Leukocytes that have lost the cell surface expression of an HLA class I allele due to somatic HLA gene mutations or 6p loss of heterozygosity (LOH) are often present in patients with immune aplastic anemia (AA), consistent with T cell-mediated bone marrow cell destruction. A recent NIH study suggested that the clinical significance and underlying pathology of HLA loss may differ according to the type of HLA allele affected (Zaimoku et al. Blood 2021) because (1) the significant correlation of HLA loss with clonal evolution was primarily attributed to patients who lost HLA-B*14:02 and A*02:01; (2) all patients who lost B*40:02 and A*02:06 achieved a hematologic response to immunosuppressive therapy (IST), though HLA loss in general was not correlated with the IST response; (3) loss of B*07:02 and B*40:01 was associated with older age of onset; (4) patients who retained B*14:02, B*07:02, and B*40:01 also showed clinical features of patients who lost respective HLA alleles. We aimed to confirm the HLA class I allele-related clinical manifestations in Japanese AA patients. HLA loss was assessed by SNP array, HLA flow cytometry, digital PCR, or deep nucleotide sequencing. An HLA allele was classified as "affected by loss" in individual patients when the single allele was lost or inactivated by somatic mutations or when both HLA-A and HLA-B alleles in a haplotype were lost due to 6p LOH with no evidence of loss or inactivation of a single HLA allele. Clinical characteristics were compared according to the affected HLA alleles. In 496 Japanese patients with AA (severe, n = 368; non-severe, n = 128; median age, 59 [range, 1-89] years), there were 146 (29%) patients with HLA loss. HLA-B*40:02, A*02:06, A*31:01, and B*54:01 were more frequently affected in our cohort than in the NIH cohort, where B*14:02, A*02:01, B*08:01, and B*07:02 held the majority of lost alleles (Figure 1). The median age of patients with B*40:01 loss (75 years) was highest among the groups of patients with HLA allele loss, consistent with the NIH study. The prevalence of ≥0.003% paroxysmal nocturnal hemoglobinuria (PNH) clones was significantly reduced in patients with B*40:02 loss but not in others. Sex and severity of AA were not markedly different across the groups of patients with HLA allele loss. Outcomes after anti-thymocyte globulin (ATG)-based IST were evaluable in 263 of the 496 patients. Although HLA loss was correlated with the IST response at 6 months (as reported by our group), loss of HLA-A*02:01 did not contribute to the high response, consistent with the NIH study. Loss of frequently affected HLA alleles in Japanese patients (B*40:02, A*02:06, B*54:01, etc.) was generally correlated with the IST-response. Further, patients who carried 5 HLA-B alleles (B*40:01, B*40:02, B*44:03, B*54:01, and B*56:01) without loss still displayed higher response rates than other patients. In contrast, the favorable effect of A*02:06 loss was not seen in patients who carried A*02:06 without loss. A*02:06 loss, the presence of the 5 HLA-B alleles (irrespective of loss), and previously reported HLA class II alleles (DRB1*13 and DRB1*15) were independently correlated with the IST response. The incidence of clonal evolution to myeloid neoplasms after ATG-based IST was significantly higher in patients with HLA-A*02:01 loss than other patients (Figure 2), as observed in the NIH study; A*02:01 was affected in 4 of 7 patients showing both HLA loss and clonal evolution in our cohort; B*40:02, B*44:02, and B*52:01 were affected in other 3 patients. In summary, our study in Japanese AA patients broadly confirmed the associations of loss of specific HLA alleles with clinical manifestations observed in American patients, including older age of onset in patients with HLA-B*40:01 loss, high IST response rates in patients who lost B*40:02 and A*02:06 but not in patients with A*02:01 loss, and frequent clonal evolution in patients with A*02:01 loss. We also revealed the correlation of 5 HLA-B alleles with IST response, irrespective of loss, similar to the relationship of B*14:02 to clonal evolution in the United States, and the reduced prevalence of PNH clones in patients with B*40:02 loss. These findings strongly suggest that the heterogeneous pathology of AA can be classified by affected HLA alleles. This should be useful for studies to unravel the pathophysiology and improve the outcomes of patients with immune AA. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal