PCR (polymerase chain reaction) amplifies DNA using a combination of primers, dNTP, and DNA polymerase. Specifically, quantitative PCR captures real-time amplification of targets using fluorescently assigned probes. ddPCR furthermore allows for unbiased amplification of a known target by isolating reactions in each of thousands of droplets (Martínez-Ricarte et al., 2018; Panditharatna et al., 2018). ddPCR is a sensitive and accurate method that can detect rare ctDNA mutations with a limit of detection of approximately .001% in CNS tumors (Hindson et al., 2011; Izquierdo et al., 2021; Li et al., 2021). To demonstrate the utility of this technique, one group used blood ddPCR to amplify ctDNA and found that it was possible to detect the H3.3K27M mutation in 85% of diffuse intrinsic pontine glioma (DIPG) patients at diagnosis and 100% of patients post-radiation or during therapy (Mueller et al., 2019).

ddPCR can only detect known targets via the addition of targeted primers and probes. NGS circumvents this issue with its capacity to seek innumerable unknown genetic alterations (Buermans and den Dunnen, 2014).

NGS (e.g, Illumina) sequences by synthesis, in which a 500-base-pair adaptor is ligated and fluorescent dNTPs are added to the strand to determine the sequence of the target strand (Voelkerding et al., 2009; Buermans and den Dunnen, 2014; Fox et al., 2014). Illumina can be used for whole-genome, whole-exome, or targeted generation sequencing. Results are obtained between 29 h and 4 days (Kanzi et al., 2020). A recent study utilized the NGS Memorial Sloan Kettering-Integrated Molecular Profiling of Actionable Cancer Targets (MSK-IMPACT) assay. Using CSF from pediatric patients with varied histologies and a sequencing depth of 2x to 1,368x, researchers screened patients with existing somatic alterations (Miller et al., 2022).

Nanopore sequencing utilizes miniature pores to sequence strands of DNA (Feng et al., 2015). As a strand of DNA passes through the pore, an electrical current is generated that base-calls nucleotides to sequence the strand. The MinION, a handheld nanopore device, is fast and more affordable than NGS with the ability to reuse parts. The nanopore device can provide same-day results and allows users to make genetic calls within the first few minutes of sequencing (Euskirchen et al., 2017).

Proteomic characterization of CNS tumors is possible via liquid chromatography, which separates compounds in a sample based on interactions between mobile and stationary phases. Mass spectrometry is then utilized to convert molecules to the ionized state allowing compound identification by measuring the speed of movement through a vacuum chamber (Pitt, 2009). With only 10 μL of plasma, 862 proteins were identified within 12 h using LC-MS (Xue et al., 2018). Using CSF LC-MS, six proteins were successfully identified that could discriminate the metastatic status of CNS tumors in children compared to controls (Spreafico et al., 2017).

LC-MS can also be used to analyze the metabolites in CSF created by the intake of anti-cancer drugs to assess penetrance beyond the blood-brain barrier (BBB), important for identifying new therapies for clinical trials. A study of seven oral anti-cancer drugs in pediatric CNS tumor patients successfully demonstrated the ability of LC-MS to identify medications with the highest BBB penetrance (Guntner et al., 2020).

Cell-free DNA (cfDNA) are 50–200 base-pair DNA fragments that are released from cells via apoptosis, necrosis, and budding DNA release (Stroun et al., 2001). Circulating tumor DNA (ctDNA) are mutation-carrying fragments released by tumor cells. Both total cfDNA quantification and the mutations identified in ctDNA can be used as biomarkers for tumor diagnosis and progression. Due to its short half-life of 16 min to 2.5 h cfDNA could be a real-time marker of treatment effect and measurable residual disease (MRD) (Mayo et al., 2022).

