Tyrosine kinase inhibitor response of ABL-class acute lymphoblastic leukemia: the role of kinase type and SH3 domain

Abstract

Acute lymphoblastic leukemia (ALL) with fusions of ABL-class tyrosine kinase genes other than BCR::ABL1 occurs in ∼3% of children with ALL. The tyrosine kinase genes involved in this BCR::ABL1-like (Ph-like) subtype include ABL1, PDGFRB, ABL2, and CSF1R, each of which has up to 10 described partner genes. ABL-class ALL resembles BCR::ABL1-positive ALL with a similar gene expression profile, poor response to chemotherapy, and sensitivity to tyrosine kinase inhibitors (TKIs). There is a lack of comprehensive data regarding TKI sensitivity in the heterogeneous group of ABL-class ALL. We observed variability in TKI sensitivity within and among each ABL-class tyrosine kinase gene subgroup. We showed that ALL samples with fusions for any of the 4 tyrosine kinase genes were relatively sensitive to imatinib. In contrast, the PDGFRB-fused ALL samples were less sensitive to dasatinib and bosutinib. Variation in ex vivo TKI response within the subset of samples with the same ABL-class tyrosine kinase gene was not associated with the ALL immunophenotype, 5' fusion partner, presence or absence of Src-homology-2/3 domains, or deletions of IKZF1, PAX5, or CDKN2A/B. In conclusion, the tyrosine kinase gene involved in ABL-class ALL is the main determinant of TKI sensitivity and relevant for specific TKI selection.

Graphical abstract

Figure 1.

Ex vivo TKI sensitivity of ABL-class ALL samples. (A-C) Z-score of the area under the dose-response curve (AUC) for imatinib, dasatinib, and bosutinib sensitivity, respectively, of primary and primary xenografted ALL samples with ABL-class fusions compared with primary and primary xenograft ALL samples of pediatric BCR::ABL1-positive ALL. (D-E) Correlation between sensitivity to imatinib (ima) and dasatinib (dasa) or bosutinib (bosu). The Spearman correlation coefficient (Rho) is calculated for the reference cohort of BCR::ABL1-positive ALL samples. The x = y line is displayed in gray. (F-G) Correlation between imatinib and dasatinib or bosutinib in ABL1- or PDGFRB-fused ALL samples, respectively. The Spearman correlation coefficient (Rho) and P value are calculated and are displayed in the tables below.

Figure 2.

Ex vivo TKI sensitivity per ABL-class tyrosine kinase gene colored by the 5′ fusion partner. (A) Dose-response curves for imatinib treatment of ZMIZ1::ABL1- (dark blue), NUP214::ABL1- (medium blue), or RCSD1::ABL1-fused (light blue) primary or primary xenografted ALL samples. (B) Dose-response curves for imatinib in primary or primary xenografted ALL samples with EBF1::PDGFRB fusion (pink) or other PDGFRB fusions (purple), JAKMIP2, or CCDC88C). Leukemic cell survival relative to untreated controls, mean ± standard error of the mean (SEM) of 1 to 3 biological replicates.

Figure 3.

Heat map of patient characteristics and secondary lesions. The primary, primary xenografted, and serial xenografted ALL samples are displayed by the ABL-class tyrosine kinase gene from the most sensitive (left) to the least sensitive (right) based on the area under the dose-response curve (AUC) for imatinib measured using an ex vivo imatinib sensitivity assay. Deletions and other aberrations in leukemia-associated genes, PAX5, IKZF1, CDKNA2A/B, and pseudoautosomal (PAR) regions, based on MLPA assays or whole-exome sequencing are shown. IKZF1 deletions are subdivided into deletions resulting in haploinsufficiency (in dark green, affecting the start codon in exon 2) and deletions that affect the DNA-binding domain, exerting a dominant-negative effect on the unaffected allele (in dark red, deletions of exons 4-7). A white square in the middle denotes no mutations but whole-exome sequencing data available, a diagonal white line denotes no deletions but MLPA data available, and gray squares denote no data available.

Figure 4.

Comparison of sample characteristics with sensitivity to imatinib. (A) Box plots of diagnostic and relapse samples and Z-score of the area under the dose-response curve (AUC). (B) Box plots of primary ALL samples, primary xenograft ALL samples, and serial PDX-transplanted ALL samples and the Z-score of the AUC for imatinib. The mean AUC of imatinib was compared using the Kruskal-Wallis test with Dunn post hoc test. (C) Box plots of ALL cells from B-lineage (indicated by “B”) or T-lineage (indicated by “T”) and the Z-score of the AUC for imatinib. The box plots show the median Z-score ± 1.5× interquartile range in the Tukey style. The red dots represent patient 2, from whom ALL cells were obtained from day 33 after 2 weeks of TKI treatment.

Figure 5.

TKI sensitivity of transduced Ba/F3 cell lines expressing ABL-class fusions. (A) Dose-response curves of transduced Ba/F3 cell lines with various ABL-class fusions for imatinib. (B) Bar graph of half-maximal inhibitory concentration (IC₅₀) values of imatinib obtained from dose-response curves. (C) Dose-response curves of transduced Ba/F3 cell lines with various ABL-class fusions of dasatinib. (D) Bar graph of IC₅₀ values of dasatinib obtained from dose-response curves. (E) Dose-response curves of transduced Ba/F3 cell lines with various ABL-class fusions for bosutinib. (F) Bar graph of IC₅₀ values of bosutinib obtained from dose-response curves. For the dose-response curves, the values are normalized against untreated controls for each cell line and represent the mean ± SEM (n = 3). Bar graphs represent the mean ± SEM.

Figure 6.

TKI sensitivity of cells with ABL-class fusions with and without (intact) SH3 and SH2 domains. (A-C) Dose-response curves for imatinib, dasatinib, and bosutinib, respectively, of Ba/F3 models expressing SH3 domain-lacking RCSD1::ABL1 (in gray) or modified constructs including an intact SH3 and SH2 domain (RCSD1::ABL1wSH3_1/2, in blue). Values are normalized against untreated controls for each cell line and represent mean ± SEM (n = 3). (D-F) Dose-response curves for imatinib, dasatinib, and bosutinib, respectively, of primary or primary xenografted ABL-class ALL samples either with SH3 domain (in blue) or without SH3 domain (in gray). Leukemic cell survival relative to untreated controls represents the mean of 1 to 3 biological replicates.

Figure 7.

Comprehensive overview of TKI sensitivity based on the involved tyrosine kinase in ALL samples and transduced Ba/F3 models relative to BCR::ABL1-positive cells. For primary or primary xenografted ALL samples, the percentage of ALL samples by tyrosine kinase gene (ABL1, PDGFRB, CSF1R, or ABL2) with an area under the dose-response curve (AUC) above the median AUC of BCR::ABL1-positive ALL samples was calculated for each TKI (imatinib, dasatinib, or bosutinib). Colors were assigned based on the calculated percentages; green means ≤75% of the samples are above the median AUC of BCR::ABL1-positive ALL samples. Red indicates that >75% of the samples are above the median AUC of the BCR::ABL1-positive ALL samples. For CSF1R and ABL2, only 1 ALL sample was included in this study. This is denoted in the figure. For the transduced Ba/F3 models, green indicates that the half-maximal inhibitory concentration (IC₅₀) is below the IC₅₀ of the transduced Ba/F3 model expressing the BCR::ABL1 fusion, orange indicates an IC₅₀ above that of the BCR::ABL1 Ba/F3 model, and light green indicates a >10-fold lower IC₅₀ than that of the BCR::ABL1 Ba/F3 model.