High detection rate of circulating-tumor DNA from cerebrospinal fluid of children with central nervous system germ cell tumors

Abstract

Central nervous system germ cell tumors (CNS-GCT) are malignant neoplasms that arise predominantly during adolescence and young adulthood. These tumors are typically sensitive to treatment, but resulting long-term health deficits are common. Additional clinical challenges include surgical risks associated with tumor biopsy, and need to determine treatment response for adapting radiotherapy protocols. The aim of this study was to establish the detectability of circulating-tumor DNA (ctDNA) from cerebrospinal fluid (CSF) of children with CNS-GCT as a potential biomarker. We obtained CSF from patients with CNS-GCT by lumbar puncture or intra-operatively. Cell-free DNA (cfDNA) was extracted and subjected to low-pass whole genome sequencing (LP-WGS). Copy-number alterations (CNAs) were inferred and served as a marker of measurable residual disease (MRD). Comparisons with imaging findings and tumor marker levels were made. A total of 29 CSF samples from 21 patients (16 with germinoma, 5 with non-germinomatous GCT) were sequenced. Twenty samples from 19 patients were collected at diagnosis, and 9 samples from 7 patients were collected during or after therapy. Among the diagnostic samples, CNAs were detected in samples from 17/19 patients (89%), which included 8 with marker-negative tumors. Specific clinical scenarios suggested that serial cfDNA analysis may carry utility in tracking treatment responses as well as clarifying indeterminate imaging findings. Our results provide evidence for the high-sensitivity in detecting ctDNA from CSF of CNS-GCT patients using LP-WGS, with potential utility for non-invasive diagnosis and disease monitoring in upcoming CNS-GCT studies.

Keywords: Cell-free DNA; Central nervous system germ cell tumors; Circulating-tumor DNA; Liquid biopsies; Measurable residual disease.

Fig. 1

A Measurable disease positivity from cerebrospinal fluid (CSF) samples collected at diagnosis or initial staging. B Exemplary case (GCT003) depicting tumor-concordant copy-number alterations inferred from low-pass whole-genome sequencing (LP-WGS) of cell-free DNA (cfDNA) extracted from CSF at diagnosis. C cfDNA concentrations in CSF samples collected at diagnosis, during therapy and after therapy, demonstrating a decreasing trend with therapy in our cohort

Fig. 2

A Circulating-tumor DNA (ctDNA) was detected 6 months prior to the borderline elevation of cerebrospinal fluid (CSF) β-hCG in a patient (GCT011) with pituitary stalk thickening and presumed germinoma, offering the potential for early non-invasive diagnostics. B Liquid biopsy finding mirrors clinical course in a patient (GCT009) with bifocal germinoma where serial CSF samples were available. Presence of ctDNA in CSF sample collected after completion of chemotherapy (middle panel) suggested residual active disease in spite of equivocal enhancing signal from resolving ventricular disease on imaging (yellow arrow heads). C Liquid biopsy detected residual disease and clarified ambiguous radiologic findings after 2 cycles of chemotherapy in a patient with relapsed germinoma (GCT014). With the resolution of ventricular disease (yellow arrowhead) and hemorrhaging temporal lobe lesion (asterisk), repeat liquid biopsy on therapy indicated persistence of molecular disease at both copy-number and mutational (KRAS) levels (middle panel). The third CSF collected after completion of chemotherapy, including high-dose chemotherapy, and resection of the left temporal lesion showed no measurable residual disease