5-Azacytidine Augments the Cytotoxicity of Mylotarg toward AML Blasts In Vitro and In Vivo.

Abstract

Binding of CD33 on AML cells by monoclonal antibodies (mAb) mediates cytotoxicicity on AML cells modulated by the protein tyrosine kinase Syk and the phosphatase SHP-1. We have demonstrated that Syk- negative AML cells are relatively resistant to the effects CD33 ligation, but after exposure to the hypomethylating agent 5-azacytidine (5-aza) they become sensitive to the effects of both unconjugated and chemically-conjugated (gemtuzumab ozogamycin, GO) anti-CD33 mAb. Here we tested a panel of 40 primary AML samples for the expression of SHP-1. In 13% of the cases SHP-1 was undetectable. Anti-CD33 mAb and GO induced growth inhibition more effectively in AML cells that expressed SHP-1. Among SHP-1-positive samples, 69% demonstrated significant growth inhibition in response to CD33 ligation. In contrast, none of the SHP-1-negative samples responded to anti-CD33 mAb. These results show a correlation between SHP-1 expression and responsiveness of AML cells to CD33 ligation. However, 5-aza treatment restored SHP-1 expression and, therefore, increased the anti-proliferative effects of anti-CD33 mAb and GO. In 40% of SHP-1-negative samples, the AML cells were only marginally inhibited by 5-aza or anti-CD33 mAb alone, whereas the combination produced a more than additive effect in AML cells where 5-aza induced re-expression of SHP-1. The effect of GO was more than doubled by 5-aza in these cells. Total inhibition of DNA synthesis in the presence of 5-aza plus anti-CD33 mAb or GO reached 60–70% and was similar in SHP-1-positive and SHP-1-negative cells. Moreover, 5-aza significantly enhanced of GO-mediated cytotoxicity in AML progenitor cells. In a NOD/scid mouse model, which permits growth of human AML cells and allows measurement of in vivo therapeutic effects of therapeutic strategies for AML, we tested whether combined treatment of the mice with 5-aza would enhance the cytotoxicity of anti-CD33 mAb and GO. Suboptimal doses of 5-aza by itself, as well as treatment with murine anti-CD33 mAb alone did not cause significant cytoreduction. However, combined treatment of mice with anti-CD33 mAb and 5-aza, resulted in a significant response. Treatment with GO mediated up to 60% inhibition of AML cell proliferation. Combined treatment of mice with GO and 5-aza resulted in reduction of leukemia cells by >80%. These data show an interaction of 5-aza and anti-CD33/GO in an in vivo AML model. Based on these data, we hypothesize that the combination of 5-aza and GO may be a potent therapy for patients with AML. Moreover, Syk and SHP-1 may serve as biomarkers of leukemia cell response. Therefore, we initiated a clinical trial of 5-aza and GO combined therapy. Six patients with relapsed AML have been treated in a dose escalation of 5-aza preceeding GO (6 mg/m2 times two). All 6 were Syk-positive, while SHP-1 expression was detected in 4 samples and absent in two. 2 days of 5-aza treatment in vivo induced re-expression of SHP-1 in both previously SHP-1 negative patient cells. Moreover, significant increases were observed in the levels of Syk protein in one baseline positive sample and 1 SHP-baseline positive sample. Study of the effects of 5-aza alone ex vivo on the baseline patient cells showed no significant effect on leukemia cell proliferation. However, importantly, in all 6 samples 5-aza more than doubled the AML cells' response to cytotoxic effects of GO and naked anti-CD33 mAb. These results suggest that 5-aza may augment the effects of anti-CD33 mAb therapy through demethylation of SYK, SHP-1, and possibly other genes. The clinical efficacy of the combined therapy requires further study.