Extracellular vesicles for acute kidney injury in preclinical rodent models: a meta-analysis

Introduction Extracellular vesicles (EVs), especially stem cell-derived EVs, have emerged as a potential novel therapy for acute kidney injury (AKI). However, their effects remain incompletely understood. Therefore, we performed this meta-analysis to systematically review the efficacy of EVs on AKI in preclinical rodent models. Methods We searched PubMed, EMBASE, and the Web of Science up to March 2019 to identify studies that reported the treatment effects of EVs in a rodent AKI model. The primary outcome was serum creatinine (Scr) levels. The secondary outcomes were the blood urea nitrogen (BUN) levels, renal injury score, percentage of apoptotic cells, and interleukin (IL)-10 and tumour necrosis factor (TNF)-α levels. Two authors independently screened articles based on the inclusion and exclusion criteria. The meta-analysis was conducted using RevMan 5.3 and R software. Results Thirty-one studies (n = 552) satisfied the inclusion criteria. Pooled analyses demonstrated that the levels of Scr (SMD = − 3.71; 95% CI = − 4.32, − 3.10; P < 0.01), BUN (SMD = − 3.68; 95% CI = − 4.42, − 2.94; P < 0.01), and TNF-α (SMD = − 2.65; 95% CI = − 4.98, − 0.32; P < 0.01); the percentage of apoptotic cells (SMD = − 6.25; 95% CI = − 8.10, − 4.39; P < 0.01); and the injury score (SMD = − 3.90; 95% CI = − 5.26, − 2.53; P < 0.01) were significantly decreased in the EV group, and the level of IL-10 (SMD = 2.10; 95% CI = 1.18, 3.02; P < 0.01) was significantly increased. Meanwhile, no significant difference was found between stem cell-derived EVs and stem cells. Conclusion The present meta-analysis confirmed that EV therapy could improve renal function and the inflammatory response status and reduce cell apoptosis in a preclinical rodent AKI model. This provides important clues for human clinical trials on EVs.


Background
Acute kidney injury (AKI) is a major kidney disease characterised by a rapid decline in renal function and is associated with an increase in mortality and hospitalisation [1]. However, the prognosis of this disease, which may occur under various circumstances, has not been significantly improved since the mid-1990s [2]. Due to the lack of efficient therapeutic methods, patients with renal ischaemia reperfusion injury (IRI) are mostly treated by supportive manoeuvres, such as renal replacement therapy [3].
Many studies have confirmed that mesenchymal stem cell (MSC) therapy can effectively improve AKI [4,5], but most of these studies have not found that MSCs colonise in the kidneys to play a direct role [4,6]. Moreover, MSC therapy may have certain risks, such as inducing tumours, and its safety remains questionable [7].
Recently, data in the literature have highlighted that the delivery of MSC-derived EVs can ameliorate AKI in preclinical models [3,6,8]. EVs are secreted by almost all types of cells and can be subdivided into exosomes, microvesicles, and apoptotic bodies [9]. Exosomes are the smallest vesicles (30-100 nm) released by the fusion of multivesicular bodies containing intraluminal vesicles with the plasma membrane. Microvesicles are vesicular structures (0.1-1.0 μm) shed by outward blebbing of the plasma membrane. The largest EVs (1-5 μm) are apoptotic bodies that are formed during the late stages of apoptosis [10]. EVs contain proteins, lipids, carbohydrates, mRNAs, and miRNAs and may influence different cell types acting on physiological processes such as proliferation and immune escape [11]. Compared with MSCs, the small size of MSC-derived EVs allows them to avoid the pulmonary first-pass effect and to penetrate deep inside most body barriers [3]. Therefore, MSCderived EVs are expected to be an effective treatment for AKI.
Many animal studies have been performed to investigate the efficacy of EVs on an AKI model with various cell origins and different injection doses, delivery routes, and therapy times [3,12]. To provide the most recent available evidence for clinical studies, we performed this meta-analysis to investigate the efficacy of EVs on preclinical rodent models.

Materials and methods
Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) was used to perform this meta-analysis [13].

Search strategy
We searched PubMed, EMBASE, and the Web of Science from database inception to March 2019. The search terms were as follows: ("extracellular vesicles" or "EVs" or "micro vesicles" or "micro-vesicles" or "microvesicles" or "microparticle" or "exosome" or "MVs" or "shedding vesicles") and ("AKI" or "acute kidney injury" or "renal ischaemia-reperfusion" or "acute renal failure"). The search was limited to rodent models with no language restrictions. The reference lists of selected studies were searched by hand to identify potentially relevant citations. Ethical approval was not required because the meta-analysis was based on published articles.

Study selection
Two independent investigators (CL and JW) conducted the study selection. Disagreements between the investigators were resolved in meetings or adjudicated by a third reviewer (XS).

