Efficacy and Mode of Action of Mesenchymal Stem Cells in Non-Ischemic Dilated Cardiomyopathy: A Systematic Review

Non-ischemic dilated cardiomyopathy (NIDCM) constitutes one of the most common causes to non-ischemic heart failure. Despite treatment, the disease often progresses, causing severe morbidity and mortality, making novel treatment strategies necessary. Due to the regenerative actions of mesenchymal stem cells (MSCs), they have been proposed as a treatment for NIDCM. This systematic review aims to evaluate efficacy and mode of action (MoA) of MSC-based therapies in NIDCM. A systematic literature search was conducted in Medline (Pubmed) and Embase. A total of 27 studies were included (3 clinical trials and 24 preclinical studies). MSCs from different tissues and routes of delivery were reported, with bone marrow-derived MSCs and direct intramyocardial injections being the most frequent. All included clinical trials and 22 preclinical trials reported an improvement in cardiac function following MSC treatment. Furthermore, preclinical studies demonstrated alterations in tissue structure, gene, and protein expression patterns, primarily related to fibrosis and angiogenesis. Consequently, MSC treatment can improve cardiac function in NIDCM patients. The MoA underlying this effect involves anti-fibrosis, angiogenesis, immunomodulation, and anti-apoptosis, though these processes seem to be interdependent. These encouraging results calls for larger confirmatory clinical studies, as well as preclinical studies utilizing unbiased investigation of the potential MoA.


Introduction
Non-ischemic dilated cardiomyopathy (NIDCM) is a disease affecting the myocardial tissue and represents one of the most common causes to non-ischemic heart failure [1]. It is characterized by systolic dysfunction and left ventricle (LV) or biventricular dilatation, in the absence of factors normally involved in global systolic impairment, including coronary artery disease, hypertension and valve disease [2]. The prevalence of NIDCM remains uncertain, though estimates suggest the number to be between 1:250 and 1:2500 [3,4]. Two electronic databases were searched; Medline (PubMed) and Embase. The last search in both databases, was performed on 20 February 2020. The literature search was restricted to articles published in English. In PubMed, the applied MeSH terms were Mesenchymal Stem Cells AND Cardiomyopathy, Dilated OR Heart Failure, and all entry terms were included as free text. For the search in Embase, the following medical subject headings were included: Mesenchymal stem cell AND Congestive cardiomyopathy OR Heart failure. All narrower terms were included using the explode function. Both search syntaxes are provided in Appendix C.
Using Covidence online software, two independent reviewers screened all titles/abstracts, retrieved from the initial search, and subsequently all full texts (CH, BF). Discrepancies regarding inclusion were resolved by a third reviewer (SF). Two electronic databases were searched; Medline (PubMed) and Embase. The last search in both databases, was performed on 20 February 2020. The literature search was restricted to articles published in English. In PubMed, the applied MeSH terms were Mesenchymal Stem Cells AND Cardiomyopathy, Dilated OR Heart Failure, and all entry terms were included as free text. For the search in Embase, the following medical subject headings were included: Mesenchymal stem cell AND Congestive cardiomyopathy OR Heart failure. All narrower terms were included using the explode function. Both search syntaxes are provided in Appendix C.
Using Covidence online software, two independent reviewers screened all titles/abstracts, retrieved from the initial search, and subsequently all full texts (CH, BF). Discrepancies regarding inclusion were resolved by a third reviewer (SF).

Study Criteria
Published clinical and animal studies analyzing the use of MSC therapy in NIDCM were included. For animal studies, both medically induced NIDCM, genetic and inflammatory NIDCM models were included to represent the heterogeneity of the disease.
The predefined exclusion criteria for title/abstract screening were: (1) irrelevant to the subject of the study, (2) reviews and meta-analysis, and (3) letters to the editors and editorials. For full text screening they included: (a) wrong animal model or a suitable model but treatment prior to the onset of NIDCM phenotype, (b) data could not be extracted separately for NIDCM patients, (c) MSCs were differentiated or altered genetically for enhancement purposes, (d) no information regarding time point for treatment/completion was reported, and (e) full text/data not available. Studies administering labelled cells were included, but data was initially extracted and analyzed separately to accommodate the risk of labelling affecting cell function.

Data Extraction
To obtain the most comprehensive knowledge on MoA, different study types and outcomes were included, emphasizing measures for cardiac function and tissue, cellular, and molecular responses to treatment. Due to the heterogeneity in reporting of outcomes among eligible studies, meta-analysis was not attempted. Thus, the present systematic review aims to give a descriptive presentation of available data.

