Swanson, Kristin R.

Glioblastoma is the most common form of brain cancer in adults and remains one of the most lethal forms of cancer. Treating glioblastoma is uniquely challenging due to its aggressive nature and its critical location, the brain, which must remain largely undamaged. Due to these difficulties, there is a lack of effective treatment options. Computer models simulating tumors and their response to therapy, known as digital twins, provide a promising new strategy. These model systems are mathematical equations that capture the biological knowledge of how cells interact during tumor growth and in response to treatment. By solving these equations, we can create virtual patients that provide accurate predictions of real patients’ tumor progression. This approach has the potential to improve outcomes by designing therapies for the specific needs of each patient, moving beyond the limitations of the “one-size-fits-all” approach that is currently used.

Abstract Glioma stem cells (GSCs) are considered a major driver of glioblastoma (GBM) progression and are highly resistant to standard cytotoxic treatments. BMP4 has been shown to promote GSC differentiation, enhance radiosensitivity, slow tumor growth, and extend survival in animal models. Despite this promise, BMP4 has yet to achieve clinical impact, owing largely to heterogeneous and nonlinear responses across preclinical experimental systems. To elucidate how BMP4 could function as an effective targeted differentiation therapy in GBM, we develop a mathematical model that describes the growth of a GBM tumor via a hierarchy of GSCs, progenitor and terminally differentiated cells. We parameterize our model using new radiotherapy and proliferation assay data (with and without BMP4 exposure) from twelve patient-derived GSC lines. This integration of model and data allows us, for the first time, to quantitatively capture patient-specific heterogeneity in BMP4 sensitivity. We perform global sensitivity analysis on the model, identifying proliferation rate and GSC self-renewal sensitivity as key determinants of BMP4 efficacy. These parameters act as model-derived biomarkers that distinguish BMP4-responsive tumors from non-responsive ones. In an in silico analysis across a broad cohort of virtual patients, we find that continuous BMP4 delivery from surgical resection through radiotherapy consistently outperforms a single-dose strategy. Virtual clinical trials further show that, without stratification using these model-derived biomarkers, BMP4 yields little observable therapeutic benefit. In contrast, selecting patients with more proliferative, BMP4-responsive GSCs markedly increases the likelihood of observing a significant treatment effect. Citation Format: Nicholas Harbour, Lee Curtin, Loizos Michaelides, Matthew E. Hubbard, Pamela Jackson, Vinitha Rani, Rajappa Kenchappa, Virginea Farias, Anna Carrano, Markus Owen, Alfredo Quinones-Hinojosa, Kristin Swanson. Virtual clinical trials of BMP4 differentiation therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 6831.

Abstract Description Many cancers, including glioblastoma (GBM), exhibit a male-biased incidence and outcome disparity, partially driven by sex hormones. While androgens are known to suppress anti-tumor T cell immunity in certain solid tumors, we found that androgens instead enhance anti-tumor immunity in brain tumors. Upon castration, tumor growth was accelerated in brain tumor models, while the opposite effect was observed for tumors located outside the brain. Decreased T cell function was observed in both tumors and peripheral lymphoid organs in castrated mice, indicated by reduced anti-tumor cytokines and increased T cell exhaustion. Mechanistically, this systemic immunosuppression was driven by hyperactivation of the hypothalamus-pituitary-adrenal (HPA) axis in castrated mice with brain tumors. Blocking glucocorticoid receptors reversed the accelerated tumor growth in these mice, confirming the role of HPA axis dysregulation. This mechanism was brain-specific rather than GBM-specific, as hyperactivation of the HPA axis was also observed with intracranial implantation of non-GBM tumors. Furthermore, elevated production of proinflammatory cytokines (IL-1β, TNF) in the brain of castrated mice was identified as a driver of HPA axis activation. Taken together, these findings reveal unique endocrine-mediated mechanisms by which brain tumor modulate systemic immunity in the androgen-deprived setting and highlight the importance of organ-specific effects of anti-tumor immunity. Funding Sources NIH P01CA245705; NIH R35NS127083; American Brain Tumor Association; Cleveland Clinic; Case comprehensive cancer center Topic Categories Tumor Immunology: Cellular Responses and Tumor Microevironment (TIME)

