GBM Model Research Articles

Prediction of bio-oil yield by machine learning model based on 'enhanced data' training

Bio-oil is widely used and has great application potential. With the development of artificial intelligence, machine learning has been gradually applied in the field of biomass, the data augmentation method is a common operation method for training models in the field of computer or data processing. In this study, the concept of 'enhanced data' was proposed for the first time. According to the composition characteristics of biomass raw materials, the sample data were divided into three different sample sets. Based on the light gradient boosting machine (Light GBM) and deep neural network (DNN) algorithm, a prediction model of bio-oil yield was established. Combined with the analysis of partial correlation, the influence of 'enhanced data' on the prediction accuracy of the model was explored. The results showed that the prediction accuracy of the model was improved to a certain extent after adding 'enhanced data'. The Light GBM model was more suitable for predicting bio-oil yield, and the Light GBM _ c model performed best, with R2 of 0.894, MAE of 3.622, and RMSE of 4.445. At the same time, this study also has certain reference significance for other prediction studies in the field of biomass thermochemical conversion.

Evaluating the pedestrian level of service for varying trip purposes using machine learning algorithms

The pedestrians’ feeling of comfort while walking on footpaths varies according to the time of day, environment, and the purpose of the trip. The quality of service offered by pedestrian facilities such as walkways, intersections, and public places is evaluated by the Pedestrian level of service (PLOS) and has been measured from time to time, to upgrade and maintain the sustainable travel choice of people. This paper aims to focus on the level of service based on three main trip purposes such as work, education, and recreation, while considering various path characteristics and pedestrian flow characteristics that affect the pedestrian’s feeling of comfort on the walkways. The data has been collected using pedestrian questionnaire surveys and pedestrian sensors in the Melbourne central business district and the significant factors that influence the PLOS for each trip purpose will be chosen using the Mutual Information gain, which is found to be different for each trip purpose. The major influencing factors that affect the PLOS will be used to develop machine learning models for three trip purposes separately using Random Forest and Light-GBM algorithm in Python. The accuracy of prediction using the light GBM model is 0.74 for education, 0.80 for recreation, and 0.70 for work trip purposes. It is found using SHAP which stands for Shapely Additive explanations that the factors such as interpersonal distance, distance from vehicles, construction sites, vehicle volume, traffic noise, and footpath surface are the most influencing variables that affect the PLOS based on three different trip purposes.

Radiomics-based model for prediction of TGF-β1 expression in head and neck squamous cell carcinoma.

To explore the connection between TGF-β1 expression and the survival of patients with head and neck squamous cell carcinoma (HNSCC), as well as whether non-invasive CT-based Radiomics can predict TGF-β1 expression in HNSCC patients. Data on transcriptional profiling and clinical information were acquired from the TCGA database and subsequently categorized based on the TGF-β1 expression cutoff value. Based on the completeness of enhanced arterial phase CT scans, 139 HNSCC patients were selected. The PyRadiomics package was used to extract radiomic features, and the 3D Slicer software was used for image segmentation. Using the mRMR_RFE and Repeat LASSO algorithms, the optimal features for establishing the corresponding gradient enhancement prediction models were identified. A survival analysis was performed on 483 patients, who were divided into two groups based on the TGF-β1 expression cut-off. The Kaplan-Meier curve indicated that TGF-β1 was a significant independent risk factor that reduced patient survival. To construct gradient enhancement prediction models, we used the mRMR_RFE algorithm and the Repeat_LASSO algorithm to obtain two features (glrlm and ngtdm) and three radiation features (glrlm, first order_10percentile, and gldm). In both the training and validation cohorts, the two established models demonstrated strong predictive potential. Furthermore, there was no statistically significant difference in the calibration curve, DCA diagram, or AUC values between the mRMR_RFE_GBM model and the LASSO_GBM model, suggesting that both models fit well. Based on these findings, TGF-β1 was shown to be significantly associated with a poor prognosis and to be a potential risk factor for HNSCC. Furthermore, by employing the mRMR_RFE_GBM and Repeat_LASSO_GBM models, we were able to effectively predict TGF-β1 expression levels in HNSCC through non-invasive CT-based Radiomics.

