High-content stimulated Raman pathology imaging and transcriptomics reveal leukemia subtype-specific lipid metabolic heterogeneity.

INTRODUCTION: Leukemia, a heterogeneous group of hematological malignancies, is characterized by abnormal proliferation of immature hematopoietic cells. Current diagnostics primarily rely on morphological evaluation for subtype classification, methods that are subjective and labor-intensive. To overcome these limitations, a High-Content Spectral Raman Pathology Imaging platform (H-SRPI) was introduced.

METHODS: H-SRPI imaging enables profiling of proteins, nucleic acids, saturated and unsaturated lipids in leukemia. We analyzed leukemia samples from 12 patients with six distinct subtypes, alongside CD34+, B, T cells, monocytes and granulocytes from 3 healthy donors, by conducting high spatial resolution Raman imaging on 324 cells. We developed a single-cell phenotyping algorithm (incorporating cellular area, protein, nucleic acid, saturated and unsaturated lipid content) to distinguish leukemia subtypes. Finally, using H-SRPI and RNA-seq transcriptomics, we uncovered the critical role of lipid composition in leukemia cells across subtype classifications.

RESULTS: The single-cell phenotyping algorithm to distinguish leukemia subtypes, achieving 88.21% accuracy. H-SRPI and RNA-seq transcriptomes revealed elevated saturated and unsaturated lipid levels in acute myeloid leukemia (AML); AML-M3 favored lipid desaturation, whereas AML-M5 upregulated saturated lipid synthesis and elongation. ALL had weaker lipid metabolism characteristics than AML.

CONCLUSIONS: Our study establishes H-SRPI as a label-free tool for metabolic profiling, enabling precise leukemia subclassification and revealing lipid metabolic heterogeneity as a potential therapeutic target.