Lucas, Calixto‐Hope G.

Abstract INTRODUCTION Glioblastoma is the most common malignant adult brain tumor with poor prognosis, largely due to its heterogeneous landscape. The AI-RANO and RANO-RGP groups organized the BraTS-Pathology 2024 challenge, providing a publicly-available dataset and a benchmarking environment for AI models to automatically identify six clinically-relevant histopathologic sub-regions in H&E-stained whole slide images (WSIs). METHODS We reclassified the TCGA-GBM and TCGA-LGG data according to the WHO 2021 criteria and identified diagnostic WSIs of 188 glioblastoma (IDH-wt, CNS WHO Gr.4) patients. A well-defined annotation protocol was communicated to eight neuropathologists, across eight institutions, who created “ground truth” annotations for Cellular Tumor (CT), Geographic Necrosis, Cortical Infiltration, Pseudopalisading Necrosis (PN), Microvascular Proliferation, and Penetration into White Matter (PWM). Comprehensive WSI curation for elimination of artifactual content and follow up partitioning resulted in approximately 300,000 data samples/patches. FINDINGS Inter-observer variability assessment highlighted difficulties in establishing a “gold standard”. The least annotated sub-regions (PWM and PN) were the most challenging to the participating methods, whereas all methods achieved their highest accuracy in the most annotated sub-region (CT). The highest average performance across sub-regions was obtained by a foundation model-based method. However, this top-ranked solution achieved an F1 score of 0.51 (Accuracy=0.65, Sensitivity=0.56, Specificity=0.93). CONCLUSIONS Our findings suggest that achieving state-of-the-art performance, in automatically identifying clinically-relevant glioblastoma sub-regions, remains an unsolved research problem. Analysis of the ground-truth annotations underscores the critical need for more precisely defined and neuropathologically standardized annotation protocol. Current AI algorithms are facing difficulties in addressing glioblastoma’s heterogeneous histologic landscape in the absence of large, comprehensively, and consistently annotated data. Building upon these findings, and towards developing more accurate AI algorithms, the 2025 iteration of the BraTS-Path challenge is currently on-going by leveraging two million data samples, aiming to enhance our disease understanding and improve diagnostic accuracy, ultimately contributing to improved patient outcomes.

Abstract Computer vision now enables deep learning models to link histopathologic features to molecular features in a quantitative manner not previously achievable by human pathologists. However, human interpretability is often limited. We trained a deep neural network to predict gene expression from digital whole slide images (WSIs) using 848 central nervous system (CNS) tumors with paired WSI and RNA-sequencing data. We used inferred RNA expression levels as a surrogate for protein expression of the CNS tumor lineage markers GFAP, OLIG2, and SSTR2, as well as the proliferation marker MKI67 (Ki67). We established thresholds for categorizing the inferred expression levels as positive or negative based on levels observed in cross-validation testing in select tumor types to approximate routine clinical interpretation of immunohistochemical (IHC) staining. We tested the sensitivity and specificity of our ‘virtual’ IHC on an independent, multi-institutional cohort of over 2,000 CNS tumors with objective diagnostic labels derived from DNA methylation-based tumor classification. As a dichotomous variable, inferred GFAP expression showed a sensitivity of 74% and a specificity of 99% as a glial marker in a cohort of gliomas and meningiomas. In the same cohort, OLIG2 showed a sensitivity of 75% and a specificity of 97% as a glial marker, while SSTR2 showed a sensitivity of 83% and a specificity of 97% as a marker of meningioma. In a cohort of ependymomas and gliomas, virtual IHC for OLIG2 showed a sensitivity of 75% and a specificity of 82% in distinguishing ependymomas from gliomas. Finally, in a cohort of gliomas, inferred expression of MKI67 demonstrated a trend consistent with what would be expected by actual IHC, with increasing MKI67 expression with increasing tumor grade. Our model provides the basis for a human interpretable and clinically applicable deep neural network to aid human pathologists in the diagnosis and grading of CNS tumors.