Protein-Protein Interaction Network
To elucidate key factors in MSC MoA, proteins significantly altered by treatment were subjected to Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) v11, applying a minimum required interaction score of 0.400 [24]. Data on protein expression from both clinical and preclinical studies were pooled, based on the assumption that the biological function of specific proteins are the same between species. The database comprises interaction records from curated websites, including Reactome, BioCyc, KEGG, and Gene Ontology as well as legacy data from PID and BioCarta [24]. The analysis shows known and predicted protein-protein interactions (PPI), in which the proteins are denoted by nodes and interactions by edges. To further explore the PPIs, biological processes associated with the protein expression were color coded.

Assessment of Study Quality and Publication Bias
Due to the inclusion of both clinical and animal studies, a modified SYRCLE's risk of bias (RoB) tool was applied [25]. Random sequence generation, baseline characteristics (animal studies), allocation concealment, blinding of participants and personnel, blinded outcome assessment, incomplete outcome data, and other bias were included in the tool. All eligible studies were assessed on all parameters and rated with either high RoB, low RoB, or not reported. Two independent reviewers rated all studies (CH and BF) and discrepancies regarding RoB assessment were resolved by a third reviewer (AQ).

Study Characteristics and Quality Assessment
The literature search initially yielded 956 studies, from which 134 were duplicates and thus removed. First, titles and abstracts were screened for relevancy and study type. Patient populations with ischemic heart disease or animal models of acute myocardial infarction were the main reasons for exclusion, leaving 66 studies for full-text evaluation. Ultimately, a total of 31 studies met the predefined inclusion criteria. Seven studies reporting data on the use of MSCs in clinical trials with NIDCM were included, though data from five of the studies originated from the same trial [12,[26][27][28][29]. To ensure that original data was only reported and analyzed once, publications from the same trial was considered as one study for the remaining analysis, giving a total number of 27 included studies.
The number of participants in the clinical trials ranged from 27 to 53 patients and follow-up ranged from one-week [30], to 12 months [12,23] post treatment. A total of 24 preclinical studies were included, with rodent models being the primary species used for NIDCM induction (Appendix D). Three studies used an autoimmune phenotype [31][32][33] and two a genetic phenotype [34,35]. The remaining studies applied a medically induced NIDCM phenotype.

Risk of Bias
A quality assessment was performed on all included studies and is presented in Appendix E. Regarding the preclinical studies, most studies report baseline characteristics, thus minimizing RoB in functional data. However, preclinical studies tend to omit information on random sequence generation and allocation concealment. Additionally, limited information was available on blinding of outcome assessment.
Five of the included animal studies had generally high RoB, as they scored low RoB in ≤1 of the assessed parameters [36][37][38][39][40]. Despite this, they reported the same treatment effect as the remaining studies. As study design generally tends to be insufficiently described in animal studies, none of them were excluded. If an outcome, however, was only reported in a high RoB study, it is noted in the text.
For the three included clinical trials, more information was available on study design, including random sequence generation and blinded outcome assessment, thus reducing RoB. Information on allocation concealment is here too sparse, which can be explained by some of the studies being open-labelled.

Cell and Transplant Type
A total of 21 of the included trials tested bone marrow derived MSCs (BM-MSCs), four tested human umbilical cord blood MSCs (hUCB-MSCs) and three tested adipose tissue-derived mesenchymal stem cells (AT-MSCs) (Appendix D). Two studies compared AT-MSCs and BM-MSCs [32,41]. Due to the frequency of BM-MSCs across studies, the results presented in this review primarily reflects the properties of this cell type. Despite this, both BM-MSCs, AT-MSCs, and hUCB-MSCs improved cardiac function and initiated anti-fibrotic, angiogenic, and immunomodulatory mechanisms [32,35,41]. Additionally, BM-MSCs and hUCB-MSC displayed anti-apoptotic properties [35]. As all MSC subtypes exhibited comparable features, it is reasonable to suggest that MoA is the same independent of cell origin.
Another aspect to consider is transplant type. A total of fourteen studies used syngeneic transplants, six used xenogeneic transplants, four used autologous transplants and three used allogeneic transplants. Recently, allogeneic MSCs have gained focus as a potential off-the-shelf therapy as they are immune evasive [12,20]. The POSEIDON-DCM trial compared autologous and allogeneic BM-MSCs and found allogeneic transplants to be superior in increasing LVEF and decreasing tumor necrosis factor α (TNF-α) [12]. Furthermore, autologous MSCs may have a reduced regenerative capacity, due to the underlying aetiology of the patient. This obstacle is avoided with allogeneic transplants from healthy donors. Despite this, most animal studies have used syngeneic and autologous transplants, while the three clinical trials have applied autologous and allogeneic transplants.