Metformin-induced reduction of CCR8 enhances the anti-tumor immune response of PD-1 immunotherapy in glioblastoma

Immunotherapy strategies targeting the programmed cell death 1 (PD-1) in clinical treatments have shown limited success in controlling glioblastoma malignancies. Metformin exserts antitumor function, yet the underlying mechanisms remain unclear. Here, we investigated whether metformin could enhance the effectiveness of anti-PD-1 therapy by activating the immune system. Whether combination of an anti-PD-1 antibody or not, metformin significantly increased tumor-infiltrating CD4+ T cells while decreased regulatory T (Treg) cells in a mouse GBM model. Additionally, metformin reduced CC motif chemokine receptor CCR8 and elevated Interleukin 17 A (IL-17 A) expressions. Mechanistically, metformin reduces histone acetylation at the CCR8 promotor and inhibits CCR8 expression by upregulating AMP-activated protein kinase (AMPK)-activated sirtuin 2 (SIRT2). Metformin enhances the effectiveness of anti-PD-1 immunotherapy by reducing CCR8 expression on tumor-infiltrating Treg cells, suggesting that metformin has an antitumor effect by alleviating immunosuppression and promoting T cell-mediated immune response.

Simultaneous Modulation of Hypoxia And Metabolism in Glioblastoma for Enhanced Radio‐Immunotherapy

AbstractGlioblastoma (GBM) is the most invasive and lethal primary brain melanoma. The existing treatment modality is unable to achieve thorough elimination of GBM due to the aggressive nature, blood‐brain barrier (BBB), hypoxic environment, and heterogeneous cellular components. Herein, a radio‐immunotherapy regimen based on a versatile nanoplatform (G/APH‐M) is proposed to effectively kill quiescent cancer stem cells (CSCs) and proliferative cancer cells in GBM in a simultaneous manner. Among o ur prepared G/APH‐M, the coating of GL261 cell membrane guarantees the BBB‐penetrable delivery and homologous GBM‐targeting of the hollow Prussian blue loaded with oxidative phosphorylation inhibitor Gboxin and catalase‐mimetic nanozymes. After substantial tumor accumulation, the released Gboxin inhibits mitochondrial oxidative phosphorylation to kill CSCs, while the nanozymes catalyze the production of oxygen to enhance radiotherapy. Consequently, potent immunogenic cell death (ICD) of GBM is induced, which in combination with immune checkpoint inhibitors (αPD‐L1), achieving a potent therapeutic effect with an 80% survival rate in the orthotopic GBM model even at 60 days after the treatment. The synergistic modulation of hypoxia and metabolism based on G/APH‐M greatly intensifies the radio‐immunotherapy of GBM, which would inspire more comprehensive strategies targeting the multiple characteristics of GBM cells for clinical benefits.

Analysis of the Practical Path of Vocational Education Serving Rural Revitalization by Digital Means and Philosophy of Technology

Abstract In this paper, based on digital means and technical philosophy, the histogram algorithm in the Light GBM model is used to calculate the floating point values of the raw data for the analysis of vocational education service for rural revitalization so that each of its features is converted into a histogram. In order to prevent the feature training overfitting problem of the Light GBM model, the LightGBM model is optimized by the iterative tree MPA algorithm, and the prediction model of returning to poverty risk based on IMPA-LightGBM is constructed. Starting from the current situation of vocational education service for rural revitalization, we put forward research hypotheses to realize the research design of vocational education accurate poverty alleviation service for rural revitalization and carry out an example analysis of vocational education service for rural revitalization combined with digital technology. The results show that in terms of model performance, the WAPE values of the return-to-poor risk values obtained from the prediction of the IMPA-LightGBM model are all lower than 5.5%, so the prediction effect is relatively satisfactory, and the return-to-poor risk values of poverty-eradicating households can be effectively predicted. On the practice road analysis, the standard deviation (SD) of the rural revitalization development index in China as a whole decreased from 0.63 to 0.52, which means that the differences in rural revitalization among provinces are decreasing. This study explores the synergistic development of the community of interest between vocational education and rural revitalization through the cultivation of new vocational farmers.

Focused ultrasound combined with radiotherapy for malignant brain tumor: a preclinical and clinical study.

Blood-brain barrier (BBB) remains to be the major obstacle to conquer in treating patients with malignant brain tumors. Radiation therapy (RT), despite being the mainstay adjuvant modality regardless of BBB, the effect of radiation induced cell death is hindered by the hypoxic microenvironment. Focused ultrasound (FUS) combined with systemic microbubbles has been shown not only to open BBB but also potentially increased regional perfusion. However, no clinical study has investigated the combination of RT with FUS-BBB opening (RT-FUS). We aimed to provide preclinical evidence of RT-FUS combination in GBM animal model, and to report an interim analysis of an ongoing single arm, prospective, pilot study (NCT01628406) of combining RT-FUS for recurrent malignant high grade glioma patients, of whom re-RT was considered for disease control. In both preclinical and clinical studies, FUS-BBB opening was conducted within 2h before RT. Treatment responses were evaluated by objective response rate (ORR) using magnetic resonance imaging, progression free survival, and overall survival, and adverse events (AE) in clinical study. Survival analysis was performed in preclinical study and descriptive analysis was performed in clinical study. In mouse GBM model, the survival analysis showed RT-FUS (2Gy) group was significantly longer than RT (2Gy) group and control, but not RT (5Gy) group. In the pilot clinical trial, an interim analysis of six recurrent malignant high grade glioma patients underwent a total of 24 RT-FUS treatments was presented. Three patients had rapid disease progression at a mean of 33days after RT-FUS, while another three patients had at least stable disease (mean 323days) after RT-FUS with or without salvage chemotherapy or target therapy. One patient had partial response after RT-FUS, making the ORR of 16.7%. There was no FUS-related AEs, but one (16.7%) re-RT-related grade three radiation necrosis. Reirradiation is becoming an option after disease recurrence for both primary and secondary malignant brain tumors since systemic therapy significantly prolongs survival in cancer patients. The mechanism behind the synergistic effect of RT-FUS in preclinical model needs further study. The clinical evidence from the interim analysis of an ongoing clinical trial (NCT01628406) showed a combination of RT-FUS was safe (no FUS-related adverse effect). A comprehensive analysis of radiation dosimetry and FUS energy distribution is expected after completing the final recruitment.

High-Resolution PM2.5 Concentrations Estimation Based on Stacked Ensemble Learning Model Using Multi-Source Satellite TOA Data

Nepal has experienced severe fine particulate matter (PM2.5) pollution in recent years. However, few studies have focused on the distribution of PM2.5 and its variations in Nepal. Although many researchers have developed PM2.5 estimation models, these models have mainly focused on the kilometer scale, which cannot provide accurate spatial distribution of PM2.5 pollution. Based on Gaofen-1/6 and Landsat-8/9 satellite data, we developed a stacked ensemble learning model (named XGBLL) combined with meteorological data, ground PM2.5 concentrations, ground elevation, and population data. The model includes two layers: a XGBoost and Light GBM model in the first layer, and a linear regression model in the second layer. The accuracy of XGBLL model is better than that of a single model, and the fusion of multi-source satellite remote sensing data effectively improves the spatial coverage of PM2.5 concentrations. Besides, the spatial distribution of the daily mean PM2.5 concentrations in the Kathmandu region under different air conditions was analyzed. The validation results showed that the monthly averaged dataset was accurate (R2 = 0.80 and root mean square error = 7.07). In addition, compared to previous satellite PM2.5 datasets in Nepal, the dataset produced in this study achieved superior accuracy and spatial resolution.

TMIC-51. THE RELEASE OF INFLAMMATORY CYTOKINES ARE REGULATED BY PYK2 SIGNALING IN GBM CELLS

Abstract Glioblastoma (GBM) is a highly aggressive brain cancer that is typically fatal within a year of diagnosis. Previous research has shown a positive relationship between the activation of a protein called proline-rich non-receptor tyrosine kinase (Pyk2) in GBM cells in addition to the expression of cytokines by tumor-infiltrating myeloid (TIM cells). Leading to a sustained increase in the expression levels of monocyte chemoattractant protein 1 (CCL2), granulocyte-macrophage colony-stimulating factor (GM-CSF), IL10, IL12, and vascular endothelial growth factor (VEGF), as well as an increase in the CD86+/CD206+ TIM population. We hypothesize that Pyk2 signaling in GBM cells is involved in the regulation of the release of inflammatory cytokines, which in turn suppresses the immune response in the tumor microenvironment. We used CRISPR/Cas9 knockout Pyk2 (Pyk2KO) and wild type (Pyk2WT) GL261 mouse glioma cells in this study. Analysis of the cytokines present in the cell culture medium conditioned by Pyk2KO and Pyk2WT cells identified a decrease in the release of CCL2 and CCL5 in the Pyk2KO cells compared to the Pyk2WT cells. RT-PCR analysis of myeloid cell population, isolated from Pyk2WT and Pyk2KO tumors in the GL261-C57Bl/6 GBM model, identified a significant down-regulation of key pro-tumorigenic factors such as CCL2, CCL12, CCL5, tumor necrosis factor (TNF), VEGF and epidermal growth factor (EGF) in the Pyk2KO tumors compared to the Pyk2WT tumors. Additionally, flow cytometry analysis revealed a reduction in myeloid-derived suppressor cells (MDSCs), an increase in lymphoid dendritic cells (DC), and up-regulation in TNF/IFNg expressing and antigen-specific CD8+ T cells in Pyk2KO tumors. Furthermore, a significant decrease in pro-tumorigenic Ly6C+/CD206+ and an up-regulation of inflammatory Ly6C+/CD86+ myeloid cells were found in the Pyk2KO tumors. These findings suggest that Pyk2KO tumors display an immunocompetent microenvironment with enhanced phagocytic and cytotoxic function compared to Pyk2WT tumors. NIH Grant 1SC1GM122691 and PRSTRT2022 supported this study

