Therapeutic efficacy of mesenchymal stem cells for abdominal aortic aneurysm: a meta-analysis of preclinical studies

Background Abdominal aortic aneurysm (AAA) is life-threatening, surgical treatment is currently the only clinically available intervention for the disease. Mesenchymal stem cells (MSCs) have presented eligible immunomodulatory and regenerative abilities which showed favorable therapeutic efficacy in various cardiovascular diseases. However, current evidence summarizing the effectiveness of MSCs for AAA is lacking. Thus, a meta-analysis and systematic review was necessary to be performed to assess the therapeutic efficacy of MSCs for AAA in preclinical studies. Methods Comprehensive literature search restricted in English was conducted in PubMed, Cochrane Library, EBSCO, EMBASE and Web of Science from inception to Oct 2021. The primary outcomes were parameters about aortic diameter change during MSCs intervention. The secondary outcomes included elastin content and expression level of inflammatory cytokines, matrix metalloproteinases (MMPs) and their inhibitors (TIMPs). Data were extracted and analyzed independently by two authors. The meta package with random effects model was used to calculate the pooled effect size and 95% confidence intervals in R (version 4.0.2). Results Meta-analysis of 18 included studies demonstrated that MSCs intervention has significant therapeutic effects on suppressing aortic diameter enlargement compared with the control group (diameter, SMD = − 1.19, 95% CI [− 1.47, − 0.91]; diameter change ratio, SMD = − 1.36, 95% CI [− 1.72, − 1.00]). Subgroup analysis revealed differences between MSCs and control group regarding to cell type, intervention route and cell compatibility. Moreover, the meta-analysis also showed that MSCs intervention had a significant effect on preserving aortic elastin content, reducing MCP-1, TNF-α, IL-6, MMP-2/9 and increasing TIMP-1/2 expression level compared with control group. Conclusion Our results suggested that MSC intervention is effective in AAA by suppressing aortic diameter enlargement, reducing elastin degradation, and modulating local immunoinflammatory reactions. These results are important for the systemic application of MSCs as a potential treatment candidate for AAA in further animal experiments and clinical trials. Supplementary Information The online version contains supplementary material available at 10.1186/s13287-022-02755-w.


Introduction
Abdominal aortic aneurysm (AAA) is permanent pathologic dilatation of the abdominal aorta, of which the maximum diameter of lesion area is 1.5 folds greater than the normal segment or more than 3 cm regardless of differences in patient gender and stature [1,2]. In recent years, the level of clinical treatment of AAA has been improving with the continuous advancements of endovascular treatment techniques [3]. However, to date, no effective drugs or non-surgical therapies have been developed [4]. It is also expected that the reliable medical treatment will inevitably slow disease progression and improve clinical prognosis, until the current indications for surgical intervention are met. Therefore, the exploration of nonsurgical treatment is of great significance for the clinical intervention of AAA.
The pathogenesis of AAA mainly includes infiltration of inflammatory cells such as macrophages, lymphocytes and neutrophils, degradation of the extracellular matrix mediated by matrix metalloproteinases, as well as apoptosis and phenotypic transition of medial vascular smooth muscle cells (VSMCs) [5,6]. Cellular therapies, especially those based on mesenchymal stem cells (MSCs), have recently demonstrated inspiring repair capabilities in diseases such as spinal cord injury [7], cardiovascular diseases [8][9][10], and Crohn's disease [11,12]. MSCs were reported to exert reparative capabilities by secreting various cytokines or exosomes to modulate local inflammatory reactions and mediate intercellular communications, and being capable of migrating to the lesion sites and differentiating into functional cells [13,14].
Preclinical studies have been conducted to investigate the mechanisms and efficacy of MSCs intervention in AAA. For example, Sharma et al. [15] proved that experimental AAA was attenuated by human placenta derived MSCs. Moreover, our previous study found intravenous injection with human umbilical cord derived MSCs could halt aneurysm enlargement, suppress elastin degradation, inhibit MMP-2/9 and TNF-α expression, and preserve/restore VSMC contractile phenotype [16]. Although results of preclinical studies are rigorous and inspiring, given the limitations of individual studies and the heterogeneity among studies, a pooled analysis about the overall therapeutic effects of MSCs for AAA in preclinical studies is necessary.
Therefore, this systematic review and meta-analysis was implemented to assess the efficacy of MSCs treatment in animals with AAA. Our findings will provide a theoretical basis and guide the clinical application of MSCs-based therapy for AAA.

