Detection of Malignant Brain Tumors by NGS Panel of Circulating MicroRNA In Plasma

Early detection and progression of glioblastomas(GBM) and melanoma brain metastasis(MBM) remains a significant clinical problem in detection and monitoring. GBM is the most common primary brain tumor whereby cutaneous melanoma is the most frequent metastasis of all non brain origin tumors. Cell-free nucleic acids(cfNA) are a potential biomarkers of brain tumor detections. CfDNA mutations or amplifications have limited frequency in these tumors. Our approach was to assess a panel of cf-microRNA(cfmiR) which have a defined signatures for both tumor types. Plasma cfmiR are detectable and have advantage over cfDNA; frequency, longer half-life, unique tumor signatures, and low degradation in drawn blood. The limitation of cfmiR has been limited single/few cfmiR that are accurate for brain tumors, PCR specificity of cfmiR assays, as well as isolation techniques are inefficient. To address these issues we utilized an assay that does not require isolation, assesses 2100 miR by NGS, artificial intelligence(AI) modeling of defining a cfmiR panel that is specific to GBM and MBM. We procured plasma(100ul) from pre-surgery GBM(n= 91) and MBM(n=23) patients identified by MRI/pathology. Normal donor blood(n=57 females, 48 males) were assessed. The blood was processed in a direct NGS library preparation(EdgeSeq nuclease probe based) using automated workflow followed by MiSeq analysis. A cfmiR pattern was determined for detection and distribution in normal donors compared to GBM and MBM patients using a t-SNE clustering of the top 20 cfmiR. Using PCA distribution we developed an AI model to identify significant separation of normal donors from GBM or MBM patients, and MBM patients from GBM patients. MiR and cfmiR correlated in respective paired tumors and plasma(p<0.001). AI analysis demonstrated the accuracy of a cfmiR clustering of brain tumor patients. We are now validating and developing a learning model in a larger sample size of brain tumor patients to improve accuracy.