IV aGVHD. But unlike the original publication, the patient genotype was associated with a 35% decrease in risk (P = .031) in the dominant model after MRD HCT (Table 3, Figure 2A).

Discussion

We conducted an independent replication analysis of 14 candidate SNP genotypes previously documented in the published literature to be associated with the risk of aGVHD. To our knowledge, this is the first attempt to replicate such a large number of published genetic associations with the use of genotyped and imputed genome wide array data generated from a relatively large cohort of HCT patients and donors. Our analytical approach and the results of these analyses raise several key questions regarding candidate gene genetic association studies, some of which are of particular importance to the field of HCT.

The idea of an agnostic genome-wide approach toward discovery of risk variants is particularly challenging for HCT because of the cost of comprehensively screening both the patient and donor genomes and the need to accumulate a sample size of at least 5000 transplantations (10 000 total samples from patients and donors) to attain > 80% power for detecting significant associations with phenotype of > 30% frequency and odds ratios ranging from 1.5 to 1.8 for SNPs with minor allele frequencies > 20%.45  As we await the assembly of such a cohort, a bridging approach might be to use genome wide data from an intermediate sized cohort to conduct candidate gene studies, using both genotyped and imputed data.

The imputation approach, which has been widely adopted in genetic association studies,46  facilitates the use of actual genotypes for a limited set of SNPs to access information for a much larger number of SNPs across the entire genome. Our validation analysis of the imputed genotypes for rs1800871 and rs1800872 in IL10 demonstrate that imputed genotype data can be robust and correlates well with actual genotype data. This approach opens the possibility of evaluating candidate genes in pathways of high biologic relevance. Although this approach is limited in scope, it is financially and statistically more realistic for the limited sample size of most HCT cohorts. However, this approach is restricted by the quantity of usable imputed data, which depends on the extent and quality of the genome-wide SNP data generated by the commercial array. In our study, the SNP data generated by the Affymetrix 5.0 array only had a success rate of 40% for useable high-quality actual and imputed genotypes for the specified candidate SNPs, thus highlighting the limitation of this approach, at least with this array.

Our IL6 results suggest that the association with the IL6 SNP is likely worthy of more in-depth analysis. We found significant associations with the IL6 SNP genotype in both the patient and donor genomes, as well as with both aGVHD phenotypes. Associations with the donor and patient IL6 genotypes have been previously reported, all with risks that are similar in magnitude in comparison with our findings.10,18,19  In particular, the association of patient genotype has been validated under our multivariate analysis, in terms of all factors such as genome, donor type, and association model. However, the association of donor genotype is only arguably validated because the original findings were discrepant with regards to the assignment of the minor allele. Although both studies identified the association to be with the donor's G allele, Mullighan et al10  assigned the G allele as the major allele and Karabon et al18  assigned the G allele as the minor allele. The complementary nature of the C and G alleles, in conjunction with their similar allele frequencies, may have resulted in these differences in minor allele assignment in the genotyping process. However, because the minor allele's frequency is close to 50% in both study populations and this appears to vary among different ethnic groups in the HapMap database, it is possible that because of regional ethnic/racial differences in previous and current study populations, the minor allele frequencies may vary to the point where the minor allele assignment can differ between studies.

To judge whether these IL6 findings are significant, the totality of the IL6 association data should be considered. An association with polymorphisms in the promoter region of IL6 is biologically plausible. SNP rs1800795 is located at position −174 in the 5′ flanking region of the IL6 gene.47  This SNP has been documented to result in up-regulation of IL6 gene transcription and higher circulating IL6 levels.47  This SNP has also been reported to influence the risk for many other phenotypes, including juvenile arthritis,47  type 2 diabetes,48  bronchiolitis obliterans syndrome after lung transplantation,49  and kidney allograft survival.50  Multiple studies have documented increased plasma or serum concentrations of IL6 in patients with aGVHD.18  IL6 is a pleiotropic interleukin that in the context of an allogeneic HCT functions as both an acute phase mediator of inflammation and also as a critical factor in the differentiation of regulatory and Th17 effector cells. These effects may be mediated through variable IL6 expression encoded by both the host and donor genomes. It is unclear why donor cells genetically predisposed to lower circulating IL6 levels will increase the risk for aGVHD. It is feasible that a patient predisposed to chronically elevated IL6 levels (the phenotype associated with the major allele) may experience down regulation of the IL6 receptor and thereby reduce their responsiveness to IL6 and risk for aGCHD. This hypothesis requires further investigation.