Administration Route
Different administration routes have been applied in the included studies. 14 studies used intramyocardial (IM) injection, 10 intravenous (IV) injection, two used intracoronary (IC) injection and one used injection in the hind limb muscle. Studies using IM and IV injections were equally distributed across outcomes related to cardiac function, fibrosis, angiogenesis, immunomodulation, and apoptosis. However, only one study using IV injections showed that MSC therapy led to altered fibrotic gene expression [22]. In spite of this, IV injections led to reduced cardiac fibrosis in five studies, suggesting that molecular alterations were present.
Overall, the included studies do not reflect any significant changes between IM and IV injections. There is, however, clinical evidence of improved retention and functional outcome in NIDCM patients with IM delivery compared to IC administration of CD34+ cells [42]. This might be similar for MSCs. However, direct meta-analysis and comparison of delivery routes was not within the scope of this review.

Cell Labelling
In order to track MSCs in vivo, cell labelling was applied in 13 out of the 27 included studies. Only one study stated that labelling did not affect cell viability and function [31,35,[38][39][40][41][43][44][45][46][47][48][49]. As the effect of labelling was not addressed in 12 out of 13 studies, the initiating analysis was performed on labelled and unlabeled cells separately. Prior to the final synthesis of results, we evaluated outcomes in all studies using labelled cells, and compared data to studies using unlabeled cells. All included studies, except two, reported improved cardiac function, suggesting that labelling did not affect the overall MSC function [45,50]. Approximately half of the studies reporting anti-fibrotic and angiogenic properties of MSCs used labelled cells, indicating that these properties remained intact [31,35,38,39,41,43,44]. Two out of three studies investigating apoptosis, reference [35,49] and two out of three investigating immunomodulation [35,41] used labelled cells and reported similar tendencies. Consequently, there was no evidence of labelling affecting cell function, thus data was assessed coherently for the remaining analysis.

Clinical Evidence of Functional Effect
Fatkhudinov et al. [30] evaluated the effect of allogeneic MCSs in 27 patients with NIDCM, advanced heart failure, and LVEF < 35%. A total of 14 patients were treated conservatively, and 13 underwent surgical procedure. Both groups were subdivided into a group receiving IC cell transplantation and a control group. All patients were followed for 12 months. MSC transplantation was associated with improved 6-min walk test and NYHA class, reaching maximum effect by month three (numerical values not provided, p < 0.05). Increased LVEF was noted in the MSC group but did not reach statistical significance. No change in left ventricular size or volume was present.
Xiao, et al. [23] compared the efficacy of IC administration of bone marrow mononuclear cells (BM-MNCs) or BM-MSCs in patients with NIDCM and LVEF < 40%. A total of 53 patients were randomized into three groups receiving IC infusion of BM-MNCs (n = 16), BM-MSCs (n = 17) or normal saline (n = 20). Patients in the BM-MSC group exhibited a significant improvement in cardiac function, as LVEF increased from 34.1 ± 3.6 to 41.4 ± 5.1 at three-month follow-up (p < 0.05) and to 41.0 ± 6.7 at 12-month follow-up (p < 0.05). Furthermore, NYHA class decreased from 2.7 ± 0.7 to 1.7 ± 0.7 and 1.9 ± 1.1 at three-and 12-month follow-up, respectively (p < 0.05). Patients receiving BM-MNCs also presented with improvement in LVEF and NYHA class, but less prominent and only statistically significant at three-month follow-up.
Hare et al. [12] performed the first randomized trial directly comparing the effects of autologous and allogeneic BM-MSCs therapy in NIDCM (POSEIDON-DCM: Percutaneous Stem Cell Injection Delivery Effects on Neomyogenesis in Dilated Cardiomyopathy). A total of 37 patients with stable heart failure and LVEF < 40% were randomized in a 1:1 ratio to receive transendocardial injections of a fixed dose (100 × 10 6 ) of either autologous or allogeneic MSCs. After 12 months, LVEF of patients receiving allogeneic MSCs had significantly improved by 8.0 percentage points (p = 0.004) and the 6-min walk-test had improved by 37 m (p = 0.04). Patients receiving autologous MSCs did not improve to the same extent, as LVEF increased 5.4 percentage points (p = 0.116). Furthermore, the 12-months all-cause rehospitalization rates and the rate of major adverse cardiovascular events were significantly lower with allogeneic MSC therapy (28.2% and 20.3%, respectively) (p < 0.05) compared to autologous (70% and 57.1%, respectively). A sub-analysis by Florea et al. [29] demonstrated that the effects of MSC therapy on cardiac function and clinical outcomes are comparable in male and female patients. This finding was present despite differences in baseline clinical characteristics. In addition, genetic sequence analysis revealed that the effect of MSC treatment was associated with genetic variants, including mutations in the cytoskeleton, nuclear membrane, sarcomere, and mitochondria. At 12-month follow-up, LVEF increased by 13.6% in the patients with no pathological variants (n = 6, p = 0.002), compared to variants of uncertain significance (+6.5%, n = 20, p = 0.005), and patients positive for pathological variants (−5.9%, n = 8, p = 0.2).
This suggests that the genetic profile of NIDCM patients plays a role in responsiveness to MSC therapy, and that genetic testing can be used before considering this therapy [26].

