Diagnosing blood cancers like leukemias and lymphomas requires two fundamentally different types of information. First, a pathologist examines cells under a microscope to assess their visible structure - the shape of the nucleus, the texture of the chromatin, the appearance of the cytoplasm, and the size of various cellular compartments. Second, a laboratory uses flow cytometry to measure which protein markers are present on the cell surface, revealing the cell's lineage and any abnormalities.
The problem is that these two information types currently come from separate, incompatible tests that must be mentally assembled by an expert hematopathologist. Neither test alone is sufficient for many diagnoses, and combining them is a heavily manual, time-intensive process that requires years of specialized training. Furthermore, current flow cytometers can measure at most about 10 markers simultaneously, requiring complex multi-tube panels that take days to analyze.
A specific challenge is that cancer cells often abnormally express or lack the surface markers that are supposed to identify their cell type. This study's dataset includes a CD20+ NK/T cell lymphoma, a CD3- T cell lymphoma, three CD19+ myeloid leukemias, and a CD56+ B cell leukemia - all situations where surface marker identity conflicts with true lineage and could mislead a diagnosis based on markers alone.
Additionally, immunotherapies like anti-CD19 CAR-T cell therapy can cause leukemia cells to down-regulate the very surface markers used to track them, making standard monitoring methods unreliable in the post-treatment setting. There is therefore a need for a marker-independent way to assess cell type and identify cancer cells that can complement or substitute for surface antibody-based approaches.
The researchers developed a concept called single-cell morphometric profiling: instead of looking at cell structure visually, they measure the molecular components that give rise to those visible structures. For example, rather than noting whether a cell has a round or irregularly shaped nucleus under the microscope, they measure the protein lamin B1, whose abundance determines nuclear mechanical properties and shape.
Eleven morphometric targets (called scatterbodies) were identified, each acting as a molecular surrogate for a specific morphological feature visible to pathologists: VAMP-7 for granularity (cytoplasmic side scatter), lamin B1 and lamin A/C for chromatin quality and nuclear shape, rRNA for nucleolar size and cytoplasmic basophilia, beta-actin for cytoskeletal content, WGA lectin for cell surface area, serpin B1, lysozyme, MPO, lactoferrin, and HP1beta for granule composition and cell maturity.
These scatterbodies were measured using mass cytometry (CyTOF), which can quantify over 40 antibody markers simultaneously on millions of individual cells using metal isotope-labeled probes. This eliminates the need to split a diagnostic panel across multiple tubes and enables all measurements to be made from a single sample in a single experiment, completable within the standard 24-hour clinical turnaround time.
The study tested this approach across 71 diverse patient samples spanning a wide range of hematologic diagnoses, including 26 acute myeloid leukemias (AML), 5 B-lymphoblastic leukemias (B-ALL), 3 T-lymphoblastic leukemias (T-ALL), 3 acute leukemias of ambiguous lineage (MPAL), 9 mature B cell lymphomas, 3 myelomas, 5 mature T cell lymphomas, and 10 non-acute myeloid neoplasms. This diversity stressed the universality of the approach.
In healthy human bone marrow, the major cell types - granulocytes, lymphocytes, monocytes, erythroid precursors, and hematopoietic blasts - showed distinctive and highly consistent morphometric profiles. Granule markers (VAMP-7, serpin B1, lysozyme, MPO, lactoferrin) separated granulocytes and monocytes from other cell types. Lamin proteins distinguished blasts from mature cells. rRNA levels tracked cell immaturity and nucleolar prominence.
Across 54 clinical patient samples spanning diverse cancer diagnoses, normal background cells showed consistent morphometric profiles despite variable storage times and despite the samples containing mixtures of malignant and normal cells. This robustness is essential for clinical utility: a diagnostic test that gives different results depending on how long the sample sat in the refrigerator before processing cannot be reliably used.
Pearson correlation analysis of morphometric profiles across patients confirmed that the same cell type clusters together regardless of which patient it comes from - even when the cancer cells in that sample had highly abnormal surface marker expression. This confirms that morphometric profiling provides a patient-independent, lineage-based classification that is more stable than surface marker-based classification.
Unsupervised t-SNE dimensionality reduction, fed only the 11 morphometric targets plus CD45, successfully separated all the major morphologically distinct cell populations in healthy bone marrow. Importantly, it did NOT separate B cells from T/NK cells - a correct outcome, since these cell types are morphologically indistinguishable and their separation requires surface immunophenotyping rather than morphometry.
Lamin B1 as a universal blast marker: All leukemic blast cells - regardless of lineage (myeloid, lymphoid, or ambiguous) - consistently expressed high levels of lamin B1. This held even for diagnostically challenging leukemias that lack the typical blast surface markers CD34 and CD117. In a panel of 39 samples with blast populations of over 20 cells, median lamin B1 in blasts was significantly higher than every other hematopoietic population, making it the most consistent blast identifier described to date.
