Key Points
-
•
Next-generation sequencing–based RNA-sequencing methods can detect differential HLA AST levels.
-
•
Higher levels of HLA transcription are associated with increased risk of acute graft-versus-host disease after UR-HCT.
Visual Abstract
Abstract
Expression levels of HLA are associated with susceptibility to various diseases and outcomes after allogeneic hematopoietic cell transplantation (HCT). However, there has been no comprehensive analysis of the effect of the expression levels of the HLA-A, -B, -C, and -DRB1 alleles in unrelated HCT (UR-HCT). Using the capture RNA-sequencing method, we analyzed the gene expression of HLA alleles in 443 healthy donors and determined allele-specific transcription (AST) levels. We assigned median AST (M-AST) levels to HLA typing data of patients who received a transplant from unrelated donors matched for HLA-A, -B, -C, and -DRB1 alleles, using transplant registry data. All 6084 patients were divided into low, middle, and high tertile groups according to the distribution of the sum of the 2 alleles of the M-AST level for each HLA locus and the sum of the 8 alleles of the M-AST level for all 4 loci. The risk of grade 2 to 4 acute graft-versus-host disease was significantly higher in the middle group (hazard ratio [HR], 1.11; P = .044) and high group (HR, 1.20; P < .001) than in the low group for the sum of the HLA-A, -B, -C, and -DRB1 loci. Similar results were observed at the HLA-A, -B, -C, and -DRB1 loci. Higher transcription levels were also associated with a lower risk of relapse at the HLA-B, -C, and -DRB1 loci. Our data suggest that high transcription levels of patient and/or donor HLA may evoke strong alloimmune responses and affect UR-HCT outcomes.
Introduction
HLA genes are highly polymorphic, and HLA class I and class II molecules play major roles in regulating the immune system by presenting antigenic peptides to T-cell receptors. HLA has been associated with susceptibility to many immune-related diseases, including autoimmune diseases, infections, and cancers.1 In allogeneic hematopoietic cell transplantation (allo-HCT), HLA plays an important role in discriminating self from nonself, and the matching status of patient and donor HLA has a critical impact on transplant outcomes.2,3
Several studies have shown that the differential expression levels of HLA molecules affect alloimmune responses. HLA-C expression levels vary by allele, and HLA-C expression levels have been associated with the progression of HIV infection.4 In unrelated HCT (UR-HCT), increasing expression of patient mismatched HLA-C is significantly associated with an increased risk of grades 3 to 4 acute graft-versus-host disease (GVHD) and mortality.5 The single-nucleotide polymorphism rs9277534, located in the 3′ untranslated region (UTR) of the HLA-DPB1 gene, is associated with HLA-DP cell surface expression and the spontaneous clearance of hepatitis B.6 In UR-HCT, the risk of acute GVHD is increased when the patient’s mismatched HLA-DPB1 is linked to the rs9277534G allele, which is associated with high expression of HLA-DP.7 We previously demonstrated that rs9277534 represents a highly conserved HLA-DPB1 gene region from exon 3 to the 3′ UTR and may provoke acute GVHD through mechanisms distinct from those captured by the T-cell epitope mismatching algorithm,8 which is based on exon 2 polymorphisms.9 Although the impact of HLA expression at individual loci on the outcomes of UR-HCT has been analyzed, no comprehensive analysis has assessed the combined effect of the expression of all HLA-A, -B, -C, and -DRB1 alleles, which are routinely typed for patient-donor HLA compatibility.
Recent developments in next-generation sequencing (NGS) methods have enabled accurate typing of highly diverse HLA genes and further analysis of allele-specific HLA expression levels. We have developed a capture RNA-sequencing (RNA-Seq) NGS method that allows genotyping of all 12 classical HLA loci (HLA-A, -B, -C, -DPA1, -DPB1, -DQA1, -DQB1, -DRA, -DRB1, -DRB3, -DRB4, and -DRB5) and quantification of their transcription levels.10 Using this method, we demonstrated significant differences in allele-specific transcription (AST) levels for all HLA class I and class II genes except HLA-DPA1 and HLA-DRA. We hypothesized that this method could be used to determine representative values of AST levels and incorporate them into a registry of HLA typing data from UR-HSCT, allowing us to analyze how HLA transcription levels affect transplant outcomes. Against this background, in this study, we retrospectively analyzed the effect of HLA AST levels on transplant outcomes after allo-HCT in a large cohort of patients who received HLA-matched unrelated transplants.
Methods
Determination of AST levels
To measure HLA AST levels of peripheral blood mononuclear cell (PBMC) samples, we selected 443 donors from a large number of high-resolution genotyped samples obtained from 2344 UR-HCT donors through the Japan Marrow Donor Program. These donors constituted an independent cohort from the patient-donor pairs whose transplant outcomes were analyzed. Customized biotinylated nucleotide capture probes were designed and synthesized using Roche proprietary method for use with the SeqCap RNA Enrichment System (Roche NimbleGen Inc, Madison, WI). This probe set includes DNA probes designed to capture 172 HLA alleles of genomic DNA regions representative of the Japanese population,10 as well as the exon region of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (accession number NM_002046.7; 1285 base pairs). Because we designed the sequences of the RNA capture probes based on previously determined HLA allele sequences in the Japanese population, we believe that our method minimizes potential biases, such as the selective amplification or sequencing of specific HLA alleles. Methods for the isolation of RNA samples and measurement, sequence capture, NGS, and allele assignment of HLA genotypes were according to the capture RNA-Seq method reported previously.10 Mapping of sequence reads and HLA allele and GAPDH sequences was performed using GS Reference Mapper version 3.0 (Roche) according to the previously reported capture RNA-Seq method.10 The number of mapped reads was corrected per kilobase, and a correction factor was calculated by dividing 10 000 reads by the number of reads in GAPDH. Finally, AST levels were determined by multiplying the number of reads for each HLA allele by the correction factor. In this study, we compared median AST (M-AST) levels between HLA loci and between HLA alleles using HLA alleles observed in ≥3 individuals.
