Personalized Oral Maintenance Therapy with Decitabine/Cedazuridine (ASTX727) Combined with a Molecularly Targeted Agent (Venetoclax, Gilteritinib, Ivosidenib, or Enasidenib) in Acute Myeloid Leukemia in First Remission

Background: Most patients with acute myeloid leukemia (AML) experience relapse after initial remission, especially when unable to complete standard consolidation strategies such as high-dose cytarabine or allogeneic stem cell transplantation (SCT). Ideal maintenance therapy consists of the prolonged administration of effective lower intensity agents with the goal of delaying or preventing relapse. Oral azacitidine (QUAZAR AML-001) and 3-day IV decitabine (ECOG ACRIN E2906) have demonstrated efficacy as maintenance. In this phase 1b multi-arm study (NCT05010772), we evaluated the combination of ASTX727 (decitabine/cedazuridine), an oral formulation of decitabine with equivalent AUC to IV decitabine, plus physician choice of an oral targeted agent as personalized maintenance therapy in AML.

Methods: This study enrolled patients ≥ 18 years with AML in first complete remission (CR) or CR with incomplete blood count recovery (CRi) unable to complete standard therapy and not immediately eligible for SCT. Patients who had received intensive induction (at least 2 cycles of intermediate/high dose cytarabine-based chemotherapy) were enrolled in cohort 1. Patients who had received at least 3 cycles of low-intensity therapy (hypomethylating agent or low-dose cytarabine-based regimens) were eligible for cohort 2. ECOG ≤ 3 and adequate bone marrow function (neutrophils > 0.5 x 10⁹/L, and platelets > 50 x 10⁹/L) were required. All patients received ASTX727 35/100 mg PO on D1-3. Patients were assigned to one of 5 arms per physician decision. Arm A consisted of ASTX727 alone. A second agent was added for the remaining arms (Arm B: venetoclax 400 mg (adjusted for CYP3A4 inhibitor use) PO on D1-5; Arm C: gilteritinib 120 mg PO on D1-28; Arm D: enasidenib 100 mg PO on D1-28; Arm E: ivosidenib 500 mg PO on D1-28). Each arm included an initial safety lead-in cohort using a 3+3 dose de-escalation design in the event of dose limiting toxicity (DLT). Cycles were 28 days and therapy continued for up to 24 cycles. The primary objective was safety/tolerability. Secondary objectives included overall survival (OS) and relapse-free survival (RFS) starting from the time of enrollment. Time-to-event endpoints were censored at the time of SCT.

Results: 31 patients have been enrolled (15 in cohort 1, 16 in cohort 2). 4 patients have been enrolled in arm A, 25 in arm B, 1 in arm C, and 1 in arm D. The median follow-up time is 15.6 m. The median age for the full cohort was 68 years (33-83). ELN 2022 risk was favorable in 13 patients (42%), intermediate in 3 (10%), and adverse in 15 (48%). TP53 mutations were identified in 1 (3%) patient and complex cytogenetics in 3 (10%) patients. 4 (13%) patients had antecedent MDS/MPN and 3 (10%) had therapy-related AML. 9 (29%) patients had detectable measurable residual disease (MRD) at the time of enrollment.

The median number of cycles given is 5 (1-24). No DLT’s have been noted in cycle 1. The most common grade 3/4 adverse events were neutropenia (100%), leukopenia (100%), thrombocytopenia (83%), and anemia (57%). Neutropenic fever occurred in 20% of patients. 1 (3%) patient went off protocol due to toxicity (prolonged thrombocytopenia after 6 cycles) and 1 death occurred while on maintenance.

For the full cohort, the median OS is not reached (NR; 82% at 1 year) and the median RFS is 22.4 months (m) (59% at 1 year). In cohort 1 (intensive induction), the median OS is NR (100% at 1 year) and the median RFS is NR (76% at 1 year). In cohort 2 (low-intensity induction), the median OS is NR (69% at 1 year) and the median RFS is 11.7 m (43% at 1 year). In arm A (ASTX727 alone), the median OS is NR (100% at 1 year) and the median RFS is 16.8 m (50% at 1 year). In arm B (ASTX727 plus venetoclax), the median OS is NR (78% at 1 year) and the median RFS is NR (65% at 1 year).

Of the 9 patients with detectable MRD at enrollment, 2 (22%) became MRD-negative while on maintenance. The median OS was 11.3 m in the MRD-positive patients versus NR in the MRD-negative patients (n=22; p < 0.01). The median RFS was 6.6 m in the MRD-positive patients versus NR in the MRD-negative patients (p < 0.001).

Conclusions: Combination targeted oral maintenance is feasible in AML and demonstrates encouraging RFS. Myelosuppression is universal in subsequent cycles. Prophylactic anti-infectives are encouraged. Further enrollment exploring dose/schedule modifications of ASTX727 on subsequent cycles and longer follow-up are required to confirm the efficacy of these regimens.