Paper Details

ABSTRACT

Background 
Additional somatic mutations beyond the pathognomonic BCR::ABL1 fusion are increasingly recognized in chronic myeloid leukemia (CML). Prospective serial sequencing data in newly diagnosed patients remain limited

Methods
The TCGA-GTA project, initiated in November 2020, is an ongoing prospective, real-world, non-interventional registry of newly diagnosed patients with chronic myeloid leukemia (CML) in the Greater Toronto Area, Canada. Following informed consent, clinical and molecular data are collected longitudinally. A CML-specific single-molecule molecular inversion probe (smMIP) panel (limit of detection ≈0.2%) targeting 40 genes was applied to peripheral-blood or bone-marrow samples obtained at diagnosis, at 6, 12 months, and beyond 12 months of follow-up. Ninety-three patients were sequenced at diagnosis. Longitudinal Mutational Pattern was further analyzed using unsupervised clustering of serial variant allele frequency (VAF) patterns across available time points.

Results
Median age at diagnosis was 48 years (19–84); 64% of patients were male. Sokal risk was low, intermediate, and high in 26%, 50%, and 23%. Disease phase was chronic in 96%. Front-line TKI therapy comprised imatinib (30%), nilotinib (30%), dasatinib (27%), asciminib (10%), or bosutinib (2%). At diagnosis, 24 of 93 patients (26%) harbored at least one mutation (27 mutations in total; three patients carried two concurrent mutations). The most frequent were TET2 (n=15) and ASXL1 (n=5); additional mutations included STAG2 (n=2) and single instances of WT1, RUNX1, CBL, CDKN2A, and PPM1D.

In serial samples, TET2 clones displayed high and relatively stable variant allele frequencies (VAFs; commonly 25–50%) across diagnosis, 6, and 12 months. Achievement of major molecular response (MMR) and MR4.5 occurred irrespective of persistent TET2 status. ASXL1 was present at diagnosis in five patients (VAF 3–53%). One patient treated with dasatinib exhibited a clear declining pattern (53% → 26% → 20%) and achieved MMR and MR4.5. Two additional ASXL1 cases (VAF ≈10% and 15%) receiving bosutinib / asciminib also attained MMR, whereas two low-level clones (VAF ≈3%) on dasatinib/imatinib did not.

Serial sequencing identified emergent mutations (absent at diagnosis but detectable at 6 and/or 12 months) in 42 patients. The most common were ASXL1, KRAS, and SF3B1 (n=6 each), followed by ABL1, PPM1D, and PTPN11 (n=4 each). Emergent ASXL1 clones (n=6; VAF 8–48%) arose on imatinib (n=2), dasatinib (n=2), or nilotinib (n=2). Low-level emergence (VAF ≤18%; n=2) was compatible with subsequent MMR without requiring TKI switch. In contrast, among higher-VAF or rising ASXL1 clones (VAF >18%; n=4), 2/4 did nor achieve MMR and required treatment modification. One patient with emergent ASXL1 on imatinib (peak VAF ≈20%) subsequently developed the T315I ABL1 kinase-domain mutation. Overall, two patients developed T315I: one (high-Sokal, chronic-phase, front-line imatinib) also carried emergent ASXL1; the second (high-Sokal, accelerated-phase, front-line nilotinib).

Unsupervised clustering of 62 serial mutations identified five evolutionary patterns that grouped into three major biological categories: (1) declining treatment-sensitive clones (Cluster 1, 22.6%); (2) emergent or delayed-emergent clones arising during therapy (Clusters 2 and 3, 51.6% combined); and (3) persistent CHIP-like clones (Cluster 5, 17.7%), with a small group of transient mutations (Cluster 4, 8.1%). TET2 mutations were predominantly assigned to persistent and delayed-emergent patterns, consistent with CHIP-like biology, whereas ASXL1 mutations were distributed across all pattern classes. All ABL1 mutations clustered within emergent groups, supporting their association with treatment-driven clonal evolution.

Conclusions
Additional somatic mutations are detectable at diagnosis in approximately one-quarter of newly diagnosed CML patients accrued prospectively, with TET2 and ASXL1 predominating. Longitudinal clustering revealed five distinct patterns of clonal evolution. TET2 mutations predominantly demonstrated stable CHIP-like persistence, whereas ASXL1 mutations exhibited heterogeneous behavior spanning declining, transient, and emergent trajectories. More than half of all mutation patterns belonged to emergent clusters, supporting active clonal evolution during TKI therapy. These findings confirmed the presence of longitudinal mutation dynamics, which will provide clinically relevant prognostic information.