Extracellular vesicles as source for the identification of minimally invasive molecular signatures in glioblastoma

Seminars in Cancer Biology 87 (2022) 148–159 Available online 11 November 2022 1044-579X/© 2022 Elsevier Ltd. All rights reserved. Extracellular vesicles as source for the identification of minimally invasive molecular signatures in glioblastoma Elisabeth Rackles a , * , Patricia Hern´ andez Lopez a , Juan M. Falcon-Perez a , b , c , d , * a Exosomes Laboratory, Center for Cooperative Research in Biosciences (CIC bioGUNE), Basque Research and Technology Alliance (BRTA), Derio, Spain b Metabolomics Platform, CIC bioGUNE, Bizkaia Technology Park, 48160 Derio, Spain c Centro de Investigaci´ on Biom´ edica en Red de Enfermedades Hep´ aticas y Digestivas (Ciberehd), Madrid, Spain d Ikerbasque, Basque Foundation for Science, Bilbao, Spain ARTICLE INFO Keywords: Extracellular vesicle Exosome Biomarker Liquid biopsy Glioblastoma ABSTRACT The analysis of extracellular vesicles (EVs) as a source of cancer biomarkers is an emerging field since lowinvasive biomarkers are highly demanded. EVs constitute a heterogeneous population of small membranecontained vesicles that are present in most of body fluids. They are released by all cell types, including cancer cells and their cargo consists of nucleic acids, proteins and metabolites and varies depending on the biologicalpathological state of the secretory cell. Therefore, EVs are considered as a potential source of reliable biomarkers for cancer. EV biomarkers in liquid biopsy can be a valuable tool to complement current medical technologies for cancer diagnosis, as their sampling is minimally invasive and can be repeated over time to monitor disease progression. In this review, we highlight the advances in EV biomarker research for cancer diagnosis, prognosis, and therapy monitoring. We especially focus on EV derived biomarkers for glioblastoma. The diagnosis and monitoring of glioblastoma still relies on imaging techniques, which are not sufficient to reflect the highly heterogenous and invasive nature of glioblastoma. Therefore, we discuss how the use of EV biomarkers could overcome the challenges faced in diagnosis and monitoring of glioblastoma. 1. Introduction Cancer is a complex disease because each tumor has a unique cellular composition and, therefore, shows varying progression rates and response to therapies. For example, glioblastoma (GBM) has a remarkably diverse cellular composition including differentiated tumor cells, glioma stem cells, and non-tumor cells that form the tumor microenvironment. This high intra-tumor heterogeneity as well as its invasive nature are reasons for its aggressiveness [1]. This characteristic of tumors shows the relevance of developing methods that allow to obtain a holistic image of the heterogeneous tumor and to follow therapy efficacy in a minimally invasive manner. Liquid biopsy can be a valuable tool to complement current medical technologies for cancer diagnosis as it is minimally invasive and allows repeated longitudinal sampling. The liquid biopsy sample contains various sources for biomarkers like for example circulating tumor cells, circulating cell-free nucleic acids, tumor educated platelets, and tumor derived extracellular vesicles (EVs) [2]. The analysis of EVs as a source of cancer biomarkers is an emerging field. They are small membrane-contained vesicles, which are released by all cell types. There are three types of EVs, which are classified depending on their biogenesis, way of release and size: exosomes, microvesicles and apoptotic bodies [3]. Exosomes are small extracellular vesicles (50–100 nm of diameter), which are formed within the endosomal network and are released into the extracellular medium upon fusion of multi-vesicular bodies with the plasma membrane [4]. Microvesicles are formed by the outward budding and pinching of the plasma membrane and are more heterogenous in size than exosomes (0.1–1 µm of diameter) [5]. Apoptotic bodies generally are bigger in size and are originated from cells undergoing apoptosis [6]. Additionally, there is a subpopulation of EVs that seem to be unique to cancer cells, the so-called large oncosomes. They are non-apoptotic EVs that originate through membrane shedding from cancer cells [7]. EVs contain a cargo including nucleic acids, proteins, lipids, and metabolites [3,4] that varies between the different EVs subtypes [3,4,8,9]. EVs are involved in intercellular communication and the cargo can be transferred to a recipient cell [3]. In cancer patients, exosomes can transfer bioactive molecules between the parent cancer cell and recipient cells in the local * Correspondence to: Exosomes Laboratory, CIC bioGUNE-BRTA, Parque Tecnologico Bizkaia, Bldg. 800, Derio 48160, Bizkaia, Spain. E-mail addresses: [email protected] (E. Rackles), [email protected] (P.H. Lopez), [email protected] (J.M. Falcon-Perez). Contents lists available at ScienceDirect Seminars in Cancer Biology journal homepage: www.elsevier.com/locate/semcancer https://doi.org/10.1016/j.semcancer.2022.11.004 Received 1 July 2022; Received in revised form 21 October 2022; Accepted 8 November 2022 Seminars in Cancer Biology 87 (2022) 148–159 149 tumor microenvironment as well as to distant tissues, thereby participating in cancer progression and metastasis [10–14]. Although EVs are classified by their origin and size, it is difficult to separate and characterize the subpopulations of EVs in practical applications because there is not yet a consensus on specific markers for the subcellular origin of the different subtypes. Therefore, the international society of extracellular vesicles (ISEV) recommends using terms that describe the studied EVs population according to their properties (size, density, biochemical composition) or descriptions of conditions or cell of origin [15]. In this review, the terms exosome and microvesicle are used if it is clarified in the referenced publication and the term EV is used if the differentiation is unclear. The EV’s cargo varies depending on the cell type and the biologicalpathological state of the secretory cell. Thus, the EV’s cargo is considered as a potential source of reliable biomarkers for different diseases. Additionally, as exosomes can be isolated from all body fluids, they have been proposed as a noninvasive source of biomarkers [16]. In this review, we will highlight the advances in EV biomarker research for cancer diagnosis, prognosis, and therapy monitoring using glioma as an example. First, we will give an overview of EV isolation methods and the different cargos of EVs. This section will include some selected examples how EVs can serve as biomarkers for various cancer types. Finally, we will discuss the great potential of EV-derived biomarkers compared to traditional biomarkers using the example of glioma. We searched the PubMed database using various combinations of the following keywords “glioblastoma”, “glioma”, “extracellular vesicle”, “exosome”, “biomarker”, “liquid biopsy”. Additionally, the references of the papers were also screened for further publications of interest. We focused on publications that used liquid biopsies as a source for EVs since low-invasive biomarkers are highly demanded. To focus on biomarkers that are mostly present in EVs, we excluded publications that use precipitation methods for EV isolation. These EV preparations contain co-isolated material that also can have value for biomarker identification but is not EV-contained and therefore not in the scope of this review (for reviews including precipitation-based methods refer to [17–21]). The review combines a broad overview about the EV’s cargo and EV isolation methods with a valuable summary of the latest publications on the application of EVs for diagnosis of glioma. Thus, the review joins publications on EV and glioma research making it relevant for readers of both fields. 