Exploring the Unseen Effects of Epigenetic Age Acceleration in Low-Risk MDS: Associations with Unique Cytokine Profiles

Introduction Ageing is a major risk factor for the development of Myelodysplastic Syndromes/Neoplasms (MDS). Epigenetic alterations are common in MDS, especially aberrant DNA methylation. On a different level, DNA methylation at specific genomic loci can be used to capture the epigenetic age of cells or tissues. Epigenetic age acceleration (EAA) can thus be evaluated in individuals comparing biological age, obtained by specific tools called epigenetic clocks, with their chronological age. We assessed EAA in lower-risk MDS (LR-MDS) patients in parallel with serum cytokine levels and clinical/molecular variables, in the attempt to correlate biological ageing with inflammation and disease progression. Methods We evaluated 101 MDS cases diagnosed according to WHO 2022, selected per lower risk IPSS-R score. IPSS-M score was also calculated. Complete clinical annotations were collected. DNA was extracted from bone marrow mononuclear cells and assessed for DNA methylation using the Infinium MethylationEPIC v2.0 array. Data preprocessing was conducted in R using the Bioconductor package SeSAMe. Clock estimates of well-established epigenetic clocks (Horvath1, Hannum, DNAmPhenoAge, GrimAgeV1, and GrimAgeV2) and their principal-component counterparts were calculated using online tools. Unsupervised hierarchical clustering with Ward D2 algorithm was used to identify clusters of differential EAA. These clusters were compared across clinical variables, cytokine levels, and mutational landscape. Results Median age at diagnosis was 75 years (yrs) (36-91) (M:F ratio 2:4). Per WHO classification: 67 patients were MDS-with low blasts (LB); 13 MDS del(5q); 11 MDS SF3B1mut; 1 MDS-hypoplastic; 6 MDS-increased blasts-1; 1 MDS-increased blasts-2; 2 MDS-RS. IPSS-R scores were: very low = 38; low = 50; intermediate = 13. IPSS-M scores: very low = 24; low = 42; moderate low = 12; moderate high = 3; high = 5; not available = 15). Patients exhibited increased EAA in several epigenetic clocks: Horvath1 (mean 6.54 yrs), Hannum (2.71 yrs), GrimAgeV1 (2.39 yrs), and GrimAgeV2 (2.70 yrs), but not in DNAmPhenoAge (-0.60 yrs). Principal component clock counterparts also revealed increased EAA: PCHorvath1 (7.40 yrs), PCHannum (6.90 yrs), PCPhenoAge (7.65 yrs), and PCGrimAge (2.57 yrs). DNAmPhenoAge EAA positively correlated with IPSS-M (Spearman r=0.28, p = 0.013; Very low = -7.09 ± 13.55 yrs; low = -0.45 ± 11.68 yrs; moderate low = 4.16 ± 12.94 yrs). None of the epigenetic clocks correlated with IPSS-R scoring. Through unsupervised clustering analysis, we identified three distinct clusters based on EAA: Cluster A, that we defined as slow-agers, exhibited the lowest EAA across all evaluated clocks (n = 38; M:F ratio 1.7); cluster B displayed heterogeneous EAA (n = 42; M:F ratio 3.2); and Cluster C, termed fast-agers, showed the highest EAA across all clocks (n = 21; M:F ratio 2.5). Fast-agers show significantly lower serum CXCL10 (20.75 vs. 38.08 ng/mL) levels (p = 0.041, Wilcoxon rank sum) and a trend towards lower levels of senescence-associated secretory phenotype (SASP) factors compared to slow-agers, including CCL3 (51.54 vs. 69.48 ng/mL), CCL4 (108.00 vs. 219.75 ng/mL), IL6 (2.41 vs. 3.07 ng/mL), IL8 (12.92 vs. 34.89 ng/mL), TNF-α (4.08 vs. 6.32 ng/mL), and VEGF (51.94 vs. 62.89 ng/mL). By clinical variables (e.g., Hb, WBC, platelets, MCV) we could not define significant differences comparing slow - and fast-agers (cluster A and C). MDS SF3B1mut were predominantly fast-agers, while MDS with del(5q) and MDS-LB were more commonly slow-agers. There was no difference in progression to HR-MDS. With the limitation of few cases observed, fast-agers seemed to have a significant shorter time to progress to acute myeloid leukemia [slow-agers (n = 3) = 803.00 ± 43.41 days; fast-agers (n = 2) = 137.50 ± 53.03 days]. Conclusion In our analysis of 101 LR-MDS patients, increased epigenetic age acceleration (EAA) was observed in almost all epigenetic clocks evaluated. IPSS-M scoring correlated with EAA, confirming the relevance of somatic mutations. Cluster analysis revealed 3 EAA distinct groups. Patients with the highest EAA, that we termed fast-agers, had lower levels of pro-inflammatory cytokines compared to slow-agers. Overall, EAA effectively identifies clusters of patients with distinct cytokine profiles, stressing the modulating role of inflammation (and genetic alterations) in the pathophysiology of LR-MDS.