GLIOresolve: TRAINING THE NEXT-GENERATION OF EUROPEAN GLIOBLASTOMA (TRANSLATIONAL) RESEARCHERS, TO RESOLVE PRECISION TARGETING OF THE BRAIN TUMOUR MICROENVIRONMENT

Our group focuses on the identification of urgently needed new treatment options and effective precision medicine therapies.

Glioblastoma (GBM) is the most frequent, aggressive and lethal brain tumour. It has a universally fatal prognosis with 85% of patients dying within two years. New treatment options and precision medicine therapies are required. This must be achieved by multi-sectoral industry-academia collaborations in newly emerging, innovative research disciplines. Glioresolve is an EU Horizon funded Doctoral Networks project which follows on from the recently completed Gliotrain. The Glioresolve consortium brings together leading European academics, clinicians, private sector and not-for-profit partners, and incorporates disruptive research methods including multiomics, ex-vivo ‘tumour-on-a-chip’ assay development, computational modelling and systems biology. Overall Glioresolve provides a comprehensive translational research strategy that goes significantly beyond the current state-of-the-art in neuro-oncology, to establish a new tumour micro-environment (TME)-targeting precision medicine platform for GBM.

ADVANCING A NOVEL THERAPEUTIC STRATEGY FOR A NEWLY IDENTIFIED SUBTYPE (TME-LOW) OF GLIOBLASTOMA

In addition to tumour cells, the glioblastoma (GBM) tumour microenvironment (TME) consists of many non-cancer cells, which exist in several niches and exhibit varying levels of interaction. We recently established a novel molecular subtyping approach based on TME composition and published our data this year in the leading journal Annals of Oncology. In summary, tumours were classified as TMELow (25%), TMEMedium(Med) (38%), or TMEHigh (37%). Molecular analyses identified mutations and uniquely altered biological pathways across subtypes. Please see our Flyer (https://www.glioresolve.eu/s/GLIORESOLVE-Flyer_FINAL1.pdf) which explains this work in lay language. Following re-analysis of data available from clinical trials, it is apparent that immunotherapies may be effective in TMEHigh patients but less effective in the TMELow and TMEMed patients. Here, we focus on studying novel TMELow subtype-specific treatments. Specifically, we will investigate whether TMELow tumours can be ‘primed’ to become more vulnerable to immunotherapies. We will employ faithful pre-clinical models (representing the TMELow subtype) to establish treatment toxicity and efficacy. Tissue collected during these studies will undergo cytokine spatial proteomic analyses to mechanistically assess novel treatment response patterns and resistance mechanisms. Project outputs will include an extended characterization of the TMELow GBM subtype, and will provide new approaches to sensitise TMELow patients to immunotherapeutic interventions. Our overall objective is to generate sufficient data to support initiation of a Phase 2 trial at Beaumont hospital.

INTERROGATION OF NOVEL GLIOBLASTOMA SUBTYPES TOWARDS AN IMPROVED PRECISION MEDICINE APPROACH FOR BRAIN TUMOUR PATIENTS

Glioblastoma (GBM) is the most frequent and aggressive adult brain tumour. Sadly, 85% of patients die within two years, despite surgery and chemo/radiotherapy. Treatment resistance is related to cell types that make up the tumour (“tumour microenvironment” or “TME”). Specifically, the behaviour of TME cells such as blood vessel and immune cells often determines therapeutic response. We believe that studying the TME will provide information on which drugs would be best for specific patients, an approach known as ‘precision medicine’. To do this, a reliable method of selecting the right patients for the drug of choice is required. In a previous project GLIOTRAIN (https://cordis.europa.eu/project/id/766069), we analysed GBM tumours  and used bioinformatics to group the tumours into three “subtypes” based on TME composition. In this project, using tumour material from GBM patients we will investigate the underlying genetic differences between these TME subtypes. We will do this at the ‘tissue’ and ‘single-cell’ level. Next, we will merge all of the datasets to try and uncover TME subtype specific vulnerabilities and resistance mechanisms. Finally, we will develop preclinical models with tumours representing each subtype and will use these models to study ‘precision medicine’ treatment strategies. Specifically, we will choose drugs that should work in each subtype, based on ‘targets’ expressed by specific tumours. Importantly, we will test immunotherapy drugs, which have to date not been successful in GBM clinical trials. Overall, our goal is to develop a new TME-based precision medicine approach, which will provide new treatments for GBM, which is a difficult-to-treat and largely fatal cancer.  Following successful completion of this study and in collaboration with clinical colleagues at The National Centre for Neurosurgery, Beaumont Hospital Dublin, we aim to initiate Ireland’s first Phase 2 trial in GBM patients.

Development of improved preclinical models

Our group also focuses on the development of improved, clinically relevant, orthoxenograft models of GBM.

Better and more accurate GBM models are required, particularly those mimicking the aggressive mesenchymal GBM subtype. The NFpp10-GBM cell line has been shown to closely mirror the heterogenous and aggressive growth pattern seen in the mesenchymal subtype (Marumoto et al., 2009 ) (Allen et  al 2017). To date this model has not been fully characterized nor widely employed in pre-clinical studies. Instead, the GL261 syngeneic model has been employed. This model fails to accurately predict patient outcomes (eg Reardon  et al 2016). Indeed, promising preclinical results using the GL261 model failed to predict outcome in a recent (negative) ICI clinical trial (NCT 02017717). Moreover, as discussed above, GBM occurs more commonly in older patients (>60 years). It is known that age negatively correlates with patient outcome, with older patients often responding worse to therapy (Perry et  al 2017). Moreover, younger GBM patients frequently harbour isocitrate dehydrogenase (IDH) mutations, which creates a more favourable tumor phenotype (Kleihues et al 2000 ). Physiological differences which exist between animals of different ages must therefore be considered

We have established the novel NFPp10a syngeneic model of mesenchymal GBM which incorporates surgical resection (Sweeney  et al 2014).  and displays significant chemo-resistance to TMZ and ICI monotherapy. We have moreover established the effect of these interventions in both young (6-8 week) and aged (18month old) mice. Indeed, our observations resonate closely with what is seen in the clinic. This model will now allow us to study novel neoadjuvant and adjuvant therapeutic approaches in a clinically relevant manner Importantly these new models will more faithfully recapitulate the clinical disease and can be used to study novel treatment strategies in a clinically relevant way.  Post-mortem analyses are implemented to interrogate ICI mediated treatment effects on the TME of mesenchymal GBM via multiplex immunohistochemistry and scRNA sequencing.

Establishment of radiomic pipelines to interrogate imaging data sets

Our research group implements preclinical imaging modalities including bioluminescence, Contrast enhanced CT and MRI imaging to interrogate response to therapy

Despite magnetic resonance imaging (MRI) being the gold standard imaging modality in this disease setting, the availability of preclinical MRI is limited. Indeed, bioluminescence imaging (BLI) represents the default modality in the pre-clinic, notwithstanding its lack of clinical relevance. Nevertheless, many models including patient derived xenografts (PDX), are not amenable to BLI due to the absence of luciferase expression. CT is an alternative clinically relevant imaging modality employed in the initial brain diagnosis of brain tumours and which is more-widely available (than MRI) in the pre-clinical setting.

In an effort to assess the role of CT in GBM orthoxenograft modelling, we have optimised CT protocols on two instruments, (IVIS-SPECTRUM-CT;TRIUMPH-CT), with and without  co-delivery of iodine based contrast agents. Furthermore, we are employing radiomic analyses of extracted CT data to generate radiomic signatures, which will provide for deeper analyses of tumour features including texture, density and cellularity. This work will uncover disease characteristics that are not otherwise easily detected or detectable by the naked eye. These Radiomic analyses study both agnostic tumour features such as skewedness, kurtosis and entropy, as well as semantic features including shape and vascularity.

Our radiomic work will define features which we hope will expand the preclinical utility of CT imaging in the brain tumour setting. This approach will further support the identification of prognostic radiomic signatures which may be subsequently assessed in clinical CT datasets from GBM patients.