Analysis of Single Nucleotide Polymorphisms in the Gamma Block of the Major Histocompatibility Complex in Association with Clinical Outcomes of Hematopoietic Cell Transplantation: A CIBMTR Study

Abstract

HLA haplotype mismatches have been associated with higher risk of acute graft versus host disease (aGVHD) in patients receiving HLA matched unrelated donor (URD) hematopoietic cell transplant (HCT) unrelated donors. The Gamma block (GB) is located in the central MHC region between beta and delta blocks (encoding C & B and DR & DQ antigens, respectively). GB contains many inflammatory and immune regulatory genes such as Bf, C2 and C4 genes. A single center study showed that mismatches in SNP c.2918+98G, c.3316C, and c.4385C in the GB block (C4 SNP) were associated with higher risk of grades 3–4 aGVHD. We investigated the association of GB SNP (GBS) mismatches with outcomes after 10/10 & 9/10 URD HCT (n=714). The primary outcome was aGVHD. OS, DFS, TRM, relapse, cGVHD, and engraftment were also analyzed. DNA samples were GBS genotyped by identifying 338 SNPs across 20kb using the Illumina NGS platform.

The overall 100 day incidence of aGVHD grades 2–4 & 3–4 were 41% and 17%, respectively. The overall incidence of matching at all GBS tested and at C4 SNP were 23% & 81%, respectively. Neither being matched across all GB SNP tested (vs. mismatched) nor having higher number of GBS mismatches was associated with transplant outcomes. There was no association between C4 SNP mismatches and outcomes except an unexpected significant association between having 2 C4 SNP mismatches and higher hazard ratio for relapse (association seen in 15 patients only, HR: 3.38, 95% CI: 1.75–6.53, p=0.0003). These data did not support the hypothesis that mismatching at GB is associated with outcomes after HCT.

INTRODUCTION

Human leukocyte antigen (HLA) mismatches have the strongest impact on clinical outcomes following unrelated hematopoietic cell transplantation (HCT).^{1, 2} In addition, in otherwise genotypically HLA matched unrelated donor HCT; major histocompatibility complex (MHC) haplotype mismatches have been associated with higher risk of graft versus host disease (GVHD) and with lower risk of disease relapse.³ The MHC region spans approximately 3.8 megabases (Mb) of the short arm of chromosome 6 at 6p21.3 and includes over 269 loci.^4,5 Petersdorf et al identified 12 single nucleotide polymorphisms (SNPs) as transplantation determinants.⁶ The risks associated with these SNPs were significant in a multivariate analysis that adjusted for HLA mismatching and non-genetic variables and were conferred by either patient or donor SNP genotype or by patient-donor SNP mismatching.

These SNPs may be in linkage disequilibrium (LD) with non-HLA transplant determinants. Similarly, a recent study using a combination of integrative computational approaches and random forest analysis of donor/recipient whole exome sequencing data identified 65 non-HLA variants associated with risk of antibody mediated rejection (ABMR) in kidney transplant recipients.⁷

These variants are functionally relevant to the rejection process in the kidney as they relate to genes enriched in genes expressed in kidney and vascular endothelium and underlie the immunobiology of graft rejection. However, the current standard of care in clinical HCT considers matching only at HLA-A, B, C, DRB1, DQB1 and DPB1 loci.⁸ Family pedigree analysis of HLA typing has shown that recombination occurs at specific locations within the MHC, leading to a block like structure of 4 major genomic blocks. Dawkins and colleagues referred to these blocks as the alpha, beta, gamma and delta blocks.⁹ The alpha block contains HLA-A, the beta block contains HLA-C and HLA-B, the gamma block contains the complement proteins C2, C4, and factor B (Bf) and the delta block contains the HLA DR and DQ genes.¹⁰

The Gamma block (GB) is located in the central MHC region between beta and delta blocks.¹¹ The clinical relevance of GB donor/recipient mismatches remains unclear. Investigating the clinical relevance of mismatches at GB in the clinical context of HCT is an important question because of its genomic location between the MHC class I and class II loci and the numerous inflammatory and immune regulatory genes it contains such as complement components C2, C4 and Bf genes.^12–15 In multivariate analysis, C4 mismatch was a significant risk factor associated with the development of acute GVHD in MUD HCT.¹⁶ In addition, a single center study has shown that mismatches in SNPs c.2918+98G, c.3316C, and c.4385C (reference sequence C4A NG_011638.1) in the GB block (C4A SNP) were associated with grades III - IV acute GVHD.¹⁷ In this study we tested the hypothesis that GB mismatches are associated with increased risk of GVHD after URD transplantation.

