Extracellular vesicles as potential biomarkers and treatment options for liver failure: A systematic review up to March 2022

Introduction Extracellular vesicles (EVs) carrying functional cargoes are emerging as biomarkers and treatment strategies in multiple liver diseases. Nevertheless, the potential of EVs in liver failure remains indistinct. In this systematic review, we comprehensively analyzed the potential of EVs as biomarkers of liver failure and the therapeutic effects and possible mechanisms of EVs for liver failure. Methods We conducted a systematic review by comprehensively searching the following electronic databases: PubMed, Web of Science, Embase and Cochrane Central Register of Controlled Trials from inception to March 2022. The used text words (synonyms and word variations) and database-specific subject headings included “Extracellular Vesicles”, “Exosomes”, “Liver Failure”, “Liver Injury”, etc. Results A total of 1479 studies were identified. After removing 680 duplicate studies and 742 irrelevant studies, 57 studies were finally retained and analyzed. Fourteen studies revealed EVs with functional cargoes could be used to make the diagnosis of liver failure and provide clues for early warning and prognostic assessment of patients with liver failure. Forty-three studies confirmed the administration of EVs from different sources alleviated hepatic damage and improved survival through inhibiting inflammatory response, oxidative stress as well as apoptosis or promoting hepatocyte regeneration and autophagy. Conclusions EVs and their cargoes can be used not only as superior biomarkers of early warning, early diagnosis and prognostic assessments for liver failure, but also as potentially effective treatment options for liver failure. In the future, large-scale studies are urgently needed to verify the diagnostic, predictive and therapeutic value of EVs for liver failure.


Introduction
Liver failure is a severe liver disease syndrome caused by multiple precipitating factors, which accompanies by grievous liver dysfunction or decompensation. The typical clinical manifestations of this syndrome include jaundice, coagulation dysfunction, ascites, hepatorenal syndrome, and hepatic encephalopathy (1,2). At present, effective treatment for liver failure is lacking except for liver transplantation, although it still faces challenges of graft rejection, high cost and donor shortage. The transplant-free survival in patients with liver failure is rather low, usually less than 50% (3)(4)(5)(6). Especially, diverse etiologies, complex clinical manifestations, undefined pathogenesis, high mortality and lack of effective treatments make it more difficult to accurately diagnose and properly treat patients with liver failure. In this context, it is extremely important to seek candidate biomarkers for early warning, early diagnosis and prognostic assessment of liver failure (7,8). On the other hand, exploring potentially effective therapies for liver failure is also essential for improving survival rate, which is the scientists have been working on all the time (9)(10)(11).
Extracellular vesicles (EVs) are nanoscale vesicles with proteolipid bilayers. They can be secreted by almost all cells into the extracellular milieu, which makes them widely distributed in various biological fluids, such as blood, urine, milk, sputum, ascites, and cerebrospinal fluid (12). EVs are broadly divided into two main subgroups depending on their biogenesis: exosomes (Exos, 50-150 nm in diameter) and microvesicles or microparticles (MVs or MPs, 50-500 nm in diameter, up to 1000 nm) (13, 14). EVs have been considered metabolic waste. Over the past decade, scientists have demonstrated that EVs contain various bioactive substances or signal transduction molecules that exert crucial roles in homeostasis maintenance, antigen presentation, gene regulation, and so on (15,16). Considering that EVs can reflect the pathophysiological state of the cells from which they are derived, EVs have the great potential to emerge as non-invasive biomarkers for early diagnosis and prognosis assessment of liver diseases (17-21). In addition, EVs possess the following advantages: stable m e m b r a n e s t r u c t u r e , l o w i m m u n o g e n i c i t y , g o o d histocompatibility, easy chemical and genetic programming. Therefore, it is feasible to utilize EVs as drugs or drug carriers to treat refractory diseases such as cancer, neurological and cardiovascular disorders (22-24).
So far, there isn't a universal ideal animal model for liver failure in view of multiple etiologies and complex mechanisms related to this syndrome (25, 26). Currently available animal models of liver failure have their strengths and weaknesses. Animal models of acute liver failure (ALF) enrolled in the present systematic review mainly involve hepatotoxic drug-induced, surgically induced, and mixed models. Virus-induced ALF models are rare and unsatisfactory (27). The commonly used drugs and chemical reagents for ALF induction include D-galactosamine (D-GalN) with or without lipopolysaccharide (LPS) or tumor necrosis factor-a (TNF-a), acetaminophen (APAP), carbon tetrachloride (CCl 4 ), concanavalin A (Con A) and thioacetamide (TAA) (28, 29). The surgical ALF models can be divided into hepatectomy (total or partial), devascularization (total or partial) and a combination of both (30,31). Hepatic ischemia-reperfusion injury (IRI), as an inevitable local sterile inflammatory response following surgery, is one of the main causes of early organ dysfunction and failure after liver transplantation (32). And hepatic IRI model accounts for a large proportion of surgically induced ALF models. Animal models of acute-on-chronic liver failure (ACLF) are usually induced by acute insult [such as LPS and/or D-GaIN, or ethyl alcohol (EtOH)] in the setting of chronic liver injury (for example, CCl 4 or bile duct ligation) (33).
Till now, studies on EVs as biomarkers or therapeutic options for liver failure are still in the infancy. This systematic review summarized the current available studies including preclinical animal studies and clinical studies in this field, and hopes to provide novel ideas and directions for the early warning, diagnosis, treatment and prognostic assessment of liver failure.