Due to the high cell turnover rate in malignancies, elevated levels of overall cfDNA are often observed in CSF (Bagley et al., 2020; Nakamura et al., 2020). A study of brainstem gliomas including DIPG demonstrated the ability to reliably detect primary tumor alterations in CSF ctDNA (Pan et al., 2019). Our group demonstrated correlation between the concentration of characteristic mutant K27M copies in CSF ctDNA and contrast-enhancing tumor area on MRI, as well as with cell proliferation in an in vitro model (Stallard et al., 2018). We further applied this technique in a larger cohort as part of a prospective clinical trial, demonstrating that change in H3.3K27M variant allele fraction (VAF) in both CSF and plasma was associated with increased progression-free survival. VAF increase of >25% was found to precede tumor progression in many cases, and cases of radiographic progression without an increase in VAF were later found to reflect pseudo-progression. These data show that CSF or plasma ctDNA H3.3K27M VAF could be a reliable biomarker for prediction of progression and for differentiation from pseudoprogression (Cantor et al., 2022).

When CSF was collected serially in children with medulloblastoma (MB), low-coverage WGS detected MRD in 54% and 85% of patients with localized and metastatic disease, respectively (Liu et al., 2021). Among 64 CSF samples from a diverse range of pediatric and young adult patients, cfDNA positivity correlated strongly with disseminated disease (Miller et al., 2022). While these data are promising, a large-scale attempt to standardize liquid biopsy for DNA biomarker isolation across multiple tumor histologies using ultra-low pass NGS was fairly unsuccessful with copy number alterations only detected in 20% of CSF samples and fewer in other biofluids (Pagès et al., 2022).

microRNAs (miRNAs) are non-coding molecules 18–24 nucleotides in length (Eibl and Schneemann, 2021). They are integral to the stability and translation of messenger RNA (mRNA) and can impact tumor angiogenesis, growth, and invasiveness (Floyd and Purow, 2014). miRNAs have been isolated from CSF, blood, and urine and are stable in biofluids. However, the standardization of control and tumor-specific miRNA panels is in its infancy.

Considering the difficulty in obtaining CSF specimens from children, liquid biopsy of serum and urine miRNA in pediatric brain tumor patients would be preferred, however, there is a relative paucity of information in this subject. Serum miRNA in pediatric astrocytomas has been found to demonstrate characteristic miRNA signatures; miRNA levels were used to successfully predict tumor size and therapeutic response (López-Aguilar et al., 2017; Bookland et al., 2018). Furthermore, reverse transcription of cell-free tumor RNA into complementary DNA has successfully detected EGFR(v)III amplification in pediatric high-grade gliomas (Figueroa et al., 2017; Manda et al., 2018).

In MB (most common cerebellar tumor in pediatric patients) miRNA expression changes have been identified that directly regulate tumorigenic MB pathways including SHH and WNT (Braoudaki and Lambrou, 2015; Wang et al., 2022). These biomarkers could provide MB genetic classification without surgical biopsy.

Extracellular vesicles (EV) are small, membrane-bound structures secreted from both normal and cancerous cells. EVs have been found to hold tumor-specific DNA, various RNAs, and proteins (Doyle and Wang, 2019). Because of their small size and tumor-specific cargo, EVs have been suggested as a biomarker for liquid biopsy in adult gliomas (Nikoobakht et al., 2022).

The diagnostic potential of EVs in pediatric brain tumors is not well studied. Proteomic signatures isolated from pediatric medulloblastoma cell lines and patient sera revealed both a brain tumor-specific profile as well as a MB-specific profile (Epple et al., 2012). Furthermore, spike-in media-derived exosomes were found to be cell attractants, increasing cell migration in vitro. Isolation of EVs from the cell media supernatant of 3 DIPG and 4 MB cell lines identified EV-derived miRNAs unique to DIPG and MB respectively (Magaña et al., 2022). Additionally, the study identified EV-specific miRNAs when compared to miRNAs isolated from parent tumor cells, suggesting targeted loading of miRNAs into EVs.

Further research may unlock important mechanistic discoveries and therapeutic targets. However, the lack of standard protocols for EV isolation and lack of liquid patient samples are barriers to further discovery.

