Machine learning is effective for differentiating chronic myeloid leukemia from polycythemia vera, primary myelofibrosis, and essential thrombocythemia.
Researchers from the Munich Leukemia Laboratory in Germany have developed a model that uses 12 genetic markers to accurately stratify patients with myeloproliferative neoplasms (MPNs), according to a study published online ahead of print in Leukemia.
The WHO categorizes classical MPNs—using cytomorphology, bone marrow biopsy, grading of fibrosis, blood counts, and a handful of molecular markers—into four individual entities: chronic myeloid leukemia (CML) and the BCR::ABL1 negative MPNs polycythemia vera (PV), primary myelofibrosis (PMF), and essential thrombocythemia (ET).
“However, overlaps, borderline findings, or transitions between MPN subtypes occur, and incomplete clinical data often complicates diagnosis,” Manja Meggendorfer, PhD, and the study coauthors wrote.
The researchers analyzed 355 patients with MPN to use the results to stratify MPN entities and provide prognostic information. The investigation revealed the presence of genetically distinct subgroups with different cytogenetic abnormalities, mutations, and JAK2 allele statuses.
“Notably, differences in JAK2 allele status (heterozygous/homozygous) correlated with diverse EFS [event-free survival] and OS [overall survival] outcomes, potentially due to additional prognostic mutations,” the researchers reported. “In contrast, groups with cytogenetic aberrations and additional mutations generally had shorter EFS and poorer OS regardless of the diagnosed entity, aligning with studies on the impact of karyotype and mutation count on survival.”
Determining Risk for Transformation
A machine-learning model that relies on 12 genetic markers for differentiating patients with CML, PV, PMF, and ET was developed using data from the analysis. The final molecular markers were mutation status of ASXL1, BCR::ABL1, CALR, JAK2, SF3B1, SRSF2, TET2, TP53, and U2AF1 and the binary variant allele frequency values CALR greater than 35%, JAK2 greater than 35%, and JAK2 greater than 60%. The researchers translated the findings into a user-friendly decision tree for routine classifying patients with MPN.
A comparison of samples at chronic and blast phase revealed that mutations in SRSF2, TET2, and RUNX1 mark the transition to the blast phase. TP53 mutations, too, are prevalent at the shift.
“Our model, incorporating these markers, can determine patients’ risk for transformation, highlighting that [patients with ET], typically considered low risk, may be high-risk genetically,” Dr. Meggendorfer and colleagues wrote. “Consequently, expanding genetic analysis beyond JAK2, CALR, and MPL at diagnosis is crucial for accurate MPN classification, early high-risk patient identification, and timely intervention.”
