Prediction of Graft-Versus-Host Disease in Recipients of Single Mismatched Unrelated Hematopoietic Cell Transplantation Donor Using a Highly Multiplexed Proteomic Assay, MHC-Pepseq

Graft-versus-host disease (GVHD) remains a major cause of treatment failure after allogeneic hematopoietic cell transplantation (alloHCT). In HLA-mismatched donor setting, indirect presentation of allogeneic peptides from recipient's mismatched HLA class I or II proteins by donor or recipient antigen presenting cells can be an immunogenic driver of GVHD. However, the potential diversity of such antigens is large, and predicting them in a systematic manner has proven challenging. Using a novel, highly-multiplexed peptide-MHC binding assay (MHC-PepSeq) we sought to 1) identify allogeneic peptides derived from mismatched HLA protein that can be efficiently presented by HLA-DR, and 2) explore the possibility that the frequency of these HLA-DRB1 binding allopeptides may be predictive of clinical GVHD in HLA-DPB1 mismatched donor/recipient pairs.

Using publicly-available population allele frequency data (allelefrequencies.net), we identified a set of class I and II sequences that cover >95% of alleles at each of 9 human HLA-loci (-A, -B, -C, -DRA1,-DRB1, -DQA1, -DQB1, -DPA1, -DPB1) in 3 major US populations (European Caucasian, African American, Mexican Chicano). When represented in the form of densely overlapping tiled 15-mer peptides, 7,744 unique 15mers were identified. We encoded these peptides into DNA oligonucleotides and used the PepSeq parallel synthesis protocol to generate a library of the corresponding DNA-barcoded peptides. The library was incubated with recombinantly-expressed full-length HLA proteins, washed, eluted, amplified, and sequenced to identify the various HLA-derived peptides that bind to the assayed HLA proteins (Figure 1). In the current study, DPB1-derived allopeptides in the setting of HLA-A, B, C, DRB1, and DQB1 (10/10) matched unrelated (MUD) HCT donors with a mismatch in DPB1 were investigated. The peptide library was assayed for binding to the DRB1*07:01 protein, selected since it was the common allele in this cohort.

We identified 327 patients who were transplanted at our center and met these criteria. For each case, we used comprehensive in silico tiling to identify HLA-DPA and DPB-derived peptides present in the recipient but absent in the donor. This set was intersected with the peptides identified as binders to HLA-DRB1*07:01 in the 7,744-plex MHC-PepSeq assay, to arrive at a donor-recipient pair-specific set of 'allopeptides' Overall, we identified such allopeptide at the median of 0 (range: 0-8) across the 327 cases.

Next, we examined an association between the number of allopeptides and acute GVHD in the cohort of 94 patients with positive HLA-DRB1*07:01. Median age at alloHCT was 60 years (range: 19-78), 53% males, 1.% bone marrow graft and only 7% female to male donors. Ablative (TBI) conditioning was delivered to 34%) pts. 83% received Tacrolimus/Sirolimus-based, and 9% received post-transplant cyclophosphamide-based GVHD prophylaxis. Patient/HCT characteristics are summarized in Table 1.

In this cohort, 18% had no DPB1 mismatch, 54% had a single and 28% had double mismatches, with 21% pts carrying non-permissive DPB1 mismatches. Allopeptide score was 0 in 75% of pts. Non-permissive mismatch 9 (39%) vs. 11 (16%) were more likely to have allopeptide score ≥1 and similarly double mismatches 11 (48%) vs. 15 (21%) were more likely to have allopeptide score of ≥1. Among pts with ≥1 allopeptide score 14 (61%) had DPB1 matched or permissive mismatch.

The cumulative incidence of grade 2-4 acute GVHD was 40.8% (range: 29-52) in pts with no allopeptides from DPB1 compared with 56% (range: 34-74) in those with ≥1 allopeptides (p=0.259) (Figure 2). The cumulative incidence of grade 3-4 acute GVHD and chronic GVHD were similar between allopeptide 0 vs. ≥1.

Together, we show that the "MHC-PepSeq" assay can identify novel candidate HLA-derived allopeptides in 10/10 MUD HCTs. The number of such peptides are relatively low - with a majority having no allopeptide. In an exploratory analysis in a selected cohort of patients with HLA-DRB1*0701 in the setting of 10/10 MUD HCT, the number of allopeptides in our assay may be predictive of GVHD. The expanded analyses on other HLA-DRB1 restriction elements are underway.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal