Spatial cellular architecture predicts prognosis in glioblastoma

Tumour hypoxia in driving genomic instability and tumour evolution

Intratumour hypoxia is a feature of all heterogenous solid tumours. Increased levels or subregions of tumour hypoxia are associated with an adverse clinical prognosis, particularly when this co-occurs with genomic instability. Experimental evidence points to the acquisition of DNA and chromosomal alterations in proliferating hypoxic cells secondary to inhibition of DNA repair pathways such as homologous recombination, base excision repair and mismatch repair. Cell adaptation and selection in repair-deficient cells give rise to a model whereby novel single-nucleotide mutations, structural variants and copy number alterations coexist with altered mitotic control to drive chromosomal instability and aneuploidy. Whole-genome sequencing studies support the concept that hypoxia is a critical microenvironmental cofactor alongside the driver mutations in MYC, BCL2, TP53 and PTEN in determining clonal and subclonal evolution in multiple tumour types. We propose that the hypoxic tumour microenvironment selects for unstable tumour clones which survive, propagate and metastasize under reduced immune surveillance. These aggressive features of hypoxic tumour cells underpin resistance to local and systemic therapies and unfavourable outcomes for patients with cancer. Possible ways to counter the effects of hypoxia to block tumour evolution and improve treatment outcomes are described.

Spatial omics shed light on the tumour organisation of glioblastoma

The glioblastoma tumour microenvironment is characterised by immense heterogeneity, with malignant and non-malignant cells that interact in a complex ecosystem. Emerging evidence suggests that the tumour microenvironment is key in facilitating rapid proliferation, invasion, migration and cancer cell survival, crucial for treatment resistance. Spatial omics technologies have enabled the molecular characterisation of regions or individual cells within their spatial context, providing previously unattainable insights into the complex organisation of the glioblastoma tumour microenvironment. Understanding this organisation is crucial for the development of new therapeutics and novel diagnostic tools that guide patient care. This review explores spatial omics technologies and how they have contributed to the development of a model outlining the architecture of the glioblastoma tumour microenvironment.

Paediatric strategy forum for medicinal product development in diffuse midline gliomas in children and adolescents ACCELERATE in collaboration with the European Medicines Agency with participation of the Food and Drug Administration

Fewer than 10 % of children with diffuse midline glioma (DMG) survive 2 years from diagnosis. Radiation therapy remains the cornerstone of treatment and there are no medicinal products with regulatory approval. Although the biology of DMG is better characterized, this has not yet translated into effective treatments. H3K27-alterations initiate the disease but additional drivers are required for malignant growth. Hence, there is an urgent unmet need to develop new multi-modality therapeutic strategies, including alternative methods of drug delivery. ONC201 (DRD2 antagonist and mitochondrial ClpP agonist) is the most widely evaluated investigational drug. Encouraging early data is emerging for CAR T-cells and oncolytic viruses. GD2, B7-H3 and PI3K signalling are ubiquitous targets across all subtypes and therapeutics directed to these targets would potentially benefit the largest number of children. PI3K, ACVR1, MAPK and PDGFRA pathways should be targeted in rational biological combinations. Drug discovery is a very high priority. New specific and potent epigenetic modifiers (PROTACS e.g. SMARCA4 degraders), with blood-brain penetrance are needed. Cancer neuroscience therapeutics are in early development. Overall survival is the preferred regulatory endpoint. However, the evaluation of this can be influenced by the use of re-irradiation at the time of progression. An efficient clinical trial design fit for regulatory purposes for the evaluation of new therapeutics would aid industry and facilitate more efficient therapy development. Challenges in conducting clinical trials such as the need for comparator data and defining endpoints, could be addressed through an international, first-in-child, randomised, complex innovative design trial. To achieve progress: i) drug discovery; ii) new multi-modality, efficient, collaborative, pre-clinical approaches, possibly including artificial intelligence and, iii) efficient clinical trial designs fit for regulatory purposes are required.

An integrated perspective on single-cell and spatial transcriptomic signatures in high-grade gliomas

High-grade gliomas (HGG) are incurable brain malignancies in children and adults. Breakthrough advances in transcriptomic technologies unveiled the intricate diversity of cellular states and their spatial organization within HGGs. We qualitatively integrated 55 neoplastic transcriptomic signatures described in 17 single-cell and spatial RNA sequencing-based studies. Our review delineates a spectrum of cellular states, represented by the expression of specific genes, which can be conceptualized along a "reactive-developmental programs" axis. Additionally, we discussed the potential cues influencing these cellular states, including how spatial organization may impact transcriptomic dynamics. Leveraging these insightful discoveries, we discussed a novel, evolutive way to integrate the different transcriptomic signatures in two or three dimensions, incorporating developmental states, their proliferative capacity, and their possible transition towards reactive states. This integrated analysis illuminates the diverse cellular landscape of HGGs and provides a valuable resource for further elucidation of malignant mechanisms, and for the design of therapeutic endeavors.

Peripheral blood-derived immune cell counts as prognostic indicators and their relationship with DNA methylation subclasses in glioblastoma patients

Glioblastomas are known for their immunosuppressive tumor microenvironment, which may explain the failure of most clinical trials in the past decade. Recent studies have emphasized the significance of stratifying glioblastoma patients to predict better therapeutic responses and survival outcomes. This study aims to investigate the prognostic relevance of peripheral immune cell counts sampled prior to surgery, with a special focus on methylation-based subclassification. Peripheral blood was sampled in patients with newly diagnosed (n = 176) and recurrent (n = 41) glioblastoma at the time of surgery and analyzed for neutrophils, monocytes, leukocytes, platelets, neutrophil-lymphocyte ratio, lymphocyte-monocyte ratio, and platelet-lymphocyte ratio. Peripheral immune cell counts were correlated with patients' survival after combined radiochemotherapy. In addition, 850 k genome-wide DNA methylation was assessed on tissue for defining tumor subclasses and performing cell-type deconvolution. In newly diagnosed glioblastoma, patients with higher peripheral neutrophil counts had an unfavorable overall survival (OS) (p = 0.01, median overall-survival (mOS) 17.0 vs. 10.0 months). At the time of first recurrence, a significant decrease of peripheral immune cell counts was observed, and elevated monocyte (p = 0.03), neutrophil (p = 0.04), and platelet (p = 0.01) counts were associated with poorer survival outcomes. DNA methylation subclass-stratified analysis revealed a significant survival influence of neutrophils (p = 0.007) and lymphocytes (p = 0.04) in the mesenchymal (MES) subclass. Integrating deconvolution of matched tumor tissue showed that platelets and monocytes were correlated with a more differentiated, tumor-progressive cell state, and peripheral immune cell counts were most accurately reflected in tissue of the MES subclass. This study illustrates a restricted prognostic significance of peripheral immune cell counts in newly diagnosed glioblastoma and a constrained representation in matched tumor tissue, but it demonstrates a more pertinent situation at the time of recurrence and after DNA methylation-based stratification.