MODL-50. CLONAL EXPANSION OF SPONTANEOUS AND INJECTED TUMORS IN SYNGENEIC MOUSE MODELS OF GLIOBLASTOMA

Abstract INTRODUCTION Glioblastoma (GBM) tumors are highly heterogenous and plastic. To better understand how tumors respond to treatment, it is important to use immunocompetent mouse models. We've created various mouse GBM models with common genetic mutations found in different subtypes of GBM, including classical, proneural, mesenchymal, and BRAF-driven. METHODS Using a combination of CRISPR and PiggyBac Transposon constructs, we injected plasmid mixtures into the ventricles of the mouse brain during the post-natal period. Electroporation was then used to enable the plasmids to enter nearby neural stem cells, which led to the spontaneous formation of tumors (spontaneous model). We collected tumor cells from these models and injected them into the adult mouse brain to create the injected mouse GBM model. Here we use spatial and single cell transcriptome profiling approaches to characterize these models at endpoint. Spatial transcriptome data was collected using the Visium 10X platform and tumor spots were deconvoluted from non-malignant spots using consensus non-negative matrix factorization. Copy number alterations (CNAs) were inferred using non-malignant spots and single-cell clusters as a reference. RESULTS We found that copy number states are largely homogenous within each sample, supporting an early population expansion in vivo. CNAs found in spontaneous tumors are also found in the matched injected tumors, which in turn harbour additional CNAs, suggesting that further selection occurs in vitro. Some of the CNAs include relevant cancer genes, suggesting the observed clonal expansions are potentially driven by additional somatic alterations that cooperate with the engineered driver mutations. Studying these models using orthogonal methods will allow us to discover alterations that synergize with the original drivers and delineate principles of GBM initiation and progression in an immune-competent environment.

TMIC-52. HISTONE H3K27ME3 LEVEL GOVERN TUMOR-ASSOCIATED MICROGLIA REPROGRAMMING IN RESPONSE TO GBM

Abstract Glioblastoma (GBM) is the most common malignant primary brain tumor and its invasive nature makes complete dissection difficult. However, how GBM invade in the brain is poorly understood. Although, microglia are the resident immune cells in the brain that can support GBM invasion, their roles are yet elusive. Here, we retraced early steps of GBM invasion and their interaction with tumor-associated microglia (TAM) in a highly infiltrative murine GBM model in immunocompetent background. We show that microglia are activated in a wide tumor field ahead of GBM invasion, followed by glial net formation, encircling invading tumors. Next, physical contacts with GBM cells initiate an astounding morphological, and functional transformation of microglia and marrow-derived macrophages to form collectively organized migration streams with intertwined GBM cells. Here we show that Kdm6b, a H3K27me3 demethylase, is induced in TAMs during this process. Intriguingly, GBM disseminated more widely in juvenile host brains, which was associated with highly induced Kdm6b expression. Moreover, we show that Kdm6b inhibition with small chemical inhibitor GSKJ-4, perturbs TAM reprogramming, and GBM invasion. Together, our data reveal that microglia reprogramming provide directional cues for GBM invasion through epigenetic mechanisms and this may give insights to halt GBM invasion.

