Classification of tumors in neuro-oncology today relies on molecular patterns (mostly DNA methylation) and their machine learning-supported interpretation. Understanding the process of algorithmic interpretation is essential for safe application in clinical routine. This is paradigmatically true for the most common primary intracranial tumor in adults, meningioma. Here, by applying multiomic profiling and multiple lines of orthogonal computational evaluation in multiple independent datasets, we found that not only tumor cell characteristics but also incremental changes in the tumor microenvironment (TME) have impact on epigenetic meningioma classification and clinical outcome. Besides revealing the decisive role of non-neoplastic cells in the CNS methylation classifier, this challenges the model of distinct meningioma subgroups toward a TME-determined risk continuum. This refines current controversies in molecular meningioma subtyping. In addition, we apply these learnings to devise and validate a simple diagnostic approach for increased clinical prediction accuracy based on immunohistochemistry, which is also applicable in resource-limited settings.

Background: Classification of central nervous system (CNS) tumors has become increasingly complex, raising concerns about the sustainability of comprehensive molecular diagnostics. We have evaluated nanopore whole genome sequencing (nWGS) as a single workflow to replace multiple diagnostic assays. Methods: We performed nWGS on DNA extracted from 90 adult CNS tumor samples (58 retrospective, 32 prospective) and compared the results to findings from standard of care (SoC) diagnostic work-up. Analysis was done through an automated workflow that consolidated diagnostically and therapeutically relevant genomic alterations, including copy-number variation, structural, and single-nucleotide variants, chromosomal aberrations, gene fusions, and methylation-based classification. Results: nWGS supported final diagnostic classification in all samples with >15% tumor cell content, requiring ~3 hours of hands-on library preparation, parallel sample processing, and sequencing times within 72 hours. Methylation-based classification was available within 1 hour and was concordant with the integrated final diagnosis in 89% of cases (80/90). All diagnostically relevant copy-number variations, single-nucleotide variants, and gene fusions were concordant with SoC testing. MGMT promoter methylation status matched in 94% of cases. In addition, nWGS identified prognostic and potentially actionable variants that were not reported or covered by SoC. Conclusions: nWGS delivers comprehensive genetic and epigenetic results with a fast turn-around compared to standard methods. This enables efficient, accurate, and scalable molecular diagnostics of CNS tumors using a single platform. This data supports its implementation in routine clinical practice and may be extended to other cancer types requiring complex genomic profiling. ### Competing Interest Statement F.S. is a co-founder and shareholder of Heidelberg Epignostix GmbH. A.P. became a full-time employee of Heidelberg Epignostix GmbH in December 2024. A.P. and F.S. are inventors on a patent application related to a nanopore sequencing-based method for cancer characterization, filed by Deutsches Krebsforschungszentrum (DKFZ), Universitat Heidelberg, and Oxford Nanopore Technologies PLC (patent application number: 18682016). S.H., H.L. and E.O.V.M. have received reimbursement for travel, accommodation and conference fees to speak at events organized by ONT. The other authors declare no competing interests. ### Clinical Protocols ### Funding Statement This work was supported by grants from Norwegian South-Eastern regional health authorities [grant numbers 2021039, 2023059 to S.H., H.L. and E.O.V.M]. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The study was approved by the Regional Ethics Committee for South East Norway, approvals number 388359 and 853700). I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced in the present study are available upon reasonable request to the authors

Molecular testing is essential for classifying central nervous system (CNS) tumors, with methylation profiling providing the highest diagnostic granularity. However, this requires more resources and time than conventional hematoxylin and eosin (H&E) histopathology, which is widely available globally. Here we propose Hetairos, an artificial intelligence algorithm that predicts 102 methylation-based CNS tumor subtypes from digital H&E slides. Built and validated on 9,606 patients and over 11,000 slides from 11 centers across four continents, Hetairos identified 50-70% of cases with high confidence, achieving an accuracy of 0.87 for its highest-rated predictions. Hetairos outperformed five board-certified neuropathologists in a direct histology-only comparison (0.68 versus 0.30). Prospective evaluation in routine diagnostics confirmed its performance, reducing turnaround time from 12 days (molecular testing) to 12 min. Hetairos supports diagnostic decision-making across the full spectrum of pediatric and adult CNS tumors by narrowing differential diagnoses and guiding efficient testing.

