The Influence of HLA Matching of the Winning and Losing Units in Relapse of Patients Receiving Double Cord Blood Transplantation for Acute Leukemia or Myelodysplastic Syndrome

Background

HLA mismatches adversely impact transplant outcomes by increasing the risk of graft-versus-host disease (GVHD) and rejection. In leukemic relapse after haplo-identical transplant, the mismatched HLA-haplotype is deleted, indicating that HLA might be involved in graft-versus-leukemia (GVL) responses. In adult cord blood transplants (CBT), relapse is reduced after HLA-mismatched compared to matched grafts. During double CBT, one cord emerges as the “winning unit” (WU). ) Based on earlier translational observations (Lamers CH et al BLOOD 2011 (117) and Cornelissen JJ et al Stem Cell Investigation 2017 (26), we hypothesised that unit predominance may be HLA-restricted and if the losing unit (LU) and recipient share the same HLA mismatch with the WU, the same T-cell response that mediates the rejection of the LU will also be directed at the leukemic clone, reducing the relapse risk.

Methods

Retrospective study by Eurocord, Royal Manchester Children’s Hospital and the Cellular Therapy & Immunobiology Working Party of EBMT.

Eligibility: Patients (children and adults) with acute leukemia or myelodysplastic syndrome (MDS) who engrafted after unrelated double CBT transplant at EBMT centers between 2005 and 2022, with no in-vivo T cell depletion, available HLA data and known WU. We analysed the impact of disease, patient and CBT characteristics on relapse and other CBT outcomes. The influence of HLA-mismatch on CBT outcomes considered classical typing (HLA-A,-B and -DRB1), with HLA-C, and whether the recipient and LU shared an HLA mismatch not present in the WU. HLA typing was compared at low resolution for class I HLA and high resolution for HLA-DRB1.

Results

In 383 eligible CBTs, median follow-up of surviving patients was 71.5 months and median year of CBT was 2012. Median age at CBT was 42.9 years, 59.8% of patients were male; most had acute leukemia (n=344; AML n=102; ALL n=232), 57.5% and 30.4% were in CR1 or CR2, respectively. Most received reduced intensity conditioning (RIC) regimens (56.5%) and total body irradiation (90%).

Considering the most mismatched unit, 64% transplants were 4/6, 3% were 5/6 and only 1% were 6/6 (HLA-A, -B and -DRB1). Including HLA-C, 41% were 6/8, 32% were 5/8, 12.5% were 7/8 and 0.5% were 8/8. In 49% cases there was an HLA-mismatch in the WU to a specificity shared between the LU and the patient. This mismatch was at HLA-A in 17.8% transplants, at -B in 15.1%, at -C in 20.1% and at -DRB1 in 12.3%, and possibly at multiple loci simultaneously.

Relapse was increased in multivariate analysis (MVA) in those with leukemia rather than MDS, in those with AML rather than ALL, in those with RIC regimen, and in those transplanted before 2013. In univariate analysis (UVA), relapse was reduced where the recipient and LU shared the same mismatch with the WU at HLA-A (16% vs 28%, p=0.08) or HLA-B (15% vs 28%, p=0.07), but not HLA-C (22% vs 27%, p=0.48) or -DRB1 (28% vs 26%, p=0.82). If the mismatch was at either HLA-A or -B (combined in the same variable) relapse was reduced compared to other mismatches (16% vs 30%; p=0.027). The same effect (not statistically significant), was observed in MVA with a 35.5% decrease in risk where there was a shared mismatch in HLA-A or -B between patient and LU.

In MVA, HLA-mismatches showed no effect in acute GVHD which was only increased in those receiving myeloablative conditioning or were cytomegalovirus (CMV) positive. A shared mismatch between LU and recipient did not influence GVHD risk. For chronic GVHD, there was no influence of HLA mismatch, but a decreased risk was observed in older subjects and those transplanted before 2013. Of note, UVA showed decreased incidence of chronic GVHD where the recipient and LU shared an HLA-DRB1 mismatch against the WU.

Discussion

CBT is associated with reduced relapse in those with acute leukaemia compared to other cell sources, likely mediated by an augmented GVL response in a transplant that is more often HLA-mismatched and is performed T-cell replete. In this series of double CBT, we provide data that support an HLA-restricted response directed at the LU by the WU, mediated by either HLA-A or HLA-B. This will reduce relapse if the disease bears the same mismatch (LU primes the immune response against the disease). There is no increase in GVHD with HLA mismatching in CBT, but there is a tendency for decreased chronic GVHD if the LU and patient share the same HLA-DRB1 mismatch against the WU.