Abstract Molecular testing has become critical components in the diagnostic classification of the central nervous system (CNS) tumors. However, these methods require substantial resources, limiting accessibility for many patients. Recent advances in artificial intelligence (AI) and computer vision empower deep learning models to infer molecular features from histopathology images, which may often be more readily available than molecular testing, to classify CNS tumors. We trained pan-CNS tumor models to predict DNA methylation and gene expression from whole-slide images (WSIs). These molecular predictions were then used in a hierarchical machine learning framework, termed Neuropath-AI, to predict nine broad tumor families and 52 tumor types using a large diverse cohort of 5,715 histopathology samples, of which 2,988 had paired DNA methylation profiling and 848 had paired RNA-sequencing. We evaluated Neuropath-AI on another large independent multi-institutional cohort of 5,516 CNS tumors, for which Neuropath-AI predicted tumor class with an associated confidence score. Neuropath-AI designated 46% of test samples as predictable with high-confidence, which it then accuracy classified in 97% of samples. It made moderate-confidence (and above) predictions for 87% of test samples, for which it achieved a diagnostic accuracy of 80% for the top-1 prediction and 86% accuracy for a top-2 predictions. A comparison with human neuropathologists showed that Neuropath-AI is comparable at classifying tumors. A second test showed that human neuropathologists were more accurate when the Neuropath-AI classifier results were provided to them, testifying to the potential translational benefit of integrating AI classification tools into the pathologist’s workflow. Our model provides the basis for a clinically applicable deep learning assistant to improve human efficiency and accuracy of CNS tumor diagnosis. The model will be made publicly available and can be readily implemented to assist human pathologists in clinical workflows, in further expanded prospective studies.

Abstract Glioblastoma, the most common malignant primary adult brain tumor, poses significant diagnostic and treatment challenges due to its heterogeneous molecular and micro-environmental profiles. To this end, we organize the BraTS-Path challenge to provide a public benchmarking environment and a comprehensive dataset to develop and validate AI models for identifying distinct histopathologic glioblastoma sub-regions in H&E-stained digitized tissue sections. We identified 188 multi-institutional diagnostic slides of glioblastoma (IDH-wt, Gr.4) cases, from the TCGA-GBM and TCGA-LGG data collections, following their reclassification according to the 2021 WHO classification criteria. Sub-regions were selected according to distinctive morphology of histopathologic features and included aggressive tumor biology and areas consistent with potential treatment effect. Selected sub-region annotations included cellular tumor, geographic necrosis, cortical infiltration, pseudopalisading necrosis, microvascular proliferation, white matter penetration, regions dense with macrophages, leptomeningeal infiltration, and presence of lymphocytes. We obtained 107,340 patches of size 512x512 from the 9 sub-regions. A global network of board-certified expert neuropathologists defined and followed a systematic annotation protocol based on clinical definitions and only delineated sub-regions with high confidence, thus ensuring high-quality standardized data. Each tissue section was assigned to an annotator-approver pair, with the annotator delineating sub-regions and the approver ensuring the consistency of the annotations. By crowdsourcing annotations, the BraTS-Path challenge harnesses the collective expertise of clinical neuropathologists and fosters a collaborative environment to advance the neuro-oncology field. The anticipated developed algorithms are expected to integrate state-of-the-art computational methods, achieving high accuracy in identifying diverse histopathologic features and advancing clinical decision-making processes. The BraTS-Path challenge aims to bridge the gap between research and clinical practice by promoting the development of AI-driven tools for precise tumor characterization. This collaborative effort can significantly enhance our understanding of glioblastoma, improve diagnostic accuracy, and inform treatment strategies, thereby contributing to better patient outcomes.

Abstract Targeted gene expression profiling improves risk stratification for meningiomas, but the utility of this approach for low grade meningiomas is incompletely understood. To address this, 466 meningiomas that were initially used to validate a 34-gene expression biomarker were combined with 270 previously unpublished samples for an analysis of biomarker performance in newly diagnosed (n=691) or recurrent (n=45) WHO grade 1 meningiomas treated with gross total (GTR) or subtotal resection (STR), and newly diagnosed WHO grade 2 meningiomas treated with GTR. Prediction of local freedom from recurrence (LFFR) or overall survival (OS) was measured using Cox proportional hazards models and the Kaplan-Meier method. The cohort included 736 patients, 69.4% female, with a median age at diagnosis of 58 years. There were 635 WHO grade 1 meningiomas, 498 with GTR and 133 with STR, and 101 WHO grade 2 meningiomas with GTR. The median follow-up was 4.6 years, the local recurrence rate was 12.0% and the rate of death was 8.4%. The biomarker classified 51.5% of meningiomas as intermediate risk and 9.6% as high risk. Combining the biomarker with extent of resection identified 19.7% as unfavorable (intermediate molecular risk with STR or high molecular risk), including 7.4% of newly diagnosed WHO grade 1 meningiomas with GTR. The risk score was prognostic for LFFR overall (hazard ratio per 0.1 increase in risk score 1.89, 95% confidence interval 1.58-2.25) and across WHO grades, extents of resection, and newly diagnosed or recurrent settings. Gene expression risk groups were prognostic for LFFR, with improved outcomes among low molecular risk compared to intermediate molecular risk (Log-rank, p=0.0002) and high molecular risk meningiomas (Log-rank, p<0.0001). These data suggest that a targeted gene expression biomarker can identify low grade meningiomas that are more likely to recur after surgery, possibly changing management for approximately one-fifth of cases.

