PLASMA PROTEOMICS AND MACHINE LEARNING FOR NOVEL BIOMARKER DISCOVERY IN ADULT MALIGNANT GLIOMA

Abstract AIMS Plasma is a valuable source for identifying non-invasive biomarkers, and when combined with an examination of the highly dynamic proteome, it has the potential to lead to the identification of novel biomarkers in glioma. The aim of this study was to uncover plasma-based protein biomarkers for adult malignant glioma. METHOD Mass spectrometry-based proteomics with tandem mass tag (TMT) labelling of plasma from healthy individuals and adult malignant gliomas was performed. A differential abundance analysis was carried out to identify proteins that were deregulated in primary gliomas compared to healthy individuals. Machine learning was employed to identify a diagnostic biomarker panel. RESULTS When comparing plasma from healthy individuals to that of primary gliomas, several high and low abundant proteins were found to be deregulated. To improve the accuracy and ability of biomarkers to detect malignant gliomas, machine learning was employed and led to a development of a classifier, which performed with high accuracy, specificity, and sensitivity. CONCLUSIONS The discovery of a plasma-based protein classifier, once validated, may facilitate an earlier diagnosis of glioma patients, and thereby reduce time-to-treatment.