Abstract A056: Growth rates of patient-derived tumor organoids reflect heterogeneity in tumor plasticity among glioblastoma patients

Abstract Glioblastoma exhibits extensive intra-tumoral heterogeneity and plasticity, which significantly compromise treatment outcomes. Patient-derived glioblastoma organoids (GBOs) are emerging as a novel preclinical model for glioblastoma. Here, we analyzed the extent to which these GBOs replicate patients’ tumour plasticity and clinical behaviors. We found that GBOs derived from 32 patients exhibit heterogeneous growth rates in culture, as determined by analysis of GBO phase-contrast microscopy images over time. Kaplan-Meier survival analysis showed that GBO growth rates correlated with patients' overall survival, with faster-growing GBOs originating from resected tumor samples of patients who presented shorter overall survival (n=31 patients with available survival data, p=0.006, HR=3.1). Bulk RNA-seq analysis of resected tumor tissues (n=6) that generated fast-growing GBOs revealed lower expression of cell-cycle genes compared to those that generated slow-growing GBOs. This characteristic, however, was reversed when analyzing GBO RNA-seq data (n=6), suggesting a higher capacity for adaptation (plasticity) in glioblastoma tumours in the fast group. Supporting this concept, results from a comparative analysis of tumours and matched GBOs using bulk RNA-seq (n=6) and scRNA-seq (n=7) revealed greater transcriptional changes from tumor to GBO in the fast group (number of differentially expressed genes (nDEG) fast group= 1,848; nDEG slow=706), characterized by an increased number of cells in neural precursor (NPC)-like and oligodendrocyte precursor (OPC)-like tumour cell states and a reduction of cells in mesenchymal-like (MES) tumour cell states. Analysis of differentially expressed genes between fast- and slow GBOs (and matched resected tumour tissues) identified nicotinamide N-methyltransferase 1 (NNMT1) as a key enzyme potentially driving increased plasticity of fast tumours. In contrast, slow-growing GBOs activate NF2L2-dependent gene expression and inflammatory cytokine production upon culture, suggesting a stress-response mechanism of adaptation. DEGs analysis further identified 12 and 5 genes as biomarkers for fast and slow glioblastoma groups. These results suggested that GBO growth rates reflect heterogeneity in patients' tumour plasticity, with highly plastic tumor better equipped to resist and adapt to therapeutic treatment and therefore leading to poor survival outcomes. We confirmed this hypothesis by analysing clinical trial RNA bulk-seq data from two independent studies and found that biomarkers associated with the fast group were highly enriched in trial arms with poor survival outcomes. In summary, our results further highlight new biomarkers and potential targets against glioblastoma tumours stratified by their plasticity capacity and prognostic outcomes. Citation Format: Guillermo Gomez, Kaitlin Scheer, Erica Yeo, Chloe Shard, Helen Palethorpe, Anahita Fouladzadeh, Nga Truong, Sakthi Lenin, Rebecca Orsmby, Niclas Skarne, Bryan Day, Minh-Son To, John Toubia, Parul Mittal, Clifford Young, Conor Ryan, Annika Mascarenhas, Vanesa Tomatis, Anna Brown, Melinda Tea, Manuela Klinger-Hoffmann, Peter Hoffmann, Yusha Sun, Fadi Jacob, Xin Wang, MacLean Nasrallah, Zev Binder, Donald O'Rourke, Guo-Li Ming, Hongjun Song, Bryan Day, Stuart Pitson, Michael Brown, Santosh Poonnoose, Tessa Gargett, Ashwini Patil, Lisa Ebert. Growth rates of patient-derived tumor organoids reflect heterogeneity in tumor plasticity among glioblastoma patients [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A056.