Study population for analyzing transplant outcomes
Data on transplant patients were collected from a national survey database called the Transplant Registry Uniform Management Program, which is managed by the Japanese Hematopoietic Cell Transplant Data Center. We included 6697 patients with acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), chronic myeloid leukemia (CML), or myelodysplastic syndrome (MDS) who received a first transplant from an unrelated donor matched for HLA-A, -B, -C, and -DRB1 alleles between 1993 and 2018. Patients in whom antithymocyte globulin was administered for GVHD prophylaxis were excluded from the analysis. Because GVHD prophylaxis using posttransplant cyclophosphamide (PTCy) is not currently used in clinical practice for HLA-matched UR-HCT in Japan, patients who received PTCy as GVHD prophylaxis were not included in this study. HLA data for patient-donor pairs before 2010 in the database incorporate HLA data from the study in which HLA genotyping was performed retrospectively.3 We applied the M-AST values for HLA alleles, which were calculated in an independent cohort, to the HLA allele typing data of these patients. After assigning M-AST levels to alleles at the HLA-A, -B, -C, and -DRB1 loci of the patients, we excluded 613 patients with ≥1 HLA allele that could not be assigned an M-AST level. Finally, a total of 6084 patients were included in this study. The study was approved by the data management committees of Transplant Registry Uniform Management Program and by the institutional review board of the Central Japan Cord Blood Bank. The study was conducted in accordance with the Declaration of Helsinki.
Statistical analysis
First, the sums of the M-AST levels of the 2 alleles at each locus and those of the 8 alleles at the HLA-A, -B, -C, and -DRB1 loci were calculated (supplemental Figure 1). Because no previous studies have analyzed the association of the expression levels of the HLA-A, -B, -C, and -DRB1 alleles with transplant outcomes, there are no available reference cutoff values. All 6084 patients were divided into low, middle, and high tertile groups according to the sum of the 2 alleles for each HLA locus and the sum of the 8 alleles for all 4 loci. Thus, the patients were divided into 3 groups of nearly equal sizes, with the low expression group serving as the reference for analyzing the risk of transplant outcomes.
The primary end point of this study was the impact of HLA transcription levels on grade 2 to 4 acute GVHD. Other end points included grade 3 to 4 acute GVHD, overall survival (OS), relapse, and nonrelapse mortality (NRM). Physicians who performed the transplants at each center diagnosed and graded acute GVHD based on traditional criteria.11 The cumulative incidences of acute GVHD and NRM were assessed using a method that eliminates the effect of competing risks.12 A competing event for acute GVHD was defined as death without acute GVHD. For NRM, relapse was the competing event. OS was defined as the time from transplantation to death, and patients who were alive at last follow-up were censored. The probability of OS was calculated using the Kaplan-Meier method.
A detailed description of methods of multivariate analysis is available in the supplemental Methods. A P value < .05 was considered significant. All analyses were performed using Stata version 17 (StataCorp LLC, College Station, TX).
Results
Sequence read information for PBMC samples
Sequence read information was obtained for all captured and enriched sequence libraries constructed from the 443 PBMC samples. The total number of draft reads was 705 445 938, with individual read counts ranging from 556 352 to 3 589 356 (mean ± standard deviation, 1 588 842 ± 496 283). These reads were of high quality, with quality values >30 and an average quality value of 34.0 ± 0.8. The draft read bases totaled 176.4 gigabases, ranging from 139.1 to 897.3 megabases (average, 397.2 ± 124.1). Therefore, these high-quality reads had sufficient sequencing volume for further HLA genotyping and transcription analyses.
AST levels
Based on the HLA genotyping, a total of 85 alleles were identified without phase ambiguity. Of these, 67 (10 HLA-A, 22 HLA-B, 14 HLA-C, and 21 HLA-DRB1) were alleles with M-AST levels determined in ≥3 individuals (Table 1; Figure 1). These alleles accounted for >95.3% of the cumulative allele frequency of known HLA alleles in the Japanese population at the field-2 resolution level (an allele resolution level that distinguishes sequences differing by a nonsynonymous substitution) across the 4 loci (Table 1).
Table 1.
M-AST levels of HLA alleles determined in 3 or more individuals
| HLA-A |
HLA-B |
HLA-C |
HLA-DRB1 |