Abstract Background Classification of central nervous system (CNS) tumors has become increasingly complex over the past decade, raising concerns about the availability, feasibility and sustainability of comprehensive molecular diagnostics. We have evaluated nanopore whole genome sequencing (nWGS) as a single workflow to replace multiple diagnostic assays. Methods We performed nWGS on DNA extracted from 90 adult CNS tumor samples (58 retrospective, 32 prospective) and compared the results to findings from standard of care (SoC) diagnostic work-up. Analysis was done through an automated workflow that consolidated diagnostically and therapeutically relevant genomic alterations, including copy-number variation, structural, and single-nucleotide variants, chromosomal aberrations, gene fusions and methylation-based classification. Results Nanopore WGS enabled final diagnostic classification in all samples with >15% tumor cell content, requiring ∼3 hours of hands-on library preparation, parallel sample processing, and sequencing times within 72 hours. Methylation-based classification was available within 1 hour and was concordant with the integrated final diagnosis in 89% of cases (80/90). All diagnostically relevant copy-number variations, single-nucleotide variants, and gene fusions were concordant with standard-of-care testing, and MGMT promoter methylation status matched in 94% of cases. In addition, nWGS identified prognostic and potentially actionable variants that were not reported or covered by SoC. Conclusions Nanopore WGS delivers comprehensive genetic and epigenetic results with a fast turn-around compared to standard methods. This enables efficient, accurate, and scalable molecular diagnostics of CNS tumors using a single platform. Its broad applicability supports its implementation in routine clinical practice and may be extended to other cancer types requiring complex genomic profiling.

BACKGROUND:Meningiomas are the most common adult brain tumors. While homozygous deletions of CDKN2A/B are linked to early recurrence and hence serve as CNS WHO grade 3 criterion, the clinical impact of hemizygous deletions remains unclear-especially since distinguishing between hemi- and homozygous losses can be technically challenging. METHODS:DNA methylation data, copy-number, and mutation data were evaluated on a multicenter cohort of 970 meningiomas. Each sample's CDKN2A/B status was manually classified by visual inspection in relation to whole chromosomal losses and gains in the copy number profile generated from global methylation array data in relation to other copy number events. Progression probabilities were determined using the Kaplan-Meier method. RESULTS:Among 970 meningiomas, n = 30 had homozygous, n = 114 hemizygous (n = 31 segmental; n = 83 focal), and n = 826 CDKN2A/B balanced status. In cases with hemizygous deletions in general, an association with increased progression risk compared to balanced cases was observed, although this did not reach statistical significance (log-rank P = .074; HR = 1.36, 95% CI [0.97, 1.90]; P = .07). However, segmental hemizygous losses were linked to a significantly worse prognosis (log-rank P = .0023), but focal hemizygous deletions were not (log-rank P = .523). Segmental hemizygous CDKN2A/B deletions were more frequently associated with a higher amount of high-risk copy number variations than focal losses. CONCLUSION:Our findings suggest that hemizygous CDKN2A/B deletions overall do not confer worse risks for progression in meningiomas. The signal for segmental deletions may not be locus-specific but just one representation of the generally instable genome of aggressive meningiomas.

Array-based DNA methylation profiling is the gold standard for central nervous system (CNS) tumor molecular classification, but requires over 100 ng input DNA from surgical tissue. Cell-free tumor DNA (cfDNA) in cerebrospinal fluid (CSF) offers an alternative for diagnosis and disease monitoring. This study aimed to test the utilization of enzymatic DNA methylation sequencing (EM-seq) methods to overcome input DNA limitations. We used the NEBNext EM-seq v2 kit on various amounts of cfDNA, as low as 0.1 ng, extracted from archival CSF samples of 10 patients with CNS tumors. Tumor classification was performed via MNP-Flex using CpG sites overlapping those on the MethylationEPIC array. EM-seq provided sufficient genomic coverage for 10 and 1 ng input DNA samples to generate global DNA methylation profiles. Samples with 0.1 ng input showed lower coverage due to read duplication. Methylation levels for CpG sites with at least 5× coverage were highly correlated across various input DNA amounts, indicating that lower input cfDNA can still be used for tumor classification. The MNP-Flex classifier, trained on tissue DNA methylation data, successfully predicted CNS tumor types for 7 out of 10 CSF samples using EM-seq methylation data with only 1 ng of input cfDNA, consistent with diagnoses based on tissue MethylationEPIC classification and/or histopathology. Additionally, we detected focal and arm-level copy number alterations previously identified via clinical cytogenetics of tumor tissue. This study demonstrated the feasibility of CNS tumor molecular classification based on CSF using the EM-seq approach, and establishes potential sample quality limitations for future studies.

