Mass spectrometry‐based extracellular vesicle micromolecule detection in cancer biomarker discovery: An overview of metabolomics and lipidomics

Extracellular vesicles (EVs) become one of the most important sources of cancer biomarkers during the past decade due to their wide distribution in body fluids and physiological stability. Previous studies mainly focused on nucleic acids and proteins, while lipids and metabolites were largely neglected. Noticeably, many of those micromolecules exhibited a high abundance in EVs. Revealing the metabolomics and lipidomics of EVs would provide more comprehensive information for biomarker discovery. With the rapid development of mass spectrometry (MS) facilities, MS‐based micromolecule detection has become an emerging technique for EVs studies. Increasing evidence demonstrated the presence of EV‐associated metabolites and lipids in different types of samples (e.g., cell, urine, serum, stool), which exhibited promising performance in cancer diagnosis, prognosis, and prediction of treatment responses. This review aims to summarize advances in micromolecule profiling of EVs for cancer biomarker discovery, with an emphasis on MS‐based metabolomic and lipidomic analytical techniques. Challenges in this field, including the minimum sample quantity, normalization methods, and compound identifications, are also discussed along with possible solutions.

Studies before 2020 mainly focused on the proteins and nucleic acids of EVs, 17 while few studies have investigated micromolecules such as metabolites and lipids.Generally, metabolites and lipids are end products of cellular processes, making them more robust as biomarkers in clinical applications as compared to nucleic acids.Also, lipids and metabolites are important components of the EV membranes, 18 providing a relatively high abundance for its direct detection in biofluids.Mass spectrometry (MS) is the most important analytical technique to analyze micromolecules. 19,20During the past decade, MS-based metabolomics and lipidomics have been proposed and used in various clinical scenarios.2][23][24] Those advanced technologies largely promoted the development of EV-based cancer biomarker discovery.However, there remain a lot of challenges.The aim of this review is (1) to summarize the development of the MS technology and its clinical application, with emphases on methodology, clinical relevance, and deficiencies in MS-based metabolomic and lipidomic studies of EVs; (2) to highlight the main challenges in this field and propose potential solutions.

