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

Paper Details

Abstracts, authors, affiliations, and disclosures.

Monday, December 14, 2026
10:30 a.m. - 12:00 p.m.
MCCNO - Hall C
Technology and Procedures, Omics Technologies, Bioinformatics, Computational Biology, Biological Processes, Pathogenesis, Immune Mechanism, Molecular Biology, Diseases, Myeloid Malignancies, Acute Myeloid Malignancies, Research, AML, Translational Research, Study Population, Human, Adult, Pediatric, Young Adult
PAPER

780 Single-cell transcriptomic and epigenomic longitudinal profiling reveals treatment-induced divergent stem cell states in TP53-mutant AML

Yoonkyu Lee, PhD, Yuki Nishida, MD, PhD2, Edward Ayoub, PhD3, Vakul Mohanty4, Li Li, MD, PhD5, Muharrem Muftuoglu, MD6, David Lehotzky, PhD7, Jake S Leighton, PhD8, Mahesh Basyal, BSc9, Andrea Bedoy, MS10, Torsten Haferlach, MD, PhD11, Koichi Takahashi, MD,PhD12, Nicholas Navin, PhD13, Naval Daver, MD14, Courtney DiNardo, MD, MSCE15, Ghayas C. Issa, MD16, Gaddy Getz, PhD17, Ken Chen, PhD18, Michael Andreeff, MD, PhD19
1 Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
2 Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
3 Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
4 Department of Bioinformatics and Computational Biology, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
5 Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
6 Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
7 Broad Institute of Massachusetts Institute of Technology (MIT) and Harvard, Cambridge, Massachusetts, United States
8 Department of Genetics, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
9 Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
10 Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
11 MLL Munich Leukemia Laboratory, Munich, Germany
12 Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
13 Department of Systems Biology, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
14 Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
15 Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
16 Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
17 Broad Institute of Massachusetts Institute of Technology (MIT) and Harvard, Cambridge, Massachusetts, United States
18 Department of Bioinformatics and Computational Biology, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
19 Section of Molecular Hematology and Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States

Introduction:

TP53-mutant (TP53mut) acute myeloid leukemias (AML) respond poorly to both intensive chemotherapy (Chemo) and the combination of hypomethylating agents with venetoclax (HMA+VEN). The mechanism driving leukemia stem cell (LSC) persistence under each treatment remains unclear, and investigating it could identify novel targets in TP53mut AML. We hypothesized that these two therapies exert distinct selective pressures on LSCs, driving regimen-specific, transcriptional and epigenetic programs contributing to the persistence of LSCs. We analyzed longitudinal single-cell RNA and ATAC sequencing data (scRNA-seq and scATAC-seq) across Chemo- and HMA+VEN-treated TP53mut AML patients within a shared analytical framework to define how LSCs persist under each regimen, and whether the resulting post-therapy disease states reveal distinct therapeutic vulnerabilities.

Methods:

We profiled 70 longitudinal scRNA-seq samples (175,308 cells) from 34 AML patients (27 TP53mut, 7 TP53wt), including 27 TP53mut patients with paired pre-/post-treatment samples (12 chemotherapy, 11 HMA+VEN, 4 other). We also integrated 4 paired TP53mut AML patient samples treated with HMA+VEN (Waclawiczek, Cell Stem Cell 2026). We further profiled scATAC-seq data from 15 TP53mut AML patients treated with HMA+VEN+Magrolimab (HVM; 26 samples, 133,511 cells).

Results:

AML cells in scRNA-seq were annotated against a normal bone marrow reference (Zeng, Blood Cancer Discovery 2025) with malignant cells identified via copy-number variation inference (Numbat and CopyKAT). We used scVI to integrate cells across patients while preserving treatment-associated biological signal. Clustering analysis of the myeloid compartment identified 19 transcriptionally distinct subgroups aligning with normal reference cell types. We also identified LSC clusters enriched for an experimentally validated LSC gene expression signature (Ng, Nature 2016).

The two regimens diverged in LSC abundance and transcriptional states. HMA+VEN reduced LSC clusters and expanded an erythroid-differentiated cluster, whereas Chemo resulted in no significant shift in LSC cluster composition. This showed that only HMA+VEN achieves a measurable reduction of the refractory LSCs. Optimal transport (Moscot) mapping of pre- and post-treatment cell states showed diagnostic LSC clusters retained high potential to develop into pre-existing monocytic and erythroid cell clusters regardless of therapy. This showed a preserved differentiation hierarchy rather than therapy-specific new downstream cell identities.

Using differentially expressed gene and gene set enrichment analyses, we found chemotherapy-persistent LSCs maintained downregulation of DNA replication/cell-cycle machinery, cholesterol/sterol biosynthesis, and NF-κB/TNF-inflammatory signaling. In contrast, HMA+VEN-persistent LSCs showed significant enrichment of proliferation, OXPHOS/mitochondrial translation, DNA repair, and interferon signaling. This suggests that Chemo-persistent LSCs enter a dormant state, while HMA+VEN-persistent LSCs exit quiescence and rely on active energy metabolism. 

With further investigation of HVM arm effects using scATAC-seq, persistent LSCs showed two candidate resistance programs. First, HSC/MPP-like LSC clusters gained GATA-family motif accessibility by scATAC-seq with concordant upregulation of GATA1 and ANK1 across scRNA-seq and scATAC-seq data, suggesting erythroid lineage program bias within the LSC compartment. Second, GMP-like LSC clusters gained AP-1 (JUN, FOS, ATF3)/NF-κB motif accessibility, suggesting an inflammatory/stress-responsive chromatin state in this compartment. Correlating scRNA-seq and scATAC-seq data is in progress. The cross-modality concordance of these programs strengthens the confidence of these resistance programs.

Conclusion:

Persistent LSCs under Chemo and HMA+VEN retain a shared differentiation hierarchy but adopt distinct transcriptional and metabolic states, with HMA+VEN achieving greater reduction of the LSC-enriched compartment. HMA+VEN-persistent LSCs show a proliferative, DNA-repair-active, OXPHOS-dependent, erythroid-biased phenotype, in contrast with the low-proliferation, reduced metabolic state of Chemo-persistent LSCs. We identified erythroid lineage bias, alongside a candidate GMP-compartment inflammatory /stress chromatin program, as features of HMA+VEN-persistent LSCs and potential therapeutic vulnerabilities.

YL is a TRIUMPH Fellow in the CPRIT Training Program (RP210028).

Yoonkyu Lee, PhD: No relevant conflicts of interest to declare.