The results for IL2, CTLA4, HPSE, and MTHFR indicate that the definition of “replication” is worthy of additional discussion in the context of genetic association studies in HCT. For each of these genes, we were able to identify an association with the candidate SNP that met the P < .05 threshold. However, certain aspects for each association were inconsistent with the original published findings. In the strictest sense, these associations did not represent replications of the original results. For IL2, our association was with the donor genome, which was inconsistent with the patient genome association in the original publication. For CTLA4, our association was observed with URD HCT and not with MRD HCT, which was the transplantation type for which the original association was documented.

For HSPE, we identified a SNP association with aGVHD, but this association was not reported as statistically significant in the original study. In fact, Ostrovsky et al reported that rs4364254 trended toward a lower cumulative risk for grades II-IV and III-IV aGVHD in the recessive model.12  Unfortunately, we were not able to impute the genotypes for rs4693608, the key HSPE SNP that was associated with aGVHD risk. For MTHFR, we observed an association with the same genome and donor type, but the observed effect was opposite to that of the original publication. At best, the associations we observed can be considered as new discoveries; at worst, they represent false-positive associations. Thus, our results cannot be used alone to judge whether a previously published association is valid. Instead, our results should be combined with previously published association studies to judge the likelihood that an association remains worthy of additional investigation.

If our IL6 findings are regarded as the only true replication of previously reported findings, then the replication rate in our analysis is only 7%. This low rate of replication is consistent with the observed replication rate of 3.6% in a comprehensive review of 166 candidate SNPs with putative associations that had been evaluated 3 or more times in the published literature.51  Guidelines for replicating genotype-phenotype associations are now well defined. These guidelines indicate that the likelihood of replicating a genetic association depends on many variables that can be related to the study population, genetic approach, phenotype definition and statistical power. We believe our results are valid because we took every measure to remain faithful to these guidelines.

Although we are unable to ensure that the ethnic background of our cohort is comparable with those of the published studies, we incorporated the principal component approach to account for the potential confounding effects of population stratification. Because there is little that we can do to control for center-to-center differences in clinical practice, we attempted to minimize imbalance in clinical risk factors by stratifying our analyses according to donor type, which provides some degree of standardization regarding treatment regimens. We also attempted to minimize center-specific differences in phenotype ascertainment by analyzing each SNP with 2 aGVHD phenotypes, increasing the likelihood that our aGVHD phenotypes approximate those described at other centers. Finally, our cohort was larger than all of the previously studied cohorts, and hence was at least equally powered to detect the original associations.

In summary, our study demonstrates the advantages and disadvantages of a novel approach via the use of imputed SNP data in genetic analyses and sets high standards for the conduct and reporting conventions of genetic association studies in the HCT population. If HCT investigators apply these standards uniformly, their findings may be more easily interpreted and replicated by other investigators in this field, thereby significantly improving the credibility of future genetic association reports. This has the potential to increase the likelihood that non-HLA genetic studies can contribute meaningfully to advancing the current understanding of the biology of aGVHD and improve our management of this disease.

There is an Inside Blood commentary on this article in this issue.

The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Acknowledgments

This work was supported by National Institutes of Health grants AI33484, AI149213, CA015704, CA18029, HL087690, HL088201, HL094260, HL105914, and K23HL69860.

National Institutes of Health

Authorship

Contribution: J.W.C., L.P.Z., P.J.M., B.E.S., M.B., E.H.W., and J.A.H. were responsible for the study concept and design; X.C.Z., W.H.F., H.W., L.P.Z., and B.E.S. were responsible for data acquisition, statistical analysis, imputation, and informatics analyses; and J.W.C., X.C.Z., L.P.Z., P.J.M., B.E.S., M.B., E.H.W., and J.A.H. were responsible for data interpretation and writing the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Jason W. Chien, MD, MSc, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, Suite D5-280, Seattle, WA 98109-1024; e-mail: jchien@fhcrc.org.