Purpose:Sarcomas pose a severe diagnostic challenge. A wide variety of these distinct entities need to be distinguished from each other and from less aggressive types of mesenchymal tumors, to ensure correct clinical management. A machine learning based classifier for sarcomas utilizing DNA methylation data from 1077 tumors recognizing 62 sarcoma types has already been developed and termed the sarcoma classifier, which we published in 2021. Here we present a major advancement of the scale and precision of the sarcoma classifier. Methods:DNA methylation profiles and histologic data from an unprecedented multi-institutional cohort of mesenchymal tumors were collected and analyzed. Utilizing a machine learning approach, the classifier was rigorously validated through five-fold nested cross-validation, achieving a 98% class-level accuracy and a Brier score of 0.017, indicative of well-calibrated probability estimates. Results:The sarcoma classifier v13.1 was developed based on a training set of 4377 methylation profiles from sarcomas and less aggressive mesenchymal tumors comprising 116 tumor sub-classes and 4 control groups forming 93 distinct methylation classes. Performance was validated using four independent cohorts, comprising a total of 1547 mesenchymal tumors. A methylation-based classifier prediction was obtained in 73% of cases in the validation sets, of which 91% matched the original histopathology diagnosis, thereby increasing diagnostic confidence. The classifier enabled a definitive molecular diagnosis or tumor reclassification in 6% of cases with inconclusive or ambiguous histological findings. Conclusion:Adding new sarcoma types and expanding tumor sample numbers in each methylation class in the new sarcoma classifier decisively increased the number of diagnostic predictions and improved match with histologic evaluation. This substantial advancement will promote clinical implementation of the tool for the diagnosis of mesenchymal tumor lesions.

Machine learning-based molecular classifications, particularly those using DNA methylation data, have greatly advanced diagnostics for meningioma, the most common type of primary intracranial tumor. Meningiomas have historically been classified into NF2-mutant and NF2-wild-type groups, while additional mutations and copy-number variations associated with progression risk have been incorporated into WHO grading. Several genome-wide methylation-based classification systems have been proposed. The systems, such as the random forest Brain Tumour Classifier, have been incorporated into diagnostic guidelines. However, while a number of core archetypes are shared among the different classifications, discrepancies on the definition and granularity of subtypes remain an obstacle to their clinical application. Understanding the underlying heterogeneity driving these classifications is therefore crucial. Through an integrated analysis of single-nuclei and spatially resolved transcriptomic data, as well as DNA methylation array data from multiple meningioma cohorts, we identified cell types and epigenetic signatures that are associated with increased aggressiveness in meningiomas. The results demonstrated that incremental changes in the tumor microenvironment (TME), particularly shifts in compositions and epigenetic-transcriptomic signatures in tumor-associated monocytes/macrophages and microglia-like cells, have a decisive impact on epigenetic classifications alongside tumor cells, and significantly affect clinical outcome. Therefore, we refine the previously proposed distinct molecular subtypes with a TME-determined risk continuum model for NF2-mutant meningiomas. Based on these discoveries, we additionally designed an immunohistochemistry-based diagnostic approach, which also captures intra-tumoral heterogeneities.

BACKGROUND:Advances in our technological capacity to interrogate CNS tumor biology have led to the ever increasing use of genomic sequencing in diagnostic decision making. Presently, CNS tumors are classified based on their epigenetic signatures, leading to a paradigm shift in diagnostic pathways. Such testing can be performed so rapidly using nanopore sequencing that results can be provided intraoperatively. This information greatly improves the fidelity of smear diagnosis and can help surgeons tailor their approach, balancing the risks of surgery with the likely benefit. Nevertheless, full integrated diagnosis may require subsequent additional assays to detect pathognomonic somatic mutations and structural variants, thereby delaying the time to final diagnosis. METHODS:Here, we present ROBIN, a tool based on PromethION nanopore sequencing technology that can provide both real-time, intraoperative methylome classification and next-day comprehensive molecular profiling within a single assay. ROBIN utilizes 3 methylation classifiers to improve diagnostic performance in the intraoperative setting. RESULTS:We demonstrate classifier performance on 50 prospective intraoperative cases, achieving a diagnostic turnaround time under 2 hours and generating robust tumor classifications within minutes of sequencing. Furthermore, ROBIN can detect single nucleotide variants, copy number variants, and structural variants in real time, and is able to inform a complete integrated diagnosis within 24 hours. Classifier performance demonstrated concordance with final integrated diagnosis in 90% of prospective cases. CONCLUSION:Nanopore sequencing can greatly improve turnaround times for standard-of-care diagnostic testing and is furthermore able to reliably provide clinically actionable intraoperative tumor classification.