1.1. Liquid biopsy as a low-invasive source for extracellular vesicles Liquid biopsies consist of the isolation and analysis of biological material present in body fluids of individuals. In contrast to traditional biopsies of tissues, they constitute a non-invasive manner of diagnosis and monitoring [2]. EVs are present in all body fluids [16]. Furthermore, from the cancer perspective, the secretion of EVs is increased in cancer cells compared to non-malignant cells [22–25]. Another important characteristic of exosomes for biomarker research is that they can cross the blood brain barrier [26]. García-Romero et al. showed that all EVs subtypes derived from glioblastoma can cross the blood brain barrier and can be detected in the peripheral blood in an orthotopic xenotransplant mouse model [27]. This characteristic of EVs allows the detection of brain tumor biomarkers by blood sampling, which is less invasive compared to liquid biopsies obtained from cerebrospinal fluid (CSF). Liquid biopsies of CSF and blood have been analyzed for the detection of biomarkers for glioblastoma [21] and some biomarkers, like for example epidermal growth factor receptor (EGFR) RNA, was identified in EVs derived from both body fluids [28,29]. Blood plasma and urine are useful sources of EVs for the detection of prostate cancer [30]. For example, a study analyzing exosomes isolated from plasma of healthy controls, prostate cancer patients and patients with benign prostatic hyperplasia showed statistically significance between the groups in the expression of the tetraspanin CD81 and Prostate Specific Antigen (PSA). The study was able to distinguish prostate cancer patients from healthy individuals with 100% specificity and sensitivity [31]. A targeted proteomic analysis of urinary EVs in prostate cancer revealed a signature of five proteins to significantly distinguish between highand low-grade prostate cancer. This signature also included a tetraspanin, CD63, and PSA [32]. Furthermore, the changes in metabolism in prostate cancer patients are reflected by the cargo of urinary EVs. A panel of 76 metabolites is significantly changed in prostate cancer patients compared to benign prostate hyperplasia [33]. Saliva is another promising liquid biopsy that is even less invasive than blood sampling. Small EVs derived from saliva were proposed to carry biomarkers for oral squamous carcinoma [34]. Further examples for liquid biopsies that have been used for the identification of EV-derived cancer biomarkers are pleural effusion and bronchoalveolar lavage which carry biomarkers for lung cancer [35,36]. 1.2. Isolation methods of extracellular vesicles for cancer biomarker identification For the detection of biomarkers, the EVs need to be isolated from the liquid biopsies. Different methods for EVs isolation based on their physical properties or the expression of surface proteins exist. Traditional separation methods are differential ultracentrifugation, size exclusion chromatography, ultrafiltration, polymer precipitation and immunoaffinity-based capture [37]. In 2019, differential ultracentrifugation alone or in combination with other methods was the most widely used technique for EV isolation [38]. Differential ultracentrifugation exploits differences in the density, shape, and size of vesicles and other particles. Being the most common method, ultracentrifugation has the advantage of being a robust method. Furthermore, it can be applied to large volumes of samples [37]. However, the yield is low and EVs can be ruptured during the process [37,39]. In contrast to ultracentrifugation, the use of size exclusion chromatography to isolate exosomes from urine leads to a higher yield and higher biological activity of the exosomes [39]. This might be one of the reasons why size exclusion chromatography is gaining popularity. This method is based on the size of the EVs, which are excluded from pores in the stationary phase of a column and, therefore, are eluted earlier than proteins and other contaminants [40]. However, EV preparations obtained by size exclusion chromatography show higher contaminations of proteins and lipoproteins than preparations by ultracentrifugation [41]. Another EV isolation method that depends on size is ultrafiltration, which is based on membrane filters with different pore sizes or predefined molecular weight cutoffs [42,43]. An easy and fast approach to isolate EVs is the use of polyethylene glycol or commercial kits to precipitate the EVs and recover them by centrifugation [44]. However, the EV pellet obtained by polymer precipitation contains various contaminants [45]. In contrast, the isolation of exosomes by immunoaffinity-based capture yields a highly pure exosome preparation. This method is based on the binding of antibody-coated beads to specific antigens only present on exosomes [46]. The high specificity is also a drawback of this method, since only exosomes presenting the antigen will be captured and not all exosomes present in the sample. In addition to the traditional separation methods, new techniques are emerging. One example is the use of microfluidic devices, which have the advantage of being fast, low-cost and showing high recovery rates [47]. Like the traditional methods, EVs are separated in these devices based on their size and density or based on specific surface markers [37]. For the use of liquid biopsies as a source of biomarkers, it is essential to be aware of the influence of pre-analytical conditions on the composition and stability of EVs [48,49]. First, the EV population from blood samples differs if it is isolated from serum or plasma and depending on the anti-coagulant used for plasma preparation. Furthermore, factors like transportation, storage temperature and storage time of the plasma has an influence on the samples [50]. Also after isolation of the EVs, storage time and temperature as well as repeated freeze-thaw cycles have an impact on the EVs [51]. Furthermore, the isolation of EVs E. Rackles et al. Seminars in Cancer Biology 87 (2022) 148–159 150 from liquid biopsies requires the optimization of the method for each body fluid. The chosen isolation method can have an influence on the obtained biomarker profile. For example, we have compared five different methods to isolate urinary EVs from healthy individuals and found that every method yielded a different composition of protein markers and that differential ultracentrifugation is the most efficient in isolating CD63-containing EVs [52]. The variety of isolation methods and a lack of consensus affects the reproducibility and comparability of different studies. For the selection of a method for EVs isolation it is crucial to consider factors like sample nature, sample volume, desired yield and purity, and final use of the EVs. Remarkably, different methods could be needed to analyze the different EVs existing in a liquid biopsy. Furthermore, a proper characterization of the obtained EVs population is mandatory and will ensure that the observed biomarkers are indeed associated with EVs and not with co-isolated material. Therefore, the ISEV published guidelines for EV characterization: The experiment should include information about the amount of the EV source (e.g. volume of body fluid), the abundance of EVs, and the presence of specific markers for EVs or potential co-isolated material [15]. 