METHODS

Study Design and Population

The study population included 714 unrelated donor HCT cases receiving a HLA-A, B, C, DRB1 and DQB1 matched (10/10) (n=551) or 9/10 HLA-B mismatch only (n=163). We included adult and pediatric patients reported to the Center for International Blood and Marrow Transplant Research (CIBMTR) between 1999 and 2011 and had genomic DNA available at the CIBMTR research repository for GB SNP genotyping of recipients and their corresponding donors.

The study was limited to patients receiving a first myeloablative unrelated donor HCT for acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), and myelodysplastic syndrome (MDS). The 9/10 mismatched HCT were limited to HLA-B mismatches to maximize the frequency of GB SNP mismatches by the virtue of the genomic location of GB between HLA-B locus and class II HLA loci. All HLA typing was verified using DNA-based methods at high resolution, as previously described.¹⁸

Outcome Definitions

The primary outcomes of interest in this study were acute (Grades II-IV, III-IV and GI) and chronic (limited and extensive) GVHD. Grades II to IV and III to IV aGVHD were defined by the Glucksberg scale, and chronic GVHD (cGVHD) was defined as limited or extensive cGVHD according to the Seattle criteria.^{19, 20} Secondary outcomes analyzed included overall survival (OS), disease relapse, disease free survival (DFS), treatmentrelated mortality (TRM) and neutrophil engraftment. OS was defined as time from HCT to death from any cause. Relapse and DFS were defined per CIBMTR criteria.²¹ TRM was defined as death in continuous remission from the primary malignancy.

Engraftment was defined as achieving an absolute neutrophil count of 500/μL for 3 consecutive days.

GB SNP Genotyping and Analysis

GBS genotyping was determined by next generation sequencing platform and reagents from Illumina, Inc (San Diego, CA). DNA samples were obtained from the NMDP Research Repository. The GB region was amplified in 4 long range fragments (1.5–5.5 kbp in length), using primers and PCR mix from Conexio Genomics (Perth, Australia). The resulting amplicons were purified using the Agencourt AMPure^® XP protocol according to the manufacturers procedure. Illumina’s Nextera^® XT DNA library preparation kit and protocol was used for library creation prior to sequencing on the MiSeq^® sequencing platform. The resulting data was analyzed in Assign^™ MPS from Conexio Genomics. The genomic locations of the SNPs tested and minor allele frequencies (MAF) are shown in supplemental table S1. Out of the 338 SNPs, 10 were excluded from LD analysis due to the presence of more than two alleles in the files. This ambiguity was produced by gene overlap at these particular SNPs. Although the kit used for testing can distinguish clearly whether donor/recipient pairs were matched or mismatched at these SNPs, it is not designed to determine copy number. Four of the analyzed SNPs were 4bp insertions and deletions (InDels) and was considered 1 x variant event. A total of 325 SNPs were included in the LD and MAF analysis. Input files were formatted in the Plink pedigree format and the software Haploview was used to generate the LD plots.²²

Statistical Analysis

Variables Analyzed

Sequence mismatches were considered across all SNPs tested (fully matched vs. mismatched at 1+ SNPs), the number of GB SNP mismatches (treating GB mismatches as a continuous variable) and mismatches at C4A SNP previously reported were analyzed for different endpoints. Patient-related variables included age at time of transplant, gender and Karnofsky score. Disease-related variables included disease (ALL, AML, or MDS) and disease status (early vs. intermediate vs. advanced vs. others). Transplant-related variables included source of hematopoietic cells (bone marrow vs. peripheral blood stem cells), donor age, year of transplant, gender match (M-M vs. M-F vs. F-M vs. F-F), donor/recipient CMV status (−/− vs. −/+ vs. +/− vs. +/+ vs. unknown), conditioning regimen (myeloablative vs. reduced intensity/non-myeloablative) and GvHD prophylaxis (Tacrolimus +/−others vs. CSA +/−others vs. others). Variables considered in multivariate analyses included patient, disease and transplant characteristics, such as recipient and donor age, race/ethnicity and other factors.

