The Role of Extracellular Vesicles in Disease Progression and Detection of Hepatocellular Carcinoma

Simple Summary Extracellular vesicles (EVs) are particles naturally released from cells and mediate intercellular communication. Recently, emerging studies have shown that EVs play a crucial role in regulating progression of hepatocellular carcinoma (HCC), which is one of the leading causes of cancer-related death worldwide. With the advances of technologies in isolating EVs from patients’ blood, EVs are regarded as promising biomarkers for detecting HCC at an earlier stage. This review provides an overview of the current EVs isolation methods, the biological roles of EVs in mediating disease progression, and the feasibility of EVs’ use for detection of HCC. Abstract Hepatocellular carcinoma (HCC) is the most common primary liver malignancy and one of the leading causes of cancer-related death worldwide. Despite the improvements in surveillance and treatment, the prognosis of HCC remains poor. Extracellular vesicles (EVs) are a heterogeneous group of phospholipid bilayer-enclosed particles circulating in the bloodstream and mediating intercellular communication. Emerging studies have shown that EVs play a crucial role in regulating the proliferation, immune escape, and metastasis of HCC. In addition, because EVs are present in the circulation at relatively early stages of disease, they are getting attention as an attractive biomarker for HCC detection. Over the past decade, dedicated efforts have been made to isolate EVs more efficiently and make them useful tools in different clinical settings. In this review article, we provide an overview of the EVs isolation methods and highlight the role of EVs as mediators in the pathogenesis and progression of HCC. Lastly, we summarize the potential applications of EVs in early-stage HCC detection.


Introduction
Hepatocellular carcinoma (HCC) is the most common primary liver malignancy (>80% cases) and ranks sixth for cancer incidence and third for cancer-related death worldwide [1,2]. Main risk factors for hepatocarcinogenesis include chronic hepatitis B virus By centrifugation at the speeds of 10,000-20,000 g and 100,000-120,000 g, large and small EVs are separated, respectively. Density gradient ultracentrifugation can be used for further EVs purification. (B) Filtration. Using a series of membrane filters with different pore diameters enables isolation of EVs with a specified size. (C) Size exclusion chromatography. The chromatography column consists of porous beads only allowing smaller particles, such as proteins, to enter. EVs are larger than the pore size; therefore, they migrate at a higher speed than the smaller particles and are isolated. (D) Precipitation. Precipitants occupy the solution and make less soluble particles, including EVs, exceed their solubility limit and precipitate. (E) Immunoaffinity capture. Antibodies targeting surface proteins on EVs are used to positively or negatively select specific subpopulations of EVs. (F) Microfluidics. For example, EV Click Chip: (1) the multimarker antibody cocktails enable targeting HCC-derived EVs, (2) nanostructured substrates increase the surface interacting with EVs, and (3) click chemistrymediated EVs capture (TCO/TZ interaction) and disulfide cleavage lead to DTT-driven EVs release, which results in isolation of HCC-derived EVs with high purity. ASGPR1, asialoglycoprotein receptor 1; DTT, 1,4-dithiothreitol; EpCAM, epithelial cell adhesion molecule; EVs, extracellular vesicles; HCC, hepatocellular carcinoma; TCO, trans-cyclooctene; Tz, tetrazine.

