The Role of Liquid Biopsies in Pediatric Brain Tumors

Abstract

Early detection and serial therapeutic monitoring for pediatric brain tumors are essential for diagnosis and therapeutic intervention. Currently, neuropathological diagnosis relies on biopsy of tumor tissue and surgical intervention. There is a great clinical need for less invasive methods to molecularly characterize the tumor and allow for more reliable monitoring of patients during treatment and to identify patients that might potentially benefit from targeted therapies, particularly in the setting where diagnostic tissue cannot be safely obtained. In this literature review, we highlight recent studies that describe the use of circulating tumor DNA, circulating tumor cells, circulating RNA and microRNA, and extracellular vesicles as strategies to develop liquid biopsies in pediatric central nervous system tumors. Liquid biomarkers have been demonstrated using plasma, urine, and cerebrospinal fluid. The use of liquid biopsies to help guide diagnosis, determine treatment response, and analyze mechanisms of treatment resistance is foreseeable in the future. Continued efforts to improve signal detection and standardize liquid biopsy procedures are needed for clinical application.

INTRODUCTION

Pediatric brain tumors represent the most common solid tumors in children and are the leading cause of cancer-related mortality in childhood (1). The management of pediatric brain tumors relies on tumor resection for tissue biopsy. However, due to the anatomical location of the tumor, it may often be difficult to obtain tumor tissue for biopsy and oftentimes only a limited amount of tumor tissue can be surgically resected. Surgical biopsy is invasive and carries the risk of surgical complications including sedation risks with anesthesia and risks of bleeding and infection. Radiographic imaging such as MRI is also limited in that it does not provide molecular information about the tumor. Liquid biopsies are minimally invasive and can provide more information about the molecular biology of the tumor and has the potential to aid in diagnosis and disease monitoring. The molecular profiling of the tumor obtained via liquid biopsies can also advance precision medicine and further the development of targeted therapy (2). Biofluids used for liquid biopsies contain tumor cells or circulating tumor molecules and include whole circulating tumor cells (CTCs), circulating tumor DNA (ctDNA), circulating tumor RNA, micro RNA (miRNA), and proteins or peptide fragments (2). Few reviews in the literature describe the utility of liquid biopsies in pediatric brain tumors (2−4). This review highlights and summarizes the current literature on the different liquid biomarkers for pediatric brain tumors, the technical challenges of developing liquid biomarker assays for pediatric subjects, and its future directions.

History of Liquid Biomarkers in Neuro-Oncology

The use of tumor cytology has been used to monitor pediatric brain tumors dating back to the 19th century (4). The process of obtaining cerebrospinal fluid (CSF) samples for cytology has been used to provide qualitative information. This is often used as the standard of care for evaluating pediatric patients with primary central nervous system (CNS) tumors, such as medulloblastoma, as part of staging and to determine disease recurrence as well as detect presence of leptomeningeal metastases (5−7). A significant portion of patients with atypical teratoid and rhabdoid tumors (AT/RTs), which are rare highly malignant embryonal CNS tumors that affect young children, present with dissemination to the CSF (8). CSF cytology in these AT/RT patients is often also examined for assessment of disease recurrence, and tumor cells found in the CSF have been shown to be in clusters and have unique eccentric nuclei, abundant cytoplasm, and prominent nucleoli (8). One case report describes the detection of rhabdoid cells with loss of INI1 reactivity on cytological examination of CSF in a patient with diffuse leptomeningeal AT/RT, illustrating the feasibility of obtaining molecular genetic information from CSF analysis in these types of tumors (9).

CSF cytology, which is routinely collected at diagnosis for newly diagnosed pediatric patients with CNS tumors, also has its limitations. CSF cytology cannot be used to serially monitor patients in a quantitative way to assess for treatment response nor does it allow for molecular analysis and provide information about the tumor phenotype. Despite the high specificity (>95%) of CSF cytology, it has low sensitivity (<50%) (10).

