Phase I study of decitabine alone or in combination with valproic acid in acute myeloid leukemia.

To determine an optimal biologic dose (OBD) of decitabine as a single agent and then the maximum-tolerated dose (MTD) of valproic acid (VA) combined with decitabine in acute myeloid leukemia (AML). Twenty-five patients (median age, 70 years) were enrolled; 12 were untreated and 13 had relapsed AML. To determine an OBD (based on a gene re-expression end point), 14 patients received decitabine alone for 10 days. To determine the MTD, 11 patients received decitabine (at OBD, days 1 through 10) plus dose-escalating VA (days 5 through 21). The OBD of decitabine was 20 mg/m(2)/d intravenously, with limited nonhematologic toxicity. In patients treated with decitabine plus VA, dose-limiting encephalopathy occurred in two of two patients at VA 25 mg/kg/d and one of six patients at VA 20 mg/kg/d. Drug-induced re-expression of estrogen receptor (ER) was associated with clinical response (P < or = .05). ER promoter demethylation, global DNA hypomethylation, depletion of DNA methyltransferase enzyme, and histone hyperacetylation were also observed. In an intent-to-treat analysis, the response rate was 44% (11 of 25). Of 21 assessable patients, 11 (52%) responded: four with morphologic and cytogenetic complete remission (CR; each had complex karyotype), four with incomplete CR, and three with partial remission. In untreated AML, four of nine assessable patients achieved CR. Clinical responses appeared similar for decitabine alone or with VA. Low-dose decitabine was safe and showed encouraging clinical and biologic activity in AML, but the addition of VA led to encephalopathy at relatively low doses. On the basis of these results, additional studies of decitabine (20 mg/m(2)/d for 10 days) alone or with an alternative deacetylating agent are warranted.