Can cell-based therapies bridge the gap between research and reality in the treatment of myocarditis? A systematic review and meta-analysis

Abstract

Introduction

Myocarditis is a non-ischaemic myocardial injury caused by infectious agents, immune-mediated diseases, cardiotoxic drugs, and vaccines. Treatment options are largely supportive, with emphasis on close monitoring, exclusion of other diseases explaining symptoms, adherence to heart failure treatment recommendations, and consideration of etiology-directed interventions, if applicable, such as antiviral or immunosuppressive agents, steroids, discontinuation of antineoplastic therapy, and strategies to increase left ventricular (LV) ejection fraction (LVEF). In this context, cell-based therapies (CBTs) have emerged as a new therapeutic strategy and were believed to fill this gap. However, results were inconsistent across studies, which necessitates the need for rigorous systematic reviews to analyze available evidence comprehensively.

Methods

The review protocol was registered on the PROSPERO website, and the study was conducted in accordance with PRISMA regulations. Searches were conducted in Embase, CINAHL Plus via EBSCO host, Web of Science, Scopus, and PubMed. A total of 60 original papers published between 2004 and 2022 in 48 peer-reviewed journals were included in the meta-analysis to determine the efficacy of CBTs on the LV fractional shortening (LVFS), LVEF, capillary density (CD), inflammatory cell infiltration rate (ICIR), and fibrotic area (FA). The risk of bias (RoB) and study quality were assessed using the SYRCLE RoB tool and CAMARADES checklist, respectively. As the preliminary assessment indicated substantial heterogeneity among the included studies, a random-effects model was applied to pool effect sizes. Subgroup analyses were performed to explore potential sources of heterogeneity across studies. Publication bias was assessed by visual inspection of funnel plots for asymmetry and statistically evaluated using Egger's regression test.

Results

In total, the adapted and optimized search strategy retrieved 14,503 records from five databases. The majority of studies exhibited an unclear or high risk of bias in several domains, particularly selection bias and performance bias. Quality assessment revealed that only four studies (6.7%) were classified as high quality and four (6.7%) as low quality, while the remaining 52 studies (86.7%) were rated as moderate quality, with scores ranging from 4 to 6. CBTs improved LVFS (%) by 7.17 [95 % CI: 5.67, 8.66], LVEF (%) by 9.00 [95 % CI: 7.03, 10.97], CD (capillaries/mm²) by 300.50 [95 % CI: 45.01, 555.99] and decreased ICIR (cells/mm²) and FA (%) by −178.99 [95 % CI: −225.91, −132.08] and −6.04 [95 % CI: −6.83, −5.25], respectively. However, significant heterogeneity between studies was maintained at I² = 84-99%.

Conclusion

Despite significant heterogeneity and moderate publication bias, the results were very encouraging, as reflected in the consistent effect directions across all studies in terms of cardiac function and histology. Overall, our results demonstrated the need for well-designed studies with adequate animal sample sizes, a standardized approach to reporting, and mechanistic investigations that directly link structural remodeling to functional recovery. Addressing these issues will be critical steps for conducting large-scale clinical trials.

Graphical abstract

Highlights of the study

• •CBTs improved LVFS and LVEF by 7.17% [95% CI: 5.67, 8.66] and 9.00% [95% CI: 7.03, 10.97], respectively.
• •CBTs reduced FA by 6.04 % [95% CI: −6.83, −5.25] and ICIR by 178.99 cells/mm² [95% CI: −225.91, −132.08].
• •CBT-recieved LVs showed an increase of 300.5 capillaries/mm² [95% CI: 45.01, 555.99] compared with controls.
• •Substantial to considerable heterogeneity was observed across pooled analyses (I² ranging from 84% to 99%).

Non standard abbreviations and acronyms

ADSCs

adipose-derived stem cells

AIC

antracycline-induced cardiomyopathy

AMI

acute myocardial injury

BMM2

bone marrow-derived M2-like macrophages

BMNCs

bone marrow mononuclear cells

CAMARADES

Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies

CAR

chimeric antigen receptor

CBTs

cell-based therapies

CD

capillary density

CDCs

cardiosphere-derived cell

cMRI

cardiac magnetic resonance imaging

CST

cell-sheet transplantation

CVB3

coxsackievirus B3

DCM

dilated cardiomyopathy

DCs

dendritic cells

DOX

doxorubicin

EAM

experimental autoimmune myocarditis

ECM

extracellular matrix

EHT

engineered heart tissue

EMB

endomyocardial biopsy

FA

fibrotic area

GSDM

gasdermin

HF

heart failure

hs-cTn

high sensitive cardiac troponins

iACS

scaffold-free-induced adipocyte cell-sheet

ICIR

inflammatory cell infiltration rate

ICIs

immune checkpoint inhibitors

iDCM

ischaemic dilated cardiomyopathy

IHD

ischaemic heart disease

JAK

Janus kinase

LVEF

left ventricular ejection fraction

LVFS

left ventricular fractional shortening

MD

mean difference

miDCM

myocarditis-induced dilated cardiomyopathy

MSCs

mesenchymal stem cells

NETs

neutrophil extracellular traps

niDCM

non-ischaemic dilated cardiomyopathy

PICO

participants, investigation, comparator, outcome

PRISMA

Preferred Referred Items for Systematic Reviews and Meta-Analyses

QQMNCs

Quality- and Quantity-controlled peripheral blood mononuclear cells

QQMNCs + F-sheet

skin fibroblast sheet with QQMNCs

RA

route of administration

RoB

risk-of-bias

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

SD

standard deviation

SMCs

smooth muscle cells

SMD

standardized mean difference

SMI

subacute mycordial injury

SR

systematic review

SYRCLE

SYstematic Review Centre for Laboratory animal Experimentation

T. cruzi

Trypanosoma cruzi

TRCs

T regulatory cells

1. Introduction

Myocarditis is a non-ischaemic myocardial injury caused by infectious agents, immune-mediated diseases, cardiotoxic drugs, and vaccines [1]. Traditionally, children and young adults were considered more susceptible to myocarditis [2]. However, this notion may have lost relevance with the expanding spectrum of patients exposed to modern anti-cancer therapies, such as immune checkpoint inhibitors (ICIs) and chimeric antigen receptor (CAR) T-cell therapies [3].

Timely diagnosis and adequate treatment of myocarditis are often hindered by multiple factors. Because myocarditis can be triggered by diverse etiological agents, its clinical presentation is highly variable, ranging from chest pain and flu-like symptoms to cardiogenic shock and sudden cardiac death [1]. Even when acute or fulminant myocarditis is definitively diagnosed, treatment options are largely supportive, with emphasis on close monitoring, exclusion of other diseases explaining symptoms, adherence to heart failure (HF) treatment recommendations, and consideration of etiology-directed interventions, if applicable, such as antiviral or immunosuppressive agents, steroids, discontinuation of antineoplastic therapy, and strategies to increase left ventricular (LV) ejection fraction (LVEF) [[4], [5], [6]]. This therapeutic limitation may explain why approximately one in five cases progresses to dilated cardiomyopathy (DCM) [1] and necessitates the development of new treatment strategies.

In this context, the emergence of stem cells was believed to be able to fill not only this gap, but also other, less treatable nosologies, thanks to their safety and potential efficacy. Therefore, over the past two decades, cell-based therapies (CBTs) have been introduced actively into nearly all fields of biomedical research [7], including cardiovascular diseases [8,9]. Numerous preclinical studies and clinical trials have demonstrated the safety and efficacy of CBTs in the treatment of cardiomyopathies, while other studies remain controversial [10]. Although some studies have reported functional and structural improvements, as well as a reduction in the incidence of malignant arrhythmias and mortality, other studies have found neither prevention of adverse cardiac remodeling nor a decline in mortality; on the contrary, a link with life-threatening complications has even been identified [11]. This inconsistency in study results may be due to significant heterogeneity in terms of sample size, disease model, cell type, cell delivery route, treatment duration, follow-up time, and outcome measurement [10].

The above concerns about CBTs underscore the need for rigorous systematic reviews (SRs) that analyze available evidence comprehensively to avoid over- or underestimation of therapeutic efficacy and to guide health professionals in the right direction. While several high-quality SRs have evaluated CBTs in myocardial infarction and in ischemic, non-ischemic, congenital, and idiopathic DCM (Table 1), none have focused specifically on myocarditis or myocarditis-induced dilated cardiomyopathy (miDCM). Furthermore, consensus is lacking regarding the optimal timing of cell delivery, particularly whether early intervention is more effective than delayed treatment in mitigating myocardial injury and preserving cardiac function in myocarditis and miDCM.

Table 1.

Systematic reviews on CBTs addressing cardiovascular diseases. ∗† Studies primarily focused on the efficacy of mesenchymal stem cells (MSCs) and adipose-derived stem cells (ADSCs), respectively, in various pathological conditions, rather than solely on cardiac diseases.

Study	Study design	PRISMA consistency	Target pathology	Types of reports included	Number of reports included
Alvarez PA et al., 2013 [11]	SR	Not mentioned	MI/iDCM	Clinical	35
Bai Y et al., 2010 [12]	SR&MA	Compatible	MI	Clinical	10
Wang C et al., 2019 [10]	SR&MA	Not mentioned	iDCM/niDCM	Clinical	20
Delewi R et al., 2013 [13]	SR&MA	Not mentioned	MI	Clinical	24
Diaz-Navarro R et al., 2019 [14]	SR&MA	Compatible	niDCM	Clinical	13
Fan M et al., 2019 [15]	SR&MA	Compatible	HF	Clinical	9
Fisher SA et al., 2013 [16]	SR&MA	Compatible	MI	Clinical	9
Fisher SA et al., 2014 [17]	SR&MA	Compatible	IHD/HF	Clinical	23
Fisher SA et al., 2016 [18]	SR&MA	Compatible	IHD/HF	Clinical	38
Fu H et al., 2020 [19]	SR&MA	Compatible	HF	Clinical	6
Gho JMIH et al., 2013 [20]	SR	Not mentioned	niDCM	Clinical& Preclinical	44
Gorjipour F et al., 2021 [21]	SR&MA	Compatible	HF	Preclinical	3
Hoeeg C et al., 2020 [22]	SR	Compatible	niDCM	Clinical& Preclinical	27
Jeyaraman MM et al., 2017 [23]	SR&MA	Compatible	MI	Clinical	42
Jiao R et al., 2014 [24]	SR&MA	Compatible	DCM	Clinical	7
Kalou Y et al., 2023 [25]	SR&MA	Compatible	iDCM/niDCM	Clinical	6
Kang S et al., 2008 [26]	SR&MA	Not mentioned	MI	Clinical	6
∗Lalu MM et al., 2012 [27]	SR&MA	Not mentioned	MI/iDCM/niDCM	Clinical	6
Li Y et al., 2019 [28]	SR&MA	Compatible	iDCM/niDCM	Clinical	5
Lipinski MJ et al., 2007 [29]	SR&MA	Not mentioned	MI	Clinical	10
Lopes GM et al., 2020 [30]	SR&MA	Compatible	niDCM (AIC)	Preclinical	22
Lotfi F et al., 2020 [31]	SR&MA	Compatible	HF	Clinical	57
Lu Y et al., 2016 [32]	SR&MA	Compatible	niDCM	Clinical	7
Marquis-Gravel G et al., 2014 [33]	SR&MA	Compatible	niDCM	Clinical	3
Poulin MF et al., 2016 [34]	SR	Not mentioned	iDCM/niDCM	Clinical	29
Rong SL et al., 2019 [35]	SR&MA	Compatible	iDCM/niDCM	Clinical	8
†Sukho P et al., 2018 [36]	SR	Not mentioned	MI/niDCM	Preclinical	11
Tian T et al., 2014 [37]	SR&MA	Compatible	IHD	Clinical	11
Tripathi A et al., 2021 [38]	SR&MA	Compatible	niDCM/HF	Clinical	11
Van Der Spoel TIG et al., 2011 [39]	SR&MA	Not mentioned	iDCM	Preclinical	52
Wang Y et al., 2019 [40]	SR&MA	Compatible	iDCM	Clinical	14
Wen Y et al., 2018 [41]	SR&MA	Compatible	niDCM	Clinical	7
Xia L et al., 2020 [42]	SR&MA	Not mentioned	niDCM	Clinical	12
Xu Z et al., 2022 [43]	SR&MA	Compatible	HF	Clinical	43
Zwetsloot PP et al., 2016 [44]	SR&MA	Not mentioned	MI	Preclinical	80

AIC - antracycline-induced dilated cardiomyopathy; CBTs - cell-based therapies; HF - heart failure (if the underlying cause of heart failure cannot be determined, this statement is true); iDCM - ischaemic dilated cardiomyopathy; IHD - ischaemic heart disease; MI - myocardial infarction; niDCM - non-ischaemic dilated cardiomyopathy; PRISMA - Preferred Reporting Items for Systematic Reviews and Meta-Analyses; SR&MA - systematic review and meta-analysis.

In light of these considerations, the present meta-analysis systematically investigates preclinical mammalian data to evaluate the efficacy and safety of CBTs in experimental myocarditis models, as well as the limitations and challenges associated with large-scale clinical translation. In addition, particular attention is devoted to recent advances that offer new insights into preventive, diagnostic, and therapeutic strategies for myocarditis and miDCM.

2. Methods

2.1. Protocol and study registration

This meta-analysis was conducted according to the recommendations described in Cochrane Handbook for Systematic Reviews of Interventions [45] and reported in adherence with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines (Supplementary Table 1) [46,47]. The review protocol was registered in PROSPERO (CRD42023451561).

