Molecular Heterogeneity Research Articles

Single-cell expression and immune infiltration analysis of polyamine metabolism in breast cancer.

Breast cancer is one of the most threatening women health diseases worldwide and its molecular heterogeneity offers a range of response to therapy. The role of polyamine metabolism is receiving increasing attention. Polyamine metabolism not only plays an important role in the occurrence and development of breast cancer, but also interacts with tumor immune microenvironment. In this work, we applied single-cell RNA-sequencing (scRNA-seq) and systems immunological approaches to interrogate immune cell infiltration gene-to-gene co-expressions in the bulk tumor transcriptomes of breast cancer. We acquired breast cancer sample data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO), evaluated the infiltration status of 22 immune cell types using CIBERSORTx tool, respectively. By leveraging the Retrospective Breast sample of various technologies including gene expression and methylation, we identified 46 breast cancer proliferation-associated co-expression modules using weighted gene coexpression network analysis (WGCNA) approach along with machine learning models which in turn delineated single cell level expressions features that these selected module possessed. We observed substantial cellular heterogeneity in the breast cancer microenvironment, where lineage-specific gene expression patterns were highly associated with tumor progression. Moreover, we also identified the gene modules correlated with immune cell infiltration level that could function as regulators in response to tumors for immune therapy. Moreover, risk scores were correlated with immune cell function in different patient groups defined by high- and low-risk. The findings of this study shed a new light upon molecular classification prognostic assessment and personalized treatment in breast cancer.

Abstract B007: Unlocking antiviral immunity against endogenous retroviruses in skin squamous cell carcinomas

Abstract Known as molecular parasites that invaded our genome through ancestors, transposons (mainly retrotransposons) are interspersed genomic repeats that constitute ∼40% of the mammalian genome. Primarily residing in the heterochromatin, retrotransposons exhibit dynamic expressions to orchestrate embryonic development and shape adult functions. Although rare, a cohort of evolutionarily young retrotransposons resisted extinction, whose selfish activities underlie polymorphic variations and genetic diseases including cancer. While the extraordinary molecular and functional heterogeneity of retrotransposons has been long recognized, mechanistic dissection remains challenging. To tackle this challenge, we use murine skin as our model, given its well-characterized, abundant, and highly accessible adult stem cells, not only mediating postnatal remodeling and tissue repair, but also serving as cell of origin for squamous cell carcinomas. By analyzing a genetic model lacking a known retrotransposon suppressor, histone methyltransferase Setdb1, which we found to be essential in the adult skin, we saw hair loss and hair follicle stem cell exhaustion phenotype, accompanied by a robust and selective surge of endogenous retroviruses (ERVs). When combined with squamous cell carcinoma drivers, Setdb1 loss significantly blocked tumor progression in vivo. We detected abundant retroviral peptides originated from its full-length copies through mass spectrometry and viral-like particles through transmission electron microscopy that are immunoreactive to previously reported ERV surface envelope (env) antibody. Similar viral-like particles were broadly evident across several epithelial tissues beyond skin, implicating its conserved function across squamous cancers. Mechanistically, epithelial originated ERVs elicit tissue wide inflammatory response, and can be ameliorated by pharmacological or genetic inhibition of retroviral activity. Moreover, ERV reactivation induced replication stress functionally accounts for stem cell exhaustion, therefore serving as a specific therapeutic vulnerability in squamous cancers. As an evolutionary conundrum, viral-coding ERVs pose a threat to host fitness and yet, they’ve resisted extinction persisting in the mammalian genome. Our findings suggest ERVs wrestle with the host surveillance programs to regulate adult tissue physiology and pathology, thus providing a key target in eliciting innate immunity in squamous cell carcinomas. Citation Format: Ying Lyu, Yejing Ge. Unlocking antiviral immunity against endogenous retroviruses in skin squamous cell carcinomas [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: RNAs as Drivers, Targets, and Therapeutics in Cancer; 2024 Nov 14-17; Bellevue, Washington. Philadelphia (PA): AACR; Mol Cancer Ther 2024;23(11_Suppl):Abstract nr B007.

A novel mouse model recapitulating the MMR-defective SCLC subtype uncovers an actionable sensitivity to immune checkpoint blockade.