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Allele | Freq, % | M-AST level | n | Allele | Freq, % | M-AST level | n | Allele | Freq, % | M-AST level | n | Allele | Freq, % | M-AST level | n |
| A∗26:03 | 2.5 | 108 549 | 18 | B∗15:07 | 0.7 | 146 299 | 5 | C∗03:03 | 13.1 | 112 525 | 68 | DRB1∗14:05 | 2.1 | 31 444 | 7 |
| A∗33:03 | 7.4 | 116 122 | 87 | B∗44:02 | 0.4 | 172 473 | 3 | C∗08:03 | 1.5 | 116 554 | 5 | DRB1∗04:10 | 2.1 | 34 134 | 6 |
| A∗26:02 | 2.0 | 126 775 | 11 | B∗44:03 | 6.6 | 183 845 | 81 | C∗05:01 | 0.4 | 120 718 | 3 | DRB1∗04:03 | 3.0 | 38 394 | 20 |
| A∗02:01 | 11.4 | 132 491 | 54 | B∗15:01 | 7.5 | 186 434 | 55 | C∗03:02 | 0.7 | 124 205 | 5 | DRB1∗14:54 | 3.4 | 38 560 | 13 |
| A∗31:01 | 8.8 | 132 866 | 42 | B∗67:01 | 1.2 | 206 987 | 9 | C∗14:03 | 6.7 | 133 171 | 78 | DRB1∗13:02 | 6.1 | 39 131 | 77 |
| A∗02:06 | 9.3 | 135 799 | 39 | B∗48:01 | 2.9 | 208 590 | 14 | C∗08:01 | 7.4 | 133 517 | 38 | DRB1∗13:01 | 0.6 | 40 210 | 4 |
| A∗11:01 | 8.9 | 144 596 | 72 | B∗40:06 | 4.6 | 242 275 | 24 | C∗12:02 | 11.3 | 144 148 | 171 | DRB1∗08:03 | 8.2 | 40 443 | 73 |
| A∗26:01 | 7.4 | 147 317 | 46 | B∗39:01 | 3.5 | 243 893 | 25 | C∗01:02 | 16.9 | 147 703 | 128 | DRB1∗04:05 | 13.1 | 41 715 | 109 |
| A∗02:07 | 3.4 | 156 803 | 31 | B∗40:02 | 7.6 | 245 059 | 51 | C∗03:04 | 12.0 | 157 356 | 69 | DRB1∗16:02 | 0.9 | 41 818 | 6 |
| A∗24:02 | 35.7 | 181 594 | 249 | B∗54:01 | 7.5 | 247 910 | 72 | C∗15:02 | 3.0 | 170 788 | 8 | DRB1∗04:06 | 3.3 | 44 185 | 31 |
| B∗52:01 | 11.2 | 248 145 | 163 | C∗07:02 | 13.0 | 173 340 | 138 | DRB1∗15:01 | 7.6 | 45 454 | 28 | ||||
| B∗46:01 | 4.6 | 255 425 | 36 | C∗14:02 | 7.2 | 186 187 | 34 | DRB1∗12:01 | 3.7 | 46 055 | 10 | ||||
| B∗35:01 | 8.2 | 271 169 | 46 | C∗06:02 | 0.9 | 233 114 | 3 | DRB1∗15:02 | 10.6 | 46 466 | 157 | ||||
| B∗07:02 | 5.8 | 279 239 | 102 | C∗04:01 | 4.5 | 235 793 | 31 | DRB1∗12:02 | 1.8 | 47 376 | 6 | ||||
| B∗58:01 | 0.6 | 281 009 | 5 | DRB1∗04:01 | 1.0 | 47 669 | 4 | ||||||||
| B∗40:01 | 5.4 | 283 607 | 22 | DRB1∗09:01 | 14.7 | 49 516 | 91 | ||||||||
| B∗56:01 | 1.0 | 289 001 | 3 | DRB1∗11:01 | 2.6 | 54 693 | 8 | ||||||||
| B∗51:01 | 9.0 | 301 093 | 37 | DRB1∗14:06 | 1.3 | 55 416 | 10 | ||||||||
| B∗59:01 | 2.0 | 315 057 | 25 | DRB1∗01:01 | 5.9 | 55 566 | 104 | ||||||||
| B∗55:02 | 2.5 | 367 337 | 5 | DRB1∗08:02 | 4.2 | 57 711 | 20 | ||||||||
| B∗13:01 | 1.3 | 431 309 | 6 | DRB1∗14:03 | 1.6 | 62 775 | 15 | ||||||||
| B∗15:18 | 1.5 | 579 702 | 3 | ||||||||||||
| 10 | 96.9∗ | 153 242† | 649 | 22 | 95.3∗ | 251 422† | 792 | 14 | 98.4∗ | 155 017† | 779 | 21 | 97.8∗ | 45 581† | 799 |
Freq, frequency.
Indicates the cumulative frequency of HLA alleles. The HLA allele frequency data are available from the HLA laboratory webpage (https://hla.or.jp/med/frequency_search/en/allele/).
Indicates the M-AST level at each HLA locus.
Figure 1.
AST levels of the HLA-A, -B, -C, and -DRB1 loci in PBMC samples and measured by capture RNA-Seq. (A-D) Box-and-whisker diagrams; the vertical axis indicates the AST level (log2) calculated as described in the “Methods”; the horizontal axis indicates the alleles for each of the HLA loci: HLA-A (A), HLA-B (B), HLA-C (C), and HLA-DRB1 (D). (E) M-AST levels at the HLA-A, -B, -C, and -DRB1 loci.
Patient characteristics
Patient and transplantation characteristics are shown in Table 2. The median age at transplantation was 44 years (range, 0-77). Among the 6084 patients, the most common disease was AML (n = 2850 [47%]), followed by ALL (n = 1690 [28%]), MDS (n = 1010 [16%]), and CML (n = 534 [9%]). Most patients underwent bone marrow transplantation (n = 5871 [96%]), with just 213 patients (4%) undergoing peripheral blood stem cell transplantation. The median follow-up of survivors was 5.2 years (range, 0.1-5.2).
Table 2.
Patient characteristics
| Patient variable | n (%) |
|---|---|
| Patient age, median (range), y | 44 (0-77) |
| Donor age, median (range), y | 36 (20-55) |
| Donor-to-patient sex mismatch | |
| Match | 3586 (59) |
| Male to female | 1523 (25) |
| Female to male | 925 (16) |
| Missing | 20 (0) |
| Disease | |
| AML | 2850 (47) |
| ALL | 1690 (28) |
| CML | 534 (9) |
| MDS | 1010 (16) |
| Leukemia risk | |
| Standard | 4559 (75) |
| Advanced | 1462 (24) |
| Missing | 63 (1) |
| Performance status | |
| 0-1 | 4865 (80) |
| 2-4 | 324 (5) |
| Missing | 895 (15) |
| HCT-CI | |
| 0 | 2605 (43) |
| 1-2 | 1005 (16) |
| ≥3 | 592 (10) |
| Missing | 1882 (31) |
| Source of stem cells | |
| Bone marrow | 5871 (96) |
| Peripheral blood | 213 (4) |
| Conditioning regimen | |
| Myeloablative | 4757 (78) |
| Reduced intensity | 1201 (20) |
| Missing | 126 (2) |
| GVHD prophylaxis | |
| Cyclosporine based | 1628 (27) |
| Tacrolimus-based | 4360 (72) |
| Others | 96 (2) |
| Year of transplantation | |
| 1993-2000 | 840 (14) |
| 2001-2009 | 1898 (31) |
| 2010-2018 | 3346 (55) |
HCT-CI, HCT-specific comorbidity index.
