What defines a leukemia’s course? Genotype or cellular state.What defines a leukemia’s course?Cellular stateCellular stateGenotypeGenotype

A longitudinal map of AML cell states

Two patients with AML can carry the same genetic mutations yet have different disease courses. Through a longitudinal, multiomic study of AML, we aim to understand which molecular features are shared across genetic backgrounds, which change through treatment, and which remain essential to leukemic identity.

A leukemia diagnosis often begins with a genetic description: mutations, chromosome abnormalities and patterns of gene activity that classify AML. Yet patients whose leukemia shares a molecular diagnosis may respond differently, and even a successful treatment can leave behind cells capable of driving relapse. What is missing from the genetic description?

We think part of the answer lies in the cell state. Genes are switched on or off, regulatory DNA becomes more or less accessible, and proteins appear or disappear from the cell surface. Together these properties influence whether a cell differentiates, survives, adapts to stress or persists after therapy.

The centerpiece of our research is a proposed longitudinal, single-cell multiomic atlas of AML, examining leukemia from diagnosis through treatment and relapse. We will integrate targeted DNA sequencing from routine clinical testing with complementary single-cell approaches, including RNA-seq to profile gene expression, ATAC-seq to examine chromatin accessibility, and CITE-seq to characterize cell-surface proteins. By computationally integrating these molecular profiles with clinical, pathologic, and treatment data, we aim to identify features shared across genetic backgrounds, understand how they change or persist through treatment, and explore their relationship to patient outcomes.

Our goal is to determine whether genetically distinct AMLs with similar clinical behavior share identifiable molecular signatures, which molecular features best predict disease course and treatment response, and which change or persist through therapy. While this atlas may reveal common patterns, shared signatures do not necessarily imply shared mechanisms. The most reproducible findings will guide functional studies to validate their biological significance, their relationship to leukemic stemness and persistence, and their potential as biomarkers or therapeutic targets.

Shared KMT2A-Associated Regulatory Dependencies in AML

Different genetic forms of AML can rely on some of the same molecular pathways to survive. One example is the KMT2A–HOX/MEIS1 regulatory axis, which helps sustain leukemia across several AML subtypes. The development of menin inhibitors has shown how targeting a shared dependency can lead to effective therapies, although not all leukemias involving this pathway respond equally.

We are building on this understanding through a collaboration investigating how different AML subtypes maintain HOX/MEIS1 activity and whether other parts of this pathway can be targeted. Of particular interest is the interaction between KMT2A and WDR5, which may offer an alternative therapeutic approach in leukemias such as SETBP1-driven AML that are less responsive to menin inhibition. This work aims to understand not only what genetically distinct leukemias share, but also why they may respond differently to therapies targeting the same regulatory network.

Myelodysplastic Neoplasms: From Population Health to Disease Progression

Our research in myelodysplastic neoplasms (MDS) examines the broader health burden of these diseases, differences in patient outcomes, and the biological changes underlying progression.

Using national mortality and epidemiologic data, we are investigating medical conditions associated with MDS, geographic patterns, and differences in outcomes. We are also examining our local patient population to identify opportunities to improve care and support the development of a dedicated myeloid malignancies program.

In parallel, we are establishing a longitudinal collection of bone marrow and blood samples from patients with MDS and related myeloid disorders. By studying molecular changes through treatment and disease progression, including transformation to AML, we hope to improve our understanding of disease evolution and explore whether multiomic profiling can refine risk prediction beyond existing tools such as IPSS-M.

Artificial Intelligence in the research process

Kupalab is building a research workflow that incorporates Artificial Intelligence across data management, instrumentation and analysis. We are exploring how these tools can support careful, reproducible leukemia research, testing their limits and sharing what we learn as the lab develops.

Follow the Artificial Intelligence lab build at kupalab.ai.