LAUSR

Liquid biopsy (LB) by cerebrospinal fluid (CSF) can be useful to identify circulating tumour DNA (ctDNA), thus offering information about the heterogeneity of the neoplastic genome. The aim of our study is to assess the effectiveness of LB of the CSF in detecting ctDNA which mirrors the genetic profile of the tumoural tissue, and to investigate the clinical and radiological aspects influencing the availability of ctDNA. Tumoral tissue and CSF samples of 13 GBM patients undergoing surgery was collected. CSF was withdrawn from the very proximity of the tumoural surface before the excision. DNA extracted from tissue samples was analysed by qPCR to identify typical genetic alterations such as copy number variations (EGFR, PDGFRA, CDK4, MDM2, CDKN2A), and point mutations (TP53, PTEN, IDH, NRAS, PI3K1, pTERT). CtDNA extracted from CSF was analysed by droplet digital PCR to assess the presence of the alterations found in the matching tissue. Both contrast-enhanced (CE) and FLAIR volumes of the lesions were measured in the pre-surgical MRI. Linear and logarithmic regressions were employed for the statistical analysis. From June 2016 to February 2017 we prospectively collected 13 GBM patients. Median age was 73 years. All lesions showed CE at the MRI; other radiological findings included necrosis (84.6%), oedema (76.9%), cortical, ventricular or meningeal involvement (76.9%, 30.8%, and 15.4%). Median volumes of CE and FLAIR lesions were 28.6 and 25.5 cm3, with a median FLAIR/CE ratio of 72.9. Surgery was subtotal (<95%) in all patients. All GBM tissues were tested for the following alterations: EGFR, PDGFRA, CDK4, MDM2, CDKN2A; 76.9% were tested for TP53, PTEN, and IDH mutations; 38.5% for NRAS and pTERT mutations; 30.8% for PI3KR1 mutation. MGMT methylation was assessed in 12 cases (92.3%) and found in 7 (58.3%). Median CSF volume, ctDNA quantity and concentration were 0.45 mL, 59.64 ng, and 0.42 ng/μL. Processable DNA was found in 11 CSF specimens (84.6%), in 8 of which (61.5%) it carried the same alteration expressed by the tumoural cells of the matched tissue, while in 3 cases (23.1%) it seemed to have a different genetic profile; finally, in 2 cases it was not possible to detect any circulating DNA in the CSF. Preliminary data on 13 patients suggest that the ctDNA concentration in the CSF could be related to the FLAIR/CE ratio as measured in the MRI before surgery (p = 0.02). Other correlations between the molecular and the radiological features are still being exploring. Our study confirms that LB of CSF can detect ctDNA carrying the same molecular profile harboured in the tumour. Therefore, it seems to be an accurate method to identify markers useful for the diagnosis and the monitoring of the disease. Additionally, our ongoing study is trying to demonstrate a potential correlation between radiological features of the tumour and availability of ctDNA in CSF.