Introduction
Molecular diagnosis of diffuse gliomas is conventionally performed using surgically obtained tumor tissues; however, the 2016 central nervous system (CNS) World Health Organization (WHO) classification defined a novel tumor [diffuse midline glioma,
H3 K27M mutation (DMG,
H3 K27M-mut)], diagnosis of which was confirmed by molecular findings [
1]. A potential diagnostic issue is that biopsies need to originate from deep tissues, such as the brainstem, and surgery to obtain tissue samples for molecular diagnosis represents a high-risk intervention. Accordingly, the development of less invasive methods for detecting tumor-specific mutations is essential.
Recent progress in genetic analysis technologies such as digital PCR systems allows precise detection of genetic alterations from small samples [
2,
3] and has promoted molecular-diagnostics applications that analyze bodily fluids (liquid biopsy), specifically in various cancers, as a less invasive method for diagnosis and monitoring [
4,
5]. Recent reports revealed the presence of cell-free tumor (ct)DNA in the cerebrospinal fluid (CSF) obtained from patients with CNS tumors [
6‐
8], and ctDNA is also used as a target for liquid biopsy of diffuse glioma [
9‐
11]. This promising approach promotes the utilization of molecular findings as diagnostic markers of diffuse gliomas for monitoring disease state and ultimately the development of non-operative diagnostic procedures. In the present study, we established a novel assay that applies liquid biopsy targeting the ctDNA in CSF using chip-based digital PCR (dPCR) for detecting the glioma-specific diagnostic driver mutations such as
IDH1 R132H (
IDH1),
H3F3A K27M (
H3 K27M), and
TERT promoter (p
TERT) mutations. Additionally, we evaluated the feasibility of this assay using CSF samples from patients with diffuse glioma.
Methods
Study population and sample collection
Intracranial CSF samples were collected from adult and adolescent patients between July 2017 and November 2019 during surgical resection of intracranial tumors under a preoperative diagnosis of diffuse glioma. We collected 3–10 mL of CSF from the operative field immediately after opening the dura under general anesthesia to prevent contamination with DNA released from the resected tumor during surgical manipulation. We also preoperatively collected 1–6 mL of CSF by lumbar puncture from enrolled adult and adolescent patients who underwent tumor resection of the intracranial tumor under preoperative diagnosis of diffuse glioma without highly elevated intracranial pressure between March 2019 and February 2020. We excluded non-glioma patients after confirmation of pathological diagnosis.
We centrifuged CSF samples at 5000
g for 10 min within 1 h of collection to remove residual cells and stored them at − 80 °C as described with slight modification [
12]. DNA was extracted from 1 to 3 mL CSF using the QIAamp circulating nucleic acid kit (Qiagen, Hilden, Germany) according to manufacturer instructions and eluted in a final volume of 60 µL. Quantification of cell-free DNA (cfDNA) was determined using the Qubit 2.0 fluorimeter with the Qubit dsDNA HS assay kit (Thermo Fisher Scientific, Foster City, CA, USA). Extracted cfDNA was stored at 4 °C and analyzed within 1 week.
CSF samples were obtained and analyzed in accordance with the Declaration of Helsinki, with the approval of the Ethics Committee of our institute, and with written consent from patients.
dPCR
Genotyping of cfDNA and tumor DNA was performed using the QuantStudio® 3D Digital PCR System (Life Technologies, Carlsbad, CA, USA). The DNA template, QuantStudio® master mix, and the assay containing primer/probe were mixed according to the manufacturer protocol to obtain a dPCR reaction.
The TaqMan dPCR liquid biopsy assay (Life Technologies) was used for pTERT C228T and pTERT C250T analyses, and a custom TaqMan SNP genotyping assay was used for IDH1 and H3 K27M mutation analyses. The primer and probe sequence data for each assay are described in the Supplemental Digital Content (see Table 1 in Supplemental Digital Content). The final 14.5-µL dPCR mixture was loaded onto a QuantStudio® 3D Digital PCR chip version2 and subjected to PCR amplification using the QuantStudio® 3D GeneAmp PCR system 9700 (see Table 2 in Supplemental Digital Content). Thermal cycling was performed according to the manufacturer’s instructions (see Table 3 in Supplemental Digital Content). Following the reaction, data were analyzed using QuantStudio® 3D Analysis Suite (v.3.1).
Conventional genotyping of tumor-tissue DNA
We isolated and purified DNA from snap-frozen (− 80 °C) intraoperative tumor samples obtained from all patients whose CSF was analyzed, using QIAmp DNA mini kits (Qiagen). We confirmed the presence of
IDH1, p
TERT C228T or C250T, and
H3 K27M point mutations by high-resolution melt analysis and subsequent Sanger sequencing, as described previously [
13,
14]. Loss of heterozygosity (LOH) on chromosome 1p and 19q was confirmed by a PCR-based LOH assay using microsatellite markers, as described previously [
15].
Gel electrophoresis
Tris–acetate-EDTA (pH 8.0) was used as electrophoresis buffer, and a 1.5% agarose gel was used for the electrophoresis of a sample prepared by mixing 5 µL of DNA template with 6X loading buffer (TAKARA, Shiga, Japan). Electrophoresis was performed using a Mupid R-exu system (Mupid Co., Ltd., Tokyo, Japan) at 100 V for 30 min, followed by staining with EtBr solution and detection and imaging using a Limited-STAGE-UV system (AMZ Systems Science, Osaka, Japan).
