A Proteomics-Driven Machine Learning Tool for Distinguishing ET from pre-PMF
Uncover the hidden differences between essential thrombocythemia and pre-myelofibrosis with a cutting-edge proteomic tool that outperforms clinical predictors, paving the way for more accurate early-stage MPN diagnosis.
This study leveraged data-independent acquisition (DIA)-based proteomic profiling of bone marrow samples from 434 essential thrombocythemia (ET) and 91 pre-myelofibrosis (pre-PMF) patients. A 9-protein classifier was developed using machine learning, achieving superior diagnostic performance (AUC=0.895) compared to clinical variables alone (AUC=0.499). The proteomic model was particularly robust in distinguishing JAK2V617F+ and CALR+ subtypes, offering a reproducible, molecular-based approach for early-phase MPN diagnosis. This innovative tool has the potential to transform the clinical management of these phenotypically similar yet biologically distinct myeloproliferative neoplasms.