Citations

Abstract Glioblastoma is the most aggressive malignant brain tumor with the poor survival due to its invasive nature driven by cell migration. There is still no effective treatment targeting glioma cell migration based on their transcriptomic information, due to the lack of understanding in the links to the fundamental mechanics of cell migration. In the previous studies, we developed a physics-based motor-clutch model, the cell migration simulator (CMS), to identify the basic mechanisms of cell migration in various cell types. In this study, we found that glioblastoma patient-derived (xenograft) (PD(X)) cell lines across mesenchymal (MES), proneural (PN), classical (CL) subtypes and two institutions (N = 13 patients) had optimal motility and traction force on stiffnesses around 9.3kPa, with otherwise heterogeneous and uncorrelated motility, traction, and F-actin flow. We applied the CMS as a consistent framework to describe the physical parameters of glioblastoma patient cell migration. Using single cell migration and force generation data on compliant 2D surfaces, we found all glioma PD(X) lines (11 patients) had balanced myosin/adhesion ratios resulting in effective migration. Mesenchymal (MES) lines had enhanced F-actin polymerization resulting in higher motility. Adherent PDX lines had increased myosin-adhesion amount resulting in higher traction force. The CMS also predicted the differential cell migration under cytoskeletal drugs between glioma subtypes. Finally, we found 18 genes in actin cytoskeleton and focal adhesion pathways significantly correlated with the glioma physical parameters, which were being used to estimate the physical parameters of a larger cohort (66 patients) and to strategize targeting their cell migration. Overall, we described a consistent physics-based framework for parameterizing individual glioblastoma patients, connecting to clinical transcriptomic data, and predicting drug effects, that can potentially be used to develop subtype and patient-specific anti-migratory therapeutic strategies.

Abstract Glioblastoma (GBM) is aggressive form of brain cancer that is associated with severe peripheral immunosuppression. This immunosuppression is a critical barrier to patient survival and the success of immunotherapies. Immunocompetent mice harboring intracranial gliomas present with the hallmark features of peripheral immunosuppression observed in GBM patients, including T cell lymphopenia. Using parabiosis, we determined that T cell lymphopenia can be transferred from glioma-bearing to healthy mice via circulation This led to the identification of a high molecular weight immunosuppressive factor in sera of glioma-bearing mice that inhibits T cell proliferation ex vivo. Proteomics analysis identified 119 proteins uniquely upregulated in the sera of glioma-bearing mice compared to controls. These proteins are associated with pathways involved in DNA response elements, cell death, and DNA-histone complexes which imply DNA release. Additional analysis revealed serum isolated from glioma-bearing mice harbor significantly higher concentrations of double stranded cell free DNA (cfDNA) compared to controls. Reducing DNA content using DNase partially reverses the proliferation defects observed in T cells when exposed to serum of glioma-bearing mice. This is while, addition of cfDNA purified from serum of glioma-bearing mice, but not genomic DNA, potently inhibited T cell function. These results were AIM2 independent implicating a novel pathway of cfDNA sensing in T cells. We contend that cfDNA induces immunosuppressive effects in experimental GBM. Devising strategies to reduce circulating cfDNA is therefore a therapeutic approach that could improve outcomes in GBM patients.

Abstract BACKGROUND Treatment of high-grade glioma (HGG) remains a challenge. Several groups have developed ex vivo drug sensitivity assays to evaluate susceptibility of the tumor tissue to specific chemotherapeutics. This tool, in addition to molecular analysis of tumors might provide another layer of confidence in choosing treatments for patients with high-grade gliomas. METHODS 3D PredictTM platform (Kiyatec, Greenville, SC, USA) was tested at three academic medical centers in the United States. Twelve patients with newly diagnosed and recurrent HGGs were included. Tumor spheroids created ex vivo were exposed to twelve pre-selected drugs. Responses were reported as: no response, moderate response and response. We analyzed: demographics, histopathology, stage of disease, success rate and results of the ex vivo assay and clinical implications. RESULTS Majority of patients were women (67%). Median age was 58.5. Histology: GBM-83%, HGGs-17%. Eighty three percent of the samples represented recurrent disease. Sixty percent represented first recurrence while the remaining 40% represented 2nd or 3rd recurrence. Eight out of 12 tissue samples (67%) were successfully tested ex vivo. In six cases (75%), the information was used for immediate clinical decision making. In 2 cases, the decision was deferred until the next progression of the disease. The drugs chosen for treatment based on assay results included: olaparib (the assay was positive for response to rucaparib), temozolomide, lomustine, carboplatin and irinotecan. CONCLUSIONS This platform is easy to use in a standard clinical setting, the success rate for creating viable spheroids for drug sensitivity testing is high and the information obtained can be useful in clinical practice. Addition of other agents to the assay, correlation with molecular testing of the tumor tissue and attempt to assess response to drug combinations should be considered in further development of this technology. Prospective clinical trial is needed to validate this approach for routine clinical use.

