Allo-HSCT mitigates the prognostic differences between bZIP and non-bZIP CEBPA mutations in AML: A long-term follow-up study of lu daopei hospital

Abstract Background CEBPA-mutated (CEBPAmu) AML, particularly double mutations (CEBPAdm) or those involving the bZIP domain, is classified as favorable-risk in both WHO 2022 and ICC. However, prognosis remains heterogeneous—patients with single mutations (CEBPAsm), non-bZIP variants often show inferior outcomes, with long-term survival as low as 40–60%. This study aims to evaluate the long-term outcomes of CEBPAmu AML patients undergoing allo-HSCT and to assess the prognostic impact of mutation zygosity, bZIP domain involvement. Methods We retrospectively analyzed 219 AML patients with CEBPAmu who underwent first allo-HSCT between December 2012 and December 2023 at Hebei Yanda Lu Daopei Hospital and Beijing Lu Daopei Hospital. CEBPAmu were identified by next-generation sequencing (NGS). Results Patient Characteristics and Transplant Outcomes A total of 219 CEBPAmu AML patients (137 males, 82 females) with a median age of 32 years (range, 2–68) underwent first allo-HSCT, with donors (median age 35, range 3–72) predominantly haploidentical (HID, 80.4%), followed by MUD (10.5%) and MSD (9.1%). At transplant, 163 (74.4%) patients were in CR (36 MRD positivity), and 56 (25.6%) in NR/PR. Co-mutations were present in 87.7%, most frequently in GATA2 (n=51, 23.3%), WT1 (n=51, 23.3%), FLT3-ITD (n=46, 21.0%), NRAS (n=34, 15.5%), and CSF3R (n=28, 12.8%). Conditioning was mainly BU-based (97%), and GVHD prophylaxis CsA-based (84%). Median infused cell doses were: MNC 9.24×10⁸/kg (range, 2.58–29.3), CD34⁺ 4.6×10⁶/kg (1.14–21). Neutrophil and platelet engraftment rates were 96.3% (median 13 days) and 93.2% (median 13 days), respectively. Median follow-up was 38.3 months (range, 0.5–120.3) as of Jan 1, 2025. Impact of CEBPA Mutation Type on Long-Term Outcomes Among 219 CEBPAmu patients, 132 (60.3%) had CEBPAdm and 87 (39.7%) had CEBPAsm. The 5-year OS (72.5% vs 66.9%, p=0.38) and LFS (70.4% vs 63.2%, p=0.17) were slightly higher in the CEBPAdm group, though not statistically significant. Similarly, 5-year NRM (20.7% vs 24.6%, p=0.60) and CIR (10.5% vs 20.1%, p=0.09) were lower in CEBPAdm, but the differences did not reach statistical significance. Prognostic Impact of bZIP Domain Involvement in 163 CR Patients 126 (77.3%) harbored bZIP-involved mutations and 37 (22.7%) did not. The 5-year OS and LFS were comparable between patients with vs. without bZIP involvement (79.8% vs 75.1%, p=0.61; and 76.2% vs 69.2%, p=0.52). 5-year NRM (18.6% vs 24.9%, p=0.46) and CIR (6.97% vs 8.64%, p=0.81) also showed no significant difference. Within the bZIP-involved group, outcomes were similar between CEBPAsm (n=21) and CEBPAdm (n=105): 5-year OS (81% vs 80%), LFS (76.2% vs 76.3%), NRM (19.1% vs 18.4%), and CIR (9.5% vs 9.3%) (all p>0.05). Outcomes remained comparable when compared to patients without bZIP-involved mutations. Prognostic Impact of bZIP Domain Involvement in 56 NR/PR Patients 31 (55.4%) had bZIP-involved mutations and 25 (44.6%) did not. The 5-year OS and LFS were similar between groups: 46.7% vs 47.3% (p=0.92) and 49.8% vs 45.3% (p=0.55), respectively. 5-year NRM (25.8% vs 31.8%, p=0.79) and CIR (26.9% vs 29.1%, p=0.51) were also not significantly different. Outcomes were comparable between bZIP-involved CEBPAsm (n=9) vs CEBPAdm (n=22), and similar to those without bZIP involvement (23 CEBPAsm and 2 CEBPAdm). However, all NR/PR subgroups had significantly worse post-transplant outcomes compared to their corresponding CR counterparts (all p<0.05). Outcomes by Donor Type in CEBPA-Mutated AML Among the 163 CR patients, 127 received haplo-HSCT and 36 received MUD/MSD transplants. 5-year OS (78.3% vs 80%, p=0.85), LFS (80% vs 73%, p=0.40), NRM (17.2% vs 20.9%, p=0.63), and CIR (2.8% vs 8.6%, p=0.23) were comparable. In NR/PR patients (n=56), outcomes were also similar between haplo and MUD/MSD groups, but all survival indicators were significantly inferior to those in CR patients (all p<0.05). Conclusions Allo-HSCT provides favorable long-term outcomes in patients with CEBPAmu AML in CR. Notably, even those with non-bZIP domain mutations achieve survival comparable to patients with favorable genotypes. These findings indicate that allo-HSCT may help mitigate the adverse prognostic impact of mutation type, bZIP domain involvement, and co-mutations, and support its potential role as a curative option across CEBPA-mutated AML subgroups.