2.2. Search strategy and data sources

Search strategy objectives, including controlled vocabulary (if applicable) and keywords, were developed using PICO criteria by two authors (UY and BI). Subsequent searches were conducted in Embase, CINAHL Plus via EBSCO host, Web of Science, Scopus, and PubMed. The search strategy was adapted and optimized for each database. The last search was performed on December 24, 2023 and all searches were duplicated by an independent researcher. Additional publications were also obtained by manually searching for references to all included primary studies, relevant narrative and systematic reviews. Our search was limited to studies available in English only. Although there were no publication date restrictions, publication status and report type were taken into account in the search process, as unpublished materials and ongoing studies, as well as review and abstract articles, were excluded. All search syntaxes are provided in Supplementary Table 2.

2.3. Study selection

All retrieved reports were first imported into EndNote Online [48] to remove duplicates and then exported to Rayyan [49] for further processing. Two authors (UY and BI) independently checked the titles and abstracts in the first stage and then the full text in the second stage against the eligibility criteria in an unblinded, standardized manner. To resolve the disagreement between the two reviewers, either examination of the full text, consensus, or the opinion of the third author (KY) was considered. The following exclusion criteria were applied: non-miDCM etiologies (genetic, ischaemic, idiopathic, etc.); clinical and non-in vivo studies (ex vivo, in vitro, and in silico); studies involving relevant comorbidities or co-interventions; studies with missing data or lacking echocardiographic or histological outcomes; sham- or placebo-only groups; non-English publications; review articles, abstracts, unpublished material, letters to the editor, and ongoing studies.

2.4. Data extraction

Data from the selected studies were extracted independently by two authors (UY and BI) using predefined data fields. When results were presented only graphically and numerical data were unavailable, values were independently estimated by both authors using WebPlotDigitizer (version 4.6), and mean values were used for analysis. If key data were not reported in the text but were referenced elsewhere in the manuscript, the corresponding authors were contacted by email on two occasions (with a second email sent 4-8 weeks after the initial request if no response was received).

Extracted data included study design (randomization and blinding); participant characteristics (species, strain, sex, weight, age, disease model, and sample size); and intervention characteristics (cell source, type, and number; route, timing, and frequency of cell administration; and time to outcome assessment).

The primary outcome was changes in LV fractional shortening (LVFS, %) rather than LVEF (%), as non-uniform LV remodeling may result in incomplete or misleading estimates of LV function, particularly in small-animal models, making LVFS a more reliable measure [50,51]. Secondary outcomes included LVEF, inflammatory cell infiltration rate (ICIR, cells/mm²), fibrotic area (FA, %), and capillary density (CD, vessels/mm²).

For studies reporting multiple repeated measures of LVFS or LVEF, endpoint values (after the final cell injection and/or before sacrifice) were extracted. To adopt a conservative analytical approach, the highest values from control groups and the lowest values from treatment groups were selected for meta-analysis when multiple control and/or treatment groups were present within the same study. Discrepancies were resolved by consensus or, when necessary, by arbitration from a third author (KY).

2.5. Merging and contrasting the data

We conducted both meta-analysis and structured narrative synthesis of the included studies, taking into account study and control group characteristics, intervention types, and outcome measures. Heterogeneity in study design was evaluated to determine the appropriateness of data pooling. Studies were considered sufficiently homogeneous if they employed comparable species, disease models, cell types, and outcome measures. Forest plots were used to visually present pooled effect estimates and between-study variability.

2.6. Outcomes of interest

All outcomes of interest were continuous variables representing key functional and structural cardiac parameters, including LVFS, LVEF, ICIR, CD, and FA. Intervention effects in individual studies were summarized, and mean differences (MDs) were calculated to quantify these effects for continuous outcomes. Given that effect sizes derived from animal studies are not directly transferable to humans, the interpretation of results focused primarily on the direction, consistency, and heterogeneity of effects rather than on the absolute magnitude of effect sizes.

2.7. Risk of bias and quality assessment

The risk of bias (RoB) of included studies was assessed using the SYstematic Review Centre for Laboratory Animal Experimentation (SYRCLE) RoB tool for animal intervention studies (SYRCLE risk-of-bias tool), which comprises 10 items covering six domains of bias: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other sources of bias [52]. Study quality was additionally evaluated using the Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies (CAMARADES) checklist [53,54].

Studies were assessed against the ten CAMARADES criteria − sample size calculation [55], randomization [56], blinding [57], specification of inclusion/exclusion criteria [58], temperature control [59], compliance with animal welfare regulations [58], use of objective outcome measures [60], appropriate statistical analysis [56], declaration of conflicts of interest [61], and measures to prevent publication bias [61]. One point was awarded for each criterion met. Based on the total score, studies were provisionally classified as high quality (7-10 points), moderate quality (4-6 points), or low quality (0-3 points) [56].

2.8. Statistical analysis models

As the preliminary assessment indicated substantial heterogeneity among the included studies, a random-effects model was applied to pool effect sizes. This approach accounts for between-study variability inherent to preclinical animal research, which arises from differences in experimental conditions and exploratory study designs. Meta-analyses were conducted by calculating effect sizes for individual comparisons, estimating pooled effects, and assessing heterogeneity. All analyses were performed using R (version 4.3.0) with the meta package [62,63].

2.9. Heterogeneity

Between-study heterogeneity was assessed using the τ² (tau-squared) and I² statistics. Higher τ² values indicate greater between-study variance, while I² describes the proportion of total variability attributable to heterogeneity rather than chance. I² values of 0-40% were considered low or potentially unimportant heterogeneity, 30-60% moderate heterogeneity, 50-90% substantial heterogeneity, and 75-100% considerable heterogeneity [45]. Potential sources of heterogeneity were explored through predefined subgroup analyses.

2.10. Subgroup analysis

Subgroup analyses were conducted according to predefined criteria, including disease stage (acute, subacute, and chronic), disease model (viral, autoimmune, chemotherapy-induced myocarditis, protozoan-induced myocarditis, and miDCM), cell delivery method, cell type, animal species, and follow-up period. These analyses were performed to explore potential sources of heterogeneity across studies. Subgroup findings were interpreted cautiously and considered hypothesis-generating rather than confirmatory.

2.11. Publication bias and sensitivity analysis

Publication bias was assessed by visual inspection of funnel plots for asymmetry and statistically evaluated using Egger's regression test. Sensitivity analyses were performed using the trim-and-fill method to estimate the potential impact of unpublished or missing studies. This approach involved imputing potentially missing studies and recalculating pooled effect estimates to assess the robustness of the meta-analysis results to publication bias and selective reporting.

3. Results

3.1. Study selection

In total, the adapted and optimized search strategy retrieved 14,503 records from five databases (Supplementary Fig. 1) [64]. After removal of 7,395 duplicates, titles and abstracts of the remaining 7,108 records were independently screened by two authors (UY and BI). Records that did not fulfil the inclusion criteria described in the Methods section were excluded, leaving 116 studies after initial screening. Of these, 98 articles were assessed in full text, as 18 records were abstracts, posters, non-English publications, unavailable in full text, or otherwise inaccessible. Full-text screening was performed independently by the same two authors (UY and BI). Reference lists of included articles and relevant websites were also manually searched to identify additional eligible studies. Ultimately, 49 studies from the primary database search and 11 studies from supplementary searches were included in the analysis (Supplementary Fig. 1). Overall, 31, 37, 10, 5, and 29 studies reported outcomes for LVFS, LVEF, ICIR, CD, and FA, respectively.

