Background: Recent advances in acute myeloid leukemia(AML) targeted therapy improve overall survival. While these targeted therapies can achieve prolonged remissions, most patients will eventually relapseunder therapy. Our recent studies suggest that relapse most often originates from several sub-clones of leukemic stem cells (LSCs), present before therapy initiation, and selected due to several resistance mechanisms. Eradication of these LSCs during treatment induction /remission could thus potentially prevent relapse.
The overall goal of the current study was to identify drugs which can be safely administrated to patients at diagnosis and that will target LSCs. Since simultaneously testing multiple drugs in vivo is not feasible, we used an in vitrohigh throughput drug sensitivity assay to identify new targets in primary AML samples.
Methods: Drug sensitivity and resistance testing (DSRT) was assessed in vitro (N=46 compounds) on primary AML samples from patients in complete remission (N=29). We performed whole exome sequencing and RNAseq on samples to identify correlations between molecular attributes and in vitro DSRT.
Results:Unsupervised hierarchical clustering analysis of in vitro DSRT, measured by IC50, identified a subgroup of primary AML samples sensitive to various tyrosine kinase inhibitors (TKIs). In this subgroup, 52% (9/17) of AML samples displayed sensitivity to dasatinib (defined as a 10-fold decrease in IC50 compared to resistant samples). Dasatinib has broad TKI activity, and is safely administered in the treatment of leukemia. We therefore focused our analysis on predicting AML response to dasatinib, validating our results on the Beat AML cohort.
Enrichment analysis of mutational variants in dasatinib-sensitive and resistant primary AML samples identified enrichment of FLT3/ITD (p=0.05) and PTPN11(p=0.05) mutations among dasatinib responders. Samples resistant to dasatinib were enriched with TP53 mutations (p=0.01). No global gene expression changes were observed between dasatinib-sensitive and resistant samples in our cohort, nor in the Beat AML cohort. Following this, we tested the differential expression of specific dasatinib-targeted genes between dasatinib-responding and resistant samples. No significant differences were identified. However, unsupervised hierarchical clustering of dasatinib targeted genes expression in our study and in the Beat AML cohort identified a subgroup of AML samples (enriched in dasatinib responders) that demonstrated overexpression of three SRC family tyrosine kinases:FGR, HCK and LYN as well as PTK6, CSK, GAK and EPHB2.
Analysis of the PTPN11 mutant samples revealed that the IC50 for dasatinib in 23 carriers of the mutant PTPN11 was significantly lower compared to the IC50 of PTPN11 wild type samples (p=0.005). LYN was also upregulated (p<0.001) in the mutant samples. We therefore hypothesized that gene expression of dasatinib-targeted genes could be used as a predictive biomarker of dasatinib response among FLT3/ITD carriers. We found that among FLT3/ITD AML carriers in the Beat AML cohort LYN, HCK, CSK and EPHB2 were significantly over-expressed in the dasatinib responding samples (N=27) as compared to the dasatinib resistant samples (N=35). To predict response to dasatinib among FLT3/ITD carriers we used a decision tree classifier based on the expression levels of these four genes. Our prediction model yielded a sensitivity of 74% and specificity of 83% for differentiating dasatinib responders from non-responders with an AUC of 0.84. Based on our findings, we selected FLT3/ITD AML samples and injected them to NSG-SGM3 mice. We found that in a subset of these samples, dasatinib significantly inhibited LSCs engraftment. This subset of FLT3/ITD AML samples expressed higher levels of LYN, HCK,FGR and SRC as compared to the FLT3/ITD samples that were not sensitive to dasatinib therapy in vivo.
In summary, we identified a subgroup of AML patients sensitive to dasatinib, based on mutational and expression profiles. Dasatinib has anti-leukemic effects on both blasts and LSCs. Further clinical studies are needed to demonstrate whether selection of tyrosine kinase inhibitors, based on specific biomarkers, could indeed prevent relapse.
Tavor:Novartis: Membership on an entity's Board of Directors or advisory committees; Abbvie: Membership on an entity's Board of Directors or advisory committees; Astellas: Membership on an entity's Board of Directors or advisory committees; Pfizer: Membership on an entity's Board of Directors or advisory committees; BMS companies: Membership on an entity's Board of Directors or advisory committees.
Author notes
Asterisk with author names denotes non-ASH members.