Abstract Glioblastoma (GBM) tumors contain a subpopulation of glioma stem cells (GSC) which can escape therapeutic effects through self-renewal and invasion. Epidermal growth factor receptor (EGFR) mutations are the most common alteration in GBM. EGFRvIII (deletion of exons 2-7) is observed in GBM (20-25%) resulting in a constitutively activated EGFR receptor. In addition, extracellular domain missense point mutations, (e.g. G598V, A289V, R108K) are observed in GBM (14.4%). These missense mutations in the EGFR extracellular domain also induces a highly active receptor by enhancing ligand binding. Earlier reports have shown that these EGFR missense point mutations have exhibited therapeutic resistance; however, the molecular mechanism of these missense point mutations in GBM remains unknown. Previous studies in our lab showed that EGFRvIII promotes GSC maintenance and therapeutic resistance via oligodendrocyte transcription factor 2 (OLIG2) signaling. OLIG2 is universally expressed in glioma and essential for reprogramming GBM cells into stem-like tumor cells. Increased OLIG2 expression promotes invasion, growth, proliferation, and survival. Here we report an increase in OLIG2 protein expression in GSCs harboring EGFR-G598V or EGFR-A289V mutations, suggesting these mutations may support stemness properties in GBM tumors. GSC expressing these point mutations display increased self-renewal, growth, and survival. In addition, EGFR-G598V and EGFR-A289V trigger distinct downstream signaling pathways in GSCs to regulate OLIG2 function. Hence, we hypothesize that GBM tumors harboring EGFR missense mutations (G598V, A289V) influence stem cell properties via multiple pathways to evade therapeutic effects and inhibiting these different signaling pathways will decrease therapeutic resistance in GBM. Understanding the mechanism of these EGFR mutations may lead to improved therapeutic strategies for GBM patients.

Abstract Methylation profiling of brain tumors has revealed novel clusters whose biological and clinical significance is unknown. These include three entities termed by the 12.5 version of the Heidelberg methylation classifier as adult-type diffuse high grade glioma, IDH-wildtype, subtype B (HGG_B), subtype E (HGG_E), and subtype F (HGG_F). We have identified 75 cases belonging to these groups: HGG_B (25 cases, 32% female), HGG_E (40 cases, 42% female), and HGG_F (18 cases, 33% female). Although originally characterized as “adult-type,” we find that 70% of HGG_E cases are pediatric, with a median age at diagnosis of 11 years. HGG_B cases are predominantly seen in young adults, and HGG_F in older adults. The histology varies among methylation classes, with HGG_B cases demonstrating high-grade, predominantly astrocytic, morphology, while HGG_E cases demonstrate high-grade glial histology with distinct nuclear features, and HGG_F cases show highly infiltrative but lower grade glial morphology. Patient outcome data obtained to date shows a collective median survival exceeding 5 years, suggesting re-consideration of the proposed grading for these tumors. MGMT promoter methylation differs across the HGG_B (0%), HGG_E (12%), and HGG_F (0%) subtypes. Mutational analyses show enrichment of TERT promoter mutations in the HGG_F group, while HGG_E samples are enriched for tumors with POLE mutations and elevated tumor mutation burden. HGG_E tumors show epigenetic similarity to pediatric high-grade diffuse gliomas (pedHGG_RTK1A); we therefore propose that HGG_E be considered a pediatric-type HGG. HGG_B tumors show demographic and genomic similarities to high-grade astrocytomas with piloid features, including a subset of cases with CDKN2A/2B loss and ATRX mutation. HGG_F separate into two groups dictated by brain location, and despite the presence of TERT promoter mutations, appear distinct from GBMs. Overall, our work represents an initial effort to understand the morphologic, genomic and clinical characteristics of these previously uncharacterized subtypes of gliomas.

Abstract INTRODUCTION Extracellular Vesicles (EVs) are lipid bilayer particles present in almost all biofluids and carry tumor-specific molecular cargo. Plasma EVs are promising sources of biomarker discovery for glioblastoma (GBM). The clinical relevance of non-neoplastic plasma EVs in GBM remains unclear. In this study, we determine if associations exist between plasma EV subpopulation frequencies in GBM and clinical factors. METHODS Platelet-free plasma samples of 63 GBM, 43 Grade 2 glioma, and 45 healthy donors were included. All glioma plasma samples were collected before surgery. Plasma EV subpopulation frequency based on surface expression of EV-associated tetraspanins (CD9/CD81/CD63) and markers of cell origin (CD45/CD41a/CD11b/CD31) was determined by spectral flow cytometry. Clinical information including age, surgery, MGMT methylation status, corticosteroid usage, tumor volume, and overall survival data was recorded. RESULTS Compared to normal donors, GBM patients had increased CD9+, CD63+, CD81+, and CD11b+ EVs but decreased CD41a+ EVs. GBM patients also had the highest frequency of myeloid-derived CD9+CD11b+ EVs (20.17%) followed by grade 2 gliomas (13.78%) and normal donors (2.76%, p< 0.001). More GBM patients (29/63, 46.0%) received preoperative corticosteroids than grade 2 glioma patients (9/43, 20.9%; p = 0.008). Corticosteroid use was not associated with differences in EV subpopulation frequencies in GBM or grade 2 glioma patients. Pearson correlation analysis showed weak but statistically significant correlations between GBM total tumor volume and CD9+ EVs (p = 0.014), CD41a+ EVs (p = 0.037), and CD9+/CD63-/CD81- EVs (p = 0.007). In GBM patients, younger age (< 60 yrs, p = 0.01), gross total resection (p = 0.02), MGMT methylation (p = 0.002), high CD9+ EV expression (p = 0.011), low CD9+CD63-CD81- EV expression (p = 0.009) and low CD9+CD11b+ EV expression (p = 0.006) correlated with longer overall survival. CONCLUSION Differentially expressed non-neoplastic plasma EV subpopulations have significant clinical relevance in GBM including association with tumor volume and predicting survival. Validation of those markers in larger cohorts could reveal new prognostic biomarkers for GBM.

Abstract H3 K27-altered diffuse midline gliomas (DMG) are uniformly lethal tumors for which no therapy has been found to be effective. The blood-brain barrier (BBB) restricts uptake and accumulation of systemically administered drugs in the brain. Convection-enhanced delivery (CED) circumvents the BBB by directly infusing therapeutics into the tumor, facilitating homogenous distribution of drugs at high local concentrations with minimal systemic toxicity. However, most drugs given by CED are rapidly cleared from the brain, severely limiting its therapeutic potential. Here, we determine the impact of drug characteristics rather than mechanism of action on compatibility with CED. Various medicinal chemistry prediction models were interrogated to define drug categories predicated on inherent physicochemical properties of the drugs. A panel of FDA-approved drugs with biologically validated activity in DMG were selected for each group: panobinostat/ponatinib, everolimus/topotecan, tolnaftate/tretinoin, and carmofur/eltrombopag. In vivo pharmacokinetic studies of these drugs delivered by CED into the pons of tumor-naïve mice showed that drugs expected to be less likely to undergo active efflux persisted at the site of infusion over prolonged periods of time. Conversely, efflux transporter (ET) substrate drugs were quickly excreted from the brain and readily detectable in the plasma following CED. Based on the observed brain and plasma concentration-time profiles, the brain exit rate constant and clearance out of the brain were much higher for ET substrate drugs compared with ET non-substrates, resulting in a significant increase in the residence time (the time spent by drugs in the brain) in the absence of ET-mediated clearance. These properties were further validated in efficacy studies in DMG animal models. The proclivity of many drugs to be rapidly effluxed from the brain after CED generates a vulnerability that can be exploited for the development of novel therapeutic regimens, such as with pharmacokinetics-informed drug selection or concomitant dosing of ET-inhibiting drugs.

Abstract PURPOSE Previous analysis of NRG-CC001 suggested pre-treatment WMI volume was a significant imaging-biomarker predictor of post-treatment neurocognitive decline at 4-and 12-months following HA-WBRT+Memantine. This suggested patients with greater pre-treatment WMI were more susceptible to neurocognitive decline, specifically when undergoing HA-WBRT, but not standard-WBRT. The current analysis examined the relationship between changes in MR-determined WMI and NCF 6-months following WBRT+memantine+/-HA. METHODS NCF-testing was performed at baseline, 2, 4, 6, and 12 months post-WBRT, and included Hopkins Verbal Learning Test–Revised (HVLT-R), Trail Making Test (TMT) Parts A and B, and Controlled Oral Word Association (COWA). Pre-treatment WMI was measured by FLAIR-volume corrected for whole brain volume and corrected for the FLAIR volume associated with metastases (met vol) (FLAIR-volume/(whole brain volume–met-vol). Pearson correlation coefficients were used to assess the correlation between pre-treatment WMI and change from baseline to 6-months for each standardized NCF score. RESULTS Of the 518 randomized patients, 443 (217 patients WBRT+Memantine and 226 HA-WBRT+Memantine) had WMI data available at baseline, and of those, 162 patients (89 patients WBRT+Memantine and 73 HA-WBRT+Memantine) had both baseline and 6-month imaging available. For these patients, there was only a trend toward significance for change in corrected FLAIR-volume from baseline to 6-months (p = 0.174) but no significant change in met-vol (p = 0.438) or whole brain volume (p = 0.977). The change in standardized TMT Part A and B and the change in composite 6-month change scores were moderately correlated with change in corrected FLAIR volume adjusted for met-vol but only in the WBRT+Memantine arm (rho=-0.35, -0.34, -0.39 and p-value = 0.003, 0.006, 0.001). CONCLUSIONS Despite an increase in post-WBRT MR-WMI at 6-months, analysis did not show strong correlation between NCF and MR-WMI as determined by FLAIR-volume change between baseline and 6-months. Similar analyses are underway for the NRG-CC003 imaging data.

Abstract There have been several attempts to predict the MGMT promoter methylation status of gliomas using deep learning (DL), with mixed results. These studies did not account for IDH mutation status and did not utilize a uniform methylation assay, introducing potential confounders in the experimental design. We assessed the performance of DL applied to brain MRI to predict the MGMT promoter methylation status of IDH-wildtype gliomas tested with a standardized assay. We included 539 IDH-wildtype gliomas tested with real-time methylation-specific PCR from 2017-2022; 205 (38%) were MGMT promoter methylated. The assay targets 8 downstream MGMT CpG sites using ACTB as control; methylation status is determined based on the difference of PCR crossing-point values: methylated if difference < 5, unmethylated if difference > 7, indeterminate otherwise. We used post-contrast T1 and T2 MRIs as inputs for our models trained using stratified 5-fold cross-validation, 400 epochs, batch size of 16, a learning rate scheduler starting at 10-3, weighted binary cross-entropy loss, and AdamW optimizer. We repeated the experiments using two 3D DL architectures (DenseNet121 and EfficientNetB0) and two image preprocessing pipelines: the BraTS protocol and one that uses padding instead of registration to an MRI atlas, keeping constant other experimental variables. For Densenet121, the best cross-validation AUROCC and F1-score (median, median absolute deviation) were 0.67 (0.06), 0.59 (0.02) using the BraTS protocol, and 0.67 (0.06), 0.58 (0.01) using padding. For EfficientNetB0, were 0.67 (0.02), 0.66 (0.01) using the BraTS protocol and 0.67 (0.01), 0.68 (0.01) using padding. Our findings suggest that it is unlikely that DL models trained on brain MRI can predict the MGMT promoter methylation of IDH-wildtype gliomas. Importantly, our study is the first to focus on an IDH-wildtype adult diffuse glioma cohort tested with a single, standardized MGMT methylation assay, eliminating the variability introduced by different assays and IDH status.

Abstract Incomplete MRI exams reduce the possibility of applying AI models. We previously trained two generative adversarial networks (GANs) on 135 subjects from the BraTS-2017 dataset to generate T1 from post-contrast T1 and FLAIR from T2 sequences (baseline models). To test the impact of the training set size, here, we trained the same models using 1251 subjects from the BraTS-2021 dataset. We used the same architecture and hyperparameters from the baseline models: batch size of 1, learning rate 2x10-4, Adam optimizer, cross-entropy and L1-loss. We trained two versions of the updated models: one stopped at an early checkpoint (early models) and one after 50 epochs (late models). We tested all models on institutional MRIs from 487 IDH-wt gliomas. Synthetic MRIs were compared to the originals using the structural-similarity-index (SSI) and MSE. We simulated scenarios where either the T1, FLAIR, or both were missing and used their synthetic version as inputs of a segmentation model. We compared the segmentations using the dice score (DSC) and used Friedman and Dunn’s test to compare the scores and correct for multiple comparisons. Median SSI for the T1 were 0.957, 0.947, 0.947, and median MSE were 0.006, 0.014, 0.008 for the baseline, early, and late models. For the FLAIR, median SSI were 0.924, 0.908, 0.915, and median MSE were 0.016, 0.016, 0.019 for the baseline, early, and late models. The DSC ranges for the baseline and updated models were (0.655 - 0.953), (0.420 - 0.952), and (0.610 - 0.952). The baseline and late models showed no statistically significant differences in DSC, but both performed significantly better than the early models. We show that GANs trained to synthesize brain MRIs on a small cohort perform similarly to those trained on an x10-larger cohort, making them a viable option for rare diseases or institutions with limited resources.