Abstract Despite growing insights into the heterogeneous cellular composition of brain tumors, the interplay between this heterogeneity and tumor progression remains largely unexplored in pediatric patients. Here, we present an in-depth longitudinal single-nucleus and spatial transcriptomic characterization of an extensive cohort of pediatric high-grade gliomas (pHGG) in children, adolescents and young adult patients. Our cohort encompasses 76 paired pHGG samples, including diffuse midline glioma, IDH-mutant or IDH-wild type high-grade glioma, and infant-type hemispheric glioma samples, that were collected throughout tumor progression, representing the most comprehensive pediatric glioma cohort profiled to date. By resolving transcriptional intra- and inter-tumor heterogeneity, we identified common and subtype-specific patterns of disease plasticity, characterized by distinct cellular programs and developmental hierarchies across pHGG subtypes. Longitudinally, progressive pHGGs exhibited notable similarity in gene expression profiles to their primary counterparts. Temporal transcriptional changes included an increase in the proportion of the neural progenitor cell-like state (NPC-like) and, on a broader level, upregulated expression of pathways related to the extracellular matrix cells in recurrent IDH-wildtype and IDH-mutant high-grade gliomas. In contrast, infant-type hemispheric gliomas showed upregulated expression of genes related to antigen presentation and downregulated pathways associated with vasculature. Spatial transcriptomic analysis of 77 tumor sections derived from 35 pHGG samples further revealed distinct global architectural features and common as well as subtype-specific cellular niches across the four included pHGG subtypes. Overall, this study dissects the transcriptional composition and spatial architecture across different pHGG subtypes and provides an unprecedented landscape of cellular and spatial intratumoral heterogeneity across molecular subtypes and during disease progression, laying the groundwork for the identification of potential novel therapeutic targets.
Danielli et al. (2025) studied this question.