e14027 Background: Primary glioblastoma (GBM) tumor cells exhibit plasticity and can adopt different cellular states that exhibit dynamic transitions between states. This could be one of the mechanisms for treatment escape and leads to poor prognosis in GBM. In this regard, we sought to uncover transcriptional regulatory mechanisms responsible for this phenotypic plasticity in GBM. Methods: Standard of care (SOC) treatment (radiation and temozolomide) modulates transcriptional regulatory networks (TRNs) of the tumor cells and drives treatment resistance. Since TRNs could be key mechanistic drivers for state transitions, integrated single-cell RNA sequencing (scRNAseq) data of primary and relapse patient samples (11) were analyzed with scMINER. Our method uses a systems biology framework to identify cell states (St_1 to St_n) in terms of regulatory networks and their activities. Clustering of co-regulated gene modules (regulons), enabled the identification of transcriptional programs and regulatory mechanisms that specify tumor cell states. Results: Activity patterns of 38 transcriptional programs clustered tumor cells from the patients into 21 states. Preliminarily, we observed common regulatory network cell states (St_1, St_2, St_5, St_6) across patient tumors. Additionally, primary tumor cells of a patient (UW14) transitioned from initial cell state(s) observed across multiple patients (e.g., St_1) to a recurrent state (e.g., St_9) through a trajectory driven by transcription factors (TFs). Herein, scMINER regulatory network inference method also defines master TFs (TFs that regulate the majority of the other TFs in the network) that were used to characterize initial (St_1) or recurrent (St_9) cell states. Moreover, certain transcription factors (22) and their co-expressed target genes (regulons) also exhibit variation in expression across the pseudotime trajectory (from St_1 to St_9) including masterTFs such as SOX9 and MECOM. Conclusions: Therefore, we observed that shared cell states may exist across patients (primary and relapse) although in different proportions. Our analysis suggests a de-differentiation trajectory from neuronal cells to developing GBM stem cells and predicts transcription factors that could be playing an important role therein. Moreover, similar transcriptional regulatory mechanisms may promote transitions between primary and recurrent tumor cell states that exhibit phenotypic plasticity.