References

References
1
Lin
MT
Storer
B
Martin
PJ
et al. 
Relation of an interleukin-10 promoter polymorphism to graft-versus-host disease and survival after hematopoietic-cell transplantation.
N Engl J Med
2003
349
23
2201
2210
2
Lin
MT
Storer
B
Martin
PJ
et al. 
Genetic variation in the IL-10 pathway modulates severity of acute graft-versus-host disease following hematopoietic cell transplantation: synergism between IL-10 genotype of patient and IL-10 receptor beta genotype of donor.
Blood
2005
106
12
3995
4001
3
Martin
PJ
McDonald
GB
Sanders
JE
et al. 
Increasingly frequent diagnosis of acute gastrointestinal graft-versus-host disease after allogeneic hematopoietic cell transplantation.
Biol Blood Marrow Transplant
2004
10
5
320
327
4
Flowers
ME
Inamoto
Y
Carpenter
PA
et al. 
Comparative analysis of risk factors for acute graft-versus-host disease and for chronic graft-versus-host disease according to National Institutes of Health consensus criteria.
Blood
2011
117
11
3214
3219
5
Price
AL
Patterson
NJ
Plenge
RM
Weinblatt
ME
Shadick
NA
Reich
D
Principal components analysis corrects for stratification in genome-wide association studies.
Nat Genet
2006
38
8
904
909
6
Cavanagh
G
Chapman
CE
Carter
V
Dickinson
AM
Middleton
PG
Donor CD31 genotype impacts on transplant complications after human leukocyte antigen-matched sibling allogeneic bone marrow transplantation.
Transplantation
2005
79
5
602
605
7
Goodman
RS
Ewing
J
Evans
PC
et al. 
Donor CD31 genotype and its association with acute graft-versus-host disease in HLA identical sibling stem cell transplantation.
Bone Marrow Transplant
2005
36
2
151
156
8
Perez-Garcia
A
De la Camara
R
Roman-Gomez
J
et al. 
CTLA-4 polymorphisms and clinical outcome after allogeneic stem cell transplantation from HLA-identical sibling donors.
Blood
2007
110
1
461
467
9
Middleton
PG
Norden
J
Cullup
H
et al. 
Oestrogen receptor alpha gene polymorphism associates with occurrence of graft-versus-host disease and reduced survival in HLA-matched sib-allo BMT.
Bone Marrow Transplant
2003
32
1
41
47
10
Mullighan
C
Heatley
S
Doherty
K
et al. 
Non-HLA immunogenetic polymorphisms and the risk of complications after allogeneic hemopoietic stem-cell transplantation.
Transplantation
2004
77
4
587
596
11
Kornblit
B
Masmas
T
Petersen
SL
et al. 
Association of HMGB1 polymorphisms with outcome after allogeneic hematopoietic cell transplantation.
Biol Blood Marrow Transplant
2010
16
2
239
252
12
Ostrovsky
O
Shimoni
A
Rand
A
Vlodavsky
I
Nagler
A
Genetic variations in the heparanase gene (HPSE) associate with increased risk of GVHD following allogeneic stem cell transplantation: effect of discrepancy between recipients and donors.
Blood
2010
115
11
2319
2328
13
Bogunia-Kubik
K
Uklejewska
A
Dickinson
A
Jarvis
M
Lange
A
HSP70-hom gene polymorphism as a prognostic marker of graft-versus-host disease.
Transplantation
2006
82
8
1116
1117
14
Bogunia-Kubik
K
Lange
A
HSP70-hom gene polymorphism in allogeneic hematopoietic stem-cell transplant recipients correlates with the development of acute graft-versus-host disease.
Transplantation
2005
79
7
815
820
15
Cullup
H
Dickinson
AM
Cavet
J
Jackson
GH
Middleton
PG
Polymorphisms of interleukin-1alpha constitute independent risk factors for chronic graft-versus-host disease after allogeneic bone marrow transplantation.
Br J Haematol
2003
122
5
778
787
16
Bertinetto
FE
Dall'Omo
AM
Mazzola
GA
et al. 
Role of non-HLA genetic polymorphisms in graft-versus-host disease after haematopoietic stem cell transplantation.
Int J Immunogenet
2006
33
5
375
384
17
MacMillan
ML
Radloff
GA
Kiffmeyer
WR
DeFor
TE
Weisdorf
DJ
Davies
SM
High-producer interleukin-2 genotype increases risk for acute graft-versus-host disease after unrelated donor bone marrow transplantation.
Transplantation
2003
76
12
1758
1762
18
Karabon
L
Wysoczanska
B
Bogunia-Kubik
K
Suchnicki
K
Lange
A