Eligibility criteria
The inclusion criteria were as follows: (1) population-rodent models with AKI; (2) intervention-various cell-derived EVs; (3) comparison-placebo; and (4) outcome measure-the primary outcome was the level of serum creatinine (Scr). The secondary outcomes were the renal injury score, percentage of  apoptotic cells, and levels of blood urea nitrogen (BUN), interleukin (IL)-10, and tumour necrosis factor (TNF)-α. The exclusion criteria were as follows: (1) AKI was not performed on rodent models, (2) repeated data, (3) insufficient information, and (4) review, letter, commentary, correspondence, case report, conference abstract, expert opinion, or editorial.

Data extraction
Data extraction was performed by two independent reviewers (CL and JH) using a standardised form. The following data were collected: first author, country or region, publication year, number of animals, type of AKI model, species, treatment time, measurement time and EV cell origins, diameter, and dose. For studies that had not shown the corresponding results, Engauge Digitizer version 4.1 software was used to extract data from the graphics [14,15].

Quality assessment
The methodological quality of each included study was evaluated by two independent authors (JW and ZM) with a Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies (CAMARADES) 10-item checklist [16]: A, peerreviewed journal; B, temperature control; C, animals were randomly allocated; D, blind established model; E, blinded outcome assessment; F, use of anaesthetic without significant intrinsic vascular protection activity; G, appropriate animal model (diabetic, advanced age, or hypertensive); H, calculation of the sample size; I, statement of compliance with animal welfare regulations; and J, statement of potential conflicts of interest.

Statistical analysis
All statistical analyses were conducted using RevMan version 5.3 and R statistical software version 3.4.1. Statistical significance was set at P < 0.05 (two-tailed). Continuous outcomes are expressed as the standardised mean difference (SMD) with the 95% CI. Heterogeneity was analysed among studies using the I 2 statistic. I 2 > 50% indicated significant heterogeneity [17]. Subgroup, sensitivity, and meta-regression analyses were performed to investigate potential betweenstudy heterogeneity and to explore other potentially confounding factors. A cumulative meta-analysis was performed to explore changes in the results over time. Funnel plots and Egger's test were conducted to detect publication bias. If publication bias was indicated, we further evaluated the number of missing studies by the Trimfill method and recalculated the pooled risk estimation with the addition of those missing studies.

Quality assessment
All the included records were peer-reviewed publications, and all animals were allocated randomly to a treatment group and a control group; however, most studies did not report sample size calculation, blinded induction of the model, or blinded assessment of outcome. The details of the study quality assessment are shown in Additional file 1: Table  S1.

Primary outcome
All studies reported the level of Scr. The pooled analysis showed that EVs can significantly reduce the Scr level when compared with the control (SMD = − 3.71; 95% CI = − 4.32, − 3.10; P < 0.01; I 2 = 73%; Fig. 2). The subgroup analysis showed that all cell-derived exosomes are effective in reducing the Scr level (Fig. 2). The cumulative meta-analysis showed that the result did not change over time (Additional file 2: Figure S1). The sensitivity analysis showed that none of the single studies significantly influenced the result (Additional file 3: Figure S2). The multivariable meta-regression analysis showed that the delivery dose (P < 0.05) and cell origin of EVs (P < 0.05) were independent influential factors of SCr reduction.

Secondary outcomes
The level of BUN was significantly decreased in the EV group (SMD = − 3.68; 95% CI = − 4.42, − 2.94; P < 0.01; I 2 = 82%; Fig. 3). A subgroup analysis was performed according to the origin of the EVs, and the results indicated that all kinds of EVs included in this meta-analysis would reduce the level of BUN. The cumulative meta-analysis showed that the result did not change over time (Additional file 4: Figure  S3). The sensitivity analysis showed that none of the single studies significantly influenced the result (Additional file 5: Figure S4). The meta-regression analysis showed that the cell origin of the EVs (P < 0.05) was an independent influential factor of BUN reduction.

Publication bias
Significant publication bias was observed (P < 0.01; Additional file 6: Figure S5). We used the Trimfill method to recalculate the pooled risk estimation with the addition of missing studies (Additional file 7: Figure S6). However, the overall results were not significantly changed. Therefore, publication bias may have little effect on the meta-analysis outcomes (data not shown).