Methods
Our systematic review was prepared according to the Cochrane recommendations for study methodology and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (34).

Literature search strategy
The search strategy was developed by two informatics specialists. We comprehensively and systematically searched the following electronic databases: PubMed, Web of Science, Embase, and Cochrane Central Register of Controlled Trials from inception to March 2022. The used text words (synonyms and word variations) and database-specific subject headings included the followings: "Extracellular Vesicles", "Exosome", "Liver Failure", "Liver Injury", and so on. The search strategy for PubMed was summarized in Table S1, and it was also applicable to all other databases.

Eligibility criteria
We enrolled the studies according to the following criteria: (1) studies on EVs or modified EVs. (2) studies on preclinical or clinical studies of liver failure. (3) studies on EVs as potential biomarkers and treatment options for liver failure.
We excluded the studies according to the following criteria: (1) studies unrelated to EVs. (2) studies on the application of EVs in diseases other than liver failure and liver injury we described in the Introduction. (3) conference abstracts, reviews, mechanism studies, non-English articles and in vitro studies.

Data extraction
The following data were extracted from each study:  experimental model, EV source, EV separation techniques, EV  administration including dosage and route, EV cargoes, EV  functions and others (See Tables 1-6 for details). To select articles potentially eligible for inclusion, two authors independently screened the title and abstract of each article, then reviewed the full texts of all retaining studies and analyzed the related information. The divergence between these two authors was judged by the corresponding author. We did not conduct a metaanalysis due to the limited number of available randomized  (43) controlled trials. All included studies recruited patients that provided informed consent before enrollment.

Quality assessment
We used the Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE) tool to evaluate the risk of bias for preclinical studies (94). And the risk of bias graph was drawn using the RevMan 5.3.2 software provided by the Cochrane Collaboration Network. The risk of bias for clinical studies was assessed using the Newcastle-Ottawa scale (NOS). The Confidence in the Evidence from Reviews of Qualitative research (CERQual) tool was used to assess the evidence quality of outcomes in this systematic review (95). The quality assessment was done independently by two authors and the divergence between these two authors was judged by the corresponding author. A PRISMA figure was created.

Literature selection
A total of 1479 articles were identified using our search strategy. After removing 680 duplicates, 799 articles were submitted to the title, abstract or full-text assessment by two independent authors. Among these, 742 irrelevant articles were further excluded, including conference abstracts, reviews, studies unrelated to EVs, EVs in other diseases, mechanism research, non-English articles and in vitro studies. Finally, 57 articles were retained and analyzed in this systematic review ( Figure 1).

