KS02.5.A RAPID AND OPTIMIZED INTRAOPERATIVE CLASSIFICATION OF CNS TUMORS BY NANOPORE SEQUENCING AND EPIGENETIC PROFILING

Abstract BACKGROUND Intraoperative profiling of CNS tumors remains challenging due to limitations in current diagnostic strategies. While intraoperative sequencing combined with epigenetic classification demonstrates high accuracy, existing workflows are time-consuming and labor-intensive. Here, we developed a high-speed intraoperative sequencing protocol that optimizes quality control (QC) steps without compromising data integrity and accelerates library preparation using transposase adapters. MATERIAL AND METHODS Our optimized workflow includes rapid DNA extraction using intraoperative sample size optimization, and real-time tissue lysis (chemical and mechanical), followed by DNA isolation and quantification using a Qubit fluorometer. Libraries were prepared using a rapid sequencing protocol with transposase adapters, replacing traditional ligation-based methods. Real-time data analysis incorporated AI-assisted base-calling and modification detection, which combined with tumor classification using a neural network-based algorithm trained for methylation profiling. A total of 144 patients were sequenced, including 8 runs using the rapid-UKER protocol and 15 runs with the ligation-based protocol. RESULTS The mean duration of the rapid-UKER protocol was 50.54 minutes (n=144 patients), with DNA isolation and quantification accounting for two-thirds of the time. Library preparation and sequencer loading were completed within 10 minutes in 87.5% of cases, compared to a mean of 28.5 minutes for the ligation-based method. Both protocols demonstrated comparable performance, achieving 92% correct tumor detection within 5 minutes of sequencing and accurate copy number variation (CNV) profiling within approximately 25 minutes. No significant differences in sequencing accuracy were observed between transposase and ligation-based adapters. CONCLUSION The rapid-UKER protocol enables accurate CNS tumor classification within 1 hour, significantly reducing library preparation time without compromising sequencing quality. Transposase adapters represent a robust and efficient alternative to traditional ligation-based workflows for intraoperative tumor profiling.