Albert G. Tsai

In single-cell datasets, patient labels indicating disease status (e.g., “sick” or “not sick”) are typically available, but individual cell labels indicating which of a patient’s cells are associated with their disease state are generally unknown. To address this, we introduce mixture modeling for multiple-instance learning (MMIL), an expectation-maximization approach that trains cell-level binary classifiers using only patient-level labels. Applied to primary samples from patients with acute leukemia, MMIL accurately separates leukemia from nonleukemia baseline cells, including rare minimal residual disease (MRD) cells; generalizes across tissues and treatment time points; and identifies biologically relevant features with accuracy approaching that of a hematopathologist. MMIL can also incorporate cell labels when they are available, creating a robust framework for leveraging both labeled and unlabeled cells. MMIL provides a flexible modeling framework for cell classification, especially in scenarios with unknown gold-standard cell labels.

Lymphoid specification in human hematopoietic progenitors is not fully understood. To better associate lymphoid identity with protein-level cell features, we conduct a highly multiplexed single-cell proteomic screen on human bone marrow progenitors. This screen identifies terminal deoxynucleotidyl transferase (TdT), a specialized DNA polymerase intrinsic to VDJ recombination, broadly expressed within CD34⁺ progenitors prior to B/T cell emergence. While these TdT⁺ cells coincide with granulocyte-monocyte progenitor (GMP) immunophenotype, their accessible chromatin regions show enrichment for lymphoid-associated transcription factor (TF) motifs. TdT expression on GMPs is inversely related to the SLAM family member CD84. Prospective isolation of CD84^lo GMPs demonstrates robust lymphoid potentials ex vivo, while still retaining significant myeloid differentiation capacity, akin to LMPPs. This multi-omic study identifies human bone marrow lymphoid-primed progenitors, further defining the lympho-myeloid axis in human hematopoiesis.

The Multiplexed Ion Beam Imager (MIBI) is a next-generation mass spectrometry-based microscope which generates 40+ plex images of protein expression in histologic tissues, enabling detailed dissection of cellular phenotypes and histoarchitectural organization. A key bottleneck in its operation occurs when users select the physical locations on the tissue for imaging. As the scale and complexity of MIBI experiments have increased, the manufacturer-provided interface and third party tools have become increasingly unwieldy for imaging large tissue microarrays and tiled tissue areas. Thus, we have developed a web-based graphical interface layer for users to set imaging locations using intuitive mouse gestures such as drag-and-drop, click-and-drag, and polygon draw. Written according to web standards already built into modern web browsers, it requires no installation of external programs, extensions, or patches. Of interest to the hundreds of current MIBI users, our interface simplifies setup of complex experiments and has drastically reduced our setup time.

Lymphoid specification in human hematopoietic progenitors is not fully understood. To better associate lymphoid identity with protein-level cell features, we conducted a highly multiplexed single-cell proteomic screen on human bone marrow progenitors. This screen identified terminal deoxynucleotidyl transferase (TdT), a specialized DNA polymerase intrinsic to VDJ recombination, broadly expressed within CD34+ progenitors prior to B/T cell emergence. While these TdT+ cells coincided with granulocyte-monocyte progenitor (GMP) immunophenotype, their accessible chromatin regions showed enrichment for lymphoid-associated transcription factor (TF) motifs. TdT expression on GMPs was inversely related to the SLAM family member CD84. Prospective isolation of CD84 ^lo GMPs demonstrated robust lymphoid potential ex vivo , while still retaining significant myeloid differentiation capacity, akin to LMPPs. This multi-omic study identifies previously unappreciated lymphoid-primed progenitors, redefining the lympho-myeloid axis in human hematopoiesis.

