Myeloid malignancies such as acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS) are typically sporadic, but germline predisposition involving genes like CEBPA, DDX41, and RUNX1 is increasingly recognized. However, the interplay between germline and somatic events in these genes remains incompletely understood.

To investigate the prevalence, co-variant patterns, and clonality of pathogenic and likely pathogenic variants in CEBPA, DDX41, and RUNX1 in patients with myeloid malignancies.

We analyzed 2,437 patients diagnosed with AML, MDS, MDS/AML, or myeloproliferative neoplasms (MPN). DNA from peripheral blood or bone marrow was sequenced using targeted NGS, covering 63 genes. Variants were processed using validated bioinformatic pipelines and classified per CAP/AMP/ASCO guidelines.

Among the cohort, AMP/ASCO Tier-1 and Tier-2 variants were most frequently detected in RUNX1 (9.97%, n = 243), followed by CEBPA (3.94%, n = 96) and DDX41 (2.63%, n = 64). RUNX1 mutations were commonly associated with co-occurring variants in ASXL1 and SRSF2, particularly in cases with double RUNX1 hits (37.5%). VAF distribution in double-mutated RUNX1 (RUNX1 highest hit: minimum: 10.9, maximum: 47.2, median: 38.5) with variants in ASXL1 (ASXL1 highest hit: min.: 4.3, max.: 49.5, median: 32.6 and/or SRSF2 (min.: 37.5, max.: 49.5, median: 46.1) were frequently consistent suggestive of a clonal process. CEBPA alterations with bi-allelic CEBPA variants (bZIP domain in-frame variant + an N-terminal loss-of-function variant). Co-occurring variants in GATA2 and WT1 were enriched in bi-allelic cases (60%) and were often seen at similar VAF (CEBPA bi-allelic highest hit: min.: 6, max.: 59.2, median: 41.9, GATA2 highest hit: min.: 5, max.: 50, median: 42,6; WT1 highest hit: min.: 5, max.: 94.1, median: 46.8), supporting a shared clonal origin and suggesting a distinct molecular signature potentially driving leukemogenesis. DDX41 variants showed a bimodal VAF distribution in double-mutated cases, with clusters likely suggestive of acquired (min.: 3, max.: 47, median: 6) or inherited (min.: 7, max.: 93, median: 49) variants. Unlike CEBPA and RUNX1, DDX41 variants were not associated with consistent co-variant patterns, suggestive of a different pathobiology. Of note biallelic disruptive DDX41 variants have been associated with hematologic malignancies with unique AML/MDS like features.

Our findings reveal distinct mutational patterns in CEBPA, DDX41, and RUNX1, genes linked to hereditary hematologic cancers. Bi-allelic CEBPA mutations, formed a molecularly coherent subgroup frequently co-mutated with GATA2 and WT1 at similar VAFs, suggesting a shared clonal origin. Similarly, RUNX1 variants often co-occurred with ASXL1 and SRSF2 mutations, also showing consistent VAFs, though RUNX1 mutations were distributed across the gene. In contrast, DDX41 cases lacked co-variant patterns but exhibited a bimodal VAF distribution, with higher VAFs (>35%) suggestive of germline variants and lower VAFs (<20%) likely representing secondary somatic events. This study highlights the distinct biological and clinical profiles of germline and somatic variants in CEBPA, DDX41, and RUNX1, underscoring the need for further research into germline predisposition and its role in the pathogenesis of myeloid neoplasms.