3.2. Study characteristics

The main characteristics of the included studies are summarized in the Supplementary Material. Experimental animals were predominantly male; however, female animals were used either alone or in combination with males in some studies, and sex was not reported in several cases.

Analysis of publication trends indicates that early experimental applications of CBTs in myocarditis and miDCM began in the early 2000s. A total of 60 studies published between 2004 and 2022, authored by 55 research groups and appearing in 48 peer-reviewed journals, were included in this review [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [95], [96], [97], [98], [99], [100], [101], [102], [103], [104], [105], [106], [107], [108], [109], [110], [111], [112], [113], [114], [115], [116], [117], [118], [119], [120], [121], [122], [123], [124]. The journals Stem Cell Research & Therapy and Journal of Molecular and Cellular Cardiology published the highest number of studies, accounting for 5 (8.3%) and 3 (5%) publications, respectively, while Cell Cycle, Circulation, Circulation Journal, Frontiers in Immunology, Molecular Medicine Reports, and Stem Cells International each published two studies (3.3%). The remaining 40 journals accounted for 40 publications (66.7%).

Nearly half of the included studies (29/60) were published within a five-year period, with notable publication peaks in 2010, 2012, 2017, 2018, and 2022 (Fig. 1a). Geographically, China, Brazil, Japan, and Germany accounted for the majority of publications, contributing 24, 13, 10, and 4 studies, respectively (85% overall; Fig. 1a).

Fig. 1.

Characteristics of included studies: quantity of published studies by year and the contribution of each country (a); by cell administration methods and cell types (b); by disease model and stage (c); and by animal species and four follow-up periods (d). EHT was generated from neonatal rat cardiomyocytes (5 × 10⁶ cells) embedded in a three-dimensional extracellular matrix composed of type I collagen and Matrigel, cast into ring-shaped constructs and matured in vitro prior to transplantation [83]. iACS were constructed by differentiating syngeneic adipose tissue-derived stromal vascular fraction cells on temperature-responsive culture dishes, allowing spontaneous sheet detachment prior to implantation in rats with EAM [80]. QQMNCs + F-sheets consisted of Quality- and Quantity-controlled peripheral blood mononuclear cells co-cultured with syngeneic dermal fibroblasts on temperature-responsive dishes to form scaffold-free, multilayered cell sheets for transplantation [102]. AMI - acute myocardial injury; BMM2 - bone marrow derived M2-like macrophages; BMNCs - bone marrow mononuclear cells; CBTs - cell-based therapies; CDCs - cardiosphere derived cells, CMs - cardiomyocytes; CVB3 - coxsackievirus B3; DCM - dilated cardiomyopathy; DCs - dendritic cells; DOX - doxorubicin; EAM - experimental autoimmun myocarditis; EHT - engineered heart tissue; FRZ - furazolidone; iACS - scaffold-free-induced adipocyte cell-sheet; MSCs - mesenchymal stem/stromal cells; PPCs - prominin-1 expressing progenitor cells; QQMNCs - Quality- and Quantity-controlled peripheral blood mononuclear cells; SMCs - smooth muscle cells; SMI - subacute myocardial injury;T. cruzi- Trypanosoma Cruzi; TRCs - T regulatory cells.

Considerable heterogeneity was observed in both cell type and route of cell administration. Among the six identified delivery routes, intramyocardial and intravenous administration were most frequently used (Fig. 1b). Less commonly employed approaches included engineered heart tissue (EHT) transplantation and intramuscular injection, as reported by Leontyev et al. and Mao et al., respectively [83,89].

In line with our previous findings [1], mesenchymal stem cells (MSCs) and bone marrow mononuclear cells (BMNCs) were the most frequently investigated cell types, used in 36 (60%) and 10 (16.7%) studies, respectively. However, the scope of investigation was not limited to stem cells alone. Other cell types, including dendritic cells (DCs), regulatory T cells (TRCs), and bone marrow-derived M2-like macrophages (BMM2), were evaluated in four studies (Fig. 1b).

Regarding disease triggers and stages, doxorubicin (DOX)-induced and Trypanosoma cruzi (T. cruzi)-induced DCM accounted for 39 (65%) studies, while experimental autoimmune myocarditis (EAM)-induced acute myocardial injury (AMI) was investigated in 10 (16.7%) studies. Only one study evaluated MSCs in a DOX-induced subacute myocardial injury (SMI) model (Fig. 1c).

Small rodents predominated among experimental species, with rats and mice used in 31 (51.7%) and 20 (33.3%) studies, respectively, while rabbits and sheep were used less frequently (Fig. 1d). To harmonize follow-up durations across studies, observation periods were categorized as ≤1 month, 1–3 months, 3–6 months, or >6 months. Most studies reported relatively short follow-up periods of ≤1 month (25%) or 1–3 months (50%) (Fig. 1d).

3.3. Risk of bias and study quality

RoB and study quality were independently assessed by two authors (UY and BI) using the six domains of the SYRCLE risk-of-bias tool and the ten-item CAMARADES checklist, respectively, as described in the Methods section. The majority of studies exhibited an unclear or high risk of bias in several domains, particularly selection bias − due to inadequate reporting of random sequence generation, allocation concealment, and baseline characteristics − and performance bias, mainly related to the lack of randomized housing and blinding of caregivers or investigators (Fig. 2). In contrast, most studies demonstrated a low risk of bias in the domains of detection, attrition, reporting, and other sources of bias (Fig. 2).

Fig. 2.

Risk of bias (RoB) assessment by SYstematic Review Centre for Laboratory animal Experimentation (SYRCLE) risk-of-bias tool for animal intervention studies. Six domains of bias were evaluated in each study. Almost all included studies demonstrated either unclear or high risk in the areas of Selection and Performance bias. Only one paper by Jin B. et al., 2010 [79] described sequence generation, while 5 out of 60 studies provided detailed baseline characteristics. Studies were heterogeneous in terms of the other four areas of bias: Detection, Attrition, Reporting, and Other sources of bias.

Assessment using the CAMARADES checklist yielded consistent results: only four studies (6.7%) were classified as high quality and four (6.7%) as low quality, while the remaining 52 studies (86.7%) were rated as moderate quality, with scores ranging from 4 to 6 (Fig. 3). Although randomization was reported in most studies, none included a sample size calculation or addressed strategies to prevent publication bias. In addition, reporting of blinded outcome assessment, inclusion/exclusion criteria, and temperature control was unclear in the majority of studies. Overall, however, most studies were classified as moderate quality, largely due to adequate reporting of objective outcome measures, compliance with animal welfare regulations, appropriate statistical analyses, and declarations of conflicts of interest (Fig. 3).

Fig. 3.

Quality assessment across studies by the Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies (CAMARADES) checklist. Ten criteria were used to assess the quality of the studies. Overall, the majority of studies (52 out of 60) were of moderate quality, rated between 4 and 6 points, while the remaining 8 studies demonstrated high (4 studies) and low (4 studies) quality. None of the studies reported sample size calculations or took measures to prevent publication bias.

