Liquid Biopsy in Gliomas- A Review
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.304105
Source of Support: None, Conflict of Interest: None
Keywords: Glioma, liquid biopsy, relevance
Glioblastoma (GBM) is the most common primary malignant brain tumor. In spite of all the advances in medicine the average survival is 14 months, with less than 5% surviving more than 5 years. The treatment of brain tumors present several unique challenges- as it is housed in the skull even a simple needle biopsy requires at least a burr hole and accurate localization with the help of a CT or MRI scan. In certain published series, this carries a mortality of up to 3% and a morbidity of up to 30%. Though, in the absence of the magic bullet, which happens to be the case in the majority, radical surgery and post-operative radiation with concurrent and adjuvant oral Temozolomide chemotherapy remains the management option of choice. In malignant gliomas, it rarely results in cure, but merely a prolongation of survival. In addition, sampling errors caused by tumor heterogeneity lead to diagnostic errors and pseudo-progression following chemoradiation is often difficult to differentiate from actual progression of disease.
Until 2016, the management of GBM was based on the histological architecture of the tumor alone with very few exceptions., But in 2016, CNS WHO and a newly formed body, The Consortium to Inform Molecular and Practical Approaches to CNS Tumor Taxonomy (cIMPACT-Now), defined diagnostic entities based on both histopathological and molecular criteria to more accurately diagnose, prognosticate and treat gliomas. cIMPACT-NOW, which has been mandated to provide guidelines to practicing clinicians, has identified a handful of genes that play an important diagnostic, prognostic and a possible therapeutic role in gliomas. In midline gliomas, histone alterations notably H3K27M also plays an important role in determining the tumor prognosis and novel therapeutic interventions against this mutation being tested currently in clinical trials.[6–9] Other subgroups are also emerging that respond to targeted therapies- the 1p19q oligodendrogliomas have long been known to respond to chemotherapy, Isocitrate dehydrogenase (IDH) mutant gliomas have a 'hypermethylated' subtype that respond better to Temozolomide. As well as exploring IDH inhibitor therapies and tumors exhibiting the BRAFv600e mutation can now be targeted specifically with therapies such as dabrafenib and vemurafenib. Clinical trials using. MEK inhibitors to target BRAF-1549KIAA mutations in pilocytic astrocytomas.
In spite of burgeoning research, the outcome for the majority of gliomas still remains abysmal. This, therefore, is an opportunity to learn from the success that has been achieved in the other cancers- early diagnosis, accurate monitoring of the disease process and an early detection of recurrences. Diagnosis of gliomas today is limited by access to high-end imaging and tertiary care neurosurgery services, monitoring is hampered by the lack of specific and accurate biomarkers and the detection of recurrence by the dilemmas of pseudo-progression. Under these circumstances, liquid biopsy is a term that has become commonplace in lung cancer, is becoming more relevant in glioma cancer diagnostics. Based on the testing of biofluids, it provides for a fairly accurate picture of the malignant potential and possible molecular targets of the tumor. All tumor release within the blood and its adjacent biofluid (CSF in brain tumors and so on) circulating tumor cells [CTCs], cell-free tumor DNA (ctDNA), exosomes, microRNA (miRNA- noncoding RNA), proteins and Tumor Educated Platelets (TEP). The role of liquid biopsy in gliomas is being investigated in multiple studies and the US-FDA has already approved a liquid biopsy test for the diagnosis of lung cancer. The present article reviews the different modalities for liquid biopsy in gliomas and its possible role in the clinic.
