Leveraging archival cerebrospinal fluid samples for genetic insights from cell-free DNA

Abstract

Cerebrospinal fluid (CSF) cytology has long held an important role in the care of patients with brain tumors and is routinely used for diagnostic and staging purposes as well as to guide treatment decisions. Although numerous diagnostic studies can be performed on malignant cells obtained from CSF, downstream analysis is often precluded by the absence or scarcity of tumor cells. It is now well established that CSF samples are often a rich source of tumor-derived cell-free DNA (cfDNA).^1-5 Many studies have demonstrated that somatic alterations in tumor-derived cfDNA can be detected in CSF in the majority of patients who have central nervous system (CNS) involvement of primary and metastatic brain tumors.^{2, 3, 6, 7} Despite low cfDNA yields after extraction from CSF, the relative fraction of tumor-derived cfDNA is generally high, enabling mutation detection even at relatively low sequence coverage and in the absence of malignant cells on cytologic evaluation.² Meanwhile, circulating tumor DNA from CNS tumors in plasma is typically undetectable, catapulting CSF cfDNA liquid biopsies into the clinical arena in recent years.^{3, 4, 8} Because of the challenges in obtaining CSF samples, most of this work has been done on found samples, it is often limited to a single tumor type, and the numbers of patients included in each study have been low. This has precluded rigorous analyses of preanalytic variables, which are critical to understanding the optimal conditions for processing of these samples to ensure maximum clinical utility.

In this issue of Cancer Cytopathology, Neil and colleagues describe their experience using a targeted next-generation sequencing assay on archival CSF samples from samples previously received and processed in the cytology laboratory.⁹ In these samples, they demonstrate a high degree of success in detecting tumor mutations in archival CSF samples. Of note, in their experience, an effect of storage time (the median storage time before isolation was 37 days) or volume was not noticeable. Furthermore, no additional stabilizing agents or storage requirements (beyond standard storage at 4°C before extraction) appear to have been required. This is of particular importance for practice settings considering barriers to the implementation of CSF cfDNA testing. In addition, the extended period of time before processing and the dual use of samples for both cytology and CSF cfDNA sequencing have tremendous practical utility considering the paucity of CSF able to be obtained clinically. Although cfDNA testing has been known to be subject to a host of preanalytic factors affecting sample yield, quality, and results^10-12 (and optimization of these factors is not without importance), their findings support the ready integration of testing within existing clinical workflows. This will hopefully encourage more laboratories to validate this specimen type. The findings are particularly timely given the recent publication of best-practice recommendations for validating, reporting, and publishing clinical circulating tumor DNA assays by the Association for Molecular Pathology.¹³

Because molecular diagnostics has now become an essential part of CNS tumor classification and an increasingly important consideration for optimal treatment selection, cfDNA from CSF offers many advantages over traditional tissue biopsy. Brain tumors are often genomically highly heterogeneous and multifocal; CSF sampling may be performed for surgically inaccessible sites and is more readily obtained at multiple clinically relevant time points,^{8, 14-17} revealing a more accurate representation of tumor heterogeneity and evolution.^{3, 17} CSF cfDNA sequencing can confirm the presence of CNS disease, enable risk stratification, demonstrate resistance mutations, and even diagnose new and unexpected primary tumors.⁵ Its role as a potential prognostic and predictive clinical biomarker remains an area of active study as sequencing cfDNA from CSF becomes increasingly adopted in routine practice and in ongoing and future clinical trials.³

The authors declared no conflicts of interest