Intracranial tumor serum biomarkers began to develop in the 1960s and 1970s. Excesses in alfa-fetoprotein (AFP) and beta human chorionic gonadotropic (beta-HCG) were found to be produced within germ cell tumor tissues (11, 12). It was also noted that after radiographical evidence of successful response to therapy, levels of AFP and beta-HCG were found to have normalized. Despite the successful use of AFP and beta-HCG biomarkers clinically, the discovery and clinical use of liquid biomarkers for other pediatric CNS tumors are scarce. However, research continues in the attempt to find more reliable liquid biomarkers for other pediatric CNS tumors to improve clinical management and find valuable therapeutic targets.

Circulating Cell-Free Tumor DNA

ctDNA are fragments of DNA shed by the primary tumor tissue and is the fraction of cell-free DNA (cfDNA) that arises from the tumor cells. Cell turnover and apoptotic and necrotic tumor cells contribute to release of ctDNA in biofluids. One advantage of the use of ctDNA as opposed to CTC is that a low amount of biofluid is required to assess tumor burden (2). There have been great advances in the use of ctDNA as a liquid biopsy specifically for pediatric diffuse midline gliomas, which has one of the highest mortality rates among all pediatric brain tumors. Despite the advancement of surgical approaches to biopsy brainstem tumors, the difficulty of performing repeated surgical biopsies is a barrier to the ability to assess clinical response. CSF is more easily obtainable than brain tumor tissue and genetic alterations detected in CSF may serve as an alternative to tissue biopsy. The inability to assess and monitor disease response and treatment related-molecular changes has led to the need to develop sensitive and specific biomarker assays such as ctDNA.

Diffuse midline gliomas, including diffuse intrinsic pontine gliomas (DIPG) are aggressive malignancies that have a median overall survival of 11 months (13). The inaccessible location of these brainstem tumors makes it particularly challenging from a diagnostic and therapeutic perspective. This elevates the clinical need for a less invasive and reliable method to obtain molecular information about the tumor and assess disease status. A study conducted by Huang et al demonstrated the feasibility of H3K27M detection in DNA found in CSF from a cohort of 11 children with diffuse midline gliomas using targeted Sanger sequencing and nested PCR mutation-specific primers. The authors were able to confirm the presence of H3K27M mutation in tumor tissue (14). In a pivotal study conducted by Panditharatna et al, it was found that there was efficacy to using ctDNA for treatment surveillance and to better characterize the biology of the tumor. This study collected whole blood, CSF, cyst fluid, and tumor tissue samples from a subset of 84 enrolled subjects (48 patients with diffuse midline glioma and 36 control patients with non-CNS disease). Authors were able to quantify ctDNA using droplet digital PCR (ddPCR) for H3K27M gliomas and identify the mutation in both CSF and in plasma in 88% of patients with diffuse midline gliomas (15). It was also noted that there was a significant decrease in H3K27M plasma ctDNA after radiotherapy in 83% of the patients (15). There was also a significantly higher amount of ctDNA noted in the CSF compared with plasma, which the authors stated was most likely attributed to the location of the biofluids in relation to the tumor site and the presence of the blood-brain barrier (15). In 4 subjects, there was a lack of detection of ctDNA at initial biopsy, but following radiotherapy, ctDNA was subsequently detected in 2 of the 4 subjects. This suggests that radiotherapy can potentially disrupt the blood-brain barrier, increase cell turnover and the quantity of apoptotic and necrotic cells, and lead to the release of ctDNA into the biofluid (15). The authors of this study concluded that both plasma and CSF were suitable for sampling and detection of ctDNA and were promising sources for disease monitoring and to assess for treatment response.