1.3. The molecular components of EVs represent potential non-invasive biomarkers The EV’s molecular composition varies depending on the cell type and the biological-pathological state of the secretory cell (Fig. 1). The EV’s membrane has a specific lipid composition and contains specific membrane proteins. In vitro studies showed that EVs are enriched in cholesterol, sphingomyelin, glycosphingolipids, and phosphatidylserine compared to their parental cell lines [53]. The most abundant lipid classes in EVs from human plasma and serum are sphingomyelin, phosphatidylcholine, and ether-type phosphatidylcholine [54]. Exosomes contain membrane proteins like tetraspanins, glycoproteins, antigen presenting molecules, and adhesion molecules. The tetraspanins CD9, CD63 and CD81 are often used as specific markers for the isolation of exosomes [3]. The EV’s cargo consists of nucleic acids (DNA, RNA, non-coding RNA), proteins, and metabolites. Although the molecular mechanisms of the cargo sorting into EVs is poorly understood, accumulating evidence indicates that the cargo is sorted and secreted by regulated pathways [55]. Thus, the EV’s cargo is considered as a potential source of reliable biomarkers for cancer diagnosis and monitoring. In the past years, proteins were the most studied EV cargo in the field of EV derived cancer biomarker research [20]. The methods for detection of EV protein biomarkers in liquid biopsies are diverse. Mass spectrometry analysis of cancer patient and control samples reveal high numbers of differentially expressed proteins, which can be validated as biomarkers by follow-up experiments. For example, a combination of untargeted proteomics and validation by targeted proteomics revealed three potential urinary EV biomarkers (HSP90, SDC1, and MARCKS) for prostate cancer [56]. Another way to screen for EV biomarkers is the use of an EV protein expression array [57]. Using this method, Zhang et al. identified 42 candidate EV proteins for monitoring the immunotherapeutic outcomes of gastric cancer, of which a panel of four proteins was validated as biomarkers [58]. A disadvantage of most EV isolation methods is that samples can be contaminated by proteins [38]. A new method using a thermophoretic aptasensor overcomes this issue and has been used successfully to discriminate between metastatic and non-metastatic breast cancer patients and healthy individuals based on a signature of eight proteins [59]. Ferguson and colleagues used another alternative approach and showed recently that a single-EV analysis of mutated KARS and P53 proteins can identify early-stage pancreatic cancer [60]. Cell-free nucleic acids like circulating tumor DNA or extracellular RNA have the potential for serving as biomarkers for cancer. However, their levels in the liquid biopsy are very low, and the nucleic acids are subjected to degradation [61]. In contrast, the analysis of nucleic acids present in EVs has the advantage that the nucleic acids are protected by the EV’s membrane and can be enriched by the isolation of the EVs from the liquid biopsy. Few studies directly compared the performance of circulating tumor DNA and EV-DNA as biomarkers for cancer. For example, EV-DNA outperforms cell-free DNA in plasma in the detection of mutations in non-small-cell lung cancer and pancreatic cancer [62, 63]. In contrast, the mutation detection for colon cancer was not more sensitive using EV-DNA compared to cell-free DNA in plasma [64]. Besides DNA, EVs carry all types of RNA, including mRNA, miRNA, long non-coding RNA (lncRNA), and circular RNA (circRNA) [19]. Non-coding RNAs show great potential for the detection of novel cancer biomarkers due to their importance in the pathogenic mechanisms of malignant cells and their enrichment in EVs compared to their parental cell. The biomarker research on non-coding RNAs in cancer EVs has a major focus on miRNAs, while the analysis of EV-derived lncRNA and circRNA as potential cancer biomarkers is a relatively novel field and Fig. 1. Molecular composition of extracellular vesicles (EVs). A heterogenous population of EVs derived from malignant and healthy cells is present in liquid biopsy. The EV´s cargo is heterogeneous and is comprised of nucleic acids, proteins, metabolites, and lipids. The EV’s membrane is enriched in specific lipids, displays an individual glycosylation profile and characteristic membrane proteins. All these molecules can serve as potential low-invasive biomarkers for cancer. E. Rackles et al. Seminars in Cancer Biology 87 (2022) 148–159 151 further studies are needed to assess their clinical value [65–67]. Also coding RNA can serve as a biomarker and a very promising example is the quantification of mRNA levels of urinary exosomes from prostate cancer patients [68]. The analysis can effectively detect high-grade prostate cancer in patients and is currently at the stage of a clinical trial. The test supports the clinical decision-making process and improves patient stratification [69]. In contrast to the above-mentioned cargos, the knowledge about the EV’s metabolome as a potential biomarker for cancer is still limited. The metabolism directly reflects the underlying biochemical activity and state of the cells. Since the metabolism of cancer and healthy cells differ, metabolomics analysis of the cells and their EVs best represents their molecular phenotype. Several studies have shown that the fatty acid or lipid composition of exosomes can distinguish cancer patients from healthy individuals [70–73]. Also, metabolites of many other classes were identified in the EV’s cargo of cancer patients [33,35,74,75]. For example, a recent study compared the metabolome of whole serum and serum-derived exosomes of head and neck cancer patients and healthy individuals. It showed that in both, whole serum and EVs, metabolites present in energy production pathways and inositol metabolism differ in cancer patients and controls. Furthermore, metabolites present in the oxidation of fatty acids and ketone body metabolism were specific to serum-derived exosomes [74]. Changes in the glycosylation profile is a common feature of malignant cells. Interestingly, also