Impact of M-AST levels on grade 2 to 4 acute GVHD
The impact of the sum of M-AST levels at the HLA-A, -B, -C, and -DRB1 loci on grade 2 to 4 acute GVHD are shown in Table 3. At the HLA-A locus, the risk of grade 2 to 4 acute GVHD was significantly higher in the middle (hazard ratio [HR], 1.18; P = .001) and high groups (HR, 1.10; P = .048) than in the low group. At the HLA-B locus, the risk of grade 2 to 4 acute GVHD did not significantly differ between the low and middle groups but was significantly higher in the high group than in the low group (HR, 1.18; P = .001). At the HLA-C locus, the risk of grade 2 to 4 acute GVHD was significantly higher in the middle (HR, 1.14; P = .009) and high groups (HR, 1.16; P = .005) than in the low group. At the HLA-DRB1 locus, the risk of grade 2 to 4 acute GVHD was significantly higher in the middle group than in the low group (HR, 1.13; P = .016), whereas there was no significant difference between the high and low groups.
Table 3.
Impact of M-AST levels on grade 2 to 4 and 3 to 4 acute GVHD
| Grade 2-4 acute GVHD |
Grade 3-4 acute GVHD |
||||||
|---|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | n | |
| HLA-A | |||||||
| Low | 1.00 | 1.00 | 2102 | ||||
| Middle | 1.18 | 1.07-1.31 | .001 | 1.23 | 1.02-1.48 | .027 | 2040 |
| High | 1.11 | 1.00-1.23 | .048 | 1.22 | 1.01-1.47 | .040 | 1942 |
| HLA-B | |||||||
| Low | 1.00 | 1.00 | 2044 | ||||
| Middle | 1.08 | 0.98-1.20 | .117 | 1.21 | 1.01-1.47 | .042 | 2103 |
| High | 1.18 | 1.07-1.31 | .001 | 1.31 | 1.08-1.58 | .005 | 1937 |
| HLA-C | |||||||
| Low | 1.00 | 1.00 | 2168 | ||||
| Middle | 1.14 | 1.03-1.26 | .009 | 1.21 | 1.02-1.45 | .033 | 2211 |
| High | 1.16 | 1.04-1.29 | .005 | 1.11 | 0.92-1.35 | .280 | 1705 |
| HLA-DRB1 | |||||||
| Low | 1.00 | 1.00 | 2118 | ||||
| Middle | 1.13 | 1.02-1.25 | .016 | 1.11 | 0.92-1.33 | .292 | 1995 |
| High | 1.01 | 0.91-1.12 | .813 | 1.10 | 0.91-1.32 | .320 | 2011 |
| HLA-A, -B, -C, and -DRB1 | |||||||
| Low | 1.00 | 1.00 | 2043 | ||||
| Middle | 1.11 | 1.00-1.23 | .044 | 1.14 | 0.94-1.37 | .189 | 2013 |
| High | 1.20 | 1.09-1.33 | <.001 | 1.29 | 1.08-1.56 | .006 | 2028 |
HR adjusted for clinical factors as listed in Table 2.
We also examined the impact of the sum of M-AST levels of the 8 alleles at the HLA-A, -B, -C, and -DRB1 loci on grade 2 to 4 acute GVHD. The risk of grade 2 to 4 acute GVHD was significantly higher in the middle (HR, 1.11; P = .044) and high groups (HR, 1.20; P < .001) than in the low group. The cumulative incidences of grade 2 to 4 acute GVHD in the low, middle, and high groups of the sum of the HLA-A, -B, -C, and -DRB1 loci were 34.5% (95% confidence interval [CI], 32.4-36.5), 37.2% (95% CI, 35.1-39.3), and 39.6% (95% CI, 37.5-41.7), respectively (Figure 2A).
Figure 2.
Grade 2 to 4 and grade 3 to 4 acute GVHD classified by the sum of the AST levels of HLA-A, -B, -C, and -DRB1. (A) Cumulative incidence of grade 2 to 4 acute GVHD in patients in the low, middle, and high groups. (B) Cumulative incidence of grade 3 to 4 acute GVHD in patients in the low, middle, and high groups.
We performed subgroup analysis to confirm the absence of interactions between the effect of HLA expression on acute GVHD and various clinical factors (patient age, disease type, conditioning regimen, GVHD prophylaxis, and disease status at transplantation). There was a trend toward a higher risk of GVHD with higher transcription levels in various subgroups. Disease stratification showed that allele expression affected the risk of acute GVHD in AML and acute ALL but not in CML and MDS (Figure 3).
Figure 3.
Subgroup analysis of the effects of the AST level on grade 3 to 4 acute GVHD classified by the sum of the AST levels of HLA-A, -B, -C, and -DRB1. The risk of grade 2 to 4 acute GVHD in the middle and high groups was compared with that in the low group. The results are adjusted using multivariate analysis. CyA, cyclosporine; MAC, myeloablative conditioning; RIC, reduced-intensity conditioning; Tac, tacrolimus.
Impact of M-AST levels on grade 3 to 4 acute GVHD
The impact of M-AST levels on grade 3 to 4 acute GVHD was similar to that on grade 2 to 4 acute GVHD, although there were some differences (Table 3). At the HLA-A locus, the risk of grade 2 to 4 acute GVHD was significantly higher in the middle (HR, 1.23; P = .027) and high groups (HR, 1.22; P = .040) than in the low group. At the HLA-B locus, the risk of grade 3 to 4 acute GVHD was significantly higher in the middle (HR, 1.21; P = .042) and high groups (HR, 1.31; P = .005) than in the low group. At the HLA-C locus, the risk of grade 3 to 4 acute GVHD was significantly higher in the middle group than in the low group (HR, 1.21; P = .033). At the HLA-DRB1 locus, the risk of grade 3 to 4 acute GVHD did not significantly differ between the middle and low groups or between the high and low groups. Regarding the sums of M-AST levels at the HLA-A, -B, -C, and -DRB1 loci, the high group had a significantly higher risk of grade 3 to 4 acute GVHD than the low group (HR, 1.29; P = .006). The cumulative incidences of grade 3 to 4 acute GVHD in the low, middle, and high groups of the sums of the HLA-A, -B, -C, and -DRB1 loci were 10.1% (95% CI, 8.8-11.4), 11.3% (95% CI, 9.9-12.7), and 12.8% (95% CI, 11.4-14.3), respectively (Figure 2B).
