A specific miRNA signature in the peripheral blood of glioblastoma patients

The prognosis of patients afflicted by glioblastoma remains poor. Biomarkers for the disease would be desirable in order to allow for an early detection of tumor progression or to indicate rapidly growing tumor subtypes requiring more intensive therapy. In this study, we investigated whether a blood-derived specific miRNA fingerprint can be defined in patients with glioblastoma. To this end, miRNA profiles from the blood of 20 patients with glioblastoma and 20 age- and sex-matched healthy controls were compared. Of 1158 tested miRNAs, 52 were significantly deregulated, as assessed by unadjusted Student's t-test at an alpha level of 0.05. Of these, two candidates, miR-128 (up-regulated) and miR-342-3p (down-regulated), remained significant after correcting for multiple testing by Benjamini-Hochberg adjustment with a p-value of 0.025. The altered expression of these two biomarkers was confirmed in a second cohort of glioblastoma patients and healthy controls by real-time PCR and validated for patients who had received neither radio- nor chemotherapy and for patients who had their glioblastomas resected more than 6 months ago. Moreover, using machine learning, a comprehensive miRNA signature was obtained that allowed for the discrimination between blood samples of glioblastoma patients and healthy controls with an accuracy of 81% [95% confidence interval (CI) 78-84%], specificity of 79% (95% CI 75-83%) and sensitivity of 83% (95% CI 71-85%). In summary, our proof-of-concept study demonstrates that blood-derived glioblastoma-associated characteristic miRNA fingerprints may be suitable biomarkers and warrant further exploration.