Identification of Somatic Mutation Contributing to Chemotherapy Resistance in Acute Myeloid Leukemia

Acute myeloid leukemia (AML) is a heterogeneous disease with various underlying genetic alterations. After conventional induction chemotherapy with cytarabine and anthracycline, approximately 20% of AML patients fail to achieve complete remission, associated with poor prognosis. Despite the recent identification of novel recurrent driver mutations and advances in the understanding of the molecular pathogenesis, little is known about the relationship between genetic abnormalities and chemoresistance in AML. Herein, we aimed to reveal somatic mutations contributing to chemoresistance in primary refractory AML patients. Firstly, we performed whole-exome sequencing (WES) analysis using paired samples at diagnosis and refractory phase from 6 primary refractory AML patients. All patients were failed to achieve complete remission by at least two cycles of standard induction chemotherapy with cytarabine and idarubicin. Buccal mucosa or CD3⁺ T cells were used as germline control. All single nucleotide variants detected by WES were validated by targeted sequencing. As a result, we identified total of 50 somatic mutations including previously described mutated genes in AML patients such as ASXL1, ASXL2, BCOR, DSCAM, GBP4, KCNH2, KIT, LRP1B, MUC5B, PTPRN, PTPN11, SMAD9, U2AF1, and VASH1. The average number of somatic mutations was 8.3 per case (range 2-17), and there were no recurrent mutations among these 6 patients. Next, targeted sequencing of all coding sequence (CDS) of these 50 genes, in which mutations were found by WES, was performed using 18 primary refractory AML samples at refractory phase. Interestingly, we identified BCOR nonsense or frameshift mutations in 4 samples (22%). Although BCOR disruptive mutations were known to be associated with various hematologic malignancies including AML, the frequency of BCOR mutations in our cohort was higher than that of previously reported historical controls from TCGA of 200 de novo AML samples. Therefore, we performed targeted sequencing of all CDS of BCOR using additional 50 samples at diagnosis including 15 primary refractory AML and 35 responder AML who achieved CR after 1 or 2 cycles of induction chemotherapy. As a result, BCOR disruptive mutations were detected in 3/15 (20%) cases in primary refractory AML samples and 1/35 (3%) cases in responder AML samples, revealing that BCOR mutations were enriched in refractory phase samples. These results indicate that BCOR disruptive mutations are related to chemoresistance in AML. To explore the biological effect of BCOR disruptive mutations in chemoresistance, we generated BCOR knockout THP-1 human AML cell lines using CRISPR/Cas9 system. When we examined the effect of BCOR nullizygosity on cell proliferation, BCOR-null THP-1 cells showed decreased proliferation capacity compared with BCOR-intact control cells in liquid culture condition. Importantly, in vitro administration with cytarabine or idarubicin showed decreased percentage of apoptotic cells in BCOR-knockout THP-1 cells compared with BCOR-intact cells. Finally, we established murine subcutaneous xenograft model of BCOR-null THP-1 cells or BCOR-intact control cells to investigate the chemoresistance nature of BCOR nullzygosity in vivo. As expected, BCOR-null mice showed increased relative tumor volume compared with BCOR-intact control mice after idarubicin administration, consistent with low complete remission rate of BCOR-mutated AML in human cases. Taken together, our findings demonstrated that BCOR disruptive mutations are frequently associated with primary refractory AML and confer resistance to conventional chemotherapy with cytarabine and anthracycline, highlighting the clinical and biological significance of BCOR mutation in AML. Further understanding of the molecular basis of chemoresistance in BCOR-mutated AML may provide novel insights into mechanism of chemoresistance and new therapeutic target of primary refractory AML.