Abstract High-grade glioma (HGG) continues to have a dismal prognosis owing in part to its heterogeneous and invasive nature both within and between patients. The most aggressive form of high-grade glioma, called glioblastoma, portends a median survival of around 15 months. Like many cancers, high-grade glioma rewires metabolism towards its own benefit, shifting away from the typical reliance of brain tissue on glucose towards other sources such as fatty acids and acetate. Using a cohort of image-localized high-grade glioma biopsies collected from diverse imageable regions with associated transcriptomics (58 patients, 202 biopsies), we are able to assess the spatial transcriptomic landscape of HGG. We have previously shown that the transcriptome of HGG can be characterized into a continuous transition between states, and we have shown that the cellular ecosystem follows a similar pattern, using a combination of trajectory inference and cellular deconvolution. These patterns are strongly associated with the non-enhancing and contrast-enhancing regions of the tumors, which are typically considered for treatment planning and progression assessment. Here, we applied differential expression of hallmark pathways to determine how metabolism factors into this paradigm. As expected, we find overall higher levels of metabolism-associated pathway expression in the contrast enhancing regions compared to the non-enhancing regions. Interestingly, we found metabolic expression differs between the ecologies associated with transcriptional subtypes of high-grade glioma. For example, fatty acid (p=0.003), xenobiotic (p<0.0001) and heme (p<0.0001) metabolism pathways were significantly upregulated for mesenchymal tumor samples, while significantly downregulated for proliferative tumor samples (p=0.02, p<0.0001, p<0.0001, respectively). As we continue to further understand the complex spatial and temporally heterogeneous landscape of these aggressive tumors, it is important to consider the metabolites on which they depend, to better inform prognosis and develop new treatments.

Abstract High-grade gliomas (HGG) have significant spatial heterogeneity, complicating therapeutic approaches. While multiparametric MRI (mpMRI) enables noninvasive assessment of this heterogeneity, most predictive models lack biological interpretability when used to classify or predict outcomes. Here, we developed an unsupervised framework that identifies novel mpMRI feature patterns through dimensional reduction, linking them to clinically relevant molecular signatures. We analyzed 18 advanced mpMRI features from 73 stereotactic biopsies across 20 HGG patients. Imaging data were normalized, followed by dimensional reduction using an iterative UMAP across varying parameters. Density-based clustering was used to generate stable image-derived clusters through consensus co-clustering. RNA-seq data from matched biopsies were analyzed using DESeq2 and GSEA to identify differentially expressed genes and enriched pathways. We then constructed a network linking specific imaging features to these transcriptional signatures. Validation was performed in an independent cohort, allowing us to extend clusters across full brain mpMRI. Unsupervised clustering of mpMRI features produced nine stable image-based clusters. These clusters were not fully explained by contrast enhancement, metabolic class, or tumor purity, shown by intraclass correlation (ICC) analyses, suggesting they capture novel biological variation. Transcriptomic analysis showed each cluster was associated with distinct gene expression patterns and enriched biological pathways, including neuronal differentiation, immune chemotaxis, and chromatin remodeling. Key imaging features (sRCBV, MTT, K2)—combined in cluster-specific patterns—enabled construction of an interpretable imaging–transcriptomic network. This was validated in an independent cohort. We then applied to full brain mpMRI, mapping spatially distinct regions to biological trends and patient outcomes. Addressing a research gap in imaging HGG, we present an interpretable network that connects specific patterns of mpMRI to clinically meaningful molecular signatures. This approach advances our understanding of tumor heterogeneity and supports future translational work in precision patient care.

Abstract High-grade gliomas (HGGs) are aggressive, infiltrative brain tumors with poor survival outcomes and limited therapeutic options. Their spatial and molecular heterogeneity necessitates large, well-curated datasets integrating imaging, clinical, and genomic information to enable precision medicine and drive novel treatment development. Using the Medtronic StealthStation neuronavigation system, we collected over 1,000 spatially localized intraoperative biopsy samples from more than 250 patients undergoing resection for glioma at Mayo Clinic and partnering sites. Each biopsy site was recorded with 3D coordinates, labeled in real time, flash-frozen, and annotated with metadata including collection time, anatomical location, and surgical plan. Samples were processed and entered into a secure database along with the corresponding sampling coordinates, frozen pathology, whole-exome and RNA sequencing data, 5-ALA fluorescence status, patient demographics, and T1, T1Gd, T2, and T2-FLAIR MRI. This multimodal dataset supports radiogenomic research by enabling precise mapping of imaging features to molecular characteristics. A subset of this data, including 268 biopsies from 52 patients (25 male, 27 female) with reported intraoperative 5-ALA fluorescence, was used to develop a radiomics model predicting locoregional patterns of 5-ALA positivity based on preoperative MRI. Image features were extracted from the biopsy sample coordinates across multiple MRI sequences and paired with corresponding 5-ALA fluorescence status to train the model for regional fluorescence prediction. The final model achieved 83% accuracy on the training set and 86% on the validation set using a 70-15-15 data split; this included 208 samples for training and 42 for validation, stratified by patient to ensure no overlap of samples from the same individual across sets. These findings demonstrate the feasibility of predicting 5-ALA fluorescence using radiomic features from standard preoperative MRI, offering a promising noninvasive strategy for improving intraoperative decision-making and enabling more precise, personalized resections in patients with high-grade gliomas.