IL-6 and IL-10 promoter gene polymorphisms of patients and donors of allogeneic sibling hematopoietic stem cell transplants associate with the risk of acute graft-versus-host disease.
Hum Immunol
2005
66
6
700
710
19
Ambruzova
Z
Mrazek
F
Raida
L
et al. 
Association of IL6 and CCL2 gene polymorphisms with the outcome of allogeneic haematopoietic stem cell transplantation.
Bone Marrow Transplant
2009
44
4
227
235
20
Nordlander
A
Uzunel
M
Mattsson
J
Remberger
M
The TNFd4 allele is correlated to moderate-to-severe acute graft-versus-host disease after allogeneic stem cell transplantation.
Br J Haematol
2002
119
4
1133
1136
21
Cavet
J
Middleton
PG
Segall
M
Noreen
H
Davies
SM
Dickinson
AM
Recipient tumor necrosis factor-alpha and interleukin-10 gene polymorphisms associate with early mortality and acute graft-versus-host disease severity in HLA-matched sibling bone marrow transplants.
Blood
1999
94
11
3941
3946
22
Middleton
PG
Taylor
PR
Jackson
G
Proctor
SJ
Dickinson
AM
Cytokine gene polymorphisms associating with severe acute graft-versus-host disease in HLA-identical sibling transplants.
Blood
1998
92
10
3943
3948
23
Stark
GL
Dickinson
AM
Jackson
GH
Taylor
PR
Proctor
SJ
Middleton
PG
Tumour necrosis factor receptor type II 196M/R genotype correlates with circulating soluble receptor levels in normal subjects and with graft-versus-host disease after sibling allogeneic bone marrow transplantation.
Transplantation
2003
76
12
1742
1749
24
Sivula
J
Turpeinen
H
Volin
L
Partanen
J
Association of IL-10 and IL-10Rbeta gene polymorphisms with graft-versus-host disease after haematopoietic stem cell transplantation from an HLA-identical sibling donor.
BMC Immunol
2009
10
24
25
Socié
G
Loiseau
P
Tamouza
R
et al. 
Both genetic and clinical factors predict the development of graft-versus-host disease after allogeneic hematopoietic stem cell transplantation.
Transplantation
2001
72
4
699
706
26
Elmaagacli
AH
Koldehoff
M
Landt
O
Beelen
DW
Relation of an interleukin-23 receptor gene polymorphism to graft-versus-host disease after hematopoietic-cell transplantation.
Bone Marrow Transplant
2008
41
9
821
826
27
Gruhn
B
Intek
J
Pfaffendorf
N
et al. 
Polymorphism of interleukin-23 receptor gene but not of NOD2/CARD15 is associated with graft-versus-host disease after hematopoietic stem cell transplantation in children.
Biol Blood Marrow Transplant
2009
15
12
1571
1577
28
Wermke
M
Maiwald
S
Schmelz
R
et al. 
Genetic variations of interleukin-23R (1143A>G) and BPI (A645G), but not of NOD2, are associated with acute graft-versus-host disease after allogeneic transplantation.
Biol Blood Marrow Transplant
2010
16
12
1718
1727
29
Ambruzova
Z
Mrazek
F
Raida
L
et al. 
Possible impact of MADCAM1 gene single nucleotide polymorphisms to the outcome of allogeneic hematopoietic stem cell transplantation.
Hum Immunol
2009
70
6
457
460
30
Robien
K
Bigler
J
Yasui
Y
et al. 
Methylenetetrahydrofolate reductase and thymidylate synthase genotypes and risk of acute graft-versus-host disease following hematopoietic cell transplantation for chronic myelogenous leukemia.
Biol Blood Marrow Transplant
2006
12
9
973
980
31
Sugimoto
K
Murata
M
Onizuka
M
et al. 
Decreased risk of acute graft-versus-host disease following allogeneic hematopoietic stem cell transplantation in patients with the 5,10-methylenetetrahydrofolate reductase 677TT genotype.
Int J Hematol
2008
87
5
451
458
32
Rocha
V
Porcher
R
Fernandes
JF
et al. 
Association of drug metabolism gene polymorphisms with toxicities, graft-versus-host disease and survival after HLA-identical sibling hematopoietic stem cell transplantation for patients with leukemia.
Leukemia
2009
23
3
545
556
33
Murphy
N
Diviney
M
Szer
J
et al. 
Donor methylenetetrahydrofolate reductase genotype is associated with graft-versus-host disease in hematopoietic stem cell transplant patients treated with methotrexate.
Bone Marrow Transplant
2006
37
8
773
779
34
Holler
E
Rogler
G
