Low-grade Gliomas Research Articles

Somatosensory Mapping Using a Novel Sensory Discrimination Task: Technical Note.

Although diffuse gliomas in the primary somatosensory cortex (S1) are often considered resectable, gliomas in the primary motor cortex require motor mapping to preserve motor function. Recent evidence indicates that some somatosensory cortex neurons may trigger motor responses, necessitating refined somatosensory mapping techniques. Using piezoelectric tactile stimulators on patients' faces and hands, we delivered 25 Hz vibrations and prompted patients to discriminate between dermatomes. Testing included areas contralateral to tumor-infiltrated and to non-tumor-infiltrated cortical regions. Sensory thresholds were determined by reducing stimulus intensity based on performance. Intraoperatively, electrocorticography electrode arrays were used to map sensory responses, and postoperative assessments evaluated sensory outcomes. The high-grade glioma case involved a 61-year-old man with right-sided weakness and numbness with a left parietal mass on MRI. Preoperative testing showed that the average vibratory detection threshold of the hand contralateral to the suspected tumor site was significantly higher than that of the hand contralateral to healthy cortex (P < .001). Intraoperative mapping confirmed the absence of functional involvement in cortical structures overlying the tumor. Postoperative imaging confirmed gross total resection, and sensory vibratory thresholds were normalized (P = .51). The low-grade glioma case included a 54-year-old man with a left parietal nonenhancing mass on MRI. No baseline sensory impairments were found on preoperative testing. Intraoperative mapping identified motor and sensory cortices, guiding tumor resection while preserving motor function. Postoperative MRI confirmed near-total resection, but new sensory impairments were noted in the hand and face contralateral to the resection site (P < .001). These deficits resolved by postoperative day 11, with no evidence of tumor progression on follow-up imaging. The sensory discrimination task provides a quantifiable method for assessing sensory changes and functional outcomes related to glioma. This technique enhances our understanding of how glioma infiltration remodels sensory systems and affects clinical outcomes in patients.

Diffusion-weighted MR image analysis based on gamma distribution model for differentiating benign and malignant brain tumors.

Considering the invasiveness of the biopsy method, we attempted to evaluate the ability of the gamma distribution model using magnetic resonance imaging images to stage and grade benign and malignant brain tumors. A total of 42 patients with malignant brain tumors (including glioma, lymphoma, and choroid plexus papilloma) and 24 patients with benign brain tumors (meningioma) underwent diffusion-weighted imaging using five b-values ranging from 0 to 2000 s/mm2 with a 1.5 T scanner. The gamma distribution model is expected to demonstrate the probability of water molecule distribution based on the apparent diffusion coefficient. For all tumors, the apparent diffusion coefficient, shape parameter (κ), and scale parameter (θ) were calculated for each b-value. In the staging step, the fractions (ƒ1, ƒ2, ƒ3) expected to reflect the intracellular, and extracellular diffusion and perfusion were investigated. Diffusion <1 × 10-4 mm2/s (ƒ1), 1 × 10-4 mm2/s < Diffusion > 3 × 10-4 mm2/s (ƒ2), and Diffusion >3 × 10-4 mm2/s (ƒ3); in the grading step, fractions were determined to check heavily restricted diffusion. Diffusion lower than 0.3 × 10-4 mm2/s (ƒ11). Diffusion lower than 0.5 × 10-4 mm2/s (ƒ12). Diffusion lower than 0.8 × 10-4 mm2/s (ƒ13). The findings were analyzed using nonparametric statistics and receiver operating characteristic curve diagnostic performance. Gamma model parameters (κ, ƒ1, ƒ2, ƒ3) showed a satisfactory difference in differentiating meningioma from glioma. For b value = 2000 s/mm2, ƒ1 had a better diagnostic performance than κ and apparent diffusion coefficient (sensitivity, 88%; specificity, 68%; P < .001). The best diagnostic performance was related to ƒ3 in b = 2000 s/mm2 (area under the curve = 0.891, sensitivity = 83%, specificity = 80%, P < .001). In the grading step, ƒ12 (area under the curve = 0.870, sensitivity = 92%, specificity = 72%, P < .001) had the best diagnostic performance in differentiating high-grade from low-grade gliomas with b = 2000 s/mm2. The findings of our study highlight the potential of using a gamma distribution model with diffusion-weighted imaging based on multiple b-values for grading and staging brain tumors. Its potential integration into routine clinical practice could advance neurooncology and improve patient outcomes through more accurate diagnosis and treatment planning.