Impact of M-AST levels on relapse, chronic GVHD, NRM, and OS
The impacts of M-AST levels on relapse, chronic GVHD, NRM, and OS are shown in Table 4. At the HLA-A locus, there was a nonsignificant trend toward a lower risk of relapse in the high group than the low group (HR, 0.90; P = .081). At the HLA-B and HLA-DRB1 loci, the high group had a significantly lower risk of relapse than the low group (HLA-B: HR, 0.88 [P = .046]; HLA-DRB1: HR, 0.88 [P = .030]). At the HLA-C locus, the risk of relapse was significantly lower in the middle group than in the low group (HR, 0.87; P = .026). Regarding the effect of all sums of M-AST levels at the HLA-A, -B, -C, and -DRB1 loci, there were no significant differences in the risk of relapse in the high and middle groups compared with the low group.
Table 4.
Impact of M-AST levels on relapse, chronic GVHD, NRM, and OS
| Relapse |
Chronic GVHD |
NRM |
OS |
n | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | HR | 95% CI | P value | HR | 95% CI | P value | ||
| HLA-A | |||||||||||||
| Low | 1.00 | 1.00 | 1.00 | 1.00 | 2102 | ||||||||
| Middle | 0.96 | 0.85-1.08 | .468 | 1.00 | 0.91-1.11 | .919 | 0.94 | 0.82-1.07 | .347 | 0.98 | 0.90-1.09 | .812 | 2040 |
| High | 0.90 | 0.79-1.01 | .081 | 1.08 | 0.98-1.20 | .127 | 0.98 | 0.86-1.11 | .745 | 0.94 | 0.85-1.03 | .177 | 1942 |
| HLA-B | |||||||||||||
| Low | 1.00 | 1.00 | 1.00 | 1.00 | 2044 | ||||||||
| Middle | 0.94 | 0.83-1.05 | .272 | 1.06 | 0.96-1.18 | .235 | 1.07 | 0.94-1.22 | .324 | 1.02 | 0.93-1.12 | .621 | 2103 |
| High | 0.88 | 0.78-0.99 | .046 | 1.06 | 0.96-1.18 | .258 | 1.10 | 0.96-1.25 | .186 | 1.02 | 0.93-1.12 | .659 | 1937 |
| HLA-C | |||||||||||||
| Low | 1.00 | 1.00 | 1.00 | 1.00 | 2168 | ||||||||
| Middle | 0.87 | 0.78-0.98 | .026 | 0.99 | 0.90-1.10 | .893 | 1.02 | 0.90-1.16 | .707 | 0.94 | 0.86-1.03 | .212 | 2211 |
| High | 0.99 | 0.88-1.12 | .870 | 0.94 | 0.85-1.05 | .291 | 1.02 | 0.90-1.18 | .695 | 1.07 | 0.98-1.19 | .137 | 1705 |
| HLA-DRB1 | |||||||||||||
| Low | 1.00 | 1.00 | 1.00 | 1.00 | 2118 | ||||||||
| Middle | 0.93 | 0.82-1.05 | .215 | 1.01 | 0.91-1.12 | .915 | 1.00 | 0.88-1.15 | .946 | 1.00 | 0.91-1.10 | .984 | 1995 |
| High | 0.88 | 0.78-0.99 | .030 | 1.06 | 0.96-1.17 | .265 | 1.03 | 0.91-1.18 | .624 | 0.99 | 0.90-1.08 | .780 | 2011 |
| HLA-A, -B, -C, and -DRB1 | |||||||||||||
| Low | 1.00 | 1.00 | 1.00 | 1.00 | 2043 | ||||||||
| Middle | 0.94 | 0.83-1.06 | .286 | 0.97 | 0.88-1.08 | .609 | 1.01 | 0.88-1.15 | .916 | 0.98 | 0.89-1.08 | .712 | 2013 |
| High | 0.91 | 0.81-1.03 | .140 | 1.02 | 0.92-1.13 | .716 | 1.08 | 0.95-1.23 | .251 | 1.05 | 0.95-1.15 | .342 | 2028 |
HR adjusted for clinical factors as listed in Table 2.
There were no significant differences in chronic GVHD, NRM, and OS at all HLA-A, -B, -C, and -DRB1 loci in the middle and high groups compared with the low group.
Discussion
In this study, we analyzed the effects of the transcription levels of HLA genes on transplant outcomes after UR-HCT. To detect the effects of the transcription levels of HLA genes themselves without excluding the effects of patient-donor HLA incompatibility, we analyzed patients who received a transplant from unrelated donors matched for HLA-A, -B, -C, and -DRB1 alleles. Because peripheral blood samples from a large sample of patients who underwent UR-HCT were not available for direct analysis of HLA transcription level, we used the M-AST levels determined in an analysis of HLA-genotyped PBMC samples collected from 443 bone marrow transplant donors from the Japan Marrow Donor Program as the HLA transcription levels of patient/donor HLAs in the large-scale registry data of UR-HCT. We found that a high transcription level of HLA genes was associated with a higher risk of acute GVHD.
Polymorphisms in the noncoding region of HLA genes have been associated with variations in HLA expression levels.13 For instance, sequence polymorphisms in the 5′ UTR for the HLA-DQB1 gene are associated with HLA-DQB1 transcription levels.14 In addition, allelic lineage-specific methylation patterns within the promoter region of the HLA-A gene correlate with HLA-A expression levels.15 Moreover, variations in the microRNA miR-148 binding site within the 3′ UTR of the HLA-C gene are related to the expression levels of HLA-C via posttranscriptional regulation.16
There have been numerous reports of associations between HLA variations and disease susceptibility.17 HLA expression has also been linked to disease susceptibility, particularly for autoimmune diseases and infectious diseases. For example, a high expression level of HLA-C is associated with reduced progression of HIV infection but an increased risk of Crohn disease.18 In addition, upregulation of HLA class I is observed in Graves disease.19 Finally, disease-susceptibility HLA-DRB1 alleles for rheumatoid arthritis are highly expressed in patients with this condition.20 These multiple lines of evidence suggest that high HLA expression may induce strong immune responses. In allo-HCT, high HLA-C5 and HLA-DPB17 expression levels of mismatched patient HLA alleles have been shown to be associated with acute GVHD. In these previous studies, methods such as quantitative polymerase chain reaction assays and flow cytometry were applied to quantify HLA expression at the mRNA and protein levels. Although these methods provide valuable data, they do not give allele-specific, high-resolution information on HLA expression.