Liquid biopsy modalities
Circulating tumor cells (CTCs)
Ever since Ashworth discovered the presence of tumor cells in the circulation of a cancer patient, there has been a search to isolate tumor cells in the circulation of patients suffering from cancer, now called Circulating Tumor Cells (CTCs). These cells have been initially assessed as non-leucocytic and implicated in the development of distant metastases. CTCs are also known to form clusters with fibroblasts, parent tumor cells, endothelial cells and platelets to protect against oxidative and immune stress and confer a survival advantage when compared to solitary cells. One of the problems of detection, is its rare occurrence in peripheral blood- a frequency of 1 in 109 blood cells. To overcome this infrequent incidence, these cells are captured using antibody-based isolation techniques that have targeted specific cell surface proteins (EpCAM). Though partly successful, such antibody specific isolation does not capture the CTC heterogeneity caused by the expression of different proteins on its cell surface. In addition, such EpCAM cell surface proteins are downregulated during the epithelial-mesenchymal transition, making cell detection even more difficult. The frequency of release of cells into the blood stream has also not been established and it may not be ubiquitous throughout therapy. Though the role of CTCs in brain tumors is still being defined, CTCs have been extensively studied in breast cancer- its role in prognosis, and functions of clonal subtypes including roles of EpCAM positive and negative cells have been very well validated. Negative selection such as erythrocyte lysis and density gradient separation has also been used for CTC separation. Muller, et al. identified CTCs in the blood of glioma patients using Glial Fibrillary Acidic Protein [GFAP] as a marker in 20% of glioma patients. Because of their major difference in size when compared with other circulating cells, dielectric properties and plasticity of CTCs, multiple microfluidic systems are now available to isolate circulating tumor cells, though they are hampered by their low specificity. Sullivan et al. used a similar micro-fluidic system using three antibodies- anti CD14, anti CD16 and anti CD45 to capture CTCs in 13 out of 39 patients. Also, Gao, et al. detected circulating tumor cells in 77% of glioma patients using selective enrichment and Fluorescent-In Situ Hybridization (FISH) while Macarthur described a telomerase based system to identify circulating cells in patients. In spite of this early promise, the fact that CTCs are extremely sensitive and may require almost immediate processing may prove to be a barrier in adoption.
The role of circulating cells in CSF in the management of primary and secondary tumors is much less clear. However, the presence of circulating tumor cells in CSF have now been elucidated in several studies. Nayak et al. in a study of 51 CSF samples in patients with solid tumors found the evidence of leptomeningeal disease in 16 patients, and all 15 who had clinical suspicion of leptomeningeal spread. The specificity and sensitivity of the tests were 100% and 92.7% respectively, far more than any other prevailing modalities. In the largest study of 95 patients with various epithelial tumors, using the cut off of 1 CTC per ml CSF, the test had a specificity and sensitivity of 95% and 93%, establishing the role of CSF CTCs in the leptomeningeal spread of epithelial cancers.
Cell-free DNA [cfDNA] and circulating tumor DNA (ctDNA)
The first report of cell-free DNA in the blood was reported by Mandel and Mathias in 1948. However, only in 1977 was the topic discussed again when Leon and his colleagues demonstrated an increased concentration of ctDNA in a patient with pancreatic cancer. The clinical potential of ctDNA was recognized further when Sorensen and colleagues detected a KRAS gene mutation in a patient with pancreatic cancer. Under normal conditions, cell-free DNA (cfDNA) is released into the blood stream from apoptotic and necrotic cells. In addition, DNA is also spontaneously released in the blood stream and is known as 'metabolic DNA', which may form a transcriptional template for RNA or bind to glycoproteins as messengers. Any such free-floating DNA is rapidly phagocytosed by macrophages. In malignant conditions, the macrophage phagocytosis is exhausted and increasing amounts of nucleosomes accumulate in the blood, which is usually the result of a large number of necrotic and apoptotic cells. There is some evidence to suggest that ctDNA may be released into the bloodstream by an active process. Bergsmedh also suggested that this active transfer of ctDNA is capable of mediating metastasis and generating the necessary genetic instability for cancer progression.