the glycosylation profile of lipids and membrane proteins of EVs differs if they are derived from a malignant cell [76]. One example for the diagnosis of cancer based on glycans is the measurement of free sialyl Lewis A antigen (CA19–9) in serum of patients, which is currently used as a biomarker for pancreatic cancer [77]. A recent study showed that CA19–9 is enriched in exosomes isolated from whole blood. Noteworthy, in one case the measurement of free CA19–9 in serum showed a false negative while the exosomes were positive for the biomarker suggesting that CA19–9 in exosomes could represent a more sensitive biomarker [78]. Another study showed that EVs presenting specific O-glycans are significantly increased in pancreatic cancer patients and this potential biomarker can even identify early stages of pancreatic cancer when the measurement of free CA19–9 is still negative [79]. Furthermore, the capture of specific glycans present on exosomes in plasma samples can differentiate between metastatic and non-metastatic pancreatic cancer patients as well as healthy individuals [80]. Although there are many promising studies on EV-derived biomarkers, the field still faces difficulties in translating the results into clinical biomarkers. Therefore, many studies focus on the development of easy, low-cost and efficient methods that are suitable for routine clinical applications [59,81]. Multiplexing platforms have been developed, since a single biomarker is in most cases not sufficient for a precise diagnosis [81,82]. Raman spectroscopy might be another method to overcome this problem since it provides information on the entire molecular cargo of the EVs. The bioinformatic analysis of Raman spectra of EVs isolated from serum of patients with different brain tumors was able to discriminate between the different patient groups [83]. Furthermore, several studies show that the analysis of different EV cargos like nucleic acids and proteins combined with cell-free DNA leads to the identification of more reliable biomarkers [84–87]. 2. The potential of EVs as low-invasive biomarkers for glioma Gliomas are a good example for the great potential of EVs to improve diagnosis and monitoring of cancer compared to traditional methods. Gliomas are tumors that arise from glial or precursor cells and are mostly comprised of astrocytomas, oligodendrogliomas and ependymomas [88]. The world health organization (WHO) classified glioma in four grades according to their histological features and predicted clinical behavior. Since the last update in 2021, the grading also includes molecular features of the tumor. GBM are astrocytoma that are classified as grade 4 [89]. GBM is a rare disease with poor prognosis. It represents the most commonly occurring primary malignant tumors of the central nervous system (CNS) [88]. The high intra-tumor heterogeneity as well as its invasive nature of GBM are reasons for its aggressiveness [1]. These features of GBM are also promoted by EVs. It has been shown that the molecular heterogeneity between the proneural or mesenchymal subtypes of glioma stem-like cells is reflected by the EV’s cargo. Interestingly, in vitro treatment of proneural glioma stem-like cells with EVs derived from the mesenchymal subtype, led a significant increase in their growth and viability. Furthermore, an EV-mediated transfer of EGFR from the mesenchymal to the proneural glioma stem-like cells was observed [90]. EVs not only contribute to heterogeneity but also promote invasive capacities of glioma cells. Treatment of glioma cells with exosomes isolated from glioma stem-like cells led to increased growth and invasiveness in a Notch1 dependent manner [91]. Current diagnostic tools cannot completely reflect the heterogeneity of GBM. The diagnosis is mostly based on magnetic resonance imaging (MRI). However, conventional MRI has several limitations since the findings can be difficult to interpret. Especially the differentiation of pseudoprogression from true progression is challenging. In case of pseudoprogression, changes of the tumor can be seen by MRI and be falsely interpreted as tumor progression leading to unnecessary therapy suspension or changes. Furthermore, MRI is done after symptom onset and is therefore not suitable for early detection. In addition to MRI, a tissue biopsy is normally done at the time point of surgery and its histological and molecular profile is analyzed. Given the high heterogeneity of GBM, a tissue biopsy can never reflect all characteristics of the tumor. Moreover, a drawback of both techniques is that longitudinal monitoring of the tumor is not possible. Especially repeated tissue sampling confers high risks for the patient due to its invasiveness [92]. However, the temporal heterogeneity of the tumor and its high invasiveness requires longitudinal monitoring to adapt the treatment to the current status of the tumor. To overcome these problems, intense research is done on liquid biopsies. Sources for liquid biopsies for GBM diagnosis are CSF and blood sampling. Several biomarkers based on circulating tumor cells and cell-free nucleic acids have been described [93]. However, also these biomarkers have limitations. The number of biomarkers is reduced by the blood-brain barrier. The number of glioma derived circulating tumor cells is very low in blood, with only a few cells per 10 milliliters [94]. Similarly, the amount of detectable cell-free DNA in glioblastoma is very low compared to other primary tumors [95]. Piccioni et al. showed a detection rate of circulating tumor DNA in only 55% of GBM patients. Previous studies have showed even lower rates and Piccioni and colleagues argue that this could be due to assay performance or histopathology [96]. This makes these biomarkers not suitable for clinical application where sample volumes are small and precise detection methods are needed. The isolation of EVs can overcome the problem of low amounts of biomarkers. Glioma cells shed high numbers of EVs, and the cargo is protected by the EV’s membrane. Furthermore, EVs derived from GBM differ in their composition from normal glial cells. A few milliliters of sample are already sufficient to get detectable amounts of EV-derived biomarkers. Since EVs can cross the blood brain barrier, GBM-derived EVs can also be found in liquid biopsies of the CSF and blood [21]. Especially the analysis of blood EVs provides non-invasive biomarkers that allow repeated sampling to follow the tumor’s temporal heterogeneity and to monitor treatment efficacy. Furthermore, the molecular diversity of the EV’s cargo allows analysis of nucleic acids and proteins and, therefore, provides more robust biomarkers as it represents a more detailed picture of the tumor’s status (Fig. 2). 