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

Paper Details

Abstracts, authors, affiliations, and disclosures.

Sunday, December 13, 2026
06:00 p.m. - 08:00 p.m.
MCCNO - Hall F
Technology and Procedures, Assays, Imaging, Molecular Testing, Omics Technologies, Pathology, Bioinformatics, Clinical Procedures, Computational Biology, Emerging Technologies, Gene Editing, Profiling, Treatment Considerations, Drug Development, Therapy Sequence, Biological Therapies, Transplantation, (CAR)-T Cell Therapies, Workforce, Adverse Events, Artificial Intelligence (AI), Computer Vision, Deep Learning, Generative AI, Machine Learning, Natural Language Processing, Biological Processes, Pathogenesis, CHIP, Genomics, Hematopoiesis, Immune Mechanism, Metabolism, Microbiome, Molecular Biology, Multi-Systemic Interactions, Clinical Practice (Health Services and Quality), Quality Improvement, Systems-Based Hematology, Diseases, Bone Marrow Failure Syndromes, Lymphoid Malignancies, Lymphoid Leukemias, Lymphomas, Plasma Cell Disorders, Myeloid Malignancies, Acute Myeloid Malignancies, Chronic Myeloid Malignancies, Metabolic Disorders, Measurable Residual Disease, ALL, CLL, LGL, Hodgkin Lymphoma, Research, Fundamental Science, AML, APL, Translational Research, MDS, MPN, Clinical Research, Clinical Trials, Epidemiology, Health Outcomes Research, Real-World Evidence, Registries, Survivorship, Study Population, Human, Adult, Elderly
PAPER

4352 Regularized mixture cure models identify gene expression features associated with short- and long-term overall survival (OS) in adult acute myeloid leukemia (AML) patients (pts)

Justice Ameyi, M.S., Deedra Nicolet, MS2, Michael Walker3, Preem Geeta Torlapati, MD, MPH4, Krzysztof Mrózek, MD, PhD5, Ann-Kathrin Eisfeld, MD6, Kellie J. Archer, Ph.D.7
1 The Ohio State University Comprehensive Cancer Center, Columbus, Ohio, United States
2 The Ohio State University Comprehensive Cancer Center, Columbus, Ohio, United States
3 The Ohio State University Comprehensive Cancer Center, Columbus, Ohio, United States
4 The Ohio State University Comprehensive Cancer Center, Columbus, Ohio, United States
5 The Ohio State University Comprehensive Cancer Center, Columbus, Ohio, United States
6 The Ohio State University Comprehensive Cancer Center, Columbus, Ohio, United States
7 The Ohio State University, Columbus, Ohio, United States

Introduction Recent advances in high-throughput genomic technologies have enabled the measurement of thousands of molecular features, such as gene expression levels, in a single pt sample. However, only a small subset of the measured features is expected to be clinically informative. Mixture cure models (MCMs) provide an attractive framework for studying AML because they distinguish pts who achieve long-term survival (LTS) (incidence) from those who remain at excess risk of AML-related mortality (latency). We developed a high-dimensional excess hazards MCM that incorporates population-expected survival (PES), defining LTS as pts whose survival reached or surpassed that expected for the general population without AML, rather than assuming indefinite survival as in classical MCMs, thereby making LTS modeling more clinically relevant, especially for elderly pts. We applied our model to identify gene expression features associated with LTS and susceptibility in adult AML. Accurate identification of such predictors can deepen our understanding of AML biology, help identify potentially novel targets for drug development, reduce wasted resources, and enhance the reproducibility of studies.

Methods We analyzed 1,064 pre-treatment adult pts with de novo AML aged 18-92 years (y; median=49y), diagnosed between 1987 and 2013, treated with intensive 7+3-based induction chemotherapy on CALGB/Alliance frontline protocols, with complete demographics (age, sex, and year of diagnosis) and RNA-seq data. There were 24% of older adults (≥ 60y), 55% male, and 76% of pts were White, 15% Black, and 4% Hispanic. The primary endpoint was OS, with 27% (n=289) pts censored. Because all pts, especially older adults, may experience mortality from causes unrelated to AML, we adjusted for PES by matching pts on demographics using US lifetables from the Human Mortality Database. The MCM included the top 1000 most variable candidate RNA-seq predictors that were expressed in at least 10% of all pts. To assess generalizability of our model, the data were split into age-stratified (< 60y vs  ≥ 60y) training (80%) and test (20%) sets. Prediction performance in the test set was evaluated using a modified area under the curve (AUC) statistic for the incidence (Asano et al., Pharm Stat. 2014;13:357) and a modified C-statistic for latency.

Results Median follow-up was 9.3y (range, 0.2-21.5) with a clear LTS plateau beyond approximately 5 years, suggesting that a non-negligible subgroup of LTS was present in the data and thus, MCMs are appropriate. The estimated proportion of LTS was 33%. Our regularized MCM identified 74 unique genes, including 39 associated with LTS and 44 with latency, with 9 genes shared by both components (ALDOC, ATL1, CCDC78, EPCAM, MYH11, MYO1B, PDE8A, SEMA7A, and U8). In all, 11 genes have previously been associated with AML, and 3 in other hematologic malignancies (LGMN (incidence), RPL10, and  IL7R), leaving 60 potentially novel genes including PDE8A and CYTH3 which showed the largest incidence and latency effects, respectively. MYH11 is well-established driver in AML that forms CBFB:MYH11 fusion in pts with inv(16)/t(16;16). ALDOC is directly involved in glycolysis (carbohydrate metabolism), ATL1 plays an important role in the cytoplasm, PDE8A hydrolyzes cyclic AMP, a key second messenger, regulating downstream metabolic pathways like glycogen and lipid metabolism, and SEMA7A is also involved in immune cell regulation. Similarly, MEN1, which was unique to the latency, encodes Menin, a therapeutic target in KMT2A-rearranged and NPM1-mutated AML. We identified numerous genes whose expressions were strongly correlated with MEN1, including NUDCD3 (r=0.72), ZC3H7B (r=0.72) and B3GALT6 (r=0.7). The independent test set demonstrated moderate discrimination performance (AUC=0.76; C-statistic=0.65).

Conclusion Using adult AML data, we identified established and potentially novel transcriptomic markers associated with LTS and disease latency in the remaining cohort, while accounting for competing background mortality. These findings support further pathway analysis, validation in external AML data, and the incorporation of clinical, mutational, cytogenetic, and methylation information to improve patient-specific risk prediction. The identification of several metabolism-related genes further supports the growing evidence that metabolic pathways contribute to AML biology and may represent promising therapeutic targets.

Justice Ameyi, M.S.: No relevant conflicts of interest to declare.