Sensitive Detection and Discrimination of Intracranial Tumors Using Plasma Cell-free DNA Methylomes. (16th November 2020)

• Record Type: Journal Article
• Authors: Nassiri, Farshad; Chakravarthy, Ankur; Feng, Shengrui; Shen, Shu Yi; Nejad, Romina; Zuccato, Jeffrey A; Voisin, Mathew; Horbinski, Craig; Aldape, Kenneth; de Carvalho, Daniel; Zadeh, Gelareh
• Is Part Of: Neurosurgery. Volume 67(2010)Supplement 1
• Publication Date: 2020-11-16
• Subjects: Nervous system -- Surgery -- Periodicals; 617.48005
• DOI: 10.1093/neuros/nyaa447_818
• Languages: English
• ISSNs: 0148-396X

Abstract

Abstract: INTRODUCTION: A major challenge in management of intracranial pathologies is accurate diagnosis of lesions identified on imaging that have a broad differential diagnosis ranging from indolent tumors to aggressive cancers. Current practice necessitates invasive surgery to obtain tissue for diagnosis and molecular subtyping introducing risks associated with surgery. Reliable non-invasive strategies to diagnosis and subtype tumours according to molecular alterations would be transformative for patient care by avoiding the need for invasive procedures or by providing diagnostic information pre-operatively that can facilitate neurosurgical planning. METHODS: We generated plasma methylomes of 608 patients with cancer (n = 389 systemic cancers and n = 219 intracranial tumors) and healthy controls using a cell-free methylated DNA immunoprecipitation technique. Matched tumor methylation profiling was performed using the EPIC array. We harnessed machine-learning approaches and split our cohort into 50 training and testing sets to generate random forest classifiers that could discriminate intracranial tumors from extracranial cancers, as well as classifiers that could discriminate common extracranial tumors and intraaxial tumors. Model performance was evaluated on held-out samples by computing the area under the receiver operating characteristic curve (AUC) in each iteration of training and testing. RESULTS: We observed high sensitivity for our models to classify gliomas among other cancerous and healthy patients (AUC = 0.99, 95%CI 0.96-1) with similar performance in IDH-mutant (AUC = 0.992, 95%CI 0.949-1) and wildtype (AUC = 0.993, 95%CI 0.966-1) gliomas as well as in lower- (AUC = 0.999, 95%CI 0.990-1) and high-grade (AUC = 0.983, 95%CI 0.903-1) gliomas. Moreover, we observed reliable discrimination of common extra-axial tumors (meningioma AUC = 0.89, 95%CI 0.80-0.97; hemangiopericytoma AUC = 0.95, 95%CI 0.73-1) as well as intra-axial tumors ranging from low-grade indolent glial-neuronal tumors (AUC 0.93, 95%CI 0.80-1) to diffuse intra-axial gliomas with distinct molecular composition (IDH-mutant glioma AUC = 0.82, 95%CI 0.66-0.98; IDHwildtype AUC = 0.71, 95%CI 0.53 - 0.9). We observed that plasma methylation correlated well with corresponding tumor tissue DNA methylation values overall (r = 0.42;p<2.2e-16), suggesting that the signatures we identified to detect tumors were primarily derived from tumor DNA. CONCLUSION: Plasma methylomes can reliably detect and distinguish clinically relevant tumors of the central nervous-system.