Small cell lung cancer (SCLC) has an extremely poor prognosis. Despite high initial response rates to chemotherapy and modest survival improvements with the addition of immune checkpoint inhibitors (ICI), almost all patients experience relapse and fatal outcomes. Recent genomic insights uncovered extensive molecular heterogeneity in addition to the almost uniform loss of RB1 and TRP53. Additionally, defective DNA mismatch repair (MMR) has recently been described in some SCLC cases. Here, we generated a novel SCLC mouse model capturing MMR deficiency and assessed immunotherapy responses. We developed an MMR-deficient genetically engineered mouse model (GEMM) of SCLC by introducing a conditional Msh2 gene, crucial for maintaining MMR integrity, into the standard Rb1fl/fl;Trp53fl/fl (RP) model. Genomic characteristics and preclinical therapy responses were evaluated by focusing on overall survival and whole exome sequencing (WES) analyses. MMR-defective SCLC tumors (Rb1fl/fl;Trp53fl/fl;Msh2fl/fl (RPM)) developed later than tumors in MMR-proficient mice. However, the time from tumor manifestation to death of the affected animals was substantially shortened (median survival 55 days in RP vs. 46.5 days in RPM), indicating increased aggressiveness of MMR-defective tumors. RPM tumors exhibited MMR deficiency, high tumor mutational burden (TMB), and an elevated load of candidate neoantigens, compared to RP lesions (p = 0.0106), suggesting increased immunogenicity. Importantly, the overall survival of RPM animals was significantly improved when exposed to ICI. We propose a novel RPM mouse model as a suitable system to mimic MMR-defective SCLC and tumors with high TMB. We provide in vivo evidence that Msh2 deficiency enhances ICI sensitivity. These findings could contribute to stratifying SCLC patients to immunotherapy, thereby improving treatment outcomes.

Single-cell RNA sequencing reveals the landscape of the cellular ecosystem of primary hepatocellular carcinoma.

Hepatocellular carcinoma (HCC) cells, along with multiple nonmalignant stromal cells, such as fibroblasts, endothelial cells and immune cells, comprise an intricate cellular ecosystem, undergo dynamic phenotypic changes and present complicated cellular interactions, thus synergistically facilitating HCC initiation and progression and leading to treatment resistance. Clarifying the heterogeneity, cell plasticity and complexity of the cellular ecosystem of HCC will be highly beneficial for understanding HCC development and identifying novel therapeutic targets. Single-cell RNA sequencing (scRNA-seq) refers to profiling the transcriptome at single-cell resolution, and the development of scRNA-seq technology and analysis algorithms has greatly promoted the analysis of cell composition, cell subpopulation heterogeneity, development trajectory and cell-to-cell interactions in cell populations. In this review, we systematically summarized and discussed scRNA-seq in treatment-naive primary HCC and revealed the global cell composition of HCC; the widespread molecular heterogeneity of HCC cells; the molecular subtypes of fibroblasts; the cell composition, functional states and development trajectory of immune cells; and the frequent interactions between different cell types to systematically draw the landscape of the cellular ecosystem of primary HCC.

Identification of Molecular Subtypes and Prognostic Traits Based on Chromosomal Instability Phenotype-Related Genes in Lung Adenocarcinoma

Lung adenocarcinoma (LUAD) exhibits significant molecular heterogeneity; however, previous studies have not fully explored its classification into distinct molecular subtypes. Here, we identified LUAD-significant chromosomal instability (CIN) phenotype genes (n = 24) using a TCGA-LUAD cohort (n = 592) and evaluated their ability to predict pathologic grade. Unsupervised clustering and principal component analysis revealed that LUAD patients could be classified into CIN phenotype-related subtypes (GroupLow, GroupModerate, and GroupHigh), each exhibiting distinct transcriptomic patterns. Notably, the GroupHigh showed significantly poor overall survival [OS; hazard ratio (HR) = 1.43, p-value < 10−3] and disease-free survival (DFS; HR = 1.27, p-value < 10−3). Univariate and multivariate analysis confirmed that its expression status was an independent prognostic predictor (p-value < 10−3, HR = 2.18, 95% C.I = 1.26–3.76) of the clinical outcomes, outperforming pathologic grade (p-value < 10−3, HR = 1.2, 95% C.I = 1.08–1.33). Moreover, analysis of surfactant metabolism-related genes revealed higher expression in the GroupLow, which was associated with a favorable prognosis. By integrating multiple independent cohorts (n = 779), we validated these findings and confirmed that CIN phenotype gene status serves as a critical prognostic marker in LUAD. Furthermore, genomic profiling showed that the GroupHigh exhibited frequent mutations in key genes such as KEAP1, LYST, SETD2, and TP53, with oncogenes in this group preferentially showing copy number gains. Our study highlights the significance of CIN phenotype gene status as a predictor of LUAD prognosis and its association with transcriptomic and genomic alterations, paving the way for further clinical validation and potential therapeutic interventions.