MS-BASED METABOLOMICS AND LIPIDOMICS IN EV STUDIES
The most common tools for metabolomics and lipidomics analysis are MS and nuclear magnetic resonance (NMR). 25uantitative MS has been widely adopted for omics studies due to its high speed and sensitivity in the classification and quantification of micromolecules by analyzing the mass-to-charge ratio (m/z) of ions. 26In this review, we mainly focus on MS-based methods since they have the advantage of higher speed and sensitivity, and can be combined with various chromatographic methods to measure a large diversity of compounds. 27Generally, the MS techniques used in metabolomics and lipidomics studies can be classified into two categories: separation-based MS and separation-free MS. 28 With the advantages of high separation ability, sensitivity, and accuracy, separation-based MS especially chromatography coupled to MS has become the most widely used technique applied in EV metabolomic and lipidomic studies. 29However, this method has several disadvantages, including the long detection time, cumbersome steps, and high cost. 302][33] Nevertheless, it is reported that much fewer fatty acids (FAs) are identified in the MALDI-MS compared with gas chromatography (GC)/MS analysis. 34herefore, the use of MALDI in the analysis of FAs is challenging.
GC-MS and liquid chromatography-MS (LC-MS) are the most widely used methods in separation-based MS techniques. 25As the preferred method for volatile substances analysis, GC-MS has the merits of low cost, easy-to-operate, and owning comprehensive databases. 35owever, for the detection of nonvolatile compounds, tedious sample derivatization is required.Meanwhile, LC-MS does not require the metabolites/lipids to be volatile 36 and has a broader detection range of analytes.The ultrahigh-performance LC (UPLC) developed in 2004 significantly improved the speed, resolution, and sensitivity of traditional LC methods. 37There are 20% more metabolites that can be detected by UPLC-MS compared with common LC-MS. 38The protocols for large-scale metabolic profiling from serum or plasma samples using GC-MS and UPLC-MS were proposed by Dunn et al. in 2011. 39Subsequently, scholars began to use these methods to identify potential metabolites or lipids in EVs as disease biomarkers, especially for human cancer.Among the 34 included EVs studies in this review, four, fourteen, and eight studies used GC-MS, LC-MS, and UPLC, respectively.Recently, great progress has been made in the development of stool metabolome based on GC-MS and LC-MS. 40Kim et al. adopted GC-MS for the metabolomic determination of stool EVs from patients with CRC and demonstrated gut flora dysbiosis may result in an alternation of amino acid (AA) metabolism. 41esides GC-MS and LC-MS, capillary electrophoresis-MS (CE-MS) is also an attractive chromatographic separation technique showing considerable potential for polar and charged molecule analyses. 42Hayasaka et al. employed CE-MS to analyze anionic metabolites of EVs derived from pancreatic cancer cells. 43However, those methods including GC-MS, LC-MS, and CE-MS fail to enable a comprehensive analysis of the metabolome, due to the large differences in physicochemical properties between metabolite classes.Supercritical fluid chromatography (SFC) represents a promising alternative approach to improve the metabolome and lipidome coverage, further increasing the analysis throughput. 44Some studies employed SFC to perform targeted quantitative lipidomic analyses in EVs derived from different cancer cells.The results demonstrated that the different lipid compositions existed not only between EVs and cells of their origin but also between high-and low-metastatic cell lines. 43,45eparation-free MS technique refers to MS analysis without prior chromatographic or electrophoretic separation, which mainly includes direct infusion-MS (DI-MS) and MALDI-MS. 28Anand et al. employed DI-MS to detect the serum lipidomic biomarkers for preeclampsia, providing a short-time and high-throughput method for lipidomic studies without the chromatographic separation process. 46However, DI-MS cannot distinguish the isomers and may be suffer from ion suppression.MALDI-MS is another powerful technique for separation-free MSbased metabolomics and lipidomics but is hindered by the significant fluctuations of the signal intensities of the analytes. 47Lobasso et al. first conducted a lipidomic analysis of EVs derived from melanoma cells by matrix-assisted laser desorption ionization-time-of-flight/mass spectrometry (MALDI-TOF/MS) in positive and negative ion modes and revealed potential lipidomic biomarkers of metastatic melanoma. 31ith the improvement of MS technologies and methodology development of EV studies, 48,49 considerable literature has accumulated on metabolomics and lipidomics of the EVs derived from cancer cells or clinically obtained biofluids.The advanced technologies for MSbased metabolomics and lipidomics of EVs in those studies are summarized in Figure 1.Most of these studies identified diagnostic biomarkers by collecting samples from cancer versus control, 21,41,[50][51][52][53][54][55][56] or pre-versus post-operation. 57[64][65]

EV METABOLOMICS IN HUMAN CANCERS
According to the Human Metabolome Database (HMDB), there are more than 8000 endogenous metabolites in the human body. 66Here, we classified the included studies based on different metabolite subcategories defined by HMDB and summarized related reports with a specific emphasis on their potential clinical application (Table 1).

AAs and derivatives
Many studies have reported changes in the levels of AAs and derivatives in human cancers, but a consensus on a shared AAs set for different malignancies is still absent.The first study using MS methods to explore the metabolomics of EVs from plasma and cell culture media was reported in 2016. 21Even though the standard MSbased identification of some AAs and derivatives is still absent, their presence in plasma EVs of cancer patients was widely reported. 21,50,67ysine, as one of the essential AAs in humans, is also an important growth factor for cancer cells.It has been proved that the reduction of L-lysine can inhibit tumor cell growth in vitro. 68Eylem et al. utilized GC-MS to investigate the metabolomics of CRC cell-derived EVs and found that lysine was significantly increased in EVs from CRC cells compared to noncancerous cells. 51Similarly, Kim and colleagues observed increased lysine in stool EVs from CRC patients. 41he tryptophan catabolism pathway is critical in tumor cell evasion of both innate and adaptive immune systems. 67,69Luo et al. compared the metabolic characteristics of large EVs (lEVs) and small EVs (sEVs) in malignant and tuberculous pleural effusion (PE) samples.In lEVs, tryptophan was more decreased in malignant PE than tuberculous PE. 23 A similar alteration trend of tryptophan was also found in serum-derived EVs from CRC patients, 51 whereas the EVs from glioblastoma (GBM) cells showed a significant increase in tryptophan concentration. 52ecently, it has been demonstrated that proline metabolism plays an important role in the process of metabolic reprogramming, which may influence the progression of cancer. 70Palviainen et al. demonstrated that proline was elevated in the EVs from three cancer types (PCa, cutaneous T-cell lymphoma, and colon cancer). 50ifferently, Eylem et al. reported that proline was significantly decreased in both CRC patients' serum and CRC cell supernatant compared to noncancerous controls. 51oreover, acetylglycine, an immunoregulation-associated AA, is significantly higher in EVs from PCa patients relative to the healthy controls. 71esides those proteinogenic AAs, the nonproteinogenic AA ornithine also plays an essential role in cancer cell activity as well as tumor development. 72For instance, accumulation of ornithine was observed in cancer-related EVs (upregulated in colon cancer urine and plasma 57 ; downregulated in malignant PE 23 ) in contrast to those of healthy controls.