Reliability-enhanced data cleaning in biomedical machine learning using inductive conformal prediction

Accurately labeling large datasets is important for biomedical machine learning yet challenging while modern data augmentation methods may generate noise in the training data, which may deteriorate machine learning model performance. Existing approaches addressing noisy training data typically rely on strict modeling assumptions, classification models and well-curated dataset. To address these, we propose a novel reliability-based training-data-cleaning method employing inductive conformal prediction (ICP). This method uses a small set of well-curated training data and leverages ICP-calculated reliability metrics to selectively correct mislabeled data and outliers within vast quantities of noisy training data. The efficacy is validated across three classification tasks with distinct modalities: filtering drug-induced-liver-injury (DILI) literature with free-text title and abstract, predicting ICU admission of COVID-19 patients through CT radiomics and electronic health records, and subtyping breast cancer using RNA-sequencing data. Varying levels of noise to the training labels were introduced via label permutation. Our training-data-cleaning method significantly enhanced the downstream classification performance (paired t-tests, p ≤ 0 . 05 among 30 random train/test partitions): significant accuracy enhancement in 86 out of 96 DILI experiments (up to 11.4% increase from 0.812 to 0.905), significant AUROC and AUPRC enhancements in all 48 COVID-19 experiments (up to 23.8% increase from 0.597 to 0.739 for AUROC, and 69.8% increase from 0.183 to 0.311 for AUPRC), and significant accuracy and macro-average F1-score improvements in 47 out of 48 RNA-sequencing experiments (up to 74.6% increase from 0.351 to 0.613 for accuracy, and 89.0% increase from 0.267 to 0.505 for F1-score). The improvement can be both statistically and clinically significant for information retrieval, disease diagnosis and prognosis. The method offers the potential to substantially boost classification performance in biomedical machine learning tasks without necessitating an excessive volume of well-curated training data or strong data distribution and modeling assumptions in existing semi-supervised learning methods.

Spatially resolved transcriptomics and graph-based deep learning improve accuracy of routine CNS tumor diagnostics

The diagnostic landscape of brain tumors integrates comprehensive molecular markers alongside traditional histopathological evaluation. DNA methylation and next-generation sequencing (NGS) have become a cornerstone in central nervous system (CNS) tumor classification. A limiting requirement for NGS and methylation profiling is sufficient DNA quality and quantity, which restrict its feasibility. Here we demonstrate NePSTA (neuropathology spatial transcriptomic analysis) for comprehensive morphological and molecular neuropathological diagnostics from single 5-µm tissue sections. NePSTA uses spatial transcriptomics with graph neural networks for automated histological and molecular evaluations. Trained and evaluated across 130 participants with CNS malignancies and healthy donors across four medical centers, NePSTA predicts tissue histology and methylation-based subclasses with high accuracy. We demonstrate the ability to reconstruct immunohistochemistry and genotype profiling on tissue with minimal requirements, inadequate for conventional molecular diagnostics, demonstrating the potential to enhance tumor subtype identification with implications for fast and precise diagnostic workup.

Network analyses of brain tumor multiomic data reveal pharmacological opportunities to alter cell state transitions

Glioblastoma Multiforme (GBM) remains a particularly difficult cancer to treat, and survival outcomes remain poor. In addition to the lack of dedicated drug discovery programs for GBM, extensive intratumor heterogeneity and epigenetic plasticity related to cell-state transitions are major roadblocks to successful drug therapy in GBM. To study these phenomenon, publicly available snRNAseq and bulk RNAseq data from patient samples were used to categorize cells from patients into four cell states (i.e., phenotypes), namely: (i) neural progenitor-like (NPC-like), (ii) oligodendrocyte progenitor-like (OPC-like), (iii) astrocyte-like (AC-like), and (iv) mesenchymal-like (MES-like). Patients were subsequently grouped into subpopulations based on which cell-state was the most dominant in their respective tumor. By incorporating phosphoproteomic measurements from the same patients, a protein-protein interaction network (PPIN) was constructed for each cell state. These four-cell state PPINs were pooled to form a single Boolean network that was used for in silico protein knockout simulations to investigate mechanisms that either promote or prevent cell state transitions. Simulation results were input into a boosted tree machine learning model which predicted the cell states or phenotypes of GBM patients from an independent public data source, the Glioma Longitudinal Analysis (GLASS) Consortium. Combining the simulation results and the machine learning predictions, we generated hypotheses for clinically relevant causal mechanisms of cell state transitions. For example, the transcription factor TFAP2A can be seen to promote a transition from the NPC-like to the MES-like state. Such protein nodes and the associated signaling pathways provide potential drug targets that can be further tested in vitro and support cell state-directed (CSD) therapy.

Potassium ion channel modulation at cancer-neural interface enhances neuronal excitability in epileptogenic glioblastoma multiforme

The central nervous system (CNS) is increasingly recognized as a critical modulator in the oncogenesis of glioblastoma multiforme (GBM), with interactions between cancer and local neuronal circuits frequently leading to epilepsy; however, the relative contributions of these factors remain unclear. Here, we report a coordinated intratumor shift among distinct cancer subtypes within progenitor-like families of epileptic GBM patients, revealing an accumulation of oligodendrocyte progenitor (OPC)-like subpopulations at the cancer-neuron interface along with heightened electrical signaling activity in the surrounding neuronal networks. The OPC-like cells associated with epilepsy express KCND2, which encodes the voltage-gated K+ channel KV4.2, enhancing neuronal excitability via accumulation of extracellular K+, as demonstrated in patient-derived ex vivo slices, xenografting models, and engineering organoids. Together, we uncovered the essential local circuitry, cellular components, and molecular mechanisms facilitating cancer-neuron interaction at peritumor borders. KCND2 plays a crucial role in mediating nervous system-cancer electrical communication, suggesting potential targets for intervention.

Genetically Engineered Brain Organoids Recapitulate Spatial and Developmental States of Glioblastoma Progression

Glioblastoma (GBM) is an aggressive form of brain cancer that is highly resistant to therapy due to significant intra-tumoral heterogeneity. The lack of robust in vitro models to study early tumor progression has hindered the development of effective therapies. Here, this study develops engineered GBM organoids (eGBOs) harboring GBM subtype-specific oncogenic mutations to investigate the underlying transcriptional regulation of tumor progression. Single-cell and spatial transcriptomic analyses revealed that these mutations disrupt normal neurodevelopment gene regulatory networks resulting in changes in cellular composition and spatial organization. Upon xenotransplantation into immunodeficient mice, eGBOs form tumors that recapitulate the transcriptional and spatial landscape of human GBM samples. Integrative single-cell trajectory analysis of both eGBO-derived tumor cells and patient GBM samples reveal the dynamic gene expression changes in developmental cell states underlying tumor progression. This analysis of eGBOs provides an important validation of engineered cancer organoid models and demonstrates their utility as a model of GBM tumorigenesis for future preclinical development of therapeutics.

Multidisciplinary quantitative and qualitative assessment of IDH-mutant gliomas with full diagnostic deep learning image reconstruction

Rationale and Objectives: Diagnostic accuracy and therapeutic decision-making for IDH-mutant gliomas in tumor board reviews are based on MRI and multidisciplinary interactions.

Materials and Methods

This study explores the feasibility of deep learning-based reconstruction (DLR) in MRI for IDH-mutant gliomas. The research utilizes a multidisciplinary approach, engaging neuroradiologists, neurosurgeons, neuro-oncologists, and radiotherapists to evaluate qualitative aspects of DLR and conventional reconstructed (CR) sequences. Furthermore, quantitative image quality and tumor volumes according to Response Assessment in Neuro-Oncology (RANO) 2.0 standards were assessed.

Results

All DLR sequences consistently outperformed CR sequences (median of 4 for all) in qualitative image quality across all raters (p < 0.001 for all) and revealed higher SNR and CNR values (p < 0.001 for all). Preference for all DLR over CR was overwhelming, with ratings of 84 % from the neuroradiologist, 100 % from the neurosurgeon, 92 % from the neuro-oncologist, and 84 % from the radiation oncologist. The RANO 2.0 compliant measurements showed no significant difference between the CR and DRL sequences (p = 0.142).

Conclusion

This study demonstrates the clinical feasibility of DLR in MR imaging of IDH-mutant gliomas, with significant time savings of 29.6 % on average and non-inferior image quality to CR. DLR sequences received strong multidisciplinary preference, underscoring their potential for enhancing neuro-oncological decision-making and suitability for clinical implementation.