Single-Sample Networks Reveal Intra-Cytoband Co-Expression Hotspots in Breast Cancer Subtypes

Breast cancer is a heterogeneous disease comprising various subtypes with distinct molecular characteristics, clinical outcomes, and therapeutic responses. This heterogeneity evidences significant challenges for diagnosis, prognosis, and treatment. Traditional genomic co-expression network analyses often overlook individual-specific interactions critical for personalized medicine. In this study, we employed single-sample gene co-expression network analysis to investigate the structural and functional genomic alterations across breast cancer subtypes (Luminal A, Luminal B, Her2-enriched, and Basal-like) and compared them with normal breast tissue. We utilized RNA-Seq gene expression data to infer gene co-expression networks. The LIONESS algorithm allowed us to construct individual networks for each patient, capturing unique co-expression patterns. We focused on the top 10,000 gene interactions to ensure consistency and robustness in our analysis. Network metrics were calculated to characterize the topological properties of both aggregated and single-sample networks. Our findings reveal significant fragmentation in the co-expression networks of breast cancer subtypes, marked by a change from interchromosomal (TRANS) to intrachromosomal (CIS) interactions. This transition indicates disrupted long-range genomic communication, leading to localized genomic regulation and increased genomic instability. Single-sample analyses confirmed that these patterns are consistent at the individual level, highlighting the molecular heterogeneity of breast cancer. Despite these pronounced alterations, the proportion of CIS interactions did not significantly correlate with patient survival outcomes across subtypes, suggesting limited prognostic value. Furthermore, we identified high-degree genes and critical cytobands specific to each subtype, providing insights into subtype-specific regulatory networks and potential therapeutic targets. These genes play pivotal roles in oncogenic processes and may represent important keys for targeted interventions. The application of single-sample co-expression network analysis proves to be a powerful tool for uncovering individual-specific genomic interactions.

Abstract 4138334: Identification and Immunological Characterization of Arrhythmia-Related Molecular Clusters in Ischemic Cardiomyopathy

Ischemic cardiomyopathy, a severe cardiac condition resulting from prolonged myocardial ischemia, is characterized by ventricular dilation, dysfunction, and an increased risk of life-threatening arrhythmias. Arrhythmia, a common complication in ischemic cardiomyopathy, is associated with poor clinical outcomes. Recent studies have indicated that the molecular heterogeneity of ischemic cardiomyopathy may contribute to the development of arrhythmias. Therefore, our study aimed to explore arrhythmia-related molecular clusters in ischemic cardiomyopathy and establish a prediction model for risk stratification. Based on gene expression profiles of ischemic cardiomyopathy patients, we analyzed the dysregulation of arrhythmia-associated genes and the associated immune characteristics (Figure 1A-C). Using unsupervised clustering analysis, we identified distinct arrhythmia-related molecular clusters in ischemic cardiomyopathy patients (Figure 1D-F). Cluster-specific differentially expressed genes were further identified using weighted gene co-expression network analysis (WGCNA). Functional enrichment analysis revealed significant differences in biological processes and pathways between the clusters (Figure 1G-H). Subsequently, we developed a prediction model using a random forest algorithm, which showed superior performance in distinguishing patients with different arrhythmia risks (Figure 1I-J). This forest model demonstrated excellent performance in one external validation datasets with AUC values of 0.891. In conclusion, our study provides novel insights into the molecular basis of arrhythmias in ischemic cardiomyopathy and establishes a prediction model with potential clinical utility for risk stratification and targeted therapies. Further investigation into the specific mechanisms underlying the identified molecular clusters may reveal new therapeutic targets for the management of arrhythmias in ischemic cardiomyopathy patients.

STEM-13. NUCLEAR LOCALIZATION OF ALPHA-1-ANTICHYMOTRYPSIN INCREASES IN GLIOBLASTOMA FOLLOWING STANDARD OF CARE TREATMENT

Abstract Glioblastoma (GBM) is the most common malignant primary brain tumor in adults with a median survival of 8-14 months and 5-year survival rate of 6.9%. Its aggressive nature arises mostly from molecular heterogeneity and resistance to standard of care (SOC), including radiation and temozolomide (TMZ). Recurrence is nearly inevitable, driven by highly invasive brain tumor initiating cells (BTICs). Our previous studies revealed that exposing BTICs to cerebrospinal fluid (CSF) from GBM patients increases cell migration and proliferation compared to non-cancer CSF. Transcriptomic analysis identified the upregulation of the SERPINA3 gene, which encodes alpha-1-antichymotrypsin (ACT). Silencing of SERPINA3 reduces BTIC migration and proliferation and abolishes the promigratory response to GBM-derived CSF, highlighting SERPINA3 as a potential therapeutic target. ACT has been reported to bind to DNA, although the implications of this interaction are unknown. We have previously observed higher levels of intranuclear ACT in human GBM compared to non-cancer brain tissues. We hypothesize that the nuclear localization of SERPINA3 in GBM contributes to increased malignancy, opening the possibility of therapeutically targeting its intranuclear transport. Here, we determined the intracellular localization of ACT in BTICs and GBM and compared it to non-cancer tissues. Additionally, we evaluated whether SOC affected intranuclear localization of ACT. Patient-derived BTICs were treated with TMZ, radiation, and a combination of both, to assess changes in ACT expression via Western Blot, PCR, and ICC. A decrease in cell confluency was correlated with a significant decrease (p<0.05) in SERPINA3 mRNA and overall protein expression when BTICs were treated with TMZ alone or in combination. Interestingly, confocal microscopy revealed a significant increase (p<0.05) in its nuclear localization, particularly with TMZ treatment, indicating a change in its intracellular distribution. Overall, nuclear translocation of SERPINA3 increases in response to SOC with potential implications as a treatment resistance mechanism.