Organic acids
Among all detected metabolite categories in serum EVs from esophageal squamous cell carcinoma (ESCC) patients, organic acids (OAs) and their derivatives were the most abundant, which accounts for 16.41%. 59As a particular group of metabolites, OAs consist of many intermediate  metabolites of critical metabolic pathways, such as the tricarboxylic acid cycle (TCA) and glycolysis. 73he status of the TCA cycle in cancer cells is different from normal cells. 744][65] For instance, a higher concentration of succinic acids was detected in both serum-derived EVs from breast cancer patients 65 and stool-derived EVs from CRC patients. 41As for cell-derived EVs, Palviainen et al. reported an elevated level of succinic acids in cancer patients. 50Contrary to the trend of succinic acids, citric acid was found to be downregulated in serum-derived EVs from head and neck cancer (HNC) patients. 64n addition to the changes in TCA, cancer cells generally exhibit elevated glycolysis for adenosine triphosphate generation, that is, Warburg effect. 75Lactic acid, as the end product of glycolysis, can suppress the activity of human cytotoxic T lymphocytes and accelerate the growth of cancer cells. 76Joshi et al. has focused on the relationship between EV metabolomics and the response of neoadjuvant chemotherapy (NAC) in patients with breast cancer.They found that L-lactic acid level was significantly upregulated in plasma-derived EVs of patients with residual disease compared with those with pathologic complete response after NAC. 65Moreover, glycolic acid, a by-product of glycolysis, was also found to be increased in cell-derived EVs from CRC patients. 51tudies also reported other differentially accumulated OAs including monomethyl phosphate and several microbe-associated OAs in EVs from CRC patients' stool or serum. 41,51For instance, Kim and colleagues found that among all detected OAs, only aminoisobutyric acid (a microbial metabolite) was downregulated while others were all upregulated. 41Although their study connected the gut flora to CRC through microbe-associated metabolites, further functional studies are still needed to investigate the effect of bacteria on cancer progression.

Nucleotides and derivatives
Nucleotide metabolism is one of the key pathways that generate purine and pyrimidine molecules for DNA replication, RNA synthesis, and cellular bioenergetics. 77ncreased nucleotide metabolism facilitates the uncontrolled growth of tumors, which is considered a hallmark of cancer. 77Several lines of evidence suggest that the nucleotides in EVs hold great potential for cancer diagnosis, prognosis, and individualized treatment strategies.Hayasaka et al. observed a massive accumulation of intermediates related to nucleotide metabolism in sEVs from pancreatic cancer cells.They detected increased levels of inosine, guanosine, hypoxanthine, and xanthine (which were involved in purine metabolism), as well as high concentrations of cytidine and uridine (which were involved in pyrimidine metabolism).Moreover, inosine, uridine, and cytidine were among the top 20 metabolites with the highest amounts in sEVs.Nevertheless, these findings might not be sufficiently convincing due to the lack of noncancerous control cells. 43hu and colleagues compared the metabolites of plasma-derived EVs from recurrent ESCC patients and those from nonrecurrent patients and healthy controls.Their results showed that 3′-uridine monophosphate (3′-UMP) was the most important biomarker for distinguishing the recurrent group and nonrecurrent group with the highest area under the curve (AUC) of 0.90.This study is the first to reveal the potential relationship between 3′-UMP and ESCC. 59As for metabolites in pyrimidine metabolism, Liu et al. observed low levels of dihydrothymine in EVs from PCa patients compared to noncancerous controls and proposed that PCa may prefer to maintain high levels of dihydrothymine in internal tumor cells instead of releasing them via EVs. 71dditionally, two other EV metabolomic studies based on the lung cancer population also demonstrated changes in metabolites related to nucleotide metabolism in EVs from urine or PE. 23,62To be noted, the group sets in a study by Hinzman et al. were preradiation therapy (RT) and post-RT. 62Therefore, the changes in xanthylic acid, imidazole, and dTTP may implicate signs of DNA damage and/or impaired DNA synthesis upon irradiation.Their study suggests that the nucleotide-related metabolites in EVs from body liquids are not only sensitive biomarkers for early detection of radiation damage but also valuable for monitoring the response to RT. 62