Methods
The implementation of this systematic review followed the guiding principles presented in the Preferred Reporting Items for Systematic Reviews and Metaanalyses (PRISMA) criteria [17]. The details about PRISMA checklist for this meta-analysis is presented in Additional file 1. In addition, the protocol for this systematic review was registered on PROSPERO (registered ID: CRD42020218430) and can be accessed at www. crd. york. ac. uk/ PROSP ERO/ displ ay_ record. asp? ID = CRD42020218430.

Literature search strategy
We conducted a comprehensive search process to evaluate the therapeutic efficacy of MSCs therapy for AAA in preclinical studies. Literatures in PubMed, EMBASE, Cochrane Library, EBSCO and Web of Science databases were searched from start to Oct 2021 by two independent authors (HW and JW). The keywords used in the search process include "stem cell*", "stromal cell*", "cell transplantation", "progenitor cell*", "precusor cell*", "cell* therap*" and combination with "abdominal aortic aneurysm*", the detailed search strategy was presented in Additional file 2. In addition, we also conducted manual search for references of relevant reviews and included studies eventually. All literatures retrieved from the above databases and manual retrieval process were imported into Endnote (version X9.3.3), a literature management software, for identifying and removing duplicates, conference abstracts, reviews and irrelevant articles.

Inclusion and exclusion criteria
Eligible preclinical studies with outcomes including assessment of the efficacy of MSCs therapy in AAA were included. Studies should be originally published in peerreviewed journals in English. The primary outcomes for inclusion in this review was the absolute final value (mm) or change ratio (% increase) of maximum aortic diameter following MSCs therapy. Secondary outcomes included elastin content, changes in cytokines expression levels and indicators of inflammatory responses. All articles reporting primary outcomes were included, regardless of whether secondary outcomes were reported. Exclusion criteria for this study were non-animal studies, non-MSC intervention, not AAA animal model, in-vitro studies, the experimental group did not receive stem cell therapy or was not simply receiving stem cell therapy alone, and the reported outcomes did not include indicators of maximum aortic diameter.

Data extraction
The process of data extraction in this study was carried out by two independent researchers (JYL and HW). For data extraction with disagreement, the decision was made by a third researcher (SQG) or after group discussion, and the results were summarized afterwards. Following the carefully conducted data extraction, we summarized and documented the details of each study including author, year, country, and details of animal models induction including animal species, sex, and modeling method, as well as details of intervention including cell type, compatibility, route, frequency, total dose, intervention duration (defined as time duration from initial cell intervention to sacrifice). Data at different time points during the follow-up period of the same study were included for analysis. If data results were presented in picture form only, we attempted to obtain data by contacting the corresponding authors, otherwise, the WebPlotDigitizer [18] software was implemented for the measurements of continuous values.

Assessment of risk of bias
The risk of bias (RoB) in this review was evaluated following the principles from the Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE) RoB tool [19]. The risk of bias was assessed by two independent authors (JZM and BYL) and disagreement was settled by the third author (SQG). Terms of risk of bias in this review includes selection bias (1. sequence generation, 2. baseline characteristics, 3. allocation concealment), performance bias (4. random housing, 5. blinding of investigators), detection bias (6. random animals assessment, 7. blinding of outcome assessor), attrition bias (8. incomplete outcome data), reporting bias (9. selective outcome reporting), and 10. other sources of bias. For each included study, the risk of bias was scored as high, low, or unclear.

Statistical analysis
Data analysis in this study were performed using meta package (version 4.18-2) [20] in R software (version 4.0.2). The results of continuous variables were expressed using standard mean difference (SMD) with the 95% confidence interval (CI). SMD were calculated by the mean of outcome, standard deviation (SD), and case number of different arms in each study. The pooled SMD was processed using the random effects model to generate forest plot. Between-study heterogeneity were analyzed by the Q test and I 2 statistics. I 2 < 40% with P > 0.05 represents low heterogeneity, 30% to 60% with P < 0.05 represents moderate heterogeneity, 50% to 90% represents substantial heterogeneity, 75% to 100% represents considerable heterogeneity. Subgroup analysis was implemented to identify potential source of heterogeneity and explore other influencing factors. Differences with a two-tailed P < 0.05 were considered statistically significant.