EVs as potential biomarkers for early warning, diagnosis and prognostic assessment of liver failure
We summarized the studies on EVs as potential biomarkers of liver failure. Tables 1, 2 listed the relevant preclinical and clinical studies, respectively. This review revealed that EVs not only could    be utilized as the early warning or diagnosis markers of liver failure, but also as biomarkers for prognostic assessment of this syndrome.
3.2.1 EVs as potential biomarkers for early warning or diagnosis of liver failure

Preclinical studies
Conde-Vancells J et al. investigated the proteome of urinary vesicles derived from D-GalN-treated rats and attempted to identify potential biomarkers for acute liver injury (ALI). They found several proteins normally present in urinary vesicles including CD26, SLC3A1 and CD81 were dramatically reduced in urinary samples obtained from D-GalN-treated rats. And the authors believed that these three proteins had the potential to be used as candidate non-invasive urinary indicators of acute liver damage (35). CD133 and CD39 are expressed by hematopoietic stem cells (HSCs) and mobilized after liver injury. In APAP-induced experimental ALI, HSC and plasma CD133 + MPs levels were increased in a CD39-dependent manner. And differentially increased plasma CD39 + CD133 + MPs were helpful for monitoring critically ill patients with hepatic dysfunction and identifying patients who urgently needed for liver transplantation (36). In IRI, platelet-and neutrophil-derived MPs were acutely elevated following injury, and could serve as markers of inflammatory injury. In contrast, MPs derived from endothelial cells increased after injury response during the reparative phase, suggesting angiogenesis in the regenerating liver. Hence, MPs might be regarded as markers of acute inflammatory injury or regeneration in IRI (37). Eva Rodrıǵuez-Suaŕez et al. analyzed the proteome of EVs derived from primary hepatocytes, and they found CES3, SLC27A2, HSP90, HSP70, FRIL1, CPS1, MAT and COMT increased in D-GalN-induced ALI through in vivo and in vitro experiments (38). Some unique proteins in EVs reflect the identity and tissue-specific origin of EVs. For instance, liver-specific proteins such as CES1, ADH1, GST, APOA1, ALB, HP and FGB in the EVs increased after hepatotoxin-induced liver injury. And ALB and HP in the circulating EVs were also confirmed to be increased in the alcohol-induced liver injury of the rodent model (39). Cytochrome P450 2D1 was also upregulated in hepatic EVs after GalN-induced injury, and this hepatocyte-specific enzyme could serve as a novel candidate marker to specifically detect and follow DILI (41).
Recently, exosomal miRNAs have emerged as promising biomarkers with diagnostic value. Receiver operating characteristic (ROC) analysis revealed exosomal miR-122a-5p exhibited superior diagnostic performance with an earlier diagnostic potential and a wider diagnostic time window compared to the corresponding serum counterpart in two animal models of ALI. In addition, exosomal miRNAs showed a higher correlation with ALT activity. Notably, exosomal miRNAs-122a-5p, 192-5p and 193a-3p manifested an injury-specific signature in ALI, and could be used not only as diagnostic tools but also to differentiate between different etiologies of hepatic injury (42). Furthermore, the levels of liver-specific miRNAs such as miR-122, miR-192 and miR-155 in circulating Exos were reported to be elevated in APAP-induced liver injury, but significantly decreased and returned to basal levels after treatment with antioxidant N- Frontiers in Immunology frontiersin.org acetyl-cysteine (NAC), suggesting the levels of exosomal miR-122, miR-192 and miR-155 mirrored the severity of hepatocyte damage and might be used as potential sensitive diagnostic biomarkers for liver injury. High levels of circulating miRNAs are produced within certain cells in a tissue-specific manner, making them good candidate biomarkers for particular types of tissue injury. In this regard, nine miRNAs were identified as signatures of ALI. Of which, five miRNAs (miR-21a-5p, miR-92a-3p, miR-194-5p, miR-17-5p and miR-19b-3p) were increased, four miRNAs (miR-451a, miR-27a-3p, miR-26a-5p and miR-223-3p) were decreased (43).