Gamma Block SNP analysis:

SNP haplotypes were estimated using an accelerated EM algorithm similar to the partition/ligation method described elsewhere.²² This creates highly accurate population frequency estimates of the phased haplotypes based on the maximum likelihood as determined from the unphased input.

Descriptive Statistics, Univariable and Multivariable Analyses

Descriptive statistics included medians and ranges for continuous variables and frequencies for categorical variables. Multivariable models were built for OS, DFS, relapse, TRM, aGVHD (Grades II-IV and III-IV), and cGVHD using the Cox proportional hazards model. All the clinical variables were tested for the affirmation of the proportional hazards assumption (P<0.01). Factors violating the proportional hazards assumption were adjusted through stratification. All outcome events were defined based on competing risks, censored as appropriate and included in the cox models. Fine & Gray’s models were used for modeling the sub-distribution hazard. A stepwise variable selection procedure was then used to select adjusted clinical variables for each outcome with a threshold of 0.05 for both entry and retention in the model. Then the association of GB SNP mismatches with clinical outcomes was tested with adjustments for the selected clinical variables. The „center‟ effect was also adjusted in all the models. A p value of < 0.01 was considered significant to adjust for multiple testing. SAS version 9.4 (SAS Institute, Cary, NC) was used for all the analyses.

RESULTS

Study Population

A cohort of 714 patients was included in the analysis. The distribution of patient, donor and transplant related variables among the cohort split by GB SNP matched vs. mismatched pairs is summarized in Table 1. The majority received peripheral blood stem cell (PBSC) grafts (75%) following myeloablative conditioning (70%) and were male (53%) and Caucasian (91%) with a median age of 49 years old. AML (55%) was the most common transplant indication followed by MDS (29%) and ALL (16%), and most patients had early stage disease (70%). All patients received calcineurin inhibitorbased GVHD prophylaxis, with a Tacrolimus-based regimen as the most common (76%).

Table 1.

Summary for SNPs demographics split by All SNPs Matched vs. SNPs Mismatched

Variable	SNP Matched N (%)	SNP Mismatched N (%)	p-valuea
Number of patients	163	551
Number of centers	59	98
Recipient age at transplant			0.78
10–19 years	4 (02)	16 (03)
20–29 years	31 (19)	90 (16)
30–39 years	17 (10)	67 (12)
40–49 years	37 (23)	109 (20)
50–59 years	74 (45)	269 (49)
Median (range)	47 (18–74)	50 (18–75)	0.63
Recipient race / ethnicity			0.49
Caucasian, non-Hispanic	152 (95)	498 (92)
African-American, non-Hispanic	0	10 (02)
Asian, non-Hispanic	2 (01)	5 (01)
Native American, non-Hispanic	0	2 (<1)
Hispanic, Caucasian	5 (03)	24 (04)
Hispanic, African-American	0	1 (<1)
Hispanic, race unknown	1 (01)	1 (<1)
Unknown	3 (N/A)	10 (N/A)
Recipient sex			0.76
Male	88 (54)	290 (53)
Female	75 (46)	261 (47)
Karnofsky score			0.61
10–80	50 (34)	187 (36)
90–100	97 (66)	328 (64)
Unknown	16 (N/A)	36 (N/A)
High-resolution HLA matches available out of 10			<0.001
9/10	16 (10)	147 (27)
10/10	147 (90)	404 (73)
Graft type			0.74
Bone marrow	39 (24)	139 (25)
Peripheral blood	124 (76)	412 (75)
Conditioning regimen			0.63
Myeloablative	117 (72)	383 (70)
RIC	36 (22)	116 (21)
Nonmyeloablative	6 (04)	32 (06)
Other	4 (02)	20 (04)
Donor age at donation			0.004
10–19 years	7 (04)	17 (03)
20–29 years	68 (42)	168 (30)
30–39 years	56 (34)	175 (32)
40–49 years	25 (15)	137 (25)
50+ years	7 (04)	54 (10)
Median (range)	31 (18–56)	35 (18–61)	<0.001
Disease at transplant			0.89
AML	91 (56)	307 (56)
ALL	27 (17)	84 (15)
MDS	45 (28)	160 (29)
Disease status at transplant			0.68
Early	114 (70)	385 (70)
Intermediate	1 (01)	5 (01)
Advanced	44 (27)	137 (25)
Other	4 (02)	24 (04)
Donor / recipient CMV serostatus			0.80
Negative / negative	51 (31)	150 (27)
Negative / positive	55 (34)	188 (34)
Positive / negative	17 (10)	59 (11)
Positive / positive	36 (22)	143 (26)
Unknown	4 (02)	11 (02)
GvHD prophylaxis			0.68
Tacrolimus + MMF ± others	33 (20)	89 (16)
Tacrolimus + MTX ± others (except MMF)	92 (56)	329 (60)
CSA + MMF ± others (except Tacrolimus)	12 (07)	44 (08)
CSA + MTX ± others (except Tacrolimus, MMF)	26 (16)	89 (16)
Donor / recipient sex match			0.14
Male / male	62 (38)	200 (36)
Male / female	54 (33)	149 (27)
Female / male	26 (16)	90 (16)
Female / female	21 (13)	112 (20)
Year of transplant			0.11
2000	4 (02)	8 (01)
2001	2 (01)	12 (02)
2002	7 (04)	15 (03)
2003	7 (04)	35 (06)
2004	8 (05)	45 (08)
2005	18 (11)	80 (15)
2006	21 (13)	76 (14)
2007	31 (19)	75 (14)
2008	26 (16)	66 (12)
2009	21 (13)	63 (11)
2010	16 (10)	43 (08)
2011	2 (01)	33 (06)
Follow-up among survivors, months
N Eval	68	214
Median (range)	48.5 (12.4–140.1)	60.0 (3.3–147.6)	0.19

