Abstract
Background: The pathogenesis of acute myeloid leukemia (AML) is complex, with cytogenetic and molecular abnormalities currently regarded as the most critical prognostic factors. The European LeukemiaNet (ELN) guidelines established the ELN-2022 genetic risk stratification based on data from patients receiving intensive chemotherapy, and later proposed the ELN-2024 risk stratification for patients undergoing low-intensity therapy. However, large-scale real-world data are needed for validation. Additionally, certain genetic abnormalities that may influence AML prognosis require further investigation.
Objective: To evaluate the prognostic value of ELN-2022 and ELN-2024 risk stratification in newly diagnosed AML patients and to develop an improved risk model incorporating genetic features.
Methods:
A retrospective analysis was conducted on 428 adult patients with newly diagnosed AML from Qilu Hospital of Shandong University. Patients were divided into two groups based on treatment: intensive chemotherapy (n=258) and venetoclax combined with azacitidine (VA, n=170). The primary endpoints were overall survival (OS) and progression-free survival (PFS). Kaplan-Meier analysis was used to estimate survival rates, and the log-rank test was employed to compare differences between groups. Logistic regression analyzed factors influencing treatment efficacy, while the Cox proportional hazards model identified variables significantly associated with survival. A novel genetic risk model was constructed based on key genetic variants, and its predictive performance was evaluated using time-dependent receiver operating characteristic curves (td-ROC) with the area under the curve (AUC).
Results:
The median age of the entire cohort was 53.0 years (range: 40.0–61.0 years). High-frequency mutated genes included FLT3 (31.8%), DNMT3A (25.2%), NRAS (25.0%), and NPM1 (23.4%). The ELN-2022 risk stratification significantly distinguished OS and PFS among favorable-, intermediate-, and adverse-risk patients in the intensive chemotherapy group (p=0.004) but showed no statistical significance in the VA group. Multivariate analysis identified CEBPA bZIP, DNMT3A, GATA2, IDH2, KIT, NPM1, TP53, U2AF1, and STAG2 as independent prognostic factors. A novel genetic risk scoring model was developed: ELN-2022 favorable/intermediate/adverse-risk groups were assigned 2/5/7 points, respectively, with 1 point added for each risk factor (DNMT3A, GATA2, KIT, TP53, U2AF1, STAG2) and 1 point subtracted for each protective factor (CEBPA bZIP, IDH2, NPM1). Patients were categorized into favorable-risk (≤3 points), intermediate-risk (4–6 points), and adverse-risk (≥7 points) groups. The new model significantly differentiated OS (p=0.003) and PFS (p=0.014) among risk groups in the intensive chemotherapy group. Combined with ELN-2024 stratification, the VA group still showed no significant differences in OS (p=0.051) or PFS (p=0.154). ROC curve analysis demonstrated superior discriminatory ability for the new model (AUC: 0.620) compared to ELN-2022 (0.606) and ELN-2024 (0.531).
Conclusion:
The genetic background of AML influences prognosis under different treatment regimens. ELN-2022 is only applicable for risk stratification in the intensive chemotherapy group, with no significant discriminatory power in the VA group or for ELN-2024. The novel genetic risk model integrating nine key gene mutations demonstrated improved prognostic discrimination in both intensive chemotherapy and VA treatment groups, offering a potential foundation for precise AML risk stratification and therapeutic decision-making.
Ethics statement： This study was approved, and the written informed consent was waived by The Medical Ethics Committee of Qilu Hospital of Shandong University (KYLL-202502-055) due to the retrospective nature of the review, and confirmed that the data was anonymized and maintained with confidentiality. The study was conducted in accordance with the Declaration of Helsinki.
