Multi-Dimensional Reduction Clustering of Complex Carbohydrates Reveal Tissue Metabolism, Heterogeneity and Histopathology

Abstract

Formalin-fixed paraffin-embedded (FFPE) human tissues represent the world's largest collection of accessible clinical specimens with matched, well-annotated clinical course for disease progression. Currently, FFPE sections are limited to low throughput histo- and immunological assessments. Extracting largescale molecular information remains a major technological barrier to uncover the vast potential within FFPE specimens for translation and clinical research. Two critical but understudied facets of glucose metabolism are anabolic pathways for glycogen and N-linked glycan biosynthesis. Together, these complex carbohydrates represent bioenergetics, protein-structure function, and tissue architecture in human biology. Herein, we report the high-dimensional Metabolomics-Assisted Digital pathology Imaging (Madi) workflow that combines matrix-assisted laser desorption ionization mass spectrometry imaging (MALDI-MSI) with machine learning for the comprehensive assessment of tissue heterogeneity, histopathology, and metabolism in human FFPE sections. In normal human tissue sections, Madi accurately identifies anatomical regions within liver and the brain. In human lung diseases, Madi accurately predicts major lung pathologies such as honeycomb change, late-stage fibrosis, diffuse alveolar damage (DAD), and acute fibrinous and organizing pneumonia (AFOP) from idiopathic pulmonary fibrosis (IPF) and COVID-19 pneumonia specimens with precision. In depth pathway enrichment analyses reveal unique metabolic pathways are associated with distinct pathological regions, which highlight aberrant complex carbohydrate metabolism as a previously unknown molecular event associated with disease progression that could hold key to future therapeutic interventions.