In the normal population, there is a high variability in the amount of cfDNA found in the blood, with a relatively short half-life. High levels of cfDNA are not necessarily correlated with malignant disease but are often associated with other conditions like trauma and inflammation, but there does seem to be an increase in the ctDNA with an increase in the size of the tumor. Even in cancer patients, ctDNA is often a very small proportion of the total cf DNA, though the ctDNA fragments are substantially shorter than the cfDNA found in normal subjects. In addition, blood stored at room temperature results in the rapid lysis of WBC, with the release of germline DNA into the serum. Recent technological advances have however overcome the limitations of isolating small fragments of DNA. Targeted approaches to find the specific gene mutation or specific re-arrangements in the gene structure have been accurately detected using variations of the 'PCR' technology - Droplet Digital PCR (ddPCR or BEAMing) using a combination of Beads, Emulsion, Amplification and Magnetics. Targeted regions are amplified and measured by fluorescent probes that bind to the amplified regions. Where specific targets are not known, or multiple targets are suspected, Whole Exome Sequencing (WES) or Whole Genome Sequencing (WGS) have also been used with a high degree of success.
The presence of ctDNA in the serum of patients with glioma was first shown by Lavon and then by Marchrzak-Celinska. In an elegant study of 70 patients of high-grade gliomas and oligodendrogliomas, Lavon et al. were able to demonstrate the presence of either a loss of heterozygosity of 1p19q or a methylation of MGMT or PTEN in 62 of the 70 serum samples. When compared to the results obtained from the tumor, a sensitivity of 55% for methylation and 51% for LOH respectively with a specificity of 100% for both tests were demonstrated. Marchrzak-Celinska on the other hand collected serum from patients with meningioma and gliomas. Of the 17 glioma patients in her study, she successfully detected methylation of MGMT, RSSFIA, P15INK4B and P14ARF with a specificity of 100% and a sensitivity of 50%. Studying the ctDNA of all primary brain tumors using the Gardent360 cell-free DNA detection assay, Piccioni et al. demonstrated that more than one somatic mutation was observed in 53% of glioblastoma patients, though the percentage was much lower for other gliomas. In a recently published study, Nassiri et al. using cell-free methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq) showed that methylation profiling of ctDNA could detect and discriminate different intracranial tumors, some of which had a similar cells of origin. The presence of ctDNA in the CSF of glioma patients has also been shown repeatedly. Pan et al. showed tissue concordant mutations in NF2, AKT, BRAF, NRAS, KRAS AND EGFR in the CSF of 7 patients, underlying its diagnostic potential. De-Mattos Arruda also demonstrated the presence of mutations of EGFR, PTEN, ESR1, IDH1, ERBB2 and FGRR2 in glioma 12 patients, the specificity and sensitivity of the CSF observations being more than the plasma. In a study of pediatric patients, Huang et al. showed that histone mutations (H3K27M) associated with Diffuse Intrinsic Pontine Glioma (DIPG) was detected in the CSF of four out of six patients. Mutations in histones were also demonstrated in the midline supratentorial tumors. In spite of a fair degree of evidence, ctDNA is yet to be unequivocally established in clinical practice.
Endovesicles and exosomes
Exosomes are nano-vesicles between 40-100 nm in diameter and are the end product of the recycling endosomal pathway that originate from the inward budding of the plasma membrane. Though they were initially considered to be cellular waste, it has now been demonstrated that exosomes play a very important role in the inter-cellular communication between distant cells. Exosomes also take part in a number of pathological processes vital to the spread of cancer including proliferation, migration, invasion, angiogenesis and even Epithelial-Mesenchymal-Transformation (EMT). Exosomes consist of a lipid bilayer which contains both transmembrane and non-membrane proteins, as well as miRNAs, mRNAs, and either single or double-stranded DNA. Though the exosomes were also found to be characterized by a conserved set of proteins independent of the cell of origin, the general protein composition roughly resembles their cell of origin, suggesting exosomal tissue-specific signature. Some of the available protocols for exosomal separation are based on their biophysical properties including size, morphology and density while others are based on immunoaffinity capture or altering exosome solubility and improving their precipitation.