Lamin A/C for T cell lymphoma: Mature T cell lymphomas expressed uniformly high and tightly distributed lamin A/C, reflecting their clonal origin. Normal T cells show a wide spectrum of lamin A/C levels, but clonal neoplastic T cells - having all descended from a single transformed ancestor - expressed a narrow, homogeneous range. This provided a new proteomic marker for detecting T cell clonality, a historically difficult diagnostic challenge in flow cytometry.
VAMP-7 as a CyTOF side scatter substitute: Flow cytometry's CD45 vs. side scatter (SSC) plot is the foundation of standard hematopathology workflow, organizing cells into characteristic regions that guide subsequent analysis. But mass cytometry cannot measure light-based side scatter. VAMP-7 levels were found to be strikingly similar to SSC patterns, providing a molecular surrogate that allows the familiar gating strategy to be replicated in CyTOF-based diagnostics.
rRNA for tracking differentiation: rRNA levels decreased linearly as cells matured from blasts to differentiated granulocytes, erythroid cells, and lymphocytes, providing a single-antigen 'pseudo-time' axis tracking developmental stage across multiple lineages simultaneously - a capability not previously available in cytometric diagnosis.
One of the most clinically important - and most difficult - tasks in leukemia diagnosis is blast enumeration: counting what percentage of bone marrow cells are malignant blast cells. This percentage determines whether acute leukemia is diagnosed (threshold: 20% blasts) and guides treatment intensity. Currently, this requires manual microscopic counting of at least 200-500 cells by an expert, a tedious and subjective process with known inter-observer variability.
Using the morphometric data combined with machine learning (linear discriminant analysis for dimensionality reduction and automated cell classification), the researchers developed an automated blast enumeration method. This was benchmarked against counts performed by expert hematopathologists on the same samples.
The morphometric machine learning approach achieved blast enumeration accuracy comparable to expert pathologists, demonstrating that the combination of molecular morphometry and computation can automate one of the most skill-dependent tasks in hematopathology. This is significant because it offers a path toward consistent, automated blast counting that does not depend on the availability of an expert microscopist.
Integrating morphometric markers with deep surface immunophenotyping - simultaneously measuring both the scatterbodies and standard diagnostic CD surface markers - produced a versatile platform compatible with traditional cytometry gating strategies while offering dramatically expanded diagnostic capability and the potential for systematic, automated analysis.
Two clinical vignettes in the paper illustrate the real diagnostic value of morphometric profiling. In one, a patient with T-cell lymphoma that was initially challenging to distinguish from T-lymphoblastic leukemia was resolved by lamin A/C measurement - the mature lymphoma cells had uniformly high lamin A/C (reflecting clonality), while T-ALL blasts did not. This distinction, which is difficult and sometimes impossible with standard flow cytometry, was made clearly and quantitatively.
In another vignette, a post-CAR-T therapy patient was assessed for residual leukemia. Because CAR-T therapy targets CD19, any remaining leukemia cells had down-regulated CD19 - making standard immunophenotyping-based detection unreliable. Lamin B1 and rRNA measurement identified the residual blast population independently of their surface marker expression, demonstrating the power of morphometric approaches for monitoring treatment response in immunotherapy-treated patients.
The CyTOF platform can process samples within standard clinical turnaround times and uses sample multiplexing (barcoding multiple patients into one run) to reduce costs and standardize workflows. These practical features address key barriers to clinical adoption that have limited other mass cytometry applications.
By eliminating the need for multi-tube flow cytometry panels and replacing manual microscopy interpretation with quantitative morphometric profiling, this platform has the potential to reduce diagnostic variability, speed up diagnosis, and improve standardization across hematopathology laboratories worldwide.
Single-cell morphometric profiling represents a fundamental advance in hematopathology diagnostics: it merges two previously orthogonal diagnostic modalities - light microscopy morphology and flow cytometric immunophenotyping - into a single, quantitative, multiplexed, single-cell assay.
The 11 morphometric targets identified here provide a practical molecular representation of the features that pathologists have used for over a century to classify blood cells under a microscope. By making these features quantitative, reproducible, and measurable alongside dozens of other markers simultaneously, the approach moves hematopathology toward a more objective and scalable future.
The versatility of the platform is a particular strength: it is compatible with existing clinical cytometry workflows (using VAMP-7 to replicate SSC gating), adds novel capabilities (lamin B1 for lineage-agnostic blast identification, lamin A/C for T cell clonality), and enables automation of difficult quantitative tasks (blast enumeration). These features make clinical adoption more feasible than approaches requiring wholesale workflow replacement.
As the complexity of hematology diagnoses continues to grow - driven by increasingly targeted therapies, immunotherapy-induced marker changes, and complex genetic subclassifications - tools that can provide more information from a single test in an automated, standardized way will be essential. Single-cell morphometric profiling points the way toward that future.