Digital profiling of gene expression from histology images with linearized attention

Cancer is a heterogeneous disease requiring costly genetic profiling for better understanding and management. Recent advances in deep learning have enabled cost-effective predictions of genetic alterations from whole slide images (WSIs). While transformers have driven significant progress in non-medical domains, their application to WSIs lags behind due to high model complexity and limited dataset sizes. Here, we introduce SEQUOIA, a linearized transformer model that predicts cancer transcriptomic profiles from WSIs. SEQUOIA is developed using 7584 tumor samples across 16 cancer types, with its generalization capacity validated on two independent cohorts comprising 1368 tumors. Accurately predicted genes are associated with key cancer processes, including inflammatory response, cell cycles and metabolism. Further, we demonstrate the value of SEQUOIA in stratifying the risk of breast cancer recurrence and in resolving spatial gene expression at loco-regional levels. SEQUOIA hence deciphers clinically relevant information from WSIs, opening avenues for personalized cancer management.

Evolution of Molecular Biomarkers and Precision Molecular Therapeutic Strategies in Glioblastoma

Glioblastoma is the most commonly occurring malignant brain tumor, with a high mortality rate despite current treatments. Its classification has evolved over the years to include not only histopathological features but also molecular findings. Given the heterogeneity of glioblastoma, molecular biomarkers for diagnosis have become essential for initiating treatment with current therapies, while new technologies for detecting specific variations using computational tools are being rapidly developed. Advances in molecular genetics have made possible the creation of tailored therapies based on specific molecular targets, with various degrees of success. This review provides an overview of the latest advances in the fields of histopathology and radiogenomics and the use of molecular markers for management of glioblastoma, as well as the development of new therapies targeting the most common molecular markers. Furthermore, we offer a summary of the results of recent preclinical and clinical trials to recognize the current trends of investigation and understand the possible future directions of molecular targeted therapies in glioblastoma.

Radiomic Features as Artificial Intelligence Prognostic Models in Glioblastoma: A Systematic Review and Meta-Analysis

BACKGROUND

Glioblastoma, the predominant primary tumor among all central nervous systems, accounts for around 80% of cases. Prognosis in neuro-oncology involves assessing the disease's progression in different individuals, considering the time between the initial pathological diagnosis and the time until the disease worsens. A noninvasive therapeutic approach called radiomic features (RFs), which involves the application of artificial intelligence in MRI, has been developed to address this issue. This study aims to systematically gather evidence and evaluate the prognosis significance of radiomics in glioblastoma using RFs.

METHODS

We conducted an extensive search across the PubMed, ScienceDirect, EMBASE, Web of Science, and Cochrane databases to identify relevant original studies examining the use of RFs to evaluate the prognosis of patients with glioblastoma. This thorough search was completed on 25 July 2024. Our search terms included glioblastoma, MRI, magnetic resonance imaging, radiomics, and survival or prognosis. We included only English-language studies involving human subjects, excluding case reports, case series, and review studies. The studies were classified into two quality categories: those rated 4-6 were considered moderate-, whereas those rated 7-9 were high-quality using the Newcastle-Ottawa Scale (NOS). Hazard ratios (HRs) and their 95% confidence intervals (CIs) for OS and PFS were combined using random effects models.

RESULTS

In total, 253 studies were found in the initial search across the five databases. After screening the articles, 40 were excluded due to not meeting the eligibility criteria, and we included only 14 studies. All twelve OS and eight PFS trials were considered, involving 1.639 and 747 patients, respectively. The random effects model was used to calculate the pooled HRs for OS and PFS. The HR for OS was 3.59 (95% confidence interval [CI], 1.80-7.17), while the HR for PFS was 4.20 (95% CI, 1.02-17.32).

CONCLUSIONS

An RF-AI-based approach offers prognostic significance for OS and PFS in patients with glioblastoma.

Integration of multi-omics data revealed the orphan CpG islands and enhancer-dominated c is-regulatory network in glioma

The complex transcriptional regulatory network leads to the poor prognosis of glioma. The role of orphan CpG islands (oCGIs) in the transcriptional regulatory network has been overlooked. We conducted a comprehensive exploration of the cis-regulatory roles of oCGIs and enhancers by integrating multi-omics data. Direct regulation of target genes by oCGIs or enhancers is of great importance in the cis-regulatory network. Furthermore, based on single-cell multi-omics data, we found that the highly activated cis-regulatory network in cluster 2 (C2) sustains the high proliferative potential of glioma cells. The upregulation of oCGIs and enhancers related genes in C2 results in glioma patients exhibiting resistance to radiotherapy and chemotherapy. These findings were further validated through glioma cell line related experiments. Our study offers insight into the pathogenesis of glioma and provides a strategy to treat this challenging disease.

Spatiotemporal modeling reveals high-resolution invasion states in glioblastoma

BACKGROUND

Diffuse invasion of glioblastoma cells through normal brain tissue is a key contributor to tumor aggressiveness, resistance to conventional therapies, and dismal prognosis in patients. A deeper understanding of how components of the tumor microenvironment (TME) contribute to overall tumor organization and to programs of invasion may reveal opportunities for improved therapeutic strategies.

RESULTS

Towards this goal, we apply a novel computational workflow to a spatiotemporally profiled GBM xenograft cohort, leveraging the ability to distinguish human tumor from mouse TME to overcome previous limitations in the analysis of diffuse invasion. Our analytic approach, based on unsupervised deconvolution, performs reference-free discovery of cell types and cell activities within the complete GBM ecosystem. We present a comprehensive catalogue of 15 tumor cell programs set within the spatiotemporal context of 90 mouse brain and TME cell types, cell activities, and anatomic structures. Distinct tumor programs related to invasion align with routes of perivascular, white matter, and parenchymal invasion. Furthermore, sub-modules of genes serving as program network hubs are highly prognostic in GBM patients.

CONCLUSION

The compendium of programs presented here provides a basis for rational targeting of tumor and/or TME components. We anticipate that our approach will facilitate an ecosystem-level understanding of the immediate and long-term consequences of such perturbations, including the identification of compensatory programs that will inform improved combinatorial therapies.

An initial game-theoretic assessment of enhanced tissue preparation and imaging protocols for improved deep learning inference of spatial transcriptomics from tissue morphology

The application of deep learning to spatial transcriptomics (ST) can reveal relationships between gene expression and tissue architecture. Prior work has demonstrated that inferring gene expression from tissue histomorphology can discern these spatial molecular markers to enable population scale studies, reducing the fiscal barriers associated with large-scale spatial profiling. However, while most improvements in algorithmic performance have focused on improving model architectures, little is known about how the quality of tissue preparation and imaging can affect deep learning model training for spatial inference from morphology and its potential for widespread clinical adoption. Prior studies for ST inference from histology typically utilize manually stained frozen sections with imaging on non-clinical grade scanners. Training such models on ST cohorts is also costly. We hypothesize that adopting tissue processing and imaging practices that mirror standards for clinical implementation (permanent sections, automated tissue staining, and clinical grade scanning) can significantly improve model performance. An enhanced specimen processing and imaging protocol was developed for deep learning-based ST inference from morphology. This protocol featured the Visium CytAssist assay to permit automated hematoxylin and eosin staining (e.g. Leica Bond), 40×-resolution imaging, and joining of multiple patients' tissue sections per capture area prior to ST profiling. Using a cohort of 13 pathologic T Stage-III stage colorectal cancer patients, we compared the performance of models trained on slide prepared using enhanced versus traditional (i.e. manual staining and low-resolution imaging) protocols. Leveraging Inceptionv3 neural networks, we predicted gene expression across serial, histologically-matched tissue sections using whole slide images (WSI) from both protocols. The data Shapley was used to quantify and compare marginal performance gains on a patient-by-patient basis attributed to using the enhanced protocol versus the actual costs of spatial profiling. Findings indicate that training and validating on WSI acquired through the enhanced protocol as opposed to the traditional method resulted in improved performance at lower fiscal cost. In the realm of ST, the enhancement of deep learning architectures frequently captures the spotlight; however, the significance of specimen processing and imaging is often understated. This research, informed through a game-theoretic lens, underscores the substantial impact that specimen preparation/imaging can have on spatial transcriptomic inference from morphology. It is essential to integrate such optimized processing protocols to facilitate the identification of prognostic markers at a larger scale.

