PATH-45. INTRAEPIGLIOM - RAPID SEQUENCING-BASED DIAGNOSIS OF BRAIN TUMORS. (14th November 2022)

• Record Type: Journal Article
• Authors: Kubelt, Carolin; Brändl, Björn; van Bömmel, Alena; Rohrandt, Christian; Wang, Gaojianyong; Evers, Maximilian; Kolkenbrock, Stephan; Wong, Derek; Danso, Dominik; Maicher, Andre; Friedrichsen, Sönke; Proescholdt, Martin; Riemenschneider, Markus; Jetzek, Ulrich; Meissner, Alexander; Yip, Stephen; Kretzmer, Helene; Synowitz, Michael; Müller, Franz-Josef
• Is Part Of: Neuro-oncology. Volume 24(2022)Supplement 7
• Page Start: vii160
• Page End: vii161
• Publication Date: 2022-11-14
• Subjects: Brain Neoplasms -- Periodicals; Brain -- Tumors -- Periodicals; Brain -- Cancer -- Periodicals; Nervous system -- Cancer -- Periodicals; 616.99481
• DOI: 10.1093/neuonc/noac209.618
• Languages: English
• ISSNs: 1522-8517

Abstract

Abstract: The intraoperative diagnosis of brain tumors remains a clinical challenge despite recent technological advances. The current clinical practice differentiates non-surgical brain tumors from those preferably treated with cytoreductive surgery employing intraoperative frozen section diagnostics. A detailed molecular diagnosis required for this classification task within the timeframe of a routine neurosurgical procedure is currently unavailable. We have analyzed a clinical cohort of several brain tumor entities using Nanopore long-read sequencing on two Oxford Nanopore Technologies sequencing platforms (MinION, PromethION). Since currently available molecular cancer classifiers such as the DKFZ methylation profiling classifier cannot be readily adapted to real-time sequencing analysis, we implemented a novel algorithm (MethyLYZR) to predict the underlying cancer type. Publicly available Illumina Infinium array data were used to train the classifier to distinguish 91 brain tumor classes. For validation of classification accuracy, we conducted a comprehensive validation strategy. Both nanopore platforms could sequence more than 5, 000 pre-selected CpG within less than 20 minutes for most of our samples. When combining an optimized library preparation protocol with the time used for sequencing the minimal number of CpGs needed for classification, we saw sample-to-answer times of less than 1 hour – in many cases within 45 minutes - from receiving a fresh biopsy to a robust cancer type prediction. Comparing actual and predicted diagnoses resulted in a favorable error rate, indicating potentially highly clinical validity. Our real-time based molecular diagnostic algorithm enables, in most cases, a reliable diagnostic call within the timeframe of a typical neuro-oncological surgery. MethyLYZR as a predictive tool may allow us to adjust the surgical strategy and deliver the prognosis to our patients right after surgery, thus allowing for as-of-yet unexplored opportunities for the intraoperative application of individualized therapeutic modalities.