Genomic and transcriptomic features associated with responses and outcomes of third-generation tyrosine kinase-inhibitor therapy in chronic myeloid leukemia

Third-generation tyrosine-kinase inhibitors (3G-TKIs), including olverembatinib and ponatinib, are effective in people with chronic myeloid leukemia (CML) failing prior TKIs, but many subjects still have poor responses and outcomes. The broader genomic and transcriptomic features associated with 3G-TKI prognosis remain incompletely defined. We interrogated genomic and transcriptomic data from 202 subjects receiving olverembatinib or ponatinib at baseline and during therapy. In penalized multi-variable analyses, the high-burden ASXL1 p.G646Wfs*12 ( ASXL1 G646Wfs*12 ) variant emerged as the most consistent genomic feature associated with inferior responses and outcomes; additional lower-frequency variants, including RUNX1 , IKZF1 , PHF6 , SETBP1 and GATA2 variants, were also associated with responses and/or outcomes. Longitudinal genomic analyses showed that ABL1 variants frequently decreased during 3G-TKI therapy, whereas persistent, emerging or expanding non- ABL1 variant clones were enriched in resistant subjects, suggesting a shift toward BCR::ABL1 -independent clonal evolution in later-line resistance. Transcriptomic analyses identified three distinct gene expression profiles (GEPs) based clusters with different prognoses. Favorable prognosis was associated with lower leukemic stemness and more immune-active transcriptional features, whereas poor prognosis was characterized by stemness- and proliferation-dominant, immune-suppressed states. These findings identify genomic and transcriptomic features associated with 3G-TKI prognosis and provide a potential framework for molecular risk stratification in later-line CML that requires independent validation.