Study selection
According to the results of literature search, a total of 1283 articles were identified and 1092 were retained after duplicates removed. Next, after reviewing the titles and abstracts of all the articles, 59 articles were isolated and assessed for full-text review. Ultimately, 18 studies were included in this meta-analysis. The detailed study selection process was illustrated in Fig. 1.

Risk of bias
The assessment of Rob was summarized in Table 2. Almost all studies mentioned randomized grouping, but none specified the method. All included studies showed similar features in terms of the baseline characteristics, which reduced the risk of selection bias. No study properly described the method of allocation concealment, performance blinding and detection blinding. Two studies [21,25] were considered to have attrition bias because the reasons for missing samples were not reported. In addition, we did not identify any additional sources of bias.

Primary outcomes Maximum aortic diameter
A total of 13 original studies including 27 experimental arms reported the final value of maximum aortic diameter in the animal model of AAA after MSCs intervention [16,21,[25][26][27][28][29][30][31][33][34][35][36]. Meta analysis with random effects model showed that MSCs intervention significantly reduced the final value of maximum diameter compared with the control group (SMD = − 1.28, 95% CI [− 1.61, − 0.96]; P < 0.05; Fig. 2A). However, a moderate betweenstudy heterogeneity (I 2 = 43%, P = 0.01) was identified. Thus, Baujat plot [37], influence diagnostics [38], leaveone-out meta-analysis and GOSH plot analysis [39] were conducted to determine the outlier studies. Baujat plot  Saline revealed that the study by Zhou et al. [35] contributes the most to the overall heterogeneity (Fig. 2B). Influence diagnostics analysis revealed that the study by Zhou et al. [35] could have distorted our pooled effect estimate and partially contributed to the between-study heterogeneity we found in our initial meta-analysis (Fig. 2C). Leaveone-out meta-analysis recalculated pooled effect, with one study omitted each time, which generated two forest plots ordered by recalculated effect size and I 2 , respectively. We found that the overall effect is narrowed when Zhou's study [35] was omitted and yielded the smallest heterogeneity (Fig. 2D, E). Moreover, GOSH plot analysis also identified the study by Zhou et al. [35]as the main contributor to the overall heterogeneity (Fig. 2F, G). Thus, the meta-analysis was reconducted with Zhou's study omitted, results showed a smaller but significant overall effect size (SMD = − 1.19, 95% CI [− 1.47, − 0.91]; P < 0.05) with low between-study heterogeneity (I 2 = 21%, P = 0.17) (Fig. 3). Subgroup analysis regarded to maximum aortic diameter were conducted with Zhou's study omitted.  Fig. S3). Subgroup analysis for animal model induction methods (Additional file 6: Fig. S4), animal species (Additional file 7: Fig. S5) and control type (Additional file 8: Fig. S6) were not statistically significant.

Secondary outcomes Elastin content
A total of 9 studies including 13 experiment arms reported valid data about elastin content of aortic wall after MSCs intervention in AAA models. Random effects model showed MSCs intervention significantly enhanced aortic elastin content compared to control group (SMD = 1.39, 95% CI [0.99, 1.78]) (Additional file 18: Fig.  S16).

Inflammatory cytokines
To assess the effects of MSCs intervention on expression level of inflammatory cytokines from aortic tissue in AAA animal model, data of MCP-1, IL-6, TNF-α and

Publication bias
Funnel plots were applied to evaluate publication bias of primary outcomes including maximum aortic diameter and change ratio of maximum aortic diameter. As shown in Fig. 8, the distributions of funnel plots were apparently asymmetric, indicating the presence of potential publication bias. Moreover, trim-and-fill procedures revealed that our initial results were overestimated due to publication bias.