Carbohydrates and derivatives
Despite the widely reported association between carbohydrate consumption and cancer risk, 78 57 However, it should be noted that these changes were only detected from EVs by normalization to EV-derived factors or with metabolite ratios, rather than from the original urine samples. 57
Many studies reported the identification of alcohol metabolites in serum EVs of both CRC patients 51,58,63 and HNC patients. 64Additionally, ethanolamine, once proposed as a probe for cancer detection, 80 was significantly upregulated not only in stool-derived EVs but also in serum-derived EVs of CRC patients compared to noncancerous controls. 41,51Strybel et al. revealed that 1,4-dithiothreitol was the only metabolite showing great performance (AUC = 0.95, accuracy = 75%) for discriminating rectal cancer patients with different responses to the neo-RT. 63These studies suggest that alcohol metabolite appears to be a potential biomarker for both diagnosis and prognosis in CRC patients.

EV LIPIDOMICS IN HUMAN CANCERS
Lipids are also important molecules that support the growth of cancer cells.According to the LIPID MAPS Structure Database, 81 there are more than 45,000 unique lipid structures in human bodies.Many studies have observed the changes in EV lipidomics in different cancers.Here, we summarize the studies focusing on the 5 most abundant lipids in humans according to the LIPID MAPS comprehensive classification system for lipids, 82,83 which include FAs, glycerolipids (GLs), glycerophospholipids (GPs), sphingolipids (SLs), and sterol lipids (Table 2).

FAs and derivatives
FAs are widely recognized to be critical to cancer cells since they can maintain membrane biosynthesis during rapid proliferation and provide an important energy source under metabolic stress. 84Several studies have demonstrated FAs in EVs as potential biomarkers for cancer diagnosis and prognosis. 41,51,55,59,60For instance, 1-hydroxy FAs were more enriched in serum-derived EVs from hepatocellular carcinoma (HCC) patients compared with those from non-HCC controls and exhibited strong predictive value in logistic regression analysis. 55almitoleic acid, palmitaldehyde, and isobutyl decanoate were selected as potential marker panels for distinguishing recurrent and nonrecurrent patients with ESCC with an AUC of 0.98. 59Moreover, Paolino et al. showed that FAs in CD81-positive plasma-derived EVs from early-stage and late-stage patients were helpful for the diagnosis of different stages of melanoma and the prediction of metastatic behavior. 60Kim et al. determined four FAs, including short-chain to long-chain FAs, that altered in stool-derived EVs from CRC patients.Among them, butanoic acid was the only downregulated one, which might be explained by its close association with microbiota. 41esides, some studies focused on the role of FAs in the radiation effect. 62,63Hinzman et al. found that the FAs in urine EVs were upregulated 90 days postirradiation compared to sham irradiated rats.Furthermore, they performed a pilot study on five patients with lung cancer pre-and post-RT and also observed lipids changes associated with radiation exposure.Therefore, they believed that these data could be extrapolated for the relationship between EV lipids and tumor response to RT, once followup data became available. 62Coincidentally, Strybel et al. compared the lipidomics of serum EVs between "good responders" and "poor responders" in patients treated with RT due to rectal cancer.FAs were found to be more enriched in EVs of good responders than poor ones, revealing the potency of FAs for predicting the response to RT in different cancers. 63

GLs
GLs are necessary for membrane formation, caloric storage, and crucial intracellular signaling processes, which are mainly composed of mono-, di-, and tri-substituted glycerols.Del Boccio et al. reported that mono-, di-, and triacylglycerols (MAG/DAG/TAG) were significantly altered in renal cell carcinoma urinary EVs, accounting for 11% of the total significantly accumulated lipids. 85any studies revealed the alteration of DAGs in cancerrelated EVs.For example, Nishida-Aoki et al. demonstrated that EVs from high-metastatic breast cancer cells enriched more unsaturated DAGs compared to those from low-metastatic cells. 45Moreover, the EVs enriched with DAGs could activate the protein kinase D signaling pathway in endothelial cells, leading to stimulated angiogenesis. 45,86Since DAG-mediated protein kinase C activation occurs in many other cancer-related functions, such as cell proliferation and immune response, DAGs in cancer EVs might facilitate EV-mediated education of the surrounding cells to support tumor progression. 45Similarly, Altadill et al. found that DAG was upregulated in EVs from transforming growth factor-beta treated PANC1 cells than control cells. 21Nonetheless, ovarian cancer cellderived EVs contained lower levels of DAG than ovarian surface epithelial cell-derived EVs. 87Further studies are needed to investigate the mechanism of DAGs in EVs on tumorigenesis and metastasis.