Ultracentrifugation
Ultracentrifugation is the most commonly used method for EVs isolation [15]. EVs are isolated by sequential centrifugations at increasing centrifugal forces, based on size and density. Using low-speed centrifugation, cells, platelets, apoptotic bodies, and cell debris are first eliminated from the EVs. Larger EVs and smaller EVs are then separated commonly at the speeds of 10,000-20,000 g and 100,000-120,000 g, respectively [16][17][18]. During ultracentrifugation, contaminants, including protein aggregates and lipoprotein particles, are also sedimented [17,18]. Additional steps of density gradient ultracentrifugation By centrifugation at the speeds of 10,000-20,000 g and 100,000-120,000 g, large and small EVs are separated, respectively. Density gradient ultracentrifugation can be used for further EVs purification. (B) Filtration. Using a series of membrane filters with different pore diameters enables isolation of EVs with a specified size. (C) Size exclusion chromatography. The chromatography column consists of porous beads only allowing smaller particles, such as proteins, to enter. EVs are larger than the pore size; therefore, they migrate at a higher speed than the smaller particles and are isolated. (D) Precipitation. Precipitants occupy the solution and make less soluble particles, including EVs, exceed their solubility limit and precipitate. (E) Immunoaffinity capture. Antibodies targeting surface proteins on EVs are used to positively or negatively select specific subpopulations of EVs. (F) Microfluidics. For example, EV Click Chip: (1) the multimarker antibody cocktails enable targeting HCC-derived EVs, (2) nanostructured substrates increase the surface interacting with EVs, and (3) click chemistry-mediated EVs capture (TCO/TZ interaction) and disulfide cleavage lead to DTT-driven EVs release, which results in isolation of HCC-derived EVs with high purity. ASGPR1, asialoglycoprotein receptor 1; DTT, 1,4-dithiothreitol; EpCAM, epithelial cell adhesion molecule; EVs, extracellular vesicles; HCC, hepatocellular carcinoma; TCO, trans-cyclooctene; Tz, tetrazine.

Ultracentrifugation
Ultracentrifugation is the most commonly used method for EVs isolation [15]. EVs are isolated by sequential centrifugations at increasing centrifugal forces, based on size and density. Using low-speed centrifugation, cells, platelets, apoptotic bodies, and cell debris are first eliminated from the EVs. Larger EVs and smaller EVs are then separated commonly at the speeds of 10,000-20,000 g and 100,000-120,000 g, respectively [16][17][18]. During ultracentrifugation, contaminants, including protein aggregates and lipoprotein particles, are also sedimented [17,18]. Additional steps of density gradient ultracentrifugation could be adopted to further purify the EVs [19][20][21]. However, ultracentrifugation is time- consuming and laborious and thus is unlikely to be applicable for routine use in clinical applications [16][17][18].

Filtration
Size-based separation of EVs from other non-EV particles can be achieved by filtration and size exclusion chromatography. The filter allows particles smaller than the pore to pass through while particles larger than the pore remain on the filter. Using a series of membrane filters with different pore diameters enables isolation of EVs fractions with specified sizes [22]. Filtration is amenable to clinical applications as it is less time-consuming and requires no special instrumentation. However, clogging and trapping of vesicles on the filter may lead to loss of EVs. Applying forces on particles to pass through filters can reduce the time of EVs isolation but may lead to deformation and breakup of large vesicles [23,24].

Size Exclusion Chromatography
Size exclusion chromatography, also known as gel filtration, separates molecules based on their size as they pass through a resin packed in a column [23,25]. The resin consists of porous beads with pores of a specific size that only allow smaller particles, such as proteins, to enter and thereby slow down their movement through the column due to increased path length. EVs are larger than the pore size and, therefore, flow through the column more quickly than soluble proteins. Therefore, the method can distinctly separate most protein contaminants from EVs [23,25]. Due to little pressure applied during the process, it preserves the structure, integrity, and biological functions of EVs [25][26][27][28]. However, there are some limitations: (1) The throughput is low, and (2) the resulting EV fraction is diluted and may require additional steps of concentration [25,29]. Some automated platforms have been developed to enable rapid and less laborious EVs isolation suitable for clinical application [30].