STEM-04. FATE-BASED CONVERSION FOR IN SITU VACCINATION IN GLIOBLASTOMA

Abstract BACKGROUND Although immune checkpoint inhibitors have shown high benefit for other solid tumors, their efficacy is limited in GBM due in large part to the “cold” tumor microenvironment (TME) of the GBM tumor. New approaches are necessary to overcome these challenges. We propose a shift in the scientific and treatment paradigm of cancer immunotherapy, calling for tumor cells to take on a central, active role in initiating their own immunity right in the TME. METHODS Using AI, we identified cell fate determinants (CFDs) to convert GBM cells directly into induced dendritic cells (iDC) inside the TME. CFDs were delivered using a viral vector and conversion assessed by immunophenotyping, scRNA-seq, and functional assays for classical DC properties. RESULTS A four-CFD subnetwork anchored by PU.1 was sufficient to convert mouse GBM cells to iDCs (CD45+MHCII+CD80/86+MHCI+). iDCs are growth arrested, exhibit 3-fold higher phagocytic activity and upregulate the canonical antigen processing and presenting machineries by 10-40-fold, resulting in 100-fold greater efficiency at processing ovalbumin and cross-presenting SIINFEKL to naive OTI-CD8+ cytotoxic T lymphocytes (CTL). In addition, iDCs efficiently cross-present exogenous antigens to naïve CD8+ CTLs and elicit >30-fold higher activation and cytotoxicity in tumor-specific T cells compared to native GBM cells, confirming their DC-like properties. Lastly, intratumoral GBM-DC conversion in a syngeneic orthotopic GBM model resulted in significant reduction in tumor burden and increase in survival compared to controls (Log-rank, p = 0.01) with synergistic efficacy with ICIs and classical DC-based tumor vaccination. We have also successfully converted human GBM cells to iDCs with similar properties using a 2-CFD combination also anchored by PU.1. CONCLUSIONS GBM-derived iDCs acquired DC-like functions and were capable of eliciting robust intratumor immune activation, thereby overcoming many current limitations of immunotherapy in GBM. Our goal is to develop a low cost, off-the-shelf gene therapy-based fate conversion tumor immunotherapy.

A pilot study of implication of machine learning for relapse prediction after allogeneic stem cell transplantation in adults with Ph-positive acute lymphoblastic leukemia

The posttransplant relapse in Ph-positive ALL increases the risk of death. There is an unmet need for instruments to predict the risk of relapse and plan prophylaxis. In this study, we analyzed posttransplant data by machine learning algorithms. Seventy-four Ph-positive ALL patients with a median age of 30 (range 18–55) years who previously underwent allo-HSCT, were retrospectively enrolled. Ninety-three percent of patients received prophylactic/preemptive TKIs after allo-HSCT. The values of the BCR::ABL1 level at serial assessments and over variables were collected in specified intervals after allo-HSCT. They were used to model relapse risk with several machine-learning approaches. GBM proved superior to the other algorithms and provided a maximal AUC score of 0.91. BCR::ABL1 level before and after allo-HSCT, prediction moment, and chronic GvHD had the highest value in the model. It was shown that after Day + 100, both error rates do not exceed 22%, while before D + 100, the model fails to make accurate predictions. As a result, we determined BCR::ABL1 levels at which the relapse risk remains low. Thus, the current BCR::ABL1 level less than 0.06% in patients with chronic GvHD predicts low risk of relapse. At the same time, patients without chronic GVHD after allo-HSCT should be classified as high risk with any level of BCR::ABL1. GBM model with posttransplant laboratory values of BCR::ABL1 provides a high prediction of relapse after allo-HSCT in the era of TKIs prophylaxis. Validation of this approach is warranted.

Study of online learning time and learning performance model based on learning platform

Abstract In the new online learning environment where information technology and education are integrated, it is important to understand the learning patterns of online learners and then predict their learning outcomes to improve learning effectiveness of online learners. In this paper, we collect students’ online learning behavior data through an online learning platform and further investigate the influence mechanism between online learning behavior characteristics and learning performance with the help of the OCCP classification model and the 3D S-F-T model. Meanwhile, the Light GBM algorithm is based on its advantages in noise and distributed processing based on its high accuracy and strong generalization ability in classification problems. The Light GBM algorithm constructs the performance prediction model for the selected learning behavior feature data. The results show that the accuracy of academic achievement prediction based on the Light GBM model is 0.85, and the accuracy of sample prediction classification is H (85%), M (73%) and L (89%), indicating that the model is effective for academic achievement prediction. The experimental results confirm that the prediction method can more accurately identify at-risk students and provide more intuitive and accurate feedback on learners’ problems and deficiencies in the learning process to support the implementation of personalized education interventions.

