Cerebrospinal Fluid Liquid Biopsy for Molecular Monitoring of Paediatric Brain Tumours: A Pilot Feasibility Study

Introduction Central nervous system (CNS) tumours are the most common solid malignancy of childhood and the leading cause of cancer-related death in this age group. Diagnosis and monitoring still rely on tissue biopsy and neuroimaging, but biopsy is invasive, captures only a single timepoint and a single region and carries substantial procedural risk, particularly for midline and brainstem lesions. Cerebrospinal fluid (CSF)–based liquid biopsy represents a minimally invasive alternative: being in direct contact with the CNS, CSF is enriched in tumour-derived cell-free DNA (cfDNA) relative to plasma. Yet standardized protocols are still lacking and most available evidence derives from retrospective, single-timepoint cohorts, with scarce longitudinal and histotype-specific data. Research objectives This pilot study aimed to validate the technical feasibility of cfDNA extraction and quantification from paediatric CSF samples (primary), to characterise cfDNA dynamics longitudinally across three clinical timepoints (secondary), to assess the feasibility of PCR amplification and Sanger sequencing for tumour-specific marker detection (tertiary) and to explore the effect of CSF input volume on cfDNA yield (quaternary). Materials and Methods In this analytical in vitro study, 41 residual CSF samples from 34 patients (aged 0–25 years; 11 histotypes), collected between September 2021 and May 2026, were analysed at three timepoints: diagnosis (T0, n=25), end of therapy (T1, n=14) and suspected relapse (T2, n=2). cfDNA was extracted with the QIAamp Circulating Nucleic Acid Kit, quantified by Qubit dsDNA HS assay (positivity criterion) and qualitatively assessed by TapeStation capillary electrophoresis. PCR and Sanger sequencing (H3F3A, H3F3B) were performed on one DMG sample, and four paired high-volume extractions were tested. Statistics were descriptive (median/IQR, Wilson score confidence intervals), with one pre-specified Fisher's exact test (T0 vs T1). Results The overall cfDNA detection rate was 29.3% (12/41; median 0.148 ng/µL). Detection was highest at diagnosis (40%, 10/25), fell markedly at end of therapy (7.1%, 1/14) and rose again at relapse (50%, 1/2); the T0-versus-T1 comparison yielded an odds ratio of 8.67 (95% CI 0.94–404; p=0.060). Embryonal tumours showed higher detectability than glial and ependymal entities. PCR and Sanger sequencing succeeded even from low-input cfDNA; H3F3A and H3F3B were wild-type at codon 27, suggesting that the underlying H3K27M alteration resides in a canonical histone gene not sequenced here. Doubling the CSF input did not recover undetectable samples, indicating that negativity reflects a genuine biological paucity of cfDNA rather than a technical extraction limitation. Conclusions CSF-based liquid biopsy is technically feasible in paediatric high-grade CNS tumours and cfDNA detectability appears to track tumour burden across the clinical course. These preliminary findings support its biological plausibility as a minimally invasive, longitudinal complement to imaging and tissue diagnosis, while underscoring the need for prospective, multicentre, histotype-specific validation and field-wide standardisation of pre-analytical and analytical procedures.