Emerging and Biological Concepts in Pediatric High-Grade Gliomas.

Primary central nervous system tumors are the most frequent solid tumors in children, accounting for over 40% of all childhood brain tumor deaths, specifically high-grade gliomas. Compared with pediatric low-grade gliomas (pLGGs), pediatric high-grade gliomas (pHGGs) have an abysmal survival rate. The WHO CNS classification identifies four subtypes of pHGGs, including Grade 4 Diffuse midline glioma H3K27-altered, Grade 4 Diffuse hemispheric gliomas H3-G34-mutant, Grade 4 pediatric-type high-grade glioma H3-wildtype and IDH-wildtype, and infant-type hemispheric gliomas. In recent years, we have seen promising advancements in treatment strategies for pediatric high-grade gliomas, including immunotherapy, CAR-T cell therapy, and vaccine approaches, which are currently undergoing clinical trials. These therapies are underscored by the integration of molecular features that further stratify HGG subtypes. Herein, we will discuss the molecular features of pediatric high-grade gliomas and the evolving landscape for treating these challenging tumors.

Anatomical and subcortical invasiveness in diffuse low-grade astrocytomas differ between IDH status and provide prognostic information.

Diffuse astrocytomas preferentially infiltrate eloquent areas affecting the outcome. A preoperative understanding of isocitrate dehydrogenase (IDH) status may offer opportunities for specific targeted therapies impacting treatment management. The aim of this study was to analyze clinical, topographical, radiological in WHO 2 astrocytomas with different IDH status and the long-term patient's outcome. A series of confirmed WHO 2 astrocytoma patients (between 2005 and 2015) were retrospectively analyzed. MRI sequences (FLAIR) were used for tumor volume segmentation and to create a frequency map of their locations into the Montreal Neurological Institute (MNI) space. The Brain-Grid (BG) system (standardized radiological tool of intersected lines according to anatomical landmarks) was used as an overlay for infiltration analysis of each tumor. Long-term follow-up was used to perform a survival analysis. Forty patients with confirmed IDH status (26 IDH-mutant, IDHm/14 IDH-wild type, IDHwt) according to WHO 2021 classification were included with a mean follow-up of 7.8 years. IDHm astrocytomas displayed a lower number of BG-voxels (P < 0.05) and were preferentially located in the anterior insular region. IDHwt group displayed a posterior insular and peritrigonal location. IDHwt group displayed a shorter OS compared with IDHm (P < 0.05), with the infiltration of 7 or more BG-voxels as an independent factor predicting a shorter OS. IDHm and IDHwt astrocytomas differed in preferential location, number of BG-voxels and OS at long follow-up time. The number of BG-voxels affected the OS in IDHwt was possibly reflecting higher tumor invasiveness. We encourage the systematic use of alternative observational tools, such as gradient maps and the Brain-Grid analysis, to better detect differences of tumor invasiveness in diffuse low-grade gliomas subtypes.

Multifaceted bioinformatic analysis of m6A-related ferroptosis and its link with gene signatures and tumour-infiltrating immune cells in gliomas.