DNA methylation-based classification is now central to contemporary neuro-oncology, as highlighted by the World Health Organization (WHO) classification of central nervous system (CNS) tumors. We present the Heidelberg CNS Tumor Methylation Classifier version 12.8 (v12.8), trained on 7,495 methylation profiles, which expands recognized entities from 91 classes in version 11 (v11) to 184 subclasses. This expansion is a result of newly identified tumor types discovered through our large online repository and global collaborations, underscoring CNS tumor heterogeneity. The random forest-based classifier achieves 95% subclass-level accuracy, with its well-calibrated probabilistic scores providing a reliable measure of confidence for each classification. Its hierarchical output structure enables interpretation across subclass, class, family, and superfamily levels, thereby supporting clinical decisions at multiple granularities. Comparative analyses demonstrate that v12.8 surpasses previous versions and conventional WHO-based approaches. These advances highlight the improved precision and practical utility of the updated classifier in personalized neuro-oncology.

Molecular data integration plays a central role in central nervous system (CNS) tumor diagnostics but currently used assays pose limitations due to technical complexity, equipment and reagent costs, as well as lengthy turnaround times. We previously reported the development of Rapid-CNS2, an adaptive-sampling-based nanopore sequencing workflow. Here we comprehensively validated and further developed Rapid-CNS2 for intraoperative use. It now offers real-time methylation classification and DNA copy number information within a 30-min intraoperative window, followed by comprehensive molecular profiling within 24 h, covering the complete spectrum of diagnostically and therapeutically relevant information for the respective entity. We validated Rapid-CNS2 in a multicenter setting on 301 archival and prospective samples including 18 samples sequenced intraoperatively. To broaden the utility of methylation-based CNS tumor classification, we developed MNP-Flex, a platform-agnostic methylation classifier encompassing 184 classes. MNP-Flex achieved 99.6% accuracy for methylation families and 99.2% accuracy for methylation classes with clinically applicable thresholds across a global validation cohort of more than 78,000 frozen and formalin-fixed paraffin-embedded samples spanning five different technologies. Integration of these tools has the potential to advance CNS tumor diagnostics by providing broad access to rapid, actionable molecular insights crucial for personalized treatment strategies. Application of a nanopore sequencing workflow for real-time analysis of brain tumors results in molecular classification within a 30-minute intraoperative window, followed by comprehensive profiling within 24 hours.

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. Ritter et al. present a spatial transcriptomics and deep learning-based approach named NePSTA (neuropathology spatial transcriptomic analysis) and leverage it to improve neuropathological diagnostics and enhance central nervous system tumor subtype classification.

DNA methylation analysis has become an essential diagnostic assay for classifying tumours of the central nervous system (CNS). However, this test requires additional resources and time compared to conventional histopathological diagnosis using haematoxylin and eosin (H&E) stained tissue sections, which are available globally. Here, we propose to reduce time and resource requirements using Hetairos, an artificial intelligence (AI) algorithm that predicts 102 methylation-based CNS tumour subtypes from digital images of H&E slides. Hetairos is built and assessed using over 11,000 slides from 10 centres across four continents. Hetairos produces well-calibrated probabilities for each subtype. Across cohorts, Hetairos identifies 50-70% of cases, which it classifies with high confidence and an accuracy of 0.87 for its highest-rated predictions. Hetairos achieved higher accuracy than four board-certified neuropathologists in a direct comparison using only histology images (0.68 vs 0.28). A prospective evaluation within a routine diagnostic setting confirmed Hetairos’ performance and highlighted its 2-day turnaround time compared to an average of 12 days for molecular testing. Hetairos can assist diagnostic decision-making by limiting the number of likely diagnoses, informing efficient further testing, and by highlighting tissue areas indicative of specific diagnoses. These capabilities suggest Hetairos is a valuable tool assisting in the diagnosis of the full spectrum of paediatric and adult CNS tumours.