Discussion
AAA is a progressive and life-threatening pathological process involving complex interactions between vessel cells and molecules, characterized by local modulation of immune reactions, infiltration of inflammatory cells and cytokines, degradation of extracellular matrix, apoptosis and phenotypic transition of VSMCs, etc. [1]. MSCs were known for their abilities of self-renewal, secretion and multipotential differentiation, which empowers them with compacities of modulating immune inflammatory responses, stimulating local cell proliferation and differentiation, and thus promoting tissue regeneration processes [40]. Recent results from preclinical and clinical studies showed MSCs based cell therapy was advantageous in cardiovascular diseases [41]. However, although randomized controlled clinical studies have been registered and implemented to investigate the safety and efficacy of MSC in AAA [42,43], relevant evidence in AAA is still lacking. To date, plenty preclinical studies have been conducted to investigate the therapeutic potential of MSCs for AAA, given the insufficient number of samples and the large heterogeneity among studies, it is necessary to summarize the present available evidence and further instruct clinical application of MSCs therapy in AAA. The fundamental mechanisms by which MSCs perform biological repair functions in-vivo mainly including immunomodulatory properties, paracrine effects, homing and differentiation potential [44]. In this metaanalysis, almost all included studies reported the immunomodulatory function of MSC administration in AAA Fig. 8 The funnel plots: contour-enhanced funnel plot for A maximum aortic diameter and C change ratio of maximum aortic diameter, respectively; Trim and fill funnel plot for B maximum aortic diameter and D change ratio of maximum aortic diameter, respectively models [15, 16, 21-25, 27, 28, 32, 36], except two studies conducted by the same author [33,34] who focused on the biomechanical changes of the aortic wall after MSC intervention. Four studies reported paracrine effects of MSC intervention in AAA models, including three studies investigated the conditioned medium of MSC [25,31,35] and one study investigated the exosome secreted by MSC [24]. A total of eight studies mentioned MSC capture in aortic tissue after MSC administration using different cell labelling methods [22,[25][26][27][28][29][30]36], which indicates MSC migration and homing properties invivo. However, only one study by Hosoyama et al. [29] presented definite evidence about MSC locally differentiating into different cell types, which highlights the differentiation abilities of MSC in-vivo. In summary, although the mechanisms of MSC intervention in AAA models have been investigated in different studies, the depth of individual study is not sufficient, subsequent relevant studies should explore the mechanisms in a more systematic and detailed manner.
The progressive increase in aortic diameter is the most apparent and dominant feature of the clinical course in AAA [45]. Notably, our meta-analysis from preclinical studies investigating therapeutic efficacy of MSCs in AAA showed MSCs intervention could suppress aneurysm enlargement and reduce maximum aortic diameter in animals. Elastin is one of the essential extracellular matrix components in aortic wall, fragmentation and degradation of elastin contributes to AAA formation and progression [16], thus maintaining the integrity of arterial elastin is crucial for the prevention of AAA initiation and development [46]. In this present study, meta-analysis showed MSC administration increase the pooled extracellular arterial elastin contents compared to control group. Up-regulation of proinflammatory cytokines and MMPs contributes to aortic immune responses and promotes aortic dilation [47]. In our meta-analysis, the expression level of TNF-α, MCP-1, IL-6, MMP-2 and MMP-9 were suppressed after MSCs intervention in AAA models. Moreover, this study attempted to uncover the different therapeutic efficacy among included studies stratified by detailed study designs, including animal model establishment strategies (animal species, model induction methods) and intervention procedures (cell type, compatibility, route, dose, control type and followup duration).
Mesenchymal stem cell has been reported to be isolated from a variety of tissues and body fluids, including bone marrow, adipose tissue, placenta, umbilical cord, blood, urine, etc. [40]. Evidence showed MSCs from different sources exhibited distinct therapeutic effects despite similar bio-characteristics and morphology [48]. In this study, adipose derived MSCs showed better therapeutic efficacy in AAA animals than other cell types regarding to maximum aortic diameter, which may be attributed to the fact that adipose derived MSCs had more potent immunomodulatory effects than other MSCs [49,50]. Conservatively, given the underlying heterogeneity of the studies included in this study, we believe that more studies are needed to be implemented.
