Clinical Relevance of HLA Supertype Matching after Myeloablative Conditioning 7/8 Unrelated Donor Hematopoietic Cell Transplantation: A CIBMTR Study

The diversity of the HLA class I and II alleles can be simplified by consolidating them into fewer supertype clusters based on functional or predicted structural similarities in epitope binding grooves of HLA molecules. HLA class I and II supertypes have been increasingly studied in association with immune susceptibility to infection and cancer with potential implications for vaccine development. However, the significance of individual allele mismatching within and outside of HLA class I or II supertypes remains unknown in the context of hematopoietic cell transplantation (HCT). We therefore studied the impact of HLA supertype disparities on clinical outcomes of 1934 patients with AML (45%), ALL (31%), CML (14%) or MDS (9%) who underwent 7/8 unrelated donor myeloablative conditioning HCT from 1999 to 2011 and were registered with CIBMTR. Median age at transplant was 35 years (range, 1-70); 53% were males; 81% Caucasian; 56% received peripheral blood grafts; 50% were ABO-mismatched; 36% had in-vivo T-cell depletion; 62% received tacrolimus- and 36% cyclosporine A-based GVHD prophylaxis; 72% male or non-parous female donors; median follow up of survivors was 54 months (3-149). Supertype assignment methods of (1) revised main HLA anchor specificities (Sydney, 2008) and (2) bioinformatics (Doytchinova, 2004-05) were used to categorize single mismatched alleles into 6 HLA-A (A01, A01A03, A01A24, A02, A03, A24), 6 HLA-B (B07, B08, B27, B44, B58, B62), 2 HLA-C (C1, C2), and 5 DRB1 (DR1, DR3, DR4, DR5, DR9) supertypes. Overall survival (OS), disease-free survival (DFS), relapse, treatment-related mortality (TRM), acute graft vs. host disease (aGVHD) and chronic GVHD were compared across matched vs. mismatched HLA-A (265 vs. 429), -B (230 vs. 92), -C (365 vs. 349), and -DRB1 (153 vs. 51) supertypes. We used predetermined α=0.01 for statistical significance as multiple exploratory analyses were conducted by Kaplan-Meier, Gray, and Cox proportional hazard methods. In the multivariable analysis, supertype B-mismatch was associated with increased risk of grade II-IV aGVHD (HR=1.78; 95% CI, 1.23-2.59, p=0.0025), however no difference was found for grade III-IV aGVHD or other clinical outcomes compared to supertype B-matches. Supertype DRB1-mismatch was associated with shorter neutrophil recovery (HR=0.51; 95% CI, 0.36-0.71, p=0.0001), yet a trend toward inferior OS (HR=1.58; 95% CI 1.04-2.38, p=0.037) and higher TRM (HR=1.64; 95% CI, 0.99-2.74, p=0.0565) compared to DRB1 matches within supertypes. There was no increased risk of GVHD with DRB1 supertype mismatch. No associations were observed between HLA-A and -C supertypes or aggregate supertype-matched vs. -mismatched groups for any outcomes. Our analysis demonstrated differential influence of HLA supertype-based allele matching within -B and -DRB1 loci on clinical outcomes after myeloablative 7/8 URD HCT.