Discussion
Our meta-analysis of 31 studies provided a comprehensive summary of the effect of EVs on the preclinical rodent AKI model. Pooled analyses confirmed that EV therapy could improve renal function and the inflammatory response status and reduce cell apoptosis in a preclinical rodent AKI model. The multivariable meta-regression analysis indicated that the delivery dose and cell origin of EVs were independent factors influencing the effect of EVs. Meanwhile, no significant difference was found between stem cellderived EVs and stem cells. Therefore, the present meta-analysis provides important clues for human clinical trials on EVs.
A previous meta-analysis focused on this topic indicated that mesenchymal stromal cell-derived EVs produce a more marked therapeutic effect on recovery from renal failure than MSC-conditioned medium [49]. Our meta-analysis contained various types of cell-derived EVs and further evaluated the effect of EVs on cell apoptosis, the tubular injury score, and inflammatory cytokines, providing useful information for further clinical trials.
Many studies have shown that RNAs carried by EVs are the pivotal mechanism for their therapeutic function [11,50], and the proteins contained in EVs are also related to many biological processes. EVs are membrane-bound vesicles released by all cell types, including stem/progenitor cells, which are important information carriers for regulating angiogenesis, extracellular matrix remodelling, gene expression, inflammation states, the cell cycle and proliferation, the phenotype of target cells, cell migration, and morphogenesis [51][52][53][54]. The surface (See figure on previous page.) Fig. 2 The forest plot shows the efficacy of EVs in reducing Scr levels in the AKI model. ADMSC, adipose-derived mesenchymal stromal cell; BMSC, bone marrow mesenchymal stromal cell; 95% CI, 95% confidence interval; EVs, extracellular vesicles; HLSC, human liver stem cell; IV, inverse variance; KMSC, kidney-derived mesenchymal stromal cell; Scr, serum creatinine; SD, standard deviation; UCMSC, umbilical cord mesenchymal stromal cell; UVEC, umbilical vein endothelial cell; WJMSC, Wharton's jelly mesenchymal stromal cell Fig. 3 The forest plot shows the efficacy of EVs in reducing BUN levels in the AKI model. ADMSC, adipose-derived mesenchymal stromal cell; BMSC, bone marrow mesenchymal stromal cell; BUN, blood urea nitrogen; 95% CI, 95% confidence interval; EVs, extracellular vesicles; HLSC, human liver stem cell; IV, inverse variance; KMSC, kidney-derived mesenchymal stromal cell; SD, standard deviation; UCMSC, umbilical cord mesenchymal stromal cell; UVEC, umbilical vein endothelial cell; WJMSC, Wharton's jelly mesenchymal stromal cell molecules of EVs permit them to be targeted to recipient cells. Once attached to a target cell, EVs can induce signalling via a receptor-ligand interaction, be internalised by endocytosis and/or phagocytosis, or even fuse with the target cell's membrane to deliver their content into its cytosol, thereby modifying the physiological state of the recipient cell [55,56].
Compared with stem cells, stem cell-derived EVs have lower immunogenicity and may reduce some of the risks associated with cellular therapy, such as cytokine release syndrome [51]. In our metaanalysis, we demonstrated that stem cell-derived EVs were equally effective as stem cells when applied to treat AKI. In one study, MSC-derived EVs were superior to MSCs in reducing global renal damage levels in a rat model of donation after circulatory death (DCD) kidney [57]. Thus, EVs appear to be a promising approach for the repair of AKI.
The multivariable meta-regression analysis showed that the delivery dose and cell origin of EVs were independent factors influencing the efficacy of EVs. This suggests that we need to consider these factors when performing clinical trials. The properties and cargoes of EVs have been summarised in databases that are continuously updated, namely, Vesiclepedia, ExoCarta, and EVpedia [58]. Interestingly, the same cell may release EVs that differ in the content of their membrane lipid composition and in their intravesicular cargo [58,59]. Therefore, further studies are urgently needed to explore the mechanism behind this phenomenon.
In our meta-analysis, various sizes of EVs were included. The large heterogeneity between EVs poses major obstacles to understanding the composition and functional properties of distinct secreted components [60]. One recent research reassessment of exosome composition established the differential distribution of protein, RNA, and DNA between small EVs and nonvesicular extracellular matter and demonstrated that small EVs are not vehicles of active DNA release [60]. It is important for further study to identify the key elements in AKI treatment.
One clinical trial tested the effects of MSC-derived EVs on the progression of chronic kidney disease (CKD) patients, and the results indicated that EVs can improve the estimated glomerular filtration rate (eGFR); decrease Scr, BUN, and TNF-α levels; and increase IL-10 levels [61]. However, significant translational challenges need to be addressed before the use of MSC-derived EVs for the clinical treatment of AKI. First, EV isolation and storage methods may potentially affect EV characteristics. It is challenging to ensure that recovered vesicles are truly from the extracellular space rather than from intracellular vesicles or artefactual particles released from cells broken during tissue harvest, processing (e.g. mechanical disruption), or storage (including freezing) [9]. Second, in most studies, the follow-up time ranged from 1 day to 2 weeks. Therefore, the long-term effects of EVs are a key issue that requires further exploration before their clinical application. Third, a development method that can be used to meet the large-scale clinical production requirement of a sufficient quantity of EVs is also a core problem [51]. Fourth, labelling EVs with lipophilic or surfacecoating fluorophores may modify the physicochemical characteristics of EVs and alter the detection mode and/or uptake by target cells [9]; thus, the development of specific tracking tools is required to further detect EVs.

Limitations
Several potential limitations to this meta-analysis should be considered. First, despite the fact that we performed subgroup and sensitivity analyses, the heterogeneity between studies cannot be remarkably reduced. This may weaken the stability of the results. Second, we included stem cell-derived EVs and other cell origin EVs, but we did not perform a direct comparison to identify the best option, which may have also increased the heterogeneity. Third, there was potential for the incomplete retrieval of identified research studies, which could have introduced publication bias. Finally, data extraction from graphics by using Engauge Digitizer software may have altered the original data, which would also affect the results.