^aThe Pearson chi-square test was used for comparing discrete variables; the Kruskal-Wallis test was used for comparing continuous variables.

Characterization of the Gamma Block SNPs

Supplemental table S1 shows the genomic positions and minor allele frequency at each SNP. Figure 1 shows a histogram of the MAF distribution of the analyzed SNPs. Table S2 shows the pairwise LD of SNPs that reached r² < 0.8. Figure 2A & B illustrate LD among all SNPs studied and of the C4 SNPs, respectively. Figure 3A shows each haplotype in a block with its population haplotype frequency and connections from one block to the next. Figure 3B shows similar analysis of haplotype blocks estimated for the 25 SNPs analyzed in the Cleveland Clinic Cohort.¹⁷

Figure 1:

Minor allele distribution of the C4 SNPs. SNP positions are based on reference sequence: NG_011638.1 (reference sequence used in this study: agaaggtagc agacagacag acggatctaa)

Figure 2a.

Linkage disequilibrium map of analyzed Gamma block SNPs based on r² values. Colors are coded as follows: r² = 0 is given in white, 0 < r² < 1 is given in shades of grey and r² = 1 is given in black. Triangles outlined in black represent haplotype blocks.

Figure 2b.

Linkage disequilibrium map of the 25 Gamma block SNPs tested previously in the Cleveland Clinic cohort based on r² values. SNP positions are listed in the top row. Colors are coded as follows: r² = 0 is given in white, 0 < r² < 1 is given in shades of grey and r² = 1 is given in black. Triangles outlined in black represent haplotype blocks.

Figure 3a.

Details of the seven haplotype blocks outlined in figure 2a. The top row lists the SNP indices (as listed in table S1). The haplotype display shows each haplotype in a block with its population haplotype frequency and connections from one block to the next. Lines show the most common crossings from one block to the next, with thicker lines showing more common crossings than thinner lines. Shown beneath the crossing lines is multilocus D’. This represents the level of recombination between the two blocks.

Figure 3b.

Haplotype blocks estimated for the 25 CC SNPs as highlighted by the black triangles in figure 2b.

Donor/Recipient SNP Mismatches

GB SNP mismatches occurred in 551 out of the 714 (77%) of the pairs, of which 404 (73%) were 10/10 and 147 (27%) were 9/10. In the 10/10 HLA (n=551) matched pairs, GB mismatches were significantly less frequent than in the 9/10 HLA (n=163) matched pairs (73% vs. 90%, respectively). The distribution of C4 SNP mismatches in different HLA/SNP matching groups is shown in table 2.