The diagnosis of lymphomas and leukemias requires hematopathologists to integrate microscopically visible cellular morphology with antibody-identified cell surface molecule expression. To merge these into one high-throughput, highly multiplexed, single-cell assay, we quantify cell morphological features by their underlying, antibody-measurable molecular components, which empowers mass cytometers to ‘see’ like pathologists. When applied to 71 diverse clinical samples, single-cell morphometric profiling reveals robust and distinct patterns of ‘morphometric’ markers for each major cell type. Individually, lamin B1 highlights acute leukemias, lamin A/C helps distinguish normal from neoplastic mature T cells, and VAMP-7 recapitulates light-cytometric side scatter. Combined with machine learning, morphometric markers form intuitive visualizations of normal and neoplastic cellular distribution and differentiation. When recalibrated for myelomonocytic blast enumeration, this approach is superior to flow cytometry and comparable to expert microscopy, bypassing years of specialized training. The contextualization of traditional surface markers on independent morphometric frameworks permits more sensitive and automated diagnosis of complex hematopoietic diseases.

Systemic mastocytosis (SM) is a highly heterogeneous disease with indolent and aggressive forms, with the mechanisms leading to malignant transformation still remaining to be elucidated. Here, we investigated the presence and frequency of genetic variants in 34 SM patients with multilineal KIT D816V mutations. Initial screening was performed by targeted sequencing of 410 genes in DNA extracted from purified bone marrow cells and hair from 12 patients with nonadvanced SM and 8 patients with advanced SM, followed by whole-genome sequencing (WGS) in 4 cases. Somatic mutations were further investigated in another 14 patients with advanced SM. Despite the fact that no common mutation other than KIT D816V was found in WGS analyses, targeted next-generation sequencing identified 67 nonsynonymous genetic variants involving 39 genes. Half of the mutations were somatic (mostly multilineal), whereas the other half were germline variants. The presence of ≥1 multilineal somatic mutation involving genes other than KIT D816V, ≥3 germline variants, and ≥1 multilineal mutation in the SRSF2, ASXL1, RUNX1, and/or EZH2 genes (S/A/R/E genes), in addition to skin lesions, splenomegaly, thrombocytopenia, low hemoglobin levels, and increased alkaline phosphatase and β2-microglobulin serum levels, were associated with a poorer patient outcome. However, the presence of ≥1 multilineal mutation, particularly involving S/A/R/E genes, was the only independent predictor for progression-free survival and overall survival in our cohort. Key Points About half of the mutations other than KIT D816V identified in SM patients were germline in nature. Somatic EZH2 gene mutations provide prognostic information in addition to that of the well-established S/A/R gene panel. Visual Abstract

Chronic eosinophilic leukemia, not otherwise specified can be difficult to distinguish from idiopathic hypereosinophilic syndrome according to the current World Health Organization guideline. To examine whether the morphological features of bone marrow might aid in the differential diagnosis of these two entities, we studied a total of 139 patients with a diagnosis of chronic eosinophilic leukemia, not otherwise specified (n=17) or idiopathic hypereosinophilic syndrome (n=122). As a group, abnormal bone marrow morphological features, resembling myelodysplastic syndromes, myeloproliferative neoplasm or myelodysplastic/myeloproliferative neoplasm, were identified in 40/139 (27%) patients: 16 (94%) of those with chronic eosinophilic leukemia and 24 (20%) of those with hypereosinophilic syndrome. Abnormal bone marrow correlated with older age (PPP=0.041), abnormal platelet count (P=0.002), organomegaly (P=0.008), elevated lactate dehydrogenase concentration (P=0.005), abnormal karyotype (PPversus not reached, PPBCR-ABL1-negative myeloproliferative neoplasm. Among patients who are currently considered to have idiopathic hypereosinophilic syndrome, abnormal bone marrow is a strong indicator of clonal hematopoiesis. Similar to other myeloid neoplasms, bone marrow morphology should be one of the major criteria to distinguish patients with chronic eosinophilic leukemia, not otherwise specified or clonal hypereosinophilic syndrome from those with truly reactive idiopathic hypereosinophilic syndrome.