The recent development of RNA-based NGS methods has enabled simultaneous expression analysis and HLA allele-specific expression quantification.21 Using our previously developed capture RNA-Seq method,10 we show here that the transcription levels of the HLA-A, -B, -C, and -DRB1 genes are associated with the risk of acute GVHD in HLA-matched UR-HCT. In HLA-compatible allo-HCT, GVHD is mainly induced by minor histocompatibility antigens on HLA molecules of patient organs.22 Different levels of HLA expression in patients may result in alloimmune responses of varying intensities in donor T cells that are elicited by minor histocompatibility antigens.7,23
Interestingly, we found that HLA-B∗15:01, which has previously been linked to a low immune response, had the fourth lowest M-AST level of the 22 HLA-B alleles analyzed in this study (Table 1; Figure 1). In the Japanese population, HLA-B∗15:01 accounts for 90% of HLA-B62 alleles. We have previously demonstrated that HLA-B62 is associated with a lower risk of acute GVHD and a higher risk of leukemia relapse in patients who received a transplant from a HLA-identical sibling donor.24 In the HLA-matched UR-HCT analyzed in this study, patients with HLA-B∗15:01 tended to have a lower risk of grade 2 to 4 acute GVHD (HR, 0.89; 95% CI, 0.77-1.09; P = .091) and had a significantly lower risk of grade 3 to 4 acute GVHD (HR, 0.73; 95% CI, 0.55-0.96; P = .023). In a study by the National Marrow Donor Program/Center for International Blood and Marrow Transplant Research, the positivity of the HLA-B62 supertype (of which HLA-B∗15:01 accounted for 80% in their study) was associated with a lower risk of NRM in 8/8 HLA-matched UR-HCT.25 In their study, there was also a trend toward fewer GVHD-related deaths among patients with the HLA-B62 supertype, as well as a significantly higher rate of deaths secondary to disease progression. The HLA-B62 supertype is associated with poor outcomes in patients with advanced melanoma treated with immune checkpoint blockade, and distinct structural and dynamic elements of HLA-B∗15:01 have the potential to impair the effective CD8+ T-cell recognition of neoantigens.26 The low immunogenicity of HLA-B∗15:01 remains consistent across different ethnic groups, which may be partly due to its low expression levels.
In our previous study, we found that the 3 common Japanese HLA haplotypes were highly conserved among unrelated individuals and that the second most frequent HLA-A∗33:03-C>∗14:03-B∗44:03-DRB1∗13:02-DQB1∗06:04-DPB1∗04:01 haplotype is associated with a lower risk of acute GVHD in UR-HCT.27 Among patients who received transplants from an HLA-A–, -B–, -C–, and -DRB1–matched unrelated donor in this study, patients with the HLA-A∗33:03-C∗14:03-B∗44:03-DRB1∗13:02 haplotype showed a significantly lower risk of grade 2 to 4 (HR, 0.87; 95% CI, 0.78-0.98; P = .017) and grade 3 to 4 acute GVHD (HR, 0.76; 95% CI, 0.62-0.94; P = .011). All 4 alleles making up this haplotype exhibited a low M-AST level at each locus: HLA-A∗33:03 was the second lowest of the 10 HLA-A alleles; HLA-B∗44:03 was the third lowest of the 22 HLA-B alleles; HLA-C∗14:03 was the fifth lowest of the 14 HLA-C alleles; and HLA-DRB1∗13:02 was the fifth lowest of the 21 HLA-DRB1 alleles (Table 1; Figure 1). A previous study demonstrated that the 3 conserved extensive haplotypes associated with autoimmune diseases showed haplotype-specific differences in gene expression.28 A more recent study revealed that polymorphisms near HLA-A, HLA-C, C4A, and HLA-DRB regions within conserved extended haplotypes exert haplotype-specific regulatory effects by mapping these sequence variations to known expression quantitative trait loci.29 The low expression of all HLA alleles making up the HLA-A∗33:03-C∗14:03-B∗44:03-DRB1∗13:02-haplotype, which is associated with a lower risk of acute GVHD, may also be due to haplotype-specific mechanisms for regulating expression.
This model should also be capable of analyzing HLA-mismatched transplants. Differences in peptide-binding motifs between patients and donors have been shown to affect transplant outcomes in HLA-DP–mismatched30 and HLA class I–mismatched transplants.31 To clarify the effects of HLA transcription levels on HLA-mismatched transplants, we need analyses that consider the impact of HLA mismatch, such as a peptide-binding motif model. In addition, the new standard GVHD prophylaxis PTCy is now widely used in haploidentical transplants. Given that PTCy facilitates the overcoming of HLA barriers in allo-HCT, our model must be reevaluated in the PTCy GVHD prophylaxis setting.