Radiation immunodynamics in patients with glioblastoma receiving chemoradiation

Introduction

This is a prospective, rigorous inquiry into the systemic immune effects of standard adjuvant chemoradiotherapy, for WHO grade 4, glioblastoma. The purpose is to identify peripheral immunologic effects never yet reported in key immune populations, including myeloid-derived suppressor cells, which are critical to the immune suppressive environment of glioblastoma. We hypothesize that harmful immune-supportive white blood cells, myeloid derived suppressor cells, expand in response to conventionally fractionated radiotherapy with concurrent temozolomide, essentially promoting systemic immunity similar what is seen in chronic diseases like diabetes and heart disease.

Methods

16 patients were enrolled in a single-institution, observational, immune surveillance study where peripheral blood was collected and interrogated by flow cytometry and RNAseq. Tumor tissue from baseline assessment was analyzed with spatial proteomics to link peripheral blood findings to baseline tissue characteristics.

Results

We identified an increase in myeloid-derived suppressor cells during the final week of a six-week treatment of chemoradiotherapy in peripheral blood of patients that were not alive at two years after diagnosis compared to those who were living. This was also associated with a decrease in CD8+ T lymphocytes that produced IFNγ, the potent anti-tumor cytokine.

Discussion

These data suggest that, as in chronic inflammatory disease, systemic immunity is impaired following delivery of adjuvant chemoradiotherapy. Finally, baseline investigation of myeloid cells within tumor tissue did not differ between survival groups, indicating immune surveillance of peripheral blood during adjuvant therapy may be a critical missing link to educate our understanding of the immune effects of standard of care therapy for glioblastoma.

Digital Spatial Profiling identifies distinct patterns of immuno-oncology-related gene expression within oropharyngeal tumours in relation to HPV and p16 status

Background

The incidence of oropharyngeal cancer (OPC) is increasing, due mainly to a rise in Human Papilloma Virus (HPV)-mediated disease. HPV-mediated OPC has significantly better prognosis compared with HPV-negative OPC, stimulating interest in treatment de-intensification approaches to reduce long-term sequelae. Routine clinical testing frequently utilises immunohistochemistry to detect upregulation of p16 as a surrogate marker of HPV-mediation. However, this does not detect discordant p16-/HPV+ cases and incorrectly assigns p16+/HPV- cases, which, given their inferior prognosis compared to p16+/HPV+, may have important clinical implications. The biology underlying poorer prognosis of p16/HPV discordant OPC requires exploration.

Methods

GeoMx digital spatial profiling was used to compare the expression patterns of selected immuno-oncology-related genes/gene families (n=73) within the tumour and stromal compartments of formalin-fixed, paraffin-embedded OPC tumour tissues (n=12) representing the three subgroups, p16+/HPV+, p16+/HPV- and p16-/HPV-.

Results

Keratin (multi KRT) and HIF1A, a key regulator of hypoxia adaptation, were upregulated in both p16+/HPV- and p16-/HPV- tumours relative to p16+/HPV+. Several genes associated with tumour cell proliferation and survival (CCND1, AKT1 and CD44) were more highly expressed in p16-/HPV- tumours relative to p16+/HPV+. Conversely, multiple genes with potential roles in anti-tumour immune responses (immune cell recruitment/trafficking, antigen processing and presentation), such as CXCL9, CXCL10, ITGB2, PSMB10, CD74, HLA-DRB and B2M, were more highly expressed in the tumour and stromal compartments of p16+/HPV+ OPC versus p16-/HPV- and p16+/HPV-. CXCL9 was the only gene showing significant differential expression between p16+/HPV- and p16-/HPV- tumours being upregulated within the stromal compartment of the former.

Conclusions

In terms of immune-oncology-related gene expression, discordant p16+/HPV- OPCs are much more closely aligned with p16-/HPV-OPCs and quite distinct from p16+/HPV+ tumours. This is consistent with previously described prognostic patterns (p16+/HPV+ >> p16+/HPV- > p16-/HPV-) and underlines the need for dual p16 and HPV testing to guide clinical decision making.

Deep learning-based multimodal spatial transcriptomics analysis for cancer

The advent of deep learning (DL) and multimodal spatial transcriptomics (ST) has revolutionized cancer research, offering unprecedented insights into tumor biology. This book chapter explores the integration of DL with ST to advance cancer diagnostics, treatment planning, and precision medicine. DL, a subset of artificial intelligence, employs neural networks to model complex patterns in vast datasets, significantly enhancing diagnostic and treatment applications. In oncology, convolutional neural networks excel in image classification, segmentation, and tumor volume analysis, essential for identifying tumors and optimizing radiotherapy. The chapter also delves into multimodal data analysis, which integrates genomic, proteomic, imaging, and clinical data to offer a holistic understanding of cancer biology. Leveraging diverse data sources, researchers can uncover intricate details of tumor heterogeneity, microenvironment interactions, and treatment responses. Examples include integrating MRI data with genomic profiles for accurate glioma grading and combining proteomic and clinical data to uncover drug resistance mechanisms. DL's integration with multimodal data enables comprehensive and actionable insights for cancer diagnosis and treatment. The synergy between DL models and multimodal data analysis enhances diagnostic accuracy, personalized treatment planning, and prognostic modeling. Notable applications include ST, which maps gene expression patterns within tissue contexts, providing critical insights into tumor heterogeneity and potential therapeutic targets. In summary, the integration of DL and multimodal ST represents a paradigm shift towards more precise and personalized oncology. This chapter elucidates the methodologies and applications of these advanced technologies, highlighting their transformative potential in cancer research and clinical practice.

Super-enhancer-driven LIF promotes the mesenchymal transition in glioblastoma by activating ITGB2 signaling feedback in microglia

BACKGROUND

The mesenchymal (MES) subtype of glioblastoma (GBM) is believed to be influenced by both cancer cell-intrinsic alterations and extrinsic cellular interactions, yet the underlying mechanisms remain unexplored.

METHODS

Identification of microglial heterogeneity by bioinformatics analysis. Transwell migration, invasion assays, and tumor models were used to determine gene function and the role of small molecule inhibitors. RNA sequencing, chromatin immunoprecipitation, and dual-luciferase reporter assays were performed to explore the underlying regulatory mechanisms.

RESULTS

We identified the inflammatory microglial subtype of tumor-associated microglia (TAM) and found that its specific gene integrin beta 2 (ITGB2) was highly expressed in TAM of MES GBM tissues. Mechanistically, the activation of ITGB2 in microglia promoted the interaction between the SH2 domain of STAT3 and the cytoplasmic domain of ITGB2, thereby stimulating the JAK1/STAT3/IL-6 signaling feedback to promote the MES transition of GBM cells. Additionally, microglia communicated with GBM cells through the interaction between the receptor ITGB2 on microglia and the ligand ICAM-1 on GBM cells, while an increased secretion of ICAM-1 was induced by the proinflammatory cytokine leukemia inhibitory factor (LIF). Further studies demonstrated that inhibition of cyclin-dependent kinase 7 substantially reduced the recruitment of SNW1 to the super-enhancer of LIF, resulting in transcriptional inhibition of LIF. We identified notoginsenoside R1 as a novel LIF inhibitor that exhibited synergistic effects in combination with temozolomide.

CONCLUSIONS

Our research reveals that the epigenetic-mediated interaction of GBM cells with TAM drives the MES transition of GBM and provides a novel therapeutic avenue for patients with MES GBM.