DDDR-46. TARGETING TRANSGLUTAMINASE 2 (TGM2) ENHANCES SENSITIVITY OF GLIOBLASTOMA CELL LINES TO RADIOTHERAPY AND TEMOZOLOMIDE

Abstract Glioblastoma (GBM), the most aggressive and common primary brain tumor in adults, remains highly resistant to conventional therapeutics, including radiation therapy (RT), largely due to its molecular heterogeneity. Few biomarkers have been identified that successfully guide or predict treatment response. Transglutaminase 2 (TGM2), a ubiquitous multifunctional enzyme involved in numerous signal transduction pathways related to tumorigenesis, has recently been suggested to drive chemoradioresistance. Here, we investigate the effects of disrupting TGM2 by pharmacologic inhibition or CRISPR knockout (TGM2-KO) on GBM cell’s response to RT and Temozolomide (TMZ) chemotherapy. Using western blotting and clonogenic assays, we show that the allosteric TGM2 inhibitor, GK921, stabilizes p53 expression and impedes reproductive ability at IC50 and IC25 concentrations. TGM2-KO and GK921 (at IC10 and lower) sensitized GBM cell lines to RT with almost equivalent efficacy, resulting in a 2- to 3-fold suppression of clonal expansion compared to scramble controls or irradiation alone. We also demonstrate increased sensitivity of GBM cell lines to TMZ upon inhibition of TGM2 by GK921. Our findings indicate that the use of TGM2 inhibitors, such as GK921, could enhance the efficacy of RT and the potency of TMZ, positioning TGM2 as a promising therapeutic target that may improve standard treatment of GBM cells expressing TGM2.

EPCO-60. EXPRESSION OF EARLY PROGENITOR MARKERS WITHIN NF2-ASSOCIATED SPINAL EPENDYMOMA

Abstract Spinal ependymomas (SP-EPNs) are intramedullary spinal tumors commonly found in patients with NF2-related schwannomatosis. Recent work using bulkRNA sequencing has suggested that SP-EPNs express a molecular signature most similar to ependymal cells. However, this type of sequencing may miss rare stem cell populations within tumors. The aim of this study was to assess the spatial heterogeneity within NF2-associated SP-EPNs and explore whether a stem cell population exists within SP-EPNs. Spatial transcriptomics (ST) was performed on a SP-EPN sample previously resected from a patient with NF2-related schwannomatosis using the 10X Genomics Visium Cytassist Platform. Libraries were sequenced on the Illumina NovaSeq 6000. 28-bp reads including spatial barcode and UMI sequences and 50-bp probe reads were generated with Illumina NovaSeq 6000 at the Center for Medical Genomics at Indiana University School of Medicine. Space Ranger 2.1 was used to process the raw files after sequencing. Data analysis was performed with R package Seurat 4.9.9.9041 following preprocessing with spaceranger. Sctransform normalized the spatial transcriptome dataset. Clusters were identified using “FindClusters” and “FindNeighbors” functions. There was diffuse expression of astrocyte and ependymal cell markers throughout the SP-EPN sample with smaller areas expressing radial glia and stem cell markers, as well as overlap between these regions. Expression of vascular and connective tissue markers varied throughout the tumor. There was minimal to no expression of neural genes in the sample. ST on our SP-EPN sample revealed a tumor with marked cellular and molecular heterogeneity. Although the primarily expressed ependymal cell and astrocyte markers, there does appear to be small, stem cell-like populations. This suggests that NF2-associated SP-EPNs may have a developmental hierarchy with radial glia-like cells and more mature progenitors within the ependymal cell lineage. This proof-of-concept study offers important findings that provide justification for further investigation using single-cell sequencing.