This study has several limitations. First, it is unclear whether AST levels determined in healthy donors will reflect HLA expression patterns in the patients’ target organs for acute GVHD. Although it has been reported that HLA expression patterns vary among tissues and cells,32 it is not easy to analyze HLA expression levels of biopsied tissues from transplant patients. Indeed, the analysis of this study revealed that the impact of acute GVHD was fairly modest and did not affect chronic GVHD or OS. In the future, an atlas of allele-specific HLA expression patterns in blood cells and various tissues will allow for more detailed analysis. In addition, further study is needed to determine how to incorporate HLA transcription levels into donor selection algorithms. Second, we were unable to analyze the effects of the transcription levels of HLA class II genes other than HLA-DRB1 because HLA-DQB1 and -DPB1 data were missing from most patients in our database. It has been reported that differences in the expression levels of the HLA-DP molecule, which are divided into high and low expression groups based on a polymorphism in the 3′ UTR of the HLA-DPB1 gene, are associated with the risk of acute GVHD.7 As mentioned above, the HLA-A∗33:03-C∗14:03-B∗44:03-DRB1∗13:02-haplotype is highly conserved up to HLA-DPB1∗04:01, which belongs to the low expression HLA-DPB1 group. Therefore, HLA-DPB1 may potentially exert a latent influence as a confounding factor in this analysis. The effect of the AST level of HLA-DPB1 on transplant outcomes requires further analysis. Third, the protein expression of HLA molecules is not determined by RNA-based NGS methods. The expression of HLA molecules on the cell surface depends on the initial transcription, the allele-specific rate of peptide-HLA complex formation, and the cell surface stability.13 Future studies also need to verify the expression at the protein level. Fourth, regarding HLA-DRB1, the M-AST calculated from the total PBMC sample might be affected by the proportion of cells expressing HLA class II in each sample. However, it was not possible to measure HLA-DRB1 expression in isolated B cells and monocytes in this study. Fifth, the probes that we developed for the capture RNA-Seq method do not guarantee capture of alleles common to ethnic groups other than Japanese. However, we believe that it would be possible to extend our method to other populations by using data on common, intermediate, and well-documented alleles.33
In conclusion, using the capture RNA-Seq method, we were able to detect differences in the transcription levels of HLA alleles that are highly prevalent in Japanese individuals and determine M-AST levels. Our data suggest that a high transcription level of patient and/or donor HLA may evoke a strong alloimmune response and affect transplant outcomes in UR-HCT. This result is in line with earlier reports that showed an association of certain alleles or haplotypes with the risk of acute GVHD. To clarify the clinical significance of HLA allele–level expression in allo-HCT, more detailed analyses, including expression in tissues and leukemic cells, are warranted.
Conflict-of-interest disclosure: The authors declare no competing financial interests.
A complete list of the members of the JSTCT HLA Working Group appears in the supplemental Appendix.
Acknowledgments
The authors thank S. Suzuki, A. Shigenari, S. Ito, C. Yahata, and K. Asami at Tokai University School of Medicine for their technical support. They also thank all physicians and data managers who contributed valuable transplantation data to the Japanese Society for Transplantation and Cellular Therapy and the members of their data management committee for their assistance.
This work was supported by grants from the Japan Agency for Medical Research and Development (JP22ek0510032 and JP25ek0510049 [S.M., T.S., and M.M.]) and the Japan Society for the Promotion of Science KAKENHI (23K07863 [S.M.]).
Authorship
Contribution: S.M., T.S., Y.M., and M.M. designed research, organized the project, performed statistical analysis, and analyzed data; F.A. and S. Tomori interpreted data; N.D., T.N., T.F., T.E., K.K., Y.H., M.T., S.O., S. Takada, Y. Katayama, N.U., H.N., Y. Kanda, N.H., J.K., M.O., and T.I. gathered and organized data; Y.A. managed the unified registry database; S.M. and T.S. wrote the first draft of the manuscript; and F.A., S. Tomori, N.D., T.N., T.F., T.E., K.K., Y.H., M.T., S.O., S. Takada, Y. Katayama, N.U., H.N., Y. Kanda, N.H., J.K., M.O., T.I., Y.A., Y.M., and M.M contributed to the final version.
Footnotes
S.M. and T.S. contributed equally to this study.
Data are available from the corresponding author, Satoko Morishima (satoko.morishima@gmail.com), on request.
The full-text version of this article contains a data supplement.
Contributor Information
Satoko Morishima, Email: satoko.morishima@gmail.com.
Takashi Shiina, Email: tshiina@tokai.ac.jp.
Supplementary Material
References
- 1.Shiina T, Hosomichi K, Inoko H, Kulski JK. The HLA genomic loci map: expression, interaction, diversity and disease. J Hum Genet. 2009;54(1):15–39. doi: 10.1038/jhg.2008.5. [DOI] [PubMed] [Google Scholar]
- 2.Sasazuki T, Juji T, Morishima Y, et al. Effect of matching of class I HLA alleles on clinical outcome after transplantation of hematopoietic stem cells from an unrelated donor. Japan Marrow Donor Program. N Engl J Med. 1998;339(17):1177–1185. doi: 10.1056/NEJM199810223391701. [DOI] [PubMed] [Google Scholar]
- 3.Morishima Y, Kashiwase K, Matsuo K, et al. Biological significance of HLA locus matching in unrelated donor bone marrow transplantation. Blood. 2015;125(7):1189–1197. doi: 10.1182/blood-2014-10-604785. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Apps R, Qi Y, Carlson JM, et al. Influence of HLA-C expression level on HIV control. Science. 2013;340(6128):87–91. doi: 10.1126/science.1232685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Petersdorf EW, Gooley TA, Malkki M, et al. HLA-C expression levels define permissible mismatches in hematopoietic cell transplantation. Blood. 2014;124(26):3996–4003. doi: 10.1182/blood-2014-09-599969. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Thomas R, Thio CL, Apps R, et al. A novel variant marking HLA-DP expression levels predicts recovery from hepatitis B virus infection. J Virol. 2012;86(12):6979–6985. doi: 10.1128/JVI.00406-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Petersdorf EW, Malkki M, O'HUigin C, et al. High HLA-DP expression and graft-versus-host disease. N Engl J Med. 2015;373(7):599–609. doi: 10.1056/NEJMoa1500140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Morishima S, Shiina T, Suzuki S, et al. Evolutionary basis of HLA-DPB1 alleles affects acute GVHD in unrelated donor stem cell transplantation. Blood. 2018;131(7):808–817. doi: 10.1182/blood-2017-08-801449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Zino E, Frumento G, Marktel S, et al. A T-cell epitope encoded by a subset of HLA-DPB1 alleles determines nonpermissive mismatches for hematologic stem cell transplantation. Blood. 2004;103(4):1417–1424. doi: 10.1182/blood-2003-04-1279. [DOI] [PubMed] [Google Scholar]