3.4. Meta-analysis of outcomes

All included studies provided extractable data in either textual or graphical form and were therefore eligible for meta-analysis. Meta-analyses were conducted for five outcomes: LVFS, LVEF, ICIR, CD, and FA. Studies grouped by outcome are summarized in Supplementary Table 3, and detailed results for each outcome are presented below.

For each study, MDs between experimental (treatment) and control groups were calculated. A random-effects model was applied to pool effect estimates and to account for between-study variability.

3.5. LVFS was significantly improved after CBTs

The meta-analysis included 31 studies comparing the effects of various CBTs on LVFS. The pooled MD across all studies was 7.17 [95% CI: 5.67, 8.66], indicating a statistically significant improvement in cardiac function in the treatment groups compared with control groups (Fig. 4). The diamond representing the overall effect was positioned to the right of the line of no effect (zero), suggesting that, on average, LVFS was 7.17% higher in the treatment groups than in the control groups.

Fig. 4.

Effect of cell-based therapies (CBTs) on left ventricular fractional shortening (LVFS). CBTs improved LVFS (%) by 7.17 [5,8,99,100]. Despite significant heterogeneity, the 95% confidence interval (CI) of most studies did not cross the midline. The sample size varied from 3 to 17 animals in the treatment and control groups.

Nearly two thirds of the included studies reported a statistically significant increase in LVFS in favor of the treatment groups, as indicated by confidence intervals that did not cross zero. In contrast, although the remaining studies also showed MDs favoring the treatment groups, their confidence intervals crossed zero, indicating no statistically significant difference between groups.

Substantial heterogeneity was observed among the included studies, with an I² value of 84%, suggesting considerable variability in effect estimates. This was further supported by the heterogeneity statistics (Tau² = 11.987; Chi² = 188.54; p < 0.01), indicating that the observed variability in effect sizes was statistically significant. To explore potential sources of heterogeneity, subgroup analyses were performed based on animal species (AS), disease models (DM), disease stages (DS), route of administration (RA), cell types (CT), and observation periods (OP) (Supplementary Fig. 2a–f). Subgroups characterized by intramyocardial injection (5.86 [95% CI: 4.27, 7.44]), BMNCs (4.14 [95% CI: 1.76, 6.52]), cardiosphere-derived cells (3.14 [95% CI: −5.33, 11.61]), and follow-up periods of 3-6 months (5.47 [95% CI: 2.21, 8.73]) exhibited non-significant heterogeneity, with I² values of 32%, 0%, 0%, and 0%, respectively.

Notably, substantial contributions to the overall pooled effect size of 7.17 [95% CI: 5.67, 8.66] were observed in several subgroups, including MSCs (7.82 [95% CI: 6.13, 9.51]); disease models induced by DOX (6.58 [95% CI: 4.68, 8.49]) and EAM (7.88 [95% CI: 4.66, 11.11]); disease stages corresponding to DCM (7.31 [95% CI: 5.69, 8.93]) and AMI (7.24 [95% CI: 3.73, 10.76]); intravenous cell administration (7.67 [95% CI: 5.22, 10.12]); follow-up periods of 1-3 months (7.77 [95% CI: 6.08, 9.47]) and up to 1 month (5.63 [95% CI: 2.90, 8.36]); and animal species including rats (8.03 [95% CI: 6.34, 9.73]) and mice (5.88 [95% CI: 3.17, 8.60]).

However, pronounced heterogeneity persisted within these subgroups, with I² values of 86% for MSCs; 87% and 80% for DOX- and EAM-induced models, respectively; 80% and 88% for DCM and AMI stages, respectively; 90% for intravenous administration; 75% and 77% for follow-up periods of 1-3 months and up to 1 month, respectively; and 78% for both mouse and rat models.

3.6. CBTs significantly improved LVEF

The effect of CBTs on LVEF was also favorable, with a pooled MD of 9.00 [95% CI: 7.03, 10.97], consistent with the results discussed above (Fig. 5). A total of 37 studies were included in this meta-analysis, and most of them (22 of 37) demonstrated a statistically significant improvement in LVEF, as their confidence intervals did not cross the line of no effect (zero). As observed for LVFS, this analysis revealed substantial heterogeneity among studies, with an I² value of 86%. This finding was supported by additional heterogeneity statistics (Tau² = 25,830; Chi² = 263.63; df = 36; p < 0.01).

Fig. 5.

Effect of cell-based therapies (CBTs) on left ventricular ejection fraction (LVEF). CBTs improved LVEF (%) by 9.00 [7.03, 10.97] with significant study heterogeneity I² = 86%. Almost 2/3 of the studies showed statistically significant results as the corresponding 95% confidence intervals did not intersect the middle zero line.

The contribution of individual studies to the overall pooled MD of 9.00 [95% CI: 7.03, 10.97] was relatively evenly distributed, with study weights ranging from 0.9% to 3.8%. While some studies, such as those by Guarita-Souza et al. [75] and Deng et al. [72] reported improvements exceeding 20% in the experimental groups, while reports by Brasil et al. [66] and Huang et al. [77] showed MDs below zero within the 95% confidence interval.

Subgroup analyses (Supplementary Fig. 3a–f) based on AS, DM, DS, RA, CT, and OP were conducted to clarify potential sources of heterogeneity and the contribution of each covariate to the pooled effect. As observed in the LVFS analysis, subgroups involving MSCs (8.92 [95% CI: 6.46, 11.38]; I² = 89%), DOX-induced models (8.30 [95% CI: 5.87, 10.74]; I² = 88%), DCM stage (9.27 [95% CI: 7.10, 11.44]; I² = 87%), intramyocardial injection (8.67 [95% CI: 6.11, 11.24]; I² = 87%), follow-up periods of 1-3 months (9.52 [95% CI: 7.15, 11.90]; I² = 81%), and rat models (10.34 [95% CI: 7.69, 12.99]; I² = 89%) were major contributors to both the pooled MD and overall heterogeneity.

The sheep and rabbit subgroups demonstrated significant effect sizes of 6.86 [95% CI: 0.01, 13.70] and 8.45 [95% CI: 6.70, 10.19], respectively, with no observed heterogeneity (I² = 0%). However, these subgroups accounted for only 4% and 16.5% of the total pooled effect, respectively (Supplementary Fig. 3a). Similarly, significant effect sizes of 8.84 [95% CI: 6.32, 11.36] and 6.10 [95% CI: 4.13, 8.07] were identified in the intraperitoneal injection and BMNCs subgroups, respectively, both with I² = 0% heterogeneity (Supplementary Fig. 3d–e).

3.7. Fewer inflammatory cells but increased capillary density in LVs treated with CBTs

Two additional histological parameters, ICIR and CD, were also analyzed to compare experimental and control groups. Compared with LVFS and LVEF, meta-analyses of ICIR and CD included a relatively small number of studies (10 and 5, respectively), as only a minority of the included studies reported these outcomes. Moreover, only data reported using the standardized units defined in the Methods section (inflammatory cells/mm² and vessels or capillaries/mm², respectively) were included in the analysis, while studies using alternative units were excluded.

