Exploring co-mutations in CML and ph(+) b-ALL with ABL1 variants: Correlations with survival and treatment response. Free

The impact of co-mutations seen with ABL1 kinase domain mutations is unclear in today's treatment landscape (Branford 2024). We conducted a retrospective study to explore the relationship between ABL1 variants and other co-mutations.

This study was exempt from IRB review (Protocol #808807). We utilized cBioPortal to identify patients with CML, Ph(+) B-ALL, and other acute leukemias who harbored ABL1 variants by NGS at our institution. We collected demographic and disease characteristics, associated co-mutations, treatment details, and outcomes data. We determined ABL1 variant and associated co-mutation frequencies. For the two most common ABL1 variants, we calculated Fisher's exact test, applying false discovery rate and continuity corrections to analyze co-mutation patterns. Cox proportional hazards model and Kaplan-Meier (KM) plots were created to assess associations between overall survival (OS) and ABL1 variants classified as E255K, T315I, and “other,” co-mutations classified as present or absent, treatment with Ponatinib or Asciminib, allogeneic hematopoietic cell transplant (HCT), or chimeric antigen receptor T-cell (CAR-T) therapy. Analysis was completed with R V4.2.

21 patients harbored NGS-identified ABL1 variants between 2015-25. At diagnosis, 12 (57%) had CML, 2 (10%) CML-Blast Phase (BP), and 7 (33%) B-ALL. 17 (81%) had BCR-ABL p210 breakpoint, while 4 (19%) had p190. The median age at diagnosis was 55 (SD 16.1, range 25-74), 11 (52%) were female, 14 (67%) were non-white, and 11 (52%) were Hispanic. 11 (52%) developed BP or differentiation to acute leukemia (myeloid 1, lymphoid 4, unk/mixed 6). On average patients received 4.25 lines of therapy, 11 (52%) had non-compliance documented and 10 (48%) had a history of medication adverse effects requiring treatment pause/discontinuation. All patients received frontline therapy that included a TKI and 18 (86%) received some form of chemotherapy or immunotherapy prior to ABL1 variant detection. After variant detection 14 (67%) received chemo/immunotherapy +/- TKI, 14 (67%) received Ponatinib, 9 (43%) received Asciminib, 3 (14%) underwent HCT, and 1 (0.05%) received CAR-T. Median (m)OS from diagnosis was 53 mo (SD 64.23), from BP 24 mo (SD 58.43), and from ABL1 variant detection 11.5 mo (21.11).

T315I was detected in 12 (57%) patients, E255K in 3 (14%), and additional variants included F359V (2), V299L (1), F317I (1), L248V (1), and Y253H (1). 3 patients, 2 with E255K, were found to develop a subsequent secondary ABL1 variant, T315I. Cox proportional hazards model found statistically significant lower OS associated with E225K compared to T315I (HR 0.09; p = 0.044*) and “other” ABL1 variants (HR 0.09; p = 0.049*). While not statistically significant (p = 0.33) a similar trend was shown with KM curves where E255K mOS was 32 mo (CI 18-NR) vs T315I mOS 53 mo (CI 24-NR) vs “other” mOS 135 mo (CI 90-NR). Non-ABL1 co-mutations were not associated with survival independently (p = 0.88) or by multivariant analysis (p = 0.285). Patients treated with Ponatinib or Asciminib had significantly better survival than those who did not receive these therapies (mOS 90 mo; CI 32-NR vs 38.5 mo; CI 12-NR; p = 0.085* and by multivariant analysis HR 0.22; p = 0.028*). HCT and CAR-T were rare events and not significant.

Co-mutations associated with T315I included KMT2A, KMT2C, NOTCH2, GNAS, WT1, CALR and ASXL2, and were associated with a strong effect size (OR = 27), however results were not significant (p = 0.2). Similarly, analysis of E255K found co-mutations NRAS, SUZ12, CUX1, NOTCH1, TP53, IKZF2, LUC7L2, PAX5, and PRPF8 were associated with a strong effect size (OR =27), also not significant (p = 0.2).

T315I followed by E255K were the most frequently identified ABL1 variants observed in our dataset and were associated with worse outcomes regardless of co-mutation status. Co-mutations occurred in the IKZF family, epigenetic modifiers, and lymphoid progenitors (Calabretta 2004; Mangum 2021). This study was limited by small sample size, thus potentially missing important associations. Our findings are exploratory, suggesting ABL1 variant status is dynamic, requires serial monitoring, and remains a dominant prognostic factor irrespective of co-mutations in today's treatment landscape, where treatment with Ponatinib or Asciminib in patients with ABL1 variants is favorable though unfortunately still does not offer extended disease control.