A comparative analysis of epigenetic age acceleration across distinct hematopoietic clones in hematologic diseases

Abstract Studiesbased on the Horvath methylation clock have demonstrated that biological age, estimated based on methylation levels at specific CpG sites strongly correlates with chronological age in healthy individuals1). The discrepancy between biological and chronological age is referred to as epigenetic age acceleration differential (AAD). Other analysis showed that AAD is indeed associated with shorter overall survival and certain exposures known to increase AAD2). Biological age, if measured in a polyclonal background, may indicate time-independent predisposition for the acquisition of clonal mutations. However, clonal vs polyclonal cells may theoretically show either concordant or discordant AAD, and this may also be gene type dependent. Given the heterogeneity of clonal burden, bulk analyses may reflect either AAD of leukemic vs normal cells, and consequently, in fully clonal disorders, AAD measured would be that of the leukemic clone rather than the host. We here investigate how AAD might differ between clones within the same individual or how it diverges from biological age in leukemia. To address this, in addition to conventional methylation arrays, we used nanopore sequencing to evaluate AAD. While our analysis of sorted samples is still ongoing, preliminary results are presented here. In patients with myeloid neoplasms (MN), separation of non-clonal vs clonal cells is not possible, and thus bulk DNA results reflect aging of the malignant clone, in particular if clonal burden is high. This limitation of our assay notwithstanding, we analyzed AAD (Horvath Clock minus chronological age in this study) in MN (n=253), aplastic anemia (AA; n=10), and healthy controls (n=322). The breakdown of MN included 191 cases of AML, 26 of MDS/CHIP, 31 of MDS/MPN, 5 of MPN. The median age was 54 years (range: 34–70) in the healthy individuals and 63 years (range: 7–85) in the MN cohort. First, we assessed the correlation between chronological and biological age. In healthy individuals using Horvath Clock (based on 353 CpG sites), the expected correlation between biological and chronologic age was reestablished (n = 322; Horvath clock: P<0.001, Pearson r = 0.81, R2=0.65). The average age acceleration was –1.22 (95% CI: –1.7-0.7). In general, among patients with hematologic diseases, the correlation between the Horvath clock and chronological age was weaker (n = 191; P<0.001, Pearson r = 0.19, R2=0.04). Next, we evaluated the AAD. In healthy individuals, AAD was low (mean: -1.22), and this parameter decreased with age, as shown by the comparison between those >50 and ≤50 years old (P<0.001). Similarly, in both MN and AA, AAD tended to decrease with increasing age. Patients with MN and AA show a higher degree of AAD compared to controls with a higher variance in patients [8.9(-55.8-125.2) in MN, 36.4 (-4.9-71.4) in AA, -1.2(-13.5-16.1) in controls; P<0.001], likely related to their somatic molecular diversity. To further investigate the factors contributing to AAD, we classified samples into a high AAD (AAD≥10 years) and a low AAD (AAD<10 years) and conducted a comparative analysis. In MN, patients with high AAD were significantly younger (P=0.001). Preliminary data indicate that high AAD may depend on somatic genetic features. For instance, IDH1 mutant cases appear to have low AAD but higher numbers of patients have to be collected for comparative analyses. In contrast, we did not find any correlation with the blast count. When we studied AA as a non-malignant disease, the presence of a PNH clone was associated with a trend toward higher AAD (P=0.057). When comparing the relationship between PNH clone size and AAD, no significant correlation was observed, and even patients with small PNH clones exhibited a high AAD. Furthermore, bone marrow transplantation was significantly associated with increased AAD in all patients (P=0.042). Regardless of clone size, AAD was observed across hematologic diseases. In AA, the presence of PNH clones appeared to be associated with increased AAD. As PNH clones are capable of immune evasion and persist under chronic cellular stress, these features may contribute to AAD. For the ASH meeting, we plan to include additional data comparing biological age across fractionated cell populations, as well as differences between nanopore sequencing and conventional methylation arrays. 1 2