Circulating Tumor Cells

Despite the promising data, ctDNA has distinct limitations. ctDNA can originate from necrotic or apoptotic tumor cells, making it difficult to discern if the ctDNA is released from cancer cells due to cell death after therapy or if the presence of ctDNA is due to resistance to therapy. Furthermore, ctDNA represents only fragments of the tumor DNA and detection requires the presence of specific mutations associated with the disease. CTCs are intact cancer cells released into circulation from the primary tumor. CTCs provide a rich source of molecular information due to complete genome and allow for characterization of the tumor phenotype using DNA, RNA, and protein analyses. CTC analysis can potentially provide more insight on the clonal evolution of the tumor and its mechanisms of tumor metastasis. CTCs are exceedingly rare (1 cell in 10⁹ blood cells) and therefore are challenging to isolate and capture (2). In adult tumors, CTCs have been captured mainly by targeting extracellularly membrane proteins such as EPCAM using positive selection. For nonepithelial cancers such as brain gliomas, it has been difficult to identify tumor-specific antibodies. Microfluidic platforms have been developed to capture CTCs using negative selection such as removal of leukocytes using magnetic beads or erythrocyte cell lysis, but only recently has this been done successfully on adult glioblastoma (GBM) malignancies. In a study by Sullivan et al, CTCs were identified in 13 of 33 GBM patients (39%) (16). Despite the blood-brain barrier, this represents a proof of concept that CTCs can be identified in peripheral blood in brain tumor patients. The frequency of release of CTCs in blood circulating from brain tumors has not been established and thus the utility of using it as a biomarker throughout treatment remains to be determined. CTCs have not yet been widely studied in pediatric brain tumors. A recent abstract describes a pilot study using vimentin as a surface marker for CTC detection in an automated method to capture CTCs. Cells that stained for Vimentin and CD45 were isolated using an automated approach. Of the 9 patients that had adequate samples, 7 patients had CTCs detected from peripheral blood (17).

Circulating miRNA

MicroRNA (miRNA) are small RNA molecules that regulate mRNA and have been shown to be involved in pediatric brain tumor biology. miRNA play key roles in regulating cellular proliferation, differentiation, and apoptosis. miRNA can inhibit translation and promote mRNA degradation and are known regulators of gene expression. It has been proposed that profiling miRNA expression facilitates disease detection. Under conditions of stress or during disease states, the function of miRNA is more pronounced.

In pediatric neuro-oncology, studies on the use of serum miRNA as a liquid biomarker are beginning to emerge. The first study used to examine the utility of serum miRNA in the detection and screening of pediatric brain tumors was done by Bookland et al (18). Authors identified characteristic miRNA (miR-21, miR15b, miR-23a, and miR-146b) in pediatric juvenile pilocytic astrocytoma patients. The elevations in levels of miRNA was able to predict tumor nodular size and response to therapy with a sensitivity of 86% and specificity of 100%.

Another study also analyzed circulating miRNA in blood serum in pediatric astrocytomas. (19). RNA was extracted from the tumor tissue and extracted from blood serum from patients with pediatric astrocytomas. Expression of miRNA was determined by quantitative PCR. miR-130a was noted to be upregulated in all samples regardless of tumor grade. miRNA 145 and 335 were noted to be downregulated in pediatric astrocytoma patients. Expression of proteoglycans (SDC4) and that of its biosynthetic enzymes (EXT1) and XYLT1 were also noted to be altered in pediatric astrocytoma patients. These results demonstrate a role of miRNAs in the biology of pediatric astrocytomas and the potential use of miRNA as a biomarker (19).

Recent discoveries have also been made on the role of miRNAs in pediatric embryonal brain tumors (20−22). The use of microRNA as biomarkers have been studied on the detection and management of pediatric brain tumors such as medulloblastoma and atypical teratoid/rhabdoid tumors (2).

Proteomic Biomarkers

Proteins and peptides have emerged as promising biomarkers as their expression correlates with tumor pathology. The discovery of proteomic biomarkers has been widely pursued, but the clinical utility of protein biomarkers in pediatric neuro-oncology needs continued investigation. There have been several CSF protein biomarkers reported in the literature including the detection and use of apolipoprotein, insulin-like growth factor binding protein, and prostaglandin (23, 24).

Due to the low abundancy of proteins and the presence of highly abundant proteins masking less abundant ones, CSF proteomic biomarker identification remains technically challenging. One study tried to overcome these challenges by processing CSF through nanoparticles in 27 children with brain tumors and 13 control patients (cohort of patients with non-Hodgkin’s lymphoma without CNS involvement) (25). This study was able to identify proteins secreted into the CSF that could be related to metastatic disease status and can distinguish metastatic cases from controls.