Precipitation
Precipitation of EVs in polyethylene glycol (PEG) solutions is, after ultracentrifugation, the second most common method for EVs isolation [31]. PEG is a superhydrophilic polymer, which occupies the water and concentrates other, less soluble, particles, including EVs, to the point of exceeding their solubility limit and precipitating [32]. The precipitate can then be pelleted by low-speed centrifugation [24]. Several commercial products have been developed based on precipitation to isolate EVs from biofluids, e.g., ExoQuick-TC™ Exosome Precipitation Solution (System Biosciences, Palo Alto, CA, USA), and Total Exosome Isolation Kit/Reagent (Invitrogen, Waltham, MA, USA). The method is quick, simple and allows for concurrent processing of numerous samples [31,32]. However, the major concern of this method is that non-EV proteins, such as lipoproteins, are coprecipitated along with EVs [33,34]. In addition, the presence of the polymer in purified EVs may interfere with down-stream analyses [33,34]. Therefore, this application is appropriate in samples thought to have sufficiently abundant EVs to reduce bias from contaminants. Additional steps before and after precipitation, including ultracentrifugation, filtration, or size exclusion chromatography, can be incorporated to reduce non-EV contaminants and PEG polymer [23,31].

Immunoaffinity Capture
Immunoaffinity capture isolates EVs based on the interaction between antibodies and surface proteins on EVs. Antibodies targeting surface proteins present on EVs are utilized to positively or negatively select specific subpopulations of EVs. These antibodies can be attached to magnetic beads [35][36][37] or microfluidic devices [38,39] (Section 3.6.), according to the desired downstream analysis. Therefore, the major benefit of this method is higher specificity and purity than those of the methods relying on physical properties [40,41]. However, selectivity can lead to lower yields [18]. Additionally, the cost of antibodies may be considerable.

Microfluidics
Microfluidic devices are designed as a network of microchannels to purify EVs by immunoaffinity and/or by physical characteristics [42]. Compared with traditional isolation methods, microfluidic-based isolation is more rapid (30 min-2 h), requires lower amounts of samples (50 µL-500 µL), and yields high-purity EVs in general [31,42]. However, some devices only allow low sample input and may therefore limit the applications of downstream analysis, such as massive sequencing [31,39,[42][43][44]. Recently, our research team developed a microfluidic device named EVClick Chip [38], which synergistically integrates (1) multimarker antibody cocktails for targeting HCC-derived EVs, (2) nanostructured substrates for increasing the surface interacting with circulating EVs, and (3) click chemistry-mediated EVs capture and disulfide cleavage-driven EVs release, to achieve purification of HCC-derived EVs with high recovery yield (82.7%) and excellent purity (90.2%). Most importantly, we demonstrated the potential application of EV Click Chip combined with downstream reverse-transcription droplet digital polymerase chain reaction (RT-ddPCR) analysis in detection of early-stage HCC.

EVs as Mediators of Chronic Liver Disease and HCC Progression
The majority of HCC diagnoses arise in patients with underlying cirrhosis, with up to 20% of HCC cases found in non-cirrhotic patients [2,45]. In a healthy liver, EVs are critical in mediating numerous signals among hepatocytes, stellate cells, and various immune cells (Kupffer cells, T and B cells, natural killer-NK-cells) to perform important functions and maintain a homeostatic state. Previous studies have demonstrated EVs play a role in the development of these predisposing liver diseases and, subsequently, in the development of HCC. EVs mediate the pathogenesis and progression process of HCC by regulating the microenvironment and multiple signaling pathways in both cancer and surrounding normal cells [46][47][48]. The studies investigating EVs' functions in HCC progression are summarized in Table 1 and

