Update of a Phase I/II Trial of 5-Azacytidine Prior to Gemtuzumab Ozogamicin (GO) for Patients with Relapsed Acute Myeloid Leukemia with Correlative Biomarker Studies

Abstract 3286

Acute myeloid leukemia (AML) cells express the cell surface antigen CD33 that is a down-regulator of cell growth when ligated by a monoclonal antibody in a Syk-dependent manner. The response of AML cells to gemtuzumab ozogamicin (GO) also depends on Syk and SHP-1 expression (Leukemia 20:2093, 2006). The hypomethylating agent 5-azacytidine (5-aza) induced re-expression of Syk in some cases, therefore increasing the sensitivity of originally Syk-negative, non-responsive cells to CD33 ligation to levels of Syk-positive cells. We initiated a phase 1/2 clinical trial examining if treatment with 5-aza prior to GO is safe, efficacious, and whether in vivo responses to GO correlated with Syk expression and induction by 5-aza. Here we update the interim results of this trial (NCI registration number NCT00766116). In Phase I, 14 patients (9 males, 5 females), age range: 39–82 years [median: 66]) were treated with 75mg/m² 5-aza daily and GO in a dose-escalation manner, 4 cohorts total. The first cohort (n=3) received 5-aza for 2 days followed by GO at 3 mg/m² on days 3 and 17; the second cohort (n=3) received 5-aza for 2 days followed by GO at 6 mg/m² on days 3 and 17; the third cohort (n=4) received 5-aza for 4 days followed by GO at 6 mg/m² on days 5 and 19; and the fourth cohort (n=4) at 5-aza for 6 days followed by GO at 6 mg/m² on days 7 and 21. There were no responses in the first 2 cohorts. One patient in cohort 3 achieved CR, and 2 in cohort 4 achieved CR and CRp. Adverse events (≥ Grade 3) included febrile neutropenia 36%, infection 14%, pancytopenia 7%, dyspnea 7%, and retinopathy 7%. Average length on study (n=14) was 45 days with a mortality rate of 14% (unrelated to treatment). No dose-limiting toxicities were encountered in phase I, therefore the MTD is the dose in cohort 4. The overall response rate in evaluable patients in phase I (n=11) is 27%. Average time to ANC recovery (n=6): 30 days (range 15–42, median 33 days). In Phase II, 10 patients (5 males, 5 females), age range: 29–64 years (median 60) were treated at the MTD: 5-aza for 6 days and GO at 6 mg/m² on days 7 and 21. 8 patients were in 1^st relapse, 1 in 2^nd and 1 in 3^rd. There were 3 responders (2 CR, 1 CRp) in this phase, all in 1^st relapse at baseline. Adverse events (≥ Grade 3) include febrile neutropenia 50%, infection 20%, increased LFTs 10%, thrombocytopenia 10%, dyspnea 10%, wheezing 10%, mucositis 10%, cough 10%, and hypoalbuminemia 10%. The average length on study (n=10) was 40 days with a mortality rate of 10% (not related to study treatment). Average time to ANC recovery in phase II (n=2): 15 days (range 12–17, median 15) with an overall response rate in evaluable patients (n=7) of 43%. The ORR for phase I/II (n=18) is 33%. 21 of the 24 patient sample pairs have been analyzed for Syk and SHP-1 expression (one patient did not have a baseline sample). Prior to therapy, Syk was expressed in 16 of 20 cases. After 5-aza treatment, Syk was re-expressed in all 4 negative cases, and increased over baseline in one case that was previously Syk +. SHP-1 was positive in 17 of the 20 cases and was re-expressed in all 3 negative cases. Leukemia cells from patients who achieved CR were Syk+ in 3 of 5 cases (the 6^th hasn't been analyzed). Syk was re-expressed in the two negative cases after 5-aza. SHP-1 was expressed in 4 of 5 cases at baseline, and re-expressed in the one negative case after 5-aza. In vitro we analyzed inhibition of proliferation (for patients 1–6) or colony formation (for patients 7–24) induced by 5-aza and GO. 5-aza alone allowed 62.3+/−3.5 survival of leukemia cells and GO alone allowed survival of 59.5+/−1.7 leukemia cells. However, exposure to both agents resulted in a survival rate of 24.8+/−1.6 (P<0.05, Students t-test). We also compared pre- and post 5-aza samples from the same patients: in all cases 5-aza treatment increased the GO-mediated cytotoxicity from 39.4+/−3.1 to 66.8+/−2.4 ((P<0.05, Students t-test). These data show that in vivo exposure to 5-aza can induce the expression of two biomarkers involved in the response to GO. This ongoing study indicates the combination of 5-aza and GO is well-tolerated, that Syk and SHP-1 are modulated by 5-aza in vivo, and that complete responses have been noted with this combination.