Table 2:

The distribution of C4 SNP mismatches in different HLA/SNP matching groups

	HLA 10/10	HLA 9/10
GB SNP Matched	147	16
GB SNP Mismatched	404	147

Association between Donor/Recipient GB SNP Mismatches and Clinical Outcomes

The overall 100 day incidence of acute GVHD grades II-IV & III-IV in the entire cohort were 41% and 17%, respectively. The overall incidence of matching at all GB SNP tested and at the C4A SNP were 23% & 81%, respectively. Neither being matched across all GB SNP tested (vs. mismatched at 1+ SNPs) nor having higher number of GB SNP mismatches (treating GB mismatches as a continuous variable) was associated with any of the analyzed outcomes (Figure 4). Table 3 shows the results of the multivariable analysis of the association between GB SNP and various study outcomes. There was no association between mismatches at C4A SNP and any of the analyzed outcomes except an unexpected significant association between having 2 SNP mismatches (n=15 pairs) with higher incidence of relapse (HR: 3.38, 95% CI: 1.756.53, p=0.0003) and lower DFS (HR: 2.76, 95% CI: 1.60–4.78, p=0.0003). In a subset analysis of the 10/10 HLA matched cohort who received PBSC grafts (N=412), the C4 SNPs previously reported to associate with a higher cumulative incidence of aGVHD grades III-IV were found to be paradoxically associated with lower incidence of aGVHD (II-IV, p = 0.0053, HR = 0.52, 95% CI: 0.32–0.82) and higher relapse (p = 0.01, HR = 1.65, 95% CI: 1.12–2.44), but not grades III-IV aGVHD (p = 0.12). Competing risk analysis did not change the results.

Figure 4:

The association between SNP mismatches and clinical outcomes

Table 3.

Multivariate results comparing Gamma block (GB) SNP matched vs. mismatched

Outcome	GB SNP N	Match (ref) HR (ref)	N	GB SNP Mismatch HR (95% Confidence Interval)	p-value
aGVHD II-IV	162	1.00	551	0.94 (0.71–1.25)	0.676
aGVHD III-IV	163	1.00	551	0.92 (0.60–1.40)	0.689
cGVHD	159	1.00	545	0.88 (0.68–1.15)	0.349
OS	162	1.00	546	1.00 (0.79–1.26)	0.991
DFS	162	1.00	543	1.09 (0.88–1.35)	0.444
Relapse	162	1.00	543	1.19 (0.89–1.59)	0.245
TRM	163	1.00	548	0.85 (0.61–1.20)	0.359
Neutrophil Engraftment	162	1.00	544	0.96 (0.79–1.16)	0.646

aGVHD II-IV - Adjusted for conditioning intensity, donor race/ethnicity, graft type, HLA match, center effects and year of transplant.

aGVHD III-IV - Adjusted for HLA match, center effects and use of TBI.

cGVHD - Adjusted for graft type, use of ATG/Campath, center effects and year of transplant.

OS - Adjusted for ABO match, CMV match, disease status, graft type, KPS, HLA match, center effects and recipient age. Stratified by year of transplant.

DFS - Adjusted for CMV match, conditioning intensity, disease status, KPS, center effects and year of transplant.

Stratified by sex match.

Relapse - Adjusted for CMV match, disease status, center effects and use of ATG/Campath. Stratified by sex match and year of transplant.

TRM - Adjusted for ABO match, disease, HLA match and recipient age. Stratified by sex match, center effects and year of transplant.

Neutrophil engraftment - Adjusted for ABO match, disease status, center effects and GVHD prophylaxis. Stratified by conditioning regimen and graft type