EVs in Promoting Progression of Chronic Liver Disease
The majority of HCC cases can be attributed to chronic viral hepatitis secondary to HBV and HCV. It has been shown that inter-and intra-cellular modulation via EVs in chronic hepatitis B and hepatitis C lead to viral propagation [49,50], fibrogenesis [51,52], and dysregulation of the immune system [50,53] through various mechanisms. For example, miR-19a in EVs derived from HCV-infected hepatocytes initiates fibrosis by activating surrounding hepatic stellate cells through the transforming growth factor beta (TGF-β) signaling pathway [51]. In addition, studies also indicated that HBV-infected hepatocytes produce EVs to suppress the expression of interleukin (IL)-21 in T cells [53] and downregulate nuclear factor kappa B (NF-κB) in NK cells [50]. This creates a microenvironment in the cirrhotic or chronically infected liver that is primed for genetic mutations and cellular dysregulation, giving the potential for the development of HCC.
There are numerous mechanisms by which EVs promote fibrosis and inflammatory processes in metabolic liver disease. At the cellular level, fatty liver disease is caused by hepatocyte dysfunction and death due to the toxic accumulation of intra-cellular lipids and the stimulation of inflammatory and fibrogenic pathways, with EVs playing a key role [54]. Hepatocytes exposed to elevated levels of lipid, such as palmitate, had increased secretion of EVs containing tumor necrosis factor (TNF)-related apoptosis inducing ligand [55], sphingosine 1-phosphate [56], miRNAs [57,58], C-X-C motif chemokine ligand 10 (CXCL10) [59], and ceramides [60,61], leading to activation of macrophages and chemotaxis. Lipotoxic conditions also release EVs containing integrin β1, which promotes monocyte adhesion to liver sinusoidal endothelial cells, resulting in inflammation and fibrosis [62]. These findings demonstrate the roles of EVs in the progression of metabolic liver disease through inflammation and fibrosis, ultimately leading to a liver microenvironment at risk for the development of HCC.
[55], sphingosine 1-phosphate [56], miRNAs [57,58], C-X-C motif chemokine ligand 10 (CXCL10) [59], and ceramides [60,61], leading to activation of macrophages and chemotaxis. Lipotoxic conditions also release EVs containing integrin β1, which promotes monocyte adhesion to liver sinusoidal endothelial cells, resulting in inflammation and fibrosis [62]. These findings demonstrate the roles of EVs in the progression of metabolic liver disease through inflammation and fibrosis, ultimately leading to a liver microenvironment at risk for the development of HCC.
Interestingly, Tian et al. first described that an acidic tumor microenvironment, attributed to the increased glycolysis in cancer cells [81], increases the levels of miR-10b and miR-21 in HCC-derived EVs compared with those produced at normal pH conditions [65]. In this study, miR-10b and miR-21 were proven to promote HCC proliferation and metastasis both in vitro and in vivo. These results highlight the role of EVs in tumor progression in response to the changing microenvironment.
EVs from surrounding stromal cells, such as tumor-associated macrophages (TAMs) [82] and cancer-associated fibroblasts (CAFs) [83], participate in the regulation of HCC progression. In a recent study, Wang et al. reported a significantly lower level of miR-125a/b in TAM-derived EVs. Subsequent functional studies showed miR-125a/b in TAM-derived EVs suppress proliferation and stem cell properties of HCC in vitro [82]. Additional studies indicated that compared with EVs derived from the para-cancer fibroblasts, the CAF-derived EVs from the same HCC patients had a significantly lower level of miR-320a [83]. miR-320a in these CAF-derived EVs acts as a suppressor of HCC proliferation and migration by directly downregulating the PBX3 oncogene.