Additionally, one of the first studies to profile proteins in CSF using mass spectrometry from DIPG patients noted that upregulation of tumor proteins including cyclophyllin A (CypA) and dimethylarginase (DDAH1) (26). With CSF proteomic analysis, the upregulation of these proteins demonstrated that these proteins may be involved in glioma formation in the brainstem and its detection may have certain clinical applications.

Osteopontin has been shown to be overexpressed in AT/RT. A study with a cohort of 39 patients showed with an enzyme-linked immunosorbent assay, that the mean osteopontin levels in plasma and in CSF in AT/RT was higher than in medulloblastoma, hydrocephalus, and epilepsy patients (27). Serum osteopontin levels were found to be correlated with therapeutic response and higher levels have been associated with worse prognosis (27). Five patients with ATRT had a decrease in osteopontin levels after treatment, but then also had an increase in levels with relapse.

In a recent study, prostaglandin D2 synthase (PDG2S), a glycoprotein commonly found in CSF was shown to be significantly reduced 6-fold in 33 medulloblastoma patients compared with 25 age-matched controls (28). This was thought to be due to host response due to presence of the tumor (28). The authors used 2D electrophoresis and mass spectrometry. CSF levels of apolipoprotein E and J were also noted to be elevated in medulloblastoma patients relative to controls.

Insulin-like growth factor has also been recently investigated in pediatric biomarker research. One study analyzed protein expression levels of insulin-like growth factor and insulin-like growth factor binding protein in CSF of 16 medulloblastoma, 4 ependymoma, and 23 controls with radioimmunoassay (29). It was noted that insulin-like growth factor binding-protein 3 was significantly higher in CSF in medulloblastoma patients compared with the ependymoma patients and controls after correcting for total CSF protein count. The authors concluded that the insulin-like growth factor system seemed to play an important role in neuronal development and increased levels may potentially be used to monitor disease.

Extracellular Vesicles

Extracellular vesicles such as exomes and microvesicles contain tumor-specific DNA, RNA, and protein and is potentially promising as a liquid tumor biomarker (30). The function of the vesicles includes cellular communication and transfer of nucleic acids, proteins, and liquids. They have been shown to be detectable in numerous biofluids including plasma, serum, CSF, saliva, and urine. It has also been thought that release of extracellular vesicles is upregulated in cancer and promotes tumor progression.

One study characterized exosomes from a medulloblastoma cell line with proteomic analysis and was able to characterize extracellular vesicles in patient serum (31). This study reported a potential role for extracellular vesicles in stimulating proliferation and tumor cell migration and suggested a role for transcription factor haptocyte nuclear factor 4 alpha (HNF4A), which may act as a tumor suppressor. Currently, Jackson et al are working on correlating extracellular vesicles with metastatic medulloblastoma cell lines and showed that metastatic cell lines produced higher quantities of exomes compared with nonmetastatic cell lines, making extracellular vesicles a potential biomarker for medulloblastoma (32).

Clinical Utility of Liquid Biopsy to Monitor Treatment Response

There have been several studies that reported the use of liquid biopsy to monitor tumor response to treatment (3). Stallard et al looked at ctDNA containing mutated H3K27M that were isolated from DIPG cells and the correlation between ctDNA with tumor cell proliferation after treatment with irradiation (33). It was noted that irradiation with 8 Gy resulted in an increase in ctDNA measured 72 − 120 hours postradiation. In the Panditharatna et al study, serial CSF ctDNA measurements were analyzed in 48 subjects with DIPG and it was noted that CSF ctDNA is increased with disease progression (15). After radiation therapy, ctDNA levels were noted to be decreased, which was correlated with a decrease in tumor size noted on MRI in 83% of the subjects. These studies illustrate that liquid biopsy not only allows for molecular characterization of the tumor, but also has the potential for surveillance of treatment response. A summary of the potential clinical applications of circulating biomarkers is shown in (Fig. 1).