Whether N6-Methyladenosine (m6A)- and ferroptosis-related genes act on immune responses to regulate glioma progression remains unanswered. Data of glioma and corresponding normal brain tissues were fetched from the TCGA database and GTEx. Differentially expressed genes (DEGs) were identified for GO and KEGG enrichment analyses. The FerrDb database was based to yield ferroptosis-related DEGs. Hub genes were then screened out using the cytoHubba database and validated in clinical samples. Immune cells infiltrating into the glioma tissues were analysed using the CIBERSORT R script. The association of gene signature underlying the m6A-related ferroptosis with tumour-infiltrating immune cells and immune checkpoints in low-grade gliomas was analysed. Of 6298 DEGs enriched in mRNA modifications, 144 were ferroptosis-related; NFE2L2 and METTL16 showed the strongest positive correlation. METTL16 knockdown inhibited the migrative and invasive abilities of glioma cells and induced ferroptosis invitro. NFE2L2 was enriched in the anti-m6A antibody. Moreover, METTL16 knockdown reduced the mRNA stability and level of NFE2L2 (both p < 0.05). Proportions of CD8+ T lymphocytes, activated mast cells and M2 macrophages differed between low-grade gliomas and normal tissues. METTL16 expression was negatively correlated with CD8+ T lymphocytes, while that of NFE2L2 was positively correlated with M2 macrophages and immune checkpoints in low-grade gliomas. Gene signatures involved in the m6A-related ferroptosis in gliomas were identified via bioinformatic analyses. NFE2L2 interacted with METTL16 to regulate the immune response in low-grade gliomas, and both molecules may be novel therapeutic targets for gliomas.

TGF-β signaling-related signature for predicting prognosis and therapeutic response in lower-grade glioma.

Low-grade glioma (LGG) is a tumor that includes World Health Organization (WHO) grade II and III glioma, the treatment of which consistently results in relapse and drug resistance. Transforming growth factor-beta (TGF-β) is a multifunctional cytokine that regulates various cellular processes, which is found to be abnormal in tumors and promotes glioma development and progression. In this study, we aimed to systematically evaluate the importance of the genes associated with TGF-β in LGG and discover the role of these genes in the prognosis and treatment response of LGG. We used the "Bioconductor Limma" and "consensusClusterplus" R packages to screen differential and prognostic TGF-β-related genes. The R package "GSVA" was used to estimate the infiltration of immune cells and metabolism signature. The drug sensitivity for each TGF-β subtype was assessed by the R package "pRRophetic". The Genomic Identification of Significant Targets in Cancer (GISTIC) algorithm was used to assess the copy number variation (CNV). The onco-print tool of the "complexheatmap package" was employed to visualize the somatic mutation and copy number alteration (CNA) among TGF clusters. We reported three subtypes (A, B, and C) of LGG according to the classification of TGF-β-related genes, where subtype A showed the best prognosis. Subtype B was highly enriched in immune cells. Somatic variations were observed to be diverse in all of the three TGF-β subtypes. Furthermore, another three genes (SHA, AC062021.1, and SNCG) related to TGF-β were identified, which can be a superior predictor of prognosis with a risk score. LGG can be divided into three subtypes based on TGF-β signaling-related genes with distinct immune infiltration, metabolism, somatic variations, and prognosis.

Insular Gliomas. Experience in a Latin American Center and Assessment of Variables Related to Surgical Management and Prognosis

To describe our experience in the resection of gliomas involving the insula and analyze the variables implicated in the management and prognosis of these tumors. This retrospective, single-center, analytic study included a cohort of 83 patients who underwent surgery for insular gliomas by the same surgeon in a third-level Argentine center, in the period between 2010 and 2023. We analyzed the population's demographic, clinical, and radiologic features and surgical variables associated with postoperative results and prognosis using multivariate regression analysis. A total of 53 patients (54% men) were included, with a mean follow-up of 40.7 months. The mean age at surgery was 41 years (range, 21-73) and 66.1% corresponded to low-grade gliomas (LGGs). Seizures were the initial symptom in most cases. There was evidence of tumor extension over the insula to the temporal or/and frontal lobe in 64.2% of patients. An extent of resection >90% was achieved in 62.3% of cases (27% of gross total resection), with an average resected volume of 89.4%. Awake craniotomy was indicated in 47% of patients and intraoperative magnetic resonance imaging was performed in 24%. Recurrence was observed in 44% of patients, with a mean progression-free survival of 31 months (42 months in LGG and 10 months in high-grade glioma [HGG]). Nine patients underwent reoperation. By the time of 2 years, survival was 100% for LGG and 46% for HGG, whereas 4-year overall survival was 92% for patients with LGG and 15.4% for those with HGG. Surgery for insular gliomas is a complex task that needs to be managed with adequate preoperative and intraoperative assessment to achieve maximum safe resection with low morbidity for better functional and oncologic outcomes. Adequate anatomic understanding, radiologic analysis, awake craniotomy, and cortical and subcortical mapping are paramount to pursue this aim.