Using automated machine learning techniques to explore key factors in anaerobic digestion: At the environmental factor, microorganisms and system levels

Exploring key factors has important guidance for understanding complex anaerobic digestion (AD) systems. This study proposed a multi-layer automated machine learning framework to understand the complex interactions in AD systems and explore key factors at the environmental factor, microorganisms and system levels. The first layer of the framework identified hydraulic residence time (HRT) as the most important environmental factor, with an optimal range of 33–45 d. In the second layer of the framework, Methanocelleus (optimal relative abundance (ORA) = 3.0%) and Candidatus_Caldatribacterium (ORA = 1.7%) were found to be the key archaea and bacteria, respectively. Furthermore, the prediction of key microorganisms based on environmental factors and remaining microbial data showed the essential roles of Methanothermobacter and Acetomicrobium. The third layer for finding the optimal combination of data variables for predicting biogas production demonstrated that combined Archaea genera and environmental factors should be achieved for the most accurate prediction (root mean square error (RMSE) = 84.21). GBM had the best model performance and prediction accuracy among all the built-in models. Based on the optimal GBM model, the analysis at the system level showed that HRT was the most important variable. However the most important microorganism, Methanocelleus, within the appropriate survival range is also essential to achieve optimal biogas production. This research explores key parameters at various levels through automated machine learning techniques, which are expected to provide guidance in understanding the complex architecture of industrial and laboratory AD systems.

P12.05.B THE ROLE OF ONCO-METABOLITE IN THE IDH MUTANT GLIOBLASTOMA MICROENVIRONMENT

Abstract BACKGROUND Recently, we found that the reactive astrocytes in the IDH wild type glioblastoma contribute to anti-tumor immunity and support pro-oncogenic signaling. Here, we characterized the transcriptomic signature of IDH1/2-mutant glioma associated astrocytes and determined a unique inflammatory transformation, profoundly different to astrocytes in IDH wildtype glioma patients due to an onco-metabolite R-2-hydroxyglutarate using Next generation sequencing and human ex-vivo slice model. MATERIAL AND METHODS We purified and transcriptionally profiled astrocytes from 9 patients with confirmed IDH1-R132H mutation, by means of RNA-sequencing and the data were analyzed using the established pipelines. We also used spatial transcriptomics to evidently show the spatial distribution of astrocytes in IDH-mutated/wildtype glioma samples. We validated our findings using human organotypic slice model inoculated with IDH-mutant cell line or treated with oncometabolite 2-hydroxy glutarate. Additionally, LC-MS was further used to give us a chart of neurotransmitters due to altered microenvironment. RESULTS Our results from RNA sequencing showed a transcriptional transformation of reactive astrocytes within the microenvironment of IDH-mutated tumors compared to wildtype glioma by means of RNA-sequencing of purified astrocytes. And, using our established human neocortical GBM model inoculated with IDH mutant tumor and R-2HG treatment, we showed that we were able to activate inflammatory transcriptional programs in astrocytes, mediated by the presence of microglia. Further, by spatially mapping the transcriptomic profiles of purified microglia, we were able to confirm that microglia also demonstrate inflammatory activation in IDH mutated glioma. This inflammatory astrocyte transformation is associated with a loss of neurotransmitter homeostasis (disrupted levels of glutamate) in the treated sections, as has been previously reported in IDH mutated tumors. Additionally, R-H2G increased neuronal spiking rate in, pointing to potential excitotoxicity. CONCLUSION Our work provides a crucial contribution towards understanding the role of R-2HG in the IDH mutant glioma microenvironment and sheds light on the significant microenvironmental differences to IDH wild-type glioma.

P10.26.B EXPLORING POTENTIAL ANTITUMOR EFFECTS OF CANNABIDIOL AND TETRAHYDROCANNABINOL IN A PATIENT-DERIVED GLIOBLASTOMA MODEL

