AUTHORS
Felipe de Almeida Sartori, Arda Durmaz, PhD2, Atsushi Marumo, MD3, Carlos Bravo-Perez, MD4, Luca Guarnera, MD5, Zachary Brady, BS6, Mark David Orland, MD, PhD7, Serhan Unlu, MD8, Aashray Mandala, BS9, Kartik Lakhotiya, MS10, Charlotte Zurowski11, Carmelo Gurnari, MD,PhD12, Valeria Visconte, PhD13, Jaroslaw P. Maciejewski, MD, PhD, FACP14, Paulo Vidal Campregher, MD, PhD15, Thomas Laframboise, PhD16
AFFILIATIONS
1 Case Western Reserve University, Cleveland, Ohio, United States, Cleveland Clinic, Cleveland, Ohio, United States, Hospital Israelita Albert Einstein, Sao Paulo, SP, Brazil
2 Cleveland Clinic, Cleveland, Ohio, United States
3 Cleveland Clinic, Cleveland, Ohio, United States, Nippon Medical School, Tokyo, Japan
4 Cleveland Clinic, Cleveland, Ohio, United States, Hospital Universitario Morales Meseguer, University of Murcia, IMIB-Pascual Parrilla, CIBERER - Instituto de Salud Carlos III, Murcia, Spain
5 Cleveland Clinic, Cleveland, Ohio, United States, University of Rome Tor Vergata, Rome, Italy
6 Cleveland Clinic, Cleveland, Ohio, United States
7 Case Western Reserve University, Cleveland, Ohio, United States, Cleveland Clinic, Cleveland, Ohio, United States
8 Cleveland Clinic, Cleveland, Ohio, United States
9 Case Western Reserve University, Cleveland, Ohio, United States, Cleveland Clinic, Cleveland, Ohio, United States
10 Case Western Reserve University, Cleveland, Ohio, United States, Cleveland Clinic, Cleveland, Ohio, United States
11 Cleveland Clinic, Cleveland, Ohio, United States
12 Cleveland Clinic, Cleveland, Ohio, United States, University of Rome Tor Vergata, Rome, Italy
13 Cleveland Clinic, Cleveland, Ohio, United States
14 Cleveland Clinic, Cleveland, Ohio, United States
15 Genesis Genomics, Sao Paulo, SP, Brazil, Hospital Israelita Albert Einstein, Sao Paulo, SP, Brazil
16 Case Western Reserve University, Cleveland, Ohio, United States
ABSTRACT
Mutational signature (MS) decomposition from WGS identifies processes underlying somatic mutagenesis from patterns of single base substitutions (SBS). Because multiple MS can coexist within an individual, their clustering may yield pathogenetic and epidemiologic insights. MS composition may reflect exogenous exposures, aging, and inherited DNA repair defects, but in myeloid neoplasia (MN) it may also be shaped by clone-intrinsic factors, including somatic mutations and proliferative rate. Thus, bulk DNA signatures may capture different contributions of host versus clonal mutagenic processes.
We applied MS analysis across clonal hematopoiesis of indeterminate potential (CHIP), aplastic anemia (AA), myelodysplastic neoplasms (MDS), and acute myeloid leukemia (AML) to define shared and disease-specific mutational processes and correlate them with clinical and genomic features. We analyzed data from the NIH All of Us (AoU) biobank (400 CHIP, 378 AA, 446 MDS, 352 AML) and compared the MS with 516 controls. Likely somatic variants were filtered from available VCFs and used for signature extraction with SigProfilerExtractor. Participants were clustered by Ward’s minimum-variance method on the Jensen-Shannon distance matrix of signature proportions. Germline and somatic variants, age, sex, blood counts, and mutation burden were compared across clusters (CL) with Benjamini-Hochberg adjustment.
Clock-like SBS MS SBS1 and SBS5 were detected in everybody. SBS17b, associated with reactive oxygen species (ROS) damage, was identified in controls, AA, MDS, and AML but not in CHIP. SBS23 was observed in CHIP and AA; SBS96 in AA and MDS; SBS89 in AA, MDS, and AML; and SBS19 only in MDS. DNA repair deficiency-associated signatures SBS3 and SBS10d were detected in AA and MDS, with SBS10d also present in AML.
Unsupervised clustering identified 7 MS-based CL, all of which contained SBS1 and SBS5. CL1, the largest, was present in all cohorts and consisted predominantly of SBS1/SBS5, with occasional contributions from SBS19, SBS23, and SBS96. CL2 and CL3 showed SBS17b; CL2 was absent in CHIP and had lower SBS17b contribution, whereas CL3 was restricted to AA, MDS, and AML and had higher SBS17b proportions. CL4 showed SBS3, SBS10d, and SBS17b and occurred in AA and MDS. CL5 shared this pattern with additional SBS96 and was also restricted to AA and MDS. CL6 showed SBS10d, SBS17b, SBS89, and SBS23 and was present in AA, MDS, and AML. CL7 showed SBS10d and SBS17b and was present in AA, MDS, and AML.
Across AA, MDS, and AML, CL1 had higher proportion of female participants than the other CL (74% vs 40%, 67% vs 28%, and 72% vs 34%, respectively; all p<0.0001), whereas CL6 had higher proportion of males (92% vs 35%, 93% vs 44%, and 92% vs 30%; all p<0.0001). CL7 contained only male participants (p=0.03 for AA and 0.001 for AML). CL1 patients were younger in AA and MDS (62 vs 65 years, p=0.04; 68 vs 71 years, p=0.04), while CL6 were older in MDS (75 vs 68 years, p=0.003).
For the disease groups, CL1 and CL3 showed higher somatic burden (e.g., CL3: 3158 vs 1675 in AA, p=8x10-6; 2282 vs 1593 in MDS, p=0.0002; 2149 vs 1665 in AML, p=0.003), whereas CL6 in AA (1635 vs 1691, p=0.04) and CL7 in AML (1254 vs 1689, p=0.003) showed lower burden.
Variant distributions also differed across CL. Germline mutations in the mitotic checkpoint BUB1B gene in AA were enriched in CL2, while SUZ12 mutations in AML were enriched in CL2 (OR 10, p=0.03) and depleted in CL1 (OR 0.09, p=0.03). In addition, the presence of any germline mutation differed across MDS clusters (p=0.05). Somatically, in AA, CEBPA mutations were enriched in CL3 (OR 50, p=0.002) and CL6 (OR 3.3, p=0.04) and depleted in CL2. In MDS, CUX1 was enriched in CL3 (OR 20, p=0.003) and PRPF8 in CL6 (OR 20, p=0.05). In AML, STAG2 mutations were enriched in CL1 (OR 5, p=0.02) and depleted in CL6, while CACNA1B was depleted in CL2 (OR 0.24, p=0.04) and enriched in CL6 (OR 2, p=0.04). AML patients in CL1 also had higher odds of cytopenia after adjustment for age and sex (OR 2.4, p=0.0006).
Our findings indicate that progression across CHIP, AA, MDS, and AML is associated with increasing diversification beyond clock-like mutagenesis, including ROS-related and DNA repair deficiency-associated processes. The resulting MS-defined clusters correlate with sex, age, mutation burden, cytopenia, and selected germline/somatic variants, supporting biologically distinct trajectories of myeloid disease evolution.
We thank AoU.
DISCLOSURES
Felipe de Almeida Sartori: No relevant conflicts of interest to declare.