Modeling glioblastoma tumor progression via CRISPR-engineered brain organoids

Glioblastoma (GBM) is an aggressive form of brain cancer that is highly resistant to therapy due to significant intra-tumoral heterogeneity. The lack of robust in vitro models to study early tumor progression has hindered the development of effective therapies. Here, we develop engineered GBM organoids (eGBOs) harboring GBM subtype-specific oncogenic mutations to investigate the underlying transcriptional regulation of tumor progression. Single-cell and spatial transcriptomic analyses revealed that these mutations disrupt normal neurodevelopment gene regulatory networks resulting in changes in cellular composition and spatial organization. Upon xenotransplantation into immunodeficient mice, eGBOs form tumors that recapitulate the transcriptional and spatial landscape of human GBM samples. Integrative single-cell trajectory analysis of both eGBO-derived tumor cells and patient GBM samples revealed the dynamic gene expression changes in developmental cell states underlying tumor progression. This analysis of eGBOs provides an important validation of engineered cancer organoid models and demonstrates their utility as a model of GBM tumorigenesis for future preclinical development of therapeutics.

Advancing precision medicine in gliomas through single-cell sequencing: unveiling the complex tumor microenvironment

Glioblastoma (GBM) displays an infiltrative growth characteristic that recruits neighboring normal cells to facilitate tumor growth, maintenance, and invasion into the brain. While the blood-brain barrier serves as a critical natural defense mechanism for the central nervous system, GBM disrupts this barrier, resulting in the infiltration of macrophages from the peripheral bone marrow and the activation of resident microglia. Recent advancements in single-cell transcriptomics and spatial transcriptomics have refined the categorization of cells within the tumor microenvironment for precise identification. The intricate interactions and influences on cell growth within the tumor microenvironment under multi-omics conditions are succinctly outlined. The factors and mechanisms involving microglia, macrophages, endothelial cells, and T cells that impact the growth of GBM are individually examined. The collaborative mechanisms of tumor cell-immune cell interactions within the tumor microenvironment synergistically promote the growth, infiltration, and metastasis of gliomas, while also influencing the immune status and therapeutic response of the tumor microenvironment. As immunotherapy continues to progress, targeting the cells within the inter-tumor microenvironment emerges as a promising novel therapeutic approach for GBM. By comprehensively understanding and intervening in the intricate cellular interactions within the tumor microenvironment, novel therapeutic modalities may be developed to enhance treatment outcomes for patients with GBM.

Network-Based Transcriptome Analysis Reveals FAM3C as a Novel Potential Biomarker for Glioblastoma

Glioblastoma (GBM) is the most common form of malignant primary brain tumor with a high mortality rate. The aim of the present study was to investigate the clinical significance of Family with Sequence Similarity 3, Member C, FAM3C, in GBM using bioinformatic-integrated analysis. First, we performed the transcriptomic integration analysis to assess the expression profile of FAM3C in GBM using several data sets (RNA-sequencing and scRNA-sequencing), which were obtained from TCGA and GEO databases. By using the STRING platform, we investigated FAM3C-coregulated genes to construct the protein-protein interaction network. Next, Metascape, Enrichr, and CIBERSORT databases were used. We found FAM3C high expression in GBM with poor survival rates. Further, we observed, via FAM3C coexpression network analysis, that FAM3C plays key roles in several hallmarks of cancer. Surprisingly, we also highlighted five FAM3C‑coregulated genes overexpressed in GBM. Specifically, we demonstrated the association between the high expression of FAM3C and the abundance of the different immune cells, which may markedly worsen GBM prognosis. For the first time, our findings suggest that FAM3C not only can be a new emerging biomarker with promising therapeutic values to GBM patients but also gave a new insight into a potential resource for future GBM studies.

A Self-Cascade Penetrating Brain Tumor Immunotherapy Mediated by Near-Infrared II Cell Membrane-Disrupting Nanoflakes via Detained Dendritic Cells

Immunotherapy can potentially suppress the highly aggressive glioblastoma (GBM) by promoting T lymphocyte infiltration. Nevertheless, the immune privilege phenomenon, coupled with the generally low immunogenicity of vaccines, frequently hampers the presence of lymphocytes within brain tumors, particularly in brain tumors. In this study, the membrane-disrupted polymer-wrapped CuS nanoflakes that can penetrate delivery to deep brain tumors via releasing the cell-cell interactions, facilitating the near-infrared II (NIR II) photothermal therapy, and detaining dendritic cells for a self-cascading immunotherapy are developed. By convection-enhanced delivery, membrane-disrupted amphiphilic polymer micelles (poly(methoxypoly(ethylene glycol)-benzoic imine-octadecane, mPEG-b-C18) with CuS nanoflakes enhances tumor permeability and resides in deep brain tumors. Under low-power NIR II irradiation (0.8 W/cm2), the intense heat generated by well-distributed CuS nanoflakes actuates the thermolytic efficacy, facilitating cell apoptosis and the subsequent antigen release. Then, the positively charged polymer after hydrolysis of the benzoic-imine bond serves as an antigen depot, detaining autologous tumor-associated antigens and presenting them to dendritic cells, ensuring sustained immune stimulation. This self-cascading penetrative immunotherapy amplifies the immune response to postoperative brain tumors but also enhances survival outcomes through effective brain immunotherapy.

Glioblastoma and brain connectivity: the need for a paradigm shift

Despite substantial advances in cancer treatment, for patients with glioblastoma prognosis remains bleak. The emerging field of cancer neuroscience reveals intricate functional interplays between glioblastoma and the cellular architecture of the brain, encompassing neurons, glia, and vessels. New findings underscore the role of structural and functional connections within hierarchical networks, known as the connectome. These connections contribute to the location, spread, and recurrence of a glioblastoma, and a patient's overall survival, revealing a complex interplay between the tumour and the CNS. This mounting evidence prompts a paradigm shift, challenging the perception of glioblastomas as mere foreign bodies within the brain. Instead, these tumours are intricately woven into the structural and functional fabric of the brain. This radical change in thinking holds profound implications for the understanding and treatment of glioblastomas, which could unveil new prognostic factors and surgical strategies and optimise radiotherapy. Additionally, a connectivity approach suggests that non-invasive brain stimulation could disrupt pathological neuron-glioma interactions within specific networks.

Glioblastoma microenvironment-from biology to therapy

Glioblastoma (GBM) is the most aggressive primary brain cancer. These tumors exhibit high intertumoral and intratumoral heterogeneity in neoplastic and nonneoplastic compartments, low lymphocyte infiltration, and high abundance of myeloid subsets that together create a highly protumorigenic immunosuppressive microenvironment. Moreover, heterogeneous GBM cells infiltrate adjacent brain tissue, remodeling the neural microenvironment to foster tumor electrochemical coupling with neurons and metabolic coupling with nonneoplastic astrocytes, thereby driving growth. Here, we review heterogeneity in the GBM microenvironment and its role in low-to-high-grade glioma transition, concluding with a discussion of the challenges of therapeutically targeting the tumor microenvironment and outlining future research opportunities.

