Somatic Loss of HLA Class I Alleles Is a Common Genetic Alteration in Acquired Aplastic Anemia and Reveals Aplastic Anemia Risk Alleles

Acquired aplastic anemia (AA) is a rare bone marrow failure syndrome. AA is believed to be immune-mediated, supported by in vitro studies and the success of empiric immunosuppressive therapy. Recently, a chromosomal alteration-copy number-neutral loss of heterozygosity of chromosome arm 6p, the site of the Major Histocompatibility Complex and the Human Leukocyte Antigen (HLA) genes-has been identified as a recurrent somatic change in AA. Clonal hematopoiesis marked by 6p CN-LOH is hypothesized to emerge by immune escape of hematopoietic cells lacking certain HLA alleles. However, because of the large size of the genomic region involved by 6p CN-LOH and the strong linkage disequilibrium among other genes in the region, specific alleles targeted by the immune selection in AA are unknown. In a previous study, we reported two patients with somatic loss-of-function mutations in HLA class I genes, leading us to hypothesize that loss of HLA alleles in AA may be common and likely defines a subset of patients with unique characteristics and disease course.

To characterize the prevalence of HLA allele loss in AA, we performed targeted next generation sequencing of HLA-A, B, and C genes, in conjunction with single nucleotide polymorphism array genotyping of bone marrow (BM) or peripheral blood DNA in 74 patients with AA. 52 patients had pediatric-onset AA, and 22 had adult-onset AA. Somatic status of mutations was confirmed by sequencing paired constitutional DNA. Eleven patients (15%) were found to have somatic loss of HLA alleles: 5 patients had 6p CN-LOH, 3 patients had loss-of-function mutations (frameshift, nonsense, or start codon loss) of the HLA class I alleles, and 3 patients were found to have both 6p CN-LOH as well as loss-of-function HLA mutations. HLA loss was more frequent in pediatric-onset AA (9 of 52 patients, 17%) as compared to adults (2 of 22 patients, 9%), although the difference did not reach statistical significance. No HLA mutations were identified in 19 patients with classical Paroxysmal Nocturnal Hemoglobinuria, nor in 20 healthy relatives (p=0.06).

Among the 11 patients with somatic HLA loss, 8 patients had evidence of oligoclonal hematopoiesis with several independent clones carrying different alterations of the same HLA allele. Among the 6 patients with loss-of-of function HLA mutations, the median number of HLA mutations per patient was 1.5 (range 1-3). Of the 8 patients with acquired 6p CN-LOH, the median number of distinct 6p CN-LOH events per patient was 2 (range 1-4). In the 3 patients harboring both 6p CN-LOH as well as the loss-of-function HLA mutations, both mechanisms led to the recurrent loss of the same allele. Strikingly, only a few distinct HLA class I alleles were targeted by mutations. The most frequently affected were HLA-B*40:02:01 (5 independent mutations in 2 patients) and HLA-B*14:02:01 (3 mutations in 2 patients, and as well as loss through polyclonal 6p CN-LOH in 2 patients). Additionally, one patient each had loss of HLA-A*68:01:01 and HLA-A*33:03:01 through mutational inactivation as well as through 6p CN-LOH.

To investigate whether HLA mutations are sufficient to cause clonal expansion or whether other somatic mutations are required, we performed comparative whole exome sequencing (WES) of paired BM and skin DNA in five patients carrying inactivating HLA mutations. Four of the five patients had no other mutations affecting protein sequence or untranslated regulatory regions. One patient had additional somatic mutations, which were subclonal to and co-segregated with the three independent inactivating mutations in the HLA-B*40:02:01 allele. Serial follow-up confirmed that HLA mutations persisted overtime, with a relative expansion of one of the HLA-B*40:02:01 mutant clones bearing a protein-altering mutation in the BCL9 gene.

Our results show that loss of HLA class I alleles is common in AA, second only to PIGA gene mutations. The affected alleles are non-random, with immune selection most commonly targeting HLA-B* 40:02:01 and HLA-B*14:02:01 alleles, providing the first evidence of specificity of immune attack in AA. The resultant hematopoiesis caused by selection of cells with HLA allele loss is typically oligoclonal and commonly occurs in the absence of other somatic mutations. Acquisition of additional mutations can lead to clonal dominance overtime. Further studies are underway to better understand the role of HLA loss in patient outcomes and AA pathogenesis.