Abstract BACKGROUND Glioblastoma (GBM) is a CNS malignancy which is in dire need of novel treatment options. Drug repurposing has become a means to circumvent major obstacles that accompany drug development for GBM. The cannabinoids cannabidiol (CBD) and tetrahydrocannabinol (THC) have garnered the interest of the scientific community for the treatment of GBM as they possess pleiotropic effects in other pathologies. In the present study we are investigating potential antitumor effects of these cannabinoids by using our patient derived GBM model which recapitulates GBM in its genetic make-up and heterogeneity. MATERIAL AND METHODS To assess potential antitumor effects in vitro, we performed a screen on 20 patient derived GBM cultures to obtain a sensitivity profile for both cannabinoids. These cultures ranged in background regarding primary/recurrence and IDH mutation. To investigate effects of these cannabinoids on GBM radio-and-chemosensitivity, combination studies with either lomustine (CCNU) or irradiation (RTx) were performed on recurrent GBM cultures. The zero-interaction potential method was used to quantify synergy between the cannabinoids. The enhancement factor was used to quantify potential radiosensitization or ratio-effects between the cannabinoids. RESULTS Screening of both cannabinoids revealed CBD to possess a higher level of cytotoxicity than THC. IC50 values of CBD ranged from 2.5 - 18.5 µM and the values of THC ranged from 7.20 - 47.60 µM. In a panel of 8 cultures, combinations of CBD with THC, CBD with CCNU and THC with CCNU was tested to gain insight in combination effects. In 1/8 cultures synergy was observed between CBD and THC. In 1/8 cultures synergy between CBD and CCNU was observed and in 4/8 cultures synergy between THC and CCNU. When combined with irradiation, no enhancement was observed in any of the 4 cultures tested. No ratio-dependent enhancement between the cannabinoids was observed when combined in a 1:1, 1:4 or 4:1 ratio of CBD:THC. CONCLUSION Screening of the cannabinoids CBD and THC revealed that CBD displays a stronger cytotoxic effect than THC. Combinations of either of the two with each other or CCNU revealed moderate additive effects in the majority of cultures and synergy in a small subset. We did not observe radiosensitizing effects in our patient derived GBM model. Relating in vitro found cytotoxic concentrations to clinically achievable concentrations, we found that CBD requires approximately 3 times higher plasma concentrations and THC >40 times to achieve an antitumor effect. Currently we are investigating long term treatment effects of CBD and its mechanism of action.

Characterization of a commercial bioluminescence tomography-guided system for pre-clinical radiation research.

Widely used Cone-beam computed tomography (CBCT)-guided irradiators have limitations in localizing soft tissue targets growing in a low-contrast environment. This hinders small animal irradiators achieving precise focal irradiation. To advance image-guidance for soft tissue targeting, we developed a commercial-grade bioluminescence tomography-guided system (BLT, MuriGlo) for pre-clinical radiation research. We characterized the system performance and demonstrated its capability in target localization. We expect this study can provide a comprehensive guideline for the community in utilizing the BLT system for radiation studies. MuriGlo consists of four mirrors, filters, lens, and charge-coupled device (CCD) camera, enabling a compact imaging platform and multi-projection and multi-spectral BLT. A newly developed mouse bed allows animals imaged in MuriGlo and transferred to a small animal radiation research platform (SARRP) for CBCT imaging and BLT-guided irradiation. Methods and tools were developed to evaluate the CCD response linearity, minimal detectable signal, focusing, spatial resolution, distortion, and uniformity. A transparent polycarbonate plate covering the middle of the mouse bed was used to support and image animals from underneath the bed. We investigated its effect on 2D Bioluminescence images and 3D BLT reconstruction accuracy, and studied its dosimetric impact along with the rest of mouse bed. A method based on pinhole camera model was developed to map multi-projection bioluminescence images to the object surface generated from CBCT image. The mapped bioluminescence images were used as the input data for the optical reconstruction. To account for free space light propagation from object surface to optical detector, a spectral derivative (SD) method was implemented for BLT reconstruction. We assessed the use of the SD data (ratio imaging of adjacent wavelength) in mitigating out of focusing and non-uniformity seen in the images. A mouse phantom was used to validate the data mapping. The phantom and an in vivo glioblastoma model were utilized to demonstrate the accuracy of the BLT target localization. The CCD response shows good linearity with<0.6% residual from a linear fit. The minimal detectable level is 972 counts for 10×10 binning. The focal plane position is within the range of 13-18mm above the mouse bed. The spatial resolution of 2D optical imaging is<0.3mm at Rayleigh criterion. Within the region of interest, the image uniformity is within 5% variation, and image shift due to distortion is within 0.3mm. The transparent plate caused<6% light attenuation. The use of the SD imaging data can effectively mitigate out of focusing, image non-uniformity, and the plate attenuation, to support accurate multi-spectral BLT reconstruction. There is<0.5% attenuation on dose delivery caused by the bed. The accuracy of data mapping from the 2D bioluminescence images to CBCT image is within 0.7mm. Our phantom test shows the BLT system can localize a bioluminescent target within 1mm with an optimal threshold and only 0.2mm deviation was observed for the case with and without a transparent plate. The same localization accuracy can be maintained for the in vivo GBM model. This work is the first systematic study in characterizing the commercial BLT-guided system. The information and methods developed will be useful for the community to utilize the imaging system for image-guided radiation research.

