2026 ASH®
Annual Meeting & Exposition December 12-15, 2026 ▪ New Orleans, Louisiana

Paper Details

Abstracts, authors, affiliations, and disclosures.

Saturday, December 12, 2026
05:30 p.m. - 07:30 p.m.
MCCNO - Hall F
Access, Representation, Population Health, Technology and Procedures, Imaging, Molecular Testing, Omics Technologies, Pathology, Bioinformatics, Blood Banking, Computational Biology, Emerging Technologies, Profiling, Treatment Considerations, Drug Development, Supportive Care, Therapy Sequence, Biological Therapies, Chimeric Antigen Receptor, (CAR)-T Cell Therapies, Women’s Health, Workforce, Adverse Events, Artificial Intelligence (AI), Computer Vision, Deep Learning, Generative AI, Machine Learning, Natural Language Processing, Biological Processes, Apoptosis, Genomics, Hematopoiesis, Molecular Biology, Clinical Practice (Health Services and Quality), Clinical Guidelines, Quality Improvement, Quality of Care, Systems-Based Hematology, Diseases, Bleeding and Clotting, Thromboembolism, Bone Marrow Failure Syndromes, Inherited Marrow Failure Syndromes, Paroxysmal Nocturnal Hemoglobinuria, Lymphoid Malignancies, Lymphoid Leukemias, Lymphomas, Plasma Cell Disorders, Myeloid Malignancies, Acute Myeloid Malignancies, Chronic Myeloid Malignancies, Metabolic Disorders, Iron Deficiency, Measurable Residual Disease, Non-Hodgkin Lymphoma, B Cell Lymphoma, Aggressive Lymphoma, Research, Fundamental Science, AML, APL, Translational Research, MDS, MPN, Clinical Research, Clinical Trials, Real-World Evidence, Study Population, Human, Adult, Pediatric
PAPER

2584 Random survival forest-based individualized prognostic assessment for newly diagnosed acute myeloid leukemia treated with venetoclax plus intensive chemotherapy

Xucan Lou, Lijie Wang2, Yiyi Yao3, Xiangie Lin4, Mingyuan Zhang5, Xiaona Zheng6, Liping Mao7, Yinjun Lou, MD8, Yanhong Tong9, Jie Jin, MD10, Zhaoxiang Wang11, Huafeng Wang, MD12
1 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
2 Research Center for Life Sciences Computing, Zhejiang Lab, China, Hangzhou, China
3 Department of Pathology, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China, The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
4 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
5 Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
6 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
7 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
8 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
9 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
10 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China
11 Research Center for Life Sciences Computing, Zhejiang Lab, China, Hangzhou, China
12 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, Zhejiang Provincial Clinical Research Center for Hematological disorders, Hangzhou, China, Zhejiang Provincial Key Lab of Hematopoietic Malignancy, Zhejiang University, Hangzhou, China, Zhejiang University Cancer Center, Hangzhou, China

Background:

Acute myeloid leukemia (AML) exhibits profound biological and clinical heterogeneity, challenging risk-adapted treatment decisions. Accumulating evidence suggests that no single prognostic framework adequately captures this diversity, particularly as different induction regimens yield distinct response and relapse dynamics, necessitating treatment-specific stratification. The ELN 2022 criteria, developed for intensive chemotherapy, and the ELN 2024 criteria, tailored to lower-intensity therapies, exemplify this need. We previously showed that venetoclax plus intensive chemotherapy (V+IC) significantly improves outcomes in newly diagnosed AML (Lancet Hematol, 2022); however, ELN 2022 failed to distinguish intermediate- from adverse-risk patients in this setting, indicating limited transferability of current genetic-risk models to novel therapeutic contexts. To address this, we developed a random survival forest (RSF) model—a machine learning approach that captures complex nonlinear interactions beyond conventional Cox regression—using a V+IC-treated cohort.

Methods:

Study design and patients: 

This study enrolled 340 newly diagnosed AML patients treated with V+IC across nine centers from 2021 to 2024, with data from five registered trials (ChiCTR2000041509, ChiCTR2100045780, ChiCTR2100053516, ChiCTR2200059663, and ChiCTR2300073897). Baseline demographics, hematological parameters, FAB subtype, cytogenetics, and targeted mutations were prospectively collected.

Model development and statistics: 

Patients were split 80:20 into training (n=272) and validation (n=68) cohorts. An RSF model was built in the training set using permutation-based variable importance, with performance evaluated by concordance index (C-index) and time-dependent area under the curve (AUC) in the validation set against Cox regression. Candidate predictors were preselected at P<0.20 in univariable analysis, supplemented by clinically relevant covariates, and refined via least absolute shrinkage and selection operator (LASSO) Cox regression. SHapley Additive exPlanations (SHAP) values enhanced interpretability.Allogeneic hematopoietic stem cell transplantation (allo-HSCT) was modeled as a time-dependent covariate. Exploratory transcriptomic and mutational analyses were performed on paired diagnosis-relapse samples.

Results:

Patient characteristics: Median age was 48 years (range, 15-71), 48.8% male. Median WBC was 12.0×10?/L, and median bone marrow blasts were 63.0%. ELN 2022 distribution: favorable 35.9%, intermediate 30.2%, adverse 33.8%. After one V+IC cycle, 87.6% achieved CR/CRh/CRi.

Prognostic performance: ELN 2022 distinguished favorable from intermediate/adverse risk for OS but failed to separate intermediate from adverse (OS: P=0.462; EFS: P=0.083). The RSF model outperformed Cox regression across ten replicates in validation C-index (0.690 vs. 0.628; P=0.003), 1-year AUC (0.777 vs. 0.724; P=0.022), and 2-year AUC (0.678 vs. 0.656; P=0.044).

Risk reclassification: RSF reduced adverse-risk patients from 33.8% to 14.1% and expanded intermediate-risk from 30.3% to 52.4%. RSF adverse group had higher NR rate than ELN 2022 adverse group (16.7% vs. 11.3%; P=0.042). Within each ELN 2022 stratum, RSF identified subgroups with distinct survival, especially adverse-risk subsets with poorer OS.

Predictors & transplant: LASSO-penalized Cox regression retained MECOM(EVI1) (HR, 17.11; P < 0.001), RUNX1 (HR, 2.32; P = 0.030), and FLT3-ITD (HR, 1.58; P = 0.039) as independent predictors, TET2 was borderline (HR, 1.62; P=0.093).SHAP confirmed FLT3-ITD, TET2, and RUNX1 as high-risk drivers, while NPM1, IDH2, CEBPA(b-Zip), and CBFβ::MYH11 as protective. Allo-HSCT improved OS in RSF adverse-risk (P=0.043), showed no benefit in intermediate-risk (P=0.404), and was linked to inferior OS in favorable-risk (P=0.010). High RSF risk was transcriptomically associated with cell migration, morphogenesis, ECM organization, and myeloid differentiation.

 

Conclusion:

In summary, our novel RSF model offers individualized, V+IC-specific survival prediction for newly diagnosed AML. It outperformed Cox regression, refined risk stratification within ELN 2022 categories, and may inform personalized HCT decision-making. As a complement rather than an alternative to ELN 2022, this model requires prospective external validation to confirm its clinical utility.

Xucan Lou: No relevant conflicts of interest to declare.