477P ENHO's protective role in lower grade glioma

477P ENHO's protective role in lower grade glioma

MR-guided focused ultrasound in pediatric neurosurgery: current insights, technical challenges, and lessons learned from 45 treatments at Children's National Hospital.

MR-guided focused ultrasound (MRgFUS) is an evolving technology with numerous present and potential applications in pediatric neurosurgery. The aim of this study was to describe the use of MRgFUS, technical challenges, complications, and lessons learned at a single children's hospital. A retrospective analysis was performed of a prospectively collected database of all pediatric patients undergoing investigational use of MRgFUS for treatment of various neurosurgical pathologies at Children's National Hospital. Treatment details, clinical workflow, and standard operating procedures are described. Patient demographics, procedure duration, and complications were obtained through a chart review of anesthesia and operative reports. In total, 45 MRgFUS procedures were performed on 14 patients for treatment of diffuse intrinsic pontine glioma (n = 12), low-grade glioma (n = 1), or secondary dystonia (n = 1) between January 2022 and April 2024. The mean age at treatment was 9 (range 5-22) years, and 64% of the patients were male. With increased experience, the total anesthesia time, sonication time, and change in core body temperature during treatment all significantly decreased. Complications affected 4.4% of patients, including 1 case of scalp edema and 1 patient with a postprocedure epidural hematoma. Device malfunction requiring abortion of the procedure occurred in 1 case (2.2%). Technical challenges related to transducer malfunction and sonication errors occurred in 6.7% and 11.1% of cases, respectively, all overcome by subsequent user modifications. The authors describe the largest series on MRgFUS technical aspects in pediatric neurosurgery at a single institution, comprising 45 total treatments. This study emphasizes potential technical challenges and provides valuable insights into the nuances of its application in pediatric patients.

Metabotropic Glutamate Receptors Type 3 and 5 Feature the "NeuroTransmitter-type" of Glioblastoma: A Bioinformatic Approach.

Glioblastoma (GBM) represents an aggressive and common tumor of the central nervous system. The prognosis of GBM is poor, and despite a refined genetic and molecular characterization, pharmacological treatment is largely suboptimal. Contribute to defining a therapeutic line in GBM targeting the mGlu3 receptor in line with the principles of precision medicine. Here, we performed a computational analysis focused on the expression of type 3 and 5 metabotropic glutamate receptor subtypes (mGlu3 and mGlu5, respectively) in high- and low-grade gliomas. The analysis allowed the identification of a particular high-grade glioma type, characterized by a high expression level of both receptor subtypes and by other markers of excitatory and inhibitory neurotransmission. This so-called neurotransmitter-GBM (NT-GBM) also shows a distinct immunological, metabolic, and vascularization gene signature. Our findings might lay the groundwork for a targeted therapy to be specifically applied to this putative novel type of GBM.