Exosomes also have an appreciable amount of nucleic acids that lend themselves amenable to biomarker investigations. In certain cases, the nucleic acid concentration in exosomes is several times that found in parent cells, suggesting active 'packing'. Noerholm et al. analyzed serum-derived exosomes from 10 patients and found significantly lower levels of 4 ribosomal function genes when compared to normal- RPL11, RPS12, TMSL3 and BM2. Skog et al. showed EGFRvIII mRNA in detectable levels in microvesicles. In a similar study of 88 patients of glioblastoma, Manda et al. compared the presence of EGFRVIII in the tumor vs the exosome in patients with glioma; where the presence of EGFRVIII was the gold standard. Using PCR for amplification of EGFR, which included its splice variant EGFRVIII, and a 2X2 table, we showed that the specificity and sensitivity of detection of EGFR and its variants were about 80% in the serum when compared to tumor. However, even more interesting, is that on follow up of two of these 'low-grade patients' who had false-negative results (EGFR + ve in Exosomes but negative in the tumor), we found that their clinical course was more in keeping with that of a glioblastoma than a Grade 2 tumor (unpublished data). A larger study looking at this group of patients more critically may be able to differentiate this small but very significant subset of patients.
Manterola et al. found increased levels of RNU6-1 and miRNA-320 and miRNA-574-3p that correlated with GBM diagnosis with a specificity and sensitivity of approximately 86%. Similar results have also been reported from the isolation of exosomes from the CSF. Figueroa et al. showed EGFRVIII can be detected in the CSF of 81 patients with GBM, with a sensitivity of 60% and a specificity of 98% when compared to the EGFRVIII presence in the tumor. In a similar study of CSF exosomes, Akers et al. demonstrated a 10-fold increase in the miR-21 levels when compared to controls. These results were then validated with a larger set of 29 patients yielding a diagnostic specificity and sensitivity of 87% and 93% respectively and went on to show that the microvesicles in the CSF are enriched with miRNA. Using unbiased high throughput next-gen' sequencing and an integrative bioinformatics platform, Ebrahimkhani et al. found 26 differentially expressed miRNAs in GBM patients when compared to healthy controls. They selected a panel of seven miRNAs-miRNA-182, miR-328-3P, miR339-5p miR340-5p, miR-486-5p and miR-543 that predicted the GBM diagnosis with a 91% accuracy. Within this multivariate model, four miRNAs -miRNA-182, miR-328-3P, miR340-5p, miR-486-5p distinguish GBMs from healthy controls with 100% accuracy. Using on-chip immunofluorescence to measure the concentration of GFAP and TAU proteins in exosomes, Lewis et al. found it was possible to differentiate the plasma of controls from that of glioblastoma patients with 60% and 94% accuracy respectively. This was later validated using another independent cohort of patients.
Serum miRNA, proteins and tumor educated platelets (TEP)
Though CTC, ctDNA and exosome remain the most popular, reliable and investigated platforms in liquid biopsy, several other platforms have also been tried. miRNA profiles have been isolated from the serum and the CSF and have separated GBM form controls with a fair degree of accuracy. Dong et al. found a 3 miRNAs- mIR-340, mIR-576-5p and mIR-626 that were overexpressed in the serum of GBM patients, whereas mIR-320, mIR-7-5p and mIR-let-7g-5p were significantly reduced. Similar results were also shown by Tang et al. - miR-85 serum levels correlated well with surgery and radiotherapy in 66 patients. Yue et al. showed that mIR-205 is significantly decreased in GBM patients and correlated with Karnofsky Performance Status, grade of tumor and survival. Teplyuk et al. investigated miRNAs in the CSF and found miR-10b and miR-21 levels were increased in the CSF of GBM patients.
Estimation of specific proteins in the blood has also generated considerable interest among researchers. GFAP, a type III intermediate filament protein that is expressed in glial tumors, has been shown to be specific and sensitive in varying degrees in several studies. Recently, van Bodegraven et al. suggest that the GFAP positive cell population contain differences in morphology, function and differentiation, suggesting GFAP is a marker of less malignant and differentiated gliomas. The authors also suggest differentiating between the two isoforms GFAPδ and GFAP± to increase the accuracy of testing. EGFR DNA amplification, one of the pathognomonic features of GBMs have been found to have increased levels of EGFR protein in the circulation.