Statistical analysis
Student’s t test was used to compare the DNA concentrations at different collection sites. Associations between ctDNA positivity and patient characteristics were assessed using non-parametric tests, including the Wilcoxon rank-sum test (continuous variables) or Fisher’s exact test (categorical variables). All statistical tests were two-sided, with an α value of ≤ 0.05 used to determine statistical significance. Statistical analysis was performed using JMP software (v.15.0; SAS Institute Japan, Tokyo, Japan).
Discussion
In this study, we describe the establishment of a novel method for liquid biopsy targeting the ctDNA in CSF using chip-based dPCR to detect glioma-specific diagnostic driver mutations. The 2016 CNS WHO classification and subsequent cIMPACT-NOW update incorporated various glioma-specific genetic alterations in order to integrate molecular diagnoses [
1,
20,
21,
23,
24]. Among these alterations, the 2016 CNS WHO classification regards the
IDH mutation and the 1p19q co-deletion as essential markers for a molecular diagnosis of diffuse glioma, and the
H3 K27M mutation as an essential marker of pediatric glioma [
1]. Additionally, p
TERT mutation is recognized as a novel essential diagnostic marker based on its presence in GBMs and oligodendrogliomas [
22]. Recent studies revealed p
TERT mutations as the most prevalent molecular markers detected in
IDH-WT GBMs and predictive of aggressive bioactivity of diffuse astrocytomas [
25]. cIMPACT-NOW (update 3) reached a consensus on designating
IDH-WT diffuse or anaplastic astrocytomas with p
TERT mutation as WHO grade IV [
22,
26]. In
IDH-mutant oligodendroglial tumors, the status of 1p19q and p
TERT are strongly correlated, with these markers interchangeable when combined with
IDH status [
20‐
22]. Although DMG,
H3 K27M-mut predominantly occurs in children and adolescents, high-grade thalamic gliomas from young adults frequently harbor
H3 F3A-K27M [
27]. The present study of six patients with
H3 K27M-mut included three adult-onset cases. Moreover, the frequency of elderly patients with DMG,
H3 K27M-mut is notable [
28,
29]. We recently described an elderly patient with the DMG,
H3 K27M-mut that mimicked a hemispheric malignant glioma [
30]. Taken together, the
H3 status of an adult hemispheric diffuse glioma affecting the midline should be confirmed even in elderly patients. It is conceivable that the majority of diffuse gliomas can be molecularly diagnosed by confirming
IDH,
H3 K27M, and p
TERT status. In the present study, we selected these genes based on this interpretation and also their status as hotspot point mutations, which enable their detection by dPCR via design of a single TaqMan probe per mutation.
Although other liquid biopsy methods for glioma diagnosis exist, sensitivity is essential for selection of analytical methods for future clinical applications. The chip-based dPCR method presented here yielded highly sensitive results using lumbar puncture CSF, especially for diagnosis high-grade diffuse gliomas. Pentsova et al. [
31] collected preoperative CSF by lumbar puncture from 53 patients with suspected brain tumors and detected ctDNA in 6 (50%) of 12 of them with diffuse glioma using next generation sequencing (NGS). Miller et al. [
32] also used NGS to analyze CSF samples from patients with diffuse gliomas when a clinical event occurred after completing adjuvant therapy, and identified at least one glioma-related genetic alteration in 42 (49.2%) of 85 tumors. Although the target genes and backgrounds differed from those in the present study, our results demonstrated a higher sensitivity in the dPCR-based method than NGS-based method.
Currently, commercially available dPCR systems are theoretically divided into droplet digital (dd) PCR and chip-based dPCR. The sensitivity of ddPCR exceeds that of NGS [
33]. Martinez et al. [
10] tested the capability of ddPCR using cfDNA samples from 20 cases of diffuse gliomas and successfully detected ctDNAs, such as p
TERT,
IDH, and
H3 K27M, in 15 (88%) of 17 patients using CSF collected from lumbar puncture. This indicated that ddPCR showed comparable or higher sensitivity relative to the dPCR assay in the present study; however, Martinez et al. [
10] used their method exclusively to validate an already confirmed mutation. Because we performed all four genotyping assays using a limited amount of CSF from each patient, we were forced to analyze further diluted DNAs from the originally extracted samples. Therefore, we speculate that the sensitivity of our results might have been reduced due to the study design, and that the sensitivity could have been improved by focusing on a target gene. Moreover, chip-based dPCR offers a higher degree of cost-effectiveness and simpler operation relative to ddPCR as a practical method [
3,
19]. Given the aim of liquid biopsy as allowing molecular diagnosis in the absence of surgery, high specificity is required for clinical applications. These results showed no false-positive determinations of mutations using cfDNAs relative to tumor-tissue DNA, indicating the extremely high specificity of the method. Accordingly, these findings suggest the efficacy of this assay for future development of a liquid biopsy acceptable for daily clinical use.
One of the major limitations of the present study is that the reliability of our method seemed to depend on the extent of CSF involvement in tumor distribution. Therefore, our method might be most appropriate for diagnosing advanced high-grade glioma involving CSF. However, we should take into consideration that lumbar puncture is not always indicated for patients with advanced high-grade glioma, because of risk of herniation by space-occupying lesions. These issues are associated with another limitation of the present study that most of our samples were obtained by craniotomy, and lumbar CSF was collected from a limited number of our patients. We detected diagnostic mutations in 28 of 34 patients with glioma using intracranial CSF and the analyzed genes in the remaining six gliomas did not harbor any diagnostic mutations even in tumor tissues. This issue is associated with another limitation that gliomas cannot always be differentially diagnosed from non-glial tumors through genotyping of diagnostic mutations. The sensitivity of our method is lower for lumbar than for intracranial CSF is another limitation. The sensitivity of our method needs to be improved, even for lumbar CSF, before less-invasive liquid biopsies could be routinely used for diagnostic purposes.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.