Preclinical Evidence of Functional Effect
One preclinical study did not investigate functional outcomes [33], and two studies found no functional effect of the MSC treatment [45,50]. The remaining 21 included preclinical studies reported a significant effect on functional parameters following MSC treatment, primarily measured as LVEF. Figure 2 illustrates the preclinical studies reporting numerical values for LVEF at follow-up. Most studies reporting both mean and standard deviation found the difference in LVEF to be between 15 and 6 percentage points, when comparing MSC groups to controls at follow-up [22,35,48,49]. Studies including baseline values reported ∆LVEF between 25.2 and 1.9 for the treatment groups, depending on the animal model, with a median at 13.6. It was not within the scope of this review to perform a meta-analysis. However, a recent meta-analysis by Lopes et al. [51] found MSC therapy to result in a weighted difference of 10.4 (7.24-12.84) percentage points in LVEF compared to controls. This is in accordance with the clinical results from POSEIDON-DCM on ∆LVEF, with allogeneic treatment of patients without pathological variants. Ventricular pressure was measured in six studies [31,33,36,41,46,48]. In five of these studies, MSC treatment resulted in significantly increased dP/dt, indicating increased cardiac contractility. This finding is consistent with the observed improvements in LVEF. Additionally, LV end diastolic pressure was significantly decreased in the three studies reporting on this outcome [31,33,36]. Arterial blood pressure was increased to normal levels in Ammar et al. [41] and tended to normalize in Psalitis et al. [43].
Taken together, the evidence suggests that MSC treatment improves functional outcomes of cardiac pump function and blood pressure. This is solid evidence for initiating phase I clinical trials. However, knowledge about MoA is necessary to move into larger clinical studies. With this in mind, we investigated the published evidence on MoA.

Effect on MSC Therapy on Cardiac Regeneration
Despite the distinct outcomes included in the present review, the majority was related to four aspects of cardiac regeneration including, fibrosis, immunomodulation, angiogenesis, and apoptosis ( Figure 3).

Effect on MSC Therapy on Cardiac Regeneration
Despite the distinct outcomes included in the present review, the majority was related to four aspects of cardiac regeneration including, fibrosis, immunomodulation, angiogenesis, and apoptosis ( Figure 3).