TMIC-46. TRANSCRIPTIONAL HETEROGENEITY AND MECHANISTIC PATHWAYS OF RECURRENT GLIOBLASTOMA: INSIGHTS FROM A PRE-CLINICAL RECURRENT TUMOR MODEL

Abstract Despite aggressive therapy combining surgical resection, radiation and chemotherapy, glioblastoma (GBM) almost always recurs. Currently, there is no standard treatment for recurrent GBM, and patient inclusion in clinical trials is based on the genetic and molecular profile of the primary tumor even though the effects of chemoradiation on the transcriptional subtype in matched primary and recurrent GBM remain largely understudied. Paralleling the GLASS consortium, we sought to perform a longitudinal assessment of GBM recurrence after therapy based on molecular subtype with the hypothesis that recurrent tumors exhibit a distinct transcriptional landscape different from treatment-naive tumors. To address this, we utilized primary tumor-derived glioma stem cells (GSCs) representing classical and proneural subtypes. We then undertook in vivo intracranial implantation studies and performed standard concomitant treatment that mimics the Stupp protocol. Upon bulk RNAseq data analyses of untreated (n=21) and recurrent tumors (n=17) we identified that while the molecular subtype of recurrent tumors does not change post therapy, their transcriptional profile is more heterogenous compared to untreated tumors for both subtypes. Hierarchical cluster analyses revealed distinct sub-clusters of recurrent tumors. Common mechanisms of recurrence include upregulation of angiogenesis pathways and PI3k/Akt/mTOR signaling pathways. Interestingly, recurrent tumors exhibit upregulation of genes involved in cellular metabolism and membrane transport/ potential. The source of heterogeneity in recurrent tumors in both subtypes appears to be due to the diversity in expression of gene sets involved in hallmark of cancer pathways such as cell cycle, Notch and Wnt signaling pathways and heterogeneity in oncogenic signatures and cell states resembling fetal or neural cells. Further research into the recurrent tumor microenvironment by spatial multi-omics approaches using this pre-clinical recurrent GBM model will elucidate the linkage between cellular as well as molecular heterogeneity and enable identification of targets that can help drive therapeutic decisions for recurrent GBM patients.

IMMU-27. INHIBITION OF THE EXTRACELLULAR MATRIX PROTEIN EFEMP1/FIBULIN-3 REDUCES IMMUNOSUPPRESIVE SIGNALING IN TUMOR STEM CELLS AND INCREASES MACROPHAGE ACTIVATION AGAINST GLIOBLASTOMA

Abstract Glioblastoma tumors remain a formidable challenge for immune-based treatments because of their molecular heterogeneity, poor immunogenicity, and growth in the largely isolated and immunosuppresive neural environment. As the tumor grows, GBM cells change the composition and architecture of the neural extracellular matrix (ECM), affecting the mobility, survival, and function of immune cells such as tumor-associated microglia and infiltrated macrophages (TAMs). We have previously described the unique expression of the ECM protein EFEMP1/fibulin-3 in GBM compared to normal brain and demonstrated that this secreted protein promotes growth and resistance of the GBM stem cell (GSC) population. Here, we describe a novel immunomodulatory role of fibulin-3 and the immuno-boosting effects of targeting this ECM protein. Mice carrying fibulin-3-deficient intracranial tumors showed increased myeloid infiltration and reduced expression of TAM pro-tumoral markers (Arginase, CD206) compared to controls. The opposite was observed in orthotopic tumors overexpressing fibulin-3. In silico dataset analysis revealed positive correlation of fibulin-3 with an immunosuppresive signature in GBM, which was validated in GBM stem cells (GSCs). Accordingly, expression of fibulin-3 by GSCs promoted anti-inflammatory polarization of co-cultured macrophages. We further demonstrated that fibulin-3 regulated the expression of immunosuppresive signals (CSF-1, TGFbeta) and the innate immune checkpoint CD47 in GSCs via autocrine activation of NF-kB signaling. Immunosuppresive signals in GSCs were downregulated by knockdown of fibulin-3 or inhibition of this protein with an anti-fibulin-3 antibody (mAb428.2). Genetic or antibody-mediated targeting of fibulin-3 in GSCs co-cultured with macrophages (U937 and THP1 lines) increased macrophage phagocytosis and reduced the viability of the tumor cells. Furthermore, local delivery of mAb428.2 in mice carrying intracranial GBM resulted in increased infiltration of TAMs expressing pro-inflammatory markers and reducing tumor viability. Our findings show that anti-fibulin-3 approaches, which impact the tumor ECM, can diminish immunosuppression in GBM and boost innate immune responses against the tumor.

