Expression profile and specific network features of the apoptotic machinery explain relapse of acute myeloid leukemia after chemotherapy

BACKGROUND: According to the different sensitivity of their bone marrow CD34+ cells to in vitro treatment with Etoposide or Mafosfamide, Acute Myeloid Leukaemia (AML) patients in apparent complete remission (CR) after chemotherapy induction may be classified into three groups: (i) normally responsive; (ii) chemoresistant; (iii) highly chemosensitive. This inversely correlates with in vivo CD34+ mobilization and, interestingly, also with the prognosis of the disease: patients showing a good mobilizing activity are resistant to chemotherapy and subject to significantly higher rates of Minimal Residual Disease (MRD) and relapse than the others. Based on its known role in patients' response to chemotherapy, we hypothesized an involvement of the Apoptotic Machinery (AM) in these phenotypic features. METHODS: To investigate the molecular bases of the differential chemosensitivity of bone marrow hematopoietic stem cells (HSC) in CR AML patients, and the relationship between chemosensitivity, mobilizing activity and relapse rates, we analyzed their AM expression profile by performing Real Time RT-PCR of 84 AM genes in CD34+ pools from the two extreme classes of patients (i.e., chemoresistant and highly chemosensitive), and compared them with normal controls. RESULTS: The AM expression profiles of patients highlighted features that could satisfactorily explain their in vitro chemoresponsive phenotype: specifically, in chemoresistant patients we detected up regulation of antiapoptotic BIRC genes and down regulation of proapoptotic APAF1, FAS, FASL, TNFRSF25. Interestingly, our analysis of the AM network showed that the dysregulated genes in these patients are characterized by high network centrality (i.e., high values of betweenness, closeness, radiality, stress) and high involvement in drug response. CONCLUSIONS: AM genes represent critical nodes for the proper execution of cell death following pharmacological induction in patients. We propose that their dysregulation (either due to inborn or de novo genomic mutations selected by treatment) could cause a relapse in apparent CR AML patients. Based on this, AM profiling before chemotherapy and transplantation could identify patients with a predisposing genotype to MRD and relapse: accordingly, they should undergo a different, specifically tailored, therapeutic regimen and should be carefully checked during the post-treatment period.