Abstract 2541


The loss or down-regulation of HLA class-I antigens often occurs in solid tumors, but it is rarely demonstrated in leukemic cells. This phenomenon may explain why leukemic cells are immunogenic to some degree with allogeneic hematopoietic stem cell transplantation (allo-SCT) or donor leukocyte infusion, because HLA class-I antigens on leukemic cells are thought to present minor histocompatibility antigens (mHAs) and leukemia-associated antigens (LAAs) to donor T cells to elicit anti-leukemic cytotoxic T-cell (CTL) responses. Recent analyses of leukemic cells that relapsed after HLA-haploidentical SCT revealed that the leukemic cells lose their unshared HLA haplotype expression as a result of acquired uniparental disomy (UPD) on the short arm of chromosome 6 (6p). The loss of HLA from leukemic cells may occur after transplantation from HLA-identical donors if CTLs specific to mHAs or LAAs play a substantial role in the eradication of leukemic cells. This study evaluated this hypothesis by investigating HLA class-I expression on leukemic cells from patients at both the time of diagnosis and relapse after allo-SCT from HLA matched and mismatched donors.


Leukemic cells were obtained from five patients with myeloid leukemia (1 CML, 4 AML) both at diagnosis and relapse after allo-SCT. HLA class-I expression on leukemic cells was determined by flow cytometry using monoclonal Abs specific for the HLA-A allele. SCT donors were an HLA-matched sibling, an HLA-A locus-mismatched mother, an HLA-B and C locus-mismatched father, and 2 HLA-matched unrelated donors. The copy number-neutral loss of heterozygosity on 6p in leukemic cells was analyzed by a single-nucleotide polymorphism (SNP) array in patients after haploidentical SCT. The origin of the patients' HLA alleles (paternal or maternal) was determined by family studies on HLA. The post-transplantation donor T-cell responses against the leukemic cells were analyzed with an IFN-γ secretion assay.


HLA-A expression of leukemic cells obtained at relapse after allo-SCT was down-regulated in all 5 patients compared to that of leukemic cell obtained at diagnosis. The patient possessing HLA-A2/A11 (Case 1) was the recipient of BM possessing HLA-A24/A11 showed that 66% of leukemic cells at relapse were deficient in HLA-A2 expression. Leukemic cells restored HLA-A2 expression when this patient relapsed after the 2nd SCT using HLA-A2-matched cord blood, but at lower level in comparison to the level at diagnosis (MFI at diagnosis 177 vs. MFI at relapse 101). In vitro, cell-surface HLA-A2 expression was completely restored when leukemic cells at the 2nd relapse were treated in culture with IFN-γ (200 U/mL) for 48 hr. HLA-A2 expression in the shared haplotype (A0206-B3902-Cw0702-DR0405) was unexpectedly missing in 12% of the relapsed leukemic cells in another patient (Case 2) with HLA-2 loci (HLA-B and -C)-mismatched BM (A3101-B5601-Cw0401-DR0901), and this proportion increased to 83% at relapse after 2nd allo-SCT from the same donor (Figure). IFN-γ failed to restore HLA-A2 expression (MFI without IFN-γ 32 vs. MFI with IFN-γ 10) in vitro, and the genomic DNA extracted from leukemic cells at relapse showed a UPD of 6p. Donor-derived T-cells stimulated with leukemia cells from case 2 at diagnosis in vitro secreted IFN-γ in response to leukemia cells at diagnosis better than leukemia cells at relapse (1.67% vs. 0.85%).


Loss of HLA-class I antigen occurs frequently in myeloid leukemia cells at relapse after allo-SCT regardless of the HLA disparities. T-cell attacks specific to the major as well as the mHAs, and LAAs may favor proliferation of leukemic cells deficient in the expression of HLA class-I.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.