JS09.6.A ADVANCING THE MINIMALLY INVASIVE CNS-TUMOR CLASSIFICATION: A COMPARATIVE STUDY OF LIQUID BIOPSY APPROACHES USING NANOPORE SEQUENCING AND PANEL-BASED NGS OF CEREBROSPINAL FLUID CTDNA

Abstract BACKGROUND Cerebrospinal fluid (CSF) liquid biopsy (LB) enables minimally invasive molecular profiling of central nervous system (CNS) tumors. We compared targeted next-generation sequencing (NGS) and whole-genome Oxford Nanopore sequencing (ONT) regarding diagnostic yield, turnaround time (TAT), and clinical utility. MATERIAL AND METHODS CSF cell-free DNA (cfDNA) from 60 patients with suspected CNS tumors was analyzed across two centers. In all cases, both NGS and ONT were performed on the same cfDNA aliquot. Matched tumor tissue was available in 91.7 % (55/60). Diagnoses included glioblastoma (GBM, n = 26), CNS lymphoma (CNSL, n = 18), astrocytoma (A, n = 5), ependymoma (EPN, n = 3), diffuse midline glioma (DMG, n = 3), and other gliomas (n = 5). NGS profiles (CNVs, SNVs, indels) were reviewed by a multidisciplinary LB board. ONT enabled CNV and methylation-based classification using Sturgeon and Rapid-CNS2. RESULTS NGS detected ctDNA in 83.3 % (50/60); among ctDNA-positive LBs, 72.0 % (36/50) received a full molecular diagnosis (A-diagnosis), and 28.0 % (14/50) a partial diagnosis (B-diagnosis). ONT was interpretable in 81.7 % (49/60), with methylation-based classification in 60.0 % (36/60) and CNV-only profiles in 21.7 % (13/60). Singleplex ONT succeeded in 90.0 % (9/10); multiplexing (n = 50) yielded valid methylation profiles in 52.0 % (26/50), depending on total cfDNA input: 64.3 % (18/28) for >2 ng vs. 36.4 % (8/22) for <2 ng. Median TAT was 4.2 days for ONT vs. 21.3 days for NGS. In 20.0 % (12/60), ONT refined or corrected NGS-based classifications—primarily in CNSL (n = 10) and GBM without a 7/10 signature (n = 2). NGS outperformed ONT in DMG (n = 3) due to detection of H3F3A mutations. CONCLUSION ONT enables rapid, clinically meaningful CSF-based classification of CNS tumors, especially in GBM and CNSL. This distinction is of high clinical and therapeutic relevance and represents a frequent diagnostic challenge in neuro-oncological practice. While performance decreases in low-input multiplexed samples, the speed and diagnostic impact of ONT supports its use in time-sensitive workflows. Singleplex ONT is preferred for clinical LB studies with limited cfDNA input.