Moderate Immunopeptidome Disparity of HLA-DRB1 in HLA-Mismatched Hematopoietic Stem Cell Transplantation Increases Acute Graft-Versus-Host Disease but Tends to Improve Survival

Introduction

In allogeneic hematopoietic stem cell transplantation (allo-HCT), the immune response of alloreactive T-cells plays a crucial role. HLAs, the key molecules presenting antigens to T-cells, are highly polymorphic and vary widely among individuals. The capacity of T-cells to present antigens, referred to as the immunopeptidome, is assessed through HLA evolutionary divergence (HED) between alleles. In HLA-mismatched allo-HCT, mismatches arise between the patient's HLA molecule/peptide complexes and the peptide structures recognized by the donor T-cells. The degree of these mismatches varies, affecting donor T-cell immune responses. While these immunopeptidome disparities can enhance antitumor effects, concerns about the increased risk of harmful graft-versus-host disease (GVHD) remain. Known risk factors for complications include the number of HLA mismatch loci and specific high-risk mismatches (Morishima, Haematologica, 2016). However, the myriad combinations of HLA mismatches make individual considerations impractical.

Aim

We examined the impact of immunopeptidome disparities between recipients and donors on HLA-mismatched allo-HCT.

Methods

Between February 2009 and February 2024, 341 patients underwent their first allo-HCT procedure at our institution. Among the 203 patients with evaluable HLA-A, -B, -C, and -DRB1 allele data, 146 had acute leukemia. We retrospectively analyzed 77 patients (AML n=53, ALL n=24) who received HLA-mismatched allo-HCT. The types of HLA mismatches were one-locus mismatch (26%), two-locus mismatch (6.5%), cord blood (27.3%), and haploidentical donor (40.3%). The median age at allo-HCT was 44 [17–68] years, and the median observation period was 646 [47–4788] days. HLA allele data were obtained from clinical records. The immunopeptidome disparity of HLA mismatch (HED-mis) was calculated for mismatched allele pairs in the graft-versus-host direction using the Grantham distance, which measures evolutionary distance based on amino acid chemistry (published in a custom Perl script by Tobias Lenz et al., https://sourceforge.net/projects/granthamdist/files/). HED-mis was calculated for HLA-A, -B, -C, and -DRB1, and denoted as HED-misA, -misB, -misC, and -misDRB1, respectively. We evaluated acute GVHD (aGVHD), non-relapse mortality (NRM), cumulative incidence of relapse (CIR), and overall survival (OS).

Result

The median [range] HED-mis values were: HED-mis A: 8.47 [0.12–13.04], HED-mis B: 7.50 [1.09–11.25], HED-mis C: 5.49 [0.69–7.58], and HED-mis DRB1: 9.47 [0.73–18.17]. After dividing each HED-mis score into quartiles;Q1, Q2, Q3, and Q4, we first compared two groups: low (Q1+Q2) and high (Q3+Q4). Pairs with matched alleles (HED-mis: 0) and known high-risk mismatched alleles (various HED-mis) were excluded from the comparison.

There was a significant increase in grade 2-4 aGVHD in the low group for HED-mis DRB1 (p=0.0043). In the multivariate analysis adjusted for graft source, the low group for HED-mis DRB1 was a risk factor for grade 2-4 aGVHD (HR 10.2, p=0.023). To further examine the effect of the degree of HED-mis DRB1, we compared Q1 and Q2 in the low group separately. Q1 and Q2 had similar incidences of grade 2-4 aGVHD (p=0.95); however, the frequency of grade 3-4 aGVHD was lower in the Q2 group (12.5% vs. 30.8% in the Q1 group). The 2-year CIR was similar between the Q1 and Q2 groups (p=0.85), but the 2-year NRM was lower in the Q2 group (0.0% vs. 30.7% in Q1 on day +100, p=0.019). Final OS was better in Q2 patients with HED-mis DRB1 (p=0.039), even when HLA-DRB1 matched patients were included in the comparison. Multivariate analysis adjusted for stage, graft source, diagnosis, conditioning, and HCT-CI suggested that Q2 for HED-mis DRB1 was associated with better OS (HR 0.27, p=0.06).

For other HLA loci, the high group for HED-mis A showed a trend towards increased grade 3-4 aGVHD at 17.4% (n=4/23) compared to 0% (n=0/13) in the low group (p=0.11) at day +100 and a trend towards lower CIR (p=0.14). However, this analysis did not show any impact on the final OS.

Conclusion

Among HLA-DRB1 mismatches, moderate disparity (Q2) in the immunopeptidome was associated with increased aGVHD but lower NRM and better OS compared to mild or severe disparity. Further validation of other HLA loci is needed, although the degree of immunopeptidome disparities may be considered in HLA-mismatched allo-HCTs.