Peripheral blood profiles reflecting progenitor lineage balance predict treatment response in chronic myeloid leukemia

Early achievement of deep remission improves patients outcome in chronic myeloid leukemia (CML) treatment, highlighting the need for predictive indicators before therapy initiation. This study aimed to develop a tool to predict CML treatment responses to guide optimal therapy selection. Using hierarchical clustering of complete blood count (CBC) data at diagnosis, patients were stratified into two clusters. Patients in Cluster 1 had higher BCR::ABL1IS mRNA levels at 3 and 6 months post-treatment and lower rates of major molecular response compared to cluster 2. Cluster 1 also showed increased granulocyte and immature white blood cell counts and decreased erythroid parameters. Flow cytometric analysis of bone marrow mononuclear cells revealed that cluster 1 had a significant increase in hematopoietic stem cell fractions and a higher ratio of granulocyte-macrophage progenitors to megakaryocyte-erythroid progenitors compared to cluster 2. These findings suggest that differences in bone marrow progenitor cell differentiation affect peripheral blood profiles. Artificial intelligence-driven ghost cytometry (GC) was evaluated for its ability to comprehensively capture these changes and successfully distinguished patients with poorer treatment responses, with GC scores at diagnosis strongly correlating with BCR::ABL1IS mRNA levels at 3 and 6 months post-treatment initiation. The study indicates that multivariate analysis of CBC or GC analysis may enable simple, early prediction of CML treatment efficacy, potentially contributing to effective and individualized CML therapy.