Cell Commun Signal. 2025 Dec 16.
Fan He,
Shuyang Lin,
Bei Gao,
Varun Ramesh,
Alexander B Kim,
Tim Kong,
Daniel A C Fisher,
Christopher T Letson,
Molly Brakhane,
Mary Fulbright,
Yanbo Yu,
Marco Sardiello,
Jorge Di Paola,
Stephen M Sykes,
Stephen T Oh.
INTRODUCTION: Myeloid malignancies, including acute myeloid leukemia (AML) and myeloproliferative neoplasms (MPN), exhibit overlapping pathophysiology. Chronic MPNs can transform into secondary AML (sAML), which is associated with poor prognosis and limited treatment options. However, the process and prognostic significance of leukemic transformation remain incompletely understood.
METHOD: Through a two-sample bidirectional Mendelian randomization (MR) analysis, we showed that genetic liability to MPN significantly predicts the risk of developing AML, establishing MPN as the precursor to leukemia. To identify mediators of this risk, we integrated population-level plasma proteomics data, identifying 55 proteins associated with MPN. Upon integrative analysis with the BEAT-AML cohort, we developed a prognostic proteogenomic gene signature, showing that higher expression of CDCP1, CRISP3, and DXCR, alongside lower MPO levels, correlates with worse AML outcomes. We further performed pharmacogenomic analysis to identify vulnerability to PI3K/AKT/mTOR signaling pathway inhibition in high-risk AML. In vitro and in vivo experiments validated the efficacy of mTOR inhibition in myeloid malignancies.
RESULTS: This gene signature effectively stratified patients by risk, with significant survival differences across the BEAT-AML and TCGA-LAML cohorts, and revealed immune alterations in high-risk groups, including elevated monocyte prevalence and cytokine signaling activity. Single-cell RNA sequencing (scRNA-seq) further suggested enrichment of these genes in progenitor cells and AML blasts. Drug sensitivity predictions suggested that high-risk AML patients may be particularly responsive to PI3K/AKT/mTOR signaling pathway inhibitors. Consistently, we observed upregulation of the genes in cell line models harboring MPN and AML mutations, which was suppressible via dual PI3K/mTOR inhibitor Omipalisib. The efficiency of PI3K/mTOR inhibition in myeloid malignancies was further corroborated by results from multiple in vivo models.
CONCLUSION: Together, our findings revealed shared molecular features across MPN and AML, identified a prognostic gene signature for risk stratification, and provided rationale for PI3K/mTOR inhibition as a promising therapeutic strategy in myeloid malignancies.
Keywords: Myeloid malignancy; PI3K/AKT/mTOR signaling; Pharmacogenomics; Prognostic model; Proteogenomic