Abstract
Abstract 2162
PDK1 is a master kinase responsible for regulating at least six kinase groups including AKT, PKC and S6K. Many of these kinases have been shown to be constitutively active in tumour tissues including leukemia suggesting PDK1 is frequently dysregulated. Here we describe the frequency and significance of PDK1 overexpression in AML and determine the potential of PDK1 as therapeutic target. Analysis of 113 AML patients showed that overexpression of PDK1 compared with normal blast cells was frequently observed in 2 groups of patients defined by FAB M1/M0 (38% PDK1Hi) and M4/M5 (44% PDK1Hi). To establish whether overexpression was a property of the entire leukemic population (including putative leukemic stem cells) we carried out multiparameter flow cytometric intracellular staining. In 7/8 patients, we found PDK1 overexpression to be uniformly expressed in all leukemic sub-populations. Ectopic overexpression of PDK1 in normal CD34+ cells promoted their survival in unsupplemented medium (155% ±38, P<0.01, n=6). Similarly, we found that primary AML blasts overexpressing PDK1 displayed significantly higher survival in vitro when cultured in the absence of growth factors than those with normal PDK1 levels (82.4%±10.1 in PDK1Hi vs 64.4%±12.2 in PDK1Norm; P<0.0001, n=71). To determine whether PDK1Hi AML blasts were more dependent on PDK1 activity for their survival, we assessed their sensitivity to PDK1 inhibition compared with PDK1Norm blasts, using a selective inhibitor, BX-795. We found that PDK1Hi blasts showed significantly increased sensitivity to PDK1 inhibition (PDK1Hi 6.4μM +/−3.07 n=30; PDK1Norm 13.4μM +/−7.4 n=37, P=0.001) indicating that oncogene addiction is a feature of PDK1 overexpression in AML. In contrast, normal bone marrow CD34+ cells were resistant to BX-795 suggesting that PDK1 inhibition selectively targets the survival of AML blasts. Western blotting of BX-795-treated AMLs revealed dose dependent knock down of all detectable PDK1 targets substantiating the specificity of this inhibitor; furthermore BX-795 showed synergistic action when used in combination with AraC (CI= 0.7). We next investigated the clinical significance of PDK1 overexpression (using a statistical model to adjust for known prognostic factors) and discovered significant heterogeneity in overall survival between M1/M0 and M4/M5 AMLs (p=0.001 for interaction) with M1/M0 PDK1Hi patients showing significantly improved overall survival (HR 0.14 (0.003-0.61) p=0.004) whereas M4/M5 PDK1Hi patients showed poorer overall survival (although this did not reach significance: HR 2.08 (0.97-4.46) p=0.06). To examine the basis of this differential survival pattern, we surveyed the targets previously associated with PDK1 hyperactivation. We discovered that PDK1 overexpression in M4/M5 patients exclusively suppressed expression of cyclin D3 (R2 =0.675). Ectopic expression of PDK1 in AML cell lines confirmed that PDK1 suppressed D3 expression in monocytic AML cell lines whereas it induced cyclin D3 in AML lines derived from undifferentiated AML. These data indicate that suppression of cyclin D3 expression by PDK1 may reduce the rate of proliferation in M4/M5 AML, decreasing their sensitivity to standard chemotherapeutic treatments (which are most effective in targeting highly proliferative cells). Taken as a whole these data suggest that therapeutic targeting of PDK1 is an effective and selective treatment for AML, but is likely to be most beneficial for M4/M5 patients. Further, these data demonstrate that developmental context can be a significant factor when establishing the role and significance of abnormalities in AML.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.