Single-cell spatial mapping reveals dynamic bone marrow microarchitectural alterations and enhances clinical diagnostics in MDS.
Overview
abstract
Myelodysplastic neoplasms (MDS) are genetically diverse hematopoietic cancers characterized by ineffective blood cell production, peripheral cytopenias, and an increased risk of acute myeloid leukemia. Diagnosis traditionally requires subjective histomorphologic assessment of a bone marrow biopsy sample. The potential biological and/or clinical relevance of subtle microarchitectural changes, unrecognizable using conventional methods, remains unknown. Here, we applied a recently developed AI-driven, whole slide imaging-based single-cell spatial proteomic profiling method to 77 annotated MDS and precursor state bone marrow tissue samples, including longitudinal cases. Compared to age-matched controls, MDS tissues showed significant changes in progenitor cell frequencies, morphologies of erythroid precursors and megakaryocytes, HSPC displacement from vasculature, abnormal progenitor cell clustering, and disrupted erythroid islands. Some alterations correlated more closely with specific mutations (e.g., SF3B1, TP53 ) than clinical risk scores (IPSS-M). Using all extracted tissue features, we developed a composite spatially informed "MDS severity score", which aligned with clinical and genetic parameters across serial samples. This work uncovers previously unrecognized, genotype-linked microarchitectural alterations in MDS, the measurement of which may enhance existing diagnostic and disease monitoring strategies.
