DNA Methylation and Gene Expression Analysis in a Phase II Randomized Study of Decitabine Vs. Decitabine Plus Valproic Acid in MDS and AML.

Abstract 3808

Poster Board III-744

DNA methylation is a common epigenetic mechanism of gene silencing in patients with the Myelodysplastic Syndrome (MDS) and Acute Myelogenous Leukemia (AML). Epigenetic therapy with drugs which inhibit DNA methylation such as 5-azacytidine and 5-aza-2'-deoxycytidine (decitabine) have proven to be clinically potent in MDS and AML. In addition to DNA methylation inhibitors, histone deacetylase inhibitors (HDACi) have activity in leukemias, and at low doses show epigenetic synergy with DNA methylation inhibitors. To test this synergy in the clinic, we designed a phase II randomized study comparing decitabine alone (20 mg/m2 IV daily x 5 every 4 weeks) to decitabine (same dose) plus valproic acid (50 mg/kg PO daily for 7 days started at the same time as decitabine). We have previously reported interim results from this study, showing an overall response rate of 64% in MDS/CMML (CR in 39%) and 46% in AML (CR in 25%) with no significant differences in response or survival between the two arms. We now report on molecular analyses in this trial. We have studied DNA methylation of ALOX12, LINE1, MapK15, miR124a-1 and 3 and P15 using bisulfite-pyrosequencing, and expression of ATM, mi124a, p15 and p21 by qPCR at baseline and at days 5, 12 and 30 after initiation of therapy in 60 (32 for expression) patients treated on the study (33 received decitabine, 27 received decitabine + valproic acid, overall there were 31CRs or HI's and 28 NRs, 1 patient was inevaluable for response). Global methylation (measured by LINE1) decreased at day 5 by an average of 6.8 ±1.8% in the DAC arm and 3.5 ± 1.2% in the DAC/VPA arm (p=0.20). At day 12, the decrease (from baseline) was by 10.2 ± 2.2% in the DAC arm and 7.0 ± 1.5% in the DAC/VPA arm (p=0.32). At day 30 we observed a decrease of 6.4 ± 1.4% in the DAC arm and 4.8 ±2.2% in the DAC/VPA arm. We found no statistical differences between the two arms in any of the other genes studied for hypomethylation. By qPCR, expression of p15 at day 5 increased by 1.2±0.6 fold in the DAC arm and by 2.5±0.7 fold in the DAC/VPA arm (p=0.01). We found no differences in the other 3 genes studied between the two arms. We next asked about correlations between epigenetic modulation and response. There was no association between LINE1 methylation change at days 5, 12 or 30 and response. By contrast, sustained hypomethylation of miR124a1 correlated with response; at day 5, miR124a methylation had changed by -17.9 ± 3.7% in responders vs. -15.2 ± 5.8% in non-responders, while at day 30, methylation decreased further to -24±6.5%% in responders, but had already partially recovered to -5.3 ±5.8% of baseline in non-responders (p=0.029 for a comparison between responders and non-responders). Additionally we found that responders hypomethylated miR124a-3 faster by a change in methylation of -34.1 ± 6.3% at day 5 compared to non-responders who had -14.8 ± 6.3% at day 5 (p= 0.039). However there was no difference at day 30. By qPCR we studied the same genes as previously listed. We found that responders had a larger induction of p15 gene expression at day 5, 2.3 ± 0.75 fold compared to non-responders who had a 0.91 ± 0.66 fold increase (p=0.018). We also found a similar pattern in expression induction in the ATM gene, where responders at day 5 had a 1.92 ± 0.51 fold increase as compared to non-responders who had a 0.3 ± 0.64 fold increase (p=0.034). Similarly, for the mature miR124a locus, responders had a 2.91 ± 0.88 fold increase in expression at day 30 compared to non-responders who had 1.1 ± 0.24 fold change in gene expression (p=0.03). In conclusion, we found that adding Valproic acid to decitabine enhances activation of P15, but also shows trends for reducing hypomethylation induction, which is consistent with in-vitro studies. These opposing trends may explain why the response rate is not dramatically different in the two arms. We also found that sustained gene specific hypomethylation correlates with response, as does induction of expression of P15, miR124 and ATM, which confirms and extends our prior studies. Thus, modulation of DNA methylation and gene expression appears to be associated with response to decitabine, and testing whether histone deacetylase inhibitors enhance this response will require non-overlapping dosing regimens and, likely, more potent HDAC inhibitors.