SURG-37. ADVANCING GLIOBLASTOMA RESEARCH THROUGH RESECTION MODELS

Abstract Regardless of its cellular and molecular heterogeneity, glioblastoma (GBM) tumors share a common hallmark: an infiltrative growth pattern that poses challenges to surgical resection and leads to incomplete tumor removal. As part of the current standard-of care treatment, patients undergo surgery in addition to temozolomide and radiation therapy. In GBM research, mice are often used to replicate tumor development and test treatments, but this approach typically omits the crucial aspects of surgical intervention, and thus fails to recapitulate important aspects of human GBM disease progression at the surgical margin. Here we address this challenge by developing a technique for tumor resection in mice. First, we implant the infiltrative patient derived GBM cell line Akaluc G16302 in the frontal cortex, specifically at anteroposterior 2mm, mediolateral 0.5mm, dorsoventral 1mm starting from bregma. After 6 weeks, we anesthetize the mouse and do a craniotomy at the injection site, approximately 3-4mm in diameter. We perform a circumferential cut using a curved Fukushima scissor at the edges of the craniotomy area and scoop the tumor core, mimicking surgical resection. Mice survive the surgery and are alive up to 37 days post-resection. Histology confirms resection of tumor cells and a surgical tumor margin remaining in the xenograft animal. In our pilot study, mice undergoing resection displayed significantly increased overall survival compared to mice without surgery (log-rank p-value = 0.0448; n=6; survival range of 91-100 days for control and 97-107 days for resected mice), recapitulating survival benefit also seen in humans. This preliminary data demonstrates the feasibility of the technique and resection-associated survival benefit. Such advancements enable researchers to more accurately replicate the biology of GBM progression and treatment response observed in human patients, bringing us closer to modeling the clinical scenario in mice. Key words: Glioblastoma, resection, biopsy

CTNI-49. PRECISION MEDICINE CLINICAL TRIAL FOR ADULT PATIENTS WITH RECURRENT GLIOBLASTOMA

Abstract Salvage therapies for recurrent glioblastoma (rGBM) have limited efficacy, with median overall survival of 9 months (OS-9) and 6-month progression-free survival (PFS-6) of 10-25%. Given GBM’s molecular heterogeneity, targeting a single molecular abnormality in isolation has consistently failed as a strategy, and precision combination approaches are needed. An initial 15-patient cohort evaluated the feasibility and safety of treating patients with rGBM with an individualized treatment regimen; we added a second 15-patient expansion cohort to improve the power of preliminary efficacy assessment. Patients underwent clinically indicated surgery with genomic profiling using the UCSF500 targeted DNA sequencing panel. Each personalized treatment regimen combined up to 4 FDA-approved drugs, including one cytotoxic agent, as determined by consensus discussion at an individualized molecular tumor board for each patient. Bevacizumab was not utilized in these combinations. The primary objective of the combined 30-patient cohort was efficacy of individualized treatment regimens by OS-9; secondary objectives included safety and additional efficacy measures including PFS-6. Median age was 53 years (range 32-71). Of 30 patients, 18 were male, 23 were enrolled at first recurrence, 29 had IDH-wildtype glioblastoma, 1 had IDH-mutant astrocytoma, and 10 had MGMT promoter methylation. A total of 12 FDA-approved drugs were used in 18 combinations; the most common regimen prescribed was lomustine, afatinib, and abemaciclib (7; 23%,); second-most common was temozolomide, olaparib, and afatinib (4; 13%). OS-9 of the combined cohort was 70% (95% CI 0.55 – 0.89) and PFS-6 was 27% (95% CI 0.15 – 0.48), with median OS of 11.6 months (95% CI 9.7 – 16.7) and median PFS of 3.7 months (95% CI 2.0 – 5.5).While overall efficacy was modest, implementation of individualized treatment regimens in a timely fashion was safe and feasible for patients with surgically resectable rGBM. Analysis is ongoing to assess whether specific genomic profiles or treatment regimens were particularly efficacious.

TMIC-52. DECODING METABOLIC INTERACTIONS IN GLIOBLASTOMA TUMOR MICROENVIRONMENT THROUGH COMPREHENSIVE SPATIAL TRANSCRIPTOMIC AND PROTEOMIC ANALYSIS

Abstract INTRODUCTION Within the GBM microenvironment, distinct micro-niches arise from independent cancer cell lineages with unique metabolic needs, characterized as Fast Cycling Cells (FCCs) and Slow Cycling Cells (SCCs). FCCs preferentially utilize aerobic glycolysis for energy, whereas treatment-resistant SCCs depend on lipid metabolism. Our study decodes the molecular and spatial heterogeneity within the GBM microenvironment, focusing on the dichotomy of immune infiltrates and the metabolic interactions between SCCs and the immune compartment in human GBM patients. METHODS We performed spatial transcriptomics and proteomics combined with geospatially resolved single-cell neighborhood analysis of IHC-stained GBM tissues using CellProfiler and the SNAQ algorithm that we developed to investigate the specific immune microenvironment of SCCs and FCCs. Holotomography and fluorescence-based time-lapse imaging coupled with flow cytometry were used to study lipid transfer between immune cells and GBM cells. Finally, the effects of pharmacological and genetic targeting of lipid transfer were investigated both in vitro and in vivo. RESULTS Our study establishes a connection between spatial metabolic heterogeneity and immune diversity, highlighting how specific tumor cell lineages correlate with characteristic immune micro-niches. In particular, SCCs exploit the metabolic properties of recruited immunosuppressive myeloid-derived cells, which metabolically support tumor cells by facilitating lipid transport and supporting SCC metabolism. Our results show that inhibiting lipid transfer affect SCCs, remodels the immune microenvironment, delays tumor growth and improves disease outcome. Our study also demonstrates a correlation between predicted sensitivity to lipid-modifying drugs, such as statins, and the SCC phenotype. This correlation could enable stratification and selection of patients most likely to benefit from such treatments. CONCLUSIONS This study establishes critical connections between the TME, cellular metabolism, anti-tumor immunity, and treatment vulnerabilities. Our research provides novel insights into how metabolic exchanges contribute to tumor progression and suggests that targeting metabolic interactions could be an effective therapeutic strategy for GBM.