Identification of a gene expression signature of vascular invasion and recurrence in stage I lung adenocarcinoma via bulk and spatial transcriptomics

Microscopic vascular invasion (VI) is predictive of recurrence and benefit from lobectomy in stage I lung adenocarcinoma (LUAD) but is difficult to assess in resection specimens and cannot be accurately predicted prior to surgery. Thus, new biomarkers are needed to identify this aggressive subset of stage I LUAD tumors. To assess molecular and microenvironment features associated with angioinvasive LUAD we profiled 162 resected stage I tumors with and without VI by RNA-seq and explored spatial patterns of gene expression in a subset of 15 samples by high-resolution spatial transcriptomics (stRNA-seq). Despite the small size of invaded blood vessels, we identified a gene expression signature of VI from the bulk RNA-seq discovery cohort (n=103) and found that it was associated with VI foci, desmoplastic stroma, and high-grade patterns in our stRNA-seq data. We observed a stronger association with high-grade patterns from VI+ compared with VI- tumors. Using the discovery cohort, we developed a transcriptomic predictor of VI, that in an independent validation cohort (n=60) was associated with VI (AUROC=0.86; p=5.42×10-6) and predictive of recurrence-free survival (HR=1.98; p=0.024), even in VI- LUAD (HR=2.76; p=0.003). To determine our VI predictor's robustness to intra-tumor heterogeneity we used RNA-seq data from multi-region sampling of stage I LUAD cases in TRACERx, where the predictor scores showed high correlation (R=0.87, p<2.2×10-16) between two randomly sampled regions of the same tumor. Our study suggests that VI-associated gene expression changes are detectable beyond the site of intravasation and can be used to predict the presence of VI. This may enable the prediction of angioinvasive LUAD from biopsy specimens, allowing for more tailored medical and surgical management of stage I LUAD.

Reprogramming Glioblastoma Cells into Non-Cancerous Neuronal Cells as a Novel Anti-Cancer Strategy

Glioblastoma Multiforme (GBM) is an aggressive brain tumor with a high mortality rate. Direct reprogramming of glial cells to different cell lineages, such as induced neural stem cells (iNSCs) and induced neurons (iNeurons), provides genetic tools to manipulate a cell's fate as a potential therapy for neurological diseases. NeuroD1 (ND1) is a master transcriptional factor for neurogenesis and it promotes neuronal differentiation. In the present study, we tested the hypothesis that the expression of ND1 in GBM cells can force them to differentiate toward post-mitotic neurons and halt GBM tumor progression. In cultured human GBM cell lines, including LN229, U87, and U373 as temozolomide (TMZ)-sensitive and T98G as TMZ-resistant cells, the neuronal lineage conversion was induced by an adeno-associated virus (AAV) package carrying ND1. Twenty-one days after AAV-ND1 transduction, ND1-expressing cells displayed neuronal markers MAP2, TUJ1, and NeuN. The ND1-induced transdifferentiation was regulated by Wnt signaling and markedly enhanced under a hypoxic condition (2% O2 vs. 21% O2). ND1-expressing GBM cultures had fewer BrdU-positive proliferating cells compared to vector control cultures. Increased cell death was visualized by TUNEL staining, and reduced migrative activity was demonstrated in the wound-healing test after ND1 reprogramming in both TMZ-sensitive and -resistant GBM cells. In a striking contrast to cancer cells, converted cells expressed the anti-tumor gene p53. In an orthotopical GBM mouse model, AAV-ND1-reprogrammed U373 cells were transplanted into the fornix of the cyclosporine-immunocompromised C57BL/6 mouse brain. Compared to control GBM cell-formed tumors, cells from ND1-reprogrammed cultures formed smaller tumors and expressed neuronal markers such as TUJ1 in the brain. Thus, reprogramming using a single-factor ND1 overcame drug resistance, converting malignant cells of heterogeneous GBM cells to normal neuron-like cells in vitro and in vivo. These novel observations warrant further research using patient-derived GBM cells and patient-derived xenograft (PDX) models as a potentially effective treatment for a deadly brain cancer and likely other astrocytoma tumors.

Network Analyses of Brain Tumor Patients' Multiomic Data Reveals Pharmacological Opportunities to Alter Cell State Transitions

Glioblastoma Multiforme (GBM) remains a particularly difficult cancer to treat, and survival outcomes remain poor. In addition to the lack of dedicated drug discovery programs for GBM, extensive intratumor heterogeneity and epigenetic plasticity related to cell-state transitions are major roadblocks to successful drug therapy in GBM. To study these phenomenon, publicly available snRNAseq and bulk RNAseq data from patient samples were used to categorize cells from patients into four cell states (i.e. phenotypes), namely: (i) neural progenitor-like (NPC-like), (ii) oligodendrocyte progenitor-like (OPC-like), (iii) astrocyte- like (AC-like), and (iv) mesenchymal-like (MES-like). Patients were subsequently grouped into subpopulations based on which cell-state was the most dominant in their respective tumor. By incorporating phosphoproteomic measurements from the same patients, a protein-protein interaction network (PPIN) was constructed for each cell state. These four-cell state PPINs were pooled to form a single Boolean network that was used for in silico protein knockout simulations to investigate mechanisms that either promote or prevent cell state transitions. Simulation results were input into a boosted tree machine learning model which predicted the cell states or phenotypes of GBM patients from an independent public data source, the Glioma Longitudinal Analysis (GLASS) Consortium. Combining the simulation results and the machine learning predictions, we generated hypotheses for clinically relevant causal mechanisms of cell state transitions. For example, the transcription factor TFAP2A can be seen to promote a transition from the NPC-like to the MES-like state. Such protein nodes and the associated signaling pathways provide potential drug targets that can be further tested in vitro and support cell state-directed (CSD) therapy.

Expression of the apelin receptor, a novel potential therapeutic target, and its endogenous ligands in diverse stem cell populations in human glioblastoma

Glioblastoma multiforme (GBM) is one of the most common and lethal forms of brain cancer, carrying a very poor prognosis (median survival of ~15 months post-diagnosis). Treatment typically involves invasive surgical resection of the tumour mass, followed by radiotherapy and adjuvant chemotherapy using the alkylating agent temozolomide, but over half of patients do not respond to this drug and considerable resistance is observed. Tumour heterogeneity is the main cause of therapeutic failure, where diverse progenitor glioblastoma stem cell (GSC) lineages in the microenvironment drive tumour recurrence and therapeutic resistance. The apelin receptor is a class A GPCR that binds two endogenous peptide ligands, apelin and ELA, and plays a role in the proliferation and survival of cancer cells. Here, we used quantitative whole slide immunofluorescent imaging of human GBM samples to characterise expression of the apelin receptor and both its ligands in the distinct GSC lineages, namely neural-progenitor-like cells (NPCs), oligodendrocyte-progenitor-like cells (OPCs), and mesenchymal-like cells (MES), as well as reactive astrocytic cells. The data confirm the presence of the apelin receptor as a tractable drug target that is common across the key cell populations driving tumour growth and maintenance, offering a potential novel therapeutic approach for patients with GBM.

Network Analyses of Brain Tumor Patients' Multiomic Data Reveals Pharmacological Opportunities to Alter Cell State Transitions

Glioblastoma Multiforme (GBM) remains a particularly difficult cancer to treat, and survival outcomes remain poor. In addition to the lack of dedicated drug discovery programs for GBM, extensive intratumor heterogeneity and epigenetic plasticity related to cell-state transitions are major roadblocks to successful drug therapy in GBM. To study these phenomenon, publicly available snRNAseq and bulk RNAseq data from patient samples were used to categorize cells from patients into four cell states (i.e. phenotypes), namely: (i) neural progenitor-like (NPC-like), (ii) oligodendrocyte progenitor-like (OPC-like), (iii) astrocyte-like (AC-like), and (iv) mesenchymal-like (MES-like). Patients were subsequently grouped into subpopulations based on which cell-state was the most dominant in their respective tumor. By incorporating phosphoproteomic measurements from the same patients, a protein-protein interaction network (PPIN) was constructed for each cell state. These four-cell state PPINs were pooled to form a single Boolean network that was used for in silico protein knockout simulations to investigate mechanisms that either promote or prevent cell state transitions. Simulation results were input into a boosted tree machine learning model which predicted the cell states or phenotypes of GBM patients from an independent public data source, the Glioma Longitudinal Analysis (GLASS) Consortium. Combining the simulation results and the machine learning predictions, we generated hypotheses for clinically relevant causal mechanisms of cell state transitions. For example, the transcription factor TFAP2A can be seen to promote a transition from the NPC-like to the MES-like state. Such protein nodes and the associated signaling pathways provide potential drug targets that can be further tested in vitro and support cell state-directed (CSD) therapy.

