Optimizing acquisition of cerebrospinal fluid liquid biopsy for next-generation sequencing in pediatric central...

Best practices for obtaining adequate cell-free DNA (cfDNA) for next-generation sequencing (NGS) are not currently well understood. This study aimed to improve the understanding of appropriate quantities of CSF extraction from an array of tumor types and locations that could be used for diagnosis of pediatric CNS tumors.

CSF was collected from patients in an IRB-approved prospective study to evaluate the collection methodology for downstream NGS. Samples were collected from control (nontumor) patients and patients with CNS tumors at various times: at dura opening during surgery, from the ventricular catheter during CSF diversion procedures (external ventricular drain or shunt surgery), or during ventricular shunt tapping. Clinical, demographic, and pathological data were gathered, as well as quantitative parameters describing cfDNA yield. A predictive model was generated.

CSF was obtained from 69 CNS tumor patients and 20 controls. The method of CSF collection included dural opening (n = 56), ventricular catheter (n = 21), and shunt tap (n = 12). The mean volume of CSF extracted for CNS tumors was 4.66 ± 3.2 ml. CNS tumors had significantly greater quantities of cfDNA obtained compared with controls (p < 0.001) with a mean cfDNA quantity of 318.4 ± 1340.7 ng (12.0 ± 14.4 ng in controls). CNS tumors had significantly greater cfDNA concentration in the CSF (mean 84.9 ± 340.9 ng/ml vs 5.7 ± 8.1 ng/ml in controls, p = 0.004). The cfDNA yield and concentration were comparable between the dural opening and ventricular catheter cohorts. CSF samples collected from optic pathway (p = 0.04), supratentorial (p = 0.006), and infratentorial (p = 0.004) tumors had greater quantities of cfDNA, with brainstem tumors demonstrating significantly lower cfDNA. Optic pathway tumors (p = 0.04) and supratentorial tumors (p = 0.01) had the greatest concentration of cfDNA. A regression model with 69 real values was used to calculate an additional 100 predicted values, suggesting that 80%, 85%, and 90% success rates for NGS assay require collection of 0.70 ml, 1.26 ml, 3.35 ml of CSF, respectively (p = 0.3).

CSF can be obtained via several methods for NGS analysis, with dural opening and ventricular catheter specimens providing comparable results. The authors show that higher grade tumors, as well as those closest to the CSF space/ventricular compartment, provide increased quantity of cfDNA. They estimate that acquisition of 3.35 ml of CSF has a 90.00% rate of allowing successful processing for NGS.