FIGURE 1.

Circulating biomarkers and its clinical applications. Analysis of circulating biomarkers as a liquid biopsy in plasma, urine, and cerebrospinal fluid has potential clinical utility to monitor therapeutic response in real time, understand the molecular basis behind tumor heterogeneity and evolution, discover therapeutic targets, and understand treatment resistance mechanisms (image created using Biorender.com).

Current Progress in Liquid Biopsy Technology and Challenges to Pediatric Brain Tumor Biomarker Development

The current workflow for biomarker analysis includes high-throughput technological assays such as droplet digital PCR as well as next-generation sequencing (30, 34). Microfluidic platforms have also been developed to try to capture CTCs, but were only recently done successfully for adult brain tumors such as GBM (16, 34). Mass spectrometry has also been used for protein biomarker discovery and quantification (23). A summary of the advantages and disadvantages of different liquid biopsy methods and a summary of available liquid biopsy techniques are illustrated in Tables 1 and 2, respectively. Despite advancements in technological liquid biopsy assay development, research on developing cost-effective platforms for capturing biofluid tumor markers is still needed in order to provide clinical utility for pediatric patients. A summary of challenges to pediatric brain tumor biomarker development is shown in Table 3.

TABLE 1.

Advantages and Limitations of Liquid Biopsy in Pediatric Brain Tumors

	Advantages	Limitations
Circulating tumor cells (CTCs)	Noninvasive Ability to help characterize the tumor phenotype Ability for broad analysis and molecularly profile RNA, DNA, and protein expression Can use techniques for immunolabeling	Rare and hard to capture cells Heterogeneity in the CTC enrichment and analytical methods Only present in late stage diseases
Cell-tumor DNA (ctDNA)	Noninvasive Technically easier to isolate that CTCs due to higher abundance Detection has previously been shown to be associated with poor prognosis	Unclear if ctDNA is being released from cancer cells dying from therapy of if due to therapeutic resistance DNA fragmentation can be an issue with analysis. Difficult to detect in early stage disease
Proteomic biomarkers	Allow for in-depth analysis of protein expression	Lack of trials to validate reliability
miRNA	Ability to analyze gene fusions and splicing variants	Lack of standardized methods for handling and storage

TABLE 2.

Current Liquid Biopsy Techniques

	ctDNA	Exosomes	CTC
Isolation	Optimized sample prep kits for ctDNA isolation	Physical separation   Immunoaffinity   Microfluidics   Precipitation	CTC enumeration   Physical separation (dielectric sorting)   Microfluidics   Immunoaffinity   Direct visualization
Analytic approaches	NGS, RT-qPCR, ddPCR	Proteomic analysis	NGSRT-qPCR, ddPCR, FISH, CTC imaging, single-cell analysis

ctDNA, circulating tumor DNA; CTC, circulating tumor cell; NGS, next-generation sequencing; RT-qPCR, quantitative reverse transcription polymerase chain reaction; ddPCR, droplet digital polymerase chain reaction; FISH, fluorescence in situ hybridization.

TABLE 3.

Challenges to Pediatric Brain Tumor Biomarker Development

Low incidence rates compared with adult cases result in smaller sample sizes. Multicentered collaborative efforts and centralized repositories are needed. Difficult to draw enough sample volume for analysis in pediatric subjects compared with adults. Selection of age-matched controls are needed for biomarker validation, but control samples are difficult to obtain. Variability in collection method, timing, biofluid type, storage.

Compared with liquid biopsies in adult neuro-oncology, research on liquid biopsies in pediatric neuro-oncology patients is only starting to emerge. A summary of recent selected literature on circulating biomarkers in pediatric brain tumors is shown in Table 4. Pediatric brain tumors are relatively rare compared with adult brain tumors, and large-scale multicenter trials using biomarker detection assays for pediatric patients are needed to generate an adequately powered study to make any meaningful conclusions. Additionally, in adult patients up to 7.5 mL of blood and CSF have been used in biomarker research and discovery, but this volume of biofluid may be difficult to obtain in young pediatric patients (4). This makes it particularly challenging to detect rare tumor biomarkers in circulation and requires techniques to process smaller volumes or requires amplification, which may impact results.