Clinical studies
ACLF is defined as an acute deterioration of liver function in patients with chronic liver diseases. It is a life-threatening clinical syndrome with a high mortality of 50-90%. Early diagnosis and recognition of patients who will die without liver transplantation are vitally important. Chen JJ and colleagues investigated differentially expressed messenger RNAs (mRNAs), long noncoding RNAs (lncRNAs), and circular RNAs (circRNAs)in circulating Exos from patients with ACLF using RNA sequencing. They found higher lncRNA but less circRNA was expressed in HBV-ACLF patients. NADPH oxidase 1 (NOX1) mRNA and lncRNA ZSCAN16-AS1 were highly expressed in patients with HBV-ACLF, and the expression levels of them were positively correlated with the progression or severity of liver injury (48). In addition, CD39 + CD133 + MPs were elevated in patients with acuteon-chronic liver decompensation, suggesting acute liver insults and/ or acute deterioration of liver function in the setting of chronic liver injury (36).

EVs as potential biomarkers for prognostic assessment of liver failure
A prospective study evaluated the predictive value of serum exosomal long noncoding RNA nuclear-enriched abundant transcript 1 (LncRNA NEAT1) for 90-day mortality in acute-onchronic hepatitis B liver failure (ACHBLF). The results displayed that lncRNA NEAT1 levels were higher in non-survivors than survivors. In the training cohort, lncRNA NEAT1 was an independent predictor for 90-day mortality of ACHBLF. Meanwhile, lncRNA NEAT1 showed a significantly higher area under the curve of receiver operating characteristic (AUC) than the MELD score in the training and validation cohort. ACHBLF patients with lncRNA NEAT1 levels above 1.92 showed poorer survival conditions than those below. Thus, serum exosomal lncRNA NEAT1 might be a better prognostic biomarker than the MELD score for 90-day mortality in ACHBLF (49). On the other hand, the assessment of liver regeneration is particularly critical for predicting prognosis and improving the quality of life in ACLF patients. Jiao Y et al. reported that the percentage of Exos with ALB, CD63 and VEGF increased in CHB, but decreased in ACLF. Among ACLF patients, the Exos with ALB, CD63 and VEGF were significantly more in the survival group than the dead group. The sensitivity and specificity of Exos with CD63, ALB and VEGF were significantly higher than other markers of liver regeneration and prognostic valuation including AFP. Therefore, the Exos with ALB and VEGF might be more accurate and specific biomarkers of liver regeneration and prognostic valuation than AFP in patients with ACLF (50).

EVs as potential biomarkers for reflecting the severity of hepatic damage in other diseases
In some cases, liver disease is one of the main contributors to the increased morbidity and mortality in other severe diseases. Under these circumstances, circulating miRNAs might be used to reflect liver damage and develop risk assessment. For example, a prospective, observational study reported serum miR-122-5p levels were significantly positively correlated with serum liver function markers in the setting of acute heart failure (44). In anti-retroviral therapies (ART)-treated, HIV-1-infected individuals, circulating levels of miR-122 and miR-200a were elevated in HIV/HCV coinfected individuals, compared to HIV mono-infected individuals. Especially, higher pre-ART levels of circulating miR-122 and miR-200a were noticed in HIV-1 positive individuals who died from liver-related diseases whilst undergoing suppressive ART, compared to matched controls. Thus, circulating miR-122 and miR-200a were considered as promising predictive biomarkers for severe liver disease in the ART-treated, HIV-1-infected populations (47). Tables 3, 4 listed the studies which exhibited the therapeutic potential of EVs for liver failure or liver injury induced by hepatotoxic drugs and surgery (mainly hepatic IRI), respectively. EVs derived from stem cells were most frequently reported to have therapeutic potential for liver failure or liver injury, including EVs from the human umbilical cord (blood) mesenchymal stem cells [hUC(B)MSCs], adipose-derived stem cells [A(D)SCs], and bone marrow-derived mesenchymal stem cells (BM-MSCs). Moreover, EVs derived from human menstrual blood-derived stem cells (MenSCs) (54), human-induced pluripotent stem cell-derived MSCs (hiPSC-MSCs) (67, 68) and human liver stem cells (HLSCs) (82) were also documented to protect against liver failure. Other sources of EVs included: hepatocytes (65, 66, 83), dendritic cells (DCs) (71), normal or damaged liver tissues (64) and red blood cells (RBCs) (90,92).