Cellular diversity through space and time: adding new dimensions to GBM therapeutic development

The current median survival for glioblastoma (GBM) patients is only about 16 months, with many patients succumbing to the disease in just a matter of months, making it the most common and aggressive primary brain cancer in adults. This poor outcome is, in part, due to the lack of new treatment options with only one FDA-approved treatment in the last decade. Advances in sequencing techniques and transcriptomic analyses have revealed a vast degree of heterogeneity in GBM, from inter-patient diversity to intra-tumoral cellular variability. These cutting-edge approaches are providing new molecular insights highlighting a critical role for the tumor microenvironment (TME) as a driver of cellular plasticity and phenotypic heterogeneity. With this expanded molecular toolbox, the influence of TME factors, including endogenous (e.g., oxygen and nutrient availability and interactions with non-malignant cells) and iatrogenically induced (e.g., post-therapeutic intervention) stimuli, on tumor cell states can be explored to a greater depth. There exists a critical need for interrogating the temporal and spatial aspects of patient tumors at a high, cell-level resolution to identify therapeutically targetable states, interactions and mechanisms. In this review, we discuss advancements in our understanding of spatiotemporal diversity in GBM with an emphasis on the influence of hypoxia and immune cell interactions on tumor cell heterogeneity. Additionally, we describe specific high-resolution spatially resolved methodologies and their potential to expand the impact of pre-clinical GBM studies. Finally, we highlight clinical attempts at targeting hypoxia- and immune-related mechanisms of malignancy and the potential therapeutic opportunities afforded by single-cell and spatial exploration of GBM patient specimens.

Artificial intelligence in neurology: opportunities, challenges, and policy implications

Neurological conditions are the leading cause of disability and mortality combined, demanding innovative, scalable, and sustainable solutions. Brain health has become a global priority with adoption of the World Health Organization's Intersectoral Global Action Plan in 2022. Simultaneously, rapid advancements in artificial intelligence (AI) are revolutionizing neurological research and practice. This scoping review of 66 original articles explores the value of AI in neurology and brain health, systematizing the landscape for emergent clinical opportunities and future trends across the care trajectory: prevention, risk stratification, early detection, diagnosis, management, and rehabilitation. AI's potential to advance personalized precision neurology and global brain health directives hinges on resolving core challenges across four pillars-models, data, feasibility/equity, and regulation/innovation-through concerted pursuit of targeted recommendations. Paramount actions include swift, ethical, equity-focused integration of novel technologies into clinical workflows, mitigating data-related issues, counteracting digital inequity gaps, and establishing robust governance frameworks balancing safety and innovation.

Artificial intelligence in neuro-oncology: advances and challenges in brain tumor diagnosis, prognosis, and precision treatment

This review delves into the most recent advancements in applying artificial intelligence (AI) within neuro-oncology, specifically emphasizing work on gliomas, a class of brain tumors that represent a significant global health issue. AI has brought transformative innovations to brain tumor management, utilizing imaging, histopathological, and genomic tools for efficient detection, categorization, outcome prediction, and treatment planning. Assessing its influence across all facets of malignant brain tumor management- diagnosis, prognosis, and therapy- AI models outperform human evaluations in terms of accuracy and specificity. Their ability to discern molecular aspects from imaging may reduce reliance on invasive diagnostics and may accelerate the time to molecular diagnoses. The review covers AI techniques, from classical machine learning to deep learning, highlighting current applications and challenges. Promising directions for future research include multimodal data integration, generative AI, large medical language models, precise tumor delineation and characterization, and addressing racial and gender disparities. Adaptive personalized treatment strategies are also emphasized for optimizing clinical outcomes. Ethical, legal, and social implications are discussed, advocating for transparency and fairness in AI integration for neuro-oncology and providing a holistic understanding of its transformative impact on patient care.

A longitudinal single-cell and spatial multiomic atlas of pediatric high-grade glioma

Pediatric high-grade glioma (pHGG) is an incurable central nervous system malignancy that is a leading cause of pediatric cancer death. While pHGG shares many similarities to adult glioma, it is increasingly recognized as a molecularly distinct, yet highly heterogeneous disease. In this study, we longitudinally profiled a molecularly diverse cohort of 16 pHGG patients before and after standard therapy through single-nucleus RNA and ATAC sequencing, whole-genome sequencing, and CODEX spatial proteomics to capture the evolution of the tumor microenvironment during progression following treatment. We found that the canonical neoplastic cell phenotypes of adult glioblastoma are insufficient to capture the range of tumor cell states in a pediatric cohort and observed differential tumor-myeloid interactions between malignant cell states. We identified key transcriptional regulators of pHGG cell states and did not observe the marked proneural to mesenchymal shift characteristic of adult glioblastoma. We showed that essential neuromodulators and the interferon response are upregulated post-therapy along with an increase in non-neoplastic oligodendrocytes. Through in vitro pharmacological perturbation, we demonstrated novel malignant cell-intrinsic targets. This multiomic atlas of longitudinal pHGG captures the key features of therapy response that support distinction from its adult counterpart and suggests therapeutic strategies which are targeted to pediatric gliomas.

Digital profiling of cancer transcriptomes from histology images with grouped vision attention

Cancer is a heterogeneous disease that demands precise molecular profiling for better understanding and management. Recently, deep learning has demonstrated potentials for cost-efficient prediction of molecular alterations from histology images. While transformer-based deep learning architectures have enabled significant progress in non-medical domains, their application to histology images remains limited due to small dataset sizes coupled with the explosion of trainable parameters. Here, we develop SEQUOIA, a transformer model to predict cancer transcriptomes from whole-slide histology images. To enable the full potential of transformers, we first pre-train the model using data from 1,802 normal tissues. Then, we fine-tune and evaluate the model in 4,331 tumor samples across nine cancer types. The prediction performance is assessed at individual gene levels and pathway levels through Pearson correlation analysis and root mean square error. The generalization capacity is validated across two independent cohorts comprising 1,305 tumors. In predicting the expression levels of 25,749 genes, the highest performance is observed in cancers from breast, kidney and lung, where SEQUOIA accurately predicts the expression of 11,069, 10,086 and 8,759 genes, respectively. The accurately predicted genes are associated with the regulation of inflammatory response, cell cycles and metabolisms. While the model is trained at the tissue level, we showcase its potential in predicting spatial gene expression patterns using spatial transcriptomics datasets. Leveraging the prediction performance, we develop a digital gene expression signature that predicts the risk of recurrence in breast cancer. SEQUOIA deciphers clinically relevant gene expression patterns from histology images, opening avenues for improved cancer management and personalized therapies.

Deep learning links localized digital pathology phenotypes with transcriptional subtype and patient outcome in glioblastoma

BACKGROUND

Deep learning has revolutionized medical image analysis in cancer pathology, where it had a substantial clinical impact by supporting the diagnosis and prognostic rating of cancer. Among the first available digital resources in the field of brain cancer is glioblastoma, the most common and fatal brain cancer. At the histologic level, glioblastoma is characterized by abundant phenotypic variability that is poorly linked with patient prognosis. At the transcriptional level, 3 molecular subtypes are distinguished with mesenchymal-subtype tumors being associated with increased immune cell infiltration and worse outcome.

RESULTS

We address genotype-phenotype correlations by applying an Xception convolutional neural network to a discovery set of 276 digital hematozylin and eosin (H&E) slides with molecular subtype annotation and an independent The Cancer Genome Atlas-based validation cohort of 178 cases. Using this approach, we achieve high accuracy in H&E-based mapping of molecular subtypes (area under the curve for classical, mesenchymal, and proneural = 0.84, 0.81, and 0.71, respectively; P < 0.001) and regions associated with worse outcome (univariable survival model P < 0.001, multivariable P = 0.01). The latter were characterized by higher tumor cell density (P < 0.001), phenotypic variability of tumor cells (P < 0.001), and decreased T-cell infiltration (P = 0.017).