Abstract Current standard of care for IDH-wildtype glioblastoma (GBM) includes maximal safe resection, adjuvant radiation, and chemotherapy with the alkylating agent temozolomide. A subset of gliomas treated with temozolomide develop somatic hypermutation with a predominance of C>T transitions due to the alkylation effects on DNA. However, the frequency of occurrence, clinical ramifications, and predictive biomarkers for developing temozolomide-induced hypermutation (TMZ-HM) in GBM are unknown. Here we performed comprehensive histopathologic, genomic, and epigenomic evaluation of paired initial and recurrent GBM from 106 patients. Seven patients (7%) developed hypermutation at time of first recurrence, while an additional five patients (5%) developed hypermutation at a subsequent recurrence. Patients who developed TMZ-HM had a significantly longer interval between initial surgery and first recurrence (27.8 vs. 9.9 months, p<0.001). In contrast to IDH-mutant gliomas where TMZ-HM is associated with poor prognosis, those patients with IDH-wildtype GBM who developed TMZ-HM had significantly longer overall survival from both initial surgery (67.9 vs. 20.3 months, p<0.001) and from time of recurrence after development of hypermutation (30.9 vs. 9.5 months, p=0.001). There was no difference in oncogenic alteration frequency in initial GBM that subsequently developed TMZ-HM. However, 4 specific CpG sites out of 12 total interrogated sites in the promoter region of MGMT (a DNA repair enzyme) and 5 specific CpG sites in the promoter region of KCNQ1DN (a negative regulator of c-Myc) were significantly more hypermethylated in those GBM that developed TMZ-HM. We then generated a weighted risk score across these CpG sites that can be used to prospectively identify GBM likely to develop TMZ-HM associated with favorable survival. In summary, patients with GBM that develop TMZ-HM follow a more favorable clinical course, and specific MGMT and KCNQ1DN promoter hypermethylation patterns could serve as a biomarker to identify these patients who may potentially benefit from extended alkylating chemotherapy treatment.

Abstract Background Surgery is the mainstay of treatment for meningioma, the most common primary intracranial tumor, but improvements in meningioma risk stratification are needed and current indications for postoperative radiotherapy are controversial. Recent studies have proposed prognostic meningioma classification systems using DNA methylation profiling, copy number variants, DNA sequencing, RNA sequencing, histology, or integrated models based on multiple combined features. Targeted gene expression profiling has generated robust biomarkers integrating multiple molecular features for other cancers, but is understudied for meningiomas. Methods Targeted gene expression profiling was performed on 173 meningiomas and an optimized gene expression biomarker (34 genes) and risk score (0 to 1) was developed to predict clinical outcomes. Clinical and analytical validation was performed on independent meningiomas from 12 institutions across 3 continents (N = 1856), including 103 meningiomas from a prospective clinical trial. Gene expression biomarker performance was compared to 9 other classification systems. Results The gene expression biomarker improved discrimination of postoperative meningioma outcomes compared to all other classification systems tested in the independent clinical validation cohort for local recurrence (5-year area under the curve [AUC] 0.81) and overall survival (5-year AUC 0.80). The increase in area under the curve compared to the current standard of care, World Health Organization 2021 grade, was 0.11 for local recurrence (95% confidence interval [CI] 0.07–0.17, P < 0.001). The gene expression biomarker identified meningiomas benefiting from postoperative radiotherapy (hazard ratio 0.54, 95% CI 0.37–0.78, P = 0.0001) and re-classified up to 52.0% meningiomas compared to conventional clinical criteria, suggesting postoperative management could be refined for 29.8% of patients. Conclusions A targeted gene expression biomarker improves discrimination of meningioma outcomes compared to recent classification systems and predicts postoperative radiotherapy responses.

Abstract Anaplastic meningiomas can have a sarcomatous appearance on histology but true sarcomatous (metaplastic) differentiation is rare. These tumors follow an aggressive clinical course with recurrence and poor clinical outcomes. Due to significant overlap in morphology and immunohistochemical profiles, distinguishing between sarcomatous transformation of a meningioma and a true sarcoma can be challenging. Here, we outline potential diagnostic pitfalls and the utility of ancillary molecular testing in 3 patients diagnosed with sarcomatous meningiomas. We report loss of typical meningothelial markers in sarcomatous meningiomas. Ancillary molecular testing can support the diagnosis of sarcomatous meningioma when a molecular signature consistent with meningioma is seen, such as inactivation of the NF2 gene. Recognition of this rare transformation in meningioma can prevent a misdiagnosis of a primary sarcoma, whether sporadic or radiation-induced from prior treatment of a more classic meningioma.