Permissive HLA-DPB1 Mismatch and Survival after Unrelated Donor Allogeneic Stem Cell Transplantation for Hematological Malignancies: A Comparative Analysis of Different Immunogenetic Models on 422 Patients from GITMO and IBMDR

Introduction: Hematopoietic Stem Cell Transplantation (HSCT) from unrelated donors (UD) is a curative therapy for many hematologic malignancies. HLA matching plays a major role in determining HSCT outcome but the relative role of incompatibilities at the different HLA loci is still debated. In particular, over 80% of UD-HSCT are performed across HLA-DPB1 mismatches (mm): a number of previous studies have devised immunogenetic models to elucidate the impact of HLA-DPB1 mm on HSCT outcome, but a comparative analysis of these models in a recent and well-characterized cohort is lacking.

Methods: We selected 422 adult patients (pts) who received an 8/8 (HLA-A, B, -C and -DRB1) allele level-matched UD-HSCT from 2012 to 2015: of them, 382 (90%) had a mm at one or both HLA-DPB1 alleles. We classified functional HLA-DPB1 matching by four models, on the basis of: I) differential immunogenicity of alleles belonging to 3 groups of T-cell epitopes (TCE), as defined by functional studies (Zino, Blood, 2004) and refined by in silico prediction (Crivello, BBMT 2015); II) a similar model subdividing allelles in 4 TCE groups (TCE4, Crocchiolo, Blood 2009); III) differences in "delta functional distance" scores between the alleles of donor and pt, based on 12 polymorphic AA in HLA-DPB1 exon 2 (Crivello, Blood 2016); IV) mismatches in the rs9277534 single-nucleotide polymorphism in the HLA-DPB1 3′ UTR region, predicted on the basis of the DPB1 genotype (Schöne, Hum Immunol 2018), and previously shown to be associated to the expression levels of HLA-DPB1 molecules (HLAexp, Petersdorf, NEJM 2015).

Indication for HSCT was acute leukemia (55%), lymphoma and multiple myeloma (29%), myelodysplastic and myeloproliferative syndromes (16%). According to EBMT score definition, 45% of pts were in early, 26% in intermediate, and 29% in advanced disease status. Conditioning regimens were myeloablative (64%) or reduced intensity (36%). Peripheral blood was the preferred stem cell source (81%). Graft-versus-host disease (GvHD) prophylaxis was based on anti-thymocyte globulin (ATG) in 91% of pts, mostly associated with cyclosporine and methotrexate (81%). Median follow-up was 3.2 y.

Results: Among the four models adopted to classify functional HLA-DPB1 matching, the TCE4 provided the best results in predicting mm that were permissive (P) or non permissive (NP) for HSCT outcomes. By this model, P mismatched pairs (N=135) had a significantly superior 3-y overall survival (OS) and Graft-versus host disease and Relapse-Free Survival (GRFS) compared to NP pairs (N=247) (60±8% vs 49±7%, p .05; and 36±8% vs 29±5%, p .04). This was associated with a higher transplant-related mortality (TRM), 30±6% in NP mm and 21±6% in P mm, p .09 and a higher 3-y CI of extensive cGvHD in NP mm (12±4%) compared to P (4±2%), p .01 (Figure 1). No effect was found for relapse incidence. Cox multivariate analysis (adjusted for pt age, donor/host gender and CMV, disease status, Sorror score, conditioning intensity, stem cell source, ATG use, HLA matching on 5 loci, center effect), showed that a NP mm compared to P mm was associated with higher hazards for OS (HR 1.6, p .01), GRFS (HR 1.4, p .02), TRM (HR 1.9, p .01), cGvHD (HR 1.6, p .03) and extensive cGvHD (HR 3.6, p <.01). No interaction was found between HLA matching on 5 loci and HLA-DPB1 permissivity predicted by TCE4. Directionality of NP mm did not impact on clinical risk stratification.

Of the 382 transplants with HLA-DPB1 mismatches, 229 had unidirectional mismatches in GvH direction and thus could be classified by the HLAexp model. The predicted expression level of the mismatched allele in the patient was associated with 100-d CI of grade≥2 aGvHD: 32±10% in high expression (N=76) versus 16±6% in low expression (N=153) mismatched alleles, p <.01. This was also confirmed in adjusted Cox multivariate analysis for grade≥2 aGvHD (HR 2.2, p <.01). However, this did not have a significant impact on severe aGvHD, TRM and OS.

No significant associations with clinical outcomes were found for the "delta functional distance" or the TCE3 model, respectively.

Conclusions: Our study provides further proof that functional HLA-DPB1 matching is crucially associated to UD-HSCT outcome also in recent transplants, and suggest that, at least in the cohort under analysis, mainly composed of Italian pts transplanted using an ATG-based prophylaxis, the TCE4 model appears superior to other models in stratifying risk groups and predicting survival.

Figure.

View largeDownload slide

Figure.

View largeDownload slide

Close modal