TABLE 4.

Summary of Recent Selected Literature on Circulating Biomarkers in Pediatric Brain Tumors

Biomarker	Sample Source	Methodology	Tumor Type	Findings	Year	References
H3K27M mutant cfDNA	CSF, plasma	Droplet digital PCR	Diffuse midline glioma	H3K27M mutant cfDNA was correlated with response to treatment	2018	(15)
IGFBP-2,3	Blood, CSF	Radioimmune assay	Medulloblastoma   Ependymoma	IGFBP-2,3 levels were increased	2008	(29)
Osteopontin	Plasma, CSF	ELISA	AT/RT	Elevated level in AT/RT patients	2005	(27)
miRNA	CSF, serum	qRT-PCR	Juvenile pilocytic astrocytoma	Serum miRNA profile correlated with tumor volume and reversed with complete tumor resection.	2018, 2017	(18, 19)
c-Tau	CSF	ELISA	Medulloblastoma   Ependymoma   Astrocytoma	Elevated CSF c-tau levels, suggesting axonal damage	2015	(36)
Netrin-1	Urine	ELISA	Medulloblastoma	Increased levels in metastatic medulloblastoma	2014	(37)
VEGF	Serum	ELISA	High grade glioma	Higher plasma VEGF levels may reflect enhanced angiogenesis in the tumors	2017	(38)

Adapted from references (4, 24).

CSF, cerebrospinal fluid; cfDNA, cell-free tumor DNA; IGFBP, insulin growth factor binding protein; VEGF, vascular endothelial growth factor; AT/RT, atypical teratoid rhabdoid tumor.

Lack of sample standardization has been a general weakness in biomarker research (35). Well-designed studies that identify biomarkers using appropriate control cohorts and appropriately correlating results to pathological findings is of great importance. Details of storage, collection, and handling of samples needs to be mentioned. The lack of accuracy and reproducibility leads to failure of biomarkers achieving any clinical utility. There is also a lack of comparative data looking at multiple biofluids. Tumor stage and location can also influence the yield of biomarkers in the biofluid and has been demonstrated in serum and in CSF. Due to the lack of standardization it is also difficult to compare the clinical utility of different tumor biomarkers including ctDNA, CTCs, proteomes, and extracellular vesicles. These biomarkers may all be complementary, but large-scale studies need to be performed to determine which biomarkers are most suitable for molecular profiling and cost-effective detection assays with good sensitivity and specificity need to be developed. To overcome these obstacles, pediatric clinical trial consortia groups could include standardization of liquid biomarkers as part of their design, mimicking the layout and standards of CANCER-ID (https://www.cancer-id.eu/) or a similar initiative.

CONCLUSIONS

Over the past decade, there has been significant increased knowledge regarding the molecular aspects of pediatric brain tumors, largely due to more information obtained from sequencing of tumor tissue. This has led to the classification of pediatric CNS tumors into clinically distinct molecular subgroups. Greater understanding of the genetic landscape of pediatric brain tumors will also create a greater need for advancement of liquid biopsy research as this could provide more molecular information than tumor histology or radiographical imaging. Biopsy of tumor tissue may be difficult to obtain, especially in tumors that are anatomically located in regions that are difficult to access. Liquid biomarkers also make it easier to monitor treatment response and disease progression in serial way. Detection of molecular epigenetic changes identified through liquid biopsy samples will provide more information on the biology of the tumor and help identify targetable mutations. Challenges will continue to arise with liquid biopsy research for pediatric brain tumors, and optimization of isolation methods and analytical tools is needed. However, as the technological developments continue to improve, the successful implementation of liquid biopsies clinically will help improve outcomes for pediatric brain tumor patients.

 

Research is in part supported by the NYU CTSA grant (TL1 TR001447) from the National Center for Advancing Translational Sciences, NIH.

The authors have no duality or conflicts of interest to declare.