HLA evolutionary divergence (HED) is a continuous metric that quantifies the breadth of the immunopeptidome that can be presented by an individual's HLA allotype. High HED scores have been shown to be associated with better outcome after immune check point inhibitor therapy in solid tumors.1 However, the role of HED has never been assessed in autoimmune disorders, and may have an impact on disease susceptibility and outcomes. Here we explored the impact of HED on clinical features of aplastic anemia (AA) and paroxysmal nocturnal hemoglobinuria (PNH). The immunological landscape of those diseases is characterized by an HLA-restricted CD8+ and CD4+ mediated autoimmunity which leads to hematopoietic stem cell destruction, with a crucial role played by immune-dominant PNH clones and immune escape mechanisms likely associated with the down-regulation of the HLA-antigen presentation. Since higher HED scores reflect a wider antigenic spectrum and potentially more diverse peptide presentation capabilities, we hypothesize that this context not only may elicit T-cellular responses by auto-antigen presentation but also that HLA structural diversity can impact on disease phenotype and have a role within the definition of an immune-dominant (higher HED) or immune-escape (lower HED) environment.

Using an NGS-based platform targeting the entire HLA locus, we have sequenced DNA samples of a cohort of 98 well-annotated AA, PNH, or AA/PNH patients. We then calculated HED at the classical HLA class I and II loci (A, B, C, DRB1, DQA1, DQB1, DPB1, DPA1), and investigated its impact on disease course and outcomes (Fig. A, B).

Homozygosity for at least 1 out of 8 loci was frequently found (70%), especially for class II, however only 2% of patients had >1 homozygous locus). HLA-B and HLA-DQA1 genotypes demonstrated the highest HED among HLA-I and II loci, respectively (Fig. C). While for class I loci no difference was detected across diagnostic subgroups, mean HED for class II loci was significantly lower for PNH patients as compared to AA (6.031 vs 7.623 respectively p=0.047). Analysis of HED metrics across several binary outcomes within logistic regression models demonstrated that higher HED at class I and class II loci was generally associated with unfavorable disease characteristics and response outcomes. Specifically, high mean HED in class I and particularly in B and C loci tracked with disease severity (severe/very severe AA according to modified Camitta criteria: OR= 1.36 [95%CI 1.12, 1.70], p=0.003),2,3 whereas higher mean HED in class II was associated with refractoriness to immunosuppression [OR=1.18 (95%CI: 1.01-1.42), p=0.041]. In particular, a strong relationship was found for DQB1 locus [p=0.032, OR=1.12 (95%CI: 1.01-1.25)] in severe AA patients. Moreover, high HED in DPB1 was associated with early disease presentation [OR= 1.32 (95%CI 1.02, 1.80), p=0.047]. No correlation was found with overall survival.

In summary, our results indicate that higher HED scores correlate with severity, early disease onset and refractoriness to immune-suppression in AA patients. High HED may increase the probability of presenting self-peptides that elicit dominant autoimmune T cell responses. Consistent with this line of reasoning, it is not surprising that lower HED (for class II HLA) is found in PNH subjects and that disruptive HLA somatic mutations in AA are preferably found in loci with increased structural diversity, consistent with the idea that less diversity may produce more immune-escape (see Abstract #: 142501). Our results suggest that HED may be useful for risk stratification of patients with immune-mediated bone marrow failure disorders, and more generally, that the ability to present a wider antigenic peptide spectrum may produce a more pervasive autoimmune response.


Patel:Alexion: Other: educational speaker. Saunthararajah:EpiDestiny: Consultancy, Current equity holder in private company, Patents & Royalties: University of Illinois at Chicago.

Author notes


Asterisk with author names denotes non-ASH members.