EPCO-48. RESOLVING INTRATUMORAL TRANSCRIPTIONAL HETEROGENEITY ACROSS THE CORE-EDGE SPATIAL AXIS IN GLIOBLASTOMA

Abstract Intratumoral heterogeneity promotes tumor progression, treatment resistance, and recurrence in glioblastoma (GBM), and the cellular tumor core and infiltrating edge represent transcriptionally distinct cellular ecosystems. In this study, we characterize and validate the molecular differences across the core-edge spatial axis in GBM by applying whole-transcriptome GeoMx Digital Spatial Profiling (DSP) to 5 IDH-wildtype tumors. Three 3-micron tissue cores were extracted from each tumor and arranged into a tissue microarray (TMA). Core and edge regions of interest (ROIs) were chosen through coregistration of H&E and immunofluorescent TMA scans. Following DSP, data for 44 ROIs (30 core and 14 edge) and 8530 RNA targets were bioinformatically analyzed. Single-sample gene set enrichment analysis aligned the DSP core and edge regional signatures to those of publicly annotated datasets, substantiating our profiling methodology. Unsupervised analyses revealed ROI clusters unique to each tumor specimen for both pre- and post-normalized data. To mitigate a potential patient-level effect in subsequent differential gene expression and pathway analyses, we leveraged a linear mixed-effect model using core versus edge as a fixed effect and tumor specimen as a random effect. Our analyses showed upregulation of epigenomic markers in core ROIs and neuronal markers in edge ROIs. Further, core and edge signatures were incongruent with the canonical GBM subtypes (Mesenchymal, Classical, Proneural) and malignant cell states (OPC-like, NPC-like, AC-like, MES-like). Cellular deconvolution analysis highlighted differential enrichment of neoplastic cells in core ROIs and neuroglia (astrocytes and oligodendrocytes) in edge ROIs. In summary, we leverage spatial profiling to further validate the geographic molecular heterogeneity of GBM, which is characterized by divergent neuronal programs within the infiltrating edge.

CNSC-44. TARGETING OF THE YAP/TAZ-TEAD AXIS DIMINISHES VIABILITY AND DIFFERENTIATION POTENTIAL IN PEDIATRIC HIGH-GRADE GLIOMA

Abstract Pediatric high-grade gliomas (pHGGs) are invasive and fast-growing brain malignancies. They account for up to 40% of all brain tumor-related deaths in childhood. Genetic alterations that characterize these gliomas include H3 onco-histone mutations and genetic alterations in receptor-tyrosine kinases (RTKs), including amplification, mutation, and gene fusions in RTKs. However, direct targeting with tyrosine kinase inhibitors (TKIs) has shown at best partial responses in patients due to the molecular heterogeneity within these tumors and the lack of pharmaceutical penetration into the central nervous system (CNS). Thus, alternative therapeutic approaches to treat pHGGs are urgently needed. The YAP/TAZ transcription factors, which cooperate with TEAD co-factors to promote target gene expression, orchestrate stem-cell fate decisions and proliferation downstream of growth factor signaling pathways, such as RTKs. We observed that YAP expression is upregulated in RTK mutant and H3 mutant pHGGs. Depletion of YAP via RNAi methods decreased proliferation, inhibited pHGG stem cell self-renewal, and induced apoptosis of pHGG cells. Use of Verteporfin (VP), a FDA-approved YAP/TAZ small molecule inhibitor, mimicked the effects seen by YAP genetic depletion. Using an in vitro human cortical organoid-tumor model in which tumor cells are engrafted to human brain cortical organoids, we screened for responses in tumor cells and in non-tumor cells in the neural microenvironment upon VP treatment, using single-cell RNA-sequencing and immunofluorescence to visualize responses. These organoid models offer improved models for modeling how the neural CNS microenvironment shapes responses of tumor cells to YAP/TAZ inhibition and VP treatments. Results from our work showing that YAP/TAZ-TEAD inhibition is a promising therapeutic avenue for management of pHGGs will be presented. Further pre-clinical work will follow to correlate observation and understand the therapeutic efficacy of YAP/TAZ-TEAD inhibition for pHGG treatment.

