Acute myeloid leukemia, or AML, is often described as a single disease, but that label conceals a sprawling collection of biologically distinct cancers. Two patients may show similar abnormalities in blood counts and bone marrow while their leukemic cells depend on entirely different molecular circuits. That diversity helps explain why a treatment can produce a dramatic response in one person and fail quickly in another. A study by Chu, Hsiao, Wang and colleagues, published in Nature Cancer, addresses this problem by combining three powerful forms of molecular analysis—genomics, proteomics and metabolomics—to map the biological architecture of AML in greater detail. The goal is not simply to catalog mutations, but to connect genetic instructions to the proteins and chemical reactions that ultimately keep leukemia cells alive.
The distinction is crucial because DNA alone provides only a partial view of cancer. Genomic sequencing can reveal mutations, chromosome alterations and changes in gene regulation, but a mutation does not automatically tell researchers whether a pathway is active, inactive or therapeutically important. Proteomics adds another layer by measuring proteins, the molecular machines that execute most cellular functions. Metabolomics goes further downstream, examining small molecules such as amino acids, lipids, nucleotides and energy-related compounds produced or consumed by cells. Together, these measurements can expose the chain of events linking an alteration in the genome to a functional dependency inside a malignant cell. In practical terms, the approach asks not only what has changed in an AML cell, but what that cell is doing—and what it may be unable to survive without.
The researchers’ integrated strategy is designed to identify molecular subtypes that may be invisible when each data type is analyzed separately. A leukemia sample might carry a mutation that appears modest on its own, while simultaneously displaying a distinctive protein abundance pattern and an unusual metabolic state. When these signals converge, they can reveal a coherent biological program. One group of leukemias may be organized around altered signaling, another around disrupted protein production, and another around exceptional reliance on particular nutrient or energy pathways. Such classifications are potentially more informative than broad diagnostic categories because they are linked to mechanisms that can be tested in the laboratory and, ultimately, targeted with drugs.
Metabolism is especially important in AML because malignant cells must continuously generate energy and raw materials while coping with the demands of rapid growth. Leukemia cells can rewire how they process glucose, amino acids, fatty acids and nucleotides, sometimes using pathways that normal blood-forming cells use only under stress. This flexibility can help cancer cells survive in the bone marrow, where oxygen and nutrients are unevenly distributed. It can also create vulnerabilities. A cell that becomes heavily dependent on one metabolic route may be damaged when that route is blocked, even if healthy cells can switch to alternatives. By placing metabolite measurements alongside protein and genomic data, the study seeks to distinguish general features of aggressive leukemia from specific biochemical dependencies that could become therapy targets.
The same logic applies to proteins involved in signaling and gene regulation. Mutated DNA may activate a kinase cascade, stabilize a transcription factor or interfere with the machinery that controls cell maturation. Yet the therapeutic value of such a change depends on whether the resulting protein network remains active in the patient’s leukemia. Proteogenomic analysis can help answer that question by measuring both protein abundance and, where possible, chemical modifications such as phosphorylation. Phosphorylation acts as a molecular switch in many signaling pathways, turning proteins on or off or changing where they operate in the cell. Detecting abnormal phosphorylation patterns can therefore reveal active signaling circuits that sequencing alone might miss, as well as identify nodes that could be blocked pharmacologically.
A major promise of the work is the identification of therapy targets associated with particular AML subtypes. Target discovery in cancer is often difficult because a molecule may be altered without being essential, or because blocking it may harm normal tissues more than tumor cells. Integrated profiling can prioritize candidates by showing that a protein or pathway is not merely present, but connected to a broader network of genomic and metabolic abnormalities. Researchers can then test whether disrupting that candidate selectively impairs leukemia cells. The resulting targets may include enzymes, signaling proteins, regulators of protein synthesis or metabolic components. The study’s significance lies in this systems-level connection: it attempts to move from molecular description to a rational explanation of why a specific leukemia might respond to a specific therapeutic strategy.
This framework could also help explain treatment resistance, one of AML’s most persistent clinical problems. Therapy may eliminate a large fraction of leukemia cells while leaving behind a smaller population with a different metabolic program or a more resilient signaling network. Those surviving cells can expand and drive relapse. If resistant cells are distinguishable by their proteins or metabolites, clinicians may eventually be able to monitor them more directly than by relying on mutation profiles alone. A genomic test might indicate that the cancer has not changed, while proteomic or metabolic measurements could reveal that the cells have shifted into a drug-tolerant state. Such information could support combination treatments designed to attack both the original driver and the adaptive pathway that allows residual disease to persist.
The study also illustrates why precision oncology increasingly depends on combining technologies rather than searching for a single universal biomarker. Each molecular layer has limitations. Genomic data can be comprehensive but mechanistically ambiguous; proteomic data can reflect cellular activity but vary with sample handling and cell composition; metabolomic data can capture rapid physiological changes but may be especially sensitive to environmental conditions. Integration requires careful computational analysis, normalization and biological interpretation. It also demands attention to the fact that a bone-marrow sample contains more than leukemia cells, including immune cells, stromal cells and normal blood precursors. Distinguishing tumor-intrinsic signals from signals produced by the surrounding microenvironment is essential before any proposed subtype or target can be translated into clinical use.
For patients, the immediate impact of this research is likely to be indirect rather than a new treatment available overnight. Molecular subtypes and candidate targets must be validated across independent patient groups, tested in leukemia models and evaluated in clinical trials. Researchers must determine whether a proposed biomarker can be measured reliably in hospitals, whether it predicts response better than existing tests and whether targeting the associated pathway is safe. Even so, the study represents an important shift in how AML can be understood. Instead of treating the disease as a list of mutations, it presents leukemia as an interconnected system in which genetic changes, protein activity and metabolism reinforce one another. That systems view could help researchers identify vulnerabilities that remain hidden when cancer is examined through only one molecular lens.
The broader message is that the future of AML medicine may depend on measuring function as well as identity. A tumor’s DNA records its history, but its proteins and metabolites reveal how that history is being enacted in real time. By bringing these layers together, the researchers offer a route toward more biologically precise disease classification and a more disciplined way to nominate therapeutic targets. The approach will not eliminate AML’s complexity, but it may make that complexity useful: distinct molecular states could become markers for diagnosis, guides for treatment selection and warning signs for relapse. As integrated profiling becomes faster and more accessible, the most important question in leukemia care may shift from “Which mutation does this cancer carry?” to “Which molecular program is sustaining it—and how can that program be interrupted?”
Subject of Research: Integrated molecular profiling of acute myeloid leukemia to identify molecular subtypes and therapy targets.
Article Title: Integrated proteogenomic and metabolomic profiling of acute myeloid leukemias to identify molecular subtypes and associated therapy targets.
Article References: Chu, SC.A., Hsiao, Y., Wang, C. et al. Integrated proteogenomic and metabolomic profiling of acute myeloid leukemias to identify molecular subtypes and associated therapy targets. Nature Cancer 7, 993–1015 (2026). https://doi.org/10.1038/s43018-026-01175-6
Image Credits: AI Generated
DOI: 10.1038/s43018-026-01175-6
Keywords: acute myeloid leukemia, AML, proteogenomics, metabolomics, cancer metabolism, molecular subtypes, precision oncology, therapy targets, leukemia biology, cancer biomarkers
Tags: AML molecular subtypesbiochemical pathways in leukemiacancer molecular circuitrygenetic mutations in AMLgenomics in leukemiametabolomics for AMLmolecular heterogeneity in cancermulti-layered cancer diagnosticsmultiomic cancer profilingpersonalized leukemia therapyproteomics in cancer treatmenttargeted treatment in AML