Genomic profiling of pediatric brain tumors has identified targetable mutations in more than half (Ramkissoon et al., 2017; Cole et al., 2018). In high-risk pediatric glial tumors, exome and transcriptome sequencing prompted a change in therapy in more than two-thirds (Koschmann et al., 2018). This has further been shown to be reproducible using liquid biopsy, with the ability to identify targetable mutations, monitor therapeutic response, and follow tumor evolution (Miller et al., 2022). Molecular profiling is now recommended along with IHC to guide care in pediatric glioma patients (Miklja et al., 2019). Spatial subclones with unique genotypes and phenotypes have been identified in pediatric GBM and DIPG (Vinci et al., 2018; Harpaz et al., 2022) limiting the accuracy of a single stereotaxic biopsy. Liquid biopsy includes elements secreted from all parts of the tumor, with better representation that can provide a more accurate diagnosis.

Liquid biopsy could also facilitate early minimally invasive tumor identification, avoiding treatment delays where appropriate. Of incidentally discovered pediatric brain tumors, most are monitored with imaging and 10% eventually require treatment (Soleman et al., 2020). In an adult diffuse glioma pilot study, 85% of tumors were successfully diagnosed via CSF ctDNA sequencing with a 7–10 days turnaround (Martínez-Ricarte et al., 2018). However, when specimens were prospectively collected in 564 children, copy number alterations were only detected in 20% of CSF, 1.3% of plasma, and 0% of urine samples (Pagès et al., 2022). In the setting of low ctDNA levels in children and low levels of clonal aberrations, extensive further research in the pediatric population is necessary to develop ctDNA as a biomarker and lumbar puncture as an early diagnostic option.

Several pediatric CNS tumors are difficult to diagnose even using tumor morphology, imaging, immunohistochemistry, sequencing, and FISH. Of these diagnostic enigmas, 71% were successfully diagnosed using DNA methylation profiling, a parameter easily accessible using liquid biopsy nucleic acids (Pages et al., 2021). Liquid biopsy miRNA markers for medulloblastoma have been identified for differentiation from controls with high specificity (Wang et al., 2022). In primary CNS lymphoma, the DNA yield, variant allele fraction, and ability to detect MYD88 (a characteristic mutation) were actually much better in CSF cfDNA than in stereotactic biopsy cellular DNA, the current gold standard for diagnosis (Yamagishi et al., 2021).

Intraoperatively, the MinION nanopore device identifies genetic data within the first few minutes of sequencing using fresh frozen tumor tissue (Euskirchen et al., 2017), which could dictate extent-of-resection decision-making. Though Nanopore has previously been limited by a higher error rate as compared to NGS, our group demonstrated 85% sensitivity and 100% specificity for ctDNA in CSF samples from pediatric high-grade glioma patients (Slatko et al., 2018; Bruzek et al., 2020).

During and following treatment of brain tumors, patients are monitored with serial MRI imaging which often fails to differentiate true progression from pseudo-progression. Serum biomarkers have been used to monitor progression in other cancers (i.e. PSA in prostate cancer, CEA in colorectal cancer, CA125 in ovarian cancer, AFP in hepatocellular carcinoma, and beta-hCG in gestational trophoblastic disease). A correlate in pediatric brain cancers could play an important role in post-treatment monitoring and early treatment of recurrence. Serial liquid biopsy in addition to serial MRI may reveal new genetic targets for precision chemotherapy (Kim et al., 2015; Salloum et al., 2017).

Our group demonstrated that change in CSF cfDNA H3K27M variant allele fraction predicted treatment response and tumor progression following ONC201 treatment in diffuse midline glioma (Cantor et al., 2022). The same biomarker was also found to be correlated with treatment response to radiotherapy by another group (Panditharatna et al., 2018). In medulloblastoma, CSF cfDNA was used as a means to detect MRD, demonstrating that patients with persistent copy number alteration detection were at higher risk of progression. Furthermore, marker detection preceded radiographic progression in half of relapsing patients (Liu et al., 2021). DNA methylation status in CSF ctDNA has also been identified as an epigenetic prognostic marker to predict outcomes in MB (Li J et al., 2020).