EVs as the potential treatment option for liver failure/injury
The therapeutic effects of EVs on liver failure were investigated only using rodent models. Overall, different EV therapies were documented to alleviate liver damage and improve the survival of animals.

EVs as the potential treatment option for drug-induced liver failure/injury
Hepatotoxins D-GalN plus LPS were most often utilized to induce liver failure. Exos derived from hUCMSC with or without TNF-a treatment were reported to alleviate D-GalN/LPS-induced ALF and promote the repair of damaged liver tissues by reducing the activity of the NLRP3 inflammasome in macrophages (60, 61). miR-17 or lncRNA H19 in ASC-derived Exos protected animals from ALF by inhibiting TXNIP/NLRP3 inflammasome activation in macrophages or promoting hepatocyte proliferation mediated by the HGF/c-Met pathway (57,58). Exos derived from BM-MSCs or human MenSCs were also demonstrated to dramatically improve the survival of mice with lethal hepatic failure by suppressing cell apoptosis (53,54). Unexpectedly, some foods can also secrete exosome-like nanoparticles (ELNs) or vesicle-like nanoparticles (VLNs) which have protective roles. For example, shiitake mushroom-derived ELNs and honey-or garlic chive-derived VLNs were verified to protect mice against D-GalN/LPS-induced liver failure through inhibiting NLRP3 inflammasome activation (86-88) ( Table 5).
CCl 4 is another common hepatotoxin to induce liver failure. hUCMSC-Exos were shown to relieve CCl 4 -induced liver injury through antioxidant effects (55,56). And miR-455-3p-enriched hUCMSC-Exos were verified to ameliorate ALI through inhibiting inflammatory response by targeting PI3K signaling (62). Exos derived from BM-MSCs were elucidated to elicit the hepatoprotective effects by activating the proliferative and regenerative responses in CCl 4 -induced liver injury (51, 59). Human hepatocyte-derived EVs were documented to attenuate ALI through modulating inflammatory immune response (65). Interestingly, the administration of both normal and damaged liver EVs was proved to significantly accelerate the recovery of liver tissue from CCl 4 -induced hepatic necrosis by inducing the production of hepatocyte growth factor at the site of the injury (64).
In addition, hUCMSC-derived EVs were demonstrated to alleviate APAP-induced ALF through activating ERK and IGF-1R/ PI3K/AKT signaling pathway (63). Moreover, EVs derived from BM-MSC exerted beneficial protection through immunosuppression in the Con A-induced animal model of liver injury (52).