Fibrosis
Data extraction revealed that cardiac fibrosis and the anti-fibrotic effects of MSCs have received great attention. A total of 14 out of 27 studies evaluated cardiac fibrosis using immunohistochemistry (IHC) and analyzed gene and protein expression patterns related to this process ( Figure 3).
Looking at gene and protein expression patterns, MSC transplantations significantly downregulated the gene expression of collagen 1 and 3 and transforming growth factor β (TGF-β) in the cardiac tissue four weeks after treatment ( Figure 4) [34,35,38,39,52]. At 10 weeks, the gene expression of collagen 3 was upregulated, but the protein expression reduced, which might be explained by a temporal shift [22]. These findings indicate that MSC transplantations inhibit collagen transcription and subsequently collagen synthesis and deposition, resulting in the reduced CVF. In addition to this, Deng et al. [52] reported that MSC transplantation reduced TGF-β transcription with 88.8% (p < 0.05). This finding was supported by Yu et al. [53], which likewise found inhibited TGF-β transcription following MSC treatment. An increased TGF-β expression is often associated with activation of fibrotic pathways; hence, the attenuated cardiac fibrosis may partially be mediated by alterations of TGF-β signaling [49,54]. Results demonstrated that MSC therapy alters the fibrotic process in NIDCM on both a tissue and molecular level [32,43,44]. MSC treatment significantly attenuated myocardial fibrosis, by reducing collagen volume fraction (CVF) and improving myocardial fiber alignment on IHC. This is, together with the positive effect on cardiac function, the most consistent finding, reported by all but one study investigating fibrosis [22,31,32,[34][35][36][37]39,43,44,47,49,52].
Looking at gene and protein expression patterns, MSC transplantations significantly downregulated the gene expression of collagen 1 and 3 and transforming growth factor β (TGF-β) in the cardiac tissue four weeks after treatment ( Figure 4) [34,35,38,39,52]. At 10 weeks, the gene expression of collagen 3 was upregulated, but the protein expression reduced, which might be explained by a temporal shift [22]. These findings indicate that MSC transplantations inhibit collagen transcription and subsequently collagen synthesis and deposition, resulting in the reduced CVF. In addition to this, Deng et al. [52] reported that MSC transplantation reduced TGF-β transcription with 88.8% (p < 0.05). This finding was supported by Yu et al. [53], which likewise found inhibited TGF-β transcription following MSC treatment. An increased TGF-β expression is often associated with activation of fibrotic pathways; hence, the attenuated cardiac fibrosis may partially be mediated by alterations of TGF-β signaling [49,54]. Another aspect of ventricular remodeling is turnover of fibrotic tissue, which is partially regulated by matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) [38]. Several studies found that MSC treatment significantly reduced the gene and protein expression of MMP-2 and MMP-9 [31,34,38,39]. However, MMP-9 reduction was only reported in a study scoring high RoB [38,39]. Shabbir et al. [34] found reduced mRNA expression of MMP-9 (p < 0.001), MMP-13 (p < 0.01), TIMP-2 (p < 0.05) and TIMP-3 (p < 0.05), compared to the NIDCM control, thus reversing the pathological expression profile associated with NIDCM. In the failing heart, both MMPs and TIMPs contribute to adverse remodeling by degrading normal collagens, which is subsequently replaced by interstitial fibrosis comprising poorly cross-linked collagens [39]. By downregulating the expression of MMPs and TIMPs, MSCs may inhibit the progression of ventricular remodeling and dilation, thus improving cardiac function. Together these findings substantiate the anti-fibrotic properties of MSCs and the advantage of applying them therapeutically to target the fibrotic nature of NIDCM.

Immunomodulation
Only three out of the twenty-seven included studies analyzed outcomes related to the immune system; two preclinical studies and POSEIDON-DCM ( Figure 3) [12,35,41].
Using a doxorubicin induced NIDCM phenotype in diabetic rats, Ammar et al. [41] found that MSC transplantations significantly reduced % area of immune cell infiltration in the myocardium (p < 0.05). Using a genetic phenotype, Gong et al. [35] established that MSCs significantly reduced serum C-reactive protein (p < 0.05). The POSEIDON-DCM study found that treatment with allogeneic MSCs significantly decreased serum levels of TNF-α with −10.6 ± 1.6 pg/mL at six-months follow-up (p < 0.0001). Elevated serum levels of TNF-α are associated with progression of heart diseases, therefore, by reducing pro-inflammatory cytokines in the myocardium, MSCs may shift the microenvironment towards an anti-inflammatory profile [12]. The study likewise found that allogeneic MSC therapy altered the humoral lymphocyte profile by reducing subtypes of both B and T cells, normally associated with chronic inflammation. Considering the fundamental role of the immune system in Another aspect of ventricular remodeling is turnover of fibrotic tissue, which is partially regulated by matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) [38]. Several studies found that MSC treatment significantly reduced the gene and protein expression of MMP-2 and MMP-9 [31,34,38,39]. However, MMP-9 reduction was only reported in a study scoring high RoB [38,39]. Shabbir et al. [34] found reduced mRNA expression of MMP-9 (p < 0.001), MMP-13 (p < 0.01), TIMP-2 (p < 0.05) and TIMP-3 (p < 0.05), compared to the NIDCM control, thus reversing the pathological expression profile associated with NIDCM. In the failing heart, both MMPs and TIMPs contribute to adverse remodeling by degrading normal collagens, which is subsequently replaced by interstitial fibrosis comprising poorly cross-linked collagens [39]. By downregulating the expression of MMPs and TIMPs, MSCs may inhibit the progression of ventricular remodeling and dilation, thus improving cardiac function. Together these findings substantiate the anti-fibrotic properties of MSCs and the advantage of applying them therapeutically to target the fibrotic nature of NIDCM.

