AUTHORS
Alexandre Bazinet, MD, MSc, Guillermo Garcia-Manero, MD2, Farhad Ravandi, MD3, Naval Daver, MD4, Nicholas Short, MD5, Maro Ohanian, DO6, Musa Yilmaz, MD7, Kelly Sharon Chien, MD8, Elias Jabbour, MD9, Abhishek Maiti, MBBS10, Danielle Hammond, MD11, Jo Ishizawa, MD, PhD12, Courtney DiNardo, MD, MSCE13, Lucia Masarova, MD14, Nitin Jain, MD15, William G. Wierda, MD, PhD16, Eitan Kugler, MD PhD17, Fadi G. Haddad, MD18, Gautam Borthakur, MD19, Hagop Kantarjian, MD20, Tapan Kadia I, MD21
AFFILIATIONS
1 MD Anderson Cancer Center, Houston, Texas, United States
2 MD Anderson Cancer Center, Houston, Texas, United States
3 MD Anderson Cancer Center, Houston, Texas, United States
4 MD Anderson Cancer Center, Houston, Texas, United States
5 MD Anderson Cancer Center, Houston, Texas, United States
6 MD Anderson Cancer Center, Houston, Texas, United States
7 MD Anderson Cancer Center, Houston, Texas, United States
8 MD Anderson Cancer Center, Houston, Texas, United States
9 MD Anderson Cancer Center, Houston, Texas, United States
10 MD Anderson Cancer Center, Houston, Texas, United States
11 MD Anderson Cancer Center, Houston, Texas, United States
12 MD Anderson Cancer Center, Houston, Texas, United States
13 MD Anderson Cancer Center, Houston, Texas, United States
14 MD Anderson Cancer Center, Houston, Texas, United States
15 MD Anderson Cancer Center, Houston, Texas, United States
16 MD Anderson Cancer Center, Houston, Texas, United States
17 MD Anderson Cancer Center, Houston, Texas, United States
18 MD Anderson Cancer Center, Houston, Texas, United States
19 MD Anderson Cancer Center, Houston, Texas, United States
20 MD Anderson Cancer Center, Houston, Texas, United States
21 MD Anderson Cancer Center, Houston, Texas, United States
ABSTRACT
Background: Hypomethylating agent-based maintenance can delay relapse and prolong survival in patients (pts) with AML who are in remission as shown by the QUAZAR AML-001 study. We evaluated the safety and preliminary efficacy of fully oral maintenance regimens consisting of oral decitabine/cedazuridine (ASTX727) combined with a targeted agent based on AML molecular profile.
Methods: We conducted a phase 1b study of ASTX727 either alone or combined with physician’s choice of a targeted oral agent as maintenance therapy in AML. Pts ≥ 18 years with AML in first CR/CRi following intensive or lower intensity induction were eligible. Prior to maintenance, pts had to have received at least 1 intensive consolidation or 2 lower intensity consolidations and not be immediately eligible for allogeneic stem cell transplantation (SCT). Treatment consisted of ASTX727 35/100mg PO on D1-3 either alone (arm A) or in combination with venetoclax 400mg PO on D1-5 (arm B), gilteritinib 120mg PO on D1-28 (arm C), enasidenib 100mg PO on D1-28 (arm D), or ivosidenib 500mg PO on D1-28 (arm E). Treatment was administered in 28-day cycles for up to 24 cycles. A 3+3 dose de-escalation design was used. The primary outcome was safety and tolerability. Secondary outcomes included relapse-free survival (RFS; measured from enrollment), overall survival (OS), and MRD clearance. Pts were censored from analysis at the time of SCT. This trial is registered on ClinicalTrials.gov (NCT05010772).
Results: As of 7/23/2026, 43 pts have been enrolled (9 in arm A, 25 in arm B, 3 in arm C, 4 in arm D, and 2 in arm E). By ELN 2022, 15 (35%) pts were favorable, 3 (7%) were intermediate, and 25 (58%) were adverse risk. 6 (14%) pts had antecedent MDS and 3 (7%) had therapy-related AML. Complex cytogenetics and TP53 mutations were identified in 5 (12%) and 2 (5%) pts, respectively. 12 pts (28%) had NPM1 mutations. Prior to maintenance, anti-leukemic therapy was intensive in 19 (44%) pts and lower intensity in 24 (56%) pts. Overall, 34 (79%) pts had been exposed to venetoclax prior to maintenance. At enrollment, 13/41 (32%) pts were MRD-positive by flow cytometry.
No dose de-escalations were required in any of the arms. The most common grade 3/4 adverse events were myelosuppression (neutropenia 91%, leukopenia 88%, thrombocytopenia 63%, anemia 42%), pneumonia (19%), hypertension (16%), and neutropenic fever (14%). One death occurred on maintenance due to sepsis and multi-organ failure. All other deaths occurred following either AML relapse or transition to SCT.
The median follow-up time was 24 months (m) and pts received a median of 6 maintenance cycles (range 1-24). The median RFS for the full cohort was 24 m. The median RFS was 22.4 m in arm A, not reached (NR) in arm B, 10.5 m in arm C, NR in arm D, and 10.1 m in arm E. Pts who received intensive induction had a median RFS that was not reached compared to 22.4 m in those who received lower-intensity induction (p=0.17). Stratified by ELN risk, median RFS was NR in favorable risk, 24 m in intermediate risk, and 22.4 m in adverse risk. Notably, pts with NPM1 mutations (n=12, of which 10/12 were treated in arm B with ASTX727 + venetoclax) had a median RFS that was NR (81% RFS at 2 years). Pts with IDH1 mutations (n=5, 3 treated with ASTX727 + venetoclax, 2 treated with ASTX727 + ivosidenib) had a median RFS of 9.3 m (40% RFS at 1 year). Pts with IDH2 mutations (n=12, including 6 treated with ASTX727 + venetoclax and 4 with ASTX727 + enasidenib) had a median RFS that was NR (RFS 63% at 2 years) and no observed relapses beyond 1 year. The median RFS was NR in pts who were MRD-negative at enrollment compared to only 6.6 m in those who were MRD-positive (p<0.001). Only 3/13 (23%) pts who were MRD-positive cleared MRD on maintenance. The median OS for the full cohort was 33.7 m. By arm, the median OS was 41.1m in arm A, 33.7 m in arm B, 30.7m in arm C, NR in arm D, and NR in arm E.
Conclusion: Personalized oral maintenance regimens were tolerable in pts with AML in remission. Compared to historical expectations of RFS in pts in remission without SCT, these oral maintenance regimens demonstrated encouraging efficacy in preventing relapse, especially in genetically sensitive subtypes (NPM1, IDH2), but had limited impact on MRD clearance.
DISCLOSURES
Alexandre Bazinet, MD, MSc: No relevant conflicts of interest to declare.