Intro/Aims:
Despite the success of tyrosine kinase inhibitors (TKIs) in chronic myeloid leukaemia (CML), ~20% of patients develop resistance. Mechanisms include BCR::ABL1 kinase domain mutations, BCR::ABL1 overexpression and epigenetic or transcriptional alterations (e.g. RUNX1, ASXL1, TET2 mutations).
Asciminib, an allosteric inhibitor targeting the ABL1 myristoyl pocket, is active against many BCR::ABL1 mutations, including T315I, and demonstrates efficacy after TKI failure. However, patients with high-risk genetic features remain prone to resistance. Combination approaches, such as asciminib plus dasatinib, are under investigation, including in the ASCENDANCE (CML14, #ACTRN1262300133865) trial.
RUNX1 mutations are frequent in blast-phase CML and associate with poor prognosis, yet their role in resistance to asciminib-based combinations is unknown. This study models dual resistance to asciminib and dasatinib and evaluates the contribution of RUNX1 using patient-derived and functional models.
Results and Methods:
Resistant K562 intermediates were generated through exposure to escalating concentrations of asciminib (ASC) and dasatinib (DAS) in combination at a ratio of 35 to 1. Drug transporter expression was measured by flow cytometry. Early intermediates (3.5 nM ASC + 0.1 nM DAS; 35 nM + 1 nM) did not upregulate ABCG2 or ABCB1. In contrast, late intermediates (350 nM + 10 nM; 525 nM + 15 nM) demonstrated significantly increased ABCG2 expression (p<0.0001, n=3), while ABCB1 remained low. No BCR::ABL1 mutations were detected.
In a retrospective cohort of 575 CML patients, RUNX1 mutations were identified in 31 patients (6.1%) and followed three patterns: (1) present at diagnosis (median variant allele frequency, VAF ~10%) in patients with optimal TKI response; (2) acquired during TKI therapy in myeloid blast phase (n=6; median VAF ~39%), frequently co-occurring with ASXL1, TET2, or SETD1B mutations; and (3) present in lymphoid blast phase (n=6; median VAF 46%), often co-occurring with BCR::ABL1 mutations.
To assess function, CRISPR-Cas9 knockout of RUNX1 was performed in K562 cells and patient-derived variants (p.Arg166Leu, p.Arg107Leu) reintroduced to test TKI sensitivity. Functional work is ongoing.
Conclusions
This stepwise K562 model of dual ASC/DAS resistance identified ABCG2 overexpression as a mediator of acquired resistance. RUNX1 mutations were characterised in patients, and functional studies are ongoing. These models provide a platform to define mutation-informed strategies to overcome resistance in high-risk CML.