DNA methylation-based classification is integral to contemporary neuro-oncological diagnostics, as highlighted by the current World Health Organization (WHO) classification of central nervous system (CNS) tumors. We introduce the Heidelberg CNS Tumor Methylation Classifier version 12.8 (v12.8), trained using 7,495 methylation profiles, thereby expanding recognized tumor types from 91 classes in the previously published v11 (Capper et al., Nature 2018) to 184 subclasses in v12.8. This expansion was primarily driven by novel tumor types discovered in our large website-derived repository and through global collaborations, further elucidating the heterogeneity of CNS tumors. Utilizing a random forest-based methodology, the classifier was rigorously validated through five-fold nested cross-validation, achieving a 95% subclass-level accuracy and a Brier score of 0.028, indicative of well-calibrated probability estimates. The hierarchical output structure facilitates comprehensive interpretation, allowing clinicians to assess subclass and aggregate class-level probabilities for informed decision-making. Comparative analyses demonstrate that v12.8 surpasses previous versions as well as traditional WHO-based diagnostics across diverse tumor cohorts. These advancements underscore the enhanced precision and practical utility of the updated Heidelberg CNS Tumor Methylation Classifier, reinforcing the pivotal role of DNA methylation profiling in personalized neuro-oncological care.

Abstract AIMS The 2021 WHO classification of CNS tumours necessitates the use of molecular testing in order to reach a confident diagnosis of many tumour types. At present, this is achieved using array-based technology delivered by a centralised model. This approach has inherent delays that mean it is often several weeks before the treating clinical team is informed of diagnosis, delaying the start of adjuvant therapy and causing significant distress for patients. As such, there is an urgent need to rethink this pathway and improve time to diagnosis. Nanopore longread sequencing technology has the potential to enable a paradigm shift in diagnostic and therapeutic pathways, offering rapid, comprehensive molecular diagnosis that can be performed in the intra-operative setting. We aimed to test the hypothesis that intra-operative methylation-based classification can be performed using nanopore sequencing and thereby expediently inform surgical and oncological decision-making. METHOD Oxford Nanopore Technology (ONT) is a native-strand, long-read sequencing platform that can perform copy- number profiling along with parallel single-gene methylation, structural variant, mutational and methylation analyses. Capital and consumable costs are relatively low, and testing can be performed in the local setting on single samples. We have optimised DNA extraction and library preparation for rapid tissue processing. We have developed a robust bioinformatic pipeline which is able to produce intraoperative classification within minutes of sequencing and full molecular profiling after 24 hours. RESULTS We demonstrate the feasibility of using ONT to deliver intra-operative methylation-based classification of CNS tumours in ‘realtime’ within an NHS hospital setting and will discuss the challenges of implementing this testing and the opportunities to inform neuro-oncological practice. CONCLUSION By giving a summary of our novel protocol of rapid DNA extraction, library preparation and bioinformatic analysis, we will show that this approach can revolutionise the diagnosis of brain tumours and has the potential for rapid adoption by non-expert laboratories.

Background Advances in our technological capacity to interrogate brain tumour biology has led to the ever-increasing use of genomic sequencing in routine diagnostic decision making. Presently, brain tumours are routinely classified based on their epigenetic signatures, leading to a paradigm shift in diagnostic pathways. Such testing can be performed so rapidly using nanopore sequencing that results can be provided intraoperatively. This information greatly improves upon the fidelity of smear diagnosis and can help surgeons tailor their approach, balancing the risks of surgery with the likely benefit. Nevertheless, full integrated diagnosis may require subsequent additional assays to detect pathognomonic somatic mutations and structural variants, thereby delaying the time to final diagnosis. Methods Here, we present ROBIN, a tool based upon PromethION nanopore sequencing technology that can provide both real-time, intraoperative methylome classification and next-day comprehensive molecular profiling within a single assay. ROBIN uniquely integrates three methylation classifiers to improve diagnostic performance in the intraoperative setting. Findings We demonstrate classifier performance on 50 prospective intraoperative cases, achieving a diagnostic turnaround time under 2 hours and generating robust tumour classifications within minutes of sequencing. Furthermore, ROBIN can detect single nucleotide variants (SNVs), copy number variants (CNVs) and structural variants (SVs) in real-time, and is able to inform a complete integrated diagnosis within 24 hours. Classifier performance demonstrated concordance with final integrated diagnosis in 90% of prospective cases. Interpretation Nanopore sequencing can greatly improve upon the turnaround times for standard of care diagnostic testing, including sequencing, and is furthermore able to reliably provide clinically actionable intraoperative tumour classification. Funding The Jean-Shanks Foundation, the Pathological Society of Great Britain and Ireland, the British Neuropathological Society, and the Wellcome Trust. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement The Jean-Shanks Foundation, the Pathological Society of Great Britain and Ireland, the British Neuropathological Society, and the Wellcome Trust. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The National Research Ethics Committee of the East Midlands gave ethical approval for this work. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes De-identified individual participant data that underlie the results reported in this article will be made available upon reasonable request. The ROBIN code used for the analysis is available at https://github.com/looselab/robin.