LSC - 2024 - Deciphering the spatial heterogeneity of interstitial lung disease by integrative radiomics and single-nucleus transcriptomics

Spatial heterogeneity in disease patterns is a key hallmark of interstitial lung diseases (ILDs), such as systemic sclerosis (SSc)-ILD. High-dimensional image analysis (radiomics) of computed tomography (CT) scans offers quantitative insights into organ-scale pathophysiology, creating digital disease fingerprints. Here, we aimed to integrate spatially resolved radiomic profiles extracted from CT scans with matched molecular data to decipher the cellular programs underlying the spatial heterogeneity in ILDs and how they are reflected in radiomic phenotypes. We therefore systematically dissected a whole right lung of a SSc-ILD patient into 65 samples for virtual 3D alignment and association of regional radiomic profiles with CT co-localized histopathology and single-nucleus RNA sequencing (snRNAseq). Radiomic profiles significantly differed with spatial location revealing a typical apical to basal gradient characteristic for the spatial heterogeneity in SSc-ILD. To assess whether those radiomic differences translate to the molecular level, we performed snRNAseq, with 14 distal lung regions from apex to base. Analyzing 124,000 cells from more than 50 cell types we revealed region-specific alterations in cellular composition, gene expression and cell-cell communication networks. Integration of radiomics and transcriptomics data showed spatial radiomic modules with distinct cellular profiles, including an apical T cell exhaustion and a basal macrophage-driven fibrogenic program. In conclusion, we provide an in-depth imaging and molecular phenotyping of the spatial heterogeneity of SSc-ILD, highlighting the potential of radiomics as virtual organ-scale biopsies.