- 10.Yamamoto F, Suzuki S, Mizutani A, et al. Capturing differential allele-level expression and genotypes of all classical HLA Loci and haplotypes by a new capture RNA-Seq method. Front Immunol. 2020;11:941. doi: 10.3389/fimmu.2020.00941. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus conference on acute GVHD grading. Bone Marrow Transplant. 1995;15(6):825–828. [PubMed] [Google Scholar]
- 12.Gooley TA, Leisenring W, Crowley J, Storer BE. Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med. 1999;18(6):695–706. doi: 10.1002/(sici)1097-0258(19990330)18:6<695::aid-sim60>3.0.co;2-o. [DOI] [PubMed] [Google Scholar]
- 13.Petersdorf EW, O'HUigin C. The MHC in the era of next-generation sequencing: implications for bridging structure with function. Hum Immunol. 2019;80(1):67–78. doi: 10.1016/j.humimm.2018.10.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Beaty JS, West KA, Nepom GT. Functional effects of a natural polymorphism in the transcriptional regulatory sequence of HLA-DQB1. Mol Cell Biol. 1995;15(9):4771–4782. doi: 10.1128/mcb.15.9.4771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ramsuran V, Kulkarni S, O'Huigin C, et al. Epigenetic regulation of differential HLA-A allelic expression levels. Hum Mol Genet. 2015;24(15):4268–4275. doi: 10.1093/hmg/ddv158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kulkarni S, Savan R, Qi Y, et al. Differential microRNA regulation of HLA-C expression and its association with HIV control. Nature. 2011;472(7344):495–498. doi: 10.1038/nature09914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Dendrou CA, Petersen J, Rossjohn J, Fugger L. HLA variation and disease. Nat Rev Immunol. 2018;18(5):325–339. doi: 10.1038/nri.2017.143. [DOI] [PubMed] [Google Scholar]
- 18.O'Huigin C, Kulkarni S, Xu Y, et al. The molecular origin and consequences of escape from miRNA regulation by HLA-C alleles. Am J Hum Genet. 2011;89(3):424–431. doi: 10.1016/j.ajhg.2011.07.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Weider T, Richardson SJ, Morgan NG, Paulsen TH, Dahl-Jørgensen K, Hammerstad SS. HLA class I upregulation and antiviral immune responses in Graves disease. J Clin Endocrinol Metab. 2021;106(4):e1763–e1774. doi: 10.1210/clinem/dgaa958. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Houtman M, Hesselberg E, Rönnblom L, Klareskog L, Malmström V, Padyukov L. Haplotype-specific expression analysis of MHC class II genes in healthy individuals and rheumatoid arthritis patients. Front Immunol. 2021;12 doi: 10.3389/fimmu.2021.707217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Johansson T, Partanen J, Saavalainen P. HLA allele-specific expression: methods, disease associations, and relevance in hematopoietic stem cell transplantation. Front Immunol. 2022;13 doi: 10.3389/fimmu.2022.1007425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Akatsuka Y, Morishima Y, Kuzushima K, Kodera Y, Takahashi T. Minor histocompatibility antigens as targets for immunotherapy using allogeneic immune reactions. Cancer Sci. 2007;98(8):1139–1146. doi: 10.1111/j.1349-7006.2007.00521.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Martin PJ. Increased disparity for minor histocompatibility antigens as a potential cause of increased GVHD risk in marrow transplantation from unrelated donors compared with related donors. Bone Marrow Transplant. 1991;8(3):217–223. [PubMed] [Google Scholar]
- 24.Morishima S, Fukuda T, Doki N, et al. Individual HLAs influence immunological events in allogeneic stem cell transplantation from HLA-identical sibling donors. Bone Marrow Transplant. 2021;56(3):646–654. doi: 10.1038/s41409-020-01070-3. [DOI] [PubMed] [Google Scholar]
- 25.Camacho-Bydume C, Wang T, Sees JA, et al. Specific class I HLA supertypes but not HLA zygosity or expression are associated with outcomes following HLA-matched allogeneic hematopoietic cell transplant: HLA supertypes impact allogeneic HCT outcomes. Transplant Cell Ther. 2021;27(2) doi: 10.1016/j.bbmt.2020.10.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Chowell D, Morris LGT, Grigg CM, et al. Patient HLA class I genotype influences cancer response to checkpoint blockade immunotherapy. Science. 2018;359(6375):582–587. doi: 10.1126/science.aao4572. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Morishima S, Ogawa S, Matsubara A, et al. Impact of highly conserved HLA haplotype on acute graft-versus-host disease. Blood. 2010;115(23):4664–4670. doi: 10.1182/blood-2009-10-251157. [DOI] [PubMed] [Google Scholar]
- 28.Vandiedonck C, Taylor MS, Lockstone HE, et al. Pervasive haplotypic variation in the spliceo-transcriptome of the human major histocompatibility complex. Genome Res. 2011;21(7):1042–1054. doi: 10.1101/gr.116681.110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lam TH, Shen M, Tay MZ, Ren EC. Unique allelic eQTL clusters in human MHC haplotypes. G. 2017;7(8):2595–2604. doi: 10.1534/g3.117.043828. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Meurer T, Crivello P, Metzing M, et al. Permissive HLA-DPB1 mismatches in HCT depend on immunopeptidome divergence and editing by HLA-DM. Blood. 2021;137(7):923–928. doi: 10.1182/blood.2020008464. [DOI] [PubMed] [Google Scholar]
- 31.Crivello P, Arrieta-Bolaños E, He M, et al. Impact of the HLA Immunopeptidome on survival of leukemia patients after unrelated donor transplantation. J Clin Oncol. 2023;41(13):2416–2427. doi: 10.1200/JCO.22.01229. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Boegel S, Löwer M, Bukur T, Sorn P, Castle JC, Sahin U. HLA and proteasome expression body map. BMC Med Genomics. 2018;11(1):36. doi: 10.1186/s12920-018-0354-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Hurley CK, Kempenich J, Wadsworth K, et al. Common, intermediate and well-documented HLA alleles in world populations: CIWD version 3.0.0. HLA. 2020;95(6):516–531. doi: 10.1111/tan.13811. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.