Memory Tract Sparing Using Diffusion Tensor Imaging in Radiation Planning of Primary Brain Tumors

Radiation therapy (RT) is a critical treatment modality for both primary and metastatic brain tumors, yet approximately 30% of patients experience cognitive decline post-radiotherapy. This cognitive toxicity is linked to low radiation doses affecting the hippocampal dentate gyrus. Hippocampal Avoidance (HA)-Whole Brain Radiotherapy (WBRT) combined with Memantine has shown promising outcomes in preserving cognitive function and quality of life (QOL) in patients with brain metastases. Nowadays, it is the standard of care for those with good performance status and no hippocampal metastases. We conducted a prospective trial approved by the IRB (SMC0307-23), including patients aged 18 and above with primary brain tumors post-resection or biopsy. Exclusion criteria included multifocal glioma crossing to the other hemisphere. RT was delivered to a total dose of 59.4Gy in 33. Diffusion tensor imaging (DTI) was performed to map hippocampal-associated white matter tracts. Using Eclipse treatment planning software, memory fiber tracts and hippocampi were contoured and integrated into RT planning. Dosimetric analyses compared two plans with memory fiber constraints and one without. The primary endpoints were safety and dosimetric feasibility. Twelve patients with low-grade gliomas were included, and the contouring of memory fibers and hippocampi was successful. VMAT treatment plans met-dose constraints for memory fibers, with an average mean dose of 10.1 Gy. The average MoCA score before RT was 27.1 and 26.4 at eight months post-treatment, with a p-value of 0.07. Excluding one patient, the scores were 27.1 and 26.6, respectively (p=0.13). MRI planning using DTI for memory fiber detection and incorporation into RT planning via VMAT techniques enables hippocampal and associated white fiber sparing, potentially preserving cognitive function. Preliminary cognitive data are promising, supporting the need for further validation in a larger cohort.

A review of sonodynamic therapy for brain tumors.

Sonodynamic therapy (SDT) is gaining attention as a promising new noninvasive brain tumor treatment that targets and selectively kills tumor cells, with limited side effects. This review examines the mechanisms of SDT and ongoing clinical trials looking at optimization of sonication parameters for potential treatment of glioblastoma (GBM) and diffuse intrinsic pontine glioma (DIPG). The results in the first patient with recurrent GBM treated at the Mayo Clinic are briefly discussed. The authors of this literature review used electronic databases including PubMed, EMBASE, and OVID. Articles reporting relevant preclinical and clinical trials were identified by searching for text words/phrases and MeSH terms, including the following: "sonodynamic therapy," "SDT," "focused ultrasound," "5-ALA," "ALA," "brain tumors," "diffuse pontine glioma," "glioblastoma," and "high grade glioma." Preclinical and clinical trials investigating the specific use of SDT in brain tumors were reviewed. In preclinical models of high-grade glioma and GBM, SDT has shown evidence of targeted tumor cell death via the production of reactive oxygen species. Emerging clinical trial results within recurrent GBM and DIPG show evidence of successful treatment response, with minimal side effects experienced by recruited patients. So far, SDT has been shown to be a promising noninvasive cancer treatment that is well tolerated by patients. The authors present pilot data suggesting good radiological response of GBM to a single SDT treatment, with unpublished observation of a lack of off-target effects even after multiple (monthly) sonication outpatient treatments. The scope of the clinical trials of SDT is to investigate whether it can be the means by which the fatal diagnosis of GBM or DIPG is converted into that of a chronic, treatable disease. SDT is safe, repeatable, and better tolerated than both chemotherapy and radiotherapy. It has been shown to have an effect in human cancer therapy, but more clinical trials are needed to establish standardized protocols for sonosensitizer delivery, treatment parameters, and combination therapies. The most appropriate timing of treatment also remains to be determined-whether to prevent recurrence in the postoperative period, or as a salvage option in patients with recurrent GBM for which redo surgery is inappropriate. It is hoped that SDT will also be developed for a wider spectrum of clinical indications, such as metastases, meningioma, and low-grade glioma. Further clinical trials are in preparation.

468P Real-world safety of trametinib monotherapy or in combination with dabrafenib in children, adolescents, and young adults with relapsed or refractory low-grade glioma (LGG) and high-grade glioma (HGG)

468P Real-world safety of trametinib monotherapy or in combination with dabrafenib in children, adolescents, and young adults with relapsed or refractory low-grade glioma (LGG) and high-grade glioma (HGG)

Multi-pool chemical exchange saturation transfer MRI in glioma grading, molecular subtyping and evaluating tumor proliferation.

