Optimizing Donor Selection for Haploidentical Transplantation Utilizing the Four-Group T Cell Epitope (TCE-4) Algorithm for Prediction of HLA-DPB1 Non-Permissive Mismatches

An HLA-DPB1 non-permissive mismatch (npmm) has been associated with higher risks of acute graft-versus-host disease and non-relapse mortality after matched unrelated donor transplantation (MUDT) and thus avoiding HLA-DPB1 npmm is important in unrelated donor selection. In contrast, HLA-DPB1 npmm by the 3-group T cell epitope algorithm (TCE-3) has been shown to be protective in the context of haploidentical donor transplantation (HIDT) using post-transplant cyclophosphamide (PTCy) (Solomon et al. BBMT 2018). Additional HLA-DPB1 "permissiveness" models, also based on cross-reactive patterns of alloreactive T cells against HLA-DPB1 alleles, include the TCE-4 (based on 4 TCE groups) and functional distance (FD, based on net difference in distance between key amino acid polymorphisms) algorithms. Lastly, an expression model is based on the surface expression of the recipient mismatched HLA-DPB1 (RDP) allele according to variants of a biallelic SNP. The present analysis had 2 major aims: to 1) determine which HLA-DPB1 permissiveness model (TCE-3, TCE-4, FD, expression) provides the best tool for haploidentical donor selection and 2) analyze the role of vector (GVH vs. HVG direction) on the effect of an HLA-DPB1 npmm. A total of 322 patients with acute leukemia, MDS, lymphoma, CLL or CML, receiving a HIDT-PTCy from a single institution were evaluated with a median follow-up time of 45 months [range 6, 184]. Baseline characteristics included a median age of 50 years [19, 80], 47% non-white, HCT-CI ≥3 in 50%, PBSC graft in 80%, and myeloablative conditioning in 49%. The number of donor-recipient pairs having an HLA-DPB1 npmm according to the TCE-3, TCE-4, FD and expression was 82 (25%), 130 (40%), 54 (17%) and 99 (31%) respectively. In univariate analysis, HLA-DPB1 npmm identified by the TCE-3 and TCE-4 models were statistically associated with improved overall survival (OS) (p=0.041 and p=0.004 respectively), whereas FD risk and RDP expression were not (see figure). For disease-free survival (DFS), only the TCE-4 model showed a statistically significant association (p=0.022). Directionality of the HLA-DP npmm (GVH vs. HVG vector) had no significant impact on survival following HIDT-PTCy, a finding similar to the context of MUDT (Fleischhauer BMT 2017). In multivariate Cox analysis, adjusting for patient/donor age, gender, race, HLA-DR mismatch and transplant year, HLA-DPB1 npmm by the TCE-4 model had the most significant association with improved OS (HR 0.59, p=0.012), with TCE-3 being less predictive (HR 0.65, p=0.07) (see figure). Furthermore, HLA-DPB1 npmm by TCE-4 led to improved DFS (HR 0.69, p=0.046) and trends for lower cumulative incidences of relapse/progression (HR 0.73, p=0.16) and NRM (HR 0.54, p=0.09). In summary, the presence of an HLA-DP npmm in either the GVH or HVG direction continues to be associated with improved survival following HIDT-PTCy in a large single institution retrospective analysis with extended follow-up. Compared to the original TCE-3 model, a TCE-4-predicted HLA-DPB1 npmm is more strongly associated with overall survival. This fact, combined with the larger number of HLA-DPB1 npmm donors identified by the TCE-4 model, suggests that it may be a better selection tool for optimal haploidentical donor identification.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal