WG1. Non-invasive Biomarker Discovery

1) Liquid biomarker A broad range of circulating biomarkers — ctDNA, CTCs, EVs, ctRNA, miRNA, circRNA, ecDNA, and Tumour-Educated Platelet RNA — will be explored for early primary brain cancer detection with a focus on blood mRNA, miRNA, proteins (e.g., unfold), circRNA, ecDNA and DNAm biomarkers. Advanced techniques such as high-throughput proteomics (MS) and NGS will be applied for both gliomas and meningiomas. For instance, one lab will conduct circRNA-enriched RNAseq on blood samples of 100 patients with GBM, 100 patients with low-grade gliomas, 200 patients with meningiomas, and 100 gender/age-matched control samples of patients with no known neoplasms. Top biomarker candidates will be evaluated by in silico analysis and validated by RT-qPCR, ddPCR, or nanostring in cohorts from other labs. In addition, differential scanning fluorimetry (DSF) will be applied to measure protein unfolding in plasma for biomarker discovery. Studies will also be carried out for prognostic and recurrent biomarker study, e.g., comparing blood samples from patients before and after treatments; patients without and with recurrent cancers within a certain period (e.g., 5-10 years).

2) Radiomics biomarker Radiomics analysis has had remarkable progress along with advances in AI/ML and medical imaging to aid physicians by offering second opinions and providing nformation about the region or tumour of interest. It has the potential to enable quantitative measurement of intraand intertumoral heterogeneity through intensity, shape, size, volume or texture features, and therefore offers great potential to accelerate precision medicine46,47. As an alternative non-invasive approach, radiomics and deep radiomics data (with transfer learning techniques) including MRI and SRH quantitative image features, offer information on tumour phenotype and microenvironment. These image descriptors will be explored to build multiple statistical and AI/ML models for detection, as well as prediction of the pathological grade and subtype, survival probability and recurrence possibility, particularly for meningiomas in the former. Radiogenomics studies will also be performed to identify correlations between the different image properties (radiomics and dynamic parameters) and genomic expressions, such as specific genes (e.g., NF1, TP53), or a tumour genomic profile. Image data and relevant clinical information have already been collected by Action members as well as in public resources, e.g., ivgGAP, TCIA, and TCGA which will be explored for radiomics biomarker study. In addition, this Action would be a good opportunity to explore the generation of larger datasets by the synergy of the broad network of members. Moreover, the European Cancer Imaging Initiative will be set up to develop an ‘Atlas’ of cancer-related images and make anonymised images accessible, which will e an additional valuable resource for this Action. It is noted that the combination of these approaches and biomarkers will have the potential for the creation of a panel of brain cancer biomarkers with diagnostic, prognostic, and predictive values in clinical practice. For example, robust and reproducible liquid-biopsy assays can be developed based on those biomarkers identified by high throughput approaches and then be applied as point-of-care (PoC) testing/screening. In addition, performing large-scale radiomics analyses may enable us to determine patterns of imaging detection. However, any positive early diagnosis result requires confirmatory diagnostic evaluation by medically established procedures (MRI & biopsy) to confirm cancer.