To evaluate the performance of multi-pool Chemical exchange saturation transfer (CEST) MRI in prediction of glioma grade, isocitrate dehydrogenase (IDH) mutation, alpha-thalassemia/mental retardation syndrome X-linked (ATRX) loss and Ki-67 labeling index (LI), based on the fifth edition of the World Health Organization classification of central nervous system tumors (WHO CNS5). 95 patients with adult-type diffuse gliomas were analyzed. The amide, direct water saturation (DS), nuclear Overhauser enhancement (NOE), semi-solid magnetization transfer (MT) and amine signals were derived using Lorentzian fitting, and asymmetry-based amide proton transfer-weighted (APTwasym) signal was calculated. The mean value of tumor region was measured and intergroup differences were estimated using student-t test. The receiver operating curve (ROC) and area under the curve (AUC) analysis were used to evaluate the diagnostic performance of signals and their combinations. Spearman correlation analysis was performed to evaluate tumor proliferation. The amide and DS signals were significantly higher in high-grade gliomas compared to low-grade gliomas, as well as in IDH-wildtype gliomas compared to IDH-mutant gliomas (all p < 0.001). The DS, MT and amine signals showed significantly differences between ATRX loss and retention in grade 2/3 IDH-mutant gliomas (all p < 0.05). The combination of signals showed the highest AUC in prediction of grade (0.857), IDH mutation (0.814) and ATRX loss (0.769). Additionally, the amide and DS signals were positively correlated with Ki-67 LI (both p < 0.001). Multi-pool CEST MRI demonstrated good potential to predict glioma grade, IDH mutation, ATRX loss and Ki-67 LI.

Enhanced TumorNet: Leveraging YOLOv8s and U-net for superior brain tumor detection and segmentation utilizing MRI scans

Brain tumors, characterized by abnormal cell growth, pose a significant challenge in clinical imaging due to their complex and diverse structures. Early and accurate identification, classification, localization, and segmentation of these tumors are critical to reducing mortality. However, the extensive data generated by MRI scans makes manual segmentation time-consuming and impractical for clinical use. To address these challenges, we propose, a hybrid deep learning model that precisely segmented tumor regions with U-Net to enable YOLOv8s to efficiently detect, classify and localize tumors. The model was trained and validated using The Cancer Imaging Archive (TCIA) dataset, which includes MRI images of brain tumors, and the Cancer Genome Atlas (CGA) low-grade glioma dataset, which includes data from 110 patients with FLAIR aberrant segmentation masks. The proposed hybrid model was evaluated using several performance metrics, including F1 score, specificity, recall, precision, accuracy, and ROC-AUC score. Hybrid proposed performed highly, achieving a precision of 97.8 %, accuracy of 98.6 %, recall of 95.2 %, F-1 score of 96.3 %, specificity of 89.1 %, and ROC-AUC score of 98.5 %. The integration of YOLOv8s and U-Net in Enhanced TumorNet offers a powerful solution for the automated analysis of brain tumors in MRI scans, significantly improving detection and segmentation accuracy. This hybrid approach holds great potential for clinical applications, enhancing the efficiency and effectiveness of brain tumor diagnosis.

Vorasidenib-Induced Trichomegaly and Hypertrichosis: a New Side Effect in a Patient with Diffuse Astrocytoma.

Vorasidenib, an oral dual inhibitor targeting mutant enzymes isocitrate dehydrogenase 1 and 2, is utilized in the management of diffuse low-grade gliomas. Despite limited documentation of its adverse events, we present the case of a 44-year-old male who exhibited trichomegaly and hypertrichosis of body hair, eyebrows, and eyelashes following one month of vorasidenib treatment. Notably, the patient experienced diffuse hair regrowth on the scalp, including in areas affected by severe androgenetic alopecia. This report holds significance as it highlights a previously unreported side effect, thereby enhancing our understanding of emerging therapies for brain tumors and their associated adverse reactions.