EVs in Regulating Angiogenesis in HCC
In addition to tumor proliferation, EVs also modulate angiogenesis in HCC [84][85][86][87][88][89][90][91]. As a hypervascular tumor, HCC requires the formation of new blood vessels for growth. Revealing the mechanism of angiogenesis through EVs might thus help identify potential therapeutic targets to inhibit HCC progression. As proven by in vitro human umbilical vein endothelial cells (HUVECs) tube-formation assay, many molecular cargoes in EVs are key players in angiogenesis [84][85][86][87][88][89]. For example, lysyl oxidase like 4 (LOXL4) could promote angiogenesis and metastasis both in vitro and in vivo through activating the focal adhesion kinase (FAK)/Src pathway [84]. In cell line studies, miR-155, lncRNA-H19, and circRNA-100338 from HCC-derived EVs are associated with angiogenesis [85,87,88]. On the other hand, miR-200b-3p and miR-451a suppress angiogenesis by downregulating ERG and LPIN1, respectively [86,89]. Vascular endothelial growth factor (VEGF) is a signaling protein that directly induces the growth of hepatocytes, cancer cells, and epithelial cells and leads to abnormal vascular structures in HCC. Fu et al. demonstrated that in EVs secreted from a multidrug-resistant HCC cell line, Bel/5-FU, miR-32-5p was the most overexpressed miRNA [90]. miR-32-5p raised the level of VEGF in vitro and increased the microvascular density of xenograft tumors in vivo.
Previous studies have suggested the oncogenic roles of HCC-derived EVs in regulation of several immune cells [75,[93][94][95][96][97]. The 14-3-3ζ protein is highly expressed in HCC, and impairs the anti-tumor activity of tumor-infiltrating T cells via HCC-derived EVs [93]. TAMs are one of the immune cells crucial in creation of the immunosuppressive tumor microenvironment [106]. miR-146a-5p and lnc-TUC339 are enriched in HCC-derived EVs and proven to promote M2-polarization of TAMs [75,96], which can further result in T cell exhaustion [96]. In addition, Ye et al. found that the high mobility group box 1 (HMGB1) protein in EVs promoted T cell immunoglobulin and mucin domain 1 (TIM-1) + regulatory B cell expansion and suppressed CD8 + T cell proliferation as well [94]. Lastly, it was shown that circ-UHRF1 in HCC-derived EVs inhibited interferon gamma (IFN-γ) and TNF-α secretions from NK cells by suppressing miR-449c-5p and upregulating TIM-3 [97]. All these studies provide strong evidence indicating HCC-derived EVs could induce immune escape and promote metastasis.
Tumor intravasation represents a critical step for HCC metastasis and relies on the interaction of cancer and endothelial cells [107]. Of note, Fang et al. reported that miR-103 in HCC-derived EVs increased vascular permeability in vitro and in vivo by suppressing the expression of VE-cadherin, p120-catenin, and zonula occludens-1, which are endothelial adhesion molecules important in maintaining cell-cell junctions [101]. Interruption of the junction integrity eventually promotes liver and lung metastases. Similarly, miRNA-25-5p in HCC-derived EVs promotes trans-endothelial motility of HCC cells and causes tumor self-seeding in vivo [100]. Once cancer cells enter and survive in the circulation, attachment to the endothelial lining of microvasculature is essential for extravasation and consequent metastasis [108]. Fu et al. demonstrated SMAD family member 3 (SMAD3) in HCC-derived EVs-promoted adhesion of HCC cells to endothelial cells in vitro and observed a higher level of SMAD3 in EVs from patients with advanced stage HCC [99]. A non-coding RNA, lnc-H19, in EVs secreted from CD90 + HCC cells also facilitates adhesion of HCC cells to endothelium [87].

EVs as Biomarkers for Detection of HCC
Current tools for the detection and diagnosis of HCC include radiographic assessments (ultrasound for screening, computed tomography, and magnetic resonance imaging for diagnosis), and serum biomarkers (alpha-fetoprotein-AFP) [110]. Imaging techniques have limitations in identifying small tumors [111]. In cases of diagnostic uncertainty, invasive procedures, such as liver biopsy, may be necessary [110]. As a result of these limitations, development of novel diagnostic tools for the detection of HCC represents an unmet need. Currently available data suggest that EVs and their cargoes, including mRNAs, non-coding RNAs, and proteins, have the potential to serve as biomarkers for the detection of HCC (Table 2). These potential biomarkers can be isolated from plasma, serum, and, in some cases, bile.
Studies have demonstrated that measuring the amounts of EVs could be a strategy for the diagnosis of HCC [12,13]. By isolating EVs from peripheral blood using ultracentrifugation, Wang et al. found an increased quantity of EVs in HCC patients compared with those with liver cirrhosis [12]. The level of EVs was correlated to the tumor size and pathological classification of HCC and, most importantly, could be used to distinguishe early-stage (TNM Stage I) HCC from cirrhotic controls with an area under the curve (AUC) of 0.83 in receiver operating characteristic (ROC) analysis. Using the fluorescence-activated cell scanning, Julich-Haertel et al. identified a subgroup of EVs, the EpCAM + asialoglycoprotein receptor 1 + (ASGPR1) + EVs, which is capable of distinguishing HCC from cirrhosis with an AUC of 0.73 [13]. In both of these studies, blood samples after curative surgical treatment had significantly reduced levels of EVs, indicating the ability of these biomarkers to reflect tumor burden and monitor treatment response [12,13].