Immunomodulation
Only three out of the twenty-seven included studies analyzed outcomes related to the immune system; two preclinical studies and POSEIDON-DCM ( Figure 3) [12,35,41].
Using a doxorubicin induced NIDCM phenotype in diabetic rats, Ammar et al. [41] found that MSC transplantations significantly reduced % area of immune cell infiltration in the myocardium (p < 0.05). Using a genetic phenotype, Gong et al. [35] established that MSCs significantly reduced serum C-reactive protein (p < 0.05). The POSEIDON-DCM study found that treatment with allogeneic MSCs significantly decreased serum levels of TNF-α with −10.6 ± 1.6 pg/mL at six-months follow-up (p < 0.0001). Elevated serum levels of TNF-α are associated with progression of heart diseases, therefore, by reducing pro-inflammatory cytokines in the myocardium, MSCs may shift the microenvironment towards an anti-inflammatory profile [12]. The study likewise found that allogeneic MSC therapy altered the humoral lymphocyte profile by reducing subtypes of both B and T cells, normally associated with chronic inflammation. Considering the fundamental role of the immune system in NIDCM and the immunomodulatory properties of MSCs, surprisingly few studies have reported on this aspect.

Angiogenesis
From the included studies, 11 out of 27 have analyzed outcomes related to angiogenesis. Microscopically, studies have demonstrated that MSC transplantations increased number and density of vessels in the myocardium [31,32,34,35,41,43]. These findings suggest that MSC transplantations activate an angiogenic response, leading to increased myocardial neovessel formation. On a molecular level, MSCs increased the cardiac gene expression of vascular endothelial growth factor (VEGF), which translated into increased serum VEGF [32,34,35,39,44,55] VEGF is an important signaling protein secreted to stimulate neovessel formation, thus the increased vessel density may be partly due to the increased VEGF production [32]. Additionally, MSC treatment has been shown to increase the circulating levels of hepatocyte growth factor (HGF), a potent angiogenic factor [31,34,55]. This finding was also present on a transcriptional level, likely mediating the increased circulating HGF [39]. This finding was, however, only reported in a study scoring high RoB.
Another aspect of the angiogenic response is endothelial function [12]. Endothelial dysfunction is a significant feature of heart failure, leading to diminished endothelial progenitor cell function and flow-mediated vasodilation (FMD) [27]. The POSEIDON-DCM trial demonstrated that allogeneic MSC therapy significantly improved endothelial function by increasing endothelial progenitor colony forming units (p = 0.0107) and FMD% (p = 0005) at three months compared to baseline [12]. Studies revealed that MSC therapy increased the ventricular protein expression of endothelial nitric oxide synthase (eNOS) (p < 0.05), an enzyme important for proper endothelial function [47,49]. Furthermore, eNOS was significantly decreased in the NIDCM control group, which is associated with reduced myocardial neovascularization and impaired endothelium-dependent vasodilation, thus supporting the results from POSEIDON-DCM [49]. The ability of allogeneic MSCs to restore endothelial function, together with the alterations in angiogenic factors, offer new insights into MSC-induced angiogenesis.
In all, these findings provide solid evidence that MSC therapy induces angiogenesis in NIDCM, likely stimulated by an increased paracrine secretion and improved endothelial function.

Apoptosis
Three out of the included twenty-seven studies investigated outcomes related to cell survival and apoptosis [34,35,49]. Shabbir et al. [34] performed IHC on myocardial tissue sections, which showed that MSC therapy reduced apoptotic cardiomyocytes in NIDCM hearts by approximately 60% (p < 0.01) compared to controls. On a molecular level, MSC therapy increased the ventricular B-cell lymphoma 2 (Bcl-2)/Bcl-2-associated X protein (Bax) ratio (p < 0.01) and reduced protein expression of Caspase-3 compared to NIDCM controls (p < 0.05) [35,49]. Bcl-2 is an important inhibitor of apoptosis among ventricular cardiomyocytes, whereas Bax is a pro-apoptotic protein. Consequently, an increased Bcl-2/Bax ratio suggests inhibition of pathways involved in cardiac apoptosis [35]. Caspase-3 is activated during cell apoptosis and has specifically been associated with doxorubicin administration in vivo [56]. However, Mohamed et al. [49], which reported reduced Caspase-3 protein levels following MSC treatment, used an Isoproterenol-induced NIDCM phenotype. These findings suggest that the same mechanisms are active in both NIDCM models, and beyond this, that MSC-mediated inhibition of Caspase-3 may reduce cardiomyocyte apoptosis and subsequently improve cardiac function.