On average, the number of infiltrating inflammatory cells in the myocardium was 178.99 cells/mm² lower in the CBT-treated groups than in the control groups (−178.99 [95% CI: −225.91, −132.08]; Fig. 6a). In parallel, capillary density was increased by 300.50 vessels/mm² following CBTs (300.50 [95% CI: 45.01, 555.99]; Fig. 6b).

Fig. 6.

Effect of cell-based therapies (CBTs) on inflammatory cell infiltration rate (ICIR) and capillary density (CD). ICIR (a) and CD (b) were driven by CBTs towards a better pattern, as shown by the figures of −178.99 [95% CI: −225.91, −132.08] and 300.50 [95% CI: 45.01, 555.99], respectively, indicating fewer inflammatory cells and more capillaries per millimetre square of myocardium. As with the other results of interest, there was considerable heterogeneity.

Both analyses exhibited considerable heterogeneity, as reflected by Tau² = 3517.621, Chi² = 986.02, df = 9 (p < 0.01), and I² = 99% for ICIR, and Tau² = 61,893.641, Chi² = 61.89, df = 4 (p < 0.01), and I² = 94% for CD. Potential contributors to this high between-study variability were explored through subgroup analyses, and the observed patterns were generally consistent with those identified in earlier outcome analyses.

Notably, ICIR was predominantly assessed (6 of 10 studies) in investigations evaluating MSCs, BMNCs, and DCs in murine models of DCM induced by T. cruzi, with most of these studies employing follow-up periods longer than six months (Supplementary Fig. 4a–f). Among the ICIR subgroup analyses, only the EAM subgroup demonstrated robust results, with a MD of −2758.47 [95% CI: −3043.09, −2473.85] and no detectable heterogeneity (I² = 0%; Supplementary Fig. 4c).

CD was assessed exclusively in rat models of EAM (Supplementary Fig. 5a–f) following treatment with scaffold-free-induced adipocyte cell-sheet (iACS), MSCs, and peripheral blood fibroblasts (PBFs). Among these, only studies using MSCs showed consistently favorable results, with all confidence intervals fully favoring treatment (393.66 [95% CI: 226.01, 561.31]) and moderate heterogeneity (I² = 67%; Supplementary Fig. 5e). No heterogeneity was observed in the AMI (315.35 [95% CI: 220.26, 410.44]; I² = 0%; Supplementary Fig. 5b), cell-sheet transplantation (CST) (9.86 [95% CI: −9.22, 28.95]; I² = 0%; Supplementary Fig. 5d), or follow-up periods shorter than one month (329.68 [95% CI: 203.82, 455.53]; I² = 0%; Supplementary Fig. 5f) subgroups.

3.8. Smaller FA was observed in CBT-treated groups

Evaluation of FA after CBTs in animal models was also of interest in this study, as the extent and severity of interstitial myocardial fibrosis following ischaemic or non-ischaemic myocardial injury is a critical factor for predicting patient outcomes and selecting appropriate therapeutic strategies in the clinical setting. A meta-analysis of 29 preclinical studies, including 267 animals in the experimental groups and 257 animals in the control groups, demonstrated that FA decreased by 6.04% [95% CI: −6.83, −5.25] following CBTs (Fig. 7).

Fig. 7.

Effect of cell-based therapies (CBTs) on fibrotic area (FA). A similar positive therapeutic effect was observed during FA evaluation as well. CBTs could reduce FA up to 6% (−6.04 [95% CI: −6.83, −5.25]) compared with control groups. Heterogeneity was as high as 97%.

The magnitude of FA reduction varied considerably among individual studies, ranging from the smallest decrease of 1.87% to the largest decrease of 22.53%, corresponding to almost one quarter of the total myocardial area. Only one study, by Wang et al. [111] reported an increase in FA of 1.88% [95% CI: 0.31, 3.45]. Consistent with the other outcomes, heterogeneity was markedly high (I² = 97%), supported by additional heterogeneity statistics (Tau² = 3.006; Chi² = 1111.44, df = 28, p < 0.01), indicating substantial methodological diversity among the included studies.

Cell type-oriented subgroup analysis revealed that BMNCs were more effective in reducing FA than MSCs (−10.66 [95% CI: −19.89, −1.44], I² = 93% vs. −5.45 [95% CI: −6.29, −4.61], I² = 98%, respectively; Supplementary Fig. 6e). Comparison of DCM and AMI models showed relatively similar reductions in FA, with values of −6.27 [95% CI: −7.83, −4.71] and −8.63 [95% CI: −13.22, −4.03], respectively (Supplementary Fig. 6b).

Unexpectedly, contrary to the prevailing view that intramyocardial cell injection provides superior benefits compared with intravenous or intraperitoneal routes, our results demonstrated a greater reduction in FA following intravenous and intraperitoneal cell administration than after intramyocardial delivery (−7.61 [95% CI: −10.17, −5.05] and −6.77 [95% CI: −10.33, −3.21] vs. −2.37 [95% CI: −3.84, −0.91], respectively). Notably, CST provided the greatest benefit, with an estimated total FA reduction of 16.15% [95% CI: −19.39, −12.91] across two studies (Supplementary Fig. 6d).

Among the four conditional OPs, no substantial differences in FA reduction were observed, except for the 3-6 month follow-up period. This finding may reflect the rapid replacement of myocardial inflammation by interstitial fibrosis during the early stages of disease progression. Nevertheless, the greatest decrease in FA (−9.79 [95% CI: −16.71, −2.87]) occurred during the 3-6 month follow-up, although substantial heterogeneity remained (I² = 99%) (Supplementary Fig. 7f).

Regarding animal species, rat and mouse hearts responded similarly to CBTs, with FA reductions of 6.83% [95% CI: −9.47, −4.18] and 5.42% [95% CI: −6.21, −4.63], respectively. In contrast, the only study conducted in sheep estimated a smaller FA reduction of 1.34% [95% CI: −2.23, −0.45], while results from rabbit experiments were statistically unsignificant (Supplementary Fig. 6a). Among all subgroup analyses, only CST demonstrated no heterogeneity (I² = 0%), although it contributed just 3.5% to the total effect size (Supplementary Fig. 6d).

3.9. Publication bias assessment

Publication bias was assessed using funnel plots, Egger's regression test, and the trim-and-fill method for each outcome separately, provided that the corresponding meta-analysis included at least 10 studies, as fewer studies do not allow reliable assessment. Overall, moderate publication bias was observed across the meta-analyses (Supplementary Figs. 7–10).

Funnel plot asymmetry was particularly evident for LVFS outcomes, with a greater number of studies appearing on the right side of the plot, indicating larger or more positive MDs (Supplementary Fig. 7). Application of the trim-and-fill method yielded similar findings, with imputed studies added predominantly to the right side of the funnel, suggesting a potential publication bias favoring studies reporting larger effect sizes (Supplementary Fig. 7). However, Egger's regression test indicated that publication bias may not have a strong influence on the overall results, as p-values remained above the conventional 0.05 threshold, and no statistically significant bias was detected even when the probability of publication of studies with high standard errors of the mean decreased.