DDDR-43. ERAS-801 IS A SELECTIVE BRAIN-PENETRANT EGFR INHIBITOR WITH IMPROVED ACTIVITY AGAINST EGFR EXTRACELLULAR DOMAIN-MUTANT GLIOBLASTOMA

Abstract Epidermal growth factor receptor (EGFR) is a driver of nearly 50% of glioblastoma (GBM), yet EGFR inhibitors have failed in GBM (e.g., erlotinib, lapatinib, neratinib, gefitinib, and afatinib) due to poor CNS penetration and/or an inability to inhibit EGFR in its wildtype and extracellular domain (ECD) mutated forms (e.g., EGFRvIII). Here we describe ERAS-801, a reversible and selective EGFR tyrosine kinase inhibitor with unique binding properties that enable potent activity for amplified and ECD mutant EGFR. ERAS-801 demonstrated the highest nonclinical CNS penetration out of all tested EGFR inhibitor comparators. Accordingly, relative to other EGFR inhibitors tested in GBM trials (e.g., erlotinib, lapatinib, and osimertinib), ERAS-801 showed superior survival benefit in EGFR altered orthotopic PDX models. Finally, further characterization of ERAS-801 in a preclinical trial of glioma orthotopic PDX models that capture the molecular heterogeneity of GBM, ERAS-801 demonstrated a significant survival benefit at a 25 mg/kg QD dose in the vast majority of EGFR-driven orthotopic GBM PDX models, with FDG-PET imaging serving as a non-invasive biomarker of response. Based on these nonclinical data, ERAS-801 is currently under clinical investigation for GBM.

3D-environment and muscle contraction regulate the heterogeneity of myonuclei.

Skeletal muscle formation involves tight interactions between muscle cells and associated connective tissue fibroblasts. Every muscle displays the same type of organisation, they are innervated in the middle and attached at both extremities to tendons. Myonuclei are heterogeneous along myotubes and regionalised according to these middle and tip domains. During development, as soon as myotubes are formed, myonuclei at muscle tips facing developing tendons display their own molecular program. In addition to molecular heterogeneity, a subset of tip myonuclei has a fibroblastic origin different to the classical somitic origin, highlighting a cellular heterogeneity of myonuclei in foetal myotubes. To gain insights on the functional relevance of myonucleus heterogeneity during limb development, we used 2D culture and co-culture systems to dissociate autonomous processes (occurring in 2D-cultures) from 3D-environment of tissue development. We also assessed the role of muscle contraction in myonucleus heterogeneity in paralysed limb muscles. The regionalisation of cellular heterogeneity was not observed in 2D cell culture systems and paralyzed muscles. The molecular signature of MTJ myonuclei was lost in a dish and paralysed muscles indicating a requirement of 3D-enviroment and muscle contraction for MTJ formation. Tip genes that maintain a regionalized expression at myotube tips in cultures are linked to sarcomeres. The behaviour of regionalized markers in cultured myotubes and paralyzed muscles allows us to speculate whether the genes intervene in myogenesis, myotube attachment or MTJ formation.

EPCO-44. SPATIAL TRANSCRIPTOMICS ENABLES STATISTICAL MODELING OF SEX DIFFERENCE IN GENE CO-EXPRESSION NETWORK AND IMMUNE LANDSCAPE OF GLIOBLASTOMA

Abstract Glioblastoma (GBM) is characterized by inter- and intra-tumor molecular heterogeneity and demonstrated cell plasticity. Sex differences have been observed in various aspects: a male:female GBM incidence ratio is 1.6:1; standard of care treatments (debulking, radiation, and temozolomide) benefit females more than males. 14 human GBM samples from two study sites were subject to spatial profiling on the Vizgen Merscope platform to examine whether distinct biological and therapeutic niches in GBM were evident, to understand how the tumor microenvironments (TME) contributed to the heterogeneity of GBM, and to interrogate sex difference in the immune landscape of GBM. Multiple sets of spatial probes were designed to capture the cell identities (cancer stem cells, astrocytes, oligodendrocytes, microglia, macrophages, etc.) and the spatially-resolved expression patterns of genes with different functions. Two of the 14 GBM samples were also profiled on another FISH-based spatial platform (Veranome). The data from the two platforms exhibited a linear correlation. We explored the gene-gene co-expression network, and calculated the interaction probabilities between genes which were then compared for sex difference. We found that female samples showed significantly lower (two-fold) CD4+/CD8A+ T-cell count ratios than male samples. We also observed sample-specific spatial niches of different GBM cell subtypes and immune cell populations. Our large-scale spatial transcriptomics study revealed intrinsic sex-based differences in immune response to tumor and illuminated microenvironmental influences on GBM.