DISCUSSION

HLA matching has been shown in numerous studies to significantly impact clinical outcomes of HCT.^{1, 2, 23,24} HCT with HLA matched sibling donors is associated with lower rates of life threatening GVHD compared to unrelated matched donors.²⁵ One possible explanation maybe that unlike matched siblings who share HLA haplotype by decent, URD despite being HLA matched, could be haplotype mismatched and therefore mismatched at other genetic variants within the MHC such as SNP, InDels and copy number variations (CNV). It is hypothesized that mismatches at non-HLA MHC sequences contribute to the inflammatory processes seen in post-transplant complications. Petersdorf et al has put this hypothesis to the test by physically separating DNA from alternate haplotypes and showing that HLA matched URD could either be haplotype matched or mismatched.³ Being haplotype matched meant that the matched HLA alleles of the donor and patient at different loci were physically linked on the same haplotype whereas being haplotype mismatched meant that matched HLA alleles were linked to different alleles at the other loci. In this study, patients who were haplotype matched with their donors had reduced risk of severe aGVHD. If these findings are reproduced, it may aid in donor selection. However, the techniques used to separate haplotypes are not technically feasible for routine HLA testing laboratories; therefore alternative techniques are required to determine the likelihood of haplotype matching. In order to develop these techniques an understanding of the MHC structure is required. Mismatches in non-HLA loci that are located in between HLA class I and class II loci such as MICA has been associated with worse HCT outcomes.^26–28 This effect has been postulated to reflect MHC haplotype heterogeneity or to be related to regulatory elements of inflammatory processes.^{26, 29} In this study we investigated the association between GB SNP mismatches and HCT outcomes. Results of this analysis did not identify significant differences in any transplant outcomes among GB SNP matches and mismatched counterparts whether mismatches were evaluated as a dichotomous “yes” vs. “no” or quantitatively as number of mismatches. These results are consistent with those of a recent single institution study, in which gamma-type matching status was not associated with outcomes of MUD HCT adult recipients.³⁰ Likewise, there was no significant association between GB mismatching or C4 SNP mismatches and the studied outcomes either. The finding of an association between C4 SNP mismatches and malignancy relapse was unexpected. We hypothesized that these mismatches would lead to higher rates of GVHD, correlating with better relapse control due to an enhanced graft-versus-leukemia effect. We report this finding cautiously due to the small sample size in the subgroup with this end point (n=15). Nevertheless, a related finding of a significant association between higher relapse and MICA mismatches was reported before in the same cohort.³¹ Further investigation of these related findings in an independent cohort could be a subject of future studies.

Our results are consistent with the results of a recent single center study that reported no significant associations between GBSP and HCT outcomes.³⁰ The study included 66 adult recipients of allogeneic HCT and GB mismatches were considered dichotomously where donor/recipient pairs were classified as either matched or mismatched. However, our results differed from those observed in a single center study that reported a significant association between GBSP and severe aGVHD.¹⁷ Possible explanations for the discordant results include having a cohort of patients with limited diagnoses (ALL, AML, and MDS) in the present study compared with the prior study that included all diagnoses. The effect of GB SNPs particularly C4 SNP mismatching may be more pronounced in the diagnoses not included in the current study. Another possible explanation is the heterogeneity introduced by inclusion of many centers in the current CIBMTR study compared to single-center studies, particularly in the scoring of aGVHD among different centers compared to more consistent scoring at a single center. It is possible that there is an over estimation of GB matched pairs. We performed a qualitative analysis of C4 SNPs. We also encountered an unexpected lower cumulative incidence of severe acute GVHD in our study population compared to the prior single center study may have rendered our study underpowered to detect statistically significant associations.

The lack of significant association between GBSP mismatches and HCT outcomes and the unexpected findings in our study does not necessarily disprove a potential clinical relevance of GB composition. Petersdorf et al identified 12 SNPs with significant associations with OS, RFS, TRM, aGVHD (II-IV & III-IV) and cGVHD.⁶ Only 6 of those SNPs conferred risks by donor/recipient mismatching (rs2242656, rs209130, rs2523957, rs3830076, rs2071479 and rs107822). The remaining SNPs conferred risks by a given recipient genotype (rs429916, rs915654 and rs2075800) or a given donor genotype (rs2244546, rs986522 and rs394657) regardless of donor/recipient matching at these SNPs. In our study, individual donor and recipient SNP genotypes were not considered because we did not have sufficient power for these analyses. In addition, since GB polymorphism involves both SNPs and CNV and CNV was not interrogated in this study, it is conceivable that in some cases CNV might have impacted clinical outcomes and were unaccounted for in our analysis. At positions where 2 SNPs were identified we were not able to determine and compare the relative copy numbers of each SNP. It is noteworthy that CNV has not been universally accounted for in prior studies and could be the included in future studies.

In conclusion, the study results show that neither GB SNP nor C4 SNP mismatches were significantly associated with HCT outcomes.

Highlights.

• The Gamma block (GB) is located in the central MHC region between the beta (encoding HLA-C and B) and delta blocks (encoding HLA-DR and DQ) and contains inflammatory and immune regulatory genes.

• Donor and recipient GB mismatching in allogeneic hematopoietic stem cell transplantation (alloHCT) associated with increased risk of grades III-IV acute graft versus host disease in a single center study.

• No associations between GB mismatching and alloHCT outcomes were observed in this large retrospective cohort from the CIBMTR.