2903: Is pseudoprogression common in low grade glioma patients treated with radiotherapy?

2903: Is pseudoprogression common in low grade glioma patients treated with radiotherapy?

Direct DNA binding by BRCA1 on β-hCG promoter and its clinical implications

ObjectiveThe role of β-hCG in breast cancer is largely unknown, this study aims to analyse the gene expression and clinical implications of β-hCG and its isoforms in various cancers focussing particularly in Breast Invasive Carcinoma (BRCA). A mechanistic approach deciphering the transcriptional regulation of β-hCG by BRCA1 was also explored. MethodsData from various comprehensive gene expression platforms like UALCAN, GEPIA2, GENT2, TIMER2, LinkedOmics, and STRING were used to analyse the expression of β-hCG and its clinical implications; Immunohistochemistry and ELISA for β-hCG expression analysis from human breast cancer patients; Electrophoretic mobility shift assay (EMSA) to analyse the direct binding of BRCA1 on β-hCG; Immunoblotting and Luciferase assay to understand the regulation of β-hCG by p53 were performed. ResultsResults from UALCAN and GENT2 gene expression cancer database revealed that TNBC subtypes and high-grade metaplastic carcinoma shows elevated expression of β-hCG and infiltration of various immune cells were also identified in BRCA by TIMER2. It was observed that most of the isoforms of β-hCG (CGB) are upregulated in breast cancers irrespective of hormonal status when BRCA1 gene is mutated according to TIMER2. Similar results were observed with Lymphoid neoplasm diffuse large B-cell lymphoma (LGG) and DLBC (Brain lower grade glioma) when BRCA1 is mutated. These results correlate with our earlier reports indicating expression of β-hCG in BRCA1 defective condition. We have also identified direct binding of BRCA1 on β-hCG promoter. ConclusionAll these findings demonstrate the importance of β-hCG as a potential target in BRCA1-deficient carcinomas.

1085: Pencil-beam scanning proton therapy for low-grade glioma patients

1085: Pencil-beam scanning proton therapy for low-grade glioma patients

Multimodal Deep Learning Improves Recurrence Risk Prediction in Pediatric Low-Grade Gliomas.

Postoperative recurrence risk for pediatric low-grade gliomas (pLGGs) is challenging to predict by conventional clinical, radiographic, and genomic factors. We investigated if deep learning of MRI tumor features could improve postoperative pLGG risk stratification. We used pre-trained deep learning (DL) tool designed for pLGG segmentation to extract pLGG imaging features from preoperative T2-weighted MRI from patients who underwent surgery (DL-MRI features). Patients were pooled from two institutions: Dana Farber/Boston Children's Hospital (DF/BCH) and the Children's Brain Tumor Network (CBTN). We trained three DL logistic hazard models to predict postoperative event-free survival (EFS) probabilities with 1) clinical features, 2) DL-MRI features, and 3) multimodal (clinical and DL-MRI features). We evaluated the models with a time-dependent Concordance Index (Ctd) and risk group stratification with Kaplan Meier plots and log-rank tests. We developed an automated pipeline integrating pLGG segmentation and EFS prediction with the best model. Of the 396 patients analyzed (median follow-up: 85 months, range: 1.5-329 months), 214 (54%) underwent gross total resection and 110 (28%) recurred. The multimodal model improved EFS prediction compared to the DL-MRI and clinical models (Ctd: 0.85 (95% CI: 0.81-0.93), 0.79 (95% CI: 0.70-0.88), and 0.72 (95% CI: 0.57-0.77), respectively). The multimodal model improved risk-group stratification (3-year EFS for predicted high-risk: 31% versus low-risk: 92%, p<0.0001). DL extracts imaging features that can inform postoperative recurrence prediction for pLGG. Multimodal DL improves postoperative risk stratification for pLGG and may guide postoperative decision-making. Larger, multicenter training data may be needed to improve model generalizability.