EVs as the potential treatment option for liver failure/injury induced by hepatic IRI
EVs and their bioactive cargoes were confirmed to protect mice or rats against hepatic failure (Table 4) induced by hepatic IRI. Exos derived from ADSCs attenuated hepatic IRI via promoting survival mediated by ERK1/2 and GSK-3b signaling pathways (85) or regulating mitochondrial dynamics and biogenesis (84). The overexpression of miR-148a in ADSC-Exos by transfection inhibited the expressions of CaMKII and TLR4 in liver ischemiareperfusion tissues and reduced the occurrence of the inflammatory response and apoptosis (75). hUCMSC-Exos also were documented to alleviate hepatic IRI by suppressing oxidative stress and neutrophil inflammatory response (76), inhibiting Beclin1-and FAS-mediated autophagy and apoptosis (77), or modulating inflammatory immune response (78) and promoting liver regeneration by downregulating Foxg1 (81). Exosomal miR-1246 derived from hUCBMSCs exerted anti-apoptosis, pro-survival and anti-inflammatory effects by modulating the GSK3b-mediated Wnt/b-catenin signaling pathway (72) or modulating the balance between Tregs and Th17 cells (73). In addition, EVs originating from BM-MSCs were demonstrated to protect the liver against IRI through modulating inflammatory response (increased antiinflammatory NLRP12 expression) (70), improving hepatic regeneration (74), or enhancing autophagy (79). Moreover, hiPSC-MSC-derived Exos played a protective role in IRI via inhibiting inflammation, apoptosis, and oxidative stress (67) or promoting cell proliferation via the activation of sphingosine kinase and sphingosine-1-phosphate pathway (68). Furthermore, 3×10 9 HLSC-EVs were able to modulate hepatic IRI by preserving tissue integrity and reducing transaminase release and inflammatory cytokines expression (82). In addition to stem cell-derived EVs, EVs from other sources also were proved to exert hepatoprotection in IRI. For instance, EVs derived from DCs were documented to attenuate IRI by transporting HSP70 to naïve T cells and stimulating the PI3K/mTOR axis to modulate the balance between Treg and Th17 Cells (71). CD47-enriched EVs released by hepatocytes in a Yes-associated protein (YAP)-dependent manner ameliorated hepatic IRI through inhibiting dendritic cell activation (83).

Modified EVs as the potential treatment option for liver failure/injury
In some cases, EVs from the parental cells are not sufficient to effectively treat diseases. In light of this, scientists are trying their best to modify EVs or combine them with other hepatoprotective agents to optimize their functions. As a result, the protective effects of EVs against liver failure enhanced after modification. For example, the therapeutic effects of ADMSC-EVs on ALI through inhibiting inflammation or rapid senescence-like response were reinforced by combined treatment with melatonin (69) or glycyrrhetinic acid (80), loading with vitamin A and quercetin (91), or even transfection with miR-148a (75). Recently, EVs from red blood cells (RBC-EVs) are considered as preferable drug delivery vehicles because of their characteristics of low immunogenicity, easy availability and liver accumulation. RBC-EVs loaded with antisense oligonucleotides (ASOs) of miR-155 (miR155-ASOs) exhibited excellent protective and therapeutic effects against ALF by regulating macrophage polarization (90). Hybridized Ce-red blood cell vesicles (Ce-ReVs), in situ growth of cerium oxide (Ce) nanocrystals onto nano-sized red blood cell vesicles (ReVs), showed strong reactive oxygen species (ROS) elimination. Upon further hybridization with MSC-Exos, the resulting Ce-ReMeVs conferred superior repair benefit and brought about promising therapeutic outcomes even for models with more severe inflammatory damage, including D-GalN/LPSinduced ALI model (92). Furthermore, scientists have developed a hydrogel-mediated sustained systemic delivery of MSC-EVs to improve liver regeneration in chronic liver failure (89). In addition, systemic administration of exosome-mimicking Meseomes which is composed of membranes and secretome from efficacy-potentiated MSCs alleviated tissue necrosis and promoted the recovery of liver function in ALI models induced by CCl 4 (93) ( Table 6).

The quality assessment results
According to the quality assessment result by the SYRCLE tool, the majority of 57 included studies were marked as "low" or "unclear" (66.7%) risk of bias due to insufficient information regarding selection method, allocation concealment, animal housing, blinding, replacement of dropout animals, and so on (Table S2). The risk of bias summary and risk of bias graph were shown in Figure S1. All the 3 case-control studies were considered as "high" quality with a cumulative NOS score of seven points or greater. The risk of bias comprehensive assessment is shown in Table S3. The CERQual tool revealed that the review findings reached a "high" confidence appraisal (Table S4). The PRISMA checklist was completed which included further details for the review scoring (Table S6).

