Invariant patterns of global mutational signatures allow new fundamental insights in etiopathogenesis of bone marrow failure including aplastic anemia and myelodysplasia Free

WGS data enables the definition of individual somatic mutational signatures (MS) through mathematical algorithms by identifying distinct global mutational patterns¹. While classically applied to neoplasia, MS can also help explore sources of mutability and the basic pathogenesis of bone marrow failure (BMF), including aplastic anemia (AA) and myelodysplasia (MDS). Most of the cases affected by these conditions are idiopathic, and their origins may be founded in specific DNA damage. Theories to its initiating dysfunction include chemical and viral toxic exposures, immune reactions, as well as genetic deficiencies, e.g., DNA detoxifying or repair pathways.Here we applied MS in AA and MDS as a strategy to bridge gaps in knowledge, support mechanistic hypotheses, and correlate them with clinical features, e.g., severity, progression to myeloid neoplasia, and clinical vs. etiological overlaps. To the best of our knowledge, this bioanalytic strategy has not been applied to study BMF.Data was collected from two sources: our institution (n=54) and the All of Us (AoU) biobank (n=452 AA, 496 MDS, and 516 controls). All VCF files were filtered using a VAF cutoff between 2 and 30%, and population frequency <0.1%, to exclude likely germline mutations. The MS were extracted using SigProfilerExtractor² for AoU and SigProfilerAssignment³ for our samples, using the AoU-derived MS as a priori knowledge. Analyses were performed using Python, R, and Unix-based tools.Two separate aging-MS (SBS1 & SBS5) were present in all the patients regardless of condition, while SBS17b, an unknown etiology (UE) MS, was found in 7.2% of controls and 17% of AA and MDS each. A distinct UE MS (SBS23), and defective POLD1 proofreading MS (SBS10d), were present only in AA and MDS, with proportions of 0.2 & 2.4%, respectively, in AA vs. 0.8 & 1.4% in MDS. Another UE MS (SBS89), and defective homologous recombination DNA repair MS (SBS3), were found only in AA, in 11 & 15.8% of samples. At the same time, SBS8 & 39, both distinct UE MS, were present only in MDS, in 9.7 & 0.4% of cases. The presence of SBS10d in AA and MDS, and SBS3 in AA suggests possible links to the involvement of common and specific DNA repair defects in these diseases.Because of combinatorial complexity of MS in individual patients, we used clustering statistics to identify seven clusters (CL1–7) based on their intra-CL similarity and inter-CL dissimilarity, with regard to the composition of MS comprising each of them. CL1 was the most predominant in patients and controls regardless of age and etiology. CL2 & 3 included AA and MDS overlapping MS (SBS10d & 23), responsible for 1.2 & 0.2% of AA, and 1.4 & 0.8% of MDS. CL4 & 7 were AA-specific (SBS3 & 89), present in 16 & 11% of cases, while CL5 & 6 were MDS-specific (SBS8 & 39), constituting 9.7 & 0.4% of cases. It is important to note that comparisons between MS and CL should account for age as a potential confounding factor, particularly when comparing different disease groups. At the same time, individual CL may themselves correlate with age, suggesting that age could be a key variable shaping the distribution of MS. Additionally, both MS and CL may be associated with various clinical features, potentially offering prognostic insights. Moreover, different CL may enrich for invariant somatic hits, such as mutations and chromosomal aberrations. For instance, TP53 and EZH2 mutants were enriched in CL4, in contrast to DNMT3A mutants, which enriched in CL1 & 7. Interestingly, CL1 patients had significantly larger PNH clone sizes (median clone sizes: 69.9% vs. 6%, p = 0.027) indicating more rapid clonal expansion.In conclusion, MS can be used to define CL that help to subgroup MDS and AA according to their very basic DNA damage profiles. CL indicate unusual genetic background exposure in AA and MDS, which sometimes overlap, as is the case of DNA repair deficiency MS, and show signatures prematurely present in younger patients vs. controls. Currently, ongoing analysis includes larger numbers of patients and clinical and morphologic features, prognosis, and somatic lesions. Results of this analysis will be presented at the ASH meeting.We gratefully acknowledge AoU participants for their contributions, without whom this research would not have been possible. We also thank the NIH AoU Research Program for making available the participant data examined in this study.