EV Protein for Detection of HCC
Proteomic analysis of EVs is a less explored avenue in the identification of novel biomarkers, with studies demonstrating different compositions of the proteomes of HCC, cirrhotic, and healthy control patients [112]. Arbelaiz et al. analyzed the EVs proteome profiles of intrahepatic cholangiocarcinoma (iCCA) and HCC and showed the differentially expressed proteins within these EVs could separate these two groups with an AUC of 0.89 [113]. This finding suggests EV-protein as a promising biomarker for characterizing an atypical intrahepatic lesion between HCC and iCCA. As one of the molecules mediating HCC metastasis, SMAD3, to has diagnostic power for HCC as well (AUC of 0.70 for distinguishing HCC from benign hepatoma and healthy controls) [99]. However, none of the EV-protein markers has been evaluated among at-risk patients with cirrhosis or chronic hepatitis B and patients with HCC, and further investigation is needed to assess their accuracy as a test for HCC surveillance.

EV miRNA for Detection of HCC
miRNAs in EVs have shown particular promise as biomarkers for the detection of HCC. Wang et al. examined the miRNA profile of EVs derived from HCC and cirrhotic patients and found that certain upregulated miRNAs (miR-122, miR-148a, and miR-1246) outperformered AFP in distinguishing HCC from cirrhosis. The final panel, comprising of miR-122, miR-148a, and AFP, resulted in an AUC of 0.93 [114]. It is noteworthy that the authors did not restrict the HCC cases to early-stage disease, thus likely overestimated the diagnostic power of the assay. In another study, Ghosh et al. identified four miRNAs, miR-10b-5p, miR-21-5p, miR-221-3p, and miR-223-3p, in liver-specific asialoglycoprotein receptor 2 + (ASGR2) + EVs for HCC diagnosis. The combination of these four miRNAs exhibited good diagnostic power among patients with low AFP expression (<250 ng/mL), with an AUC of 0.80 [115]. Although Sohn et al. showed that clusters of miRNAs (miR-18a, miR-221, miR-222, miR-224, miR-101, miR-106b, miR-122, and miR-195) were differentially expressed among patients with chronic hepatitis B, cirrhosis and patients with HCC, no further analysis using ROC was performed to determine their diagnostic performance [116].

EV lncRNA and EV circRNA for Detection of HCC
Along with the numerous promising miRNA biomarker targets, other non-coding RNAs packaged in EVs, such as lncRNAs, have shown promise in the early detection of HCC [117][118][119][120]. One study selected six upregulated lncRNAs from The Cancer Genome Atlas by comparing 371 HCC and 50 nontumor tissues and showed that LINC00853 in EVs was particularly promising for the identification of early-stage HCC [117]. Specifically, when setting a 14-fold increase as a cutoff for the expression of LINC00853 in EVs, it could discriminate patients with a single, <2cm HCC from those with chronic hepatitis or liver cirrhosis, with an AUC of 0.96 [117]. By combining three EV-derived lncRNA and AFP, Lu et al. established an HCC diagnostic model and validated it in an independent validation cohort [120]. In this large validation cohort (n = 360), they demonstrated the model could distinguish HCC from chronic hepatitis with an AUC of 0.87 [120]. Nevertheless, failure to restrict the HCC patients to those with an early-stage disease and failure to set noncirrhotic patients as the control group are the major limitations of this study.
EV-derived circRNAs are another example of potential biomarkers for HCC detection. One study identified three upregulated circRNAs, circ_0004001, circ_0004123, and circ_0075792, in EVs isolated from HCC patients compared with those from healthy donors [121]. These three circRNAs are associated with the VEGF, PI3K/Akt, mechanistic target of rapamycin (mTOR), and Wnt pathways and the authors found the combination of these circRNAs had a potential for detection of HCC with an AUC of 0.89 [121]. In 2019, Li et al. performed EVs' long RNA sequencing in five cancers including HCC to identify biomarkers for cancer diagnosis [122]. A diagnostic model containing eight long RNAs was built and validated in two cohorts to detect HCC from healthy donors and patients with unspecified benign hepatic disorders [122]. Despite the promising results with AUCs of 0.96 in the validation cohorts, there is still a concern about overestimating the diagnostic power considering the cohort composition.

