Abstract
Histone deacetylases inhibitors (HDIs), such as valproic acid (VPA), demonstrate significant clinical activity in a proportion of patients with high risk acute myeloid leukemia (AML). However the mechanism by which HDIs selectively induce apoptosis in leukemic blasts remains unknown. We have therefore correlated the impact of VPA exposure on gene expression in leukemic cell lines and primary AML blasts with its ability to induce cell death in vitro and induce clinical responses in vivo. 14 hemato-lymphoid cell lines were tested for their sensitivity (IC50) to apoptotic cell killing by VPA. Gene expression array analyses using HGMP chip 6500 gene arrays were performed on the same cell lines prior to HDI exposure. A bioinformatics approach combined the array and IC50 data to generate a score for each gene identifying those whose elevated expression correlated with sensitivity or resistance to VPA in vitro. In a concurrent Phase I/II clinical trial 24 patients with high risk AML (relapsed n=11, newly diagnosed n=6, primary refractory n=3) with a median age of 64 yrs (41–83 yrs) received combination treatment with VPA, all trans-retinoic acid (ATRA) and theophylline. Changes in histone acetylation and expression of HDI responsive genes was measured in leukaemic blasts before and after commencement of VPA therapy. Expression of genes associated with VPA sensitivity in vitro were re-analysed with respect to their expression in pre-treatment blasts from non-responding and responding trial patients. By combining LC50 values to VPA and microarray data generated from pre-treatment RNA in 14 hematolymphoid cell lines we were able to identify candidate genes and signalling networks mediating sensitivity and resistance to VPA in vitro. Genes whose higher expression conferred sensitivity included IL-1β and those associated with VPA resistance included PLOD2, cyclin B1 (CCNB1) and ACVR2A. In the clinical trial 5 patients, all with relapsed AML, achieved clinical responses by Cheson criteria (complete remission n=1, partial remission n=4). Comparison of gene expression, as defined by microarray studies, in responding and non-responding patients with the previously identified in vitro VPA signalling networks identified that similar networks to those defined in vitro appeared to be correlated with clinical response. This study demonstrates induction of pro-apoptotic gene expression by VPA in a clinical context and identifies potential mechanisms through which its anti-leukaemic effect may be mediated in vivo. Furthermore we have identified potential VPA-signalling networks containing novel sensitivity and resistance genes whose expression correlates with VPA-mediated tumour cell killing in vitro and may predict clinical activity.
Author notes
Disclosure: No relevant conflicts of interest to declare.