CONCLUSIONS

We modify a well-known convolutional neural network architecture for glioblastoma digital slides to accurately map the spatial distribution of transcriptional subtypes and regions predictive of worse outcome, thereby showcasing the relevance of artificial intelligence-enabled image mining in brain cancer.

A novel immune cell signature for predicting glioblastoma after radiotherapy prognosis and guiding therapy

Background: Glioblastoma, a highly aggressive brain tumor, poses a significant clinical challenge, particularly in the context of radiotherapy. In this study, we aimed to explore infiltrating immune cells and identify immune-related genes associated with glioblastoma radiotherapy prognosis. Subsequently, we constructed a signature based on these genes to discern differences in molecular and tumor microenvironment immune characteristics, ultimately informing potential therapeutic strategies for patients with varying risk profiles. Methods: We leveraged UCSC Xena and CGGA gene expression profiles from post-radiotherapy glioblastoma as verification cohorts. Infiltration ratios were stratified into high and low groups based on the median value. Differential gene expression was determined through Limma differential analysis. A signature comprising four genes was constructed, guided by Gene Ontology (GO) functional enrichment results and Kaplan-Meier survival analysis. We evaluated differences in cell infiltration levels, Immune Score, Stromal Score, and ESTIMATE Score and their Pearson correlations with the signature. Spearman's correlation was computed between the signature and patient drug sensitivity (IC50), predicted using Genomics of Drug Sensitivity in Cancer (GDSC) and CCLE databases. Results: Notably, the infiltration of central memory CD8+T cells exhibited a significant correlation with glioblastoma radiotherapy prognosis. Samples were dichotomized into high- and low-risk groups based on the optimal signature threshold (2.466642). Kaplan-Meier (K-M) survival analysis revealed that the high-risk group experienced a significantly poorer prognosis (p = .0068), with AUC values exceeding 0.82 at 1, 3, and 5 years, underscoring the robust predictive potential of the signature scoring system. Independent validation sets substantiated the validity of the signature. Statistically significant differences in tumor microenvironments (p < .05) were observed between high- and low-risk groups, and these differences were significantly correlated with the signature (p < .05). Furthermore, there were significant correlations between high and low-risk groups regarding immune checkpoint expressions, Immune Prognostic Score (IPS), and Tumor Immune Dysfunction and Exclusion (TIDE) scores. Conclusion: The immune cell signature, comprising SDC-1, PLAUR, FN1, and CXCL13, holds promise as a predictive tool for assessing glioblastoma prognosis following radiotherapy. This signature also offers valuable guidance for tailoring treatment strategies, emphasizing its potential clinical relevance in improving patient outcomes.

Therapeutic Targeting of Glioblastoma and the Interactions with Its Microenvironment

Glioblastoma (GBM) is the most common primary malignant brain tumour, and it confers a dismal prognosis despite intensive multimodal treatments. Whilst historically, research has focussed on the evolution of GBM tumour cells themselves, there is growing recognition of the importance of studying the tumour microenvironment (TME). Improved characterisation of the interaction between GBM cells and the TME has led to a better understanding of therapeutic resistance and the identification of potential targets to block these escape mechanisms. This review describes the network of cells within the TME and proposes treatment strategies for simultaneously targeting GBM cells, the surrounding immune cells, and the crosstalk between them.

Cell phenotypes can be predicted from propensities of protein conformations

Proteins exist as dynamic conformational ensembles. Here we suggest that the propensities of the conformations can be predictors of cell function. The conformational states that the molecules preferentially visit can be viewed as phenotypic determinants, and their mutations work by altering the relative propensities, thus the cell phenotype. Our examples include (i) inactive state variants harboring cancer driver mutations that present active state-like conformational features, as in K-Ras4BG12V compared to other K-Ras4BG12X mutations; (ii) mutants of the same protein presenting vastly different phenotypic and clinical profiles: cancer and neurodevelopmental disorders; (iii) alterations in the occupancies of the conformational (sub)states influencing enzyme reactivity. Thus, protein conformational propensities can determine cell fate. They can also suggest the allosteric drugs efficiency.

Deep topographic proteomics of a human brain tumour

The spatial organisation of cellular protein expression profiles within tissue determines cellular function and is key to understanding disease pathology. To define molecular phenotypes in the spatial context of tissue, there is a need for unbiased, quantitative technology capable of mapping proteomes within tissue structures. Here, we present a workflow for spatially-resolved, quantitative proteomics of tissue that generates maps of protein abundance across tissue slices derived from a human atypical teratoid-rhabdoid tumour at three spatial resolutions, the highest being 40 µm, to reveal distinct abundance patterns of thousands of proteins. We employ spatially-aware algorithms that do not require prior knowledge of the fine tissue structure to detect proteins and pathways with spatial abundance patterns and correlate proteins in the context of tissue heterogeneity and cellular features such as extracellular matrix or proximity to blood vessels. We identify PYGL, ASPH and CD45 as spatial markers for tumour boundary and reveal immune response-driven, spatially-organised protein networks of the extracellular tumour matrix. Overall, we demonstrate spatially-aware deep proteo-phenotyping of tissue heterogeneity, to re-define understanding tissue biology and pathology at the molecular level.

The Overlooked Role of Specimen Preparation in Bolstering Deep Learning-Enhanced Spatial Transcriptomics Workflows

The application of deep learning methods to spatial transcriptomics has shown promise in unraveling the complex relationships between gene expression patterns and tissue architecture as they pertain to various pathological conditions. Deep learning methods that can infer gene expression patterns directly from tissue histomorphology can expand the capability to discern spatial molecular markers within tissue slides. However, current methods utilizing these techniques are plagued by substantial variability in tissue preparation and characteristics, which can hinder the broader adoption of these tools. Furthermore, training deep learning models using spatial transcriptomics on small study cohorts remains a costly endeavor. Necessitating novel tissue preparation processes enhance assay reliability, resolution, and scalability. This study investigated the impact of an enhanced specimen processing workflow for facilitating a deep learning-based spatial transcriptomics assessment. The enhanced workflow leveraged the flexibility of the Visium CytAssist assay to permit automated H&E staining (e.g., Leica Bond) of tissue slides, whole-slide imaging at 40x-resolution, and multiplexing of tissue sections from multiple patients within individual capture areas for spatial transcriptomics profiling. Using a cohort of thirteen pT3 stage colorectal cancer (CRC) patients, we compared the efficacy of deep learning models trained on slide prepared using an enhanced workflow as compared to the traditional workflow which leverages manual tissue staining and standard imaging of tissue slides. Leveraging Inceptionv3 neural networks, we aimed to predict gene expression patterns across matched serial tissue sections, each stemming from a distinct workflow but aligned based on persistent histological structures. Findings indicate that the enhanced workflow considerably outperformed the traditional spatial transcriptomics workflow. Gene expression profiles predicted from enhanced tissue slides also yielded expression patterns more topologically consistent with the ground truth. This led to enhanced statistical precision in pinpointing biomarkers associated with distinct spatial structures. These insights can potentially elevate diagnostic and prognostic biomarker detection by broadening the range of spatial molecular markers linked to metastasis and recurrence. Future endeavors will further explore these findings to enrich our comprehension of various diseases and uncover molecular pathways with greater nuance. Combining deep learning with spatial transcriptomics provides a compelling avenue to enrich our understanding of tumor biology and improve clinical outcomes. For results of the highest fidelity, however, effective specimen processing is crucial, and fostering collaboration between histotechnicians, pathologists, and genomics specialists is essential to herald this new era in spatial transcriptomics-driven cancer research.