2.1. The analysis of the EV protein cargo as biomarker for glioma diagnosis and prognosis The protein cargo of EVs is a useful source for biomarkers for diagnosis of GBM (Table 1). The total amount of protein content of exosomes E. Rackles et al. Seminars in Cancer Biology 87 (2022) 148–159 152 allows to discriminate between healthy controls and patients as well as between glioma grades [97]. Proteomics analysis reveals biomarker signatures consisting of several proteins that can serve for the discrimination of patients displaying glioma of different grades. Analysis of plasma EV samples of 41 glioma grade 2–4 patients and 11 controls revealed grade-specific protein signatures of 78, three and eight proteins for glioma grade 4, 3 and 2, respectively. Three proteins, AK2, CYB5A and GOLT1B, were significantly higher in all three glioma grades compared to controls [98]. Of note, the protein signature identified in EVs for GBM diagnosis largely depends on the cohort. Osti et al. did proteomic analysis of three different pools of plasma samples of in total 43 GBM patients and found that only a small subset consisting of 11 proteins (VWF, APCS, C4B, AMBP, APOD, AZGP1, C4BPB, Serpin3, FTL, C3 and APOE) overlapped between the pools, which they defined as ‘the GBM EV protein signature’. Interestingly, this signature was no longer present after surgery, indicating a direct correlation between the signature and the presence of the tumor [99]. Two of the signature proteins, VWF and C3, were also identified by another proteomic study. Bioinformatics analysis of the biological function of the enriched proteins in this study showed an ‘inflammatory molecular profile’ [100]. Follow-up studies are required to validate these signature proteins in a larger cohort and to implement the measurement of these proteins by methods that are suitable in the clinics. In this line, Rana and colleagues first screened plasma-derived EVs by proteomics for early detection biomarkers for glioma grade 1–3 and then validated one candidate by enzyme-linked immunoassay (ELISA), which is a commonly used laboratory test to detect biomarkers. Galectin-3 binding protein (LGALS3BP) was significantly up regulated in grade 1, 2 and 3 glioma compared to controls [101]. A proteomic analysis of EVs derived patient-derived glioma stem-like cell lines revealed that FASN was present in high levels in EVs [90]. A follow up study showed that FASN could serve as a potential biomarker for glioma. Immunoblotting and flow cytometry analysis showed that FASN levels were elevated in plasma EVs of glioma patients [102]. VWF and FCN3 are further potential biomarkers that have been identified by proteomics and confirmed by immunoblotting or ELISA [103]. Flow cytometry analysis of EV markers is a method that can relatively easily be implemented in clinical routine. Bead-assisted flow cytometry EV analysis was used to screen for glioblastoma biomarkers in vitro and then in patient serum samples. This revealed a protein signature consisting of CD29, CD44, CD81, CD146, C1QA, histone H3, which indicates tumor progression in patients. If the signature is confirmed with a larger patient number, it could help to distinguish pseudoprogression from a tumor progression in cases of equivocal MRI [104]. Other protein biomarkers for glioma were identified in studies that analyzed the function of EVs in glioma. The analysis of tumor promoted vascular leakage revealed that EVs contribute to endothelial and vascular permeability in a Sema3A/NRP1-dependent manner. In a very small cohort, the authors showed that EVs in sera of GBM patients contained high levels of Sema3A and therefore pointed it as a potential biomarker [105]. Pinet et al. studied the role of exosomes in GBM aggressiveness and proliferation. They found TrkB-containing exosomes to be involved in the transfer of glioblastoma aggressiveness to YKL-40-inactivated glioblastoma cells. TrkB was also detected in exosomes isolated from plasma of GBM patients, suggesting that it may be a useful biomarker for diagnosis [106]. The analysis of PTRF/Cavin 1 in tumor progression and intercellular communication of glioma revealed another potential biomarker for diagnosis. PTRF was detected in tumor tissue and in exosomes from serum samples of glioma patients. For both, tissue, and serum, a higher PTRF/CD63 ratio was detected in grade 4 glioma compared to grade 2 glioma patients and the higher ratio was correlated with poorer prognosis. Furthermore, the PTRF/CD63 ratio in serum samples decreased after surgery, indicating a that PTRF could serve as a biomarker for tumor volume or surgical effects [107]. Finally, analysis of the role of glioma exosomes in immune modulation revealed that exosomes of GBM patients contain TGFβ, which is a potentially immunosuppressive cytokine [108]. Since all the above-mentioned biomarkers were discovered as part of a mechanistic study, the cohort analyzed was rather small. Therefore, it is necessary that they are validated in a larger cohort to confirm their value as liquid biopsy biomarkers for diagnosis of glioma. Fig. 2. EV derived biomarkers for glioma. The main sources for EVs as biomarkers for glioma are the blood and CSF. The EVs can be isolated from the liquid biopsy by various methods. Depending on the EV’s cargo that should be analyzed, different methods are applied for biomarker identification. Studies have shown that the DNA, RNA, and protein cargo of EVs contain potential biomarkers for glioma which can be applied for diagnosis, prognosis, and treatment monitoring of glioma. E. Rackles et al. Seminars in Cancer Biology 87 (2022) 148–159 153 2.2. The analysis of the EV nucleic acid cargo as biomarker for glioma diagnosis and prognosis Many studies on EV biomarkers for glioma focus on miRNAs because of their potential function as regulators of oncogenes or tumor suppressors genes (Table 2). One example is miR-21 whose oncogenic potential has extensively been studied in glioma and other solid cancers [109]. miR-21 is increased in CSF-derived EVs of glioma patients compared to healthy individuals [110,111] and the levels of miR-21 are higher in patients with grade 4 glioma than grade 2 glioma [111]. Interestingly, the difference in miR-21 levels can only be observed in EVs present in the CSF but not in the serum [110,111]. miR-21 levels are significantly lower after surgery, which indicates that they are directly correlated with the presence of the tumor. Furthermore, miR-21 can potentially serve as a biomarker for glioma prognosis since its levels are correlated with the anatomical site of tumor reoccurrence and with patient survival [111]. Furthermore, for GBM prognosis, a panel of four miRNAs has been suggested as a potential biomarker. The miRNAs are present in CD44 positive EVs and have been shown to correlate with overall survival. Additionally, the study identified five miRNAs in CD44 Table 1 Liquid biopsy EV protein biomarkers for glioma diagnosis, prognosis, and monitoring. Glioma subtypes are based on CNS WHO grades [89]. If the study included a detection and validation phase, only the size of the validation cohort is mentioned. DU=differential ultracentrifugation, UC=ultracentrifugation, DG-UC=density gradient and ultracentrifugation, SEC=size exclusion chromatography. Glioma grade EV source EV isolation Detection method Biomarkers Study population Potential application Reference 4 Serum DU ELISA Sema3A Patient n=4 Control n=15 Diagnosis [105] 4 Serum DU Immunoblotting EGFR, EGFRvIII, and TGF-β Patient n=12 Diagnosis [108] 2, 4 Serum DU Immunoblotting PTRF/CD63 Patient n=36 Diagnosis, prognosis, surgical effects [107] 2–4 Serum DU Flow cytometry, qRT-PCR EGFR, NLGN3, PTTG1 Patient n=23 Control n=12 Diagnosis [119] 3, 4 Serum DU Flow cytometry CD9 +/SVN+, CD9 +/GFAP+/ SVN+ Patient n=8 Therapeutic effects [133] 4 Serum SEC, DU Flow cytometry CD29, CD44, CD81, CD146, C1QA, histone H3 Patient n=67 Control n=22 Prognosis [104] 4 Plasma UC NTA, proteomics Vesicle number, VWF, APCS, C4B, AMBP, APOD, AZGP1, C4BPB, Serpin3, FTL, C3, APOE Patient n=30 Control n=16 Diagnosis, surgical effects [99] 4 Plasma DU Immunoblotting TrkB Patient n=11 Control n=6 Diagnosis [106] 4 Plasma DU Proteomics VWF, FCGBP, C3, PROS1, SERPINA1 Patient n=15 Control n=10 Diagnosis [100] 1–3 Plasma DU Proteomics, ELISA LGALS3BP Patient n=40 Control n=40 Early detection [101] 4 Plasma DU Flow cytometry Annexin V+Patient n=16 Prognosis [124] 4 Plasma DU Flow cytometry Vesicle number Patient n=11 Control n=7 Therapeutic effects [127] 3–4 Plasma DU, SEC Immunoblotting, flow cytometry FASN, FASN+/CD63 +, FASN+/ CD81 + Patient n=29 Control n=17 Diagnosis [102] 4 Plasma SEC, UC Interferometry light microscope, ELISA, immunoblotting, proteomics Vesicle number, VWF, FCN3 Patient n=10 Control n=10 Diagnosis [103] 2–4 Plasma SEC Proteomics AK2, CYB5A, GOLT1B Patient n=41 Control n=11 Diagnosis [98] 4 Plasma Microfluidic platform NMR Vesicle number, CD63, EGFR, EGFRvIII, Patient n=12 Therapeutic effects [128] 4 Plasma Multiplexing Electrochemical impedance measurement EGFR, EGFRvIII, PDGFR α Patient n=10 Control n=10 Diagnosis [82] E. Rackles et al. Seminars in Cancer Biology 87 (2022) 148–159 154 positive EVs that are suitable for GBM diagnosis [112]. Expression levels of miR-210 in serum exosomes might also serve for GBM diagnosis. miR-210 levels are higher in glioma patients than in healthy individuals, reflect the glioma grade, and decrease after surgery. Additionally, they are strongly associated with markers of tumor hypoxia [113]. Deep sequencing allows the unbiased identification of EV miRNA biomarkers for glioma. miRNA sequencing of exosomes isolated from CSF identified miR-1246 as a potential a biomarker to detect tumor recurrence after surgery [114]. Using deep sequencing and subsequent bioinformatics analysis, two distinct signatures for the identification of highor low-grade glioma were identified. The signature for the diagnosis of GBM is comprised of seven miRNAs and has a predicted accuracy of 92% [115]. Stakaitis and colleagues studied the potential of miRNAs as biomarkers from a new point of view. They analyzed the association of miR-181b/d expression in tissue and serum exosome samples with their association with patients’ functional and psychological outcome. They found that the expression levels of miR-181b/d in tissue and exosomes were weakly correlated with the patient’s functioning and symptoms Table 2 Liquid biopsy EV nucleic acid biomarkers for glioma diagnosis, prognosis, and monitoring. Glioma subtypes are based on CNS WHO grades [89]. If the study included a detection and validation phase, only the size of the validation cohort is mentioned. DU=differential ultracentrifugation, UC=ultracentrifugation, DG-UC=density gradient and ultracentrifugation, SEC=size exclusion chromatography. Glioma grade EV source EV isolation Detection method Biomarkers Study population Potential application Reference 4 CSF UC qRT-PCR EGFRvIII Patient n=60 Diagnosis [29] 2–4 CSF UC dPCR IDH1 Patient n=14 Control n =2 Diagnosis [121] 4 CSF DU qRT-PCR miR-21 Patient n=15 Control n=16 Diagnosis [110] 2, 4 CSF DU qRT-PCR miR-21 Patient n=70 Control n=25 Diagnosis, prognosis [111] ? CSF DU miRNA sequencing miR-1246 Patient n=5–6 Diagnosis [114] 4 CSF, serum DU qRT-PCR miR-151a Patient n=14 Therapeutic effects [130] 1–4 Serum DU qRT-PCR miR-210 Patient n=91 Control n=50 Diagnosis, prognosis, surgical effects [113] 4 Serum DU qRT-PCR miR-1238 Patient n=26 Therapeutic effects [132] 2–4 Serum DU Flow cytometry, qRT-PCR EGFR, NLGN3, PTTG1 Patient n=23 Control n=12 Diagnosis [119] 2–4 Serum DU Fast Cold-PCR IDH1 Patient n=21 Diagnosis [27] 4 Serum SEC Deep sequencing miR-182–5p, miR-328–3p, miR-339–5p, miR340–5p, miR-485–3p, miR-486–5p, miR-543 Patient n=12 Control n=12 Diagnosis [115] 3, 4 Serum SEC dPCR circSMARCA5 and circHIPK3 Patient n=28 Control n =5 Diagnosis [117] 4 Serum SEC, immunoprecipitation qRT-PCR miR-15b-3p, miR-21–3p, miR-155–5p, let-7a5p, miR-106a-5p, miR-328–3p Patient n=55 Control n =5 Diagnosis, Prognosis [112] 4 Serum UC qRT-PCR miR-25–3p Patient n=67 Control n=15 Treatment response [131] 4 Serum Microfluidic platform qRT-PCR EPHA2, EGFR, MGMT, APNG Patient n =7 Diagnosis, therapeutic effects [118] 2–4 Serum Membrane affinity columns qRT-PCR miR-181-b/d Patient n=34 Control n=108 Prognosis [116] 4 Serum, plasma DU dPCR PD-L1 Patient n=21 Control n =5 Diagnosis [122] 3, 4 Plasma SEC qRT-PCR IL-8, TGF-β Patient n=20 Control n=10 Therapeutic effects [97] 3, 4 Plasma Membrane affinity columns dPCR EGFR, EGFRvIII Patient n=40 Control n=14 Diagnosis [120] E. Rackles et al. Seminars in Cancer Biology 87 (2022) 148–159 155 [116]. Like miRNA, circRNA is also involved in regulation of gene expression and can serve as biomarkers for glioma. circSMARCA5 and circHIPK3 are two circRNAs present in serum EVs, which are suitable to distinguish glioma patient samples from control samples. Of note, a combination of these circRNA biomarkers with inflammatory markers in GBM improved their diagnostic accuracy [117]. Two genetic alterations that are commonly present in GBM are wildtype EGFR (wtEGFR) amplification and EGFR variant (v)III mutation (EGFRvIII). It has been shown that the wtEGFR and EGFRvIII mRNA and protein are present in EVs [82,108,118,119]. The levels of EGFR mRNA in serum exosomes are heterogenous between samples but on average they are higher in GBM patients than in healthy individuals [118]. Similarly, the EGFR protein levels are increased in glioma patient EVs and allow to distinguish between glioma grades [119]. Interestingly, EGFRvIII in GBM can be diagnosed by the analysis of the RNA cargo of CSFand plasma-derived EVs with a sensitivity of 61% and a specificity of 98% and a sensitivity of 73% and a specificity of 98%, respectively [29,120]. Another genetic alteration in glioma is the mutation of isocitrate dehydrogenase (IDH) 1/2. According to the new WHO guidelines, glioblastoma is defined as glioma grade 4 that has the wild-type form of IDH [89]. Chen et al. aimed to identify the IDH mutation status in glioma patient EVs by digital PCR. They did not detect the IDH mutation in EVs isolated from serum. However, they successfully detected the mutation in five of eight EV preparations from the CSF of patients with confirmed IDH mutant tumor status and their analysis did not show any false positives [121]. In contrast, a different study detected the IDH mutation in EVs isolated from serum. The analysis of serum EVs with Fast Cold-PCR matched the mutation status detected in tissue samples of low-grade glioma patients. However, several false positives were detected for high grade glioma patients [27]. These studies show that the detection of the molecular status of glioma in EVs still needs to be improved. However, it should be kept in mind that also the analysis of the tissue sample could be wrong due to the high heterogeneity of glioma and the small size of the sample that is taken. The analysis of known factors in glioma proliferation and invasion can also provide potential new biomarkers. The mRNA levels of PTTG1, NLGN3 and EPHA2 are increased in serum EVs of glioma patients [118, 119]. While EVs of some patients also contained NLGN3 protein, the PTTG1 protein was not present [119]. The PD-L1 is a factor known to be involved in the immune suppression of GBM. PD-L1 is also present on EVs, which might represent an additional mechanism how GBM can evade the immune response. Similar to the above-mentioned factors, PD-L1 DNA was enriched in serum and plasma EVs of GBM patients while the protein was not detectable [122]. 