In addition to true publication bias, funnel plot asymmetry may also be attributable to substantial heterogeneity, which must be considered in the interpretation of these findings, given the consistently high heterogeneity observed across analyses [125]. Overall, the risk of publication bias was assessed as moderate and may have led to a slight overestimation of treatment effects, with studies reporting smaller or non-significant effects likely under-represented in the literature (Supplementary Fig. 7).

Assessment of publication bias for the remaining outcomes − LVEF, ICIR, and FA (CD was excluded due to the limited number of studies) − revealed similar patterns (Supplementary Figs. 8–10). These findings suggest that the observed treatment effects may be overestimated owing to selective publication of studies with more favorable or statistically significant results, while studies reporting little or no benefit may have been omitted from the published literature (Supplementary Figs. 8–10).

4. Discussion

The aim of this systematic review and meta-analysis was to identify the cumulative effect of CBTs on myocarditis and miDCM, as well as to discuss the problems and limitations associated with translating the results into clinical practice. Although CBTs have been considered one of the potential therapeutic strategies for the treatment of myocarditis and miDCM for decades, as of today, CBTs have not been successfully implemented in large-scale clinical trials for several reasons, such as poor migration to the heart, engraftment, and differentiation of the administered cells; lack of information on suitable cell types and their quantity; clinically inappropriate cell delivery routes (except for intravenous injection) and timing; difficulties in making an accurate diagnosis; as well as complexity of disease pathophysiology and host-injected cell interaction. Therefore, in the present study, we attempted to address these issues by systematically reviewing the available data on CBTs testing for the treatment of myocarditis and miDCM. In this context, the current meta-analysis is the first to systematically evaluate the efficacy of CBTs on myocarditis and subsequent DCM in both large and small animal models in terms of LV function and architecture.

Our findings show that CBTs led to sustained improvement in cardiac function and myocardial architecture in preclinical mammalian models of myocarditis and miDCM. Across a wide range of experimental settings, CBTs significantly improved LVFS and LVEF (Fig. 4, Fig. 5, respectively), reduced inflammatory cell recruitment into the myocardium (Fig. 6a), promoted neovascularization (Fig. 6b), and reduced FA (Fig. 7). Despite varying degrees of effectiveness observed in the included studies, the overall direction of effect consistently favored CBTs in terms of functional and structural outcomes. It is noteworthy that the significant reduction in myocardial fibrosis in the groups receiving CBTs was one of the most compelling findings of our study, suggesting that modulation of adverse cardiac remodeling may represent a key mechanism by which CBTs mediates sustained functional benefits in models of myocarditis and miDCM. While expected in experimental interventional studies, enhancements in LVFS and LVEF were confirmed by pooled estimates across multiple disease models, disease stages, cell types, cell delivery routes, and animal species, which strength the biological rationale of CBT-mediated myocardial protection. Although intravenous injection − the most clinically applicable method of cell delivery − was used in the majority of preclinical studies and was associated with beneficial effects, direct transplantation of iACS (generated by differentiating adipose tissue-derived syngeneic stromal vascular-fraction cells into adipocytes on temperature-responsive dishes) [80], QQMNCs + F-sheet (skin fibroblasts isolated from EAM rats were cultured on temperature-responsive dishes to form confluent cell sheets and then co-cultured with QQMNCs) [102], and EHT (generated from neonatal rat cardiomyocytes [5 × 10⁶ cells] embedded in a three-dimensional extracellular matrix composed of type I collagen and Matrigel, cast into ring-shaped constructs and matured in vitro prior to transplantation) [83] were associated with larger pooled effect estimates for LV function and architecture compared to other routes of administration (Supplementary figures 2d-4d, and 6d).

It is also proved that MSCs were the most preferred stem cells in the reports, probably due to their availability and well-characterized properties. However, our results did not show a large difference between early (during AMI) and late (during DCM) cell delivery schedules, as LV function remained at almost the same level (Supplementary Figs. 2b and 3b) after CBTs. In contrast, early CBTs led to a reduced myocardial fibrosis (Supplementary Fig. 6b) and less severe inflammatory cell infiltration (Supplementary Fig. 4b), rather than stimulating angiogenesis (Supplementary Fig. 5b). This confirms that CBT-mediated cardioprotection was mainly driven by paracrine mechanisms, including systemic release of potent pro-reparative, anti-fibrotic, and anti-inflammatory agents from cells, rather than myocardial migration and engraftment [8].

The results of the included studies were significantly heterogeneous, reflecting variations in disease models, disease stages, cell types, routes of administration, animal species, and follow-up time, which is an inevitable characteristics of preclinical studies of complex inflammatory diseases such as myocarditis.

Although observed heterogeneity limits direct quantitative comparisons of individual CBTs, it is important to note that the consistency in the effect direction across different models suggests that the observed benefits are unlikely to be explained solely by random variation or selective publication. This was entirely in line with the main objective of this meta-analysis, which focused on the direction and consistency of effects in various preclinical models of myocarditis and miDCM.

The assessment of RoB and study quality offers additional information for interpreting the pooled results. Most studies showed unclear or high risks in the selection and performance biases, mainly due to insufficient coverage of randomization, concealed allocation, and blinding (Fig. 2). In contrast, the risk of detection, attrition, and reporting biases were generally low (Fig. 2). The predominance of moderate-quality studies (52 of 60) highlights the exploratory nature of CBTs in the treatment of myocarditis and warns of the need for a more rigorous methodology in future preclinical studies (Fig. 3). Importantly, this approach is one of the main strengths of the present study, as a thorough assessment of the risk of systematic error and study quality not only ensures transparency but also provides a basis for improving experimental designs in future CBT research.

As predicted, even after applying random effects models and subgroup analyses, considerable heterogeneity persisted, particularly with regard to ICIR and CD. This variability is likely related to differences in immune responses, disease severity, biological characteristics of the host organism, and histological evaluation methods. Therefore, these results should be interpreted as hypothetical clues for future research rather than definitive conclusions, highlighting the need for targeted, well-designed studies focused on optimized CBTs.

Despite promising preclinical results, there are a number of issues that hinder the transfer of CBTs from the laboratory to clinical practice. Small sample sizes, methodological heterogeneity, and selective reporting likely contribute to overestimation of effect sizes. On the other hand, differences between studies may reflect the wide range of patient characteristics in clinical settings and therefore, serve as a strong point [21]. The tendency to publish only studies with favorable results may further complicate interpretation. In addition, complex therapeutic constructions such as EHT introduce additional variables related to cell composition, extracellular matrix scaffolding, and mechanical integration that cannot be directly compared to single-cell injections.

5. Emerging therapeutic, diagnostic, and preventive strategies for myocarditis

The recent therapeutic, diagnostic, and preventive strategies for myocarditis have increasingly focused on underlying pathophysiological mechanisms and/or downstream signaling pathways. One emerging strategy involves modulation of neutrophil extracellular traps (NETs) or NETosis, whose direct role in cardiac remodeling has not yet been fully elucidated, but which has been implicated in Coxsackievirus B3 (CVB3)-induced myocarditis [126]. PANoptosis − a form of programmed cell death that integrates features of pyroptosis, apoptosis, and necroptosis − may also play a role in CVB3- and SARS-CoV-2-associated myocarditis, potentially mediated by reactive oxygen species overproduction and cytokine storm, respectively [127].