GPs and SLs
][92] For instance, Elmallah et al. found the dysregulation of PC in plasma EVs and CRC cell-derived EVs.PC 34:1 was markedly increased in plasma EVs from the nonmetastatic CRC patients and significantly decreased in the metastatic CRC patients. 61Similarly, Bestard-Escalas and coworkers observed a higher abundance of PC in plasma-derived EVs from patients with adenomatous polyps or hereditary CRC, but not in patients with invasive neoplasia. 90Lysophosphatidylcholine (LysoPC) is a class of chemical compounds that is derived from PC. Buentzel et al. revealed a significantly elevated LysoPC in plasma EVs of breast cancer compared to controls and further demonstrated the association of high concentrations of LysoPCaC 26:0 with shorter overall survival. 53Another study showed that the LysoPC 22:0 in serum-derived EVs was associated with tumor stage, CA199, CA242, and tumor diameter in pancreatic cancer patients. 24s a phospholipid highly expressed on the surface of cancer cells, 93 PS is significantly dysregulated in EVs derived from urine 56 and plasma 55 of different cancer patients, and the supernatant of cancer cells. 87,94,95Moreover, cardiolipin is specially enriched in both serum EVs of HCC patients and the supernatant of HCC cells. 55,96Other GPs such as PE 24,61,85,90 and PI 87,90 were also reported as potential EV-associated biomarkers for cancer diagnosis and prognosis.

TA B L E 2 MS-based lipidomic studies of EVs for biomarker discovery in human cancer
Sphingomyelin (SM) is the most abundant sphingolipid in cell membranes.Studies have demonstrated that SM was significantly dysregulated in cell-derived EVs of CRC, 58,94 PCa, 54,95 GBM, 96 and the body liquids-derived EVs of endometrial cancer 21 and lung cancer. 23Ceramide is the central molecule in SM metabolism and plays an important role in cancer biology. 97A large amount of evidence has shown that the concentrations of ceramide were significantly changed in cancer-related EVs compared to controls. 56,61,87,91,98All of these studies together revealed the great value of GPs and SLs as EV-associated biomarkers for early cancer detection and metastasis surveillance.

Sterol lipids
Sterol lipids play an important role in the generation and progression of human cancers. 99Previous studies have compared the lipid profiles between cell-derived EVs and the parental cells of different cancers such as PCa and CRC.They revealed that the relative amounts of sterol lipids were significantly higher in cell-derived EVs than in the parental cells. 94,95Although the detection of high levels of sterol lipids in EVs might be due to the presence of contaminating lipoproteins or lipid droplets in EV isolates, 100 several studies have also observed significant differences in these lipids between cancer-related EVs and noncancerous controls. 22,54,87,91o identify lipidomic biomarkers for PCa, Brzozowski and colleagues compared the EVs lipidomics of tumorigenic cells, metastatic cells, and nontumorigenic cells.Sterol lipids, such as cholesterol ester (ChE), were more highly abundant in EVs derived from metastatic cells than in nontumorigenic cells. 54Similarly, Clos-Garcia et al. detected significantly elevated levels of steroid hormones in urine-derived EVs from PCa patients than in benign prostate hyperplasia patients.Since systemic metabolism status largely affects the analysis results, they further explored the correlations between these lipids and body mass index but failed to find any significant correlations. 22esides PCa, the alteration of sterol lipids in cancerrelated EVs has also been observed in other cancers.For example, Cheng et al. reported that ChE was more abundant in EVs from ovarian cancer cells than in ovarian surface epithelial cells. 87Smolarz et al. detected elevated levels of ChE in serum-derived EVs from lung cancer patients compared to healthy controls. 91In contrast, Nishida-Aoki et al. observed a lower level of ChE in highmetastatic breast cancer cell lines than in low-metastatic cell lines. 45Noticeably, most of the lipidomic studies above have highlighted the EV isolation methodology in their studies since it is essential to separate EVs from contaminations such as low-density lipoprotein or very low-density lipoprotein. 91,94Nevertheless, there is still a lack of gold standards regarding this issue.