2.3. The analysis of EV concentration as biomarker for glioma diagnosis and prognosis Not only the EV’s cargo but also the analysis of EV concentration in plasma samples of GBM patients was suggested as a biomarker. Studies isolated EVs by differential ultracentrifugation and found that the number of EVs is increased in plasma of GBM patients compared to healthy donors while the EV size of both groups kept similar [99,123]. Importantly, the EV concentration is only elevated in glioma patients and not in patients with other brain lesions [99]. Furthermore, comparison of EV concentration in plasma of patients before surgery, after surgery and at tumor relapse shows a correlation between EV number and tumor mass. The resection of the glioma leads to a decrease in EVs while they increase again when the tumor relapses [99,103]. However, molecular markers for glioma like wtEGFR amplification or IDH mutations do not correlate with EV concentration and measurement of EV concentration at glioma diagnosis does not provide information about overall or progression-free survival [99]. In contrast, Evans et al. analyzed microvesicles isolated by differential ultracentrifugation at the beginning of chemoradiation therapy, i.e., after surgery. They found that the microvesicle concentration can provide a prognosis of patient outcome. A decreasing number of Annexin V positive microvesicles measured by flow cytometry during the duration of the therapy is correlated with a better outcome for reoccurrence and overall survival [124]. Thus, EV concentration can provide information about the presence of glioma, surgical effects, and patient outcome. This observation has been challenged by recent studies showing no difference in EV concentration between GBM patients and healthy individuals before surgery [83,100,125]. The observed discrepancies might be due to the EV source (plasma vs. serum), isolation or quantification method and cohort size. 2.4. The analysis of EVs as biomarkers for glioma therapy monitoring GBM is an incurable tumor with a median survival of only 8 months [88]. Treatment of GBM consist of an initial of maximally safe surgical resection, which is followed by radiation therapy and concurrent Temozolomide (TMZ) chemotherapy [126]. The assessment on the tumor status by MRI after radiation therapy is especially challenging since pseudoprogression can occur [92]. Therefore, biomarkers for therapy monitoring are greatly demanded. In a preliminary study, Koch et al. addressed this problem by analyzing microvesicle concentration using flow cytometry and comparing them to MRI diagnosis. They found that patients who showed by MRI stable disease or pseudoprogression displayed lower microvesicle concentrations in plasma than patients with tumor progression [127]. This combination of imaging diagnosis with liquid biopsy could overcome the limitations of MRI in the discrimination between true tumor progression and pseudoprogression after treatment. Another study also measured EV concentration to monitor treatment efficiency. It combined the measurement of EV concentration and protein markers CD63, wtEGFR and EGFRvIII using a microfluidic platform. The obtained values were used to calculate a ‘treatment response index’ to discriminate between responders and non-responders to TMZ treatment, which was defined by MRI. An increase in the index measured before and after TMZ treatment indicates that the patients are not responding to the treatment [128]. Efficacy of TMZ in GBM patients is enhanced when the promoter of O 6 -methylguanine-DNA methyltransferase (MGMT) is methylated, which reduces the expression of the protein in cells [129]. Exosomes can reveal the MGMT methylation status of patients. A comparison of the DNA methylation status in primary GBM tissue and MGMT mRNA levels in exosomes isolated from serum revealed that mRNA levels are significantly higher in patients with negative methylation status, while the levels were similar between healthy individuals and GBM patients with positive methylation status. Thus, exosomes can serve as a non-invasive alternative to measuring promoter epigenetic methylation in GBM tissue-derived genomic DNA. Furthermore, preliminary data suggests that a longitudinal analysis of the change in MGMT and APNG expression might predict TMZ treatment outcome [118]. The analysis of miRNAs is another tool to predict TMZ treatment outcome. Loss of miR-151a in exosomes has been shown to transfer TMZ resistance to GBM cells in vitro. In this line, low levels of miR-151a in exosomes of GBM patients was correlated with poor response to TMZ treatment. Interestingly, only CSF derived exosomes were suitable for disease prognosis while no significant difference in miR-151a levels were found in serum exosomes [130]. Similarly, miR‑25–3p is overexpressed in exosomes of TMZ resistant cells in vitro and high miR-25–3p levels in serum of GBM patient is correlated with TMZ resistance and greater tumor size [131]. Furthermore, miR-1238 has been associated with TMZ resistance and miR-1238 levels are significantly higher in serum exosomes of patients with recurrent GBM compared to patients with primary GBM [132]. To summarize, the analysis of different cargos as well as the concentration of EVs isolated from liquid biopsy is a promising tool to monitor the response of glioma patients to chemoradiation therapy. In addition to the traditional treatment of glioma by chemoradiation, researchers work on the development of new strategies like the use of E. Rackles et al. Seminars in Cancer Biology 87 (2022) 148–159 156 tumor vaccines. To assess the immunological and clinical responses of patients to a new dendritic cells-based vaccine, Muller and colleagues assessed plasma derived exosomes of patients with recurrent malignant glioma enrolled in a phase I/II clinical trial. The analysis of total protein content did show a decrease between preand post-treatment samples, albeit not significantly. In contrast, exosomal mRNA analysis of selected genes revealed potential biomarkers for the treatment response. Changes in IL-8 and TGF-β expression showed a positive correlation to immunologic responses to the vaccine in the patient. Additionally, a weak