Abstract BACKGROUND Glioblastoma (GBM) is a molecularly and phenotypically heterogeneous tumor, with regional invasion patterns driving treatment resistance and universal recurrence. In infiltrative regions, GBM cells adopt at least two predominant invasive phenotypes: Neuronal-enriched (NEU), with synaptic and neurodevelopment programs, and Glycolytic/Plurimetabolic (GPM), with heightened immune and metabolic activity. These subtypes differ in therapeutic vulnerability and microenvironmental interactions; features not captured by conventional contrast-enhanced (T1+C) MRI due to minimal blood-brain barrier (BBB) disruption. Our group sought to develop predictive models to distinguish variability of NEU/GPM subtypes within infiltrative GBM, utilizing graph convolutional networks (GCN), advanced multiparametric MRI (mpMRI), and spatially-matched image-localized biopsies with transcriptomic profiling. METHODS We analyzed 18 mpMRI features —including conventional, diffusion-weighted, and Dynamic Susceptibility Contrast (DSC) perfusion MRI— of 101 image-localized biopsies from MRI non-enhancing infiltrative margins of 64 GBM patient tumors. Features were hierarchically clustered per modality; each cluster was represented as a node in a Neo4j graph. Edges connected nodes sharing overlapping samples. Nodes were annotated with NEU or GPM subtype proportion. GCNs were trained to classify subtype-enriched nodes using graph topology and node features. RESULTS GCNs achieved high predictive performance on held-out test data (AUC = 0.95 for NEU, 0.88 for GPM; 70:30 train-test split) with similar 5-fold cross-validation performance and generalized well to an independent cohort (AUC > 0.85, n = 45). Importantly, performance was robust without relying on T1+C imaging, enabling prediction beyond BBB-disrupted regions. GCNs outperformed traditional classifiers (logistic regression, random forest). CONCLUSION Our results demonstrate that GCN-modeling of mpMRI features enables non-invasive prediction of invasive GBM subtypes at the tumor margin. Unlike convolutional neural networks (CNN), which rely on spatial continuity, GCNs leverage relational structure between heterogeneous imaging features, making them ideal for modeling sparse, non-grid data. This may support non-invasive regional vulnerability detection, surgical targeting, and improved response assessment.

Abstract Many cancers, including glioblastoma (GBM), exhibit a male-biased incidence and outcome disparity. The underlying reasons for this sex bias are unclear but likely involve differences in tumor cell state and immune response. This effect is further influenced by sex hormones, including androgens, which have been shown to inhibit anti-tumor T cell immunity in other solid tumors. Using a series of pre-clinical models, we observed that androgens drive anti-tumor immunity in brain tumors, contrasting with their effect in other tumor types. Upon castration, tumor growth was accelerated in GBM and brain tumor models, while the opposite effect was observed when tumors were located outside the brain. This castration effect could be reversed by testosterone administration. Flow cytometry revealed decreased T cell function, indicated by reduced anti-tumor cytokines (IFNγ and TNFα), and increased T cell exhaustion in both tumors and peripheral tissues in castrated mice, suggesting systemic immunosuppression. Indeed, the survival difference was completely abrogated in T cell deficient mice (RAG1 knockout). Mechanistically, increased hypothalamus-pituitary-adrenal (HPA) axis activity in castrated mice was responsible for the observed systemic immunosuppression, particularly in those with brain tumors. Blocking glucocorticoid receptors reversed the accelerated tumor growth in castrated mice, indicating that elevated glucocorticoid signaling mediated the castration effect. This mechanism was brain-specific rather than GBM-specific, as hyperactivation of the HPA axis was also observed with intracranial implantation of non-GBM tumors. Neuroinflammation caused by brain tumors was stronger in castrated mice, and proinflammatory cytokines like IL-1 and TNF induced HPA axis hyperactivation. Together, these findings highlight that brain tumors drive distinct endocrine-mediated mechanisms in the androgen-deprived setting and highlight the importance of organ-specific effects on anti-tumor immunity.