Exploring CSF MicroRNA Signatures as Diagnostic Biomarkers in Adult-type Diffuse Gliomas

Abstract

Mutations in isocitrate dehydrogenase (IDH) genes (i.e., IDH1 or IDH2) occur in diffuse gliomas (DG), resulting in distinct molecular subtypes (i.e., IDH-wildtype and IDH-mutant). Abnormal expression of extracellular vesicle-derived miRNAs (EV-miRNAs) in the cerebrospinal fluid (CSF) of DG patients could serve as minimally invasive diagnostic and prognostic biomarkers. We investigated differentially expressed microRNAs (DE-miRNAs) in the CSF of patients with DG using miRNA-sequencing (miRNA-seq) and quantitative real-time PCR (qRT-PCR). Our qRT-PCR results indicate that EV-miR-21-5p can differentiate CSF in glioblastoma (GBM) vs. control subjects (p = 0.012, area under the curve [AUC] = 0.84), as well as in IDH-mutant vs. controls (p = 0.003, AUC = 0.93). Additionally, Ingenuity Pathway Analysis revealed that the largest interaction network in GBM is found in miR-16-5p and other miRNAs with seed AGCAGCA. Five miRNAs (miR-150-5p, miR-142-3p, miR-19b-3p, miR-99a-5p, and miR-27b-3p) allowed accurate distinction between IDH-wildtype and IDH-mutant gliomas under multivariable logistic regression (MLR) (AUC = 1.00). Furthermore, a decision tree-based approach revealed that two miRNAs, miR-150-5p and miR-34c-5p, can accurately distinguish CSF from patients with GBM vs. IDH-mutant gliomas. Additionally, MLR reliably differentiates CSF from patients with GBM vs. control (AUC = 1.00), with GBM CSF expressing significantly lower levels of nine miRNAs (miR-1298-5p, miR-1911-5p, miR-195-5p, miR-196a-5p, miR-26a-5p, miR-26b-5p, miR-30a-3p, miR-30a-5p, and miR-30e-5p) compared to controls. For the cut-off, the decision tree predicts diagnostic information based on four miRNAs of miR-1911-5p , miR-100-5p , miR-101-3p , and miR-483-5p . Furthermore, miR-142-3p can accurately distinguish CSF from patients with IDH-mutant gliomas vs. controls (AUC = 1.00). In summary, our study has identified a panel of miRNAs with the ability to accurately discriminate CSF in patients with IDH-mutant gliomas, GBM, or no tumor. This panel can facilitate minimally invasive characterization of CSF samples from patients with DG.