EV mRNA for Detection of HCC
A study compared the potential of LDHC mRNA level in serum vs. serum-derived EVs for HCC detection. Interestingly, the authors demonstrated the superior ability of EV-derived LDHC mRNA to distinguish the TNM stage I/II HCC patients from healthy controls compared with serum-only LDHC mRNA, with an AUC of 0.95 vs. 0.84, respectively [123]. This difference might be attributed to the fact that the mRNAs in EVs are more stable than the circulating ones due to the protection by a phospholipid bilayer.
Together, all these studies demonstrate the promising application of EVs as biomarkers for the detection of HCC. It should be noted that almost all the studies enriched total EVs rather than those specifically secreted by HCC or hepatocytes. As the HCC-derived EVs represent a small portion of the total EVs, disease-specific changes in these potential EVs biomarkers may be difficult to detect, considering elevated background noise. To address this issue, our research team has developed a streamlined HCC EVs digital scoring assay [38] that couples two very powerful technologies, i.e., EV Click Chip for purification of HCC-derived EVs and RT-ddPCR for quantification of a panel of 10 HCC-specific mRNA markers. Benefiting from the nanostructured substrates, antibody cocktails-including anti-EpCAM, anti-ASGPR1, and anti-CD147-and click chemistry-mediated EVs capture/release process, EV Click Chip enables rapid and efficient purification of HCC-derived EVs. Most important of all, thanks to quantifying the 10 HCC-specific mRNA markers [124] in these purified EVs, the resulting HCC EVs digital scores exhibited promising potential for distinguishing the Barcelona Clinic liver cancer (BCLC) stage 0-A HCC from at-risk cirrhotic patients, with an AUC of 0.93 [38].

Conclusion and Future Direction
EVs play a crucial role in intercellular communication and mediate the pathogenesis, proliferation, immune escape, and metastasis of HCC [46][47][48]. As such, EVs are regarded as potential therapeutic agents or vehicles for HCC treatment [125]. With emerging studies in the field, the EVs' cargos, including functional proteins, non-coding RNAs, and mRNAs, are promising biomarkers for the detection of early-stage HCC [126]. In parallel, dedicated efforts have been made to isolate EVs more efficiently, to facilitate the adoption of this technology for clinical applications [31,127].
Despite these encouraging results, most of the studies on detecting HCC by using EVs are still in the preclinical phase and large prospective cohort studies are warranted to validate their diagnostic value. In addition, for case-control studies to accurately estimate the diagnostic performance of EVs, it is essential to restrict the cases to early-stage HCC and the controls to at-risk patients with liver cirrhosis or chronic hepatitis B, in line with the current clinical practice guidelines which define the at-risk population to whom it is recommended to undergo screening. With the guidance of a biomarker development framework provided by the International Liver Cancer Association [128], advents of more high-quality biomarker studies on EVs for detection of early-stage HCC are expected.

Conflicts of Interest:
H.-R.T. would like to disclose his financial interests in CytoLumina Technologies Corp. and Pulsar Therapeutics Corp., given his role as the co-founder of these companies. J.D.Y. provides a consulting service for Exact Sciences, Gilead, and Eisai. The funders had no role in the interpretation of data and in the writing of the manuscript.