Dysfunctional DNA Double-Strand Break Repair Is Present in a Subset of Primary t-AML/t-MDS Myeloblasts

Abstract 2415

The chromosomal aberrations found in treatment-related acute myeloid leukemia/myelodysplastic syndrome (t-AML/t-MDS) cells suggest that disease initiation and progression may result from the inappropriate response to double-strand DNA breaks (DSBs) induced by prior exposure to radiation or chemotherapy. We hypothesized that dysregulation of DSB repair by homology-directed repair (HDR) or nonhomologous end joining (NHEJ) in t-AML/t-MDS may result from acquired mutations in HDR/NHEJ pathway genes. To test this possibility, we used next-generation sequencing technology to identify somatic genetic variants in 21 canonical HDR and 9 NHEJ DNA repair genes, as well as a subset of 7 DNA damage response genes using tumor DNA and paired normal DNA obtained from 25 t-AML/t-MDS patients. We identified 6 patients with somatic changes in 3 of these genes (RAD51L3, EME1, TP53). As dysfunctional DSB repair from epigenetic or post-translational modifications in DSB repair pathway genes or abnormalities in other DNA repair pathway genes would be missed using this approach, in parallel we performed functional studies of DSB repair using primary bone marrow cells from 16 of these t-AML/t-MDS patients and CD34+ cells from 5 normal donors. We evaluated DSB by measuring phosphorylated histone H2AX (pH2AX), a well established marker for DSB, in myeloblasts (CD45 dim, low side scatter) and lymphocytes (a surrogate for normal cells) in these samples. Baseline measurements of primary cells, coupled with a time course to measure pH2AX induction and decay after 2 Gy of irradiation (IR) were used to assess the basal DSB burden and response to acute damage, respectively. pH2AX levels were measured by flow cytometry and the geometric mean of the fluorescence intensity was converted to mean equivalent soluble fluorophore (MESF) through the use of standard beads included in each experiment.

We found that 4 of 16 t-AML/t-MDS patients had myeloblasts that displayed baseline and post-damage pH2AX levels similar to normal CD34+ controls, while 12/16 patients had abnormal pH2AX levels which fell into one of three major patterns. 1) The first subset had myeloblasts in which baseline pH2AX levels were elevated compared to normal donor CD34+ (average MESF 23,107 vs 11,490, respectively; p<=0.002) suggesting an increased basal DSB burden in these cells. Furthermore, the myeloblasts showed impaired pH2AX induction (measured at 30 min. post IR) compared to CD34+ controls (1.53 vs 2.97 fold increase in pH2AX over baseline, p<=0.002), suggesting a defect in detecting DSB. This phenotype was unique to patients harboring trisomy 8 and was tumor specific, as their lymphocytes displayed baseline and post-induction pH2AX levels similar to lymphocytes from normal controls. No somatic (tumor) sequencing variants were present in the interrogated genes, raising the possibility that trisomy 8 could be driving an abnormal DNA damage response. 2) A second subset of patients had impaired pH2AX induction compared to normal donor CD34+ cells (1.44 vs 2.97 fold increase in pH2AX over baseline, p<=0.01), again suggesting a defect in detecting DSBs. These patients also lacked somatic changes in HDR/NHEJ pathway genes. 3) The final subset of patients had delayed resolution of pH2AX levels compared to CD34+ controls post IR either at 4 hours (average MESF 39,260 vs 25,480, p<0.05) or delayed resolution over the entire 24 hour period compared to controls (p<0.001). These data are consistent with a DSB repair defect and similar to our data showing cells lacking BRCA2, a gene central to the HDR pathway, have elevated pH2AX levels at 4–24 hours post DSB induction compared to BRCA2 sufficient cells (p=0.01). One of these patients had an acquired mutation in the HDR gene RAD51L3. We are currently determining the sensitivity of primary t-AML/t-MDS cells with abnormalities in pH2AX levels to a combination of DSB inducing chemotherapy and PARP inhibition, which is synthetically lethal in the setting of HDR defects. We show cell lines lacking RAD51L3 are more sensitive to PARP inhibition compared to isogenic controls (surviving fraction (SF)₅₀ 5 nM vs 20,000 nM). In total, this study confirms that DNA repair genes are mutated in t-AML/t-MDS, suggests that dysfunctional DSB repair is present in t-AML/t-MDS myeloblasts, and provides a rationale to test whether the abnormal DNA damage response can be exploited therapeutically using a synthetic lethal approach in this disease.