Future perspectives
Several challenges should be addressed to optimize the quality of MS-based metabolomics and lipidomics EV studies.First, the optimal quantity of sample sources required for the metabolomic and lipidomic analyses should be determined.It has been reported that 10 ml of urine or 500 μL of plasma might be sufficient for generating high-quality MS data for profiling the metabolomic and lipidomic content of the EVs. 21,57Besides the common body liquids including urine and plasma, recent evidence provides new insight into EV sources such as stool and PE. 23,41Therefore, more studies are needed to develop methodological standards for different sample sources.Additionally, the great heterogeneity in isolation and characterization methods of EVs may lead to incomparability among different studies.Some experts claimed that size-exclusion chromatography is the preferred isolation method for MS analysis of urine EV. 62 Nevertheless, most studies still adopted ultracentrifugation for its great performance in discriminating EVs from nanoparticles of similar size.For the characterization of EVs, the International Society for Extracellular Vesicles (ISEV) recommends that surface biomarkers should include at least three positive protein markers (at least one transmembrane protein and cytosolic protein), and one negative protein marker. 51,96 Relative quantification of metabolites or lipids and the selection of an internal reference for normalization are also tricky.Normalization to EV-derived parameters such as CD9, metabolite ratios, 57 or lipid-to-protein ratio 94 might be a useful approach to reveal cancer-related alterations in the EV metabolomic and lipidomic profiles.Moreover, since cholesterol was abundant in all samples and exhibited no significant differences among the different EVs, it might also be a potential internal reference for EVs lipidomic analysis. 61Besides normalization, compound identification has always presented as a big challenge owing to the complexity and instability of metabolites and lipids.Observing the changes in a class of molecules rather than each molecule may help us find principles from complex MS data.For instance, the ratio Σ34:1containing species/Σ38:4-containing species was demonstrated to be potentially discriminative between healthy and pathological patients. 90As for the structure complexity caused by isomerides, other techniques such as isotope labeling and NMR may be helpful to provide further information. 52,101o deal with the instability, more efforts should be paid to the storage and preservation of EVs and timely detection.
Lastly, although several studies have demonstrated differences in metabolomics and lipidomics between EVs and their parental cells or bioliquids, 23,43,55,58,94,95,102 the superiority of EVs over their parental cells or bioliquids in biomarker discovery efficacy is rarely confirmed. 56molarz et al. reported that the performance of the serumderived EV lipid-based classification model was comparable to a similar model based on directly quantified serum metabolites. 91It should be noted that there are many factors affecting the metabolomics and lipidomics of EVs such as human diets and cell culture conditions. 41,103Therefore, results in this field should be interpreted with caution.In our opinion, the development of MS-based metabolomic and lipidomic analytical techniques with higher precision and less affected by the biochemical condition of human bodies will be crucial to establish the advantage of EV metabolomics and lipidomics over ordinary MS profiling on serum micromolecules.

CONCLUSIONS
Micromolecules such as metabolites and lipids in EVs have received great attention owing to their potential clinical application as diagnostic or prognostic biomarkers, and as targets for cancer therapies.Here, we provide an overview of the application of MS-based metabolomics and lipidomics in EV biomarker discovery for different cancers.Generally, most of the recent studies have demon-strated significantly altered levels of specific metabolites or lipids in cancer-related EVs derived from different types of samples, which exhibit promising performance in cancer diagnosis, prognosis, and prediction of treatment responses.However, despite the great progress in MS-based metabolomics and lipidomics on EVs, research in this field is still incipient, with an absence of methodology standards and clinical translations.Therefore, we believe that with the improvement of methodology in EV purification and micromolecules identification, MS-based metabolomics and lipidomics will play a crucial role in EV-based prognostic and diagnostic cancer biomarkers discovery.

A C K N O W L E D G M E N T S
This work was supported by grants from the Young Elite Scientist Sponsorship Program by China Association for Science and Technology (No. YESS20220494, 2022QNRC001), the National Natural Science Foundation of China (grant number: 82073390), and the Beijing Municipal Science and Technology Project (grant number: Z191100006619081).The study sponsors had no role in the design of the study and the preparation of this manuscript.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no conflict of interest.