Characteristics and outcomes of CEBPA bZIP in-frame mutations in de novo and treated secondary acute myeloid leukemia

Abstract Introduction: Mutations (mut) in CEBPA occur in 7-20% of patients (pts) with de novo acute myeloid leukemia (AML). In-frame mut in the CEBPA basic leucine zipper domain (bZIP-Inf) account for ~50% of theseCEBPA variants and are associated with favorable outcomes with frontline intensive chemotherapy (IC). However, data on the efficacy of frontline low intensity (LIT) regimens [including those incorporating venetoclax (VEN)] and outcomes of treated secondary (ts)-AML with CEBPA bZIP-Inf mut are limited. Methods: We conducted a retrospective study of adult pts with newly diagnosed CEBPA bZIP-Inf mut AML in two cohorts: de-novo (without antecedent myeloid neoplasm) or ts-AML (previously treated antecedent myeloid neoplasm). Outcomes were assessed based on type of induction therapy (IC or LIT). Results: A total of 34 pts treated from 2/2016 - 4/2025 were included: 28 pts with de novo AML and 6 pts with ts-AML. In the de-novo cohort, IC was used in 17 (61%) pts, LIT in 11 (39%) pts. In the de novo AML IC subgroup (n=17), the median age was 40 years (y) (range, 23 – 68). VEN was added to IC in 11 (65%) pts. Baseline adverse risk cytogenetics (CG) were present in 3 (18%) pts (complex in 2 pts, -5 in 1 pt), whereas 12 (71%) pts had a diploid karyotype. The most common baseline co-mut were RAS pathway (7, 41%), GATA2 (6, 35%), WT1 (4, 24%), TP53 (3, 18%), and FLT3-ITD (3, 18%). Responses with IC were complete remission (CR) in 15 (88%) pts, no response in 1 (6%) pt, and early death in 1 (6%) pt. Among responders, all 13 (100%) pts with available testing attained negative measurable residual disease (MRD) status by flow cytometry. 6 (35%) pts underwent allogenic stem cell transplant (ASCT) in 1st CR, with median time to ASCT of 3.6 months (m). After a median follow up of 44.7 m, 16 (94%) pts remain alive. AML relapse after initial CR occurred in 1 (7%) pt. The median relapse-free survival (RFS) and overall survival (OS) were not reached (NR) (3 y RFS 93% and 3 y OS 94%). On landmark analysis, pts who underwent ASCT in 1st CR had similar median OS compared with those who did not (NR vs NR, p>0.99). In the de novo AML LIT subgroup (n=11), the median age was 66 y (range, 44 – 82). The LIT backbone used was low dose Ara-c in 9 (82%) pts and HMA in 2 (18%) pts. VEN was added to LIT in all 11 (100%) pts. Baseline CG were adverse in 3 (27%) pts (complex in 2 pts, del17p in 1 pt) and favorable in 1 (9%) pt (inv 16); 5 (46%) pts had diploid CG. The most common baseline co-mut were DNMT3a, TET2, and GATA2, occurringin 2 (18%) pts each, whereas TP53 and WT1 mut each occurred in 1 (9%) pt. Responses with LIT were CR in 10 (91%) pts and no response in 1 (9%) pt. Among responders, MRD negativity was attained in 9 (90%) pts. 3 (27%) pts underwent ASCT in 1st CR with a median time to ASCT of 4 m. After a median follow up of 24.6 m, 6 (55%) pts remain alive (all in 1st CR). AML relapse after initial CR occurred in 2 (18%) pts, and 2 (18%) pts died in 1st CR. For the full LIT cohort, median RFS and OS were 14.2 m (3 y RFS 46%) and 15.6 m (3 y OS 39%), respectively. On landmark analysis, pts who underwent ASCT in 1st CR had similar median OS compared to those who did not (NR vs 30.6 m, p= 0.89). The ts-AML cohort included 6 pts with median age of 74 y (range 22 – 78). Pre-AML diagnoses included MDS in 3 pts, MDS/MPN overlap in 2 pts, and myelofibrosis in 1 pt. Prior therapies included HMA (n=3), VEN (n=2), and ruxolitinib (n=1). Common baseline co-mut at ts-AML diagnosis included ASXL1 (4, 66%), RUNX1 (2, 33%), and SRSF2 (2,33%). CEBPA bZIP-Inf mut was newly emergent at the time of AML transformation in all 4 (100%) pts with available longitudinal mut data. IC was used in 1 (17%) pt, LIT in 5 (83%) pts, with 2 (33%) pts attaining CRi, 1 (17%) pt with aplastic bone marrow and 3 (50%) pts with no response. 1 pt underwent ASCT. After a median follow up of 37 m from the start of AML treatment, the median OS for the ts-AML cohort was 13.6 m. Conclusion: De novo AML with CEBPA bZIP-Inf mut in our cohort had excellent MRD negative CR rates to both IC and VEN-based LIT induction regimens indicating a favorable risk profile. Despite a relatively low rate of relapse, OS in the LIT cohort was lower due to an increased frequency of non-relapse mortality. Of interest, we note that CEBPA bZIP-Inf mut in ts-AML often emerge at the time of AML transformation and may mitigate the otherwise dismal outcomes of ts-AML (OS <6 m; Senapati et al., AJH 2025).