MSC Mode of Action in NIDCM
The results demonstrate that most of the included studies have evaluated MSC efficacy and MoA approximately one month following treatment (Figure 4 and Appendix D). At this timepoint studies found improved cardiac function, reduced fibrosis, and increased myocardial capillary density. These tendencies suggest that the molecular and cellular mechanisms underlying these effects, have been initiated within the first weeks after treatment.
When reporting on MoA, studies most commonly discuss specific regenerative mechanisms as isolated processes. However, when analyzing outcomes associated with fibrosis, angiogenesis, apoptosis, and immunomodulation, it becomes evident that these processes are mutually connected. To exemplify this, the POSEIDON-DCM trial found that allogeneic MSCs reduced serum TNF-α, while Mohamed et al. [49] reported reduced ventricular Caspase-3 protein following treatment [12]. Interestingly, release of TNF-α, has been described to activate Caspase-3 and subsequently stimulate progression of cardiomyocyte apoptosis. It is therefore likely that the reduced Caspase-3 is in part mediated by a decreased serum TNF-α, thus shifting the inflammatory microenvironment and alleviating cardiomyocyte apoptosis [12,35,49]. The biological properties of TGF-β likewise exemplifies the complexity of MSC-mediated cardiac regeneration. Most of the included studies describe its involvement in cardiac fibrosis, thus suggesting that downregulation is beneficial [10,52,53]. However, TGF-β is also described to be anti-inflammatory, as it can promote differentiation of anti-inflammatory macrophages and inhibit cytotoxic T cells in the damaged heart [20,27,28]. Based on this, reporting increased or decreased TGF-β expression to be solely beneficial or detrimental may be oversimplified and not reflective of the complex processes in vivo.
To gain further knowledge on the complex biological processes initiated by MSC therapy, a STRING analysis was performed. The analysis provided a PPI enrichment p-value of < 1.0 × 10 −16 , indicating that the proteins, whose expression was altered by treatment, are biologically connected, and not randomly occurring. The connectivity also points towards that similar processes are initiated following MSC treatment, despite varying MSC types and NIDCM models. As illustrated in Figure 5, several of the proteins are implicated in numerous physiological processes including ECM organization (purple nodes) and angiogenesis (red nodes). These mechanisms are most likely accountable for the observed increase in vessel density and decrease in cardiac fibrosis. The central placement and multiple connections of VEGFA, TNF-α, and IGF-1 point towards the initiated mechanisms being conducted through regulation of these factors. However, knowledge regarding which cell populations are responsible for the changed proteins levels is poorly investigated. Due to the notoriously low retention rates of MSCs in the heart, it is likely that the examined proteins are secreted by endogenous cell populations and not MSCs themselves [42].
Though little has been reported on immunological and apoptotic markers, the STRING analysis support that MSC therapy exerts immunomodulation (green nodes) and alters apoptotic processes (yellow nodes) in NIDCM. However, the downstream effects of these remain uncertain, underpinning the need for deeper exploration of MoA. All things considered, the effect of MSC therapy cannot be ascribed to one single growth factor or limited to one physiological process, but instead is the result of different regenerative processes, which may act synergistically [57].
Despite fibrosis, angiogenesis, apoptosis, and immunomodulation being the primary focus of existing studies, other aspects of MSCs may be fundamental to the observed improvement in cardiac function. Oxidative stress has been described to be one of the major mechanisms through which the anthracycline, doxorubicin, injures the heart [10,56]. Doxorubicin interacts with eNOS, and with increasing concentrations, eNOS can switch from generation of nitric oxide (NO) to superoxide, a reactive oxygen species contributing to endothelial dysfunction [28,47,56]. Endothelial function is often measured using endothelial progenitor cell-colony forming units (EPC-CFU), and has been found to be inversely correlated with serum TNF-α. Administration of MSCs increased peripheral blood EPC-CFUs, reduced serum TNF-α, and normalized ventricular eNOS protein expression. These findings suggest that MSC therapy can alleviate anthracycline-induced endothelial dysfunction and oxidative stress, possibly by restoring eNOS function [12,28,47,49]. Thus, the functional improvement may partially be caused by improved endothelial function and reduced oxidative stress. Using the STRING analysis, it emerged that seven of the included proteins were implicated in oxidative stress and ROS regulation, strengthening this hypothesis. Though little has been reported on immunological and apoptotic markers, the STRING analysis support that MSC therapy exerts immunomodulation (green nodes) and alters apoptotic processes (yellow nodes) in NIDCM. However, the downstream effects of these remain uncertain, underpinning the need for deeper exploration of MoA. All things considered, the effect of MSC therapy cannot be ascribed to one single growth factor or limited to one physiological process, but instead is the result of different regenerative processes, which may act synergistically [57].
Despite fibrosis, angiogenesis, apoptosis, and immunomodulation being the primary focus of existing studies, other aspects of MSCs may be fundamental to the observed improvement in cardiac function. Oxidative stress has been described to be one of the major mechanisms through which the anthracycline, doxorubicin, injures the heart [10,56]. Doxorubicin interacts with eNOS, and with increasing concentrations, eNOS can switch from generation of nitric oxide (NO) to superoxide, a reactive oxygen species contributing to endothelial dysfunction [28,47,56]. Endothelial function is often measured using endothelial progenitor cell-colony forming units (EPC-CFU), and has been found to be inversely correlated with serum TNF-α. Administration of MSCs increased peripheral