Brenmoehl
J
et al. 
The role of genetic variants of NOD2/CARD15, a receptor of the innate immune system, in GvHD and complications following related and unrelated donor haematopoietic stem cell transplantation.
Int J Immunogenet
2008
35
4-5
381
384
35
Holler
E
Rogler
G
Brenmoehl
J
et al. 
Prognostic significance of NOD2/CARD15 variants in HLA-identical sibling hematopoietic stem cell transplantation: effect on long-term outcome is confirmed in 2 independent cohorts and may be modulated by the type of gastrointestinal decontamination.
Blood
2006
107
10
4189
4193
36
Holler
E
Rogler
G
Herfarth
H
et al. 
Both donor and recipient NOD2/CARD15 mutations associate with transplant-related mortality and GvHD following allogeneic stem cell transplantation.
Blood
2004
104
3
889
894
37
van der Velden
WJ
Blijlevens
NM
Maas
FM
et al. 
NOD2 polymorphisms predict severe acute graft-versus-host and treatment-related mortality in T-cell-depleted haematopoietic stem cell transplantation.
Bone Marrow Transplant
2009
44
4
243
248
38
Arora
M
Lindgren
B
Basu
S
et al. 
Polymorphisms in the base excision repair pathway and graft-versus-host disease.
Leukemia
2010
24
8
1470
1475
39
Tambur
AR
Yaniv
I
Stein
J
et al. 
Cytokine gene polymorphism in patients with graft-versus-host disease.
Transplant Proc
2001
33
1-2
502
503
40
Hattori
H
Matsuzaki
A
Suminoe
A
et al. 
Polymorphisms of transforming growth factor-beta1 and transforming growth factor-beta1 type II receptor genes are associated with acute graft-versus-host disease in children with HLA-matched sibling bone marrow transplantation.
Bone Marrow Transplant
2002
30
10
665
671
41
Takahashi
H
Furukawa
T
Hashimoto
S
et al. 
Contribution of TNF-alpha and IL-10 gene polymorphisms to graft-versus-host disease following allo-hematopoietic stem cell transplantation.
Bone Marrow Transplant
2000
26
12
1317
1323
42
Ishikawa
Y
Kashiwase
K
Akaza
T
et al. 
Polymorphisms in TNFA and TNFR2 affect outcome of unrelated bone marrow transplantation.
Bone Marrow Transplant
2002
29
7
569
575
43
Cavet
J
Dickinson
AM
Norden
J
Taylor
PR
Jackson
GH
Middleton
PG
Interferon-gamma and interleukin-6 gene polymorphisms associate with graft-versus-host disease in HLA-matched sibling bone marrow transplantation.
Blood
2001
98
5
1594
1600
44
Kim
DH
Lee
NY
Lee
MH
Sohn
SK
Vascular endothelial growth factor gene polymorphisms may predict the risk of acute graft-versus-host disease following allogeneic transplantation: preventive effect of vascular endothelial growth factor gene on acute graft-versus-host disease.
Biol Blood Marrow Transplant
2008
14
12
1408
1416
45
Hansen
JA
Chien
JW
Warren
EH
Zhao
LP
Martin
PJ
Defining genetic risk for graft-versus-host disease and mortality following allogeneic hematopoietic stem cell transplantation.
Curr Opin Hematol
2010
17
6
483
492
46
Marchini
J
Howie
B
Genotype imputation for genome-wide association studies.
Nat Rev Genet
2010
11
7
499
511
47
Fishman
D
Faulds
G
Jeffery
R
et al. 
The effect of novel polymorphisms in the interleukin-6 (IL-6) gene on IL-6 transcription and plasma IL-6 levels, and an association with systemic-onset juvenile chronic arthritis.
J Clin Invest
1998
102
7
1369
1376
48
Illig
T
Bongardt
F
Schopfer
A
et al. 
Significant association of the interleukin-6 gene polymorphisms C-174G and A-598G with type 2 diabetes.
J Clin Endocr Metab
2004
89
10
5053
5058
49
Lu
KC
Jaramillo
A
Lecha
RL
et al. 
Interleukin-6 and interferon-gamma gene polymorphisms in the development of bronchiolitis obliterans syndrome after lung transplantation.
Transplantation
2002
74
9
1297
1302
50
Muller-Steinhardt
M
Hartel
C
Muller
B
Kirchner
H
Fricke
L
The interleukin-6-174promoter polymorphism is associated with long-term kidney allograft survival.
Kidney Int
2002
62
5
1824
1827
51
Hirschhorn
JN
Lohmueller
K
Byrne
E
Hirschhorn
K
A comprehensive review of genetic association studies.
Genet Med
2002
4
2
45
61