Discussion
In this review, we systematically and comprehensively summarized published studies on the administration of EVs in experimental animal models or clinical studies related to liver failure, and assessed the potential of EVs as biomarkers and treatment options for liver failure. The results revealed that EVs carrying differentially expressed proteins or nucleic acids (especially surface membrane proteins and various RNAs) can be utilized not only as early warning or diagnostic biomarkers, but also to provide clues for prognostic evaluation of liver failure. In addition, EVs and their functional cargoes have been confirmed to exert important hepatoprotective effects and can be used to treat liver failure.
Emerging studies have documented that EVs with their cargoes can serve as biomarkers for early warning, early or differential diagnosis and prognosis assessment of liver diseases, such as druginduced liver injury, viral hepatitis, non-alcoholic fatty liver disease, alcoholic liver disease, liver cirrhosis and liver cancer (18)(19)(20)(21). Nevertheless, it remains unclear about the potential of EVs as biomarkers of liver failure. And one of the major aims of this systematic review is to clarify this issue.
Differentially expressed proteins in EVs usually reflect the functional properties and tissue-specific origin of EVs, as well as underlying pathophysiological mechanisms, thus they have the potential to act as emerging biomarkers for early warning or diagnosis of liver failure. Multiple studies in this systematic review showed that the unique proteins, including intrinsic proteins or surface membrane proteins carried by EVs, are good early warning or diagnostic biomarkers for liver failure. In terms of preclinical studies, elevated surface protein CD133 (36) or intrinsic proteins ALB, HP and FGB (39) in circulating EVs may serve as early warning biomarkers in APAP-induced liver injury. In addition, several proteins such as CES3, SLC27A2, and HSP90 were found to be more commonly expressed in EVs isolated from sera of D-GalN-induced rats (38). On the contrary, some proteins (such as CD26, SLC3A1 and CD81) that are normally present in urinary vesicles were discovered to be dramatically reduced in urinary samples obtained from D-GalN-treated rats (35). In hepatic IRI, MPs derived from platelets (with surface CD41 + ) and neutrophils (with surface Ly-6G + ) may be markers of hepatic inflammation following injury, and MPs derived from endothelial cells (with surface CD62E + ) may reflect angiogenesis during the reparative phase (37).
In addition, differentially expressed RNA, especially miRNA in EVs, can also be considered as early warning, diagnosis or prognosis evaluation biomarkers of liver failure. For instance, miR-122, miR-192 and miR-155 carried by EVs may serve as biomarkers of ALI considering that they were upregulated in APAP-induced liver injury and reversed to basal levels after NAC treatment (40). Notably, exosomal miR-122a-5p was regarded as the best diagnostic biomarker with higher diagnostic power and a wider diagnostic window for ALI (42). Furthermore, a set of miRNAs including miR-122-5p, miR-192-5p and miR-22-3p were considered as potential biomarkers which contribute mainly to liver steatosis and inflammation in ALI (43).
Consistent with findings in experimental animal studies, differentially expressed surface proteins and RNAs carried by EVs also may be considered as potential early warning or prognostic biomarkers for liver failure in clinical studies. ALB and VEGF carried by Exos may be more accurate and specific biomarkers of liver regeneration and prognostic evaluation than AFP in serum for patients with ACLF, and CD63 + ALB + Exos may be an early warning marker for ACLF patients (50). Furthermore, NOX1 mRNA and lncRNA ZSCAN16-AS1 were reported to be increased in HBV-ACLF and HBV patients compared to healthy control (48). Especially, lncRNA NEAT1 is a better prognostic biomarker than the MELD score for 90-day mortality in ACHBLF (49). Interestingly, miR-122 has also been found to reflect the severity of liver damage in patients with other diseases, such as patients with acute heart failure and ART-treated, HIV-1-infected individuals (44, 47). However, more comprehensive and large-scale clinical studies are still needed in the further to verify the role of EVs as good biomarkers for liver failure.