Gasdermin (GSDM) family proteins, particularly the GSDME-cGAS-STING axis, have also emerged as potential targets, since suppression of this pathway reduced mitochondrial DNA leakage, interferon signaling, antigen presentation, and neutrophil degranulation in Gsdme^−/− mice treated with anti-PD-1 antibodies [128]. In parallel, baricitinib, a Janus kinase (JAK) inhibitor, was shown to alleviate ICI-myocarditis in animal models by improving cardiac function and histological outcomes, primarily through inhibition of the JAK1-STAT3 pathway and polarization of macrophages toward an M2 phenotype [129].

Study investigating mitochondrial involvement in T. cruzi-induced myocarditis showed that the mitochondrial uncoupler 2,4-dinitrophenol administration exacerbates Chagas disease due to ineffective control of parasitemia and parasite load [130], while activation of peroxisome proliferator-activated receptors (PPARs), in particular PPAR-⍺, by fenofibrate, a hypolipidemic and anti-inflammatory agent in clinical practice, can affect the ratio and polarization of macrophages, inhibit pro-inflammatory cytokines and reduce cardiac fibrosis in both acute and chronic phases of T. cruzi infected murine hearts [131].

A retrospective study of 105 patients demonstrated that serum levels of matrix metalloproteinase-1 and procollagen type I C-terminal propeptide may aid in differentiating infarct-like myocarditis from non-ST-segment elevation myocardial infarction, achieving sensitivities and specificities of 94.4% and 90.9%, and 89.5% and 71.4%, respectively [132]. Preventive strategies are also evolving; notably, intranasal immunization with an attenuated CVB3(μ) strain induced robust mucosal and systemic immunity in mice, protected against severe disease and weight loss, and conferred passive vertical immunity to offspring [133].

6. Conclusion

The existing data indicate that research on the use of CBTs in the treatment of myocarditis and miDCM is still in its early stages, and more thorough preclinical studies should be considered in the future. Despite significant heterogeneity and moderate publication bias, the results were very encouraging, as reflected in the consistent effect directions across all studies in terms of cardiac function and histology, particularly myocardial fibrosis. Given the central role of fibrosis in ventricular stiffness and irreversible decline in function, this observation may be highly relevant for translational applications. Overall, our results demonstrated the need for well-designed studies with adequate animal sample sizes, a standardized approach to reporting, and mechanistic investigations that directly link structural remodeling to functional recovery. Addressing these issues will be critical steps for conducting large-scale clinical trials.

7. Study limitations

This study has several limitations. First, a possible underreporting of unfavorable results − moderate publication bias was identified, which makes it difficult to elucidate the real efficacy of CBTs in the treatment of myocarditis and miDCM. Second, the reporting quality of most of the included studies did not meet important core quality indicators such as sample size calculation, blinding of caregivers, investigators/outcome assessors, accurate inclusion/exclusion criteria, control of vital physiological parameters, declaration of conflict of interest, and measures to prevent publication bias, which may further complicate interpretation of pooled effect sizes. Third, almost all studies showed unclear or high risk of selection and outcome biases, indicating a potential impact on the reliability of the results. Fourth, since only a few reports, if any, can meet the strict criteria regarding cell type, route of administration, disease model, animal species, and follow-up time, this study covered all possible reports on CBT research in mammalian models of myocarditis and miDCM. Finally, only works in English were considered for the meta-analysis.

Ethical approval

Not applicable.

Disclosures

None.

Data availability statement

The datasets underlying current study are included in this manuscript and its Supplementary Material.

Author contribution

UY conceptualized the research question and methodology; retrieved, analyzed and interpreted the data; and wrote the initial draft of the manuscript; KY resolved disputes and critically revised the manuscript; SK developed codes for statistical analyses and critically revised the manuscript; BI retrieved, analyzed and interpreted the data; TK and SS critically edited the manuscript; SM supervised and coordinated the study and critically revised the manuscript. All authors read and approved the final manuscript for submission. The authors declare that all co-authors fulfilling the authorship criteria were listed in the appropriate order and none were omitted.

Funding

This study did not receive funding.

Declaration of competing interest

None.

Appendix A. Supplementary data

The following are the Supplementary data to this article.

Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flowchart.

Subgroup analysis of left ventricular fractional shortening results by animal species.

Subgroup analysis of left ventricular fractional shortening results by disease stage.

Subgroup analysis of left ventricular fractional shortening results by disease model.

Subgroup analysis of left ventricular fractional shortening results by cell administration methods.

Subgroup analysis of left ventricular fractional shortening results by cell types.

Subgroup analysis of left ventricular fractional shortening results by follow-up period.

Subgroup analysis of left ventricular ejection fraction results by animal species.

Subgroup analysis of left ventricular ejection fraction results by disease stage.

Subgroup analysis of left ventricular ejection fraction results by disease model

Subgroup analysis of left ventricular ejection fraction results by cell administration methods

Subgroup analysis of left ventricular ejection fraction results by cell types.

Subgroup analysis of left ventricular ejection fraction results by follow-up period.

Subgroup analysis of inflammatory cell infiltration rate results by animal species

Subgroup analysis of inflammatory cell infiltration rate results by disease stage.

Subgroup analysis of inflammatory cell infiltration rate results by disease model.

Subgroup analysis of inflammatory cell infiltration rate results by cell administration methods.

Subgroup analysis of inflammatory cell infiltration rate results by cell type.

Subgroup analysis of inflammatory cell infiltration rate results by follow-up period.

Subgroup analysis of capillary density results by animal species.

Subgroup analysis of capillary density results by disease stage.

Subgroup analysis of capillary density results by disease model.

Subgroup analysis of capillary density results by cell administration methods.

Subgroup analysis of capillary density results by cell types.

Subgroup analysis of capillary density results by follow-up period.

Subgroup analysis of fibrotic area results by animal species.

Subgroup analysis of fibrotic area results by disease stage.

Subgroup analysis of fibrotic area results by disease model.

Subgroup analysis of fibrotic area results by cell administration methods.

Subgroup analysis of fibrotic area results by cell types.

Subgroup analysis of fibrotic area results by follow-up period.

Assessment of publication bias among studies on left ventricular fractional shortening using: (a) funnel plot, (b) trim-and-fill analysis, and (c) Egger's regression test.

Assessment of publication bias among studies on left ventricular fractional shortening using: (a) funnel plot, (b) trim-and-fill analysis, and (c) Egger's regression tes.

Assessment of publication bias among studies on inflammatory cell infiltration rate using: (a) funnel plot, (b) trim-and-fill analysis, and (c) Egger's regression test.

Assessment of publication bias among studies on fibrotic area using: (a) funnel plot, (b) trim-and-fill analysis, and (c) Egger's regression test.