Studies comparing the therapeutic safety and efficacy between allogeneic and xenogeneic MSCs were limited, results of related comparative studies were controversial. María et al. [51] reported that allogeneic and xenogeneic transplantation of adipose derived MSCs shared equal efficacy in acute cerebral infarct model. Fatemeh et al. [52] showed that although there was no statistical difference, allogeneic MSCs transplantation was more efficient than xenogeneic MSCs in the abortion mouse model. One study investigated the immune response to autologous, allogeneic, and xenogeneic MSCs after intraarticular injection in horses, results showed that allogeneic MSCs elicited undetectable immune responses compared with xenogeneic MSCs [53], which highlights the different potential of triggering immune responses between allogeneic and xenogeneic MSCs. In addition, Choi et al. [54] revealed that the strongest humoral immune response was induced by xenogeneic MSCs in a murine systemic lupus erythematosus model compared to allogeneic and syngeneic MSCs. Moreover, Hwang et al. [55] addressed the diverse immunogenic properties and different patterns of immune responses of allogeneic and xenogeneic MSCs in mouse brain. In this study, our results showed significant improvement of allogeneic MSCs compared to xenogeneic MSCs in terms of maximum aortic diameter reduction and diameter change ratio. As mentioned above, the distinct therapeutic efficacy between allogeneic and xenogeneic MSCs in AAA might attribute to the different immunogenic potential since immune responses played the leading role in AAA initiation and progression [56].
Intravenous delivery is the most convenient and the least invasive route, however, the biggest concern for intravenous delivery is the retention in lung and other organs which causes insufficient cell migration and homing [57]. Kanelidis et al. [58] showed that local transendocardial injection of MSCs was superior to intravenous in myocardial infarction. Jeong et al. [59] found intraarterial injection provided increased benefits over local injection and intravenous injection. Apparently, current comparative studies are generally contradictory and insufficient to draw solid conclusion about the optimal delivery route. In our results, perivascular incubation presented the most valid therapeutic efficacy compared to intraluminal and intravenous delivery. The above evidence suggests that the effectiveness of cellular intervention modalities varies in different diseases, and it seems that the closer the first intervention is to the lesion site the more effective it is in AAA, and this observation needs to be corroborated by further studies.
Cell count is an important indicator to assess the clinical efficacy and safety of cell therapy. Kabat et al. [60] summarized MSCs based clinical trials from 2014 to 2018 and found that the minimal effective doses ranging from 70 to 190 million MSCs/patient/dose regarding to intravenous injection. In a meta-analysis investigating the therapeutic efficacy of MSCs in preclinical studies of sepsis, Lalu et al. [61] reported a favorable result of ≥ 1*10 6 cells dose compared to < 1*10 6 cells. However, opposite result was achieved in a similar meta-analysis [62]. In our results, it is revealed that ≥ 1*10 6 cells group was slightly more effective than < 1*10 6 cells group. Inevitably, current evidence is limited and more investigations should be implemented to get insights into the optimal cell dose of MSC intervention in AAA animals and human studies. Furthermore, during the translational processes from animal to human studies, it is recommended to adjust the optimal dose based on body surface area [63].
In present study, we found the therapeutic efficacy of MSC in AAA was decreasing over time duration, the most significant diameter decrease was achieved in ≤ 2 week (SMD = − 1.47) group, and it was more inferior in ≤ 4 weeks group (SMD = − 1.20), the result of > 4 weeks group was not significant. It is undeniable that all available animal models have limitations regarding to the pathological interpretation and natural history of human AAA [58], the subsequent time-dependent self-healing process exits in most AAA animal models once the stimulus stops [64]. However, their translational relevance should not be doubted, and above results underscore the necessity of developing novel AAA animal models and highlight the investigation of optimal intervention frequency in both animal and human AAA studies.
Although the results of our study are encouraging and instructive, several definite limitations exist. First, funnel plots revealed publication bias in this study, which indicates some missing data in the whole study and more relevant studies are needed. Second, the heterogeneity among studies was significant and individual studies was omitted to achieve eligible between-study heterogeneity, which might affect the accuracy of this study. Moreover, the quality assessments of included studies are relatively low since most studies lacked blinding and randomization methods. Furthermore, our meta-analysis focused on the diameter change as primary outcome, while most studies provided values of the final maximum aortic diameter, some studies reported the percentage change of maximum aortic diameter, which may contribute to the instability of our results. In addition, for secondary outcomes evaluation, too few studies provided available data about elastin contents and inflammatory cytokines, thus, it is suggested to be cautious when interpreting the results. Finally, considering the potential heterogeneity in animal studies, the random effects model was applied throughout this meta-analysis, which might overestimate the overall effects of MSC in AAA.

Conclusion
The present meta-analysis and systematic review included 18 studies, results showed that MSCs intervention is associated with suppression of aortic diameter enlargement, reduction of elastin degradation and inflammatory cytokines such as MCP-1, IL-6, TNF-α and MMP2/9, increase of TIMP1/2 expression. Regarding this, future large-scale animal studies are needed to determine the eligible protocols of transplantation details. Furthermore, clinical randomized investigations are required to improve current insight on the therapeutic efficacy of MSCs in AAA patients, thereby, potentially promoting clinical application of MSC based cell therapy in AAA.