Although preclinical and clinical studies have revealed that EVs may be potential early warning, diagnosis or prognosis evaluation biomarkers of liver failure, research in this area remains to be in its infancy and large-scale clinical studies are scarce. This may be ascribed to the followings: (1) there is no widely accepted ideal animal model of liver failure, especially ACLF; (2) high technical requirements and detection costs as well as uncertain results limit the implementation of studies related to Exos; (3) the etiologies of liver failure are diversified, which leads to great individual variation; (4) liver failure is often accompanied by complex complications, which may act as confounding factors and limit the diagnostic and prognostic value of EVs for liver failure in clinical studies.
The other objective of this systematic review is to assess the treatment potential of EVs in liver failure. To this end, we comprehensively summarized the studies in which EVs are utilized to treat liver failure. EVs derived from stem cells, especially MSCs, are most commonly used to treat liver failure. MSCs can be isolated from bone marrow, adipose tissue, umbilical cord, and so on. They have been documented to possess the strong ability of differentiation, regeneration and immunomodulation, and can be used to treat various liver diseases efficiently and safely in clinical studies (96,97). The therapeutic effect of MSCs can be attributed to paracrine, especially EVs, to a large extent. Compared with MSCs, MSC-EVs have the following advantages: (1) much smaller in size and easier to pass through biological barriers including the blood-brain barrier; (2) can be administered intravenously, and more likely accumulate in the liver; (3) cryopreservation does not affect the clinical efficacy; (4) lower immunogenicity and better histocompatibility. Therefore, EVs are emerging as a novel treatment option.
EVs and their functional cargoes turn out to play important hepatoprotective effects in liver failure by inhibiting inflammatory response, oxidative stress, apoptosis or promoting hepatocyte regeneration and autophagy.
With respect to enhancement of autophagy, Exos obtained from MSC-derived hepatocyte-like cells treatment enhanced autophagy during hepatic IRI to exert the hepatoprotective effect (79), and hUCMSC-Exos carried with miR-20a inhibited Beclin-1 and FAS to alleviate the abnormal expression of apoptosis-and autophagyrelated genes in liver ischemia-reperfusion (77).
In addition to stem cell-derived EVs, EVs derived from other cells or tissues have also been confirmed to protect against liver failure, such as hepatocytes (65, 66, 83), DCs (71) as well as normal or damaged liver tissue (64). Interestingly, researchers also extracted desired EVs from natural foods, such as mushroom, honey and garlic chive, and proved that they may have a protective effect on the liver by inhibiting the activation of inflammasome (86)(87)(88). These findings greatly broaden the source of EVs and provide new ideas for the treatment of liver failure.
In recent years, scientists have attempted to modify EVs to achieve superior therapeutic effects for liver failure. For example, red blood cells which have the features of low immunogenicity, easy Graphical Abstract to summarize the potential of biomarkers and treatment options for EVs. availability and liver accumulation were transfected with hepatoprotective miRNAs [such as miR155-ASOs (90)] or hybridized with Ce and MSC-Exos (92) to improve the targeting and therapeutic effect of EVs on ALF. Moreover, the protective effect of ADMSC-Exos on liver failure was improved by loading them with vitamin A and quercetin (91). In addition, EVencapsulated PEG hydrogels were developed to retard the clearance of MSC-EVs, ultimately improving liver regeneration in chronic liver failure (89).

Conclusion
In conclusion, EVs and their cargoes can be used not only as superior biomarkers of early warning, diagnosis and prognostic assessments for liver failure, but also as potentially effective treatment options for patients with liver failure (Figure 2). In the future, large-scale studies are urgently needed to verify the diagnostic, predictive and therapeutic value of EVs for liver failure.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.