correlation between IL-8 expression and overall survival as well as time to progression was shown [97]. The response of patients with recurrent glioma to an anti-survivin vaccine was evaluated by analysis of CD9 positive exosomes by flow cytometry. The surface marker survivin (SVN) was analyzed independently and together with GFAP protein. Glioma patients showed higher levels of SVN+/GFAP+/CD9 +exosomes than healthy individuals at study entry. After vaccination, some patients showed tumor progression sooner than other patients who displayed late, or no progression revealed by MRI. The latter group showed a decrease in serum CD9 +/SVN+and CD9 +/ GFAP+/SVN+exosomes immediately following the vaccination. In contrast, patients with early tumor progression had an increase in CD9 +/SVN+and CD9 +/ GFAP+/SVN+exosomes. Thus, the analysis of these biomarkers can predict therapy efficacy. Of note, one patient enrolled in the study displayed a detectable increase in CD9 +/GFAP+/SVN+exosomes already 16 weeks prior to the detection of progression by brain MRI scanning [133]. Taken together, these two studies suggest that vaccine-induced changes in the patients’ immune responses can be reflected by the EV’s cargo. 2.5. Limitations associated with applications of EVs in cancer As mentioned above, the isolation methods of EVs from liquid biopsies represent a challenge in cancer research due to their variety, lack of consensus and reproducibility, and possible low sample purity [15, 38]. In addition, other limitations due to the isolation and separation process have also been identified, including loss of EVs and alteration of their functionality and biological properties [134,135]. Therefore, those methods can disrupt EVs markers—for example, fragile branch glycans—hindering the EV-based glycan analysis, a promising potential area for the diagnosis of cancer [134–136]. Apart from that, isolating and separating EVs from biofluids allows to characterize EVs population—size and molecular profile—but not EVs biological environment—subcellular origin, biodistribution, release and uptake dynamics, half-life, and targeting mechanisms. That limits the understanding of EVs contribution to the pathophysiology of cancer and their clinical application as a biomarker [134]. To sum up, the understanding of the physiological and pathological roles of EVs in cancer faces multiple challenges and limitations, including the current deficient clinical validation and translation of research information into clinical practice, that should be further studied in the future. 3. Conclusion Liquid biopsies are a valuable source of EVs for the identification of minimally invasive molecular signatures as biomarkers for cancer. EV biomarkers allow to assess the status of tumors over time, which overcomes one of the greatest limitations of traditional tissue biopsy. Furthermore, analyzing different cargos of the EV gives a more detailed profile of the heterogenous tumor. These advantages of EV biomarkers show their great potential in the growing field of precision and personalized medicine (Table 3). Many potential candidates for biomarkers for glioma have been described. However, further work is needed to translate the basic research into clinical practice. The reviewed studies are very diverse. They use different body fluids of patients with different grades of glioma and varying cohort sizes. Furthermore, various methods for EV isolation and biomarker detection are applied. By this, a vast variety of different biomarkers has been identified. However, it is difficult to compare the studies and to identify reproducible biomarkers. Therefore, the biomarkers need to be confirmed in follow-up studies with larger cohorts. Additionally, it is necessary to standardize the protocols and identification methods. Especially the development of robust and easy tests is a prerequisite for the implementation of EV biomarkers in the clinical routine. Further research in collaboration with clinicians will overcome these issues and will advance the implementation of EV biomarkers in clinical routine. In future, EV biomarkers will be used to complement existing methodology leading to an improved personalized care of cancer patients. Data Availability No data was used for the research described in the article. Acknowledgements Authors would like to acknowledge the funding agencies supporting our research, mainly the EV-Glio ERAPerMed EU project (AECCPERME20733FALC; ISCIII-AC20/00024), the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation; 471840646), the Alzheimer’s Association Grant (AARG-NTF-22-968911), and the Spanish Ministry of Science and Innovation (RTI2018-094969-B-I00; PID2021-125104OB-I00). References [1] A.P. Becker, B.E. Sells, S.J. Haque, A. Chakravarti, Tumor heterogeneity in glioblastomas: from light microscopy to molecular pathology, Cancers 13 (4) (2021), https://doi.org/10.3390/cancers13040761. [2] S.N. Lone, S. Nisar, T. Masoodi, M. Singh, A. Rizwan, S. Hashem, et al., Liquid biopsy: a step closer to transform diagnosis, prognosis and future of cancer treatments, Mol. Cancer 21 (1) (2022) 79, https://doi.org/10.1186/s12943-02201543-7. [3] S. Gurung, D. Perocheau, L. Touramanidou, J. Baruteau, The exosome journey: from biogenesis to uptake and intracellular signalling, Cell Commun. Signal 19 (1) (2021) 47, https://doi.org/10.1186/s12964-021-00730-1. [4] S. Xie, Q. Zhang, L. Jiang, Current knowledge on exosome biogenesis, cargosorting mechanism and therapeutic implications, Membranes 12 (5) (2022) 498, https://doi.org/10.3390/membranes12050498. [5] J.W. Clancy, M. Schmidtmann, C. D’Souza-Schorey, The ins and outs of microvesicles, FASEB Bioadv. 3 (6) (2021) 399–406, https://doi.org/10.1096/ fba.2020-00127. [6] M. Battistelli, E. Falcieri, Apoptotic bodies: particular extracellular vesicles involved in intercellular communication, Biology 9 (1) (2020), https://doi.org/ 10.3390/biology9010021. [7] C. Ciardiello, R. Migliorino, A. Leone, A. Budillon, Large extracellular vesicles: Size matters in tumor progression, Cytokine Growth Factor Rev. 51 (2020) 69–74, https://doi.org/10.1016/j.cytogfr.2019.12.007. [8] R. Crescitelli, C. L¨ asser, T.G. Szab´ o, A. Kittel, M. Eldh, I. Dianzani, et al., Distinct RNA profiles in subpopulations of extracellular vesicles: apoptotic bodies, microvesicles and exosomes, J. Extracell. Vesic. (2013) 2, https://doi.org/ 10.3402/jev.v2i0.20677. [9] R. Crescitelli, C. L¨ asser, S.C. Jang, A. Cvjetkovic, C. Malmh¨ all, N. Karimi, et al., Subpopulations of extracellular vesicles from human metastatic melanoma tissue Table 3 Summary of the current advances and challenges in the application of EV biomarkers in cancer diagnosis and prognosis. Advances Challenges Easy and cost-effective isolation of EV biomarkers Need for standardization of isolation techniques EV biomarkers are present in various liquid biopsies Improvement of biomarker detection suitable for clinical routine Low-invasive biomarkers Confirmation of biomarkers in larger study populations Possibility of longitudinal diagnosis Possible application for diagnosis, prognosis, and treatment response E. Rackles et al.