EPCO-34. DECIPHERING THE LONGITUDINAL TRAJECTORIES OF GLIOBLASTOMA BY INTEGRATIVE SINGLE-CELL GENOMICS

Abstract

Abstract The evolution of cellular heterogeneity in IDH-wildtype glioblastoma (GBM) after standard-of-care therapy remains poorly understood. To address it, we assembled a longitudinal cohort of 121 primary and recurrent GBM specimens from 59 patients, with extensive clinical annotations, and profiled it by single-nucleus RNA-sequencing and bulk tumor DNA sequencing. In most cases, longitudinal samples diverged in their composition of cell types and cell states. However, almost all theoretical trajectories were observed in our cohort such that the overall distribution of cell types and cell states was comparable between primary and recurrent samples. The most consistent longitudinal effect (66% of patients) was a lower malignant cell fraction at recurrence and a reciprocal increase in proportions of glio-neuronal TME cell types; in some cases, this was further accompanied by a coordinated shift of malignant cells towards neuronal-like states. MGMT methylation and radiation-related small deletion phenotypes were linked to particular trajectories, with depletion of mesenchymal-like cells and enrichment of hypoxia-related malignant cells, respectively. Importantly, changes in malignant states were also associated with specific changes in TME composition. In summary, our analysis highlights diverse longitudinal GBM trajectories that are shaped by treatment response and TME interactions.