Challenges, Limitations and Future Perspectives
At this point, only a few small clinical trials have been conducted. Though the results are encouraging, there is a need for larger, international, trials, enabling inclusion of more patients. These should be performed to confirm the beneficial effects of MSC treatment in NIDCM patients and move forward in the drug development pipeline. However, to initiate larger trials, more knowledge on MoA is required.
As stated previously, 12 out of 13 studies did not analyze the effect of cell labeling on MSC viability and function. Since current literature reports labelling to affect these exact two properties, this issue should be evaluated in future studies attempting to address MoA [58,59]. If the fundamental functions are in fact altered, the results presented here may not uncover the full potential of MSC therapy in NIDCM.
The included studies show that MSC therapy improves cardiac function, ameliorates myocardial fibrosis and stimulates angiogenesis [31,32,34,35,41,43]. Despite the solid evidence of these properties, most studies build upon histochemical evaluation of cardiac tissue sections taken from animals, in which the same outcomes are addressed at approximately the same timepoint (Figures 3 and 4). This tendency elucidates the reproducibility of the results but fails to provide further mechanistic insight. Additionally, little attention has been paid towards the immunomodulation. This aspect of MSC-mediated regeneration in NIDCM, thus, seems an evident topic for future research on MoA, due to the fundamental role of the immune system in NIDCM. However, as the immunomodulatory properties are extensively described in other cardiac diseases, including ischemic heart disease, one may raise the question, if results are excluded from published articles due to non-significant findings or simply lack of focus on this topic [17,20,60]. All things considered, the existing MoA data favors a more explorative approach in future research, in which the immediate molecular and especially cellular processes should be prioritized. Furthermore, it is fundamental that the currently known MoA is being evaluated in future clinical trials, in order to translate findings into human.

Conclusions
MSC therapy has emerged as a promising treatment strategy for patients with NIDCM, due the degenerative nature of the disease and the regenerative properties of MSCs. The present systematic review provides evidence that MSC therapy has the potential to improve cardiac function, reduce myocardial fibrosis and increase angiogenesis. Further insight into MoA displays that MSCs induce both molecular and tissue alterations, initiating multiple physiological processes which act simultaneously to stimulate cardiac regeneration. However, given the limited amount of clinical trials and mechanistic data, further research is warranted to elucidate the effect in humans and the complete MoA underlying the functional improvement.  Any other kind of stem cell therapy than MSCs, also iPS-MSCs iii.
Modified MSCs (preconditioning, gene modified, differentiated etc.), unless data from a control group with "normal" MSCs can be extracted separately iv.
Any kind of co-intervention, regardless of character v.
Patients with arrythmogenic and hypertrophic cardiomyopathies vii.
Reviews viii.
Editorial comments Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale.

2-3
Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched.

2-3
Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated.

Study selection 9
State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta-analysis).

2-3
Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators.

Data items 11
List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made.

12
Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.
3 + Appendix E Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means). N/A

Synthesis of results 14
Describe the methods of handling data and combining results of studies, if done, including measures of consistency (e.g., I 2 ) for each meta-analysis.

Risk of bias across studies 15
Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective reporting within studies).

Additional analyses 16
Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta-regression), if done, indicating which were pre-specified.

Study selection 17
Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram.

Study characteristics 18
For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and provide the citations. Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta-regression (see Item 16)).

Summary of evidence 24
Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (e.g., healthcare providers, users, and policy makers).

4-12
Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of identified research, reporting bias).

Conclusions 26
Provide a general interpretation of the results in the context of other evidence, and implications for future research.

Funding 27
Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review. Intramyocardial No significant improvement in heart function.
BM-MSCs were present in the myocardium after two weeks and