Development of a rabbit human glioblastoma model for testing of endovascular selective intra-arterial infusion (ESIA) of novel stem cell-based therapeutics.

Endovascular selective intra-arterial (ESIA) infusion of cellular oncotherapeutics is a rapidly evolving strategy for treating glioblastoma. Evaluation of ESIA infusion requires a unique animal model. Our goal was to create a rabbit human GBM model to test IA infusions of cellular therapies and to test its usefulness by employing clinical-grade microcatheters and infusion methods to deliver mesenchymal stem cells loaded with an oncolytic adenovirus, Delta-24-RGD (MSC-D24). Rabbits were immunosuppressed with mycophenolate mofetil, dexamethasone, and tacrolimus. They underwent stereotactic xenoimplantation of human GBM cell lines (U87, MDA-GSC-17, and MDA-GSC-8-11) into the right frontal lobe. Tumor formation was confirmed on magnetic resonance imaging, histologic, and immunohistochemistry analysis. Selective microcatheter infusion of MSC-D24 was performed via the ipsilateral internal carotid artery to assess model utility and the efficacy and safety of this approach. Twenty-five rabbits were implanted (18 with U87, 2 MDA-GSC-17, and 5 MDA-GSC-8-11). Tumors formed in 68% of rabbits (77.8% for U87, 50.0% for MDA-GSC-17, and 40.0% for MDA-GSC-8-11). On MRI, the tumors were hyperintense on T2-weighted image with variable enhancement (evidence of blood brain barrier breakdown). Histologically, tumors showed phenotypic traits of human GBM including varying levels of vascularity. ESIA infusion into the distal internal carotid artery of 2 ml of MSCs-D24 (107 cells) was safe in the model. Examination of post infusion specimens documented that MSCs-D24 homed to the implanted tumor at 24 hours. The intracranial immunosuppressed rabbit human GBM model allows testing of ESIA infusion of novel therapeutics (eg, MSC-D24) in a clinically relevant fashion.

Development and validation of a risk prediction model for cage subsidence after instrumented posterior lumbar fusion based on machine learning: a retrospective observational cohort study.

Interbody cage subsidence is a common complication after instrumented posterior lumbar fusion surgery, several previous studies have shown that cage subsidence is related to multiple factors. But the current research has not combined these factors to predict the subsidence, there is a lack of an individualized and comprehensive evaluation of the risk of cage subsidence following the surgery. So we attempt to identify potential risk factors and develop a risk prediction model that can predict the possibility of subsidence by providing a Cage Subsidence Score (CSS) after surgery, and evaluate whether machine learning-related techniques can effectively predict the subsidence. This study reviewed 59 patients who underwent posterior lumbar fusion in our hospital from 2014 to 2019. They were divided into a subsidence group and a non-subsidence group according to whether the interbody fusion cage subsidence occurred during follow-up. Data were collected on the patient, including age, sex, cage segment, number of fusion segments, preoperative space height, postoperative space height, preoperative L4 lordosis Angle, postoperative L4 lordosis Angle, preoperative L5 lordosis Angle, postoperative PT, postoperative SS, postoperative PI. The conventional statistical analysis method was used to find potential risk factors that can lead to subsidence, then the results were incorporated into stepwise regression and machine learning algorithms, respectively, to build a model that could predict the subsidence. Finally the diagnostic efficiency of prediction is verified. Univariate analysis showed significant differences in pre-/postoperative intervertebral disc height, postoperative L4 segment lordosis, postoperative PT, and postoperative SS between the subsidence group and the non-subsidence group (p < 0.05). The CSS was trained by stepwise regression: 2 points for postoperative disc height > 14.68 mm, 3 points for postoperative L4 segment lordosis angle >16.91°, and 4 points for postoperative PT > 22.69°. If the total score is larger than 0.5, it is the high-risk subsidence group, while less than 0.5 is low-risk. The score obtains the area under the curve (AUC) of 0.857 and 0.806 in the development and validation set, respectively. The AUC of the GBM model based on the machine learning algorithm to predict the risk in the training set is 0.971 and the validation set is 0.889. The AUC of the avNNet model reached 0.931 in the training set and 0.868 in the validation set, respectively. The machine learning algorithm has advantages in some indicators, and we have preliminarily established a CSS that can predict the risk of postoperative subsidence after lumbar fusion and confirmed the important application prospect of machine learning in solving practical clinical problems.