In a study of the CSF of ten patients with diffuse pontine glioma of childhood (DIPG), Saratsis et al. found increased levels of two proteins in the CSF- Cyph A and DDAH1. These proteins were not upregulated in other lesions of the pons or in other gliomas locations. When these proteins were tested in the serum and urine, they were detected in the only (DIPG) patient in whom a serum sample was available, the two patients in whom urine samples were available but also in the urine of one of the three controls who had no documented brain lesions.
In the last few years appreciation has been increasing for the role of platelets in cancer.
Platelets have been known to support the tumor environment by helping neo-vascularisation by the secretion of several pro-angiogenic factors, reducing tumor apoptosis and inducing EMT to protect circulating tumor cells form immune responses. It has been shown that platelets can infiltrate the tumor tissue and exosomes in the serum are sequestered in them, thereby giving rise to the term TEP or Tumor Educated Platelets. In glioblastoma patients, EGFRVIII RNA molecules have been isolated from such platelets.
A summary of the major liquid biopsy modalities and their relative strengths and weaknesses in summarised in [Table 1] (adapted from Siravegna et al.).
Clinical paradigms of liquid biopsy use
Though none of the current methods of liquid biopsy are approved by the FDA yet, it is only a matter of when, not if. There is now a burgeoning body of work that points to the usefulness of liquid biopsy in the diagnosis and management of gliomas. EGFR, has been shown to be present in the serum and the CSF of glioblastoma patients in multiple large studies. In addition, there is evidence that CSF EGFR ctDNA may be a rapid way of assessing response in glioblastoma patients. Prognostic data for GBM can also be obtained by detecting miRNA in the blood and the CSF of these patients. With specificity and sensitivity rates above 80%, the use of liquid biopsy in GBM diagnosis imminent. Mutations in IDH, probably the most important marker in glioma prognostication and diagnosis, and is usually seen in lower grade gliomas has also been reliably shown in the ctDNA of low grade glioma patients. With data suggesting the higher incidence of IDH mutant tumors in India, an IDH-mutant specific target will greatly benefit to the Indian population. Mutations in BRAF, are also being detected in the blood and CSF of patients with optic pathway, pilocytic and pilomyxoid tumors. As the v600e mutation is targetable, and targeted therapy results in a dramatic clinical response, the detection of this mutation in patients can completely eliminate the need for an invasive biopsy, and already being pioneered in some centres of clinical excellence. The prognosis of midline gliomas depends on the presence or absence of H3K27M. This can be reliably detected in CSF and blood and even in patients with DIPG, where traditionally a tissue biopsy is avoided. With several clinical trials now targeting this mutation now in its final stages, liquid biopsy will become the default option for all these cases. Several other primary brain tumors have also been found to lend themselves to liquid biopsy- including meningiomas, ependymomas and medulloblastomas.
In India, our current clinical pathway is hindered by nationwide inadequacies in infrastructure and skilled manpower (baring the exceptions in the big cities), difficulties in early diagnosis, limitations in imaging resolution and available therapeutic options. Though liquid biopsy does not provide all the answers, its use can ease some of the problems in diagnosis and treatment. Liquid biopsy-based glioma diagnosis will be useful not only in the remote areas of India where infrastructure is scarce, but also in patients where a biopsy is contra-indicated either because of general debility, other life-threatening disease, or the presence of anticoagulation. It will also be useful in deep-seated lesions in the hemispheres or the brain stem where even needle biopsies are risky or in the presence of gross neurological deficit where radical excisions are usually not undertaken. The accurate diagnosis of prognostic markers before surgery does permit more meaningful discussions on informed consent, especially in the elderly where the prognosis remains grim irrespective of treatment. With the increase in targeted therapy in gliomas, liquid biopsy will not only reduce the time to the start of therapy, but also allow real-time targeting of cancer-causing pathways as the disease progresses.
A spatially and temporally limited tumor biopsy may not be the best diagnostic option against a malignancy, which to this day, has continued to frustrate most efforts. While it is easy to get carried away with all its promise, liquid biopsy in gliomas today is more a research tool with an anecdotal use in the clinic. For this to become mainstream, large prospective studies with multimodal liquid biopsy approaches is the need of the hour. Only when clinically validated, does it become an invaluable tool in our fight against gliomas.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.